Sahina Bose wrote:
On 03/28/2014 05:26 PM, Joop wrote:
> On 28-3-2014 12:38, Sahina Bose wrote:
>>
>> On 03/28/2014 04:02 PM, Joop wrote:
>>> Setting up oVirt with a separate cluster for storage using gluster
>>> we discovered that the gluster services are not set to autostart
>>> for the default runlevels.
>>> Host(s) were clean Centos-6.5 hosts with the correct repositories
>>> and then added from the webui. Hosts got all the packages and
>>> gluster was correctly started for that session but after a reboot
>>> gluster isn't running and the host was set to non-operational.
>>>
>>> Is this a known problem and if not should I file a BZ but under
>>> which category?
>>
>> Which version of gluster are you using?
>>
>> [Adding gluster-users]
>>
>>
> Sorry, the version that comes with the repos ovirt.repo and epel so
> 3.4.2. Version shouldn't matter because I expect that the hostdeploy
> from ovirt-engine to take care of making sure that glusterd and
> glusterfsd are started on system startup.
I think the chkconfig values are set by gluster rpm - which is why I
added gluster users.
AFAIU, host-deploy installs the rpm and starts the services after
installation, but does not change the chkconfig value.
Thats correct host-deploy starts the services but doesn't change the
chkconfig value and my collegues get bitten by it. They are used to
Debian/Ubuntu where packages which install services (almost) always make
sure that they are started at boot. I noticed in CentOS that is normally
not the case, oVirt being the exception :-)
Regards,
Joop