Bug #59512: Draining scrubbing seems impossible - Ceph - Ceph

Actions

Bug #59512

open

Status:

New

Priority:

Normal

Assignee:

Category:

OSD

Target version:

% Done:

Source:

Tags:

Backport:

Regression:

Severity:

3 - minor

Reviewed:

Affected Versions:

ceph-qa-suite:

Pull request ID:

Crash signature (v1):

Crash signature (v2):

Description

I would like to temporarily halt scrubbing. I ran

ceph osd set noscrub
ceph osd set nodeep-scrub

but observe in "ceph status" that

4   active+clean+scrubbing+deep

and at the same time, "iostat -x 5" indicates that there is no IO at all, so no deep-scrubbing can be going on.

This indicates that it is impossible to drain scrubbing from a cluster.

Actions

Background in my case:

I would like to drain all ongoing scrubs because I wish to change "osd_deep_scrub_stride" and https://tracker.ceph.com/issues/58240 / https://github.com/ceph/ceph/commit/db8d185288d0e9602f654c89b7f18f8be45de3d0 suggests thta in my Ceph version, I should better first stop all ongoing scrubs.

But I think in general the fact that scrubbing cannot be drained looks like a bug.

There is also the tangentially related issue https://tracker.ceph.com/issues/11202 but that is about instanteneous stopping of scrubs, not draining.

Actions