Closed bleetube closed 1 year ago
I noticed this as well yesterday and thought I was losing my mind / had broken something strange. Turns out I had upgraded from Grafana 9.2.1 to 9.2.2. Downgrading to 9.2.1 fixed the problem. Seems like a regression in the 9.2.2 release?
Please, follow my answer: https://github.com/grafana/grafana/issues/57690#issuecomment-1296787752
Thanks @xelatirdan, now it shows the graph again.
I did notice the graph looks a little weird because it isn't scaling to max 100 (max now is 100 * nr cpus), a workaround for this is to change max to 'auto' instead of the default 100, the graph looks okay again but values are still multiplied with the number of CPU's.
After doing some research on the dashboard, I made a pull request to fix this. Tested on grafana 9.2.3. Working perfectly.
Thanks @xelatirdan, now it shows the graph again.
I did notice the graph looks a little weird because it isn't scaling to max 100 (max now is 100 * nr cpus), a workaround for this is to change max to 'auto' instead of the default 100, the graph looks okay again but values are still multiplied with the number of CPU's.
It works for me! Thank you very much.
@Z10N0110 I've tested your pr on grafana 9.2.6 and it works fine for me.
The change to using irate in 6bcaee1194cc9b578d1753a0fc15a24b52d8fc44 fixed the issue for me as well. I believe the bug in Grafana itself prior to that change still exists.
hi I still having this issue with Grafana 9.4.3
The funny thing is above 12 hours. I can see CPU graph
but I change it to under 12 hours I got this
I used use node_exporter 1.2.0 I upgrade to node_exporter 1.6.0 but still not showing a graph under 12 hours
any idea?
i change my prometheus config, then it works now
#scrape_interval: 1m
scrape_interval: 30s
This was fixed on https://github.com/grafana/grafana/issues/57690#issuecomment-1329324668
CPU graphs seem borked on rev28 for me. Both of my "CPU Basic" and "CPU" graphs are blank on revision 28 with Grafana 9.2.2. If I downgrade either Grafana to 9.1.0, or the dashboard to revision 27, the charts display data again. I think this is due to some new issue and not related to the old
$__rate_interval
problems. I've tried replacing withs/$__rate_interval/5m/
and they're still blank. Otherwise all other charts are showing data as expected. Since I'm working on an ansible playbook, I tried reproducing the issue on a couple different versions and two Ubuntu releases (different Ubuntu, Prometheus, and node_exporter versions made no difference).