ngc104
ngc104
Hello @roidelapluie I'm back on this issue (replacing @Winael ). If you can come back on it too (or anybody else), please tell me how to take pprof traces if...
@prashbnair not sure to understand... You want me to run `go tool pprof -sample_index=inuse_space` instead of just `go tool pprof` ? I will try that ASAP. Are PDF files the...
Starting taking the traces. They will be posted later today. Here is my script to take traces : ``` #! /bin/bash DATE=$(date +'%Y%m%d_%H%M%S') FILEPREFIX="profile-${DATE}" go tool pprof -sample_index=inuse_space -pdf http://localhost:9093/debug/pprof/allocs...
# pprof more profiles.pdf I restarted my test (as described in a previous comment above). I took some traces right after I started Alertmanager and my loop for the alert,...
@prashbnair Yes. In fact, the main problem is Kubernetes killing the Alertmanager pod container when going OOM because of the "memory" going higher than the resource limit. What is the...
Restarted my test at 9h22 UTC. Here is some interesting result at 10h03 UTC... ``` ➜ ./alertmanager/amtool --alertmanager.url=http://localhost:9093/ silence query --expired ID Matchers Ends At Created By Comment 2fca38e1-9dc3-4fbf-8f35-291dcac02a34 alertname="foo"...
I was bad on this one... I was exploring data.retention in the Production environment. We have data.retention unset (e.g. using the default value 120h / 5 days). With amtool, I...
Hello and happy new year ! Alertmanager-0.25.0 is released. Let's give a try... Same procedure as above (in a previous comment) 12h37 [profile-20230106_155103-allocs.pdf](https://github.com/prometheus/alertmanager/files/10360911/profile-20230106_155103-allocs.pdf) 14h06 [profile-20230106_140612-allocs.pdf](https://github.com/prometheus/alertmanager/files/10360912/profile-20230106_140612-allocs.pdf) 16h55 [profile-20230106_123718-allocs.pdf](https://github.com/prometheus/alertmanager/files/10360914/profile-20230106_123718-allocs.pdf) 12h37 **14Mb** (`nodes...
Hello, I had not noticed that I was running a standalone alertmanager in cluster mode. In order to remove any noise in the investigation, I retried with this command line...
Hello, Sorry for the delay... Too much work besides this incident. And I'm not sure that the "raw" pprof files are obtained just with `-raw` option instead of `-pdf`. Please...