On Mon, May 18, 2020 at 1:33 PM Anton Louw via Users <users@ovirt.org> wrote:


Hello,

 

I am not sure if the below thread went missing somewhere. Just to add on as well, when moving the hosted engine to a standalone host, will the process look similar to the below:


I never tried that, but:
 

 

Backup and remove Hosted Engine:

 

1. Backup Hosted Engine (engine-backup --scope=all --mode=backup --file=Full --log=Log_Full)

2. Download the backup files from HE using WinSCP

3. Enable global Maintenance (hosted-engine --set-maintenance --mode=global)

4. Power down hosted engine (hosted-engine --vm-shutdown)

 

---------------------------------------------------------------------------------------------

Redeploy Hosted Engine:

 

Build new CentOS VM in another environment

*Same IP

*Same name

*Same resources

 

1. sudo yum install https://resources.ovirt.org/pub/yum-repo/ovirt-release43.rpm

2. sudo yum install -y ovirt-engine

3. Copy backup files to newly deployed CentOS VM

4. engine-backup --mode=restore --file=Full --log=Log_Full --provision-db --provision-dwh-db --restore-permissions


You might want to pass also --he-remove-storage-vm , and perhaps also --he-remove-hosts . This removes them, obviously, only from the database, does not connect to them or do anything on them.

You'll then have to add them again. You will not be able to do that if there are running VMs on them, I think.

So:

If you need to keep the VMs up, and have some of them on the hosted-engine hosts, do not use these options, but do test very well, and study the bugs for which these options were added - you can check git log for this file searching this:

https://github.com/oVirt/ovirt-engine/commits/master?after=3cd2766ae1576b22d6ba7caf082b8122d64725c1+34&path%5B%5D=packaging&path%5B%5D=bin&path%5B%5D=engine-backup.sh.in

https://github.com/oVirt/ovirt-engine/commit/542e4a318584c8601159b4bd6d576f1bed6ffe90#diff-5943b645f9ab69667611177f4bfd03ba
Bug-Url: https://bugzilla.redhat.com/1240466
Bug-Url: https://bugzilla.redhat.com/1235200
The engine will still think that it's a hosted-engine, so it will likely be confused. Perhaps --he-remove-storage-vm is enough for that.

This is likely to be problematic, see above bugs.

If you don't need to keep the VMs up, it's probably safer to just take all of them down, pass also --he-remove-hosts, and then, in the new engine, add back the hosts. Or even:

Install a new engine, do not restore the backup, add the hosts, and import the storage domains.

 

5. after retore has completed, run engine-setup

 

Or are there any additional steps I need to take? I have tried this in my  one lab environment, and it works without any issues, however when trying this on my second lab environment, everything shows as “down”, ie. Hosts, Data Center and Storage Domains.


No idea why they are down. Check engine.log. How long did you wait? Did the engine have access to the hosts? Can you manually ssh to them from the engine machine?

Best regards,
--
Didi