## Issue Addressed
Currently, if you launch the beacon node with the `--purge-db` flag and the `beacon` directory exists, but one (or both) of the `chain_db` or `freezer-db` directories are missing, it will error unnecessarily with:
```
Failed to remove chain_db: No such file or directory (os error 2)
```
This is an edge case which can occur in cases of manual intervention (a user deleted the directory) or if you had previously run with the `--purge-db` flag and Lighthouse errored before it could initialize the db directories.
## Proposed Changes
Check if the `chain_db`/`freezer_db` exists before attempting to remove them. This prevents unnecessary errors.
## Issue Addressed
Currently, running the Web3Signer tests locally without having a java runtime environment installed and available on your PATH will result in the tests failing.
## Proposed Changes
Add a note regarding the Web3Signer tests being dependent on java (similar to what we have for `ganache-cli`)
I have been in the process of debugging libp2p tasks as there is something locking our executor.
This addition adds a metric allowing us to track all tasks within lighthouse allowing us to identify various sections of Lighthouse code that may be taking longer than normal to process.
## Issue Addressed
Continuation of #2728, fix the fork choice tests for Rust 1.56.0 so that `unstable` is free of warnings.
CI will be broken until this PR merges, because we strictly enforce the absence of warnings (even for tests)
## Issue Addressed
Attempts to fix#2701 but I doubt this is the reason behind that.
## Proposed Changes
maintain a waker in the rpc handler and call it if an event is received
## Issue Addressed
e895074ba updated the altair fork and now that we are a week away this test no longer panics.
## Proposed Changes
Remove the expected panic and explanatory note.
## Issue Addressed
In the backfill sync the state was maintained twice, once locally and also in the globals. This makes it so that it's maintained only once.
The only behavioral change is that when backfill sync in paused, the global backfill state is updated. I asked @AgeManning about this and he deemed it a bug, so this solves it.
## Issue Addressed
When compiling with Rust 1.56.0 the compiler generates 3 instances of this warning:
```
warning: trailing semicolon in macro used in expression position
--> common/eth2_network_config/src/lib.rs:181:24
|
181 | })?;
| ^
...
195 | let deposit_contract_deploy_block = load_from_file!(DEPLOY_BLOCK_FILE);
| ---------------------------------- in this macro invocation
|
= note: `#[warn(semicolon_in_expressions_from_macros)]` on by default
= warning: this was previously accepted by the compiler but is being phased out; it will become a hard error in a future release!
= note: for more information, see issue #79813 <https://github.com/rust-lang/rust/issues/79813>
= note: this warning originates in the macro `load_from_file` (in Nightly builds, run with -Z macro-backtrace for more info)
```
This warning is completely harmless, but will be visible to users compiling Lighthouse v2.0.1 (or earlier) with Rust 1.56.0 (to be released October 21st). It is **completely safe** to ignore this warning, it's just a superficial change to Rust's syntax.
## Proposed Changes
This PR removes the semi-colon as recommended, and fixes the new Clippy lints from 1.56.0
## Issue Addressed
Mitigates #1096
## Proposed Changes
Add a flag to the beacon node called `--disable-lock-timeouts` which allows opting out of lock timeouts.
The lock timeouts serve a dual purpose:
1. They prevent any single operation from hogging the lock for too long. When a timeout occurs it logs a nasty error which indicates that there's suboptimal lock use occurring, which we can then act on.
2. They allow deadlock detection. We're fairly sure there are no deadlocks left in Lighthouse anymore but the timeout locks offer a safeguard against that.
However, timeouts on locks are not without downsides:
They allow for the possibility of livelock, particularly on slower hardware. If lock timeouts keep failing spuriously the node can be prevented from making any progress, even if it would be able to make progress slowly without the timeout. One particularly concerning scenario which could occur would be if a DoS attack succeeded in slowing block signature verification times across the network, and all Lighthouse nodes got livelocked because they timed out repeatedly. This could also occur on just a subset of nodes (e.g. dual core VPSs or Raspberri Pis).
By making the behaviour runtime configurable this PR allows us to choose the behaviour we want depending on circumstance. I suspect that long term we could make the timeout-free approach the default (#2381 moves in this direction) and just enable the timeouts on our testnet nodes for debugging purposes. This PR conservatively leaves the default as-is so we can gain some more experience before switching the default.
## Description
The `eth2_libp2p` crate was originally named and designed to incorporate a simple libp2p integration into lighthouse. Since its origins the crates purpose has expanded dramatically. It now houses a lot more sophistication that is specific to lighthouse and no longer just a libp2p integration.
As of this writing it currently houses the following high-level lighthouse-specific logic:
- Lighthouse's implementation of the eth2 RPC protocol and specific encodings/decodings
- Integration and handling of ENRs with respect to libp2p and eth2
- Lighthouse's discovery logic, its integration with discv5 and logic about searching and handling peers.
- Lighthouse's peer manager - This is a large module handling various aspects of Lighthouse's network, such as peer scoring, handling pings and metadata, connection maintenance and recording, etc.
- Lighthouse's peer database - This is a collection of information stored for each individual peer which is specific to lighthouse. We store connection state, sync state, last seen ips and scores etc. The data stored for each peer is designed for various elements of the lighthouse code base such as syncing and the http api.
- Gossipsub scoring - This stores a collection of gossipsub 1.1 scoring mechanisms that are continuously analyssed and updated based on the ethereum 2 networks and how Lighthouse performs on these networks.
- Lighthouse specific types for managing gossipsub topics, sync status and ENR fields
- Lighthouse's network HTTP API metrics - A collection of metrics for lighthouse network monitoring
- Lighthouse's custom configuration of all networking protocols, RPC, gossipsub, discovery, identify and libp2p.
Therefore it makes sense to rename the crate to be more akin to its current purposes, simply that it manages the majority of Lighthouse's network stack. This PR renames this crate to `lighthouse_network`
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
Part of https://github.com/sigp/lighthouse/issues/2557
## Proposed Changes
Refactor the slashing protection export so that it can export data for a subset of validators.
This is the last remaining building block required for supporting the standard validator API (which I'll start to build atop this branch)
## Additional Info
Built on and requires #2598
## Issue Addressed
NA
## Proposed Changes
- Update versions to `v2.0.1` in anticipation for a release early next week.
- Add `--ignore` to `cargo audit`. See #2727.
## Additional Info
NA
## Description
This PR updates the peer connection transition logic. It is acceptable for a peer to immediately transition from a disconnected state to a disconnecting state. This can occur when we are at our peer limit and a new peer's dial us.
## Issue Addressed
NA
## Proposed Changes
Adds some more testing for Altair to the op pool. Credits to @michaelsproul for some appropriated efforts here.
## Additional Info
NA
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
## Proposed Changes
Clone the proposer pubkeys during backfill signature verification to reduce the time that the pubkey cache lock is held for. Cloning such a small number of pubkeys has negligible impact on the total running time, but greatly reduces lock contention.
On a Ryzen 5950X, the setup step seems to take around 180us regardless of whether the key is cloned or not, while the verification takes 7ms. When Lighthouse is limited to 10% of one core using `sudo cpulimit --pid <pid> --limit 10` the total time jumps up to 800ms, but the setup step remains only 250us. This means that under heavy load this PR could cut the time the lock is held for from 800ms to 250us, which is a huge saving of 99.97%!
## Issue Addressed
Resolves#2262
## Proposed Changes
Add a new CLI flag `--beacon-nodes-tls-certs` which allows the user to specify a path to a certificate file (or a list of files, separated by commas). The VC will then use these certificates (in addition to the existing certificates in the OS trust store) when connecting to a beacon node over HTTPS.
## Additional Info
This only supports certificates in PEM format.
## Issue Addressed
#2695
## Proposed Changes
This updates discovery to the latest version which has patched a panic that occurred due to a race condition in the bucket logic.
## Issue Addressed
Resolves#2689
## Proposed Changes
Copy of #2709 so I can appease CI and merge without waiting for @realbigsean to come online. See #2709 for more information.
## Additional Info
NA
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
NA
## Proposed Changes
This is a wholesale rip-off of #2708, see that PR for more of a description.
I've made this PR since @realbigsean is offline and I can't merge his PR due to Github's frustrating `target-branch-check` bug. I also changed the branch to `unstable`, since I'm trying to minimize the diff between `merge-f2f`/`unstable`. I'll just rebase `merge-f2f` onto `unstable` after this PR merges.
When running `make lint` I noticed the following warning:
```
warning: patch for `fixed-hash` uses the features mechanism. default-features and features will not take effect because the patch dependency does not support this mechanism
```
So, I removed the `features` section from the patch.
## Additional Info
NA
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
NA
## Proposed Changes
This PR is near-identical to https://github.com/sigp/lighthouse/pull/2652, however it is to be merged into `unstable` instead of `merge-f2f`. Please see that PR for reasoning.
I'm making this duplicate PR to merge to `unstable` in an effort to shrink the diff between `unstable` and `merge-f2f` by doing smaller, lead-up PRs.
## Additional Info
NA
## Issue Addressed
Closes#2419
## Proposed Changes
Address a long-standing issue with the import of slashing protection data where the import would fail due to the data appearing slashable w.r.t the existing database. Importing is now idempotent, and will have no issues importing data that has been handed back and forth between different validator clients, or different implementations.
The implementation works by updating the high and low watermarks if they need updating, and not attempting to check if the input is slashable w.r.t itself or the database. This is a strengthening of the minification that we started to do by default since #2380, and what Teku has been doing since the beginning.
## Additional Info
The only feature we lose by doing this is the ability to do non-minified imports of clock drifted messages (cf. Prysm on Medalla). In theory, with the previous implementation we could import all the messages in case of clock drift and be aware of the "gap" between the real present time and the messages signed in the far future. _However_ for attestations this is close to useless, as the source epoch will advance as soon as justification occurs, which will require us to make slashable attestations with respect to our bogus attestation(s). E.g. if I sign an attestation 100=>200 when the current epoch is 101, then I won't be able to vote in any epochs prior to 101 becoming justified because 101=>102, 101=>103, etc are all surrounded by 100=>200. Seeing as signing attestations gets blocked almost immediately in this case regardless of our import behaviour, there's no point trying to handle it. For blocks the situation is more hopeful due to the lack of surrounds, but losing block proposals from validators who by definition can't attest doesn't seem like an issue (the other block proposers can pick up the slack).
Currently, the beacon node has no ability to serve the HTTP API over TLS.
Adding this functionality would be helpful for certain use cases, such as when you need a validator client to connect to a backup beacon node which is outside your local network, and the use of an SSH tunnel or reverse proxy would be inappropriate.
## Proposed Changes
- Add three new CLI flags to the beacon node
- `--http-enable-tls`: enables TLS
- `--http-tls-cert`: to specify the path to the certificate file
- `--http-tls-key`: to specify the path to the key file
- Update the HTTP API to optionally use `warp`'s [`TlsServer`](https://docs.rs/warp/0.3.1/warp/struct.TlsServer.html) depending on the presence of the `--http-enable-tls` flag
- Update tests and docs
- Use a custom branch for `warp` to ensure proper error handling
## Additional Info
Serving the API over TLS should currently be considered experimental. The reason for this is that it uses code from an [unmerged PR](https://github.com/seanmonstar/warp/pull/717). This commit provides the `try_bind_with_graceful_shutdown` method to `warp`, which is helpful for controlling error flow when the TLS configuration is invalid (cert/key files don't exist, incorrect permissions, etc).
I've implemented the same code in my [branch here](https://github.com/macladson/warp/tree/tls).
Once the code has been reviewed and merged upstream into `warp`, we can remove the dependency on my branch and the feature can be considered more stable.
Currently, the private key file must not be password-protected in order to be read into Lighthouse.
## Proposed Changes
This is a refactor of the PeerDB and PeerManager. A number of bugs have been surfacing around the connection state of peers and their interaction with the score state.
This refactor tightens the mutability properties of peers such that only specific modules are able to modify the state of peer information preventing inadvertant state changes that can lead to our local peer manager db being out of sync with libp2p.
Further, the logic around connection and scoring was quite convoluted and the distinction between the PeerManager and Peerdb was not well defined. Although these issues are not fully resolved, this PR is step to cleaning up this logic. The peerdb solely manages most mutability operations of peers leaving high-order logic to the peer manager.
A single `update_connection_state()` function has been added to the peer-db making it solely responsible for modifying the peer's connection state. The way the peer's scores can be modified have been reduced to three simple functions (`update_scores()`, `update_gossipsub_scores()` and `report_peer()`). This prevents any add-hoc modifications of scores and only natural processes of score modification is allowed which simplifies the reasoning of score and state changes.
## Issue Addressed
Fix `cargo audit` failures on `unstable`
Closes#2698
## Proposed Changes
The main culprit is `nix`, which is vulnerable for versions below v0.23.0. We can't get by with a straight-forward `cargo update` because `psutil` depends on an old version of `nix` (cf. https://github.com/rust-psutil/rust-psutil/pull/93). Hence I've temporarily forked `psutil` under the `sigp` org, where I've included the update to `nix` v0.23.0.
Additionally, I took the chance to update the `time` dependency to v0.3, which removed a bunch of stale deps including `stdweb` which is no longer maintained. Lighthouse only uses the `time` crate in the notifier to do some pretty printing, and so wasn't affected by any of the breaking changes in v0.3 ([changelog here](https://github.com/time-rs/time/blob/main/CHANGELOG.md#030-2021-07-30)).
## Proposed Changes
While investigating memory usage I noticed that the malloc metrics were going negative once they passed 2GiB. This is because the underlying `mallinfo` function returns a `i32`, and we were casting it straight to an `i64`, preserving the sign.
The long-term fix will be to move to `mallinfo2`, but it's still not yet widely available.
## Issue Addressed
Resolves#2563
Replacement for #2653 as I'm not able to reopen that PR after force pushing.
## Proposed Changes
Fixes all broken api links. Cherry picked changes in #2590 and updated a few more links.
Co-authored-by: Mason Stallmo <masonstallmo@gmail.com>
## Issue Addressed
Resolves#2555
## Proposed Changes
Don't log errors on resubscribing to topics. Also don't log errors if we are setting already set attnet/syncnet bits.
## Issue Addressed
Fix#2585
## Proposed Changes
Provide a canonical version of test_logger that can be used
throughout lighthouse.
## Additional Info
This allows tests to conditionally emit logging data by adding
test_logger as the default logger. And then when executing
`cargo test --features logging/test_logger` log output
will be visible:
wink@3900x:~/lighthouse/common/logging/tests/test-feature-test_logger (Add-test_logger-as-feature-to-logging)
$ cargo test --features logging/test_logger
Finished test [unoptimized + debuginfo] target(s) in 0.02s
Running unittests (target/debug/deps/test_logger-e20115db6a5e3714)
running 1 test
Sep 10 12:53:45.212 INFO hi, module: test_logger:8
test tests::test_fn_with_logging ... ok
test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
Doc-tests test-logger
running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
Or, in normal scenarios where logging isn't needed, executing
`cargo test` the log output will not be visible:
wink@3900x:~/lighthouse/common/logging/tests/test-feature-test_logger (Add-test_logger-as-feature-to-logging)
$ cargo test
Finished test [unoptimized + debuginfo] target(s) in 0.02s
Running unittests (target/debug/deps/test_logger-02e02f8d41e8cf8a)
running 1 test
test tests::test_fn_with_logging ... ok
test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
Doc-tests test-logger
running 0 tests
test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s
## Issue Addressed
N/A
## Proposed Changes
We set a `$TAG` env variable in the github actions workflow, and then re-use this name in the `publish.sh` script. It makes this check `if [[ -z "$TAG" ]]` return true, when it should return false on the first time it's hit.
## Additional Info
N/A
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
This PR addresses an issue found by @YorickDowne during testing of v2.0.0-rc.0.
Due to a lack of atomic database writes on checkpoint sync start-up, it was possible for the database to get into an inconsistent state from which it couldn't recover without `--purge-db`. The core of the issue was that the store's anchor info was being stored _before_ the `PersistedBeaconChain`. If a crash occured so that anchor info was stored but _not_ the `PersistedBeaconChain`, then on restart Lighthouse would think the database was unitialized and attempt to compare-and-swap a `None` value, but would actually find the stale info from the previous run.
## Proposed Changes
The issue is fixed by writing the anchor info, the split point, and the `PersistedBeaconChain` atomically on start-up. Some type-hinting ugliness was required, which could possibly be cleaned up in future refactors.
## Issue Addressed
This PR addresses issue #2657
## Proposed Changes
Changes `/eth/v1/config/deposit_contract` endpoint to return the chain ID from the loaded chain spec instead of eth1::DEFAULT_NETWORK_ID which is the Goerli chain ID of 5.
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
## Proposed Changes
Cut the first release candidate for v2.0.0, in preparation for testing and release this week
## Additional Info
Builds on #2632, which should either be merged first or in the same batch
## Issue Addressed
N/A
## Proposed Changes
When peers switching to a disconnecting state, decrement the disconnected peers counter. This also downgrades some crit logs to errors.
I've also added a re-sync point when peers get unbanned the disconnected peer count will match back to the number of disconnected peers if it has gone out of sync previously.
## Issue Addressed
Closes#2528
## Proposed Changes
- Add `BlockTimesCache` to provide block timing information to `BeaconChain`. This allows additional metrics to be calculated for blocks that are set as head too late.
- Thread the `seen_timestamp` of blocks received from RPC responses (except blocks from syncing) through to the sync manager, similar to what is done for blocks from gossip.
## Additional Info
This provides the following additional metrics:
- `BEACON_BLOCK_OBSERVED_SLOT_START_DELAY_TIME`
- The delay between the start of the slot and when the block was first observed.
- `BEACON_BLOCK_IMPORTED_OBSERVED_DELAY_TIME`
- The delay between when the block was first observed and when the block was imported.
- `BEACON_BLOCK_HEAD_IMPORTED_DELAY_TIME`
- The delay between when the block was imported and when the block was set as head.
The metric `BEACON_BLOCK_IMPORTED_SLOT_START_DELAY_TIME` was removed.
A log is produced when a block is set as head too late, e.g.:
```
Aug 27 03:46:39.006 DEBG Delayed head block set_as_head_delay: Some(21.731066ms), imported_delay: Some(119.929934ms), observed_delay: Some(3.864596988s), block_delay: 4.006257988s, slot: 1931331, proposer_index: 24294, block_root: 0x937602c89d3143afa89088a44bdf4b4d0d760dad082abacb229495c048648a9e, service: beacon
```
## Issue Addressed
Resolves#2552
## Proposed Changes
Offers some improvement in inclusion distance calculation in the validator monitor.
When registering an attestation from a block, instead of doing `block.slot() - attesstation.data.slot()` to get the inclusion distance, we now pass the parent block slot from the beacon chain and do `parent_slot.saturating_sub(attestation.data.slot())`. This allows us to give best effort inclusion distance in scenarios where the attestation was included right after a skip slot. Note that this does not give accurate results in scenarios where the attestation was included few blocks after the skip slot.
In this case, if the attestation slot was `b1` and was included in block `b2` with a skip slot in between, we would get the inclusion delay as 0 (by ignoring the skip slot) which is the best effort inclusion delay.
```
b1 <- missed <- b2
```
Here, if the attestation slot was `b1` and was included in block `b3` with a skip slot and valid block `b2` in between, then we would get the inclusion delay as 2 instead of 1 (by ignoring the skip slot).
```
b1 <- missed <- b2 <- b3
```
A solution for the scenario 2 would be to count number of slots between included slot and attestation slot ignoring the skip slots in the beacon chain and pass the value to the validator monitor. But I'm concerned that it could potentially lead to db accesses for older blocks in extreme cases.
This PR also uses the validator monitor data for logging per epoch inclusion distance. This is useful as we won't get inclusion data in post-altair summaries.
Co-authored-by: Michael Sproul <micsproul@gmail.com>
## Proposed Changes
Instead of checking for strict equality between a BN's spec and the VC's local spec, just check the genesis fork version. This prevents us from failing eagerly for minor differences, while still protecting the VC from connecting to a completely incompatible BN.
A warning is retained for the previous case where the specs are not exactly equal, which is to be expected if e.g. running against Infura before Infura configures the mainnet Altair fork epoch.
## Proposed Changes
Add tooling to lcli to provide a way to measure the attestation packing efficiency of historical blocks by querying a beacon node API endpoint.
## Additional Info
Since block rewards are proportional to the number of unique attestations included in the block, a measure of efficiency can be calculated by comparing the number of unique attestations that could have been included into a block vs the number of unique attestations that were actually included.
This lcli tool provides the following data per block:
- Slot Number
- Proposer Index and Grafitti (if any)
- Available Unique Attestations
- Included Unique Attestations
- Best-effort estimate of the number of offline validators for the epoch. This means we can normalize the calculated efficiency, removing offline validators from the available attestation set.
The data is outputted as a csv file.
## Usage
Install lcli:
```
make install-lcli
```
Alternatively install with the `fake_crypto` feature to skip signature verification which improves performance:
```
cargo install --path lcli --features=fake_crypto --force --locked
```
Ensure a Lighthouse beacon node is running and synced. A non-default API endpoint can be passed with the `--endpoint` flag.
Run:
```
lcli etl-block-efficiency --output /path/to/output.csv --start-epoch 40 --end-epoch 80
```
## Issue Addressed
NA
## Proposed Changes
As `cargo audit` astutely pointed out, the version of `zeroize_derive` were were using had a vulnerability:
```
Crate: zeroize_derive
Version: 1.1.0
Title: `#[zeroize(drop)]` doesn't implement `Drop` for `enum`s
Date: 2021-09-24
ID: RUSTSEC-2021-0115
URL: https://rustsec.org/advisories/RUSTSEC-2021-0115
Solution: Upgrade to >=1.2.0
```
This PR updates `zeroize` and `zeroize_derive` to appease `cargo audit`.
`tiny-bip39` was also updated to allow compile.
## Additional Info
I don't believe this vulnerability actually affected the Lighthouse code-base directly. However, `tiny-bip39` may have been affected which may have resulted in some uncleaned memory in Lighthouse. Whilst this is not ideal, it's not a major issue. Zeroization is a nice-to-have since it only protects from sophisticated attacks or attackers that already have a high level of access already.
