We noticed high CPU usage from Influxdb. Data shows that this jump in CPU started at 7:50 one morning before anyone got in the office. It has persisted through several restarts and reboots. Previous CPU usage hovered around 37%, within a few minutes CPU jumped to 93% and has pretty much stayed there ever since.
System info:
OS: CentOS 6.8
InfluxDB: 1.7.1
Mem: 32GB (Usage hovers at 10GB, and 18GB cached)
Storage: 4TB SSD, 58% Used, Disk Activity Spikes to 10GB/s but generally < 1MB/s
Steps to reproduce:
Not sure what to put here, this started on it’s own. We’ve tried restarting the service and the CPU spikes right back up.
Expected behavior: Maintain expected levels of CPU usage. As far as we’ve been able to tell nothing in the env changed. No code was released, there were no new hosts sending data.
Actual behavior: CPU spiked out of nowhere. Almost exclusively used up by InfluxDB. It’s persisted since the 8th. This has led to slow performance for grafana.
Additional info:
influx-debug.host.tar.gz - https://drive.google.com/open?id=1P7kc9Mh8NWBcRzsEhqAtWw4yyLOhahe7
iostat.txt - https://drive.google.com/open?id=1ql9T88h1bQtkiltTKo92nj16QEO3RCsb
I’ve got the vars.txt file as well, but apparently new users can only put two links in the post.
