[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [rdo-list] Newton HA galera-ready dependency error



[heat-admin overcloud-controller-0 ~]$ sudo hiera mysql_short_node_names
["overcloud-controller-0", "overcloud-controller-1", "overcloud-controller-2"]

On 12/10/2016 09:09, Marius Cornea wrote:
That's odd. I encountered the same issue and it was caused by missing
this patch.  What do you get if you do sudo hiera
mysql_short_node_names on the controller node?

On Wed, Oct 12, 2016 at 12:48 AM, Charles Short <cems ebi ac uk> wrote:
Hi,

puppet-tripleo-5.2.0-0.20161007035759.f32e484.el7.centos.noarch

The patch seems to be already present -

   grep short
/usr/share/openstack-puppet/modules/tripleo/manifests/profile/pacemaker/database/mysql.pp

   # short name which is already registered in pacemaker until we get around
   $galera_node_names_lookup = hiera('mysql_short_node_names',
hiera('mysql_node_names', $::hostname))

Charles

On 11/10/2016 19:39, Marius Cornea wrote:
I think the issue is caused by the addresses in wsrep_cluster_address
not matching the pacemaker node names:

wsrep_cluster_address=gcomm://overcloud-controller-0.internalapi.localdomain,overcloud-controller-1.internalapi.localdomain,overcloud-controller-2.internalapi.localdomain

Could you please confirm what version of puppet-tripleo you've got
installed on the overcloud nodes and if it contains the following
patch:
https://review.openstack.org/#/c/382883/1/manifests/profile/pacemaker/database/mysql.pp

Thanks,
Marius

On Tue, Oct 11, 2016 at 7:42 PM, Charles Short <cems ebi ac uk> wrote:
Ok install finished with same error
The latest pcs status etc

http://pastebin.com/ZK683gZe


On 11/10/2016 17:35, Charles Short wrote:
Deployment almost finished...so


http://pastebin.com/zE9B19XB

This shows the pcs status as the deployment nears the end, and pcs
resource show galera

Charles

On 11/10/2016 16:59, Marius Cornea wrote:
Great, thanks for checking this.

On Tue, Oct 11, 2016 at 5:58 PM, Charles Short <cems ebi ac uk> wrote:
Currently having more generic deployment issues (no valid host found
etc).
I can work around/solve these.
I don't yet have another stack to analyse, but will do soon.

Charles


On 11/10/2016 16:35, Marius Cornea wrote:
Did it succeed in bringing the Galera nodes to Master? You can ssh to
the nodes and run 'pcs resource show galera' even though the
deployment hasn't finished. I'm interested to see how the
wsrep_cluster_address is set to see if it's affected by the resource
agent issue described in
https://bugs.launchpad.net/tripleo/+bug/1628521

On Tue, Oct 11, 2016 at 5:18 PM, Charles Short <cems ebi ac uk>
wrote:
Looks similar to this bug (still waiting on deployment to finish)

https://bugzilla.redhat.com/show_bug.cgi?id=1368214


On 11/10/2016 15:25, Charles Short wrote:
Sorry for the delay.


Just redeploying to make sure I can repeat the same error. Should
not
be
long.

Charles

On 11/10/2016 14:24, Marius Cornea wrote:
Hi,

Could you also please paste the output for 'pcs resource show
galera',
it looks that all the galera nodes show up as slaves?

      Master/Slave Set: galera-master [galera]
          Slaves: [ overcloud-controller-0 overcloud-controller-1
overcloud-controller-2 ]

Thanks

On Tue, Oct 11, 2016 at 2:16 PM, Charles Short <cems ebi ac uk>
wrote:
Hi,

Here you are -

      - Heat stack error  - http://pastebin.com/E8KZa2vE
      - PCS status  - http://pastebin.com/z34gSLq6
      - mariadb.log - http://pastebin.com/APFXPBLc

Thanks

Charles


On 11/10/2016 12:07, Marius Cornea wrote:
Hi Charles,

Could you please paste the output of 'pcs status' ? The log in
/var/log/mariadb/mariadb.log might also be a good indicator.


On Tue, Oct 11, 2016 at 11:16 AM, Charles Short <cems ebi ac uk>
wrote:
To add I built my own image from






http://cloud.centos.org/centos/7/images/CentOS-7-x86_64-GenericCloud.qcow2

as the images in





http://buildlogs.centos.org/centos/7/cloud/x86_64/tripleo_images/newton/delorean/
caused sporadic ramdisk loading errors  (hung at x% loaded on
boot)

Does my image now need to be customised in any way for HA to
work?

Charles




On 11/10/2016 09:55, Charles Short wrote:
Hi,

I am installing Newton with TripleO on baremetal HP blades.
I can deploy a single controller stack overcloud no problem,
however
when
I choose three controllers the deployment fails (including






/usr/share/openstack-tripleo-heat-templates/environments/puppet-pacemaker.yaml)
.

The heat stack error first complains "Dependency
Exec[galera-ready]
has
failures" which in turn causes lots of other errors.

I have deployed Liberty and Mitaka successfully in the past on
baremetal
with three controllers, and this is the first time I have seen
this
error.

Charles

--
Charles Short
Cloud Engineer
Virtualization and Cloud Team
European Bioinformatics Institute (EMBL-EBI)
Tel: +44 (0)1223 494205

_______________________________________________
rdo-list mailing list
rdo-list redhat com
https://www.redhat.com/mailman/listinfo/rdo-list

To unsubscribe: rdo-list-unsubscribe redhat com

--
Charles Short
Cloud Engineer
Virtualization and Cloud Team
European Bioinformatics Institute (EMBL-EBI)
Tel: +44 (0)1223 494205

--
Charles Short
Cloud Engineer
Virtualization and Cloud Team
European Bioinformatics Institute (EMBL-EBI)
Tel: +44 (0)1223 494205

--
Charles Short
Cloud Engineer
Virtualization and Cloud Team
European Bioinformatics Institute (EMBL-EBI)
Tel: +44 (0)1223 494205

--
Charles Short
Cloud Engineer
Virtualization and Cloud Team
European Bioinformatics Institute (EMBL-EBI)
Tel: +44 (0)1223 494205

--
Charles Short
Cloud Engineer
Virtualization and Cloud Team
European Bioinformatics Institute (EMBL-EBI)
Tel: +44 (0)1223 494205


--
Charles Short
Cloud Engineer
Virtualization and Cloud Team
European Bioinformatics Institute (EMBL-EBI)
Tel: +44 (0)1223 494205


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]