## Issue Addressed
#4827
## Proposed Changes
This PR introduces a new build-arg to the Lighthouse Dockerfile: `CARGO_USE_GIT_CLI`. This arg will be passed into the `CARGO_NET_GIT_FETCH_WITH_CLI` [environment variable](https://doc.rust-lang.org/cargo/reference/config.html#netgit-fetch-with-cli), which instructs `cargo` to use the git CLI during `fetch` operations instead of the git library. Doing so works around [a bug](https://github.com/rust-lang/cargo/issues/10583) with the git library that causes it to go OOM during `fetch` operations on `arm64` platforms.
The default value is `false` so this doesn't affect Lighthouse builds or the CI pipeline. Running a build with `--build-arg CARGO_USE_GIT_CLI=true` will activate it, which is necessary to cross-compile the `arm64` binary when not using `cross` (i.e., when building via the Dockerfile instead of natively if you don't have a rust environment ready to go).
Special thanks to @michaelsproul for helping me repro the initial problem.
Co-authored-by: Michael Sproul <micsproul@gmail.com>
## Issue Addressed
Following the conversation on https://github.com/libp2p/rust-libp2p/pull/3666 the changes introduced in this PR will allow us to give more insights if the bandwidth limitations happen at the transport level, namely if quic helps vs yamux and it's [window size limitation](https://github.com/libp2p/rust-yamux/issues/162) or if the bottleneck is at the gossipsub level.
## Proposed Changes
introduce new quic and tcp bandwidth metric gauges.
cc @mxinden (turned out to be easier, Thomas gave me a hint)
## Issue Addressed
Related to https://github.com/sigp/lighthouse/issues/4676.
Deneb-specifc CI code to be removed before merging to `unstable`. Dot not merge until we're ready to merge into `unstable`, as we may need to release deneb docker images before merging.
Keep in mind that most of the changes in the below PR (to `unstable`) have already
been merged to `deneb-free-blobs`, so merging `deneb-free-blobs` into `unstable` would include those changes - it would be ok if the release runners are ready, otherwise we may want to exclude them before merging.
- https://github.com/sigp/lighthouse/pull/4592
## Issue Addressed
Fix OOMs caused by too many concurrent tests. The runner machine is currently liable to run `32 * 5 = 160` tests in parallel. If each test uses say 300MB max, this is 48GB of RAM!
## Proposed Changes
Reduce the number of threads per runner job to 8. This should cap the memory at 4x lower than the current limit, i.e. around 12GB. If we continue to run out of RAM, we should consider more sophisticated limits.
## Issue Addressed
Fix a deadlock in the tests that was causing tests on tree-states to run for hours without finishing: https://github.com/sigp/lighthouse/actions/runs/6491194654/job/17628138360.
## Proposed Changes
Avoid using a Mutex under the Rayon `par_iter`. Instead, use an `AtomicUsize`. I've run the new version several times in a loop and it hasn't deadlocked (it was deadlocking consistently on tree-states).
## Additional Info
The same bug exists in unstable and tree-states, but I'm not sure why it was triggering so consistently on the tree-states branch.
## Proposed Changes
Add `compare_fields(as_iter)` as a field attribute to `compare_fields_derive`. This allows any iterable type to be compared in the same as a slice (by index).
This is forwards-compatible with tree-states types like `List` and `Vector` which can not be cast to slices.
## Issue Addressed
N/A
## Proposed Changes
I saw a false positive on the link-check CI run and while investigating I noticed that this link technically 404's but is not "dead" in the strict sense. I have updated it to the correct path.
## Proposed Changes
Fix the misplacement of the total block production time metric, which occurred during a previous refactor.
Total block production times are no longer skewed low (data from Holesky + blockdreamer):
```
# HELP beacon_block_production_seconds Full runtime of block production
# TYPE beacon_block_production_seconds histogram
beacon_block_production_seconds_bucket{le="0.005"} 0
beacon_block_production_seconds_bucket{le="0.01"} 0
beacon_block_production_seconds_bucket{le="0.025"} 0
beacon_block_production_seconds_bucket{le="0.05"} 0
beacon_block_production_seconds_bucket{le="0.1"} 0
beacon_block_production_seconds_bucket{le="0.25"} 0
beacon_block_production_seconds_bucket{le="0.5"} 37
beacon_block_production_seconds_bucket{le="1"} 65
beacon_block_production_seconds_bucket{le="2.5"} 66
beacon_block_production_seconds_bucket{le="5"} 66
beacon_block_production_seconds_bucket{le="10"} 66
beacon_block_production_seconds_bucket{le="+Inf"} 66
beacon_block_production_seconds_sum 34.225780452
beacon_block_production_seconds_count 66
```
## Additional Info
Cheers to @jimmygchen for helping spot this.
## Issue Addressed
Addresses #4778, and potentially fixes the flaky deneb builder test `builder_works_post_deneb`.
The [deneb builder test](c5c84f1213/beacon_node/http_api/tests/tests.rs (L5371)) has been quite flaky on our CI (`release-tests`) since it was introduced. I'm guessing that it might be timing out on the builder `get_header` call (1 second), and therefore the local payload is used, while the test expects builder payload to be used.
On my machine the [`get_header` ](c5c84f1213/beacon_node/execution_layer/src/test_utils/mock_builder.rs (L367)) call takes about 550ms, which could easily go over 1s on slower environments (our windows CI runner is much slower than the ubuntu one).
I did a profile on the test and it showed that `blob_to_kzg_commiment` and `compute_kzg_proof` was taking a large chunk of time, so perhaps pre-generating the blobs could help stablise this test.
## Proposed Changes
Pre-generate blobs bundle for Mainnet and Minimal presets.
Before the change `get_header` took about **550ms**, and it's now reduced to **50-55ms** after the change. If timeout was indeed the cause of the flaky test, this fix should stablise it. This also brings the flaky `builder_works_post_deneb` test time from 50s to 10s. (8s if we only use a single blob)
## Issue Addressed
CI is currently blocked by persistently failing integration tests.
## Proposed Changes
Use latest Nethermind release and apply the appropriate fixes as there have been breaking changes.
Also increase the timeout since I had some local timeouts.
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
Co-authored-by: antondlr <anton@delaruelle.net>
Co-authored-by: Jimmy Chen <jchen.tc@gmail.com>
* use workspace deps in kzg crate
* delete unused blobs dp path field
* full match on fork name in engine api get payload v3
* only accept v3 payloads on get payload v3 endpoint in mock el
* remove FIXMEs related to merge transition tests
* move static tx to test utils
* default max_per_epoch_activation_churn_limit to mainnet value
* remove unnecessary async
* remove comment
* use task executor in `blob_sidecars` endpoint
* Add `blob_sidecar` event to SSE.
* Return 202 if a block is published but failed blob validation when validation level is `Gossip`.
* Move `BlobSidecar` event to `process_gossip_blob` and add test.
* Emit `BlobSidecar` event when blobs are received over rpc.
* Improve test assertions on `SseBlobSidecar`s.
* Add quotes to blob index serialization in `SseBlobSidecar`
Co-authored-by: realbigsean <seananderson33@GMAIL.com>
---------
Co-authored-by: realbigsean <seananderson33@GMAIL.com>
* Initial Commit of State LRU Cache
* Build State Caches After Reconstruction
* Cleanup Duplicated Code in OverflowLRUCache Tests
* Added Test for State LRU Cache
* Prune Cache of Old States During Maintenance
* Address Michael's Comments
* Few More Comments
* Removed Unused impl
* Last touch up
* Fix Clippy
## Proposed Changes
Instead of sending every attestation subscription every slot to every BN:
- Send subscriptions 32, 16, 8, 7, 6, 5, 4, 3 slots before they occur.
- Track whether each subscription is sent successfully and retry it in subsequent slots if necessary.
## Additional Info
- [x] Add unit tests for `SubscriptionSlots`.
- [x] Test on Holesky.
- [x] Based on #4774 for testing.
## Issue Addressed
Partially #4788
## Proposed Changes
Remove documentation on `/lighthouse/database/reconstruct` API to avoid confusion as the calling the API during historical block download will show an error in the beacon log
Add Events API about `payload_attributes`
## Additional Info
Please provide any additional information. For example, future considerations
or information useful for reviewers.
Co-authored-by: chonghe <44791194+chong-he@users.noreply.github.com>
Co-authored-by: Michael Sproul <micsproul@gmail.com>
## Issue Addressed
While reviewing #4801 I noticed that our use of `take_while` in the block replayer means that if a state root iterator _with gaps_ is provided, some additonal state roots will be dropped unnecessarily. In practice the impact is small, because once there's _one_ state root miss, the whole tree hash cache needs to be built anyway, and subsequent misses are less costly. However this was still a little inefficient, so I figured it's better to fix it.
## Proposed Changes
Use [`peeking_take_while`](https://docs.rs/itertools/latest/itertools/trait.Itertools.html#method.peeking_take_while) to avoid consuming the next element when checking whether it satisfies the slot predicate.
## Additional Info
There's a gist here that shows the basic dynamics in isolation: https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=40b623cc0febf9ed51705d476ab140c5. Changing the `peeking_take_while` to a `take_while` causes the assert to fail. Similarly I've added a new test `block_replayer_peeking_state_roots` which fails if the same change is applied inside `get_state_root`.
## Issue Addressed
Closes https://github.com/sigp/lighthouse/issues/4712
## Proposed Changes
Exit aggregation step early if no validator is aggregator. This avoids an unnecessary request to the beacon node and more importantly fixes noisy errors if Lighthouse VC is used with other clients such as Lodestar and Prysm.
## Additional Info
Related issue https://github.com/ChainSafe/lodestar/issues/5553
## Issue Addressed
- Close#4596
## Proposed Changes
- Add `Filter::recover` to handle rejections specifically as 404 NOT FOUND
Please list or describe the changes introduced by this PR.
## Additional Info
Similar to PR #3836
## Issue Addressed
This PR closes https://github.com/sigp/lighthouse/issues/3237
## Proposed Changes
Remove topic weight of old topics when the fork happens.
## Additional Info
- Divided `NetworkService::start()` into `NetworkService::build()` and `NetworkService::start()` for ease of testing.
## Issue Addressed
We've had a report of sync committee performance suffering with the beacon processor HTTP API prioritisations.
## Proposed Changes
Increase the priority of `/eth/v1/beacon/blocks/head/root` requests, which are used by the validator client to form sync committee messages, here:
441fc1691b/validator_client/src/sync_committee_service.rs (L181-L188)
Additionally, avoid loading the blinded block in all but the `block_id=block_root` case. I'm not sure why we were doing this previously, I suspect it was just an oversight during the implementation of the `finalized` status on API requests.
## Additional Info
I think this change should have minimal negative impact as:
- The block root endpoint is quick to compute (a few ms max).
- Only the priority of `head` requests is increased. Analytical processes that are making lots of block root requests for past slots are unable to DoS the beacon processor, as their requests will still be processed after attestations.
## Issue Addressed
N/A
## Proposed Changes
We were currently downscoring a peer for sending us a block that we already have in fork choice. This is unnecessary as we get duplicates in lighthouse only when
1. We published the block, so the block is already in fork choice
2. We imported the same block over rpc
In both scenarios, the peer who sent us the block over gossip is not at fault.
This isn't exploitable as valid duplicates will get dropped by the gossipsub duplicate filter
## Issue Addressed
Right now lighthouse accepts zero as enr ports. Since enr ports should be reachable, zero ports should be rejected here
## Proposed Changes
- update the config to use `NonZerou16` as an ENR port for all enr-related fields.
- the enr builder from config now sets the enr to the listening port only if the enr port is not already set (prev behaviour) and the listening port is not zero (new behaviour)
- reject zero listening ports when used with `enr-match`.
- boot node now rejects listening port as zero, since those are advertised.
- generate-bootnode-enr also rejected zero listening ports for the same reason.
- update local network scripts
## Additional Info
Unrelated, but why do we overwrite `enr-x-port` values with listening ports if `enr-match` is present? we prob should only do this for enr values that are not already set.
## Issue Addressed
https://github.com/sigp/lighthouse/issues/4543
## Proposed Changes
- Removes `NotBanned` from `BanResult`, implements `Display` and `std::error::Error` for `BanResult` and changes `ban_result` return type to `Option<BanResult>` which helps returning `BanResult` on `handle_established_inbound_connection`
- moves the check from for banned peers from `on_connection_established` to `handle_established_inbound_connection` to start addressing #4543.
- Removes `allow_block_list` as it's now redundant? Not sure about this one but if `PeerManager` keeps track of the banned peers, no need to send a `Swarm` event for `alow_block_list` to also keep that list right?
## Questions
- #4543 refers:
> More specifically, implement the connection limit behaviour inside the peer manager.
@AgeManning do you mean copying `libp2p::connection_limits::Behaviour`'s code into `PeerManager`/ having it as an inner `NetworkBehaviour` of `PeerManager`/other? If it's the first two, I think it probably makes more sense to have it as it is as it's less code to maintain.
> Also implement the banning of peers inside the behaviour, rather than passing messages back up to the swarm.
I tried to achieve this, but we still need to pass the `PeerManagerEvent::Banned` swarm event as `DiscV5` handles it's node and ip management internally and I did not find a method to query if a peer is banned. Is there anything else we can do from here?
3397612160/beacon_node/lighthouse_network/src/discovery/mod.rs (L931-L940)
Same as the question above, I did not find a way to check if `DiscV5` has the peer banned, so that we could check here and avoid sending `Swarm` events
3397612160/beacon_node/lighthouse_network/src/peer_manager/network_behaviour.rs (L168-L178)
Is there a chance we try to dial a peer that has been banned previously?
Thanks!
## Proposed Changes
- only use LH types to avoid build issues
- use warp instead of axum for the server to avoid importing the dep
## Additional Info
- wondering if we can move the `execution_layer/test_utils` to its own crate and import it as a dev dependency
- this would be made easier by separating out our engine API types into their own crate so we can use them in the test crate
- or maybe we can look into using reth types for the engine api if they are in their own crate
Co-authored-by: realbigsean <seananderson33@gmail.com>
* add processing and processed caching to the DA checker
* move processing cache out of critical cache
* get it compiling
* fix lints
* add docs to `AvailabilityView`
* some self review
* fix lints
* fix beacon chain tests
* cargo fmt
* make availability view easier to implement, start on testing
* move child component cache and finish test
* cargo fix
* cargo fix
* cargo fix
* fmt and lint
* make blob commitments not optional, rename some caches, add missing blobs struct
* Update beacon_node/beacon_chain/src/data_availability_checker/processing_cache.rs
Co-authored-by: ethDreamer <37123614+ethDreamer@users.noreply.github.com>
* marks review feedback and other general cleanup
* cargo fix
* improve availability view docs
* some renames
* some renames and docs
* fix should delay lookup logic
* get rid of some wrapper methods
* fix up single lookup changes
* add a couple docs
* add single blob merge method and improve process_... docs
* update some names
* lints
* fix merge
* remove blob indices from lookup creation log
* remove blob indices from lookup creation log
* delayed lookup logging improvement
* check fork choice before doing any blob processing
* remove unused dep
* Update beacon_node/beacon_chain/src/data_availability_checker/availability_view.rs
Co-authored-by: Michael Sproul <micsproul@gmail.com>
* Update beacon_node/beacon_chain/src/data_availability_checker/availability_view.rs
Co-authored-by: Michael Sproul <micsproul@gmail.com>
* Update beacon_node/beacon_chain/src/data_availability_checker/availability_view.rs
Co-authored-by: Michael Sproul <micsproul@gmail.com>
* Update beacon_node/beacon_chain/src/data_availability_checker/availability_view.rs
Co-authored-by: Michael Sproul <micsproul@gmail.com>
* Update beacon_node/network/src/sync/block_lookups/delayed_lookup.rs
Co-authored-by: Michael Sproul <micsproul@gmail.com>
* remove duplicate deps
* use gen range in random blobs geneartor
* rename processing cache fields
* require block root in rpc block construction and check block root consistency
* send peers as vec in single message
* spawn delayed lookup service from network beacon processor
* fix tests
---------
Co-authored-by: ethDreamer <37123614+ethDreamer@users.noreply.github.com>
Co-authored-by: Michael Sproul <micsproul@gmail.com>
Attempting to improve our CI speeds as its recently been a pain point.
Major changes:
- Use a github action to pull stable/nightly rust rather than building it each run
- Shift test suite to `nexttest` https://github.com/nextest-rs/nextest for CI
UPDATE:
So I've iterated on some changes, and although I think its still not optimal I think this is a good base to start from. Some extra things in this PR:
- Shifted where we pull rust from. We're now using this thing: https://github.com/moonrepo/setup-rust . It's got some interesting cache's built in, but was not seeing the gains that Jimmy managed to get. In either case tho, it can pull rust, cargofmt, clippy, cargo nexttest all in < 5s. So I think it's worthwhile.
- I've grouped a few of the check-like tests into a single test called `code-test`. Although we were using github runners in parallel which may be faster, it just seems wasteful. There were like 4-5 tests, where we would pull lighthouse, compile it, then run an action, like clippy, cargo-audit or fmt. I've grouped these into a single action, so we only compile lighthouse once, then in each step we run the checks. This avoids compiling lighthouse like 5 times.
- Ive made doppelganger tests run on our local machines to avoid pulling foundry, building and making lcli which are all now baked into the images.
- We have sccache and do not incremental compile lighthouse
Misc bonus things:
- Cargo update
- Fix web3 signer openssl keys which is required after a cargo update
- Use mock_instant in an LRU cache test to avoid non-deterministic test
- Remove race condition in building web3signer tests
There's still some things we could improve on. Such as downloading the EF tests every run and the web3-signer binary, but I've left these to be out of scope of this PR. I think the above are meaningful improvements.
Co-authored-by: Paul Hauner <paul@paulhauner.com>
Co-authored-by: realbigsean <seananderson33@gmail.com>
Co-authored-by: antondlr <anton@delaruelle.net>
## Issue Addressed
#4675
## Proposed Changes
- Update local ENR (**only port numbers**) with local addresses received from libp2p (via `SwarmEvent::NewListenAddr`)
- Only use the zero port for CLI tests
## Additional Info
### See Also ###
- #4705
- #4402
- #4745
