This PR renames the existing `validator_duties` to `beacon_validators`
and in doing so, names validators running inside the beacon node process
"beacon validators" while those running the VC can be referred to as
"client validators" to disambiguate the two.
The existing `validator_duties` instead takes on a new responsibility:
as a home for logic shared between beacon and client validators - ie
code that provides consistency in implementation and behavior between
the two modes of operation.
Not only does this simplify reasoning about where to put code -it also
reduces the number of dependencies the validator client has from ~5000
to ~3000 modules (!) according to `nim genDepend` significantly reducing
compile times.
When a block is introduced to the system both via REST and gossip at the
same time, we will call `storeBlock` from two locations leading to a
dupliace check race condition as we wait for the EL.
This issue may manifest in particular when using an external block
builder that itself publishes the block onto the gossip network.
* refactor enqueue flow
* simplify calling `addBlock`
* complete request manager verifier future for blobless blocks
* re-verify parent conditions before adding block
among other things, it might have gone stale or finalized between one
call and the other
Split up the `ShufflingRef` acceleration logic into generically usable
parts and attester shuffling specific parts. The generic parts could be
used to accelerate other purposes, e.g., REST `/states/xxx/randao` API.
To enable additional use cases, e.g., `/states/###/randao` beacon API,
`ShufflingRef` acceleration logic needs to be able to operate on parts
of the DAG that do not have `BlockRef`. Changing `commonAncestor` to
act on `BlockId` instead of `BlockRef` is a step toward that and also
simplifies the logic some more.
* fall back to non-fcu fork choice on epoch boundaries
* Future[bool]
* fix
* Update beacon_chain/consensus_object_pools/consensus_manager.nim
Co-authored-by: Etan Kissling <etan@status.im>
* make things consistent with Opt[void] return
---------
Co-authored-by: Etan Kissling <etan@status.im>
Post-merge blocks contain all information to directly obtain RANDAO
without having to load any additional info. Take advantage of that to
further accelerate `ShufflingRef` computation. Note that it is still
necessary to verify that `blck` / `state` share a sufficiently recent
ancestor for the purpose of computing attester shufflings.
- new: 243.71s, 239.67s, 237.32s, 238.36s, 239.57s
- old: 251.33s, 234.29s, 249.28s, 237.03s, 236.78s
Current RANDAO recovery logic is quite complex as it optimizes for the
minimum amount of database reads. Loading blocks isn't the bottleneck
though, so rather make the implementation more concise by avoiding the
complex strategy planning step. Note that this also prepares for an even
faster implementation for post-merge blocks in the future that extracts
RANDAO from `ExecutionPayload` directly if available, so even in cases
where efficiency is slightly lower, only historical data is affected.
`time nim c -r tests/test_blockchain_dag` (cached binary):
- new: 145.45s, 133.59s, 144.65s, 127.69s, 136.14s
- old: 149.15s, 150.84s, 135.77s, 137.49s, 133.89s
* Perform block pre-check before validating execution
When syncing, blocks have not been gossip-validated and are therefore
prone to trivial faults like being known-unviable, duplicate or missing
their parent.
In addition, the duplicate-block check in BlockProcessor was not
considering the quarantine flow and would therefore cause
recently-quarantined blocks to be silenty dropped when their parent
appears delaying the sync end-game and thus causing longer startup
resync time.
This PR verifies trivial conditions before performing execution
validation thus avoiding duplicates and missing parents alike.
It also ensures that the fast-sync EL mode is used for finalized blocks
even if the EL is timing out / slow to respond - this allows the CL to
complete its sync faster and switch to "normal" lock-step at the head of
the chain more quickly, thus also allowing the EL to access the latest
consensensus information earlier.
* oops
* remove unused constant
When the requestmanager is busy fetching blocks, the queue might get
filled with multiple entries of the same root - since there is no
deduplication, requests containing the same root multiple times will be
sent out.
Also, because the items sit in the queue for a long time potentially,
the request might be stale by the time that the manager is ready with
the previous request.
This PR removes the queue and directly fetches the blocks to download
from the quarantine which solves both problems (the quarantine already
de-duplicates and is clean of stale information).
Removing the queue for blobs is left for a future PR.
Co-authored-by: tersec <tersec@users.noreply.github.com>
* early exit `commonAncestor` when comparing with `finalizedHead`
As all `BlockRef` lead to `finalizedHead` (`parent == nil`),
can shortcut in that situation and immediately return `finalizedHead`
if passed as one of the arguments.
* typo in comment
* add test from #5152
Co-authored-by: tersec <tersec@users.noreply.github.com>
* add note about test complexity
* regenerate test summary
---------
Co-authored-by: tersec <tersec@users.noreply.github.com>
These tables can't be deleted from (read-only) and would be too slow to
delete from anyway due to the inefficient storage format in use.
* slow down startup clearing too
* remove unused del function
`produceSyncAggregate` is called in new slot when block is produced,
while the other functions in `sync_committee_msg_pool` are called in
previous slot. So, need to subtract 1 slot when producing sync aggregate
to accept the signatures using the old digest during fork transition.
* also pack attestations where LMD vote is orphaned
When `attestation.data.beacon_block_root` gets orphaned, attestations
with a good `attestation.data.target.root` may still be valuable.
The LMD GHOST vote is not relevant for attestation rewards.
Switch to use the FFG vote (`attestation.data.target.root`) instead,
gossip validation ensures it is an ancestor of `beacon_block_root`.
* lint
* Clarify addOrphan error/logging
addOrphan returned a bool to indicate success. Change this to a Result
so that different errors can be distinguished.
* Update beacon_chain/consensus_object_pools/block_quarantine.nim
Co-authored-by: tersec <tersec@users.noreply.github.com>
* Update beacon_chain/gossip_processing/gossip_validation.nim
---------
Co-authored-by: tersec <tersec@users.noreply.github.com>
* replace optimisticRoots table with field in BlockRef
* copyright year
* mark finalized blocks as verified on load
* Update beacon_chain/consensus_object_pools/block_dag.nim
Co-authored-by: Etan Kissling <etan@status.im>
* expand non-optimistic block checking to all pre-merge blocks; refactor markBlockVerified to use BlockRef rather than block root and remove superfluous caller in newPayload path replaced by addResolvedHeadBlock BlockRef construction
* don't treat finalized block specially; VALID status is sticky
---------
Co-authored-by: Etan Kissling <etan@status.im>
`SyncCommitteeMsgPool` grouped messages by their `beacon_block_root`.
This is problematic around sync committee period boundaries and forks.
Around sync committee period boundaries, members from both the current
and next sync committee may sign the same `beacon_block_root`; mixing
the signatures from both committees together is a mistake. Likewise,
around fork transitions, the `signing_root` changes, so those messages
also need to be segregated.
When an uncached `ShufflingRef` is requested, we currently replay state
which can take several seconds. Acceleration is possible by:
1. Start from any state with locked-in `get_active_validator_indices`.
Any blocks / slots applied to such a state can only affect that
result for future epochs, so are viable for querying target epoch.
`compute_activation_exit_epoch(state.slot.epoch) > target.epoch`
2. Determine highest common ancestor among `state` and `target.blck`.
At the ancestor slot, same rules re `get_active_validator_indices`.
`compute_activation_exit_epoch(ancestorSlot.epoch) > target.epoch`
3. We now have a `state` that shares history with `target.blck` up
through a common ancestor slot. Any blocks / slots that the `state`
contains, which are not part of the `target.blck` history, affect
`get_active_validator_indices` at epochs _after_ `target.epoch`.
4. Select `state.randao_mixes[N]` that is closest to common ancestor.
Either direction is fine (above / below ancestor).
5. From that RANDAO mix, mix in / out all RANDAO reveals from blocks
in-between. This is just an XOR operation, so fully reversible.
`mix = mix xor SHA256(blck.message.body.randao_reveal)`
6. Compute the attester dependent slot from `target.epoch`.
`if epoch >= 2: (target.epoch - 1).start_slot - 1 else: GENESIS_SLOT`
7. Trace back from `target.blck` to the attester dependent slot.
We now have the destination for which we want to obtain RANDAO.
8. Mix in all RANDAO reveals from blocks up through the `dependentBlck`.
Same method, no special handling necessary for epoch transitions.
9. Combine `get_active_validator_indices` from `state` at `target.epoch`
with the recovered RANDAO value at `dependentBlck` to obtain the
requested shuffling, and construct the `ShufflingRef` without replay.
* more tests and simplify logic
* test with different number of deposits per branch
* Update beacon_chain/consensus_object_pools/blockchain_dag.nim
Co-authored-by: Jacek Sieka <jacek@status.im>
* `commonAncestor` tests
* lint
---------
Co-authored-by: Jacek Sieka <jacek@status.im>
* Incremental pruning
When turning on pruning the first time the current pruning algorithm
will prune the full database at startup. This delays restart
unnecessarily, since all of the pruned space is not needed at once.
This PR introduces incremental pruning such that we will never prune
more than 32 blocks or the sync speed, whichever is higher.
This mode is expected to become default in a follow-up release.
Post-Capella, historical roots are computed from historical summaries
instead of being directly stored in the beacon state.
Slightly messy to pass both lists around - this is done to avoid
computing the historical root unnecessarily.
The consensus-spec-tests already cover the scenarios of our custom test
runner, so the custom tests can be removed. Also cleans up unused config
flags and related unreachable logic.
The 'peek' name was incorrect as it was actually removing from the
table. It was consequently used incorrectly in block processing: the
blobless block wasn't returned to the table when it should be.
* Simplify block quarantine blobless
The quarantine blobless table was initially keyed off of (Eth2Digest,
ValidatorSig). This was modelled off the orphan table. The presence of
the signature in the key is necessary for orphans, because we can't
verify the signature for an orphan. That is not the case for a
blobless block, where the signature can be verified.
So this PR changes the blobless block table to be keyed off a
Eth2Digest only. This simplifies the retrieval and handling of
blobless blocks.
* review feedback
Just the variable, not yet `lcDataForkAtStateFork` / `atStateFork`.
- Shorten comment in `light_client.nim` to keep line width
- Do not rename `stateFork` mention in `runProposalForkchoiceUpdated`.
- Do not rename `stateFork` in `getStateField(dag.headState, fork)`
Rest is just a mechanical mass replace
When using `--history=prune`, `dag.tail.slot` may advance beyond the
configured light client data retention period. Update the LC logic so
that the `dag.tail.slot` is no longer considered for LC pruning.
It is still considered to check whether new data can be produced.
* Support for driving multiple EL nodes from a single Nimbus BN
Full list of changes:
* Eth1Monitor has been renamed to ELManager to match its current
responsibilities better.
* The ELManager is no longer optional in the code (it won't have
a nil value under any circumstances).
* The support for subscribing for headers was removed as it only
worked with WebSockets and contributed significant complexity
while bringing only a very minor advantage.
* The `--web3-url` parameter has been deprecated in favor of a
new `--el` parameter. The new parameter has a reasonable default
value and supports specifying a different JWT for each connection.
Each connection can also be configured with a different set of
responsibilities (e.g. download deposits, validate blocks and/or
produce blocks). On the command-line, these properties can be
configured through URL properties stored in the #anchor part of
the URL. In TOML files, they come with a very natural syntax
(althrough the URL scheme is also supported).
* The previously scattered EL-related state and logic is now moved
to `eth1_monitor.nim` (this module will be renamed to `el_manager.nim`
in a follow-up commit). State is assigned properly either to the
`ELManager` or the to individual `ELConnection` objects where
appropriate.
The ELManager executes all Engine API requests against all attached
EL nodes, in parallel. It compares their results and if there is a
disagreement regarding the validity of a certain payload, this is
detected and the beacon node is protected from publishing a block
with a potential execution layer consensus bug in it.
The BN provides metrics per EL node for the number of successful or
failed requests for each type Engine API requests. If an EL node
goes offline and connectivity is resoted later, we report the
problem and the remedy in edge-triggered fashion.
* More progress towards implementing Deneb block production in the VC
and comparing the value of blocks produced by the EL and the builder
API.
* Adds a Makefile target for the zhejiang testnet
* Local sim impovements
* Added support for running Capella and EIP-4844 simulations
by downloading the correct version of Geth.
* Added support for using Nimbus remote signer and Web3Signer.
Use 2 out of 3 threshold signing configuration in the mainnet
configuration and regular remote signing in the minimal one.
* The local testnet simulation can now use a payload builder.
This is currently not activated in CI due to lack of automated
procedures for installing third-party relays or builders.
You are adviced to use mergemock for now, but for most realistic
results, we can create a simple builder based on the nimbus-eth1
codebase that will be able to propose transactions from the regular
network mempool.
* Start the simulation from a merged state. This would allow us
to start removing pre-merge functionality such as the gossip
subsciption logic. The commit also removes the merge-forcing
hack installed after the TTD removal.
* Consolidate all the tools used in the local simulation into a
single `ncli_testnet` binary.
Other changes:
Renamed the `EIP_4844_FORK_*` config constants to `DENEB_FORK_*` as
this matches the latest spec and it's already used in the official
Sepolia config.
By pre-seeding the sync committee cache when applying blocks, we avoid a
significantly expensive validator set traversal / sync committee index
construction during sync / block application - 20-30% sync speedup
post-altair.
* also cache/reload total active balance for another cool 10%
Extends fork choice state to also track slot numbers to improve accuracy
of `/eth/v1/debug/fork_choice` endpoint. Autoenable this API on devnet,
and disable some extra checks on devnet to aid focused testing efforts.
Align fork choice pruning logic with API based on checkpoints vs root.
* clean up some Nim 1.2 workarounds
* re-add notes about JS backend
* another proc/noSideEffect -> func
* revert ncli/ncli_common.nim changes; 19969 evidently wasn't backported to 1.6
To allow LC data retention longer than the one for historic states,
introduce persistent DB caches for `current_sync_committee` and
`LightClientHeader` for finalized epoch boundary blocks.
This way, historic `LightClientBootstrap` requests may still be honored
even after pruning. Note that historic `LightClientUpdate` requests are
already answered using fully persisted objects, so don't need changes.
Sync committees and headers are cached on finalization of new data.
For existing data, info is lazily cached on first access.
Co-authored-by: Jacek Sieka <jacek@status.im>
When the epoch boundary block is missed, we incorrectly assume that the
next couple blocks improve finality, leading to repeated pushes of the
same light client finality update and incorrectly ignoring some gossip.
* exit/validatorchange pool includes BLS to execution messages; REST
support for new pool
* catch failed individual futures
* increase BLS changes bound and keep BLS seen consistent with subpool
* deque capacities should be powers of 2
Distinguish between those code locations that need to be updated on each
light client data format change, and those others that should generally
be fine, as long as a valid light client object is processed.
The former are tagged with static assert for `LightClientDataFork.high`.
The latter are changed to `lcDataFork > LightClientDataFork.None` to
indicate that they depend only on presence of any valid object.
Also bundled a few minor cleanups and fixes.
Also add `Forky` type for `LightClientStore` and minor fixes / cleanups.
The light client data structures were changed to accommodate additional
fields in future forks (e.g., to also hold execution data).
There is a minor change to the JSON serialization, where the `header`
properties are now nested inside a `LightClientHeader`.
The SSZ serialization remains compatible.
See https://github.com/ethereum/consensus-specs/pull/3190
and https://github.com/ethereum/beacon-APIs/pull/287
In a future fork, light client data will be extended with execution info
to support more use cases. To anticipate such an upgrade, introduce
`Forky` and `Forked` types, and ready the database schema.
Because the mapping of sync committee periods to fork versions is not
necessarily unique (fork schedule not in sync with period boundaries),
an additional column is added to `period` -> `LightClientUpdate` table.
* correctly report ignored contributions in metrics
* avoid counting subset contributions in vmon (bring in line with
attestation aggregates)
* avoid signature checks for subset attestations
A being a non-strict subset is a sufficient condition to ignore.
Introduce (optional) pruning of historical data - a pruned node will
continue to answer queries for historical data up to
`MIN_EPOCHS_FOR_BLOCK_REQUESTS` epochs, or roughly 5 months, capping
typical database usage at around 60-70gb.
To enable pruning, add `--history=prune` to the command line - on the
first start, old data will be cleared (which may take a while) - after
that, data is pruned continuously.
When pruning an existing database, the database will not shrink -
instead, the freed space is recycled as the node continues to run - to
free up space, perform a trusted node sync with a fresh database.
When switching on archive mode in a pruned node, history is retained
from that point onwards.
History pruning is scheduled to be enabled by default in a future
release.
In this PR, `minimal` mode from #4419 is not implemented meaning
retention periods for states and blocks are always the same - depending
on user demand, a future PR may implement `minimal` as well.
With https://github.com/status-im/nimbus-eth2/pull/4420 implemented, the
checks that we perform are equivalent to those of a `SYNCING` EL - as
such, we can treat missing EL the same as SYNCING and proceed with an
optimistic sync.
This mode of operation significantly speeds up recovery after an offline
EL event because the CL is already synced and can immediately inform the
EL of the latest head.
It also allows using a beacon node for consensus archival queries
without an execution client.
* deprecate `--optimistic` flag
* log block details on EL error, soften log level because we can now
continue to operate
* `UnviableFork` -> `Invalid` when block hash verification fails -
failed hash verification is not a fork-related block issue
When not backfilling all the way to genesis (#4421), it becomes more
useful to start rebuilding the historical indices from an arbitrary
starting point.
To rebuild the index from non-genesis, a state and an unbroken block
history is needed - here, we allow loading the state from an era file
and recreating the history from there onwards.
* speed up partial era state loading
When backfilling, we only need to download blocks that are newer than
MIN_EPOCHS_FOR_BLOCK_REQUESTS - the rest cannot reliably be fetched from
the network and does not have to be provided to others.
This change affects only trusted-node-synced clients - genesis sync
continues to work as before (because it needs to construct a state by
building it from genesis).
Those wishing to complete a backfill should do so with era files
instead.
Trigger ANSI art on upgrade to Capella, similar to the merge.
Future extension could log blinking art when user successfully managed
to get BLS to Execution change included into a block for a validator.
Art created by http://beatscribe.com/ (beatscribe#1008 on Discord)
Other changes:
* More optimal search for TTD block.
* Add timeouts to all REST requests during trusted node sync.
Fixes#4037
* Removed support for storing a deposit snapshot in the network
metadata.
* Types and scaffolding for EIP-4844
This commit adds the EIP-4844 spec types, and fills in
scaffolding/boilerplate for the use of these types across the repo.
None of the actual EIP-4844 logic is introduced yet.
This follows the pattern used by @tersec when introducing Capella (#4276).
* use eth2-networks fork
* review feedback: add static check EIP4844_FORK_EPOCH == FAR_FUTURE_EPOCH
* review feedback: remove EIP4844 from /eth/v1/config/spec response
* Cleanup / review feedback
* Fix REST test
* implement several capellaImplementationMissing points
* don't register validator activity for not-active validators
* don't check validator indices already coming out of committees which exist; must be active validators, or else other deeper bugs
* Initial commit.
* NextAttestationEntry type.
* Add doppelgangerCheck and actual check.
* Recover deleted check.
* Remove NextAttestainEntry changes.
* More cleanups for NextAttestationEntry.
* Address review comments.
* Remove GENESIS_EPOCH specific check branch.
* Decrease number of full epochs for doppelganger check in VC.
Co-authored-by: zah <zahary@status.im>
We currently use `BlockError` for both beacon blocks and LC objects.
In light of EIP4844, we will likely also use it for blob sidecars.
To avoid confusion, renaming it to a more generic `VerifierError`,
and update its documentation to be more generic.
To avoid long lines as a followup, also renaming the `block_processor`'s
`BlockProcessingCompleted.completed`->`ProcessingStatus.completed` and
`BlockProcessingCompleted.notCompleted`->`ProcessingStatus.notCompleted`
This PR removes a bunch of code to make TNS aware of era files, avoiding
a duplicated backfill when era files are available.
* reuse chaindag for loading backfill state, replacing the TNS homebrew
* fix era block iteration to skip empty slots
* add tests for `can_advance_slots`
* move duty tracking code to `ActionTracker`
* fix earlier duties overwriting later ones
* re-run subnet selection when new duty appears
* log upcoming duties as soon as they're known (vs 4 epochs before)
Currently, we require genesis and a checkpoint block and state to start
from an arbitrary slot - this PR relaxes this requirement so that we can
start with a state alone.
The current trusted-node-sync algorithm works by first downloading
blocks until we find an epoch aligned non-empty slot, then downloads the
state via slot.
However, current
[proposals](https://github.com/ethereum/beacon-APIs/pull/226) for
checkpointing prefer finalized state as
the main reference - this allows more simple access control and caching
on the server side - in particular, this should help checkpoint-syncing
from sources that have a fast `finalized` state download (like infura
and teku) but are slow when accessing state via slot.
Earlier versions of Nimbus will not be able to read databases created
without a checkpoint block and genesis. In most cases, backfilling makes
the database compatible except where genesis is also missing (custom
networks).
* backfill checkpoint block from libp2p instead of checkpoint source,
when doing trusted node sync
* allow starting the client without genesis / checkpoint block
* perform epoch start slot lookahead when loading tail state, so as to
deal with the case where the epoch start slot does not have a block
* replace `--blockId` with `--state-id` in TNS command line
* when replaying, also look at the parent of the last-known-block (even
if we don't have the parent block data, we can still replay from a
"parent" state) - in particular, this clears the way for implementing
state pruning
* deprecate `--finalized-checkpoint-block` option (no longer needed)
* Allow chain dag without genesis / block
This PR enables the initialization of the dag without access to blocks
or genesis state - it is a prerequisite for implementing a number of
interesting features:
* checkpoint sync without any block download
* pruning of blocks and states
* backfill checkpoint block
When EL `newPayload` is slow (e.g., Raspberry Pi with Besu), the epoch
and shuffling caches tend to fill up with multiple copies per epoch when
processing gossip and performing validator duties close to wall slot.
The old strategy of evicting oldest epoch led to the same item being
evicted over and over, leading to blocking of over 5 minutes in extreme
cases where alternate epochs/shuffling got loaded repeatedly.
Changing the cache eviction strategy to least-recently-used seems to
improve the situation drastically. A simple implementation was selected
based on single linked-list without a hashtable.
When backfilling LC updates (`--light-client-data-import-mode=full`),
the highest participation update is computed without ensuring that the
finalized header is in the same period. Updates sharing same period for
both finalized and attested headers should be preferred.
Fixes a bug leading to suboptimal update selection.
* avoid database race-condition inconsistency after fcU `INVALID` then crash
* ensure head doesn't fall behind finalized; add more tests for head movement/reloading DAG
When the BN's head is reorged while shut down, reloading the BN will not
assign `BlockRef` to alternate branches. However, blocks from other
branches are still present in the database, leading to their descendants
incorrectly marked as `UnviableFork`. By restricting the check to blocks
that have been finalized, they should be reported as `MissingParent`
instead, eventually re-assigning a `BlockRef` to them.
Since these files may have been created in a previous run or manually,
we want to keep loading them even on nodes that don't enable the
keystore API (for example static setups)
Other changes:
* log keystore loading progressively (#3699)
* print initial fee recipient when loading validators
* log dynamic fee recipient updates
* more efficient forkchoiceUpdated usage
* await rather than asyncSpawn; ensure head update before dag.updateHead
* use action tracker rather than attached validators to check for next slot proposal; use wall slot + 1 rather than state slot + 1 to correctly check when missing blocks
* re-add two-fcU case for when newPayload not VALID
* check dynamicFeeRecipientsStore for potential proposal
* remove duplicate checks for whether next proposer
When the BN-embedded LC makes sync progress, pass the corresponding
execution block hash to the EL via `engine_forkchoiceUpdatedV1`.
This allows the EL to sync to wall slot while the chain DAG is behind.
Renamed `--light-client` to `--sync-light-client` for clarity, and
`--light-client-trusted-block-root` to `--trusted-block-root` for
consistency with `nimbus_light_client`.
Note that this does not work well in practice at this time:
- Geth sticks to the optimistic sync:
"Ignoring payload while snap syncing" (when passing the LC head)
"Forkchoice requested unknown head" (when updating to LC head)
- Nethermind syncs to LC head but does not report ancestors as VALID,
so the main forward sync is still stuck in optimistic mode:
"Pre-pivot block, ignored and returned Syncing"
To aid EL client teams in fixing those issues, having this available
as a hidden option is still useful.
tersec
Jacek Sieka
Etan Kissling
henridf
zah
Eugene Kabanov