Bug 1610997

Summary: Unable to install ceph-common during upgrade from 2.5 to 3.1 on xenial
Product: [Red Hat Storage] Red Hat Ceph Storage Reporter: Coady LaCroix <clacroix>
Component: Ceph-AnsibleAssignee: Ken Dreyer (Red Hat) <kdreyer>
Status: CLOSED ERRATA QA Contact: Coady LaCroix <clacroix>
Severity: urgent Docs Contact:
Priority: urgent    
Version: 3.1CC: aschoen, ceph-eng-bugs, clacroix, gabrioux, gmeno, hnallurv, kdreyer, nthomas, sankarshan, shan, vakulkar, vpoliset
Target Milestone: rcFlags: vakulkar: automate_bug+
Target Release: 3.1   
Hardware: Unspecified   
OS: Linux   
Whiteboard:
Fixed In Version: RHEL: ceph-ansible-3.1.0-0.1.rc17.el7cp Ubuntu: ceph-ansible_3.1.0~rc17-2redhat1 Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2018-09-26 18:23:43 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Comment 4 Christina Meno 2018-08-01 22:24:57 UTC
Coady,

Would you please provide the output of:
apt-cache showpkg ceph-base
apt-cache showpgk ceph-test

and the repository configs on the box in question?

Comment 6 Ken Dreyer (Red Hat) 2018-08-07 22:56:31 UTC
I've reproduced this. The minimal steps are:

1. set up rhcs 2.5 Ubuntu repos
2. sudo apt-get install ceph-mon ceph-test
3. switch to rhcs 3.1 Ubuntu repos
4. sudo apt-get install ceph-common

Actual results:
The following packages have unmet dependencies:
 ceph-base : Breaks: ceph-test (< 12.2.2-14) but 10.2.10-24redhat1xenial is to be installed
 ceph-mon : Breaks: ceph-test (< 12.2.2-14) but 10.2.10-24redhat1xenial is to be installed
E: Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages.


Expected results: Apt can upgrade the package(s) without error

When I run "apt-get install ceph-common ceph-test" instead, it works. This reminds me of the problem described at https://github.com/ceph/ceph-ansible/issues/2504

I'm still investigating some way we could solve this in the packaging.

Comment 7 Ken Dreyer (Red Hat) 2018-08-08 16:13:16 UTC
I've also reproduced this in the following scenarios:

A) Upgrading from 2.5 to 3.0
B) Upgrading upstream jewel to upstream luminous

In all cases, the solution is to upgrade the ceph-test deb before (or in conjunction with) ceph-common.

Comment 8 Ken Dreyer (Red Hat) 2018-08-08 17:29:30 UTC
I've found no way to resolve this in the Debian packaging. "apt-get install ceph-common" doesn't solve dependencies far enough to know to upgrade ceph-test.

I suspect we've never uncovered this bug before because most Ceph users run "apt-get dist-upgrade", or Teuthology runs "apt-get install <the entire list of packages to upgrade>".

I've pushed https://github.com/ceph/ceph-ansible/pull/2987 to run through upstream CI, but I have not personally verified that change this fixes this bug yet.

Comment 9 Ken Dreyer (Red Hat) 2018-08-08 22:46:26 UTC
I've tested PR 2987:
- upgrading a 2.5 monitor to 3.0
- upgrading a 2.5 monitor to 3.1

Guillaume or Sebastien, would you please merge that PR and tag a new stable-3.1 version?

Comment 13 subhash 2018-08-14 06:43:46 UTC
Hi Ken,

>3.0Z5 to 3.1 upgrade

Deployed 3.0z5(ubuntu) cluster with ceph_test:True parameter set in all.yml
Versions: 12.2.4-46redhat1xenial,3.0.39-2(ceph,ansible)

updated ceph-ansible to 3.1 version 

Upgraded to 3.1(by running rolling_update.yml)
Versions: 12.2.5-23redhat1xenial,3.1.0~rc17-2redhat1 

Playbook ran fine ,by updating all packages(even ceph_test).


>2.5 to 3.1 upgrade

rolling update ran fine(updating cept-test package first then later ceph-common) ,without any issues

>abt 2.5 to 3.0 upgrade
Not sure if the the fix is in rhcs3  3.0ceph-ansible?should the fix be backported ?
pl let us know

thanks

Comment 14 Coady LaCroix 2018-08-16 18:49:47 UTC
Ken,

I have also verified that the rolling update from 2.5 to 3.1 succeeds without any issues. I have the same question that subhash raised about 3.0. I have tested 2.5  to 3.0 and encountered the same issue. It seems like the fix for this issue is something we should backport to 3.0. If it is not I can mark this bz as verified, otherwise I would like to address it here before moving forward.

Thanks

Comment 17 Coady LaCroix 2018-08-16 23:21:49 UTC
Thanks for the confirmation Ken! Verified.

Comment 19 errata-xmlrpc 2018-09-26 18:23:43 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-2018:2819