fix(mpool): MempoolFilter returns the pending transactions#7250
Open
akaladarshi wants to merge 4 commits into
Open
fix(mpool): MempoolFilter returns the pending transactions#7250akaladarshi wants to merge 4 commits into
MempoolFilter returns the pending transactions#7250akaladarshi wants to merge 4 commits into
Conversation
akaladarshi
requested review from
hanabi1224 and
sudo-shashank
and removed request for
a team
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (5)
🔗 Linked repositories identifiedCodeRabbit considers these linked repositories for cross-repo context during reviews:
🚧 Files skipped from review as they are similar to previous changes (3)
WalkthroughPending-transaction filtering now uses a subscriber-backed mempool stream, buffers hashes per filter, and drains them through the Ethereum RPC path. Handler construction passes chain id and mempool subscriber through daemon and offline-server wiring. Pubsub and stateful tests are updated to match the new hash flow. ChangesPending transaction hash flow
Sequence Diagram(s)sequenceDiagram
participant RPCClient
participant MempoolFilterManager
participant pending_tx_added_hashes
participant MpoolSubscriber
participant MempoolFilter
RPCClient->>MempoolFilterManager: install()
MempoolFilterManager->>pending_tx_added_hashes: start fan-out task(eth_chain_id, subscriber)
pending_tx_added_hashes->>MpoolSubscriber: subscribe()
MpoolSubscriber-->>pending_tx_added_hashes: MpoolUpdate::Add
pending_tx_added_hashes-->>MempoolFilterManager: EthHash
MempoolFilterManager->>MempoolFilter: push(hash)
RPCClient->>MempoolFilterManager: remove()/eth_getFilterChanges
MempoolFilterManager->>MempoolFilter: drain()
MempoolFilter-->>RPCClient: Vec<EthHash>
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes Possibly related PRs
Suggested reviewers
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
✨ Simplify code
Comment |
6 tasks
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@src/rpc/methods/eth/filter/mempool.rs`:
- Around line 256-258: The async test in the mempool filter flow should not rely
on a single tokio::task::yield_now() to let the fan-out task deliver the
broadcast event. Update the test around the filter subscription/drain logic in
the mempool.rs path to poll with a small timeout until both filters have
received the hash, using the existing filter/broadcast setup instead of assuming
one scheduler yield is sufficient.
- Around line 119-128: The fan-out task startup in ensure_fanout_task currently
calls tokio::spawn unconditionally, which can panic if there is no active Tokio
runtime. Update the install path that reaches ensure_fanout_task to check for a
runtime first and return an installation error instead of spawning blindly, and
make sure the caller of ensure_fanout_task propagates that Result so the
existing error handling is preserved.
In `@src/rpc/methods/eth/filter/mod.rs`:
- Around line 142-146: `EthEventHandler::new()` is wiring pending-filter RPCs to
`MpoolSubscriber::dummy()`, so `eth_newPendingTransactionFilter` never receives
mempool adds in states built from it. Update the RPC state constructors that use
pending filters, especially the snapshot/test paths in `generate_test_snapshot`
and `test_snapshot`, to pass a real `MessagePool` subscriber
(`mpool.subscriber()`) into `EthEventHandler::from_config(...)` instead of
relying on the dummy subscriber. Make sure the `EthEventHandler` and related
`RPCState` setup still use the same `EventsConfig` and chain ID, but are
connected to the live mempool bus.
In `@src/tool/subcommands/api_cmd/stateful_tests.rs`:
- Around line 1041-1044: The pending-tx assertions are still happening after
chain inclusion because wait_pending_message() continues into StateWaitMsg, so
the tests do not actually verify pre-mining mempool behavior. Update the
stateful tests to inspect the filter immediately after the first MpoolPending
observation, or split wait_pending_message() into a helper that returns once the
CID appears so the test can stop before StateWaitMsg. Apply the same fix in both
affected test sections that use wait_pending_message() and
poll_pending_filter_until().
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: b1fab06a-d564-4511-ba0e-b6e9944153cd
📒 Files selected for processing (14)
CHANGELOG.mdsrc/daemon/mod.rssrc/message_pool/msgpool/events.rssrc/message_pool/msgpool/mod.rssrc/message_pool/msgpool/msg_pool.rssrc/message_pool/msgpool/pending_store.rssrc/message_pool/msgpool/selection.rssrc/rpc/methods/eth.rssrc/rpc/methods/eth/filter/mempool.rssrc/rpc/methods/eth/filter/mod.rssrc/rpc/methods/eth/pubsub.rssrc/tool/offline_server/server.rssrc/tool/subcommands/api_cmd/stateful_tests.rssrc/utils/task/mod.rs
🔗 Linked repositories identified
CodeRabbit considers these linked repositories for cross-repo context during reviews:
filecoin-project/lotus(manual)
Comment on lines
+119
to
+128
| fn ensure_fanout_task(&self) { | ||
| self.fanout_task.get_or_init(|| { | ||
| let filters = self.filters.shallow_clone(); | ||
| let mut hashes = | ||
| pending_tx_added_hashes(self.subscriber.subscribe(), self.eth_chain_id); | ||
| let task = tokio::spawn(async move { | ||
| while let Some(hash) = hashes.next().await { | ||
| fan_out(&filters.read(), hash); | ||
| } | ||
| }); |
Contributor
There was a problem hiding this comment.
🩺 Stability & Availability | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo '--- mempool.rs outline ---'
ast-grep outline src/rpc/methods/eth/filter/mempool.rs --view expanded || true
echo '--- ensure_fanout_task occurrences ---'
rg -n "ensure_fanout_task|fn install|eth_new_pending_transaction_filter|tokio::spawn" src/rpc/methods/eth/filter -S || trueRepository: ChainSafe/forest
Length of output: 3074
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo '--- mempool.rs relevant lines ---'
sed -n '110,150p' src/rpc/methods/eth/filter/mempool.rs | cat -n
echo '--- mod.rs relevant lines ---'
sed -n '210,340p' src/rpc/methods/eth/filter/mod.rs | cat -n
echo '--- tests around pending filter ---'
sed -n '1220,1285p' src/rpc/methods/eth/filter/mod.rs | cat -nRepository: ChainSafe/forest
Length of output: 10232
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo '--- install_filter call sites ---'
rg -n "install_filter\(" src/rpc/methods/eth -S || true
echo '--- eth_new_pending_transaction_filter implementation ---'
sed -n '230,275p' src/rpc/methods/eth/filter/mod.rs | cat -n
echo '--- FilterManager trait and implementations ---'
sed -n '70,110p' src/rpc/methods/eth/filter/mod.rs | cat -nRepository: ChainSafe/forest
Length of output: 5330
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo '--- install_filter call sites ---'
rg -n "install_filter\(" src/rpc/methods/eth -S || true
echo '--- mod.rs around install_filter and eth_new_pending_transaction_filter ---'
sed -n '220,270p' src/rpc/methods/eth/filter/mod.rs | cat -nRepository: ChainSafe/forest
Length of output: 3372
Guard the fan-out task spawn with a runtime check
ensure_fanout_task calls tokio::spawn from a synchronous install path, so a caller without an active Tokio runtime will panic instead of getting the existing Result. Propagate an installation error here instead.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/rpc/methods/eth/filter/mempool.rs` around lines 119 - 128, The fan-out
task startup in ensure_fanout_task currently calls tokio::spawn unconditionally,
which can panic if there is no active Tokio runtime. Update the install path
that reaches ensure_fanout_task to check for a runtime first and return an
installation error instead of spawning blindly, and make sure the caller of
ensure_fanout_task propagates that Result so the existing error handling is
preserved.
Comment on lines
+256
to
+258
| // The receiver is subscribed during install, so the event is already | ||
| // buffered; one yield lets the fan-out task drain it into both filters. | ||
| tokio::task::yield_now().await; |
Contributor
There was a problem hiding this comment.
🩺 Stability & Availability | 🟡 Minor | ⚡ Quick win
Avoid relying on a single scheduler yield in this async test.
One yield_now() does not guarantee the fan-out task has drained and pushed the broadcast event, so this can be flaky under load. Poll with a small timeout until both filters receive the hash.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@src/rpc/methods/eth/filter/mempool.rs` around lines 256 - 258, The async test
in the mempool filter flow should not rely on a single tokio::task::yield_now()
to let the fan-out task deliver the broadcast event. Update the test around the
filter subscription/drain logic in the mempool.rs path to poll with a small
timeout until both filters have received the hash, using the existing
filter/broadcast setup instead of assuming one scheduler yield is sufficient.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary of changes
Changes introduced in this pull request:
Reference issue to close (if applicable)
Closes
Other information and links
Change checklist
Outside contributions
Summary by CodeRabbit
New Features
Bug Fixes
eth_newPendingTransactionFilternow returns hashes for truly pending mempool transactions (not on-chain execution events).eth_getFilterChangesnow returns only newly observed pending transaction hashes per poll, avoiding duplicates across polls.