## Proposed Changes
In an attempt to fix OOM issues and database consistency issues observed by some users after the introduction of compaction in v0.3.4, this PR makes the following changes:
* Run compaction less often: roughly every 1024 epochs, including after long periods of non-finality. I think the division check proposed by Paul is pretty solid, and ensures we don't miss any events where we should be compacting. LevelDB lacks an easy way to check the size of the DB, which would be another good trigger.
* Make it possible to disable the compaction on finalization using `--auto-compact-db=false`
* Make it possible to trigger a manual, single-threaded foreground compaction on start-up using `--compact-db`
* Downgrade the pruning log to `DEBUG`, as it's particularly noisy during sync
I would like to ship these changes to affected users ASAP, and will document them further in the Advanced Database section of the book if they prove effective.
## Issue Addressed
A peer might send a lot of requests that comply to the rate limit and the disconnect, this humongous pr makes sure we don't process them if the peer is not connected
This PR adds a number of improvements:
- Downgrade a warning log when we ignore blocks for gossipsub processing
- Revert a a correction to improve logging of peer score changes
- Shift syncing DB reads off the core-executor allowing parallel processing of large sync messages
- Correct the timeout logic of RPC chunk sends, giving more time before timing out RPC outbound messages.
## Issue Addressed
- RPC Errors were being logged twice: first in the peer manager and then again in the router, so leave just the peer manager's one
- The "reduce peer count" warn message gets thrown to the user for every missed chunk, so instead print it when the request times out and also do not include there info that is not relevant to the user
- The processor didn't have the service tag so add it
- Impl `KV` for status message
- Do not downscore peers if we are the ones that timed out
Other small improvements
## Issue Addressed
Resolves#1801
## Proposed Changes
Verify queries to `attestation_data` are for no later than `current_slot + 1`. If they are later than this, return a 400.
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
#1856
## Proposed Changes
- For clarity, the router's processor now only decides if a peer is compatible and it disconnects it or sends it to sync accordingly. No logic here regarding how useful is the peer.
- Update peer_sync_info's rules
- Add an `IrrelevantPeer` sync status to account for incompatible peers (maybe this should be "IncompatiblePeer" now that I think about it?) this state is update upon receiving an internal goodbye in the peer manager
- Misc code cleanups
- Reduce the need to create `StatusMessage`s (and thus, `Arc` accesses )
- Add missing calls to update the global sync state
The overall effect should be:
- More peers recognized as Behind, and less as Unknown
- Peers identified as incompatible
## Issue Addressed
- Asymmetric pings - Currently with symmetric ping intervals, lighthouse nodes race each other to ping often ending in simultaneous ping connections. This shifts the ping interval to be asymmetric based on inbound/outbound connections
- Correct inbound/outbound peer-db registering - It appears we were accounting inbound as outbound and vice versa in the peerdb, this has been corrected
- Improved logging
There is likely more to come - I'll leave this open as we investigate further testnets
## Issue Addressed
Using `heaptrack` I could see that ~75% of Lighthouse temporary allocations are caused by temporary string allocations here.
## Proposed Changes
Reduces temporary `String` allocations when updating metrics in the `network` crate. The solution isn't perfect since we rebuild our caches with each call, but it's a significant improvement.
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
Correctly handles peer state transitions on gossipsub changes + refactors handling of peer state transitions into one function used for lighthouse score changes and gossipsub score changes.
Co-authored-by: Age Manning <Age@AgeManning.com>
## Issue Addressed
NA
## Proposed Changes
Increases the target peers for a subnet, so that subnet queries are executed until we have at least the minimum required peers for a mesh (`MESH_N_LOW`). We keep the limit of `6` target peers for aggregated subnet discovery queries, therefore the size (and the time needed) for a query doesn't change.
## Issue Addressed
#1606
## Proposed Changes
Uses dynamic gossipsub scoring parameters depending on the number of active validators as specified in https://gist.github.com/blacktemplar/5c1862cb3f0e32a1a7fb0b25e79e6e2c.
## Additional Info
Although the parameters got tested on Medalla, extensive testing using simulations on larger networks is still to be done and we expect that we need to change the parameters, although this might only affect constants within the dynamic parameter framework.
## Issue Addressed
Resolves#1809Resolves#1824Resolves#1818Resolves#1828 (hopefully)
## Proposed Changes
- add `validator_index` to the proposer duties endpoint
- add the ability to query for historical proposer duties
- `StateId` deserialization now fails with a 400 warp rejection
- add the `validator_balances` endpoint
- update the `aggregate_and_proofs` endpoint to accept an array
- updates the attester duties endpoint from a `GET` to a `POST`
- reduces the number of times we query for proposer duties from once per slot per validator to only once per slot
Co-authored-by: realbigsean <seananderson33@gmail.com>
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
Closes#1866
## Proposed Changes
* Compact the database on finalization. This removes the deleted states from disk completely. Because it happens in the background migrator, it doesn't block other database operations while it runs. On my Medalla node it took about 1 minute and shrank the database from 90GB to 9GB.
* Fix an inefficiency in the pruning algorithm where it would always use the genesis checkpoint as the `old_finalized_checkpoint` when running for the first time after start-up. This would result in loading lots of states one-at-a-time back to genesis, and storing a lot of block roots in memory. The new code stores the old finalized checkpoint on disk and only uses genesis if no checkpoint is already stored. This makes it both backwards compatible _and_ forwards compatible -- no schema change required!
* Introduce two new `INFO` logs to indicate when pruning has started and completed. Users seem to want to know this information without enabling debug logs!
Updates libp2p to the latest version.
This adds tokio 0.3 support and brings back yamux support.
This also updates some discv5 configuration parameters for leaner discovery queries
## Issue Addressed
NA
## Proposed Changes
Lets the networking behavior ignore messages of peers that are not connected. Furthermore, old peers are not removed from the peerdb based on score anymore but based on the disconnection time.
## Issue Addressed
- Related to #1691
## Proposed Changes
Adds the following API endpoints:
- `GET lighthouse/eth1/syncing`: status about how synced we are with Eth1.
- `GET lighthouse/eth1/block_cache`: all locally cached eth1 blocks.
- `GET lighthouse/eth1/deposit_cache`: all locally cached eth1 deposits.
Additionally:
- Moves some types from the `beacon_node/eth1` to the `common/eth2` crate, so they can be used in the API without duplication.
- Allow `update_deposit_cache` and `update_block_cache` to take an optional head block number to avoid duplicate requests.
## Additional Info
TBC
## Issue Addressed
Fixes head syncing
## Proposed Changes
- Get back to statusing peers after removing chain segments and making the peer manager deal with status according to the Sync status, preventing an old known deadlock
- Also a bug where a chain would get removed if the optimistic batch succeeds being empty
## Additional Info
Tested on Medalla and looking good
## Issue Addressed
NA
## Proposed Changes
As raised by @hermanjunge in a DM, the `http_api` tests have been observed taking 100+ minutes on debug. This PR:
- Moves the `http_api` tests to only run in release.
- Groups some `http_api` tests to reduce test-setup overhead.
## Additional Info
NA
## Issue Addressed
Michael's comment here: https://github.com/sigp/lighthouse/issues/1434#issuecomment-708834079Resolves#1808
## Proposed Changes
- Add query param `id` and `status` to the `validators` endpoint
- Add string serialization and deserialization for `ValidatorStatus`
- Drop `Epoch` from `ValidatorStatus` variants
## Additional Info
Please provide any additional information. For example, future considerations
or information useful for reviewers.
## Issue Addressed
Sync edge case when we get an empty optimistic batch that passes validation and is inside the download buffer. Eventually the chain would reach the batch and treat it as an ugly state.
## Proposed Changes
- Handle the edge case advancing the chain's target + code clarification
- Some largey changes for readability + ergonomics since rust has try ops
- Better handling of bad batch and chain states
## Issue Addressed
NA
## Proposed Changes
Currently a banned peer will remain banned indefinitely as long as update is called on the score struct regularly. This fixes this bug and the score decay starts after `BANNED_BEFORE_DECAY` seconds after banning.
## Issue Addressed
Closes#1504Closes#1505
Replaces #1703Closes#1707
## Proposed Changes
* Update BLST and Milagro to versions compatible with BLSv4 spec
* Update Lighthouse to spec v1.0.0-rc.0, and update EF test vectors
* Use the v1.0.0 constants for `MainnetEthSpec`.
* Rename `InteropEthSpec` -> `V012LegacyEthSpec`
* Change all constants to suit the mainnet `v0.12.3` specification (i.e., Medalla).
* Deprecate the `--spec` flag for the `lighthouse` binary
* This value is now obtained from the `config_name` field of the `YamlConfig`.
* Built in testnet YAML files have been updated.
* Ignore the `--spec` value, if supplied, log a warning that it will be deprecated
* `lcli` still has the spec flag, that's fine because it's dev tooling.
* Remove the `E: EthSpec` from `YamlConfig`
* This means we need to deser the genesis `BeaconState` on-demand, but this is fine.
* Swap the old "minimal", "mainnet" strings over to the new `EthSpecId` enum.
* Always require a `CONFIG_NAME` field in `YamlConfig` (it used to have a default).
## Additional Info
Lots of breaking changes, do not merge! ~~We will likely need a Lighthouse v0.4.0 branch, and possibly a long-term v0.3.0 branch to keep Medalla alive~~.
Co-authored-by: Kirk Baird <baird.k@outlook.com>
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
#1825
## Proposed Changes
Since we penalize more blocks by range requests that have large steps, it is possible to get requests that will never be processed. We were not informing peers about this requests and also logging CRIT that is no longer relevant. Later we should check if more sophisticated handling for those requests is needed
## Issue Addressed
NA
## Proposed Changes
Fixes a bug which causes the node to reject valid eth1 nodes.
- Fix core bug: failure to apply `YamlConfig` values to `ChainSpec`.
- Add a test to prevent regression in this specific case.
- Fix an invalid log message
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
- Panic or return error if we overflow `usize` in SSZ decoding/encoding derive macros.
- I claim that the panics can only be triggered by a faulty type definition in lighthouse, they cannot be triggered externally on a validly defined struct.
- Use `Ordering` instead of some `if` statements, as demanded by clippy.
- Remove some old clippy `allow` that seem to no longer be required.
- Add comments to interesting clippy statements that we're going to continue to ignore.
- Create #1713
## Additional Info
NA
## Issue Addressed
- Related to #1691
## Proposed Changes
- Add `DEPOSIT_CHAIN_ID` and `DEPOSIT_NETWORK_ID` to `config.yaml`.
- Pass the `DEPOSIT_NETWORK_ID` to the `eth1::Service`.
- Remove the unused `MAX_EPOCHS_PER_CROSSLINK` from the `altona` and `medalla` configs (see [spec commit](2befe90032 (diff-efb845ac2ebd4aafbc23df40f47ce25699255064e99d36d0406d0a14ca7953ec))).
- Change from compressing the whole testnet directory, to only compressing the genesis state file. This is the only file we need to compress and *not* compressing the others makes them work nicely with git.
- We can modify the boot nodes, configs, etc. without incurring an eternal binary-blob cost on our git history.
- This change is backwards compatible (i.e., non-breaking).
## Additional Info
NA
## Issue Addressed
Further testing revealed another edge case where we attempt to unban a peer that can be in a disconnected start. Although this causes no real issue, it does log an error to the user.
This PR adds a check to prevent this edge case and prevents the error being logged to the user.
This PR prevents a user-facing error.
It prevents optimistically unbanning a peer and instead checks the state of the peer before requesting the peers state to be unbanned.
As @AgeManning mentioned the newest libp2p version had some problems and got downgraded again on lighthouse master. This is an intermediate version that makes no problems and only adds a small change of allowing only one topic per message.
## Description
This downgrades the recent libp2p upgrade.
There were issues with the RPC which prevented syncing of the chain and this upgrade needs to be further investigated.
## Overview
This updates lighthouse to discovery v5.1
Note: This makes lighthouse's discovery not compatible with any previous version. Lighthouse cannot discover peers or send/receive ENR's from any previous version. This is a breaking change.
This resolves#1605
## Description
This increases the logging of the underlying UPnP tasks to inform the user of UPnP error/success.
This also decreases the batch syncing size to two epochs per batch.
## Description
Updates to the latest libp2p and includes gossipsub updates.
Of particular note is the limitation of a single topic per gossipsub message.
Co-authored-by: blacktemplar <blacktemplar@a1.net>
## Issue Addressed
Closes#800Closes#1713
## Proposed Changes
Implement the temporary state storage algorithm described in #800. Specifically:
* Add `DBColumn::BeaconStateTemporary`, for storing 0-length temporary marker values.
* Store intermediate states immediately as they are created, marked temporary. Delete the temporary flag if the block is processed successfully.
* Add a garbage collection process to delete leftover temporary states on start-up.
* Bump the database schema version to 2 so that a DB with temporary states can't accidentally be used with older versions of the software. The auto-migration is a no-op, but puts in place some infra that we can use for future migrations (e.g. #1784)
## Additional Info
There are two known race conditions, one potentially causing permanent faults (hopefully rare), and the other insignificant.
### Race 1: Permanent state marked temporary
EDIT: this has been fixed by the addition of a lock around the relevant critical section
There are 2 threads that are trying to store 2 different blocks that share some intermediate states (e.g. they both skip some slots from the current head). Consider this sequence of events:
1. Thread 1 checks if state `s` already exists, and seeing that it doesn't, prepares an atomic commit of `(s, s_temporary_flag)`.
2. Thread 2 does the same, but also gets as far as committing the state txn, finishing the processing of its block, and _deleting_ the temporary flag.
3. Thread 1 is (finally) scheduled again, and marks `s` as temporary with its transaction.
4.
a) The process is killed, or thread 1's block fails verification and the temp flag is not deleted. This is a permanent failure! Any attempt to load state `s` will fail... hope it isn't on the main chain! Alternatively (4b) happens...
b) Thread 1 finishes, and re-deletes the temporary flag. In this case the failure is transient, state `s` will disappear temporarily, but will come back once thread 1 finishes running.
I _hope_ that steps 1-3 only happen very rarely, and 4a even more rarely. It's hard to know
This once again begs the question of why we're using LevelDB (#483), when it clearly doesn't care about atomicity! A ham-fisted fix would be to wrap the hot and cold DBs in locks, which would bring us closer to how other DBs handle read-write transactions. E.g. [LMDB only allows one R/W transaction at a time](https://docs.rs/lmdb/0.8.0/lmdb/struct.Environment.html#method.begin_rw_txn).
### Race 2: Temporary state returned from `get_state`
I don't think this race really matters, but in `load_hot_state`, if another thread stores a state between when we call `load_state_temporary_flag` and when we call `load_hot_state_summary`, then we could end up returning that state even though it's only a temporary state. I can't think of any case where this would be relevant, and I suspect if it did come up, it would be safe/recoverable (having data is safer than _not_ having data).
This could be fixed by using a LevelDB read snapshot, but that would require substantial changes to how we read all our values, so I don't think it's worth it right now.
## Issue Addressed
Potentially resolves#1647 and sync stalls.
## Proposed Changes
The handling of the state of banned peers was inadequate for the complex peerdb data structure. We store a limited number of disconnected and banned peers in the db. We were not tracking intermediate "disconnecting" states and the in some circumstances we were updating the peer state without informing the peerdb. This lead to a number of inconsistencies in the peer state.
Further, the peer manager could ban a peer changing a peer's state from being connected to banned. In this circumstance, if the peer then disconnected, we didn't inform the application layer, which lead to applications like sync not being informed of a peers disconnection. This could lead to sync stalling and having to require a lighthouse restart.
Improved handling for peer states and interactions with the peerdb is made in this PR.
## Issue Addressed
NA
## Proposed Changes
Adds a `lighthouse/beacon/states/:state_id/ssz` endpoint to allow us to pull the genesis state from the API.
## Additional Info
NA
## Issue Addressed
Closes#1769Closes#1708
## Proposed Changes
Tweaks the op pool pruning so that the attestation pool is pruned against the wall-clock epoch instead of the finalized state's epoch. This should reduce the unbounded growth that we've seen during periods without finality.
Also fixes up the voluntary exit pruning as raised in #1708.
## Issue Addressed
- Resolves#1766
## Proposed Changes
- Use the `warp::filters::cors` filter instead of our work-around.
## Additional Info
It's not trivial to enable/disable `cors` using `warp`, since using `routes.with(cors)` changes the type of `routes`. This makes it difficult to apply/not apply cors at runtime. My solution has been to *always* use the `warp::filters::cors` wrapper but when cors should be disabled, just pass the HTTP server listen address as the only permissible origin.
## Issue Addressed
`node` endpoints in #1434
## Proposed Changes
Implement these:
```
/eth/v1/node/health
/eth/v1/node/peers/{peer_id}
/eth/v1/node/peers
```
- Add an `Option<Enr>` to `PeerInfo`
- Finish implementation of `/eth/v1/node/identity`
## Additional Info
- should update the `peers` endpoints when #1764 is resolved
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
Closes#1548
## Proposed Changes
Optimizes attester slashing choice by choosing the ones that cover the most amount of validators slashed, with the highest effective balances
## Additional Info
Initial pass, need to write a test for it
check for advanced peers and the state of the chain wrt the clock slot to decide if a chain is or not synced /transitioning to a head sync. Also a fix that prevented getting the right state while syncing heads
