Project

General

Profile

Actions

Bug #59681

open

[rbd_support] improve cli_generic.sh tests for recovery from blocklisting

Added by Ramana Raja almost 1 year ago. Updated 12 months ago.

Status:
New
Priority:
Normal
Assignee:
Target version:
-
% Done:

0%

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

Description

Ilya's suggestions,
"
for TrashPurgeScheduleHandler and MirrorSnapshotScheduleHandler tests, instead of just checking that both pre- and post-blocklisting schedules show up, it would be good to test that both pre- and post-blocklisting scheduled work actually gets done. For trash purge scheduler, I would suggest creating two pools with an image trashed in each, adding a short (1-2m) schedule for one pool before blocklisting and for another after blocklisting and asserting that trash purge runs after enough time passes. And similarly for mirror snapshot scheduler: a single pool with two images would do as per-image schedules can be added there.

for TaskHandler test, queue some (let's say 5) flattens on different images before blocklisting and a sixth flatten after blocklisting and assert that all six flattens complete."

Source: https://github.com/ceph/ceph/pull/49742#issuecomment-1537460028


Related issues 1 (0 open1 closed)

Related to rbd - Bug #56724: [rbd_support] recover from RADOS instance blocklistingResolvedRamana Raja

Actions
Actions

Also available in: Atom PDF