-
Hi all.
I had a continouos replication task, copying VMs from one xcp-ng server to another.
One night during copying some error occured, and the VM was not fully copied to server #2, so it was in kinda "XO importing..." state. Originannly this VM had "Auto power on" enabled on server #1, so upon copying this feature stayed turned on, and when i restarted the xcp-ng server#2 i coudn't gain access to it, because my pfsense VM was unable to start (lack of memory). Somehow this faulty VM was starting before pfsense VM (even though it shouldn't start at all - it's a cont.replication), taking necessary RAM and preventing my pfsense VM to start.
The only way to solve this was via gaining direct access to the server#2 and turning autostart off/deleting the faulty VM.
I guess the autostart attribute is removed after the VM is successfully copied which in my case didn't happenxcp-ng 8.2, xo-server 5.117.0, xo-web 5.120.0, Xen Orchestra, commit 99147
-
@dave-opc
Maybe it's possible to create a delay for autostart VM parameter, like 0 seconds start instantly, 60 seconds - to delay VM autostart.
I guess this will be useful for those who have slow HDDs and lots of VMs starting all at the same time. -
This happens when CR gives an error (does not complete backup until the end) .
Maybe an option can be activated so that Auto Start is no longer taken into account for a CR. That depends on the XO team.
The solution, for now, is that, before restarting:- delete CR with problems ( XO importing ....)
- disable auto start on CR backup, before restarting server.
Another option is to enter on ssh on the server and stop CR with problems that started.
-
I would have done those operation, if i wouldn't loose connection to xcp-ng.
I got only 1 IP address which is dedicated to pfsense, and from there i connect to xcp-ng (LAN side). -
O olivierlambert moved this topic from Feedback and requests on
