[Pacemaker] can't get pacemaker started

[Andrew Beekhof]

On Tue, Jul 31, 2012 at 1:55 AM, Dave Jiang <dave.jiang at intel.com> wrote:
> On Sun 29 Jul 2012 08:47:48 PM MST, Andrew Beekhof wrote:
>> On Fri, Jul 27, 2012 at 9:31 AM, Dave Jiang <dave.jiang at intel.com> wrote:
>>> Hi. I'm following the cluster from scratch guide to create a simple
>>> active/passive 2 node cluster. I'm using the standard packages that come
>>> with Fedora 17. I have corosync running and linked up. However I cannot
>>> seem to get Pacemaker to run correctly. I don't see all the processes
>>> loaded:
>>>
>>> 17286 ?        Ss     0:00 /usr/sbin/pacemakerd
>>> -f
>>> 17288 ?        Ss     0:00  \_ /usr/libexec/pacemaker/stonithd
>>>
>>> Looking at the log these stand out:
>>>
>>> Jul 26 16:26:02 leftnode cib[17378]:  warning: retrieveCib: Cluster
>>> configuration not found: /var/lib/heartbeat/crm/cib.xml
>>> Jul 26 16:26:02 leftnode attrd[17381]:   notice: crm_cluster_connect:
>>> Connecting to cluster infrastructure: corosync
>>> Jul 26 16:26:02 leftnode cib[17378]:  warning: readCibXmlFile: Primary
>>> configuration corrupt or unusable, trying backup...
>>> Jul 26 16:26:02 leftnode crmd[17383]:     info: crm_log_init_worker:
>>> Changed active directory to /var/lib/heartbeat/cores/hacluster
>>> Jul 26 16:26:02 leftnode cib[17378]:  warning: readCibXmlFile:
>>> Continuing with an empty configuration.
>>>
>>> I'm not running heartbeat, should I be? It wasn't talked about in the guide.
>>
>> No. Some of these paths were determined back when Pacemaker was part
>> of heartbeat.
>> The last few are being changed for 1.1.8
>>
>>>
>>> And then I noticed the qb_rb_chmod failed and a bunch of other failures.
>>> Any ideas what am I not setting up correctly?
>>
>> That looks quite odd.
>> Which user are you starting corosync as?
>>
>> Could you show the rpm version of corosync and libqb please?
>
> I was running as root. I finally got it working. It was the evil that
> is selinux. I thought I had it turned off but apparently I didn't.
> Thanks. Probably should mention that as a gotcha in the guide though.

It is mentioned in clusters from scratch.  Was that the document you
were reading?