op-node -> op-reth disconnect (op-reth v1.5.0+) #17539
Description
Describe the bug
Running base image v0.12.6 / 7 (op-reth v1.5.0 / op-reth v1.5.1) sync'ing base (mainnet)
Periodically op-reth (v1.5.0, v1.5.1) errors, logging:
ts=2025-07-18T09:12:57.487850699Z level=warn target=libmdbx message="Long-lived read transaction has been timed out" open_duration=300.001270456s backtrace=None
This aligns with op-node being unable to talk to op-reth:
t=2025-07-17T11:15:57+0000 lvl=warn msg="Engine temporary error" err="temp: failed to update insert payload: failed to execute payload: Post \"http://<reth-url>/\": context deadline exceeded"
op-reth and op-node do not subsequently resume comms, the node stalls and op-reth needs to be restarted to start sync'ing
It appears related to high rpc activity (discussing with the base team directly, esp. high vol. eth_call can contribute to this error, which has been observed for other providers).
Downgrading base to image v0.12.5 (op-reth v1.4.3) and running for 48 hours, we observe no instances of op-reth / op-node stopping comms and stalling the node. With versions v0.12.5 / v0.12.6 instances of the error are sporadic, but can be as frequent as hourly.
Steps to reproduce
Running base image v0.12.6 / 7 (op-reth v1.5.0 / op-reth v1.5.1) sync'ing base (mainnet)
Load node with rpc calls (esp. eth_call)
Node logs
Platform(s)
Linux (x86)
Container Type
Docker
What version/commit are you on?
v1.5.0-61e38f9 / v1.5.1-dbe7ee9
What database version are you on?
Current database version: 2
Local database version: 2
Which chain / network are you on?
base
What type of node are you running?
Archive (default)
What prune config do you use, if any?
No response
If you've built Reth from source, provide the full command you used
No response
Code of Conduct
- I agree to follow the Code of Conduct