Open jcpunk opened 1 week ago
Do you use swap on your machine?
The system does have a linux swap partition, but I'm not showing any memory pressure or pages swapped out.
@jcpunk According to the iotop usage help a SWAPIN of 88% means that 88% of time was spent on swapping in memory pages from disk. If not from memory pressure then what other explanation could there be for swapped out memory pages?
I wonder if it hit the pod memory limit and started swapping rather than oom kill? I'll add the memory_limiter
to my config to see if that makes a difference. Not that I can trigger this bug on purpose...
Component(s)
No response
What happened?
Description
The otelcol-contrib binary suddenly is doing a whole lot of disk IO with no real reason/cause
Steps to Reproduce
Unknown, it just sorta happens after the container has been up for a week or so since v0.107.0 (first version I ever used)
Expected Result
Actual Result
Collector version
v0.112.0
Environment information
Environment
OS: Almalinux 9 Podman: ghcr.io/open-telemetry/opentelemetry-collector-releases/opentelemetry-collector-contrib:latest
OpenTelemetry Collector configuration
Log output
Additional context
[root@mu2e-dcs-01 ~]# cat /etc/containers/systemd/otel-collector.container
strace of process for little bit:
strace a little later
lsof for dedicated user:
podman ps: