On 03/04/2014 10:19 AM, Itamar Heim wrote:
On 03/04/2014 11:15 AM, René Koch wrote:
On 03/04/2014 09:40 AM, Itamar Heim wrote:
On 03/04/2014 10:16 AM, René Koch wrote:
On 03/04/2014 05:14 AM, Sahina Bose wrote:
On 03/04/2014 01:59 AM, Itamar Heim wrote:
On 03/03/2014 07:26 PM, René Koch wrote: > Hi list, > > My hosted engine is running again, so I want to start a new thread > for > another issue with my setup. > > I have a GlusterFS storage domain, which can be mounted from CLI > without > problems. oVirt is 3.4 from ovirt-3.4.0-prerelease repository > running on > CentOS 6.5 with latest updates (both OS and oVirt). > > Both hosts, which act as hypervisors and GlusterFS nodes are in > state > "Non Operational" in oVirt because "Gluster command [Non interactive > user] failed on server ovirt-host02.dmz.linuxland.at." > > In engine.log I see the entry "glusterHostList is not supported" > (attached are the log entries when activating one of the hosts): > > 2014-03-03 18:17:11,764 ERROR > [org.ovirt.engine.core.vdsbroker.gluster.GlusterServersListVDSCommand] > > > (org.ovirt.thread.pool-6-thread-21) [6eee3cbd] Command > GlusterServersListVDSCommand(HostName = > ovirt-host02.dmz.linuxland.at, > HostId = dd399eeb-f623-457a-9986-a7efc69010b2) execution failed. > Exception: VDSNetworkException: org.apache.xmlrpc.XmlRpcException: > <type > 'exceptions.Exception'>:method "glusterHostsList" is not supported > > Can you give me a hint what this means and how I can activate my > hosts > and storage again? > Thanks a lot! > > > > > _______________________________________________ > Users mailing list > Users@ovirt.org > http://lists.ovirt.org/mailman/listinfo/users >
sahina ?
Do you have vdsm-gluster on the node?
No, I didn't had it on (both) nodes.
After installing vdsm-gluster the storage works fine again. Thanks a lot!
But there's one strange thing. According to oVirt logs and yum.log GlusterFS storage worked fine until yesterdays "yum update", where the following packages where updated:
Mar 03 10:01:09 Updated: ovirt-hosted-engine-ha-1.1.0-1.el6.noarch Mar 03 10:01:10 Updated: otopi-1.2.0-0.5.rc.el6.noarch Mar 03 10:01:11 Updated: ovirt-engine-sdk-python-3.4.0.6-1.el6.noarch Mar 03 10:01:12 Updated: ovirt-hosted-engine-setup-1.1.0-1.el6.noarch Mar 03 10:01:13 Updated: libtiff-3.9.4-10.el6_5.x86_64
According to yum.log vdsm-gluster was never installed on these hosts, but storage did work.
Shouldn't vdsm-gluster be a requirement for hosts and therefor be installed during host setup?
Do you have any clue why it storage did work until the update of these packages?
vdsm-gluster is to manage gluster storage, not consume gluster storage. did you enable the gluster mode post installing the hosts? (if you enabled it before installing them, or re-install[1] post enabling gluster mode, it should have deployed vdsm-gluster as well)
[1] after moving hos to maintenance.
Yes, I did it after installing the hosts.
Would it be possible to add a check before activating this option?
I'm thinking of the following: - Edit Cluster - Enable Gluster Service - OK checks if all hosts in the cluster have vdsm-gluster installed - if not an error message occurs that this package is required in order to enable the gluster service.
makes sense - please open a bug to track. check would probably warn if unreachable hosts exists to warn user.