## Issue Addressed
NA
## Proposed Changes
Missed head votes on attestations is a well-known issue. The primary cause is a block getting set as the head *after* the attestation deadline.
This PR aims to shorten the overall time between "block received" and "block set as head" by:
1. Persisting the head and fork choice *after* setting the canonical head
- Informal measurements show this takes ~200ms
1. Pruning the op pool *after* setting the canonical head.
1. No longer persisting the op pool to disk during `BeaconChain::fork_choice`
- Informal measurements show this can take up to 1.2s.
I also add some metrics to help measure the effect of these changes.
Persistence changes like this run the risk of breaking assumptions downstream. However, I have considered these risks and I think we're fine here. I will describe my reasoning for each change.
## Reasoning
### Change 1: Persisting the head and fork choice *after* setting the canonical head
For (1), although the function is called `persist_head_and_fork_choice`, it only persists:
- Fork choice
- Head tracker
- Genesis block root
Since `BeaconChain::fork_choice_internal` does not modify these values between the original time we were persisting it and the current time, I assert that the change I've made is non-substantial in terms of what ends up on-disk. There's the possibility that some *other* thread has modified fork choice in the extra time we've given it, but that's totally fine.
Since the only time we *read* those values from disk is during startup, I assert that this has no impact during runtime.
### Change 2: Pruning the op pool after setting the canonical head
Similar to the argument above, we don't modify the op pool during `BeaconChain::fork_choice_internal` so it shouldn't matter when we prune. This change should be non-substantial.
### Change 3: No longer persisting the op pool to disk during `BeaconChain::fork_choice`
This change *is* substantial. With the proposed changes, we'll only be persisting the op pool to disk when we shut down cleanly (i.e., the `BeaconChain` gets dropped). This means we'll save disk IO and time during usual operation, but a `kill -9` or similar "crash" will probably result in an out-of-date op pool when we reboot. An out-of-date op pool can only have an impact when producing blocks or aggregate attestations/sync committees.
I think it's pretty reasonable that a crash might result in an out-of-date op pool, since:
- Crashes are fairly rare. Practically the only time I see LH suffer a full crash is when the OOM killer shows up, and that's a very serious event.
- It's generally quite rare to produce a block/aggregate immediately after a reboot. Just a few slots of runtime is probably enough to have a decent-enough op pool again.
## Additional Info
Credits to @macladson for the timings referenced here.
## Issue Addressed
Resolves#2033
## Proposed Changes
Adds a flag to enable shutting down beacon node right after sync is completed.
## Additional Info
Will need modification after weak subjectivity sync is enabled to change definition of a fully synced node.
## Issue Addressed
Closes#2526
## Proposed Changes
If the head block fails to decode on start up, do two things:
1. Revert all blocks between the head and the most recent hard fork (to `fork_slot - 1`).
2. Reset fork choice so that it contains the new head, and all blocks back to the new head's finalized checkpoint.
## Additional Info
I tweaked some of the beacon chain test harness stuff in order to make it generic enough to test with a non-zero slot clock on start-up. In the process I consolidated all the various `new_` methods into a single generic one which will hopefully serve all future uses 🤞
## Issue Addressed
Resolves#2114
Swapped out the ctrlc crate for tokio signals to hook register handlers for SIGPIPE and SIGHUP along with SIGTERM and SIGINT.
## Proposed Changes
- Swap out the ctrlc crate for tokio signals for unix signal handing
- Register signals for SIGPIPE and SHIGUP that trigger the same shutdown procedure as SIGTERM and SIGINT
## Additional Info
I tested these changes against the examples in the original issue and noticed some interesting behavior on my machine. When running `lighthouse bn --network pyrmont |& tee -a pyrmont_bn.log` or `lighthouse bn --network pyrmont 2>&1 | tee -a pyrmont_bn.log` none of the above signals are sent to the lighthouse program in a way I was able to observe.
The only time it seems that the signal gets sent to the lighthouse program is if there is no redirection of stderr to stdout. I'm not as familiar with the details of how unix signals work in linux with a redirect like that so I'm not sure if this is a bug in the program or expected behavior.
Signals are correctly received without the redirection and if the above signals are sent directly to the program with something like `kill`.
## Issue Addressed
Resolves#2487
## Proposed Changes
Logs a message once in every invocation of `Eth1Service::update` method if the primary endpoint is unavailable for some reason.
e.g.
```log
Aug 03 00:09:53.517 WARN Error connecting to eth1 node endpoint action: trying fallbacks, endpoint: http://localhost:8545/, service: eth1_rpc
Aug 03 00:09:56.959 INFO Fetched data from fallback fallback_number: 1, service: eth1_rpc
```
The main aim of this PR is to have an accompanying message to the "action: trying fallbacks" error message that is returned when checking the endpoint for liveness. This is mainly to indicate to the user that the fallback was live and reachable.
## Additional info
This PR is not meant to be a catch all for all cases where the primary endpoint failed. For instance, this won't log anything if the primary node was working fine during endpoint liveness checking and failed during deposit/block fetching. This is done intentionally to reduce number of logs while initial deposit/block sync and to avoid more complicated logic.
## Issue Addressed
NA
## Proposed Changes
This PR expands the time that entries exist in the gossip-sub duplicate cache. Recent investigations found that this cache is one slot (12s) shorter than the period for which an attestation is permitted to propagate on the gossip network.
Before #2540, this was causing peers to be unnecessarily down-scored for sending old attestations. Although that issue has been fixed, the duplicate cache time is increased here to avoid such messages from getting any further up the networking stack then required.
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
A Discord user presented logs which indicated a drop in their peer count caused by a variety of peers sending attestations where we'd already seen an attestation for that validator. It's presently unclear how this case came about, but during our investigation I noticed that we are down-voting peers for sending such attestations.
There are three scenarios where we may receive duplicate unagg. attestations from the same validator:
1. The validator is committing a slashable offense.
2. The gossipsub message-deduping functionality is not working as expected.
3. We received the message via the HTTP prior to seeing it via gossip.
Scenario (1) would be so costly for an attacker that I don't think we need to add DoS protection for it.
Scenario (2) seems feasible. Our "seen message" caches in gossipsub might fill up/expire and let through these duplicates. There are also cases involving message ID mismatches with the other peers. In both these cases, I don't think we should be doing 1 attestation == -1 point down-voting.
Scenario (3) is not necessarily a fault of the peer and we shouldn't down-score them for it.
## Additional Info
NA
Due to the altair fork, in principle we can now subscribe to up to 148 topics. This bypasses our original limit and we can end up rejecting subscriptions.
This PR increases the limit to account for the fork.
## Issue Addressed
N/A
## Proposed Changes
Adds a cli option to disable packet filter in `lighthouse bootnode`. This is useful in running local testnets as the bootnode bans requests from the same ip(localhost) if the packet filter is enabled.
## Issue Addressed
Resolves#2524
## Proposed Changes
- Return all known forks in the `/config/fork_schedule`, previously returned only the head of the chain's fork.
- Deleted the `StateId::head` method because it was only previously used in this endpoint.
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
N/A
## Proposed Changes
Refactor discovery queries such that only `QueryType::Subnet` queries are queued for grouping. `QueryType::FindPeers` is always made regardless of the number of active `Subnet` queries (max = 2). This prevents `FindPeers` queries from getting starved if `Subnet` queries start queuing up.
Also removes `GroupedQueryType` struct and use `QueryType` for all queuing and processing of discovery requests.
## Additional Info
Currently, no distinction is made between subnet discovery queries for attestation and sync committee subnets when grouping queries. Potentially we could prioritise attestation queries over sync committee queries in the future.
## Issue Addressed
NA
## Proposed Changes
- Version bump
- Increase queue sizes for aggregated attestations and re-queued attestations.
## Additional Info
NA
## Issue Addressed
Which issue # does this PR address?
## Proposed Changes
- Add a counter metric to log when a block is received late from gossip.
- Also push a `DEBG` log for the above condition.
- Use Debug (`?`) instead of Display (`%`) for a bunch of logs in the beacon processor, so we don't have to deal with concatenated block roots.
- Add new ERRO and CRIT to HTTP API to alert users when they're publishing late blocks.
## Additional Info
NA
## Proposed Changes
* Consolidate Tokio versions: everything now uses the latest v1.10.0, no more `tokio-compat`.
* Many semver-compatible changes via `cargo update`. Notably this upgrades from the yanked v0.8.0 version of crossbeam-deque which is present in v1.5.0-rc.0
* Many semver incompatible upgrades via `cargo upgrades` and `cargo upgrade --workspace pkg_name`. Notable ommissions:
- Prometheus, to be handled separately: https://github.com/sigp/lighthouse/issues/2485
- `rand`, `rand_xorshift`: the libsecp256k1 package requires 0.7.x, so we'll stick with that for now
- `ethereum-types` is pinned at 0.11.0 because that's what `web3` is using and it seems nice to have just a single version
## Additional Info
We still have two versions of `libp2p-core` due to `discv5` depending on the v0.29.0 release rather than `master`. AFAIK it should be OK to release in this state (cc @AgeManning )
## Issue Addressed
NA
## Proposed Changes
- Bump to `v1.5.0-rc.0`.
- Increase attestation reprocessing queue size (I saw this filling up on Prater).
- Reduce error log for full attn reprocessing queue to warn.
## TODO
- [x] Manual testing
- [x] Resolve https://github.com/sigp/lighthouse/pull/2493
- [x] Include https://github.com/sigp/lighthouse/pull/2501
## Issue Addressed
- Resolves#2457
- Resolves#2443
## Proposed Changes
Target the (presently unreleased) head of `libp2p/rust-libp2p:master` in order to obtain the fix from https://github.com/libp2p/rust-libp2p/pull/2175.
Additionally:
- `libsecp256k1` needed to be upgraded to satisfy the new version of `libp2p`.
- There were also a handful of minor changes to `eth2_libp2p` to suit some interface changes.
- Two `cargo audit --ignore` flags were remove due to libp2p upgrades.
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
When testing our (not-yet-released) Doppelganger implementation, I noticed that we aren't detecting attestations included in blocks (only those on the gossip network).
This is because during [block processing](e8c0d1f19b/beacon_node/beacon_chain/src/beacon_chain.rs (L2168)) we only update the `observed_attestations` cache with each attestation, but not the `observed_attesters` cache. This is the correct behaviour when we consider the [p2p spec](https://github.com/ethereum/eth2.0-specs/blob/v1.0.1/specs/phase0/p2p-interface.md):
> [IGNORE] There has been no other valid attestation seen on an attestation subnet that has an identical attestation.data.target.epoch and participating validator index.
We're doing the right thing here and still allowing attestations on gossip that we've seen in a block. However, this doesn't work so nicely for Doppelganger.
To resolve this, I've taken the following steps:
- Add a `observed_block_attesters` cache.
- Rename `observed_attesters` to `observed_gossip_attesters`.
## TODO
- [x] Add a test to ensure a validator that's been seen in a block attestation (but not a gossip attestation) returns `true` for `BeaconChain::validator_seen_at_epoch`.
- [x] Add a test to ensure `observed_block_attesters` isn't polluted via gossip attestations and vice versa.
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Proposed Changes
* Implement the validator client and HTTP API changes necessary to support Altair
Co-authored-by: realbigsean <seananderson33@gmail.com>
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
## Issue Addressed
Resolves#2069
## Proposed Changes
- Adds a `--doppelganger-detection` flag
- Adds a `lighthouse/seen_validators` endpoint, which will make it so the lighthouse VC is not interopable with other client beacon nodes if the `--doppelganger-detection` flag is used, but hopefully this will become standardized. Relevant Eth2 API repo issue: https://github.com/ethereum/eth2.0-APIs/issues/64
- If the `--doppelganger-detection` flag is used, the VC will wait until the beacon node is synced, and then wait an additional 2 epochs. The reason for this is to make sure the beacon node is able to subscribe to the subnets our validators should be attesting on. I think an alternative would be to have the beacon node subscribe to all subnets for 2+ epochs on startup by default.
## Additional Info
I'd like to add tests and would appreciate feedback.
TODO: handle validators started via the API, potentially make this default behavior
Co-authored-by: realbigsean <seananderson33@gmail.com>
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Issue Addressed
N/A
## Proposed Changes
- Removing a bunch of unnecessary references
- Updated `Error::VariantError` to `Error::Variant`
- There were additional enum variant lints that I ignored, because I thought our variant names were fine
- removed `MonitoredValidator`'s `pubkey` field, because I couldn't find it used anywhere. It looks like we just use the string version of the pubkey (the `id` field) if there is no index
## Additional Info
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Issue Addressed
- Resolves#2169
## Proposed Changes
Adds the `AttesterCache` to allow validators to produce attestations for older slots. Presently, some arbitrary restrictions can force validators to receive an error when attesting to a slot earlier than the present one. This can cause attestation misses when there is excessive load on the validator client or time sync issues between the VC and BN.
## Additional Info
NA
## Proposed Changes
Add the `sync_aggregate` from `BeaconBlock` to the bulk signature verifier for blocks. This necessitates a new signature set constructor for the sync aggregate, which is different from the others due to the use of [`eth2_fast_aggregate_verify`](https://github.com/ethereum/eth2.0-specs/blob/v1.1.0-alpha.7/specs/altair/bls.md#eth2_fast_aggregate_verify) for sync aggregates, per [`process_sync_aggregate`](https://github.com/ethereum/eth2.0-specs/blob/v1.1.0-alpha.7/specs/altair/beacon-chain.md#sync-aggregate-processing). I made the choice to return an optional signature set, with `None` representing the case where the signature is valid on account of being the point at infinity (requires no further checking).
To "dogfood" the changes and prevent duplication, the consensus logic now uses the signature set approach as well whenever it is required to verify signatures (which should only be in testing AFAIK). The EF tests pass with the code as it exists currently, but failed before I adapted the `eth2_fast_aggregate_verify` changes (which is good).
As a result of this change Altair block processing should be a little faster, and importantly, we will no longer accidentally verify signatures when replaying blocks, e.g. when replaying blocks from the database.
## Issue Addressed
NA
## Proposed Changes
This PR addresses two things:
1. Allows the `ValidatorMonitor` to work with Altair states.
1. Optimizes `altair::process_epoch` (see [code](https://github.com/paulhauner/lighthouse/blob/participation-cache/consensus/state_processing/src/per_epoch_processing/altair/participation_cache.rs) for description)
## Breaking Changes
The breaking changes in this PR revolve around one premise:
*After the Altair fork, it's not longer possible (given only a `BeaconState`) to identify if a validator had *any* attestation included during some epoch. The best we can do is see if that validator made the "timely" source/target/head flags.*
Whilst this seems annoying, it's not actually too bad. Finalization is based upon "timely target" attestations, so that's really the most important thing. Although there's *some* value in knowing if a validator had *any* attestation included, it's far more important to know about "timely target" participation, since this is what affects finality and justification.
For simplicity and consistency, I've also removed the ability to determine if *any* attestation was included from metrics and API endpoints. Now, all Altair and non-Altair states will simply report on the head/target attestations.
The following section details where we've removed fields and provides replacement values.
### Breaking Changes: Prometheus Metrics
Some participation metrics have been removed and replaced. Some were removed since they are no longer relevant to Altair (e.g., total attesting balance) and others replaced with gwei values instead of pre-computed values. This provides more flexibility at display-time (e.g., Grafana).
The following metrics were added as replacements:
- `beacon_participation_prev_epoch_head_attesting_gwei_total`
- `beacon_participation_prev_epoch_target_attesting_gwei_total`
- `beacon_participation_prev_epoch_source_attesting_gwei_total`
- `beacon_participation_prev_epoch_active_gwei_total`
The following metrics were removed:
- `beacon_participation_prev_epoch_attester`
- instead use `beacon_participation_prev_epoch_source_attesting_gwei_total / beacon_participation_prev_epoch_active_gwei_total`.
- `beacon_participation_prev_epoch_target_attester`
- instead use `beacon_participation_prev_epoch_target_attesting_gwei_total / beacon_participation_prev_epoch_active_gwei_total`.
- `beacon_participation_prev_epoch_head_attester`
- instead use `beacon_participation_prev_epoch_head_attesting_gwei_total / beacon_participation_prev_epoch_active_gwei_total`.
The `beacon_participation_prev_epoch_attester` endpoint has been removed. Users should instead use the pre-existing `beacon_participation_prev_epoch_target_attester`.
### Breaking Changes: HTTP API
The `/lighthouse/validator_inclusion/{epoch}/{validator_id}` endpoint loses the following fields:
- `current_epoch_attesting_gwei` (use `current_epoch_target_attesting_gwei` instead)
- `previous_epoch_attesting_gwei` (use `previous_epoch_target_attesting_gwei` instead)
The `/lighthouse/validator_inclusion/{epoch}/{validator_id}` endpoint lose the following fields:
- `is_current_epoch_attester` (use `is_current_epoch_target_attester` instead)
- `is_previous_epoch_attester` (use `is_previous_epoch_target_attester` instead)
- `is_active_in_current_epoch` becomes `is_active_unslashed_in_current_epoch`.
- `is_active_in_previous_epoch` becomes `is_active_unslashed_in_previous_epoch`.
## Additional Info
NA
## TODO
- [x] Deal with total balances
- [x] Update validator_inclusion API
- [ ] Ensure `beacon_participation_prev_epoch_target_attester` and `beacon_participation_prev_epoch_head_attester` work before Altair
Co-authored-by: realbigsean <seananderson33@gmail.com>
## Proposed Changes
Update to the latest version of the Altair spec, which includes new tests and a tweak to the target sync aggregators.
## Additional Info
This change is _not_ required for the imminent Altair devnet, and is waiting on the merge of #2321 to unstable.
Co-authored-by: Paul Hauner <paul@paulhauner.com>
## Proposed Changes
Remove the remaining Altair `FIXME`s from consensus land.
1. Implement tree hash caching for the participation lists. This required some light type manipulation, including removing the `TreeHash` bound from `CachedTreeHash` which was purely descriptive.
2. Plumb the proposer index through Altair attestation processing, to avoid calculating it for _every_ attestation (potentially 128ms on large networks). This duplicates some work from #2431, but with the aim of getting it in sooner, particularly for the Altair devnets.
3. Removes two FIXMEs related to `superstruct` and cloning, which are unlikely to be particularly detrimental and will be tracked here instead: https://github.com/sigp/superstruct/issues/5
* Adjust beacon node timeouts for validator client HTTP requests (#2352)
Resolves#2313
Provide `BeaconNodeHttpClient` with a dedicated `Timeouts` struct.
This will allow granular adjustment of the timeout duration for different calls made from the VC to the BN. These can either be a constant value, or as a ratio of the slot duration.
Improve timeout performance by using these adjusted timeout duration's only whenever a fallback endpoint is available.
Add a CLI flag called `use-long-timeouts` to revert to the old behavior.
Additionally set the default `BeaconNodeHttpClient` timeouts to the be the slot duration of the network, rather than a constant 12 seconds. This will allow it to adjust to different network specifications.
Co-authored-by: Paul Hauner <paul@paulhauner.com>
* Use read_recursive locks in database (#2417)
Closes#2245
Replace all calls to `RwLock::read` in the `store` crate with `RwLock::read_recursive`.
* Unfortunately we can't run the deadlock detector on CI because it's pinned to an old Rust 1.51.0 nightly which cannot compile Lighthouse (one of our deps uses `ptr::addr_of!` which is too new). A fun side-project at some point might be to update the deadlock detector.
* The reason I think we haven't seen this deadlock (at all?) in practice is that _writes_ to the database's split point are quite infrequent, and a concurrent write is required to trigger the deadlock. The split point is only written when finalization advances, which is once per epoch (every ~6 minutes), and state reads are also quite sporadic. Perhaps we've just been incredibly lucky, or there's something about the timing of state reads vs database migration that protects us.
* I wrote a few small programs to demo the deadlock, and the effectiveness of the `read_recursive` fix: https://github.com/michaelsproul/relock_deadlock_mvp
* [The docs for `read_recursive`](https://docs.rs/lock_api/0.4.2/lock_api/struct.RwLock.html#method.read_recursive) warn of starvation for writers. I think in order for starvation to occur the database would have to be spammed with so many state reads that it's unable to ever clear them all and find time for a write, in which case migration of states to the freezer would cease. If an attack could be performed to trigger this starvation then it would likely trigger a deadlock in the current code, and I think ceasing migration is preferable to deadlocking in this extreme situation. In practice neither should occur due to protection from spammy peers at the network layer. Nevertheless, it would be prudent to run this change on the testnet nodes to check that it doesn't cause accidental starvation.
* Return more detail when invalid data is found in the DB during startup (#2445)
- Resolves#2444
Adds some more detail to the error message returned when the `BeaconChainBuilder` is unable to access or decode block/state objects during startup.
NA
* Use hardware acceleration for SHA256 (#2426)
Modify the SHA256 implementation in `eth2_hashing` so that it switches between `ring` and `sha2` to take advantage of [x86_64 SHA extensions](https://en.wikipedia.org/wiki/Intel_SHA_extensions). The extensions are available on modern Intel and AMD CPUs, and seem to provide a considerable speed-up: on my Ryzen 5950X it dropped state tree hashing times by about 30% from 35ms to 25ms (on Prater).
The extensions became available in the `sha2` crate [last year](https://www.reddit.com/r/rust/comments/hf2vcx/ann_rustcryptos_sha1_and_sha2_now_support/), and are not available in Ring, which uses a [pure Rust implementation of sha2](https://github.com/briansmith/ring/blob/main/src/digest/sha2.rs). Ring is faster on CPUs that lack the extensions so I've implemented a runtime switch to use `sha2` only when the extensions are available. The runtime switching seems to impose a miniscule penalty (see the benchmarks linked below).
* Start a release checklist (#2270)
NA
Add a checklist to the release draft created by CI. I know @michaelsproul was also working on this and I suspect @realbigsean also might have useful input.
NA
* Serious banning
* fmt
Co-authored-by: Mac L <mjladson@pm.me>
Co-authored-by: Paul Hauner <paul@paulhauner.com>
Co-authored-by: Michael Sproul <michael@sigmaprime.io>
## Issue Addressed
- Resolves#2452
## Proposed Changes
I've seen a few people confused by this and I don't think the message is really worth it.
## Additional Info
NA
## Issue Addressed
#635
## Proposed Changes
- Keep attestations that reference a block we have not seen for 30secs before being re processed
- If we do import the block before that time elapses, it is reprocessed in that moment
- The first time it fails, do nothing wrt to gossipsub propagation or peer downscoring. If after being re processed it fails, downscore with a `LowToleranceError` and ignore the message.
## Issue Addressed
NA
## Proposed Changes
Adds a metric to see how many set bits are in the sync aggregate for each beacon block being imported.
## Additional Info
NA
## Issue Addressed
NA
## Proposed Changes
Adds more detail to the log when an attestation is ignored due to a prior one being known. This will help identify which validators are causing the issue.
## Additional Info
NA
## Issue Addressed
- Resolves#2444
## Proposed Changes
Adds some more detail to the error message returned when the `BeaconChainBuilder` is unable to access or decode block/state objects during startup.
## Additional Info
NA
## Issue Addressed
Closes#2245
## Proposed Changes
Replace all calls to `RwLock::read` in the `store` crate with `RwLock::read_recursive`.
## Additional Info
* Unfortunately we can't run the deadlock detector on CI because it's pinned to an old Rust 1.51.0 nightly which cannot compile Lighthouse (one of our deps uses `ptr::addr_of!` which is too new). A fun side-project at some point might be to update the deadlock detector.
* The reason I think we haven't seen this deadlock (at all?) in practice is that _writes_ to the database's split point are quite infrequent, and a concurrent write is required to trigger the deadlock. The split point is only written when finalization advances, which is once per epoch (every ~6 minutes), and state reads are also quite sporadic. Perhaps we've just been incredibly lucky, or there's something about the timing of state reads vs database migration that protects us.
* I wrote a few small programs to demo the deadlock, and the effectiveness of the `read_recursive` fix: https://github.com/michaelsproul/relock_deadlock_mvp
* [The docs for `read_recursive`](https://docs.rs/lock_api/0.4.2/lock_api/struct.RwLock.html#method.read_recursive) warn of starvation for writers. I think in order for starvation to occur the database would have to be spammed with so many state reads that it's unable to ever clear them all and find time for a write, in which case migration of states to the freezer would cease. If an attack could be performed to trigger this starvation then it would likely trigger a deadlock in the current code, and I think ceasing migration is preferable to deadlocking in this extreme situation. In practice neither should occur due to protection from spammy peers at the network layer. Nevertheless, it would be prudent to run this change on the testnet nodes to check that it doesn't cause accidental starvation.
