Project

General

Profile

Bug #53894

vstart should disable pg autoscaler automatically when running crimson

Added by Joseph Sawaya about 2 years ago. Updated almost 2 years ago.

Status:
Resolved
Priority:
Low
Assignee:
-
Target version:
-
% Done:

0%

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

Description

When running vstart and creating a pool with 1 PG I'd get this in the ceph -s output:

data:
pools: 2 pools, 2 pgs
objects: 1 objects, 0 B
usage: 80 KiB used, 100 GiB / 100 GiB avail
pgs: 2 active+clean

Then shortly after I would get this:

data:
pools: 2 pools, 33 pgs
objects: 1 objects, 0 B
usage: 92 KiB used, 100 GiB / 100 GiB avail
pgs: 93.939% pgs unknown
31 unknown
2 active+clean

This is due to the pg autoscaler running, which apparently is broken on Crimson, so it should be disabled automatically when running vstart on Crimson OSD.

After running ceph osd pool set noautoscale I get this as the output from ceph -s after creating a pool:

data:
pools: 2 pools, 33 pgs
objects: 1 objects, 0 B
usage: 268 KiB used, 100 GiB / 100 GiB avail
pgs: 33 active+clean

History

#1 Updated by Matan Breizman almost 2 years ago

  • Status changed from New to Resolved
  • Pull request ID set to 45317

Also available in: Atom PDF