Project

General

Profile

Bug #42709

mgr/dashboard: RBD tests must use pools with power-of-two pg_num

Added by Ricardo Marques 11 days ago. Updated 8 days ago.

Status:
Pending Backport
Priority:
Normal
Category:
dashboard/qa
Target version:
Start date:
11/08/2019
Due date:
% Done:

0%

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

Description

Since https://github.com/ceph/ceph/pull/30525 cluster health will change to "WARN" if non-power-of-two pg_num pools are used.

`test_rbd.py` is creating pools with non-power-of-two pg_num which will result in a timeout while waiting for cluster to be in health "OK":

2019-11-05 17:12:49,329.329 INFO:__main__:======================================================================
2019-11-05 17:12:49,329.329 INFO:__main__:ERROR: test_list (tasks.mgr.dashboard.test_rbd.RbdTest)
2019-11-05 17:12:49,329.329 INFO:__main__:----------------------------------------------------------------------
2019-11-05 17:12:49,330.330 INFO:__main__:Traceback (most recent call last):
2019-11-05 17:12:49,330.330 INFO:__main__:  File "/ceph/qa/tasks/mgr/dashboard/helper.py", line 154, in setUp
2019-11-05 17:12:49,330.330 INFO:__main__:    self.wait_for_health_clear(20)
2019-11-05 17:12:49,330.330 INFO:__main__:  File "/ceph/qa/tasks/ceph_test_case.py", line 134, in wait_for_health_clear
2019-11-05 17:12:49,330.330 INFO:__main__:    self.wait_until_true(is_clear, timeout)
2019-11-05 17:12:49,331.331 INFO:__main__:  File "/ceph/qa/tasks/ceph_test_case.py", line 166, in wait_until_true
2019-11-05 17:12:49,331.331 INFO:__main__:    raise RuntimeError("Timed out after {0}s".format(elapsed))
2019-11-05 17:12:49,331.331 INFO:__main__:RuntimeError: Timed out after 20s

Related issues

Related to RADOS - Bug #42215: cluster [WRN] Health check failed: 1 pool(s) have non-power-of-two pg_num (POOL_PG_NUM_NOT_POWER_OF_TWO)" in cluster log New 10/07/2019
Copied to mgr - Backport #42729: nautilus: mgr/dashboard: RBD tests must use pools with power-of-two pg_num In Progress

History

#1 Updated by Ricardo Marques 11 days ago

  • Status changed from New to In Progress

#2 Updated by Ricardo Marques 11 days ago

  • Category set to dashboard/qa

#3 Updated by Ricardo Marques 11 days ago

  • Status changed from In Progress to Need Review
  • Pull request ID set to 31490

#4 Updated by Tatjana Dehler 8 days ago

  • Related to Bug #42215: cluster [WRN] Health check failed: 1 pool(s) have non-power-of-two pg_num (POOL_PG_NUM_NOT_POWER_OF_TWO)" in cluster log added

#5 Updated by Lenz Grimmer 8 days ago

  • Status changed from Need Review to Pending Backport

#6 Updated by Ricardo Marques 8 days ago

  • Copied to Backport #42729: nautilus: mgr/dashboard: RBD tests must use pools with power-of-two pg_num added

Also available in: Atom PDF