You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Deploy postgresql with grafana-agent related through the cos-agent endpoint.
Expected behavior
Grafana-agent is able to collect relevant log files from prometheus without significant logging overhead.
Actual behavior
/var/log/syslog is excessively large, growing to 17 gigabytes in just over 48 hours, with an estimated 16 gigabytes being due to grafana-agent logging about the patroni log files. More specifically, roughly every 30 seconds I'm seeing 10,080 log entries, one for each of those 10,080 per-minute patroni log files.
Each line is a little over 300 characters. Repeated 10,080 times 24602 times a day (2,880 times), that comes out to 10,080 * 300 * 2,800 = 8,467,200,000 bytes, i.e. just short of 8 gigabytes per day. This means that for the just-over-2-day-old syslog file mentioned above, roughly 16 out of the 17 gigabytes mentioned is estimated to be due to these messages. And there's also the I/O impact of the same.
I think this should be reviewed, and perhaps should be discussed with
the COS team if necessary, to see how this charm can accomplish its
goals while also not generating excessively large syslogs due to the
interaction of grafana-agent with this charm.
The text was updated successfully, but these errors were encountered:
Steps to reproduce
Deploy postgresql with grafana-agent related through the cos-agent endpoint.
Expected behavior
Grafana-agent is able to collect relevant log files from prometheus without significant logging overhead.
Actual behavior
/var/log/syslog is excessively large, growing to 17 gigabytes in just over 48 hours, with an estimated 16 gigabytes being due to grafana-agent logging about the patroni log files. More specifically, roughly every 30 seconds I'm seeing 10,080 log entries, one for each of those 10,080 per-minute patroni log files.
Each line is a little over 300 characters. Repeated 10,080 times 24602 times a day (2,880 times), that comes out to 10,080 * 300 * 2,800 = 8,467,200,000 bytes, i.e. just short of 8 gigabytes per day. This means that for the just-over-2-day-old syslog file mentioned above, roughly 16 out of the 17 gigabytes mentioned is estimated to be due to these messages. And there's also the I/O impact of the same.
Versions
Operating system: Ubuntu 22.04.5 LTS
Juju CLI: 3.6.0-genericlinux-amd64
Juju agent: 3.4.6
Charm revision: 468 (14/stable)
LXD: N/A
Log output
Example log entry for the record:
Additional context
I think this should be reviewed, and perhaps should be discussed with
the COS team if necessary, to see how this charm can accomplish its
goals while also not generating excessively large syslogs due to the
interaction of grafana-agent with this charm.
The text was updated successfully, but these errors were encountered: