Project

General

Profile

Feature #47732

Issue health warning if a performance issue is occurring especially for ceph-osd client responses.

Added by David Zafman over 3 years ago. Updated over 3 years ago.

Status:
New
Priority:
Normal
Assignee:
David Zafman
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Reviewed:
Affected Versions:
Component(RADOS):
Pull request ID:

Description

This feature would identify a false network ping warning which might occur with a very busy ceph-osd(s).

The monitor could collect load averages from hosts somehow and report health warnings if necessary

The hard part is knowing when a warning is necessary. Also, are hosts in the crush map? If not, how do we identify which averages coming from our various daemons are relevant and how do we report the warning.

Maybe each ceph-osd report its own health in terms of client load instead? And forget about daemons. Just identify too slow client responses. Each OSD could warn if it can't provide some level of client performance.


Related issues

Related to Orchestrator - Feature #47885: Add networking checks New

History

#1 Updated by David Zafman over 3 years ago

  • Assignee set to David Zafman

#2 Updated by David Zafman over 3 years ago

#3 Updated by David Zafman over 3 years ago

Include in Orchestator checks?

#4 Updated by David Zafman over 3 years ago

Look at swap to make sure memory isn't over provisioned to containers, for example.

Do containers swap or crash if not enough memory?

Also available in: Atom PDF