CentOS 7 Hyperconverged oVirt 4.2 with Self-Hosted-Engine with glusterfs with 2 Hypervisors and 1 glusterfs-Arbiter-only
Hi, I'm trying to install oVirt 4.2 as 2-Node Hyperconverged System based on glusterfs. A third node should be used as glusterfs arbiter Node and to provide Quorum for the Cluster. The third node is a small PCEngines APU2 Host, so it is not usable as Hypervisor. My question is: Is this kind of setup possible? What is the best way to install a cluster like this one? Thanks, -- : Philipp Richter : LINFORGE | Peace of mind for your IT : : T: +43 1 890 79 99 : E: philipp.richter@linforge.com : https://www.xing.com/profile/Philipp_Richter15 : https://www.linkedin.com/in/philipp-richter : : LINFORGE Technologies GmbH : Brehmstraße 10 : 1110 Wien : Österreich : : Firmenbuchnummer: FN 216034y : USt.- Nummer : ATU53054901 : Gerichtsstand: Wien : : LINFORGE® is a registered trademark of LINFORGE, Austria.
Hi, this should work according to my gluster colleagues. The recommended way to install this would be by using one of the "full" nodes and deploying hosted engine via cockpit there. The gdeploy plugin in cockpit should allow you to configure the arbiter node. The documentation for deploying RHHI (hyper converged RH product) is here: https://access.redhat.com/documentation/en-us/red_hat_hyperconverged_infrast... Best regards Martin Sivak On Mon, Feb 12, 2018 at 4:25 PM, Philipp Richter <philipp.richter@linforge.com> wrote:
Hi,
I'm trying to install oVirt 4.2 as 2-Node Hyperconverged System based on glusterfs. A third node should be used as glusterfs arbiter Node and to provide Quorum for the Cluster. The third node is a small PCEngines APU2 Host, so it is not usable as Hypervisor.
My question is: Is this kind of setup possible? What is the best way to install a cluster like this one?
Thanks, --
: Philipp Richter : LINFORGE | Peace of mind for your IT : : T: +43 1 890 79 99 : E: philipp.richter@linforge.com : https://www.xing.com/profile/Philipp_Richter15 : https://www.linkedin.com/in/philipp-richter : : LINFORGE Technologies GmbH : Brehmstraße 10 : 1110 Wien : Österreich : : Firmenbuchnummer: FN 216034y : USt.- Nummer : ATU53054901 : Gerichtsstand: Wien : : LINFORGE® is a registered trademark of LINFORGE, Austria. _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
Hi,
The recommended way to install this would be by using one of the "full" nodes and deploying hosted engine via cockpit there. The gdeploy plugin in cockpit should allow you to configure the arbiter node.
The documentation for deploying RHHI (hyper converged RH product) is here: https://access.redhat.com/documentation/en-us/red_hat_hyperconverged_infrast...
Thanks for the documentation pointer about RHHI. I was able to successfully setup all three Nodes. I had to edit the final gdeploy File, as the Installer reserves 20GB per arbiter volume and I don't have that much space available for this POC. The problem now is that I don't see the third node i.e. in the Storage / Volumes / Bricks view, and I get warning messages every few seconds into the /var/log/ovirt-engine/engine.log like: 2018-02-13 15:40:26,188+01 WARN [org.ovirt.engine.core.vdsbroker.gluster.GlusterVolumesListReturn] (DefaultQuartzScheduler3) [5a8c68e2] Could not add brick 'ovirtpoc03-storage:/gluster_bricks/engine/engine' to volume '2e7a0ac3-3a74-40ba-81ff-d45b2b35aace' - server uuid '0a100f2f-a9ee-4711-b997-b674ee61f539' not found in cluster 'cab4ba5c-10ba-11e8-aed5-00163e6a7af9' 2018-02-13 15:40:26,193+01 WARN [org.ovirt.engine.core.vdsbroker.gluster.GlusterVolumesListReturn] (DefaultQuartzScheduler3) [5a8c68e2] Could not add brick 'ovirtpoc03-storage:/gluster_bricks/vmstore/vmstore' to volume '5a356223-8774-4944-9a95-3962a3c657e4' - server uuid '0a100f2f-a9ee-4711-b997-b674ee61f539' not found in cluster 'cab4ba5c-10ba-11e8-aed5-00163e6a7af9' Of course I cannot add the third node as normal oVirt Host as it is slow, has only minimal amount of RAM and the CPU (AMD) is different than that one of the two "real" Hypervisors (Intel). Is there a way to add the third Node only for gluster management, not as Hypervisor? Or is there any other method to at least quieten the log? thanks, -- : Philipp Richter : LINFORGE | Peace of mind for your IT : : T: +43 1 890 79 99 : E: philipp.richter@linforge.com : https://www.xing.com/profile/Philipp_Richter15 : https://www.linkedin.com/in/philipp-richter : : LINFORGE Technologies GmbH : Brehmstraße 10 : 1110 Wien : Österreich : : Firmenbuchnummer: FN 216034y : USt.- Nummer : ATU53054901 : Gerichtsstand: Wien : : LINFORGE® is a registered trademark of LINFORGE, Austria.
On Tue, Feb 13, 2018 at 3:57 PM, Philipp Richter < philipp.richter@linforge.com> wrote:
Hi,
The recommended way to install this would be by using one of the "full" nodes and deploying hosted engine via cockpit there. The gdeploy plugin in cockpit should allow you to configure the arbiter node.
The documentation for deploying RHHI (hyper converged RH product) is here: https://access.redhat.com/documentation/en-us/red_hat_ hyperconverged_infrastructure/1.1/html-single/deploying_red_ hat_hyperconverged_infrastructure/index#deploy
Thanks for the documentation pointer about RHHI. I was able to successfully setup all three Nodes. I had to edit the final gdeploy File, as the Installer reserves 20GB per arbiter volume and I don't have that much space available for this POC.
The problem now is that I don't see the third node i.e. in the Storage / Volumes / Bricks view, and I get warning messages every few seconds into the /var/log/ovirt-engine/engine.log like:
2018-02-13 15:40:26,188+01 WARN [org.ovirt.engine.core.vdsbroker.gluster.GlusterVolumesListReturn] (DefaultQuartzScheduler3) [5a8c68e2] Could not add brick 'ovirtpoc03-storage:/gluster_bricks/engine/engine' to volume '2e7a0ac3-3a74-40ba-81ff-d45b2b35aace' - server uuid '0a100f2f-a9ee-4711-b997-b674ee61f539' not found in cluster 'cab4ba5c-10ba-11e8-aed5-00163e6a7af9' 2018-02-13 15:40:26,193+01 WARN [org.ovirt.engine.core.vdsbroker.gluster.GlusterVolumesListReturn] (DefaultQuartzScheduler3) [5a8c68e2] Could not add brick 'ovirtpoc03-storage:/gluster_bricks/vmstore/vmstore' to volume '5a356223-8774-4944-9a95-3962a3c657e4' - server uuid '0a100f2f-a9ee-4711-b997-b674ee61f539' not found in cluster 'cab4ba5c-10ba-11e8-aed5-00163e6a7af9'
Of course I cannot add the third node as normal oVirt Host as it is slow, has only minimal amount of RAM and the CPU (AMD) is different than that one of the two "real" Hypervisors (Intel).
Is there a way to add the third Node only for gluster management, not as Hypervisor? Or is there any other method to at least quieten the log?
Adding Sahina here.
thanks, --
: Philipp Richter : LINFORGE | Peace of mind for your IT : : T: +43 1 890 79 99 : E: philipp.richter@linforge.com : https://www.xing.com/profile/Philipp_Richter15 : https://www.linkedin.com/in/philipp-richter : : LINFORGE Technologies GmbH : Brehmstraße 10 : 1110 Wien : Österreich : : Firmenbuchnummer: FN 216034y : USt.- Nummer : ATU53054901 : Gerichtsstand: Wien : : LINFORGE® is a registered trademark of LINFORGE, Austria. _______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
participants (3)
-
Martin Sivak -
Philipp Richter -
Simone Tiraboschi