Project

General

Profile

Bug #51906

mgr/dashboard: review units in Grafana Dashboards

Added by Ernesto Puerta over 2 years ago. Updated over 1 year ago.

Status:
Triaged
Priority:
High
Category:
Monitoring
Target version:
-
% Done:

0%

Source:
other
Tags:
Backport:
Regression:
No
Severity:
2 - major
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

After the recent fix https://github.com/ceph/ceph/pull/42217, we'd need to review all the Grafana dashboards to ensure that if no metric unit is made explicit (e.g.: milliseconds in ceph_osd_apply_latency_ms), they'd be using SI units or base (seconds, bytes, bits) and not multiples/submultiples.

We should put extra emphasis on latency related metrics (*_latency_*): ceph_osd_op_r_latency_sum, etc.

An case raised by Ben England:

OSD device details

Y-axis units need to be correct.  "MBs" -> MBps or MB/s.   It would be really interesting to have a set of graphs with *all* the OSDs in them (if there are less than 20 of them).   If there are more than that, just show me the "top" OSDs with respect to each metric.

History

#1 Updated by Ernesto Puerta over 2 years ago

  • Description updated (diff)

#2 Updated by Ernesto Puerta over 2 years ago

  • Status changed from New to Triaged

#3 Updated by Loïc Dachary over 2 years ago

  • Target version deleted (v16.2.6)

#4 Updated by Ernesto Puerta over 2 years ago

  • Description updated (diff)

#5 Updated by Voja Molani over 1 year ago

Would the difference between Bps and bps be in-scope of this issue? Noticed that in the "Network Load" graph the y scale is in "B/s" such as "3.4 GB/s" but I am quite certain it is actually displaying bits per second ("b/s") and not bytes per second ("B/s").

Also available in: Atom PDF