[CEPH] Grafana misses performance-related info from OSDs
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Grafana Charm |
New
|
Undecided
|
Unassigned |
Bug Description
Using bionic-ussuri and Ceph Octopus.
My deployment runs with ceph_exporter snap v2.0.0
Also saw that problem on Focal distro.
Deployed ceph with the following bundle: https:/
After I restart prometheus because of LP #1891942,grafana comes up with only Ceph Pools dashboard.
On grafana charm logs, I can see:
2020-10-15 07:51:19 DEBUG juju-log Skipping Dashboard Template: CephCluster.json.j2 missing 2 metrics.Missing: ceph_osd_
Likewise for CephOSD dashboard:
2020-10-15 07:51:18 DEBUG juju-log Skipping Dashboard Template: CephOSD.json.j2 missing 2 metrics.Missing: ceph_osd_
I can see there was another bug reported with similar issue:
https:/
Although the bug is outdated, comments are pretty recent, from a couple of months ago.
So, looking into the missing metrics a little closer, I am not sure where this is supposed to be built, either collected directly from Ceph or calculated with existing metrics.
I cannot find them on ceph_exporter source code.
Only reference on upstream ceph_exporter I can see is on: https:/
Which only adds some tests to check for those metrics but no logic to collect or calculate them.
The commit that added those metrics to grafana charm is:
https:/
description: | updated |
description: | updated |
Following: https:/ /bugs.launchpad .net/charm- prometheus- ceph-exporter/ +bug/1867100 ceph-exporter to latest/edge partially resolved the issue
and moving prometheus-
Ceph OSD dashboard is showing "0" osds now.