Open mdschmitt opened 2 years ago
ugh, it seems that if the rules are changed, a reload is not enough, the service must be restarted. Back to the memory snapshot feature, I guess, unless I create a whole separate service.running resource for restarting instead of reloading.
https://saltstackcommunity.slack.com/archives/C7K04SEJC/p1662430673735549
Is your feature request related to a problem?
There's actually no need to restart the prometheus service if you're just updating the configs. A simple
reload
will do the trick. https://prometheus.io/docs/introduction/faq/#can-i-reload-prometheuss-configurationDescribe the solution you'd like
IF
AND
THEN
Reload the prometheus service instead of "hard" restarting it.
Describe alternatives you've considered
I guess maybe enabling
--enable-feature=memory-snapshot-on-shutdown
https://prometheus.io/docs/prometheus/latest/feature_flags/#memory-snapshot-on-shutdown would help minimize restart time, but it's technically an experimental feature which makes me think I shouldn't run it in prod..Additional context
If your prometheus installation is large-scale (i.e. scrapes a LOT of data), restarting it can take minutes while it replays the WAL logs. Reloading involves no downtime.