On 24. 3. 2022, at 18:38, Nenhum de Nos via Users <users@ovirt.org> wrote:

Hi,

I run a small test node and engine of ovirt since 4.3. Its in my home, but I kind messed up and now I can't use or reinstall it without loosing all my vm's data (at least this is the point my knowledge of ovirt is).

I got the issue where I could not update yum since CentOS 8 mirrors would not answer. And I tried some solution I found in the list archives, and got really bad here.

Hi,

I had a power outage last Saturday and my node got to nonresponsive. No matter what I did, never left this state. Kept complaining about the CPU Type, I run it on Ryzen 3400G (used to run on Intel i5, also desktop CPU, and never had an issue, when I changed to Ryzen, every outage from power here is a problem), and the node wouldn't come back. So I tried to upgrade my cluster to 4.4 level (was 4.3) and it didn't help. So I tried to get back to 4.3 and a message said just empty clusters could do that. I then removed my ryzen node from it and it got back to 4.3.
But now I could not install the node. I tried changing its name, no good. Failed on how to get packages from GLuster8 repo.
That is where my story got sad. I tried to update my repo using the steps in the last post from https://lists.ovirt.org/archives/list/users@ovirt.org/thread/D3FEXQ5KOLN3SST3MPGIHSCEF52IBTKY/. Bad idea, and another one was I didn't do a backup of the engine.

Now my engine won't start, the backup won't run:

engine-backup --mode=backup --file=/root/backup_deuruim --log=/root/backup_deuruim.log
Start of engine-backup with mode 'backup'
scope: all
archive file: /root/backup_deuruim
log file: /root/backup_deuruim.log
Backing up:
Notifying engine
Notifying engine
FATAL: Failed notifying engine

pg_dump: error: Dumping the contents of table "audit_log" failed: PQgetResult() failed.
pg_dump: error: Error message from server: ERROR: MultiXactId 1076887572 has not been created yet -- apparent wraparound
pg_dump: error: The command was: COPY public.audit_log (audit_log_id, user_id, user_name, vm_id, vm_name, vm_template_id, vm_template_name, vds_id, vds_name, log_time, log_type_name, log_type, severity, message, processed, storage_pool_id, storage_pool_name, storage_domain_id, storage_domain_name, cluster_id, cluster_name, correlation_id, job_id, quota_id, quota_name, gluster_volume_id, gluster_volume_name, origin, custom_event_id, event_flood_in_sec, custom_data, deleted, call_stack, brick_id, brick_path, custom_id) TO stdout;
2022-03-24 13:16:10 3461: FATAL: Database engine backup failed

the Admin portal won't work, and I get a message on the login screen from engine saying I have a web console at port 9090. I don't get it.

I don't get it either, it all sounds too confusing.

As my last resort I am here, trying at least to know a way I could backup it and install the engine elsewhere. All vm data is fine, but as the names are cryptographic, import them would have me guessing things, right?

The number of VM's is low, 6 or 7, so if there is a way to do it manually, I would try. Installing it again would be my nightmare here.

with just a few VMs it's indeed best to just reinstall from scratch and importing the storage domain back. You would only be missing the other non-VM configuration in oVirt that you have to do again, but the VMs should have the right name and all after import. Assuming you still have the storage domain somewhere safe.