Hi Sahina, 

Unfortunately my compute nodes and gluster nodes are on same servers. I running 
my VMs on the gluster nodes and vica-versa. 
Also, my ovirt portal is running on the first node, that is an gluster and 
compute node too. 

I have a gluster based installation with three servers, two for replica&VMs, 
one for VMs-s only. 

This is a productive environment, so I'm affraid so much of upgrading...:(.I 
don't want glusterfs mistakes, split-brains, etc..) 

Also, I could make little downtime for upgrading, but I'm need to know what is 
the trustable upgrade procedure. Is there an official upgrade documentation for 
ovirt, that is contain the glusterfs upgrde methods? 

My cluster level is 3.5. 

Thanks a lot. 

Tibor 

----- Eredeti üzenet -----

> On 02/01/2015 04:40 PM, Demeter Tibor wrote:

> > Hi,
> 

> > Is glusterfs 3.6 is neccessary for ovirt 3.5.1?
> 
> > I have 3.5.2 at this moment.
> 

> > 3.6 neccessary for upgade ?
> 

> If you only want to upgrade the engine to oVirt 3.5.1 - you don't need to
> upgrade the gluster nodes. I understand that your gluster nodes are separate
> from the virt nodes. That is, you do not have any cluster enabled with both
> virt + gluster service.

> glusterfs 3.6 is not necessary to upgrade engine to oVirt 3.5.1. It would
> only be required if you were upgrading the cluster level (the compatibility
> version) which in turn would require newer vdsm versions and gluster
> versions. But in your case, I think you already have your cluster running
> with glusterfs 3.5.2. What is the cluster compatibility version?

> > I'm worried what will happed with my gluster volumes after the upgrade...
> > :(
> 

> > Plese help me.
> 

> > Also, how can I upgrade the nodes? Need i'm to a "yum upgrade" on the
> > compute
> > nodes?
> 

> By upgrade the nodes - you mean only the compute nodes? yum upgrade on each
> node should work.
> You can also move the Host to maintenance mode and click on Reinstall or
> upgrade from engine.

> > Thanks.
> 

> > Tibor
> 

> > ----- Eredeti üzenet -----
> 

> > > Hi,
> > 
> 

> > > No, I have only one replica beetwen two nodes.
> > 
> 
> > > The third node is only a host for vms and that using this replica.
> > 
> 

> > > Thanks
> > 
> 
> > > Tibor
> > 
> 

> > > ----- Eredeti üzenet -----
> > 
> 

> > > > On 01/22/2015 08:03 PM, Demeter Tibor wrote:
> > > 
> > 
> 

> > > > > Hello,
> > > > 
> > > 
> > 
> 

> > > > > I have an ovirt 3.5.0 cluster with three nodes and we using glusterfs
> > > > > for
> > > > > serving backend storage for VM-s. Glusterfs are on same servers with
> > > > > ovirt.
> > > > 
> > > 
> > 
> 
> > > > > We have Gluster 3.5.1 on all of nodes.
> > > > 
> > > 
> > 
> 

> > > > > How can I upgrade to ovirt 3.5.1 with new glusterfs? I don't want
> > > > > data
> > > > > inconsistency/data loss/split brains?
> > > > 
> > > 
> > 
> 

> > > > Do you have a replica 3 volume setup across these 3 nodes?
> > > 
> > 
> 

> > > > Glusterfs supports rolling upgrade for replica volumes.
> > > 
> > 
> 
> > > > But there are some issues to upgrade from 3.5 to 3.6 versions
> > > > (depending
> > > > on
> > > > version used)
> > > 
> > 
> 
> > > > Adding gluster-users & Pranith for recommended procedures.
> > > 
> > 
> 

> > > > thanks
> > > 
> > 
> 
> > > > sahina
> > > 
> > 
> 

> > > > > Need I update gluster version to the latest?
> > > > 
> > > 
> > 
> 
> > > > > How I need to upgrade the nodes ? Is there any whitepaper about this?
> > > > 
> > > 
> > 
> 

> > > > > Thanks in advance,
> > > > 
> > > 
> > 
> 

> > > > > Tibor
> > > > 
> > > 
> > 
> 

> > > > > _______________________________________________
> > > > 
> > > 
> > 
> 
> > > > > Users mailing list Users@ovirt.org
> > > > > http://lists.ovirt.org/mailman/listinfo/users
> > > > 
> > > 
> > 
> 

> > > _______________________________________________
> > 
> 
> > > Users mailing list
> > 
> 
> > > Users@ovirt.org
> > 
> 
> > > http://lists.ovirt.org/mailman/listinfo/users
> > 
> 
_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users

Reply via email to