## Issue Addressed
#629
## Proposed Changes
This removes banned peers from the DHT and informs discovery to block the node_id and the known source IP's associated with this node. It has the capabilities of un banning this peer after a period of time.
This also corrects the logic about banning specific IP addresses. We now use seen_ip addresses from libp2p rather than those sent to us via identify (which also include local addresses).
## Issue Addressed
In principle.. closes#1551 but in general are improvements for performance, maintainability and readability. The logic for the optimistic sync in actually simple
## Proposed Changes
There are miscellaneous things here:
- Remove unnecessary `BatchProcessResult::Partial` to simplify the batch validation logic
- Make batches a state machine. This is done to ensure batch state transitions respect our logic (this was previously done by moving batches between `Vec`s) and to ease the cognitive load of the `SyncingChain` struct
- Move most batch-related logic to the batch
- Remove `PendingBatches` in favor of a map of peers to their batches. This is to avoid duplicating peers inside the chain (peer_pool and pending_batches)
- Add `must_use` decoration to the `ProcessingResult` so that chains that request to be removed are handled accordingly. This also means that chains are now removed in more places than before to account for unhandled cases
- Store batches in a sorted map (`BTreeMap`) access is not O(1) but since the number of _active_ batches is bounded this should be fast, and saves performing hashing ops. Batches are indexed by the epoch they start. Sorted, to easily handle chain advancements (range logic)
- Produce the chain Id from the identifying fields: target root and target slot. This, to guarantee there can't be duplicated chains and be able to consistently search chains by either Id or checkpoint
- Fix chain_id not being present in all chain loggers
- Handle mega-edge case where the processor's work queue is full and the batch can't be sent. In this case the chain would lose the blocks, remain in a "syncing" state and waiting for a result that won't arrive, effectively stalling sync.
- When a batch imports blocks or the chain starts syncing with a local finalized epoch greater that the chain's start epoch, the chain is advanced instead of reset. This is to avoid losing download progress and validate batches faster. This also means that the old `start_epoch` now means "current first unvalidated batch", so it represents more accurately the progress of the chain.
- Batch status peers from the same chain to reduce Arc access.
- Handle a couple of cases where the retry counters for a batch were not updated/checked are now handled via the batch state machine. Basically now if we forget to do it, we will know.
- Do not send back the blocks from the processor to the batch. Instead register the attempt before sending the blocks (does not count as failed)
- When re-requesting a batch, try to avoid not only the last failed peer, but all previous failed peers.
- Optimize requesting batches ahead in the buffer by shuffling idle peers just once (this is just addressing a couple of old TODOs in the code)
- In chain_collection, store chains by their id in a map
- Include a mapping from request_ids to (chain, batch) that requested the batch to avoid the double O(n) search on block responses
- Other stuff:
- impl `slog::KV` for batches
- impl `slog::KV` for syncing chains
- PSA: when logging, we can use `%thing` if `thing` implements `Display`. Same for `?` and `Debug`
### Optimistic syncing:
Try first the batch that contains the current head, if the batch imports any block, advance the chain. If not, if this optimistic batch is inside the current processing window leave it there for future use, if not drop it. The tolerance for this block is the same for downloading, but just once for processing
Co-authored-by: Age Manning <Age@AgeManning.com>
## Issue Addressed
N/A
## Proposed Changes
Prevent subscribing to core gossipsub topics until after we have achieved a full sync. This prevents us censoring gossipsub channels, getting penalised in gossipsub 1.1 scoring and saves us computation time in attempting to validate gossipsub messages which we will be unable to do with a non-sync'd chain.
## Issue Addressed
Closes#1472
## Proposed Changes
Add `--staking` ~~and`staking-with-eth1-endpoint`~~ flag to improve UX for stakers.
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
- Resolves#1080
## Proposed Changes
- Call `update_pubkey_cache` only in the `build_all_caches` method and `get_validator_index` method.
## Additional Info
This does reduce the number of places the cache is updated, making it simpler. But the `get_validator_index` method is used a couple times when we are iterating through the entire validator registry (or set of active validators). Before, we would only call `update_pubkey_cache` once before iterating through all validators. So I'm not _totally_ sure this change is worth it.
## Issue Addressed
#1437
## Proposed Changes
- Make the `--wallet-password` flag optional and creates an interactive prompt if not provided.
- Make the `--wallet-name` flag optional and creates an interactive prompt if not provided.
- Add a minimum password requirement of a 12 character length.
- Update the `--stdin-passwords` flag to `--stdin-inputs` because we have non-password user inputs
## Additional Info
This reverts commit 4fca306397.
Something in the BLST update is causing SIGILLs on aarch64 non-portable builds. While we debug the issue, I think it's best if we just revert the update.
## Issue Addressed
Closes#1098
## Proposed Changes
Add a `SafeArithIter` trait with a `safe_sum` method, and use it in `state_processing`. This seems to be the only place in `consensus` where it is relevant -- i.e. where we were using `sum` and the integer_arith lint is enabled.
## Additional Info
This PR doesn't include any Clippy linting to prevent `sum` from being called. It seems there is no existing Clippy lint that suits our purpose, but I'm going to look into that and maybe schedule writing one as a lower-priority task.
This theoretically _is_ a consensus breaking change, but it shouldn't impact Medalla (or any other testnet) because `slashings` shouldn't overflow!
## Issue Addressed
Closes#1504
Closes https://github.com/sigp/lighthouse/issues/1505
## Proposed Changes
* Update `blst` to the latest version, which is more portable and includes finer-grained compilation controls (see below).
* Detect the case where a binary has been explicitly compiled with ADX support but it's missing at runtime, and report a nicer error than `SIGILL`.
## Known Issues
* None. The previous issue with `make build-aarch64` (https://github.com/supranational/blst/issues/27), has been resolved.
## Additional Info
I think we should tweak our release process and our Docker builds so that we provide two options:
Binaries:
* `lighthouse`: compiled with `modern`/`force-adx`, for CPUs 2013 and newer
* `lighthouse-portable`: compiled with `portable` for older CPUs
Docker images:
* `sigp/lighthouse:latest`: multi-arch image with `modern` x86_64 and vanilla aarch64 binary
* `sigp/lighthouse:latest-portable`: multi-arch image with `portable` builds for x86_64 and aarch64
And relevant Docker images for the releases (as per https://github.com/sigp/lighthouse/pull/1574#issuecomment-687766141), tagged `v0.x.y` and `v0.x.y-portable`
## Issue Addressed
- Resolves#1616
## Proposed Changes
If we look at the function which persists fork choice and the canonical head to disk:
1db8daae0c/beacon_node/beacon_chain/src/beacon_chain.rs (L234-L280)
There is a race-condition which might cause the canonical head and fork choice values to be out-of-sync.
I believe this is the cause of #1616. I managed to recreate the issue and produce a database that was unable to sync under the `master` branch but able to sync with this branch.
These new changes solve the issue by ignoring the persisted `canonical_head_block_root` value and instead getting fork choice to generate it. This ensures that the canonical head is in-sync with fork choice.
## Additional Info
This is hotfix method that leaves some crusty code hanging around. Once this PR is merged (to satisfy the v0.2.x users) we should later update and merge #1638 so we can have a clean fix for the v0.3.x versions.
## Issue Addressed
N/A
## Proposed Changes
Shifts the local `metadata` to `network_globals` making it accessible to the HTTP API and other areas of lighthouse.
## Additional Info
N/A
## Issue Addressed
N/A
## Proposed Changes
Adds the following check from the spec
> A reader SHOULD NOT read more than max_encoded_len(n) bytes after reading the SSZ length-prefix n from the header.
## Issue Addressed
- Resolves#1616
## Proposed Changes
Fixes a bug where we are unable to read the finalized block from fork choice.
## Detail
I had made an assumption that the finalized block always has a parent root of `None`:
e5fc6bab48/consensus/fork_choice/src/fork_choice.rs (L749-L752)
This was a faulty assumption, we don't set parent *roots* to `None`. Instead we *sometimes* set parent *indices* to `None`, depending if this pruning condition is satisfied:
e5fc6bab48/consensus/proto_array/src/proto_array.rs (L229-L232)
The bug manifested itself like this:
1. We attempt to get the finalized block from fork choice
1. We try to check that the block is descendant of the finalized block (note: they're the same block).
1. We expect the parent root to be `None`, but it's actually the parent root of the finalized root.
1. We therefore end up checking if the parent of the finalized root is a descendant of itself. (note: it's an *ancestor* not a *descendant*).
1. We therefore declare that the finalized block is not a descendant of (or eq to) the finalized block. Bad.
## Additional Info
In reflection, I made a poor assumption in the quest to obtain a probably negligible performance gain. The performance gain wasn't worth the risk and we got burnt.
## Issue Addressed
N/A
## Proposed Changes
This will consider all gossipsub messages that have either the `from`, `seqno` or `signature` field as invalid.
## Additional Info
We should not merge this until all other clients have been sending empty fields for a while.
See https://github.com/ethereum/eth2.0-specs/issues/1981 for reference
## Issue Addressed
https://github.com/ethereum/eth2.0-specs/pull/2044
## Proposed Changes
Shifts the gossipsub message id to use the first 8 bytes of the SHA256 hash of the gossipsub message data field.
## Additional Info
We should merge this in once the spec has been decided on. It will cause issues with gossipsub scoring and gossipsub propagation rates (as we won't receive IWANT) messages from clients that also haven't made this update.
## Issue Addressed
Closes#1076
## Proposed Changes
Remove an extra unnecessary decompression of the deposit public key from `process_deposit`. The key is decompressed and used to verify the signature in `verify_deposit_signature`, making this initial decompression redundant.
## Additional Info
This is _not_ a consensus-breaking change because keys which previously failed the early decompression check will not be found in the pubkey cache (they are invalid), and will be checked and rejected as part of `verify_deposit_signature`.
## Issue Addressed
#1590
## Proposed Changes
This is a temporary workaround that prevents finalized chain sync from swapping chains. I'm merging this in now until the full solution is ready.
## Issue Addressed
Malicious users could request very large block ranges, more than we expect. Although technically legal, we are now quadraticaly weighting large step sizes in the filter. Therefore users may request large skips, but not a large number of blocks, to prevent requests forcing us to do long chain lookups.
## Proposed Changes
Weight the step parameter in the RPC filter and prevent any overflows that effect us in the step parameter.
## Additional Info
## Issue Addressed
#1431
## Proposed Changes
Added an archived zip file with required files manually
## Additional Info
1) Used zip, instead of tar.gz to add a single dependency instead of two.
2) I left the download from github code for now, waiting to hear if you'd like it cleaned up or left to be used for some tooling needs.
## Issue Addressed
Closes#1365
## Proposed Changes
Dial peers in the `cached_enrs` who aren't connected, aren't banned and satisfy the subnet predicate before making a subnet discovery query.
## Issue Addressed
- Resolves#897
- Resolves#821
## Proposed Changes
Removes references to the rust docs that we're no long maintaining.
## Additional Info
NA
## Issue Addressed
N/A
## Proposed Changes
Add a `lighthouse am wallet recover` command that recreates a wallet from a mnemonic but no validator keys. Add a `lighthouse am validator recover` command which would directly create keys from a mnemonic for a given index and count.
## Additional Info
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
Partly addresses #1547
## Proposed Changes
This fix addresses the missing attestations at slot 0 of an epoch (also sometimes slot 1 when slot 0 was skipped).
There are 2 cases:
1. BN receives the block for the attestation slot after 4 seconds (1/3rd of the slot).
2. No block is proposed for this slot.
In both cases, when we produce the attestation, we pass the head state to the
`produce_unaggregated_attestation_for_block` function here
9833eca024/beacon_node/beacon_chain/src/beacon_chain.rs (L845-L850)
Since we don't advance the state in this function, we set `attestation.data.source = state.current_justified_checkpoint` which is atleast 2 epochs lower than current_epoch(wall clock epoch).
This attestation is invalid and cannot be included in a block because of this assert from the spec:
```python
if data.target.epoch == get_current_epoch(state):
assert data.source == state.current_justified_checkpoint
state.current_epoch_attestations.append(pending_attestation)
```
https://github.com/ethereum/eth2.0-specs/blob/dev/specs/phase0/beacon-chain.md#attestations
This PR changes the `produce_unaggregated_attestation_for_block` function to ensure that it advances the state before producing the attestation at the new epoch.
Running this on my node, have missed 0 attestations across all 8 of my validators in a 100 epoch period 🎉
To compare, I was missing ~14 attestations across all 8 validators in the same 100 epoch period before the fix.
Will report missed attestations if any after running for another 100 epochs tomorrow.
## Issue Addressed
#1421
## Proposed Changes
Bounding the error_message that can be returned for RPC domain errors
Co-authored-by: Age Manning <Age@AgeManning.com>
## Proposed Changes
This is an extraction of the quoted int code from #1569, that I've come to rely on for #1544.
It allows us to parse integers from serde strings in YAML, JSON, etc. The main differences from the code in Paul's original PR are:
* Added a submodule that makes quoting mandatory (`require_quotes`).
* Decoding is generic over the type `T` being decoded. You can use `#[serde(with = "serde_utils::quoted_u64::require_quotes")]` on `Epoch` and `Slot` fields (this is what I do in my slashing protection PR).
I've turned on quoting for `Epoch` and `Slot` in this PR, but will leave the other `types` changes to you Paul.
I opted to put everything in the `conseus/serde_utils` module so that BLS can use it without a circular dependency. In future when we want to publish `types` I think we could publish `serde_utils` as `lighthouse_serde_utils` or something. Open to other ideas on this front too.
This reverts commit 2627463366.
## Issue Addressed
This is a temporary fix for #1589, by reverting #1574. The Docker image needs to be built with `--build-arg PORTABLE=true`, and we could probably integrate that into the multi-arch build, but in the interests of expediting a fix, this PR opts for a revert.
Converts the graffiti binary data to string before printing to logs.
## Issue Addressed
#1566
## Proposed Changes
Rather than converting graffiti to a vector the binary data less the last character is passed to String::from_utf_lossy(). This then allows us to call the to_string() function directly to give us the string
## Additional Info
Rust skills are fairly weak
## Issue Addressed
#1512
## Proposed Changes
Use Github Actions to automate the Docker image build, so that we can make a multi-arch image.
## Additional Info
This change will require adding the DOCKER_USERNAME and DOCKER_PASSWORD secrets in Github. It will also require disabling the Docker Hub automated build.
## Issue Addressed
#1479
## Proposed Changes
* Add an optional flag `reuse-password` in the `import` command of account_manager, allowing to use the same password for all imported keystores.
## Issue Addressed
N/A
## Proposed Changes
- Fix a wrong command in the validator generation example.
- Replace occurrences of 'passphrase' with 'password'. This is mostly because I felt that there was a lot of mixing of the two phrases in the documentation and the actual commands which is a bit confusing. Picked 'password' everywhere because it felt more appropriate but I don't mind changing it to 'passphrase' as long it's consistent everywhere.
## Issue Addressed
N/A
## Proposed Changes
Adds extended metrics to get a better idea of what is happening at the gossipsub layer of lighthouse. This provides information about mesh statistics per topics, subscriptions and peer scores.
## Additional Info
## Issue Addressed
The lighthouse user has recently changed to `lighthouse` from root.
This requires uses to change ownership of their current docker mounted volumes and the upgrade path is non-trivial.
This reverts #1502 and we will include it in a major release in the future.
## Proposed Changes
N/A
## Additional Info
N/A
…d in the Docker image
## Issue Addressed
https://github.com/sigp/lighthouse/issues/1459
## Proposed Changes
- Create new `lighthouse` user and group in Docker container
- Set user as the default user
## Issue Addressed
#1464
## Proposed Changes
Increase the slog-async log channel size from the default of 128 to 2048 to reduce the number of dropped logs.
## Additional Info
## Issue Addressed
Fixes#1509
## Proposed Changes
Exit the beacon node if the eth1 endpoint points to an invalid eth1 network. Check the network id before every eth1 cache update and display an error log if the network id has changed to an invalid one.
## Issue Addressed
#1172
## Proposed Changes
* updates the libp2p dependency
* small adaptions based on changes in libp2p
* report not just valid messages but also invalid and distinguish between `IGNORE`d messages and `REJECT`ed messages
Co-authored-by: Age Manning <Age@AgeManning.com>
## Issue Addressed
Resolve#1387
## Proposed Changes
Replace flag name **end_after_checks** to ** continue_after_checks**
Change condition to simple (remove **!**, It's no change logic.)
## Additional Info
Operation check
- [x] subcommand `eth1-sim` with ganach-cli
- [x] `./simulator eth1-sim` -> test is completes
- [x] `./simulator eth1-sim --continue_after_checks` -> test is never completes
- [x] `./simulator eth1-sim -c` -> test is never completes
- [x] `./simulator eth1-sim -c true` -> error: Found (clap)
- [x] `./simulator eth1-sim -c false` -> error: Found (clap)
- [x] subcommand `no-eth1-sim`
- [x] `./simulator no-eth1-sim` -> test is completes
- [x] `./simulator no-eth1-sim --continue_after_checks` -> test is never completes
- [x] `./simulator no-eth1-sim -c` -> test is never completes
- [x] `./simulator no-eth1-sim -c true` -> error: Found (clap)
- [x] `./simulator no-eth1-sim -c false` -> error: Found (clap)
