Thomas Lahmer

Thanks for the fast reply, I will give it a try.

Unfortunately, the same issue occurs with Matomo 5. Setting the recorders to 1 would probably resolve the deadlocks, but with that option set, I will run into time problems. Thanks!

@sgiehl Unfortunately not, all tables are stored using the InnoDB storage engine.

It is a bit hard to reproduce, but it also occurs on the current daily rotated logfiles. (`access_log.1`), so no, the data is not older than yesterday. But it often...

Unfortunately, QueuedTracking did not work with the log importing, as I get tracking errors. I think it is also not recommended to use it when importing logs (https://github.com/matomo-org/matomo-log-analytics/issues/240) Nevertheless, thanks...

No, each log (for different sites) is imported sequentially with multiple recorders. Afterwards, when all logs are imported, the archiving process is called. Pseudocode: ``` import_logs.py --recorders 5 site1_access_log.1 --idsite=1...

Thanks for the detailed analysis and report, this makes sense to me. I will give it a try!

Unfortunately, the issue remains with the recommended change. ``` | InnoDB | | ===================================== 2024-01-24 12:21:08 0x7fa7ee792640 INNODB MONITOR OUTPUT ===================================== Per second averages calculated from the last 19 seconds...