-
Notifications
You must be signed in to change notification settings - Fork 1.7k
RPC call duration skyrocketed after upgrade to v2.7.2-stable #11503
Comments
Thank you for the report, this looks worrying. Roughly how many RPC calls per second are you making here? |
@dvdplm It looks very worrying, indeed. The RPC calls we're executing can be seen in the screenshot I posted. We're executing those for every new block. In the case of That being said + considering the avg durations for the calls after the upgrade, I'd say it's way less than 1 RPC per second. |
@kwix we have changed a few rocksdb parameters from 2.5 to 2.7, it could be a background rocksdb process doing a compaction, could you show us your rocksdb log file? |
Hey @ordian ! This path |
@kwix sorry, I meant |
Here's the log you asked for: From what I can see, it starts today but the event happened 2 days ago. Thanks for looking into this! |
Right, have you restarted the node today? you can see a bunch of If you encounter this issue again, please let us know. |
Indeed the step-up in latency is correlated with a compaction at start time. We zoomed out a little and noticed that there are very similar load increases correlated with container restarts in the past as well; so I think we can safely assume there's no correlation with this upgrade :) Thanks! Closing this one 🙌 |
Configuration:
After we upgraded from v2.5.13-stable to v2.7.2-stable, we've seen a huge spike in RPC call execution durations.
Here's a screenshot from our monitoring:
The text was updated successfully, but these errors were encountered: