Upgrade to stein fails on creating pacemaker cluster

Bug #1834015 reported by Jiří Stránský
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
tripleo
Won't Fix
High
Unassigned

Bug Description

TASK [Debug output for task: Run puppet host configuration for step 1] *********

.. snip ..

        "<13>Jun 24 11:38:27 puppet-user: Error: /Stage[main]/Pacemaker::Corosync/Exec[Create Cluster tripleo_cluster]/returns: change from 'notrun' to ['0'] failed: '/sbin/pcs cluster setup tripleo_cluster cont
roller-0 addr=172.17.1.41' returned 1 instead of one of [0]",

When run manually on the node:

[root@controller-0 ~]# /sbin/pcs cluster setup tripleo_cluster controller-0 addr=172.17.1.41
Error: Host 'controller-0' is not known to pcs, try to authenticate the host using 'pcs host auth controller-0' command
Error: None of hosts is known to pcs.
Error: Errors have occurred, therefore pcs is unable to continue

After running `pcs host auth controller-0` and providing the user+password taken from hiera and re-running the upgrade step, the cluster gets created.

Tags: upgrade
Revision history for this message
Jiří Stránský (jistr) wrote :

The hiera key for the generated password is "hacluster_pwd".

Revision history for this message
Michele Baldessari (michele) wrote :

Thanks for the heads-up Jiri, so yeah pcs has this new-fangled way of setting up the cluster. It was mandatory for remotes and that is where we actually added its support via https://github.com/openstack/puppet-pacemaker/commit/fde568564d43c78bf2828d3c981d5cc685fb3369

I wonder though why it only barfs after an upgrade and not on fresh deploys

Revision history for this message
Jiří Stránský (jistr) wrote :

@Michele Before upgrading the operating system we do `pcs cluster stop` and `pcs cluster destroy`, not sure if this can somehow cause it being different from fresh deploy. It's still possible that it was environmental issue and i was missing something, i did hit it on 2 different attempts though.

Revision history for this message
Jiří Stránský (jistr) wrote :

@Michele oh and one more important thing, we utilize these overrides:

https://github.com/openstack/puppet-tripleo/blob/0ad12ff23ab79e2e25baff14c7215e21f275570b/manifests/profile/base/pacemaker.pp#L124

https://github.com/openstack/puppet-tripleo/blob/0ad12ff23ab79e2e25baff14c7215e21f275570b/manifests/profile/base/pacemaker.pp#L143

But then again, when i upgraded the last controller (controller-2), the overrides should already be cleared when Puppet runs, and it still didn't authenticate on controller-2 automatically.

Changed in tripleo:
milestone: train-2 → train-3
Changed in tripleo:
milestone: train-3 → ussuri-1
Changed in tripleo:
milestone: ussuri-1 → ussuri-2
wes hayutin (weshayutin)
Changed in tripleo:
milestone: ussuri-2 → ussuri-3
wes hayutin (weshayutin)
Changed in tripleo:
milestone: ussuri-3 → ussuri-rc3
wes hayutin (weshayutin)
Changed in tripleo:
milestone: ussuri-rc3 → victoria-1
Revision history for this message
Jiří Stránský (jistr) wrote :

Stale bug, let's reopen if necessary.

Changed in tripleo:
status: Triaged → Won't Fix
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.