HA - backup mode stop

[theheavy]

Hi folks,

I was wondering if there is any way to perform a 'stop' mode backup at HA controlled VMs or if anyone has a sweet solution for this.

Cannot execute a backup with stop mode on a HA managed and enabled Service. Use snapshot mode or disable the Service.

This is the regular status failed message.

The only way to prevent this as far as i know, would be removing the vm out of the HA configuration, perform the backup and add the vm to the HA config again.
Is there a CLI tool or do I have to use the pvesh for scripting?

Thanks in advance!

 

[fiona]

Hi,
the rationale behind this is that a HA managed enabled service is not allowed to be stopped except by the HA manager itself. It should be possible to get a working script with CLI tools, maybe something like:

1. Request shutdown with HA ha-manager set vm:<ID> --state stopped
2. Loop and wait till the service is stopped by checking the output of either qm status <ID> or ha-manager status
3. Run vzdump
4. Request start with HA ha-manager set vm:<ID> --state started

 

[theheavy]

Hi,
that's totally reasonable from the HA manager perspective.

Thanks a lot for the hints and the ha-manager cli tool definetly helped heaps. I tested your approach but tbh it's a bit too much detailed work to keep the VMID.conf files updated with the hook scripts, checking the loop and so on.

As we have static backup windows from e.g. 04:00 to 05:00ish, I will do something like this:

1. 03:50, remove all the VMs out of the HA group
2. backing up all the vms, configured by the UI
3. 05:00, adding all the VMs to the HA group

I lose 1h of the HA safety net but can still configure backup tasks at one place using the UI.


For step 1. i parse all the needed VMs out of vzdump.cron, this won't work for the "all" setting but that's another days problem.

havms=`grep ^"0 4" /etc/pve/vzdump.cron | awk -F "--" '{print $1}' | awk -F "vzdump" '{printf $2}'`

for i in ${havms}; do

ha-manager remove vm:${i}

done

 

[jieiku]

I noticed that you can use suspend but not stop when an LXC container is configured as HA.

I think this might be a bit short sighted because I can manually right click on that LXC container and select stop and it will do this:

HA 110 - Stop
CT 110 - Stop

at this point I can do a stop mode backup.

then I just have to right click on that same LXC container and select start.

I think when your configuring your scheduled backups and you select stop as your backup mode the script should be able to do what I am able to do manually with a couple clicks, suspend mode backups work but there can possibly be issues with them and stopping the container first ensures a trouble free backup.