b8a32419b8
* async batch verification When batch verification is done, the main thread is blocked reducing concurrency. With this PR, the new thread signalling primitive in chronos is used to offload the full batch verification process to a separate thread allowing the main threads to continue async operations while the other threads verify signatures. Similar to previous behavior, the number of ongoing batch verifications is capped to prevent runaway resource usage. In addition to the asynchronous processing, 3 addition changes help drive throughput: * A loop is used for batch accumulation: this prevents a stampede of small batches in eager mode where both the eager and the scheduled batch runner would pick batches off the queue, prematurely picking "fresh" batches off the queue * An additional small wait is introduced for small batches - this helps create slightly larger batches which make better used of the increased concurrency * Up to 2 batches are scheduled to the threadpool during high pressure, reducing startup latency for the threads Together, these changes increase attestation verification throughput under load up to 30%. * fixup * Update submodules * fix blst build issues (and a PIC warning) * bump --------- Co-authored-by: Zahary Karadjov <zahary@gmail.com> |
||
---|---|---|
.. | ||
README.md | ||
batch_validation.nim | ||
block_processor.nim | ||
eth2_processor.nim | ||
gossip_validation.nim | ||
light_client_processor.nim | ||
optimistic_processor.nim |
README.md
Gossip Processing
This folder holds a collection of modules to:
- validate raw gossip data before
- rebroadcasting it (potentially aggregated)
- sending it to one of the consensus object pools
Validation
Gossip validation is different from consensus verification in particular for blocks.
- Blocks: https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/phase0/p2p-interface.md#beacon_block
- Attestations (aggregated): https://github.com/ethereum/consensus-specs/blob/v1.4.0-beta.0/specs/phase0/p2p-interface.md#beacon_aggregate_and_proof
- Attestations (unaggregated): https://github.com/ethereum/consensus-specs/blob/v1.3.0/specs/phase0/p2p-interface.md#attestation-subnets
- Voluntary exits: https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/phase0/p2p-interface.md#voluntary_exit
- Proposer slashings: https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/phase0/p2p-interface.md#proposer_slashing
- Attester slashing: https://github.com/ethereum/consensus-specs/blob/v1.4.0-alpha.3/specs/phase0/p2p-interface.md#attester_slashing
There are multiple consumers of validated consensus objects:
- a
ValidationResult.Accept
output triggers rebroadcasting in libp2p- We jump into method
validate(PubSub, Message)
in libp2p/protocols/pubsub/pubsub.nim - which was called by
rpcHandler(GossipSub, PubSubPeer, RPCMsg)
- We jump into method
- a
blockValidator
message enqueues the validated object to the processing queue inblock_processor
blockQueue: AsyncQueue[BlockEntry]
(shared with request_manager and sync_manager)- This queue is then regularly processed to be made available to the consensus object pools.
- a
xyzValidator
message adds the validated object to a pool in eth2_processor- Attestations (unaggregated and aggregated) get collected into batches.
- Once a threshold is exceeded or after a timeout, they get validated together using BatchCrypto.
Security concerns
As the first line of defense in Nimbus, modules must be able to handle bursts of data that may come:
- from malicious nodes trying to DOS us
- from long periods of non-finality, creating lots of forks, attestations