[ovirt-users] oVirt-Gluster Hyperconvergence: Graceful shutdown and startup

[Bernhard Seidl]

Am 04.01.2018 um 11:56 schrieb Sahina Bose:
> 
> 
> On Wed, Jan 3, 2018 at 9:27 PM, Bernhard Seidl <info at bernhard-seidl.de
> <mailto:info at bernhard-seidl.de>> wrote:
> 
>     Hi all and a happy new year,
> 
>     I am just testing oVirt 4.2 using a three node gluster hyperconvergence
>     and self hosted engine setup. How should this setup be shutdown and
>     started again? Here is what tried an experienced:
> 
> 
> The approach you outlined is correct.
> 
> 
>     Shutdown:
> 
>     1. Shutdown all VMs
> 
>     2. Enable global ha maintenance
> 
>     3. Wait for all VM
> 
>     4. Shutdown hosted-engine using "hosted-engine --vm-shutdown"
> 
>     5. Wait for stopped hosted engine
> 
>     6. Shutdown all nodes
> 
>     Startup:
> 
>     1. Switch on all nodes
> 
>     2. Start glusterd on all nodes since it does not start by default (is
>     this a bug?) using "systemctl start glusterd*"*
> 
> 
> Can you log a bug?

There it is: https://bugzilla.redhat.com/show_bug.cgi?id=1531052

> 
> 
>     3. Check volume status using "gluster peer status" and "gluster volume
>     status all" on one of the nodes
> 
>     4. Wait for ovrt-ha-agent until "hosted-engine --vm-status" does not
>     fail anymore printing "The hosted engine configuration has not been
>     retrieved from shared storage. Please ensure that ovirt-ha-agent is
>     running and the storage server is reachable."
> 
>     4. Start hosted engine "hosted-engine --vm-start" on one of the nodes
> 
>     5. Check status using "hosted-engine --vm-status" and wait until health
>     reports good
> 
>     6. Wait until a host got the SPM role
> 

I think that I missed a step here:
6a. Disable global ha maintenance


>     7. Start VMs
> 
>     Is this the best/correct way?
> 
>     Kind regards,
>     Bernhard