Join GitHub today
GitHub is home to over 31 million developers working together to host and review code, manage projects, and build software together.
Sign upPanic in LabelValues sorting (Prometheus 2.0) #3217
Comments
grobie
added
dev-2.0
kind/bug
labels
Sep 25, 2017
This comment has been minimized.
This comment has been minimized.
|
Hmm, this looks like memory corruption and the only point where I can see that something could go wrong is this: https://github.com/prometheus/prometheus/blob/dev-2.0/storage/tsdb/tsdb.go#L251-L257 I am not sure even this could cause anything wrong as both definitions are the same. My simple tests revealed nothing wrong. This is happening in |
This comment has been minimized.
This comment has been minimized.
|
Our prometheus server scraping node_exporters in one of our datacenters has hit this problem 27 times during the last 7d. |
This comment has been minimized.
This comment has been minimized.
|
@grobie considered fixed? |
This comment has been minimized.
This comment has been minimized.
|
None of our servers have experienced such a panic since rc.0. |
grobie
closed this
Oct 10, 2017
This comment has been minimized.
This comment has been minimized.
lock
bot
commented
Mar 23, 2019
|
This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs. |
grobie commentedSep 25, 2017
What did you do?
Run Prometheus v2.0.0-beta.5 (plus a tiny patch #3215) against our node_exporter infrastructure (so very static targets, almost no time series churn)
What did you expect to see?
Stable Prometheus servers, without any gaps in data.
What did you see instead? Under which circumstances?
I first noticed that our graphs had gaps. I then realized that these gaps were the results of crashes. All crashes happen at the beginning of a compaction cycle it seems.
Environment
Linux 4.4.10+soundcloud #1 SMP Thu Jun 16 15:17:20 UTC 2016 x86_64 GNU/Linux
The full stack trace includes over 50k lines