Eth logs listener with a topic filter causes a timeout

[rvagg]

Reported by @Schwartz10, error logs provided by Protofire.

If you try to use the event “topics” in event log listeners to filter event logs by a specific criteria, it times out

Comes from:

using abigen to create go web3 bindings for a specific contract

{"level":"warn","ts":"2025-01-09T16:38:50.064Z","logger":"rpc","caller":"[email protected]/handler.go:421","msg":"error in RPC call to 'eth_getLogs': failed to get events for filter:\n    github.com/filecoin-project/lotus/node/impl/full.(*EthEventHandler).EthGetLogs\n        /go/lotus/node/impl/full/eth.go:1743\n  - failed to get events for filter from chain indexer:\n    github.com/filecoin-project/lotus/node/impl/full.(*EthEventHandler).ethGetEventsForFilter\n        /go/lotus/node/impl/full/eth.go:1789\n  - failed to wait for head to be indexed:\n    github.com/filecoin-project/lotus/chain/index.(*SqliteIndexer).GetEventsForFilter\n        /go/lotus/chain/index/events.go:467\n  - context canceled"}

So this is a plain eth_getLogs, from here, which shouldn't be erroring out on waiting for future stuff. We don't have the specific filter, perhaps it's "latest"? It would be good to figure out what the max it's aiming for. But also perhaps max doesn't matter here and just by supplying a topic that it hits this error? Either way, it shouldn't wait to find this, it should be a very quick wait to allow for the async Apply->index process happens and nothing more.

[BigLep]

2025-01-29 conversation: this is the ChainIndexer waiting for the latest epoch to get indexed. The ChainIndexer reconciliation logic should be returning quicker.

We don't know how many are impacted this.

We don't have a repro for this either. @Schwartz10: by chance, do you have additional details here to help with reproduce it?

We certainly welcome contribution on this fix if anyone wants to take it on.