## Issue Addressed
Which issue # does this PR address?
## Proposed Changes
- Add a counter metric to log when a block is received late from gossip.
- Also push a `DEBG` log for the above condition.
- Use Debug (`?`) instead of Display (`%`) for a bunch of logs in the beacon processor, so we don't have to deal with concatenated block roots.
- Add new ERRO and CRIT to HTTP API to alert users when they're publishing late blocks.
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
When testing our (not-yet-released) Doppelganger implementation, I noticed that we aren't detecting attestations included in blocks (only those on the gossip network).
This is because during [block processing](e8c0d1f19b/beacon_node/beacon_chain/src/beacon_chain.rs (L2168)) we only update the `observed_attestations` cache with each attestation, but not the `observed_attesters` cache. This is the correct behaviour when we consider the [p2p spec](https://github.com/ethereum/eth2.0-specs/blob/v1.0.1/specs/phase0/p2p-interface.md):
> [IGNORE] There has been no other valid attestation seen on an attestation subnet that has an identical attestation.data.target.epoch and participating validator index.
We're doing the right thing here and still allowing attestations on gossip that we've seen in a block. However, this doesn't work so nicely for Doppelganger.
To resolve this, I've taken the following steps:
- Add a `observed_block_attesters` cache.
- Rename `observed_attesters` to `observed_gossip_attesters`.
## TODO
- [x] Add a test to ensure a validator that's been seen in a block attestation (but not a gossip attestation) returns `true` for `BeaconChain::validator_seen_at_epoch`.
- [x] Add a test to ensure `observed_block_attesters` isn't polluted via gossip attestations and vice versa.
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Proposed Changes
* Implement the validator client and HTTP API changes necessary to support Altair
Co-authored-by: realbigsean <seananderson33@gmail.com>
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
## Issue Addressed
Resolves#2069
## Proposed Changes
- Adds a `--doppelganger-detection` flag
- Adds a `lighthouse/seen_validators` endpoint, which will make it so the lighthouse VC is not interopable with other client beacon nodes if the `--doppelganger-detection` flag is used, but hopefully this will become standardized. Relevant Eth2 API repo issue: https://github.com/ethereum/eth2.0-APIs/issues/64
- If the `--doppelganger-detection` flag is used, the VC will wait until the beacon node is synced, and then wait an additional 2 epochs. The reason for this is to make sure the beacon node is able to subscribe to the subnets our validators should be attesting on. I think an alternative would be to have the beacon node subscribe to all subnets for 2+ epochs on startup by default.
## Additional Info
I'd like to add tests and would appreciate feedback.
TODO: handle validators started via the API, potentially make this default behavior
Co-authored-by: realbigsean <seananderson33@gmail.com>
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
N/A
## Proposed Changes
- Removing a bunch of unnecessary references
- Updated `Error::VariantError` to `Error::Variant`
- There were additional enum variant lints that I ignored, because I thought our variant names were fine
- removed `MonitoredValidator`'s `pubkey` field, because I couldn't find it used anywhere. It looks like we just use the string version of the pubkey (the `id` field) if there is no index
## Additional Info
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
- Resolves#2169
## Proposed Changes
Adds the `AttesterCache` to allow validators to produce attestations for older slots. Presently, some arbitrary restrictions can force validators to receive an error when attesting to a slot earlier than the present one. This can cause attestation misses when there is excessive load on the validator client or time sync issues between the VC and BN.
## Additional Info
NA
## Proposed Changes
Add the `sync_aggregate` from `BeaconBlock` to the bulk signature verifier for blocks. This necessitates a new signature set constructor for the sync aggregate, which is different from the others due to the use of [`eth2_fast_aggregate_verify`](https://github.com/ethereum/eth2.0-specs/blob/v1.1.0-alpha.7/specs/altair/bls.md#eth2_fast_aggregate_verify) for sync aggregates, per [`process_sync_aggregate`](https://github.com/ethereum/eth2.0-specs/blob/v1.1.0-alpha.7/specs/altair/beacon-chain.md#sync-aggregate-processing). I made the choice to return an optional signature set, with `None` representing the case where the signature is valid on account of being the point at infinity (requires no further checking).
To "dogfood" the changes and prevent duplication, the consensus logic now uses the signature set approach as well whenever it is required to verify signatures (which should only be in testing AFAIK). The EF tests pass with the code as it exists currently, but failed before I adapted the `eth2_fast_aggregate_verify` changes (which is good).
As a result of this change Altair block processing should be a little faster, and importantly, we will no longer accidentally verify signatures when replaying blocks, e.g. when replaying blocks from the database.
## Issue Addressed
NA
## Proposed Changes
This PR addresses two things:
1. Allows the `ValidatorMonitor` to work with Altair states.
1. Optimizes `altair::process_epoch` (see [code](https://github.com/paulhauner/lighthouse/blob/participation-cache/consensus/state_processing/src/per_epoch_processing/altair/participation_cache.rs) for description)
## Breaking Changes
The breaking changes in this PR revolve around one premise:
*After the Altair fork, it's not longer possible (given only a `BeaconState`) to identify if a validator had *any* attestation included during some epoch. The best we can do is see if that validator made the "timely" source/target/head flags.*
Whilst this seems annoying, it's not actually too bad. Finalization is based upon "timely target" attestations, so that's really the most important thing. Although there's *some* value in knowing if a validator had *any* attestation included, it's far more important to know about "timely target" participation, since this is what affects finality and justification.
For simplicity and consistency, I've also removed the ability to determine if *any* attestation was included from metrics and API endpoints. Now, all Altair and non-Altair states will simply report on the head/target attestations.
The following section details where we've removed fields and provides replacement values.
### Breaking Changes: Prometheus Metrics
Some participation metrics have been removed and replaced. Some were removed since they are no longer relevant to Altair (e.g., total attesting balance) and others replaced with gwei values instead of pre-computed values. This provides more flexibility at display-time (e.g., Grafana).
The following metrics were added as replacements:
- `beacon_participation_prev_epoch_head_attesting_gwei_total`
- `beacon_participation_prev_epoch_target_attesting_gwei_total`
- `beacon_participation_prev_epoch_source_attesting_gwei_total`
- `beacon_participation_prev_epoch_active_gwei_total`
The following metrics were removed:
- `beacon_participation_prev_epoch_attester`
- instead use `beacon_participation_prev_epoch_source_attesting_gwei_total / beacon_participation_prev_epoch_active_gwei_total`.
- `beacon_participation_prev_epoch_target_attester`
- instead use `beacon_participation_prev_epoch_target_attesting_gwei_total / beacon_participation_prev_epoch_active_gwei_total`.
- `beacon_participation_prev_epoch_head_attester`
- instead use `beacon_participation_prev_epoch_head_attesting_gwei_total / beacon_participation_prev_epoch_active_gwei_total`.
The `beacon_participation_prev_epoch_attester` endpoint has been removed. Users should instead use the pre-existing `beacon_participation_prev_epoch_target_attester`.
### Breaking Changes: HTTP API
The `/lighthouse/validator_inclusion/{epoch}/{validator_id}` endpoint loses the following fields:
- `current_epoch_attesting_gwei` (use `current_epoch_target_attesting_gwei` instead)
- `previous_epoch_attesting_gwei` (use `previous_epoch_target_attesting_gwei` instead)
The `/lighthouse/validator_inclusion/{epoch}/{validator_id}` endpoint lose the following fields:
- `is_current_epoch_attester` (use `is_current_epoch_target_attester` instead)
- `is_previous_epoch_attester` (use `is_previous_epoch_target_attester` instead)
- `is_active_in_current_epoch` becomes `is_active_unslashed_in_current_epoch`.
- `is_active_in_previous_epoch` becomes `is_active_unslashed_in_previous_epoch`.
## Additional Info
NA
## TODO
- [x] Deal with total balances
- [x] Update validator_inclusion API
- [ ] Ensure `beacon_participation_prev_epoch_target_attester` and `beacon_participation_prev_epoch_head_attester` work before Altair
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Proposed Changes
Remove the remaining Altair `FIXME`s from consensus land.
1. Implement tree hash caching for the participation lists. This required some light type manipulation, including removing the `TreeHash` bound from `CachedTreeHash` which was purely descriptive.
2. Plumb the proposer index through Altair attestation processing, to avoid calculating it for _every_ attestation (potentially 128ms on large networks). This duplicates some work from #2431, but with the aim of getting it in sooner, particularly for the Altair devnets.
3. Removes two FIXMEs related to `superstruct` and cloning, which are unlikely to be particularly detrimental and will be tracked here instead: https://github.com/sigp/superstruct/issues/5
## Issue Addressed
#635
## Proposed Changes
- Keep attestations that reference a block we have not seen for 30secs before being re processed
- If we do import the block before that time elapses, it is reprocessed in that moment
- The first time it fails, do nothing wrt to gossipsub propagation or peer downscoring. If after being re processed it fails, downscore with a `LowToleranceError` and ignore the message.
## Issue Addressed
NA
## Proposed Changes
Adds a metric to see how many set bits are in the sync aggregate for each beacon block being imported.
## Additional Info
NA
## Issue Addressed
- Resolves#2444
## Proposed Changes
Adds some more detail to the error message returned when the `BeaconChainBuilder` is unable to access or decode block/state objects during startup.
## Additional Info
NA
## Proposed Changes
Implement the consensus changes necessary for the upcoming Altair hard fork.
## Additional Info
This is quite a heavy refactor, with pivotal types like the `BeaconState` and `BeaconBlock` changing from structs to enums. This ripples through the whole codebase with field accesses changing to methods, e.g. `state.slot` => `state.slot()`.
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
#2377
## Proposed Changes
Implement the same code used for block root lookups (from #2376) to state root lookups in order to improve performance and reduce associated memory spikes (e.g. from certain HTTP API requests).
## Additional Changes
- Tests using `rev_iter_state_roots` and `rev_iter_block_roots` have been refactored to use their `forwards` versions instead.
- The `rev_iter_state_roots` and `rev_iter_block_roots` functions are now unused and have been removed.
- The `state_at_slot` function has been changed to use the `forwards` iterator.
## Additional Info
- Some tests still need to be refactored to use their `forwards_iter` versions. These tests start their iteration from a specific beacon state and thus use the `rev_iter_state_roots_from` and `rev_iter_block_roots_from` functions. If they can be refactored, those functions can also be removed.
## Issue Addressed
`make lint` failing on rust 1.53.0.
## Proposed Changes
1.53.0 updates
## Additional Info
I haven't figure out why yet, we were now hitting the recursion limit in a few crates. So I had to add `#![recursion_limit = "256"]` in a few places
Co-authored-by: realbigsean <seananderson33@gmail.com>
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
## Proposed Changes
A user on Discord (`@ChewsMacRibs`) reported that the validator monitor was logging `WARN Attested to an incorrect head` for their validator while it was awaiting activation.
This PR modifies the monitor so that it ignores inactive validators, by the logic that they are either awaiting activation, or have already exited. Either way, there's no way for an inactive validator to have their attestations included on chain, so no need for the monitor to report on them.
## Additional Info
To reproduce the bug requires registering validator keys manually with `--validator-monitor-pubkeys`. I don't think the bug will present itself with `--validator-monitor-auto`.
## Issue Addressed
NA
## Proposed Changes
When observing `jemallocator` heap profiles and Grafana, it became clear that Lighthouse is spending significant RAM/CPU on processing blocks from the RPC. On investigation, it seems that we are loading the parent of the block *before* we check to see if the block is already known. This is a big waste of resources.
This PR adds an additional `check_block_relevancy` call as the first thing we do when we try to process a `SignedBeaconBlock` via the RPC (or other similar methods). Ultimately, `check_block_relevancy` will be called again later in the block processing flow. It's a very light function and I don't think trying to optimize it out is worth the risk of a bad block slipping through.
Also adds a `New RPC block received` info log when we process a new RPC block. This seems like interesting and infrequent info.
## Additional Info
NA
## Issue Addressed
NA
## Primary Change
When investigating memory usage, I noticed that retrieving a block from an early slot (e.g., slot 900) would cause a sharp increase in the memory footprint (from 400mb to 800mb+) which seemed to be ever-lasting.
After some investigation, I found that the reverse iteration from the head back to that slot was the likely culprit. To counter this, I've switched the `BeaconChain::block_root_at_slot` to use the forwards iterator, instead of the reverse one.
I also noticed that the networking stack is using `BeaconChain::root_at_slot` to check if a peer is relevant (`check_peer_relevance`). Perhaps the steep, seemingly-random-but-consistent increases in memory usage are caused by the use of this function.
Using the forwards iterator with the HTTP API alleviated the sharp increases in memory usage. It also made the response much faster (before it felt like to took 1-2s, now it feels instant).
## Additional Changes
In the process I also noticed that we have two functions for getting block roots:
- `BeaconChain::block_root_at_slot`: returns `None` for a skip slot.
- `BeaconChain::root_at_slot`: returns the previous root for a skip slot.
I unified these two functions into `block_root_at_slot` and added the `WhenSlotSkipped` enum. Now, the caller must be explicit about the skip-slot behaviour when requesting a root.
Additionally, I replaced `vec![]` with `Vec::with_capacity` in `store::chunked_vector::range_query`. I stumbled across this whilst debugging and made this modification to see what effect it would have (not much). It seems like a decent change to keep around, but I'm not concerned either way.
Also, `BeaconChain::get_ancestor_block_root` is unused, so I got rid of it 🗑️.
## Additional Info
I haven't also done the same for state roots here. Whilst it's possible and a good idea, it's more work since the fwds iterators are presently block-roots-specific.
Whilst there's a few places a reverse iteration of state roots could be triggered (e.g., attestation production, HTTP API), they're no where near as common as the `check_peer_relevance` call. As such, I think we should get this PR merged first, then come back for the state root iters. I made an issue here https://github.com/sigp/lighthouse/issues/2377.
## Issue Addressed
The latest version of Rust has new clippy rules & the codebase isn't up to date with them.
## Proposed Changes
Small formatting changes that clippy tells me are functionally equivalent
## Issue Addressed
N/A
## Proposed Changes
Add unit tests for the various CLI flags associated with the beacon node and validator client. These changes require the addition of two new flags: `dump-config` and `immediate-shutdown`.
## Additional Info
Both `dump-config` and `immediate-shutdown` are marked as hidden since they should only be used in testing and other advanced use cases.
**Note:** This requires changing `main.rs` so that the flags can adjust the program behavior as necessary.
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
Resolves#2186
## Proposed Changes
404 for any block-related information on a slot that was skipped or orphaned
Affected endpoints:
- `/eth/v1/beacon/blocks/{block_id}`
- `/eth/v1/beacon/blocks/{block_id}/root`
- `/eth/v1/beacon/blocks/{block_id}/attestations`
- `/eth/v1/beacon/headers/{block_id}`
## Additional Info
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Proposed Changes
Use two instances of max cover when packing attestations into blocks: one for the previous epoch, and one for the current epoch. This reduces the amount of computation done by roughly half due to the `O(n^2)` running time of max cover (`2 * (n/2)^2 = n^2/2`). This should help alleviate some load on block proposal, particularly on Prater.
## Issue Addressed
NA
## Proposed Changes
- Adds a specific log and metric for when a block is enshrined as head with a delay that will caused bad attestations
- We *technically* already expose this information, but it's a little tricky to determine during debugging. This makes it nice and explicit.
- Fixes a minor reporting bug with the validator monitor where it was expecting agg. attestations too early (at half-slot rather than two-thirds-slot).
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
- Ensure that the [target consistency check](b356f52c5c) is always performed on aggregates.
- Add a regression test.
## Additional Info
NA
This is a small PR that cleans up compiler warnings.
The most controversial change is removing the `data_dir` field from the `BeaconChainBuilder`.
It was removed because it was never read.
Co-authored-by: Paul Hauner <paul@paulhauner.com>
Co-authored-by: Herman Junge <hermanjunge@protonmail.com>
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
## Issue Addressed
Which issue # does this PR address?
## Proposed Changes
Avoids cloning the `BeaconState` each time Prometheus scrapes our metrics (generally every 5s 😱).
I think the original motivation behind this was *"don't hold the lock on the head whilst we do computation on it"*, however I think is flawed since our computation here is so small that it'll be quicker than the clone.
The primary motivation here is to maintain a small memory footprint by holding less in memory (i.e., the cloned `BeaconState`) and to avoid the fragmentation-creep that occurs when cloning the big contiguous slabs of memory in the `BeaconState`.
I also collapsed the active/slashed/withdrawn counters into a single loop to increase efficiency.
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
I noticed the following error on one of our nodes:
```
Mar 18 00:03:35 ip-xxxx lighthouse-bn[333503]: Mar 18 00:03:35.103 ERRO Unable to validate aggregate error: ObservedAttestersError(EpochTooLow { epoch: Epoch(23961), lowest_permissible_epoch: Epoch(23962) }), peer_id: 16Uiu2HAm5GL5KzPLhvfg9MBBFSpBqTVGRFSiTg285oezzWcZzwEv
```
The slot during this log was 766,815 (the last slot of the epoch). I believe this is due to an off-by-one error in `observed_attesters` where we were failing to provide enough capacity to store observations from the previous, current and next epochs. See code comments for further reasoning.
Here's a link to the spec: https://github.com/ethereum/eth2.0-specs/blob/v1.0.1/specs/phase0/p2p-interface.md#beacon_aggregate_and_proof
## Additional Info
NA
## 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](072695284f/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 🎉~~ 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](072695284f/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>
## Proposed Changes
While investigating an incorrect head + target vote for the epoch boundary block 708544, I noticed that the state advance failed to prime the proposer cache, as per these logs:
```
Mar 09 21:42:47.448 DEBG Subscribing to subnet target_slot: 708544, subnet: Y, service: attestation_service
Mar 09 21:49:08.063 DEBG Advanced head state one slot current_slot: 708543, state_slot: 708544, head_root: 0xaf5e69de09f384ee3b4fb501458b7000c53bb6758a48817894ec3d2b030e3e6f, service: state_advance
Mar 09 21:49:08.063 DEBG Completed state advance initial_slot: 708543, advanced_slot: 708544, head_root: 0xaf5e69de09f384ee3b4fb501458b7000c53bb6758a48817894ec3d2b030e3e6f, service: state_advance
Mar 09 21:49:14.787 DEBG Proposer shuffling cache miss block_slot: 708544, block_root: 0x9b14bf68667ab1d9c35e6fd2c95ff5d609aa9e8cf08e0071988ae4aa00b9f9fe, parent_slot: 708543, parent_root: 0xaf5e69de09f384ee3b4fb501458b7000c53bb6758a48817894ec3d2b030e3e6f, service: beacon
Mar 09 21:49:14.800 DEBG Successfully processed gossip block root: 0x9b14bf68667ab1d9c35e6fd2c95ff5d609aa9e8cf08e0071988ae4aa00b9f9fe, slot: 708544, graffiti: , service: beacon
Mar 09 21:49:14.800 INFO New block received hash: 0x9b14…f9fe, slot: 708544
Mar 09 21:49:14.984 DEBG Head beacon block slot: 708544, root: 0x9b14…f9fe, finalized_epoch: 22140, finalized_root: 0x28ec…29a7, justified_epoch: 22141, justified_root: 0x59db…e451, service: beacon
Mar 09 21:49:15.055 INFO Unaggregated attestation validator: XXXXX, src: api, slot: 708544, epoch: 22142, delay_ms: 53, index: Y, head: 0xaf5e69de09f384ee3b4fb501458b7000c53bb6758a48817894ec3d2b030e3e6f, service: val_mon
Mar 09 21:49:17.001 DEBG Slot timer sync_state: Synced, current_slot: 708544, head_slot: 708544, head_block: 0x9b14…f9fe, finalized_epoch: 22140, finalized_root: 0x28ec…29a7, peers: 55, service: slot_notifier
```
The reason for this is that the condition was backwards, so that whole block of code was unreachable.
Looking at the attestations for the block included in the block after, we can see that lots of validators missed it. Some of them may be Lighthouse v1.1.1-v1.2.0-rc.0, but it's probable that they would have missed even with the proposer cache primed, given how late the block 708544 arrived (the cache miss occurred 3.787s after the slot start): https://beaconcha.in/block/708545#attestations
## Issue Addressed
NA
## Proposed Changes
- Use the pre-states from #2174 during block production.
- Running this on Pyrmont shows block production times dropping from ~550ms to ~150ms.
- Create `crit` and `warn` logs when a block is published to the API later than we expect.
- On mainnet we are issuing a warn if the block is published more than 1s later than the slot start and a crit for more than 3s.
- Rename some methods on the `SnapshotCache` for clarity.
- Add the ability to pass the state root to `BeaconChain::produce_block_on_state` to avoid computing a state root. This is a very common LH optimization.
- Add a metric that tracks how late we broadcast blocks received from the HTTP API. This is *technically* a duplicate of a `ValidatorMonitor` log, but I wanted to have it for the case where we aren't monitoring validators too.
## Issue Addressed
Closes#1787
## Proposed Changes
* Abstract the `ValidatorPubkeyCache` over a "backing" which is either a file (legacy), or the database.
* Implement a migration from schema v2 to schema v3, whereby the contents of the cache file are copied to the DB, and then the file is deleted. The next release to include this change must be a minor version bump, and we will need to warn users of the inability to downgrade (this is our first DB schema change since mainnet genesis).
* Move the schema migration code from the `store` crate into the `beacon_chain` crate so that it can access the datadir and the `ValidatorPubkeyCache`, etc. It gets injected back into the `store` via a closure (similar to what we do in fork choice).
## Issue Addressed
NA
## Problem this PR addresses
There's an issue where Lighthouse is banning a lot of peers due to the following sequence of events:
1. Gossip block 0xabc arrives ~200ms early
- It is propagated across the network, with respect to [`MAXIMUM_GOSSIP_CLOCK_DISPARITY`](https://github.com/ethereum/eth2.0-specs/blob/v1.0.0/specs/phase0/p2p-interface.md#why-is-there-maximum_gossip_clock_disparity-when-validating-slot-ranges-of-messages-in-gossip-subnets).
- However, it is not imported to our database since the block is early.
2. Attestations for 0xabc arrive, but the block was not imported.
- The peer that sent the attestation is down-voted.
- Each unknown-block attestation causes a score loss of 1, the peer is banned at -100.
- When the peer is on an attestation subnet there can be hundreds of attestations, so the peer is banned quickly (before the missed block can be obtained via rpc).
## Potential solutions
I can think of three solutions to this:
1. Wait for attestation-queuing (#635) to arrive and solve this.
- Easy
- Not immediate fix.
- Whilst this would work, I don't think it's a perfect solution for this particular issue, rather (3) is better.
1. Allow importing blocks with a tolerance of `MAXIMUM_GOSSIP_CLOCK_DISPARITY`.
- Easy
- ~~I have implemented this, for now.~~
1. If a block is verified for gossip propagation (i.e., signature verified) and it's within `MAXIMUM_GOSSIP_CLOCK_DISPARITY`, then queue it to be processed at the start of the appropriate slot.
- More difficult
- Feels like the best solution, I will try to implement this.
**This PR takes approach (3).**
## Changes included
- Implement the `block_delay_queue`, based upon a [`DelayQueue`](https://docs.rs/tokio-util/0.6.3/tokio_util/time/delay_queue/struct.DelayQueue.html) which can store blocks until it's time to import them.
- Add a new `DelayedImportBlock` variant to the `beacon_processor::WorkEvent` enum to handle this new event.
- In the `BeaconProcessor`, refactor a `tokio::select!` to a struct with an explicit `Stream` implementation. I experienced some issues with `tokio::select!` in the block delay queue and I also found it hard to debug. I think this explicit implementation is nicer and functionally equivalent (apart from the fact that `tokio::select!` randomly chooses futures to poll, whereas now we're deterministic).
- Add a testing framework to the `beacon_processor` module that tests this new block delay logic. I also tested a handful of other operations in the beacon processor (attns, slashings, exits) since it was super easy to copy-pasta the code from the `http_api` tester.
- To implement these tests I added the concept of an optional `work_journal_tx` to the `BeaconProcessor` which will spit out a log of events. I used this in the tests to ensure that things were happening as I expect.
- The tests are a little racey, but it's hard to avoid that when testing timing-based code. If we see CI failures I can revise. I haven't observed *any* failures due to races on my machine or on CI yet.
- To assist with testing I allowed for directly setting the time on the `ManualSlotClock`.
- I gave the `beacon_processor::Worker` a `Toolbox` for two reasons; (a) it avoids changing tons of function sigs when you want to pass a new object to the worker and (b) it seemed cute.
## Issue Addressed
NA
## Proposed Changes
Add an optimization to perform `per_slot_processing` from the *leading-edge* of block processing to the *trailing-edge*. Ultimately, this allows us to import the block at slot `n` faster because we used the tail-end of slot `n - 1` to perform `per_slot_processing`.
Additionally, add a "block proposer cache" which allows us to cache the block proposer for some epoch. Since we're now doing trailing-edge `per_slot_processing`, we can prime this cache with the values for the next epoch before those blocks arrive (assuming those blocks don't have some weird forking).
There were several ancillary changes required to achieve this:
- Remove the `state_root` field of `BeaconSnapshot`, since there's no need to know it on a `pre_state` and in all other cases we can just read it from `block.state_root()`.
- This caused some "dust" changes of `snapshot.beacon_state_root` to `snapshot.beacon_state_root()`, where the `BeaconSnapshot::beacon_state_root()` func just reads the state root from the block.
- Rename `types::ShuffingId` to `AttestationShufflingId`. I originally did this because I added a `ProposerShufflingId` struct which turned out to be not so useful. I thought this new name was more descriptive so I kept it.
- Address https://github.com/ethereum/eth2.0-specs/pull/2196
- Add a debug log when we get a block with an unknown parent. There was previously no logging around this case.
- Add a function to `BeaconState` to compute all proposers for an epoch without re-computing the active indices for each slot.
## Additional Info
- ~~Blocked on #2173~~
- ~~Blocked on #2179~~ That PR was wrapped into this PR.
- There's potentially some places where we could avoid computing the proposer indices in `per_block_processing` but I haven't done this here. These would be an optimization beyond the issue at hand (improving block propagation times) and I think this PR is already doing enough. We can come back for that later.
## TODO
- [x] Tidy, improve comments.
- [x] ~~Try avoid computing proposer index in `per_block_processing`?~~
## Issue Addressed
Which issue # does this PR address?
## Proposed Changes
Replaces use of `format!` in `slog` logging with it's special no-allocation `?` and `%` shortcuts. According to a `heaptrack` analysis today over about a period of an hour, this will reduce temporary allocations by at least 4%.
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
Adds some metrics to track delays regarding:
- LH processing of blocks
- delays receiving blocks from other nodes.
## Additional Info
NA
## Issue Addressed
- Resolves#2064
## Proposed Changes
Adds a `ValidatorMonitor` struct which provides additional logging and Grafana metrics for specific validators.
Use `lighthouse bn --validator-monitor` to automatically enable monitoring for any validator that hits the [subnet subscription](https://ethereum.github.io/eth2.0-APIs/#/Validator/prepareBeaconCommitteeSubnet) HTTP API endpoint.
Also, use `lighthouse bn --validator-monitor-pubkeys` to supply a list of validators which will always be monitored.
See the new docs included in this PR for more info.
## TODO
- [x] Track validator balance, `slashed` status, etc.
- [x] ~~Register slashings in current epoch, not offense epoch~~
- [ ] Publish Grafana dashboard, update TODO link in docs
- [x] ~~#2130 is merged into this branch, resolve that~~
## Issue Addressed
The non-finality period on Pyrmont between epochs [`9114`](https://pyrmont.beaconcha.in/epoch/9114) and [`9182`](https://pyrmont.beaconcha.in/epoch/9182) was contributed to by all the `lighthouse_team` validators going down. The nodes saw excessive CPU and RAM usage, resulting in the system to kill the `lighthouse bn` process. The `Restart=on-failure` directive for `systemd` caused the process to bounce in ~10-30m intervals.
Diagnosis with `heaptrack` showed that the `BeaconChain::produce_unaggregated_attestation` function was calling `store::beacon_state::get_full_state` and sometimes resulting in a tree hash cache allocation. These allocations were approximately the size of the hosts physical memory and still allocated when `lighthouse bn` was killed by the OS.
There was no CPU analysis (e.g., `perf`), but the `BeaconChain::produce_unaggregated_attestation` is very CPU-heavy so it is reasonable to assume it is the cause of the excessive CPU usage, too.
## Proposed Changes
`BeaconChain::produce_unaggregated_attestation` has two paths:
1. Fast path: attesting to the head slot or later.
2. Slow path: attesting to a slot earlier than the head block.
Path (2) is the only path that calls `store::beacon_state::get_full_state`, therefore it is the path causing this excessive CPU/RAM usage.
This PR removes the current functionality of path (2) and replaces it with a static error (`BeaconChainError::AttestingPriorToHead`).
This change reduces the generality of `BeaconChain::produce_unaggregated_attestation` (and therefore [`/eth/v1/validator/attestation_data`](https://ethereum.github.io/eth2.0-APIs/#/Validator/produceAttestationData)), but I argue that this functionality is an edge-case and arguably a violation of the [Honest Validator spec](https://github.com/ethereum/eth2.0-specs/blob/dev/specs/phase0/validator.md).
It's possible that a validator goes back to a prior slot to "catch up" and submit some missed attestations. This change would prevent such behaviour, returning an error. My concerns with this catch-up behaviour is that it is:
- Not specified as "honest validator" attesting behaviour.
- Is behaviour that is risky for slashing (although, all validator clients *should* have slashing protection and will eventually fail if they do not).
- It disguises clock-sync issues between a BN and VC.
## Additional Info
It's likely feasible to implement path (2) if we implement some sort of caching mechanism. This would be a multi-week task and this PR gets the issue patched in the short term. I haven't created an issue to add path (2), instead I think we should implement it if we get user-demand.
## Issue Addressed
NA
## Proposed Changes
Copied from #2083, changes the config milliseconds_per_slot to seconds_per_slot to avoid errors when slot duration is not a multiple of a second. To avoid deserializing old serialized data (with milliseconds instead of seconds) the Serialize and Deserialize derive got removed from the Spec struct (isn't currently used anyway).
This PR replaces #2083 for the purpose of fixing a merge conflict without requiring the input of @blacktemplar.
## Additional Info
NA
Co-authored-by: blacktemplar <blacktemplar@a1.net>
## Issue Addressed
NA
## Proposed Changes
As per #2100, uses derives from the sturm library to implement AsRef<str> and AsStaticRef to easily get str values from enums without creating new Strings. Furthermore unifies all attestation error counter into one IntCounterVec vector.
These works are originally by @blacktemplar, I've just created this PR so I can resolve some merge conflicts.
## Additional Info
NA
Co-authored-by: blacktemplar <blacktemplar@a1.net>
## Issue Addressed
`test_dht_persistence` failing
## Proposed Changes
Bind `NetworkService::start` to an underscore prefixed variable rather than `_`. `_` was causing it to be dropped immediately
This was failing 5/100 times before this update, but I haven't been able to get it to fail after updating it
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Proposed Changes
`@potuz` on the Eth R&D Discord observed that Lighthouse blocks on Pyrmont were always arriving at other nodes after at least 1 second. Part of this could be due to processing and slow propagation, but metrics also revealed that the Lighthouse nodes were usually taking 400-600ms to even just produce a block before broadcasting it.
I tracked the slowness down to the lack of a pre-built tree hash cache (THC) on the states being used for block production. This was due to using the head state for block production, which lacks a THC in order to keep fork choice fast (cloning a THC takes at least 30ms for 100k validators). This PR modifies block production to clone a state from the snapshot cache rather than the head, which speeds things up by 200-400ms by avoiding the tree hash cache rebuild. In practice this seems to have cut block production time down to 300ms or less. Ideally we could _remove_ the snapshot from the cache (and save the 30ms), but it is required for when we re-process the block after signing it with the validator client.
## Alternatives
I experimented with 2 alternatives to this approach, before deciding on it:
* Alternative 1: ensure the `head` has a tree hash cache. This is too slow, as it imposes a +30ms hit on fork choice, which currently takes ~5ms (with occasional spikes).
* Alternative 2: use `Arc<BeaconSnapshot>` in the snapshot cache and share snapshots between the cache and the `head`. This made fork choice blazing fast (1ms), and block production the same as in this PR, but had a negative impact on block processing which I don't think is worth it. It ended up being necessary to clone the full state from the snapshot cache during block production, imposing the +30ms penalty there _as well_ as in block production.
In contract, the approach in this PR should only impact block production, and it improves it! Yay for pareto improvements 🎉
## Additional Info
This commit (ac59dfa) is currently running on all the Lighthouse Pyrmont nodes, and I've added a dashboard to the Pyrmont grafana instance with the metrics.
In future work we should optimise the attestation packing, which consumes around 30-60ms and is now a substantial contributor to the total.
## Issue Addressed
Closes#2048
## Proposed Changes
* Broadcast slashings when the `--slasher-broadcast` flag is provided.
* In the process of implementing this I refactored the slasher service into its own crate so that it could access the network code without creating a circular dependency. I moved the responsibility for putting slashings into the op pool into the service as well, as it makes sense for it to handle the whole slashing lifecycle.
## Issue Addressed
Closes#2028
Replaces #2059
## Proposed Changes
If writing to the database fails while importing a block, revert fork choice to the last version stored on disk. This prevents fork choice from being ahead of the blocks on disk. Having fork choice ahead is particularly bad if it is later successfully written to disk, because it renders the database corrupt (see #2028).
## Additional Info
* This mitigation might fail if the head+fork choice haven't been persisted yet, which can only happen at first startup (see #2067)
* This relies on it being OK for the head tracker to be ahead of fork choice. I figure this is tolerable because blocks only get added to the head tracker after successfully being written on disk _and_ to fork choice, so even if fork choice reverts a little bit, when the pruning algorithm runs, those blocks will still be on disk and OK to prune. The pruning algorithm also doesn't rely on heads being unique, technically it's OK for multiple blocks from the same linear chain segment to be present in the head tracker. This begs the question of #1785 (i.e. things would be simpler with the head tracker out of the way). Alternatively, this PR could just revert the head tracker as well (I'll look into this tomorrow).
## Issue Addressed
Closes#2042
## Proposed Changes
Pass blocks that fail gossip verification to the slasher. Blocks that are successfully verified are not passed immediately, but will be passed as part of full block verification.
