lighthouse/validator_client/src/initialized_validators.rs

856 lines
35 KiB
Rust
Raw Normal View History

Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
//! Provides management of "initialized" validators.
//!
//! A validator is "initialized" if it is ready for signing blocks, attestations, etc in this
//! validator client.
//!
//! The `InitializedValidators` struct in this file serves as the source-of-truth of which
//! validators are managed by this validator client.
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
use crate::signing_method::SigningMethod;
use account_utils::{
read_password, read_password_from_user,
validator_definitions::{
self, SigningDefinition, ValidatorDefinition, ValidatorDefinitions, CONFIG_FILENAME,
},
ZeroizeString,
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
};
use eth2::lighthouse_vc::std_types::DeleteKeystoreStatus;
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
use eth2_keystore::Keystore;
use lighthouse_metrics::set_gauge;
use lockfile::{Lockfile, LockfileError};
use parking_lot::{MappedMutexGuard, Mutex, MutexGuard};
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
use reqwest::{Certificate, Client, Error as ReqwestError};
use slog::{debug, error, info, warn, Logger};
use std::collections::{HashMap, HashSet};
use std::fs::{self, File};
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
use std::io::{self, Read};
use std::path::{Path, PathBuf};
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
use std::sync::Arc;
use std::time::Duration;
Optimize validator duties (#2243) ## Issue Addressed Closes #2052 ## Proposed Changes - Refactor the attester/proposer duties endpoints in the BN - Performance improvements - Fixes some potential inconsistencies with the dependent root fields. - Removes `http_api::beacon_proposer_cache` and just uses the one on the `BeaconChain` instead. - Move the code for the proposer/attester duties endpoints into separate files, for readability. - Refactor the `DutiesService` in the VC - Required to reduce the delay on broadcasting new blocks. - Gets rid of the `ValidatorDuty` shim struct that came about when we adopted the standard API. - Separate block/attestation duty tasks so that they don't block each other when one is slow. - In the VC, use `PublicKeyBytes` to represent validators instead of `PublicKey`. `PublicKey` is a legit crypto object whilst `PublicKeyBytes` is just a byte-array, it's much faster to clone/hash `PublicKeyBytes` and this change has had a significant impact on runtimes. - Unfortunately this has created lots of dust changes. - In the BN, store `PublicKeyBytes` in the `beacon_proposer_cache` and allow access to them. The HTTP API always sends `PublicKeyBytes` over the wire and the conversion from `PublicKey` -> `PublickeyBytes` is non-trivial, especially when queries have 100s/1000s of validators (like Pyrmont). - Add the `state_processing::state_advance` mod which dedups a lot of the "apply `n` skip slots to the state" code. - This also fixes a bug with some functions which were failing to include a state root as per [this comment](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/consensus/state_processing/src/state_advance.rs#L69-L74). I couldn't find any instance of this bug that resulted in anything more severe than keying a shuffling cache by the wrong block root. - Swap the VC block service to use `mpsc` from `tokio` instead of `futures`. This is consistent with the rest of the code base. ~~This PR *reduces* the size of the codebase :tada:~~ It *used* to reduce the size of the code base before I added more comments. ## Observations on Prymont - Proposer duties times down from peaks of 450ms to consistent <1ms. - Current epoch attester duties times down from >1s peaks to a consistent 20-30ms. - Block production down from +600ms to 100-200ms. ## Additional Info - ~~Blocked on #2241~~ - ~~Blocked on #2234~~ ## TODO - [x] ~~Refactor this into some smaller PRs?~~ Leaving this as-is for now. - [x] Address `per_slot_processing` roots. - [x] Investigate slow next epoch times. Not getting added to cache on block processing? - [x] Consider [this](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/beacon_node/store/src/hot_cold_store.rs#L811-L812) in the scenario of replacing the state roots Co-authored-by: pawan <pawandhananjay@gmail.com> Co-authored-by: Michael Sproul <michael@sigmaprime.io>
2021-03-17 05:09:57 +00:00
use types::{Graffiti, Keypair, PublicKey, PublicKeyBytes};
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
use url::{ParseError, Url};
use validator_dir::Builder as ValidatorDirBuilder;
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
use crate::key_cache;
use crate::key_cache::KeyCache;
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
/// Default timeout for a request to a remote signer for a signature.
///
/// Set to 12 seconds since that's the duration of a slot. A remote signer that cannot sign within
/// that time is outside the synchronous assumptions of Eth2.
const DEFAULT_REMOTE_SIGNER_REQUEST_TIMEOUT: Duration = Duration::from_secs(12);
// Use TTY instead of stdin to capture passwords from users.
const USE_STDIN: bool = false;
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
#[derive(Debug)]
pub enum Error {
/// Refused to open a validator with an existing lockfile since that validator may be in-use by
/// another process.
Lockfile(LockfileError),
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// The voting public key in the definition did not match the one in the keystore.
VotingPublicKeyMismatch {
definition: Box<PublicKey>,
keystore: Box<PublicKey>,
},
/// There was a filesystem error when opening the keystore.
UnableToOpenVotingKeystore(io::Error),
UnableToOpenKeyCache(key_cache::Error),
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// The keystore path is not as expected. It should be a file, not `..` or something obscure
/// like that.
BadVotingKeystorePath(PathBuf),
BadKeyCachePath(PathBuf),
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// The keystore could not be parsed, it is likely bad JSON.
UnableToParseVotingKeystore(eth2_keystore::Error),
/// The keystore could not be decrypted. The password might be wrong.
UnableToDecryptKeystore(eth2_keystore::Error),
/// There was a filesystem error when reading the keystore password from disk.
UnableToReadVotingKeystorePassword(io::Error),
/// There was an error updating the on-disk validator definitions file.
UnableToSaveDefinitions(validator_definitions::Error),
/// It is not legal to try and initialize a disabled validator definition.
UnableToInitializeDisabledValidator,
/// There was an error while deleting a keystore file.
UnableToDeleteKeystore(PathBuf, io::Error),
/// There was an error while deleting a validator dir.
UnableToDeleteValidatorDir(PathBuf, io::Error),
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// There was an error reading from stdin.
UnableToReadPasswordFromUser(String),
/// There was an error running a tokio async task.
TokioJoin(tokio::task::JoinError),
/// Cannot initialize the same validator twice.
DuplicatePublicKey,
/// The public key does not exist in the set of initialized validators.
ValidatorNotInitialized(PublicKey),
/// Unable to read the slot clock.
SlotClock,
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
/// The URL for the remote signer cannot be parsed.
InvalidWeb3SignerUrl(String),
/// Unable to read the root certificate file for the remote signer.
InvalidWeb3SignerRootCertificateFile(io::Error),
InvalidWeb3SignerRootCertificate(ReqwestError),
UnableToBuildWeb3SignerClient(ReqwestError),
/// Unable to apply an action to a validator because it is using a remote signer.
InvalidActionOnRemoteValidator,
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
impl From<LockfileError> for Error {
fn from(error: LockfileError) -> Self {
Self::Lockfile(error)
}
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// A validator that is ready to sign messages.
pub struct InitializedValidator {
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
signing_method: Arc<SigningMethod>,
graffiti: Option<Graffiti>,
/// The validators index in `state.validators`, to be updated by an external service.
index: Option<u64>,
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
impl InitializedValidator {
/// Return a reference to this validator's lockfile if it has one.
pub fn keystore_lockfile(&self) -> Option<MappedMutexGuard<Lockfile>> {
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
match self.signing_method.as_ref() {
SigningMethod::LocalKeystore {
ref voting_keystore_lockfile,
..
} => MutexGuard::try_map(voting_keystore_lockfile.lock(), |option_lockfile| {
option_lockfile.as_mut()
})
.ok(),
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
// Web3Signer validators do not have any lockfiles.
SigningMethod::Web3Signer { .. } => None,
}
}
}
fn open_keystore(path: &Path) -> Result<Keystore, Error> {
let keystore_file = File::open(path).map_err(Error::UnableToOpenVotingKeystore)?;
Keystore::from_json_reader(keystore_file).map_err(Error::UnableToParseVotingKeystore)
}
fn get_lockfile_path(file_path: &Path) -> Option<PathBuf> {
file_path
.file_name()
.and_then(|os_str| os_str.to_str())
.map(|filename| file_path.with_file_name(format!("{}.lock", filename)))
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
impl InitializedValidator {
/// Instantiate `self` from a `ValidatorDefinition`.
///
/// If `stdin.is_some()` any missing passwords will result in a prompt requesting input on
/// stdin (prompts published to stderr).
///
/// ## Errors
///
/// If the validator is unable to be initialized for whatever reason.
async fn from_definition(
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
def: ValidatorDefinition,
key_cache: &mut KeyCache,
key_stores: &mut HashMap<PathBuf, Keystore>,
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
) -> Result<Self, Error> {
if !def.enabled {
return Err(Error::UnableToInitializeDisabledValidator);
}
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
let signing_method = match def.signing_definition {
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
// Load the keystore, password, decrypt the keypair and create a lockfile for a
// EIP-2335 keystore on the local filesystem.
SigningDefinition::LocalKeystore {
voting_keystore_path,
voting_keystore_password_path,
voting_keystore_password,
} => {
use std::collections::hash_map::Entry::*;
let voting_keystore = match key_stores.entry(voting_keystore_path.clone()) {
Vacant(entry) => entry.insert(open_keystore(&voting_keystore_path)?),
Occupied(entry) => entry.into_mut(),
};
let voting_keypair = if let Some(keypair) = key_cache.get(voting_keystore.uuid()) {
keypair
} else {
let keystore = voting_keystore.clone();
let keystore_path = voting_keystore_path.clone();
// Decoding a local keystore can take several seconds, therefore it's best
// to keep if off the core executor. This also has the fortunate effect of
// interrupting the potentially long-running task during shut down.
let (password, keypair) = tokio::task::spawn_blocking(move || {
Result::<_, Error>::Ok(
match (voting_keystore_password_path, voting_keystore_password) {
// If the password is supplied, use it and ignore the path
// (if supplied).
(_, Some(password)) => (
password.as_ref().to_vec().into(),
keystore
.decrypt_keypair(password.as_ref())
.map_err(Error::UnableToDecryptKeystore)?,
),
// If only the path is supplied, use the path.
(Some(path), None) => {
let password = read_password(path)
.map_err(Error::UnableToReadVotingKeystorePassword)?;
let keypair = keystore
.decrypt_keypair(password.as_bytes())
.map_err(Error::UnableToDecryptKeystore)?;
(password, keypair)
}
// If there is no password available, maybe prompt for a password.
(None, None) => {
let (password, keypair) = unlock_keystore_via_stdin_password(
&keystore,
&keystore_path,
)?;
(password.as_ref().to_vec().into(), keypair)
}
},
)
})
.await
.map_err(Error::TokioJoin)??;
key_cache.add(keypair.clone(), voting_keystore.uuid(), password);
keypair
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
};
if voting_keypair.pk != def.voting_public_key {
return Err(Error::VotingPublicKeyMismatch {
definition: Box::new(def.voting_public_key),
keystore: Box::new(voting_keypair.pk),
});
}
// Append a `.lock` suffix to the voting keystore.
let lockfile_path = get_lockfile_path(&voting_keystore_path)
.ok_or_else(|| Error::BadVotingKeystorePath(voting_keystore_path.clone()))?;
let voting_keystore_lockfile = Mutex::new(Some(Lockfile::new(lockfile_path)?));
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
SigningMethod::LocalKeystore {
voting_keystore_path,
voting_keystore_lockfile,
voting_keystore: voting_keystore.clone(),
voting_keypair: Arc::new(voting_keypair),
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
SigningDefinition::Web3Signer {
url,
root_certificate_path,
request_timeout_ms,
} => {
let signing_url = build_web3_signer_url(&url, &def.voting_public_key)
.map_err(|e| Error::InvalidWeb3SignerUrl(e.to_string()))?;
let request_timeout = request_timeout_ms
.map(Duration::from_millis)
.unwrap_or(DEFAULT_REMOTE_SIGNER_REQUEST_TIMEOUT);
let builder = Client::builder().timeout(request_timeout);
let builder = if let Some(path) = root_certificate_path {
let certificate = load_pem_certificate(path)?;
builder.add_root_certificate(certificate)
} else {
builder
};
let http_client = builder
.build()
.map_err(Error::UnableToBuildWeb3SignerClient)?;
SigningMethod::Web3Signer {
signing_url,
http_client,
voting_public_key: def.voting_public_key,
}
}
};
Ok(Self {
signing_method: Arc::new(signing_method),
graffiti: def.graffiti.map(Into::into),
index: None,
})
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
/// Returns the voting public key for this validator.
pub fn voting_public_key(&self) -> &PublicKey {
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
match self.signing_method.as_ref() {
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
SigningMethod::LocalKeystore { voting_keypair, .. } => &voting_keypair.pk,
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
SigningMethod::Web3Signer {
voting_public_key, ..
} => voting_public_key,
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
}
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
pub fn load_pem_certificate<P: AsRef<Path>>(pem_path: P) -> Result<Certificate, Error> {
let mut buf = Vec::new();
File::open(&pem_path)
.map_err(Error::InvalidWeb3SignerRootCertificateFile)?
.read_to_end(&mut buf)
.map_err(Error::InvalidWeb3SignerRootCertificateFile)?;
Certificate::from_pem(&buf).map_err(Error::InvalidWeb3SignerRootCertificate)
}
fn build_web3_signer_url(base_url: &str, voting_public_key: &PublicKey) -> Result<Url, ParseError> {
Url::parse(base_url)?.join(&format!("api/v1/eth2/sign/{}", voting_public_key))
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
/// Try to unlock `keystore` at `keystore_path` by prompting the user via `stdin`.
fn unlock_keystore_via_stdin_password(
keystore: &Keystore,
keystore_path: &Path,
) -> Result<(ZeroizeString, Keypair), Error> {
eprintln!();
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
eprintln!(
"The {} file does not contain either of the following fields for {:?}:",
CONFIG_FILENAME, keystore_path
);
eprintln!();
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
eprintln!(" - voting_keystore_password");
eprintln!(" - voting_keystore_password_path");
eprintln!();
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
eprintln!(
"You may exit and update {} or enter a password. \
If you choose to enter a password now then this prompt \
will be raised next time the validator is started.",
CONFIG_FILENAME
);
eprintln!();
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
eprintln!("Enter password (or press Ctrl+c to exit):");
loop {
let password =
read_password_from_user(USE_STDIN).map_err(Error::UnableToReadPasswordFromUser)?;
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
eprintln!();
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
match keystore.decrypt_keypair(password.as_ref()) {
Ok(keystore) => break Ok((password, keystore)),
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
Err(eth2_keystore::Error::InvalidPassword) => {
eprintln!("Invalid password, try again (or press Ctrl+c to exit):");
}
Err(e) => return Err(Error::UnableToDecryptKeystore(e)),
}
}
}
/// A set of `InitializedValidator` objects which is initialized from a list of
/// `ValidatorDefinition`. The `ValidatorDefinition` file is maintained as `self` is modified.
///
/// Forms the fundamental list of validators that are managed by this validator client instance.
pub struct InitializedValidators {
/// A list of validator definitions which can be stored on-disk.
definitions: ValidatorDefinitions,
/// The directory that the `self.definitions` will be saved into.
validators_dir: PathBuf,
/// The canonical set of validators.
Optimize validator duties (#2243) ## Issue Addressed Closes #2052 ## Proposed Changes - Refactor the attester/proposer duties endpoints in the BN - Performance improvements - Fixes some potential inconsistencies with the dependent root fields. - Removes `http_api::beacon_proposer_cache` and just uses the one on the `BeaconChain` instead. - Move the code for the proposer/attester duties endpoints into separate files, for readability. - Refactor the `DutiesService` in the VC - Required to reduce the delay on broadcasting new blocks. - Gets rid of the `ValidatorDuty` shim struct that came about when we adopted the standard API. - Separate block/attestation duty tasks so that they don't block each other when one is slow. - In the VC, use `PublicKeyBytes` to represent validators instead of `PublicKey`. `PublicKey` is a legit crypto object whilst `PublicKeyBytes` is just a byte-array, it's much faster to clone/hash `PublicKeyBytes` and this change has had a significant impact on runtimes. - Unfortunately this has created lots of dust changes. - In the BN, store `PublicKeyBytes` in the `beacon_proposer_cache` and allow access to them. The HTTP API always sends `PublicKeyBytes` over the wire and the conversion from `PublicKey` -> `PublickeyBytes` is non-trivial, especially when queries have 100s/1000s of validators (like Pyrmont). - Add the `state_processing::state_advance` mod which dedups a lot of the "apply `n` skip slots to the state" code. - This also fixes a bug with some functions which were failing to include a state root as per [this comment](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/consensus/state_processing/src/state_advance.rs#L69-L74). I couldn't find any instance of this bug that resulted in anything more severe than keying a shuffling cache by the wrong block root. - Swap the VC block service to use `mpsc` from `tokio` instead of `futures`. This is consistent with the rest of the code base. ~~This PR *reduces* the size of the codebase :tada:~~ It *used* to reduce the size of the code base before I added more comments. ## Observations on Prymont - Proposer duties times down from peaks of 450ms to consistent <1ms. - Current epoch attester duties times down from >1s peaks to a consistent 20-30ms. - Block production down from +600ms to 100-200ms. ## Additional Info - ~~Blocked on #2241~~ - ~~Blocked on #2234~~ ## TODO - [x] ~~Refactor this into some smaller PRs?~~ Leaving this as-is for now. - [x] Address `per_slot_processing` roots. - [x] Investigate slow next epoch times. Not getting added to cache on block processing? - [x] Consider [this](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/beacon_node/store/src/hot_cold_store.rs#L811-L812) in the scenario of replacing the state roots Co-authored-by: pawan <pawandhananjay@gmail.com> Co-authored-by: Michael Sproul <michael@sigmaprime.io>
2021-03-17 05:09:57 +00:00
validators: HashMap<PublicKeyBytes, InitializedValidator>,
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// For logging via `slog`.
log: Logger,
}
impl InitializedValidators {
/// Instantiates `Self`, initializing all validators in `definitions`.
pub async fn from_definitions(
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
definitions: ValidatorDefinitions,
validators_dir: PathBuf,
log: Logger,
) -> Result<Self, Error> {
let mut this = Self {
validators_dir,
definitions,
validators: HashMap::default(),
log,
};
this.update_validators().await?;
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
Ok(this)
}
/// The count of enabled validators contained in `self`.
pub fn num_enabled(&self) -> usize {
self.validators.len()
}
/// The total count of enabled and disabled validators contained in `self`.
pub fn num_total(&self) -> usize {
self.definitions.as_slice().len()
}
/// Iterate through all voting public keys in `self` that should be used when querying for duties.
Optimize validator duties (#2243) ## Issue Addressed Closes #2052 ## Proposed Changes - Refactor the attester/proposer duties endpoints in the BN - Performance improvements - Fixes some potential inconsistencies with the dependent root fields. - Removes `http_api::beacon_proposer_cache` and just uses the one on the `BeaconChain` instead. - Move the code for the proposer/attester duties endpoints into separate files, for readability. - Refactor the `DutiesService` in the VC - Required to reduce the delay on broadcasting new blocks. - Gets rid of the `ValidatorDuty` shim struct that came about when we adopted the standard API. - Separate block/attestation duty tasks so that they don't block each other when one is slow. - In the VC, use `PublicKeyBytes` to represent validators instead of `PublicKey`. `PublicKey` is a legit crypto object whilst `PublicKeyBytes` is just a byte-array, it's much faster to clone/hash `PublicKeyBytes` and this change has had a significant impact on runtimes. - Unfortunately this has created lots of dust changes. - In the BN, store `PublicKeyBytes` in the `beacon_proposer_cache` and allow access to them. The HTTP API always sends `PublicKeyBytes` over the wire and the conversion from `PublicKey` -> `PublickeyBytes` is non-trivial, especially when queries have 100s/1000s of validators (like Pyrmont). - Add the `state_processing::state_advance` mod which dedups a lot of the "apply `n` skip slots to the state" code. - This also fixes a bug with some functions which were failing to include a state root as per [this comment](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/consensus/state_processing/src/state_advance.rs#L69-L74). I couldn't find any instance of this bug that resulted in anything more severe than keying a shuffling cache by the wrong block root. - Swap the VC block service to use `mpsc` from `tokio` instead of `futures`. This is consistent with the rest of the code base. ~~This PR *reduces* the size of the codebase :tada:~~ It *used* to reduce the size of the code base before I added more comments. ## Observations on Prymont - Proposer duties times down from peaks of 450ms to consistent <1ms. - Current epoch attester duties times down from >1s peaks to a consistent 20-30ms. - Block production down from +600ms to 100-200ms. ## Additional Info - ~~Blocked on #2241~~ - ~~Blocked on #2234~~ ## TODO - [x] ~~Refactor this into some smaller PRs?~~ Leaving this as-is for now. - [x] Address `per_slot_processing` roots. - [x] Investigate slow next epoch times. Not getting added to cache on block processing? - [x] Consider [this](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/beacon_node/store/src/hot_cold_store.rs#L811-L812) in the scenario of replacing the state roots Co-authored-by: pawan <pawandhananjay@gmail.com> Co-authored-by: Michael Sproul <michael@sigmaprime.io>
2021-03-17 05:09:57 +00:00
pub fn iter_voting_pubkeys(&self) -> impl Iterator<Item = &PublicKeyBytes> {
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
self.validators.iter().map(|(pubkey, _)| pubkey)
}
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
/// Returns the voting `Keypair` for a given voting `PublicKey`, if all are true:
///
/// - The validator is known to `self`.
/// - The validator is enabled.
pub fn signing_method(&self, voting_public_key: &PublicKeyBytes) -> Option<Arc<SigningMethod>> {
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
self.validators
.get(voting_public_key)
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
.map(|v| v.signing_method.clone())
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
/// Add a validator definition to `self`, replacing any disabled definition with the same
/// voting public key.
///
/// The on-disk representation of the validator definitions & the key cache will both be
/// updated.
pub async fn add_definition_replace_disabled(
&mut self,
def: ValidatorDefinition,
) -> Result<(), Error> {
// Drop any disabled definitions with the same public key.
let delete_def = |existing_def: &ValidatorDefinition| {
!existing_def.enabled && existing_def.voting_public_key == def.voting_public_key
};
self.definitions.retain(|def| !delete_def(def));
// Add the definition.
self.add_definition(def).await
}
/// Add a validator definition to `self`, overwriting the on-disk representation of `self`.
pub async fn add_definition(&mut self, def: ValidatorDefinition) -> Result<(), Error> {
if self
.definitions
.as_slice()
.iter()
.any(|existing| existing.voting_public_key == def.voting_public_key)
{
return Err(Error::DuplicatePublicKey);
}
self.definitions.push(def);
self.update_validators().await?;
self.definitions
.save(&self.validators_dir)
.map_err(Error::UnableToSaveDefinitions)?;
Ok(())
}
/// Delete the validator definition and keystore for `pubkey`.
///
/// The delete is carried out in stages so that the filesystem is never left in an inconsistent
/// state, even in case of errors or crashes.
pub async fn delete_definition_and_keystore(
&mut self,
pubkey: &PublicKey,
) -> Result<DeleteKeystoreStatus, Error> {
// 1. Disable the validator definition.
//
// We disable before removing so that in case of a crash the auto-discovery mechanism
// won't re-activate the keystore.
if let Some(def) = self
.definitions
.as_mut_slice()
.iter_mut()
.find(|def| &def.voting_public_key == pubkey)
{
if def.signing_definition.is_local_keystore() {
def.enabled = false;
self.definitions
.save(&self.validators_dir)
.map_err(Error::UnableToSaveDefinitions)?;
} else {
return Err(Error::InvalidActionOnRemoteValidator);
}
} else {
return Ok(DeleteKeystoreStatus::NotFound);
}
// 2. Delete from `self.validators`, which holds the signing method.
// Delete the keystore files.
if let Some(initialized_validator) = self.validators.remove(&pubkey.compress()) {
if let SigningMethod::LocalKeystore {
ref voting_keystore_path,
ref voting_keystore_lockfile,
ref voting_keystore,
..
} = *initialized_validator.signing_method
{
// Drop the lock file so that it may be deleted. This is particularly important on
// Windows where the lockfile will fail to be deleted if it is still open.
drop(voting_keystore_lockfile.lock().take());
self.delete_keystore_or_validator_dir(voting_keystore_path, voting_keystore)?;
}
}
// 3. Delete from validator definitions entirely.
self.definitions
.retain(|def| &def.voting_public_key != pubkey);
self.definitions
.save(&self.validators_dir)
.map_err(Error::UnableToSaveDefinitions)?;
Ok(DeleteKeystoreStatus::Deleted)
}
/// Attempt to delete the voting keystore file, or its entire validator directory.
///
/// Some parts of the VC assume the existence of a validator based on the existence of a
/// directory in the validators dir named like a public key.
fn delete_keystore_or_validator_dir(
&self,
voting_keystore_path: &Path,
voting_keystore: &Keystore,
) -> Result<(), Error> {
// If the parent directory is a `ValidatorDir` within `self.validators_dir`, then
// delete the entire directory so that it may be recreated if the keystore is
// re-imported.
if let Some(validator_dir) = voting_keystore_path.parent() {
if validator_dir
== ValidatorDirBuilder::get_dir_path(&self.validators_dir, voting_keystore)
{
fs::remove_dir_all(validator_dir)
.map_err(|e| Error::UnableToDeleteValidatorDir(validator_dir.into(), e))?;
return Ok(());
}
}
// Otherwise just delete the keystore file.
fs::remove_file(voting_keystore_path)
.map_err(|e| Error::UnableToDeleteKeystore(voting_keystore_path.into(), e))?;
Ok(())
}
/// Returns a slice of all defined validators (regardless of their enabled state).
pub fn validator_definitions(&self) -> &[ValidatorDefinition] {
self.definitions.as_slice()
}
/// Indicates if the `voting_public_key` exists in self and if it is enabled.
pub fn is_enabled(&self, voting_public_key: &PublicKey) -> Option<bool> {
self.definitions
.as_slice()
.iter()
.find(|def| def.voting_public_key == *voting_public_key)
.map(|def| def.enabled)
}
/// Returns the `graffiti` for a given public key specified in the `ValidatorDefinitions`.
Optimize validator duties (#2243) ## Issue Addressed Closes #2052 ## Proposed Changes - Refactor the attester/proposer duties endpoints in the BN - Performance improvements - Fixes some potential inconsistencies with the dependent root fields. - Removes `http_api::beacon_proposer_cache` and just uses the one on the `BeaconChain` instead. - Move the code for the proposer/attester duties endpoints into separate files, for readability. - Refactor the `DutiesService` in the VC - Required to reduce the delay on broadcasting new blocks. - Gets rid of the `ValidatorDuty` shim struct that came about when we adopted the standard API. - Separate block/attestation duty tasks so that they don't block each other when one is slow. - In the VC, use `PublicKeyBytes` to represent validators instead of `PublicKey`. `PublicKey` is a legit crypto object whilst `PublicKeyBytes` is just a byte-array, it's much faster to clone/hash `PublicKeyBytes` and this change has had a significant impact on runtimes. - Unfortunately this has created lots of dust changes. - In the BN, store `PublicKeyBytes` in the `beacon_proposer_cache` and allow access to them. The HTTP API always sends `PublicKeyBytes` over the wire and the conversion from `PublicKey` -> `PublickeyBytes` is non-trivial, especially when queries have 100s/1000s of validators (like Pyrmont). - Add the `state_processing::state_advance` mod which dedups a lot of the "apply `n` skip slots to the state" code. - This also fixes a bug with some functions which were failing to include a state root as per [this comment](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/consensus/state_processing/src/state_advance.rs#L69-L74). I couldn't find any instance of this bug that resulted in anything more severe than keying a shuffling cache by the wrong block root. - Swap the VC block service to use `mpsc` from `tokio` instead of `futures`. This is consistent with the rest of the code base. ~~This PR *reduces* the size of the codebase :tada:~~ It *used* to reduce the size of the code base before I added more comments. ## Observations on Prymont - Proposer duties times down from peaks of 450ms to consistent <1ms. - Current epoch attester duties times down from >1s peaks to a consistent 20-30ms. - Block production down from +600ms to 100-200ms. ## Additional Info - ~~Blocked on #2241~~ - ~~Blocked on #2234~~ ## TODO - [x] ~~Refactor this into some smaller PRs?~~ Leaving this as-is for now. - [x] Address `per_slot_processing` roots. - [x] Investigate slow next epoch times. Not getting added to cache on block processing? - [x] Consider [this](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/beacon_node/store/src/hot_cold_store.rs#L811-L812) in the scenario of replacing the state roots Co-authored-by: pawan <pawandhananjay@gmail.com> Co-authored-by: Michael Sproul <michael@sigmaprime.io>
2021-03-17 05:09:57 +00:00
pub fn graffiti(&self, public_key: &PublicKeyBytes) -> Option<Graffiti> {
self.validators.get(public_key).and_then(|v| v.graffiti)
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// Sets the `InitializedValidator` and `ValidatorDefinition` `enabled` values.
///
/// ## Notes
///
/// Enabling or disabling a validator will cause `self.definitions` to be updated and saved to
/// disk. A newly enabled validator will be added to `self.validators`, whilst a newly disabled
/// validator will be removed from `self.validators`.
///
/// Saves the `ValidatorDefinitions` to file, even if no definitions were changed.
pub async fn set_validator_status(
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
&mut self,
voting_public_key: &PublicKey,
enabled: bool,
) -> Result<(), Error> {
if let Some(def) = self
.definitions
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
.as_mut_slice()
.iter_mut()
.find(|def| def.voting_public_key == *voting_public_key)
{
def.enabled = enabled;
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
self.update_validators().await?;
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
self.definitions
.save(&self.validators_dir)
.map_err(Error::UnableToSaveDefinitions)?;
Ok(())
}
/// Tries to decrypt the key cache.
///
/// Returns the decrypted cache if decryption was successful, or an error if a required password
/// wasn't provided and couldn't be read interactively.
///
/// In the case that the cache contains UUIDs for unknown validator definitions then it cannot
/// be decrypted and will be replaced by a new empty cache.
///
/// The mutable `key_stores` argument will be used to accelerate decyption by bypassing
/// filesystem accesses for keystores that are already known. In the case that a keystore
/// from the validator definitions is not yet in this map, it will be loaded from disk and
/// inserted into the map.
async fn decrypt_key_cache(
&self,
mut cache: KeyCache,
key_stores: &mut HashMap<PathBuf, Keystore>,
) -> Result<KeyCache, Error> {
// Read relevant key stores from the filesystem.
let mut definitions_map = HashMap::new();
for def in self.definitions.as_slice().iter().filter(|def| def.enabled) {
match &def.signing_definition {
SigningDefinition::LocalKeystore {
voting_keystore_path,
..
} => {
use std::collections::hash_map::Entry::*;
let key_store = match key_stores.entry(voting_keystore_path.clone()) {
Vacant(entry) => entry.insert(open_keystore(voting_keystore_path)?),
Occupied(entry) => entry.into_mut(),
};
definitions_map.insert(*key_store.uuid(), def);
}
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
// Remote signer validators don't interact with the key cache.
SigningDefinition::Web3Signer { .. } => (),
}
}
//check if all paths are in the definitions_map
for uuid in cache.uuids() {
if !definitions_map.contains_key(uuid) {
debug!(
self.log,
"Resetting the key cache";
"keystore_uuid" => %uuid,
"reason" => "impossible to decrypt due to missing keystore",
);
return Ok(KeyCache::new());
}
}
//collect passwords
let mut passwords = Vec::new();
let mut public_keys = Vec::new();
for uuid in cache.uuids() {
let def = definitions_map.get(uuid).expect("Existence checked before");
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
match &def.signing_definition {
SigningDefinition::LocalKeystore {
voting_keystore_password_path,
voting_keystore_password,
voting_keystore_path,
} => {
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
let pw = if let Some(p) = voting_keystore_password {
p.as_ref().to_vec().into()
} else if let Some(path) = voting_keystore_password_path {
read_password(path).map_err(Error::UnableToReadVotingKeystorePassword)?
} else {
let keystore = open_keystore(voting_keystore_path)?;
unlock_keystore_via_stdin_password(&keystore, voting_keystore_path)?
.0
.as_ref()
.to_vec()
.into()
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
};
passwords.push(pw);
public_keys.push(def.voting_public_key.clone());
}
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
// Remote signer validators don't interact with the key cache.
SigningDefinition::Web3Signer { .. } => (),
};
}
//decrypt
tokio::task::spawn_blocking(move || match cache.decrypt(passwords, public_keys) {
Ok(_) | Err(key_cache::Error::AlreadyDecrypted) => cache,
_ => KeyCache::new(),
})
.await
.map_err(Error::TokioJoin)
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
/// Scans `self.definitions` and attempts to initialize and validators which are not already
/// initialized.
///
/// The function exits early with an error if any enabled validator is unable to be
/// initialized.
///
/// ## Notes
///
/// A validator is considered "already known" and skipped if the public key is already known.
/// I.e., if there are two different definitions with the same public key then the second will
/// be ignored.
pub(crate) async fn update_validators(&mut self) -> Result<(), Error> {
//use key cache if available
let mut key_stores = HashMap::new();
// Create a lock file for the cache
let key_cache_path = KeyCache::cache_file_path(&self.validators_dir);
let cache_lockfile_path =
get_lockfile_path(&key_cache_path).ok_or(Error::BadKeyCachePath(key_cache_path))?;
let _cache_lockfile = Lockfile::new(cache_lockfile_path)?;
let cache =
KeyCache::open_or_create(&self.validators_dir).map_err(Error::UnableToOpenKeyCache)?;
let mut key_cache = self.decrypt_key_cache(cache, &mut key_stores).await?;
let mut disabled_uuids = HashSet::new();
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
for def in self.definitions.as_slice() {
if def.enabled {
match &def.signing_definition {
SigningDefinition::LocalKeystore {
voting_keystore_path,
..
} => {
Optimize validator duties (#2243) ## Issue Addressed Closes #2052 ## Proposed Changes - Refactor the attester/proposer duties endpoints in the BN - Performance improvements - Fixes some potential inconsistencies with the dependent root fields. - Removes `http_api::beacon_proposer_cache` and just uses the one on the `BeaconChain` instead. - Move the code for the proposer/attester duties endpoints into separate files, for readability. - Refactor the `DutiesService` in the VC - Required to reduce the delay on broadcasting new blocks. - Gets rid of the `ValidatorDuty` shim struct that came about when we adopted the standard API. - Separate block/attestation duty tasks so that they don't block each other when one is slow. - In the VC, use `PublicKeyBytes` to represent validators instead of `PublicKey`. `PublicKey` is a legit crypto object whilst `PublicKeyBytes` is just a byte-array, it's much faster to clone/hash `PublicKeyBytes` and this change has had a significant impact on runtimes. - Unfortunately this has created lots of dust changes. - In the BN, store `PublicKeyBytes` in the `beacon_proposer_cache` and allow access to them. The HTTP API always sends `PublicKeyBytes` over the wire and the conversion from `PublicKey` -> `PublickeyBytes` is non-trivial, especially when queries have 100s/1000s of validators (like Pyrmont). - Add the `state_processing::state_advance` mod which dedups a lot of the "apply `n` skip slots to the state" code. - This also fixes a bug with some functions which were failing to include a state root as per [this comment](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/consensus/state_processing/src/state_advance.rs#L69-L74). I couldn't find any instance of this bug that resulted in anything more severe than keying a shuffling cache by the wrong block root. - Swap the VC block service to use `mpsc` from `tokio` instead of `futures`. This is consistent with the rest of the code base. ~~This PR *reduces* the size of the codebase :tada:~~ It *used* to reduce the size of the code base before I added more comments. ## Observations on Prymont - Proposer duties times down from peaks of 450ms to consistent <1ms. - Current epoch attester duties times down from >1s peaks to a consistent 20-30ms. - Block production down from +600ms to 100-200ms. ## Additional Info - ~~Blocked on #2241~~ - ~~Blocked on #2234~~ ## TODO - [x] ~~Refactor this into some smaller PRs?~~ Leaving this as-is for now. - [x] Address `per_slot_processing` roots. - [x] Investigate slow next epoch times. Not getting added to cache on block processing? - [x] Consider [this](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/beacon_node/store/src/hot_cold_store.rs#L811-L812) in the scenario of replacing the state roots Co-authored-by: pawan <pawandhananjay@gmail.com> Co-authored-by: Michael Sproul <michael@sigmaprime.io>
2021-03-17 05:09:57 +00:00
let pubkey_bytes = def.voting_public_key.compress();
if self.validators.contains_key(&pubkey_bytes) {
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
continue;
}
if let Some(key_store) = key_stores.get(voting_keystore_path) {
disabled_uuids.remove(key_store.uuid());
}
match InitializedValidator::from_definition(
def.clone(),
&mut key_cache,
&mut key_stores,
)
.await
{
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
Ok(init) => {
let existing_lockfile_path = init
.keystore_lockfile()
.as_ref()
.filter(|l| l.file_existed())
.map(|l| l.path().to_owned());
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
self.validators
Optimize validator duties (#2243) ## Issue Addressed Closes #2052 ## Proposed Changes - Refactor the attester/proposer duties endpoints in the BN - Performance improvements - Fixes some potential inconsistencies with the dependent root fields. - Removes `http_api::beacon_proposer_cache` and just uses the one on the `BeaconChain` instead. - Move the code for the proposer/attester duties endpoints into separate files, for readability. - Refactor the `DutiesService` in the VC - Required to reduce the delay on broadcasting new blocks. - Gets rid of the `ValidatorDuty` shim struct that came about when we adopted the standard API. - Separate block/attestation duty tasks so that they don't block each other when one is slow. - In the VC, use `PublicKeyBytes` to represent validators instead of `PublicKey`. `PublicKey` is a legit crypto object whilst `PublicKeyBytes` is just a byte-array, it's much faster to clone/hash `PublicKeyBytes` and this change has had a significant impact on runtimes. - Unfortunately this has created lots of dust changes. - In the BN, store `PublicKeyBytes` in the `beacon_proposer_cache` and allow access to them. The HTTP API always sends `PublicKeyBytes` over the wire and the conversion from `PublicKey` -> `PublickeyBytes` is non-trivial, especially when queries have 100s/1000s of validators (like Pyrmont). - Add the `state_processing::state_advance` mod which dedups a lot of the "apply `n` skip slots to the state" code. - This also fixes a bug with some functions which were failing to include a state root as per [this comment](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/consensus/state_processing/src/state_advance.rs#L69-L74). I couldn't find any instance of this bug that resulted in anything more severe than keying a shuffling cache by the wrong block root. - Swap the VC block service to use `mpsc` from `tokio` instead of `futures`. This is consistent with the rest of the code base. ~~This PR *reduces* the size of the codebase :tada:~~ It *used* to reduce the size of the code base before I added more comments. ## Observations on Prymont - Proposer duties times down from peaks of 450ms to consistent <1ms. - Current epoch attester duties times down from >1s peaks to a consistent 20-30ms. - Block production down from +600ms to 100-200ms. ## Additional Info - ~~Blocked on #2241~~ - ~~Blocked on #2234~~ ## TODO - [x] ~~Refactor this into some smaller PRs?~~ Leaving this as-is for now. - [x] Address `per_slot_processing` roots. - [x] Investigate slow next epoch times. Not getting added to cache on block processing? - [x] Consider [this](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/beacon_node/store/src/hot_cold_store.rs#L811-L812) in the scenario of replacing the state roots Co-authored-by: pawan <pawandhananjay@gmail.com> Co-authored-by: Michael Sproul <michael@sigmaprime.io>
2021-03-17 05:09:57 +00:00
.insert(init.voting_public_key().compress(), init);
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
info!(
self.log,
"Enabled validator";
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
"signing_method" => "local_keystore",
"voting_pubkey" => format!("{:?}", def.voting_public_key),
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
);
if let Some(lockfile_path) = existing_lockfile_path {
warn!(
self.log,
"Ignored stale lockfile";
"path" => lockfile_path.display(),
"cause" => "Ungraceful shutdown (harmless) OR \
non-Lighthouse client using this keystore \
(risky)"
);
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}
Err(e) => {
error!(
self.log,
"Failed to initialize validator";
"error" => format!("{:?}", e),
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
"signing_method" => "local_keystore",
"validator" => format!("{:?}", def.voting_public_key)
);
// Exit on an invalid validator.
return Err(e);
}
}
}
SigningDefinition::Web3Signer { .. } => {
match InitializedValidator::from_definition(
def.clone(),
&mut key_cache,
&mut key_stores,
)
.await
{
Ok(init) => {
self.validators
.insert(init.voting_public_key().compress(), init);
info!(
self.log,
"Enabled validator";
"signing_method" => "remote_signer",
"voting_pubkey" => format!("{:?}", def.voting_public_key),
);
}
Err(e) => {
error!(
self.log,
"Failed to initialize validator";
"error" => format!("{:?}", e),
"signing_method" => "remote_signer",
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
"validator" => format!("{:?}", def.voting_public_key)
);
// Exit on an invalid validator.
return Err(e);
}
}
}
}
} else {
Optimize validator duties (#2243) ## Issue Addressed Closes #2052 ## Proposed Changes - Refactor the attester/proposer duties endpoints in the BN - Performance improvements - Fixes some potential inconsistencies with the dependent root fields. - Removes `http_api::beacon_proposer_cache` and just uses the one on the `BeaconChain` instead. - Move the code for the proposer/attester duties endpoints into separate files, for readability. - Refactor the `DutiesService` in the VC - Required to reduce the delay on broadcasting new blocks. - Gets rid of the `ValidatorDuty` shim struct that came about when we adopted the standard API. - Separate block/attestation duty tasks so that they don't block each other when one is slow. - In the VC, use `PublicKeyBytes` to represent validators instead of `PublicKey`. `PublicKey` is a legit crypto object whilst `PublicKeyBytes` is just a byte-array, it's much faster to clone/hash `PublicKeyBytes` and this change has had a significant impact on runtimes. - Unfortunately this has created lots of dust changes. - In the BN, store `PublicKeyBytes` in the `beacon_proposer_cache` and allow access to them. The HTTP API always sends `PublicKeyBytes` over the wire and the conversion from `PublicKey` -> `PublickeyBytes` is non-trivial, especially when queries have 100s/1000s of validators (like Pyrmont). - Add the `state_processing::state_advance` mod which dedups a lot of the "apply `n` skip slots to the state" code. - This also fixes a bug with some functions which were failing to include a state root as per [this comment](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/consensus/state_processing/src/state_advance.rs#L69-L74). I couldn't find any instance of this bug that resulted in anything more severe than keying a shuffling cache by the wrong block root. - Swap the VC block service to use `mpsc` from `tokio` instead of `futures`. This is consistent with the rest of the code base. ~~This PR *reduces* the size of the codebase :tada:~~ It *used* to reduce the size of the code base before I added more comments. ## Observations on Prymont - Proposer duties times down from peaks of 450ms to consistent <1ms. - Current epoch attester duties times down from >1s peaks to a consistent 20-30ms. - Block production down from +600ms to 100-200ms. ## Additional Info - ~~Blocked on #2241~~ - ~~Blocked on #2234~~ ## TODO - [x] ~~Refactor this into some smaller PRs?~~ Leaving this as-is for now. - [x] Address `per_slot_processing` roots. - [x] Investigate slow next epoch times. Not getting added to cache on block processing? - [x] Consider [this](https://github.com/sigp/lighthouse/blob/072695284f7eff82c51f79bc921ad942fea7483a/beacon_node/store/src/hot_cold_store.rs#L811-L812) in the scenario of replacing the state roots Co-authored-by: pawan <pawandhananjay@gmail.com> Co-authored-by: Michael Sproul <michael@sigmaprime.io>
2021-03-17 05:09:57 +00:00
self.validators.remove(&def.voting_public_key.compress());
match &def.signing_definition {
SigningDefinition::LocalKeystore {
voting_keystore_path,
..
} => {
if let Some(key_store) = key_stores.get(voting_keystore_path) {
disabled_uuids.insert(*key_store.uuid());
}
}
Web3Signer support for VC (#2522) [EIP-3030]: https://eips.ethereum.org/EIPS/eip-3030 [Web3Signer]: https://consensys.github.io/web3signer/web3signer-eth2.html ## Issue Addressed Resolves #2498 ## Proposed Changes Allows the VC to call out to a [Web3Signer] remote signer to obtain signatures. ## Additional Info ### Making Signing Functions `async` To allow remote signing, I needed to make all the signing functions `async`. This caused a bit of noise where I had to convert iterators into `for` loops. In `duties_service.rs` there was a particularly tricky case where we couldn't hold a write-lock across an `await`, so I had to first take a read-lock, then grab a write-lock. ### Move Signing from Core Executor Whilst implementing this feature, I noticed that we signing was happening on the core tokio executor. I suspect this was causing the executor to temporarily lock and occasionally trigger some HTTP timeouts (and potentially SQL pool timeouts, but I can't verify this). Since moving all signing into blocking tokio tasks, I noticed a distinct drop in the "atttestations_http_get" metric on a Prater node: ![http_get_times](https://user-images.githubusercontent.com/6660660/132143737-82fd3836-2e7e-445b-a143-cb347783baad.png) I think this graph indicates that freeing the core executor allows the VC to operate more smoothly. ### Refactor TaskExecutor I noticed that the `TaskExecutor::spawn_blocking_handle` function would fail to spawn tasks if it were unable to obtain handles to some metrics (this can happen if the same metric is defined twice). It seemed that a more sensible approach would be to keep spawning tasks, but without metrics. To that end, I refactored the function so that it would still function without metrics. There are no other changes made. ## TODO - [x] Restructure to support multiple signing methods. - [x] Add calls to remote signer from VC. - [x] Documentation - [x] Test all endpoints - [x] Test HTTPS certificate - [x] Allow adding remote signer validators via the API - [x] Add Altair support via [21.8.1-rc1](https://github.com/ConsenSys/web3signer/releases/tag/21.8.1-rc1) - [x] Create issue to start using latest version of web3signer. (See #2570) ## Notes - ~~Web3Signer doesn't yet support the Altair fork for Prater. See https://github.com/ConsenSys/web3signer/issues/423.~~ - ~~There is not yet a release of Web3Signer which supports Altair blocks. See https://github.com/ConsenSys/web3signer/issues/391.~~
2021-09-16 03:26:33 +00:00
// Remote signers do not interact with the key cache.
SigningDefinition::Web3Signer { .. } => (),
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
info!(
self.log,
"Disabled validator";
"voting_pubkey" => format!("{:?}", def.voting_public_key)
);
}
}
for uuid in disabled_uuids {
key_cache.remove(&uuid);
}
let validators_dir = self.validators_dir.clone();
let log = self.log.clone();
if key_cache.is_modified() {
tokio::task::spawn_blocking(move || {
match key_cache.save(validators_dir) {
Err(e) => warn!(
log,
"Error during saving of key_cache";
"err" => format!("{:?}", e)
),
Ok(true) => info!(log, "Modified key_cache saved successfully"),
_ => {}
};
})
.await
.map_err(Error::TokioJoin)?;
} else {
debug!(log, "Key cache not modified");
}
// Update the enabled and total validator counts
set_gauge(
&crate::http_metrics::metrics::ENABLED_VALIDATORS_COUNT,
self.num_enabled() as i64,
);
set_gauge(
&crate::http_metrics::metrics::TOTAL_VALIDATORS_COUNT,
self.num_total() as i64,
);
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
Ok(())
}
pub fn get_index(&self, pubkey: &PublicKeyBytes) -> Option<u64> {
self.validators.get(pubkey).and_then(|val| val.index)
}
pub fn set_index(&mut self, pubkey: &PublicKeyBytes, index: u64) {
if let Some(val) = self.validators.get_mut(pubkey) {
val.index = Some(index);
}
}
Introduce validator definition file for VC (#1357) ## Issue Addressed NA ## Proposed Changes - Introduces the `valdiator_definitions.yml` file which serves as an explicit list of validators that should be run by the validator client. - Removes `--strict` flag, split into `--strict-lockfiles` and `--disable-auto-discover` - Adds a "Validator Management" page to the book. - Adds the `common/account_utils` crate which contains some logic that was starting to duplicate across the codebase. The new docs for this feature are the best description of it (apart from the code, I guess): https://github.com/sigp/lighthouse/blob/9cb87e93ce239bbe06a9d01e92126f978356dd13/book/src/validator-management.md ## API Changes This change should be transparent for *most* existing users. If the `valdiator_definitions.yml` doesn't exist then it will be automatically generated using a method that will detect all the validators in their `validators_dir`. Users will have issues if they are: 1. Using `--strict`. 1. Have keystores in their `~/.lighthouse/validators` directory that weren't being detected by the current keystore discovery method. For users with (1), the VC will refuse to start because the `--strict` flag has been removed. They will be forced to review `--help` and choose an equivalent flag. For users with (2), this seems fairly unlikely and since we're only in testnets there's no *real* value on the line here. I'm happy to take the risk, it would be a different case for mainnet. ## Additional Info This PR adds functionality we will need for #1347. ## TODO - [x] Reconsider flags - [x] Move doc into a more reasonable chapter. - [x] Check for compile warnings.
2020-07-22 09:34:55 +00:00
}