deadlock issue of v1.15.6 on eth_getLogs for filter & thus shutdown corruption

[jun0tpyrc]

System information

{"jsonrpc":"2.0","id":1,"result":"Geth/v1.15.9-stable-74165a8f/linux-amd64/go1.24.2"}

Expected behaviour

can answer in reasonable time

Actual behaviour

slow ,timeout for 1minute, the same call can finish on geth 1.15.3 on some api provider / erigon3.0.2 returning immediately

Steps to reproduce the behaviour

 curl 0:8545 -X POST   -H "Content-Type: application/json"   --data '{"method":"eth_getLogs","params":[{"fromBlock":"0x154ce96","toBlock":"0x154cefa","address":"0x918778e825747a892b17C66fe7D24C618262867d"}],"id":1,"jsonrpc":"2.0"}'

Backtrace

[backtrace]

When submitting logs: please submit them as text and not screenshots.

[jun0tpyrc]

other calls are fine without slowdown

[rjl493456442]

The range of block is

>>> 0x154ce96
22335126
>>> 0x154cefa
22335226

[wdsse]

+1

[22november1963]

+1

[jun0tpyrc]

think this one might be important
saw this glitch causing different l2 setups run on the l1 geth be stuck too
reproduced on 4 geth nodes of this version

[murat-deniz1]

+1

is this issue releated to #31589 ?

[fjl]

Is the timeout a repeatable issue, i.e. does it always occur with the same query?

[jun0tpyrc]

Is the timeout a repeatable issue, i.e. does it always occur with the same query?

definitely reproducible , i can see it on 4 different nodes with different filters (even change to single block range ,or a lower traffic address) , it stucks easily

feel like some deadlocks thing

May be not first appearing on 1.15.9 (our upgrade path was alltools-v1.14.13 -> alltools-v1.15.9 docker)

Sounds similar to #31589 mentioned

[rjl493456442]

@jun0tpyrc
Can you please share the query you used?

 curl 0:8545 -X POST   -H "Content-Type: application/json"   --data '{"method":"eth_getLogs","params":[{"fromBlock":"0x154ce96","toBlock":"0x154cefa","address":"0x918778e825747a892b17C66fe7D24C618262867d"}],"id":1,"jsonrpc":"2.0"}'

We can get the result instantly for this query:

> eth.getLogs({"fromBlock":"0x154ce96","toBlock":"0x154cefa","address":"0x918778e825747a892b17C66fe7D24C618262867d"})
[]

[jun0tpyrc]

@jun0tpyrc Can you please share the query you used?

 curl 0:8545 -X POST   -H "Content-Type: application/json"   --data '{"method":"eth_getLogs","params":[{"fromBlock":"0x154ce96","toBlock":"0x154cefa","address":"0x918778e825747a892b17C66fe7D24C618262867d"}],"id":1,"jsonrpc":"2.0"}'

We can get the result instantly for this query:

> eth.getLogs({"fromBlock":"0x154ce96","toBlock":"0x154cefa","address":"0x918778e825747a892b17C66fe7D24C618262867d"})
[]

it just being same as your above one , on different geth api provider (checked by rpc method) or erigon we could also get the [] (empty result ) instantly but not for the geth nodes we have now

Furthermore I just found shutdown peacefully becomes a problem on this version - took like 20min and stuck without the final peaceful Blockchain stopped message

INFO [04-24|12:36:38.215] Looking for peers                        peercount=1   tried=42  static=1
WARN [04-24|12:36:38.389] Already shutting down, interrupt more to panic. times=9
..
WARN [04-24|12:52:00.769] Beacon client online, but no consensus updates received in a while. Please fix your beacon client to follow the chain!
INFO [04-24|12:52:07.761] Looking for peers                        peercount=0   tried=64  static=1
INFO [04-24|12:52:17.835] Looking for peers                        peercount=0   tried=58  static=1
INFO [04-24|12:52:27.943] Looking for peers                        peercount=0   tried=72  static=1
INFO [04-24|12:52:37.966] Looking for peers                        peercount=1   tried=56  static=1
INFO [04-24|12:52:48.012] Looking for peers                        peercount=0   tried=45  static=1

main running flag

geth --networkid=1 --config=/var/lib/geth/geth.toml --http.port=8545 --cache=102400 --http --http.addr=0.0.0.0 --http.corsdomain=* --http.vhosts=* --gcmode="full" --snapshot=false --txlookuplimit=0 --cache.preimages --authrpc.jwtsecret="/opt/jwtsecret" --authrpc.vhosts="*" --authrpc.port="8551" --authrpc.addr="localhost"

[jun0tpyrc]

1.15.9 cannot shut down normally (after 25min forcefully stop)
1.15.9 -> 1.14.13 (cannot start, won't work for db incompatibility as mentioned)
-> 1.15.6 (can start ,cannot solve call timeout issue) (although my db-state starts to corrupt and could not move forward for new blocks, can still examine for old block)
-> 1.15.3 (can start ,solve call timeout issue,the call return immediately)
-> 1.15.4 (can start ,solve call timeout issue,the call return immediately)
-> 1.15.5 (can start ,solve call timeout issue,the call return immediately)
-> 1.15.6 (can start ,call timeout issue is back )

This sounds be
1.15.5-> 1.15.6 problem