Bug 1753178 - Stand-by MGRs are going down after enabling all mgr modules sequentially
Summary: Stand-by MGRs are going down after enabling all mgr modules sequentially
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Red Hat Ceph Storage
Classification: Red Hat Storage
Component: RADOS
Version: 3.3
Hardware: Unspecified
OS: Unspecified
unspecified
high
Target Milestone: z1
: 3.3
Assignee: Kefu Chai
QA Contact: Uday kurundwade
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2019-09-18 10:30 UTC by Uday kurundwade
Modified: 2019-10-22 13:29 UTC (History)
10 users (show)

Fixed In Version: RHEL: ceph-12.2.12-73.el7cp Ubuntu: ceph_12.2.12-66redhat1
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2019-10-22 13:29:30 UTC
Embargoed:

Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Ceph Project Bug Tracker 38460 0 None None None 2019-09-29 07:35:29 UTC
Ceph Project Bug Tracker 42086 0 None None None 2019-09-27 17:23:12 UTC
Github ceph ceph pull 27639 0 None closed luminous: mgr: drop GIL in StandbyPyModule::get_config 2020-01-30 21:26:43 UTC
Red Hat Product Errata RHBA-2019:3173 0 None None None 2019-10-22 13:29:46 UTC

Description Uday kurundwade 2019-09-18 10:30:51 UTC 
Description of problem:
Stand-by MGRs are going down after enabling all mgr modules sequentially

Version-Release number of selected component (if applicable):

ceph-base-12.2.12-70.el7cp.x86_64
ceph-mgr-12.2.12-70.el7cp.x86_64
ceph-ansible-3.2.27-1.el7cp.noarch
ceph-common-12.2.12-70.el7cp.x86_64


How reproducible:
Always

Steps to Reproduce:
1.Deploy ceph cluster with atleast 3 mgrs
2.Enable all available mgr modules one after another on active mgr


Actual results:
Stand-by mgrs are going down 

Expected results:
Stand-by mgrs should be up and running

Additional info:
Stand-by mgr status is Active and Running but not displaying in ceph status
Comment 1 Boris Ranto 2019-09-20 09:15:50 UTC 
When you enable all the ceph-mgr modules, it looks like the standby mgrs are hitting a deadlock. The daemons are still running but they are not reachable by the cluster, i.e. ceph -s does not show the standbys anymore. Re-targetting to rados since this looks like a ceph-mgr daemon issue.
Comment 3 Kefu Chai 2019-09-27 14:51:37 UTC 
i am able to reproduce it. looking
Comment 4 Kefu Chai 2019-09-27 17:23:12 UTC 
upstream PR pending on review: https://github.com/ceph/ceph/pull/30623
Comment 5 Kefu Chai 2019-09-27 17:35:14 UTC 
fix pushed to ceph-3.3-rhel-patches
Comment 14 Kefu Chai 2019-09-29 07:03:55 UTC 
Thanks Thomas!
Comment 15 Kefu Chai 2019-09-29 07:36:21 UTC 
the upstream PR is actually https://github.com/ceph/ceph/pull/27639. my change duplicates it.
Comment 18 errata-xmlrpc 2019-10-22 13:29:30 UTC 
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2019:3173
Note You need to log in before you can comment on or make changes to this bug.