whylabs/whylogs-container

Sqlite performance issues relating to disk usage

Opened this issue · 0 comments

I summarized the issue in an SO post here when I hit a wall: https://stackoverflow.com/questions/71963259/what-could-cause-a-sqlite-application-to-slow-down-over-time-with-high-load

For whatever reason, after hitting the container (in REST mode) for long enough (about a minute on the machine I was using) disk utilization caps out and throughput drops off. It doesn't get worse the longer it you use it, it just hits some bottleneck that I can't identify yet.

It would be nice to figure this out at some point because the container takes a ~50% tps hit once whatever it is that's happening happens, then it chugs along at that rate.