Bug 1572933 - infrastructure-playbooks/shrink-osd.yml leaves behind NVMe partition; scenario non-collocated
Summary: infrastructure-playbooks/shrink-osd.yml leaves behind NVMe partition; scenari...
Alias: None
Product: Red Hat Ceph Storage
Classification: Red Hat
Component: Ceph-Ansible
Version: 3.0
Hardware: Unspecified
OS: Unspecified
Target Milestone: rc
: 3.3
Assignee: Guillaume Abrioux
QA Contact: Vasishta
Erin Donnelly
Depends On: 1644847
Blocks: 1572368 1629656 1726135
TreeView+ depends on / blocked
Reported: 2018-04-29 03:27 UTC by Randy Martinez
Modified: 2019-08-21 15:10 UTC (History)
18 users (show)

Fixed In Version: RHEL: ceph-ansible-3.2.16-1.el7cp Ubuntu: ceph-ansible_3.2.16-2redhat1 Container: rhceph:ceph-3.3-rhel-7-containers-candidate-89086-20190718150813
Doc Type: Bug Fix
Doc Text:
.The `shrink-osd.yml` playbook removes partitions from NVMe disks in all situations Previously, the Ansible playbook `infrastructure-playbooks/shrink-osd.yml` did not properly remove partitions on NVMe devices when used with the `osd_scenario: non-collocated` option in containerized environments. This bug has been fixed with this update, and the playbook removes the partitions as expected.
Clone Of:
Last Closed: 2019-08-21 15:10:24 UTC
Target Upstream Version:

Attachments (Terms of Use)
File contains playbook log, inventory file (279.05 KB, text/plain)
2019-07-10 17:39 UTC, Vasishta
no flags Details

System ID Private Priority Status Summary Last Updated
Github ceph ceph-ansible pull 2896 0 None None None 2018-07-18 14:22:53 UTC
Github ceph ceph-ansible pull 2907 0 None None None 2018-07-19 13:49:26 UTC
Github ceph ceph-container pull 1119 0 None None None 2018-07-19 13:49:59 UTC
Github ceph ceph-container pull 1425 0 None None None 2019-07-16 12:36:34 UTC
Github https://github.com/ceph ceph-ansible pull 3515 0 None None None 2019-01-18 22:48:23 UTC
Red Hat Product Errata RHSA-2019:2538 0 None None None 2019-08-21 15:10:41 UTC

Description Randy Martinez 2018-04-29 03:27:46 UTC
Description of problem:

infrastructure-playbooks/shrink-osd.yml does not clean up NVME partitions w/scenario non-collocated.

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

How reproducible:

Steps to Reproduce:
1. run shrink-osd.yml on non-collocated osd(NVME journal).

Comment 3 seb 2018-07-18 14:22:54 UTC
That is not part of the scope of shrink-osd.yml.

Comment 4 seb 2018-07-18 14:23:50 UTC
Is it happening on containerized or non-containerized?

Comment 5 Randy Martinez 2018-07-18 16:15:35 UTC

Comment 6 seb 2018-07-19 11:58:12 UTC
Hum, I guess this is a bug in ceph-disk then, we use the following line to destroy and zap the OSD:

ceph-disk destroy --cluster {{ cluster }} --destroy-by-id {{ item.0 }} --zap

So I'd assume it's ceph-disk's job to do the cleanup.
I'll see if I can add a task for this since ceph-disk is in the deprecation path.

Comment 7 Ken Dreyer (Red Hat) 2018-07-24 21:59:11 UTC
Josh Durgin pointed out that shrinking clusters is not a common scenario and this should not block the 3.1 release. Re-targeting until we can resolve this.

Comment 11 Vasishta 2018-11-16 10:40:07 UTC
Hi Sebastien, 

Working fine in non-containerized scenario 
NVMe partitions are still left behind in containerized scenario.

I think that it might be because of 

Moving to ASSIGNED state.

Vasishta Shastry
QE, Ceph

Comment 18 Vasishta 2019-07-10 17:36:34 UTC
Partion on dedicated device was not removed on non-NVMe device when shrink-osd-ceph-disk.yml was used
Moving back to ASSIGNED state.

Vasishta Shastry
QE, Ceph

Comment 19 Vasishta 2019-07-10 17:39:42 UTC
Created attachment 1589176 [details]
File contains playbook log, inventory file

Comment 20 Noah Watkins 2019-07-15 20:40:04 UTC
Hey Guillaume

> fix will be present in v3.2.16

Did this make it in?

Comment 27 errata-xmlrpc 2019-08-21 15:10:24 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.


Note You need to log in before you can comment on or make changes to this bug.