Project

General

Profile

Bug #36270

The "Many more objects per PG than average" warning does not work well when objects are differently-sized

Added by Greg Farnum 10 months ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
Administration/Usability
Target version:
-
Start date:
10/01/2018
Due date:
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Component(RADOS):
Manager (RADOS bits), Monitor, pgmap
Pull request ID:

Description

See https://bugzilla.redhat.com/show_bug.cgi?id=1633221

It's rare, but some clusters have pools with very differently-sized objects. "If the amount of data is matching and the object count isn't for different pools, we either need the warning to be more useful, like how it will negatively affect the cluster performance or everything's good to keep as it is" and so the warning shouldn't trigger.

I'm not sure if the best strategy is to check the total data and silence the warning per-pool if the total data is the same, or if we should just warn based on total data amounts rather than object counts, or what, but this warning should be made more careful.

Also available in: Atom PDF