When engine_newPayloadV6 fails with a transient error (e.g. Besu's
ConcurrentModificationException), queue the envelope for retry instead
of permanently rejecting it. Matches Lodestar's behavior of retrying
on the next BlockImported event.
- Add RetryEnvelope variant to ReprocessQueueMessage
- On BlockImported, immediately dispatch any pending retry envelopes
- Fallback timeout of 1 slot in case no block arrives
- Max 3 retries per envelope to prevent infinite loops
- Only retry non-penalizing EL errors (transient failures)
Fixes a flaky CI failure in `data_column_reconstruction_at_deadline` where 2 `column_reconstruction` events are emitted instead of the expected 1.
- Change `queued_column_reconstructions` from `HashMap<Hash256, DelayKey>` to `HashMap<Hash256, Option<DelayKey>>`, where `None` indicates reconstruction was already dispatched.
- On dispatch (`ReadyColumnReconstruction`), set the entry to `None` instead of removing it. This prevents a subsequent gossip column from inserting a fresh reconstruction request into the now-vacant slot.
- Prune stale `None` entries on each dispatch to keep the map bounded.
Co-Authored-By: Josh King <josh@sigmaprime.io>
Serves envelope by range and by root requests. Added PayloadEnvelopeStreamer so that we dont need to alter upstream code when we introduce blinded payload envelopes.
Co-Authored-By: Eitan Seri- Levi <eserilev@gmail.com>
Co-Authored-By: Eitan Seri-Levi <eserilev@ucsc.edu>
Co-Authored-By: dapplion <35266934+dapplion@users.noreply.github.com>
Add a queue that allows us to reprocess an envelope when it arrives over gossip references a unknown block root. When the block is finally imported, we immediately reprocess the queued envelope.
Note that we don't trigger a block lookup sync. Incoming attestations for this block root will already trigger a lookup for us. I think thats good enough
Co-Authored-By: Eitan Seri- Levi <eserilev@gmail.com>
- Step 0 of the tree-sync roadmap https://github.com/sigp/lighthouse/issues/7678
Current lookup sync tests are written in an explicit way that assume how the internals of lookup sync work. For example the test would do:
- Emit unknown block parent message
- Expect block request for X
- Respond with successful block request
- Expect block processing request for X
- Response with successful processing request
- etc..
This is unnecessarily verbose. And it will requires a complete re-write when something changes in the internals of lookup sync (has happened a few times, mostly for deneb and fulu).
What we really want to assert is:
- WHEN: we receive an unknown block parent message
- THEN: Lookup sync can sync that block
- ASSERT: Without penalizing peers, without unnecessary retries
Keep all existing tests and add new cases but written in the new style described above. The logic to serve and respond to request is in this function `fn simulate` 2288a3aeb1/beacon_node/network/src/sync/tests/lookups.rs (L301)
- It controls peer behavior based on a `CompleteStrategy` where you can set for example "respond to BlocksByRoot requests with empty"
- It actually runs beacon processor messages running their clousures. Now sync tests actually import blocks, increasing the test coverage to the interaction of sync and the da_checker.
- To achieve the above the tests create real blocks with the test harness. To make the tests as fast as before, I disabled crypto with `TestConfig`
Along the way I found a couple bugs, which I documented on the diff.
Co-Authored-By: dapplion <35266934+dapplion@users.noreply.github.com>
All the required boilerplate for gloas gossip. We'll include the gossip message processing logic in a separate PR
Co-Authored-By: Eitan Seri- Levi <eserilev@gmail.com>
Co-Authored-By: Jimmy Chen <jchen.tc@gmail.com>
since block and blob both start with `bl`, it was not clear how to differentiate between `blbroots_queue` and `bbroots_queue`
After renaming, there also seems to be a discrepancy
Co-Authored-By: Kevaundray Wedderburn <kevtheappdev@gmail.com>
This recent PR below changes the max reconstruction delay to be a function of slot time. However it uses `seconds_from_slot_start` when comparing (and dropping `nano`), so it might delay reconstruction on networks where the slot time isn’t a multiple of 4, e.g. on gnosis this only happens at 2s instead of 1.25s.:
- https://github.com/sigp/lighthouse/pull/8067#discussion_r2443875068
Use `millis_from_slot_start` when comparing against reconstruction deadline
Also added some tests for reconstruction delay.
Co-Authored-By: Jimmy Chen <jchen.tc@gmail.com>
Part of #7866
- Continuation of #7921
In the above PR, we enabled rayon for batch KZG verification in chain segment processing. However, using the global rayon thread pool for backfill is likely to create resource contention with higher-priority beacon processor work.
This PR introduces a dedicated low-priority rayon thread pool `LOW_PRIORITY_RAYON_POOL` and uses it for processing backfill chain segments.
This prevents backfill KZG verification from using the global rayon thread pool and competing with high-priority beacon processor tasks for CPU resources.
However, this PR by itself doesn't prevent CPU oversubscription because other tasks could still fill up the global rayon thread pool, and having an extra thread pool could make things worse. To address this we need the beacon
processor to coordinate total CPU allocation across all tasks, which is covered in:
- #7789
Co-Authored-By: Jimmy Chen <jchen.tc@gmail.com>
Co-Authored-By: Eitan Seri- Levi <eserilev@gmail.com>
Co-Authored-By: Eitan Seri-Levi <eserilev@ucsc.edu>
Fix a memory leak in the reprocess queue.
If the vec of attestation IDs for a block is never evicted from the reprocess queue by a `BlockImported` event, then it stays in the map forever consuming memory. The fix is to remove the entry when its last attestation times out. We do similarly for light client updates.
In practice this will only occur if there is a race between adding an attestation to the queue and processing the `BlockImported` event, or if there are attestations for block roots that we never import (e.g. random block roots, block roots of invalid blocks).
Co-Authored-By: Michael Sproul <michael@sigmaprime.io>
#7697
If we're three seconds into the current slot just trigger reconstruction. I don't know what the correct reconstruction deadline number is, but it should probably be at least half a second before the attestation deadline
Co-Authored-By: Eitan Seri- Levi <eserilev@gmail.com>
Co-Authored-By: Eitan Seri-Levi <eserilev@ucsc.edu>
Similar to `BEACON_PROCESSOR_WORKERS_ACTIVE_TOTAL` but this metric also records the work type.
This is useful in identifying the task when a worker is stuck due to a deadlock or something else, and usually difficult to debug in production / release mode.
This PR adds a created_timestamp to the beacon processor send channel. When work items are sent through that channel `try_send` will forward the work event along with the current timestamp to the beacon processor. When the work event is completed the `Drop` impl for `SendOnDrop` will track the time it took from work event creation to its completion. Previously we only had data on how long a work event took to process, but didn't have data on how long it sat in the queue + how long it took to process.
Partially https://github.com/sigp/lighthouse/issues/6291
This PR removes the reprocess event channel from being externally exposed. All work events are now sent through the single `BeaconProcessorSend` channel. I've introduced a new `Work::Reprocess` enum variant which we then use to schedule jobs for reprocess. I've also created a new scheduler module which will eventually house the different scheduler impls.
This is all needed as an initial step to generalize the beacon processor
A "full" implementation for the generalized beacon processor can be found here
https://github.com/sigp/lighthouse/pull/6448
I'm going to try to break up the full implementation into smaller PR's so it can actually be reviewed
#6970
This allows for us to receive `SingleAttestation` over gossip and process it without converting. There is still a conversion to `Attestation` as a final step in the attestation verification process, but by then the `SingleAttestation` is fully verified.
I've also fully removed the `submitPoolAttestationsV1` endpoint as its been deprecated
I've also pre-emptively deprecated supporting `Attestation` in `submitPoolAttestationsV2` endpoint. See here for more info: https://github.com/ethereum/beacon-APIs/pull/531
I tried to the minimize the diff here by only making the "required" changes. There are some unnecessary complexities with the way we manage the different attestation verification wrapper types. We could probably consolidate this to one wrapper type and refactor this even further. We could leave that to a separate PR if we feel like cleaning things up in the future.
Note that I've also updated the test harness to always submit `SingleAttestation` regardless of fork variant. I don't see a problem in that approach and it allows us to delete more code :)
Our Holesky nodes running with the light client enabled were logging messages about full queues:
> Feb 12 22:09:28.949 ERRO Work queue is full queue: unknown_light_client_optimistic_update, queue_len: 128, msg: the system has insufficient resources for load, service: bproc
I thought this might be genuine overload, but it turns out this queue was never being read from!
- [x] Rename light-client related queues in the beacon processor for clarity.
- [x] Ensure all light-client related queues are being popped from.
Closes
- https://github.com/sigp/lighthouse/issues/6805
- Use a new `WorkEvent::GossipAttestationToConvert` to handle the conversion from `SingleAttestation` to `Attestation` _on_ the beacon processor (prevents a Tokio thread being blocked).
- Improve the error handling for single attestations. I think previously we had no ability to reprocess single attestations for unknown blocks -- we would just error. This seemed to be the case in both gossip processing and processing of `SingleAttestation`s from the HTTP API.
- Move the `SingleAttestation -> Attestation` conversion function into `beacon_chain` so that it can return the `attestation_verification::Error` type, which has well-defined error handling and peer penalties. The now-unused variants of `types::Attestation::Error` have been removed.
Addresses #6706
This PR activates PeerDAS at the Fulu fork epoch instead of `EIP_7594_FORK_EPOCH`. This means we no longer support testing PeerDAS with Deneb / Electrs, as it's now part of a hard fork.
