-
Hi, as I understand it, it's not possible to failback a VM that has failed over to another host without replicating the entire dataset (which is especially an issue if the dataset is large). Is this correct?
If so, is there a plan to implement this at some point? This is an extremely useful feature and given that other hypervisors/software already have this would big make a difference to being a real alternative.
Thanks
-
Hi,
I'm not sure to understand exactly what you mean.
- There's HA on the pool: if you have a shared storage, you can enable HA on your pool (cluster) so if a host is down, all its VMs will be restarted on the other members of that pool
- If you don't have a shared storage, you can use Xen Orchestra replication to… replicate to another storage/pool. See https://xcp-ng.org/blog/2022/07/26/vm-replication-in-xcp-ng/
-
@olivierlambert sorry I should have clarified, this is explicitly when not using shared storage, i.e. VMs on Host A are being replicated to Host B.
In either a planned or unplanned failover, I can switch everything to Host B, but when and if Host A comes back and I want things to start again on A (i.e. the performance is better on there), then it sounds like I'll need to wait for them to replicate from scratch again instead of just waiting for the changes (since the failover) to sync back to Host A (assuming of course there wasn't an issue with the storage itself etc.)
Is the above possible?
-
Not yet. But this will come in Xen Orchestra. For now, replication is "one way". There's no big technical barriers to achieve that, except complexity of configure it from a user perspective.
-
@olivierlambert Thanks - do you have any kind of timeline for that?
-
Not yet, it's not top priority until we got our new backup code ready. After that, it might be easier than before
-
O olivierlambert moved this topic from Feedback and requests on