ceph-osd adn ceph-cluster dashboard can't import due to metrics absent for osd_perf

[Zhanglei Mao]

The pattern dashboard "CephCluster.json.j2 CephOSD.json.j2" requires the metircs for
"ceph_osd_perf_apply_latency_seconds', 'ceph_osd_perf_commit_latency_seconds". And it seems the prometheus-ceph-exporter haven't provided it yet. So above dashboards wouldn't be imported.

[Zhanglei Mao (zhanglei-mao)]

A workaround is to add them to the ignore_metrics like below:

root@grafana-1:/var/log/juju# diff /var/lib/juju/agents/unit-grafana-0/charm/reactive/grafana.py /home/ubuntu/grafana.py
924c924,926
< "neutron_public_ip_usage"
---
> "neutron_public_ip_usage",
> 'ceph_osd_perf_apply_latency_seconds',
> 'ceph_osd_perf_commit_latency_seconds'
root@grafana-1:/var/log/juju#

[Zhanglei Mao (zhanglei-mao)]

This bug is duplicated to this LP#1867100. The upsteam seems have it fixed and it was waiting the snap updating for prometheus-ceph-exporter.