Not that I am aware of but someone might know a way.
The idea is .... if the VM server reboots.... there is a problem that needs to be corrected. If the server reboots on a clustered pool and you're using HA... then you have the potential of putting a bad server back into the pool of resources. If you're going to purposefully reboot a server, then you will have probably already migrated the VM guests to another server.
I am thinking about a kernel panic or some sporadic network problem which causes the OVM server to restart, in that case it is perfectly correct restart vm automatically no?
AFAIK Xen (which Oracle VM based on) has this ability to restart vm guests when dom0 restarts
Might be a nice to have as an option. Wouldn't necessarily want it to do such be default.
Either way, there would need be some type of mechanism to verify the proper availability of resources. I've not been using Oracle VM for years but I've only had one situation where a server panic caused a reboot. I'm glad it didn't automatically restart my VM Guests because I had ran out of pool storage. It would of produced a viscous cycle of restarts. I can see the file corruption spiraling now.... I ended up being able to fsck my ocfs filesystems and the VM guests. Everything was fine. Might not have been that way if there had been a dozen reboots taken place before was able to stop it.
Xen surely does that, but I tend to also think that a VM server shouldn't auto-start guests - at least if these guest hold any valuable data that cannot be restored easily.
It really depends on how you are using your OVM (cluster). OVM is a linux system, so there are of course ways to implement some auto-start functionality, but I'd only use it on single-hosts pools.
If you're at least an average Linux admin, you'll know how to accomplish this on your own - if not, you probably shouldn't look into this anyway, since whatever solution you come up with - it won't be supported by Oracle.
Perhaps I am misunderstanding what you are looking for, but if you make the VM "High Availability" then it will auto-restart if your vm server crashes or migrate to another VM server in your cluster.
The downside of that is if you shutdown the VM from inside the VM, the OVM manager will restart it until you un-set it as "High Availability".
If you only want the vm to restart in the node where it was running (dom0 restarts) you can set HA in the VM and use anti-affinity groups (you need to combine group from different OVM Servers in the pool) to achieve what you want
As many other posts point out, in a clustered environment this may be a bad idea. Also, the HA settings can make OVMM start the vm on other hosts.
But, to answer your question, put a symbolic link to the vm.cfg file in the folder /etc/xen/auto. Any VMs in the folder are automatically started at boot time. That's a good way to start the manager if you are in a one-host environment and have virtualized OVMM.
My repository UUID is 00abc00.
My firewall VM UUID is 00def00.
My webserver VM UUID is 00cba00.
To autostart both, create the links:
ln -s /OVS/Repositories/00abc00/VirtualMachines/00def00/vm.cfg /etc/xen/auto/firewall
ln -s /OVS/Repositories/00abc00/VirtualMachines/00cba00/vm.cfg /etc/xen/auto/webserver
you can add a line to the /etc/inittab of the OVM server
we use :
8:3:once:/usr/sbin/xm create /OVS/Repositories/0004fb000003000028bea91df1332359/VirtualMachines/0004fb0000060000292bfdf47a61bee1/vm.cfg
9:3:once:/usr/sbin/xm create /OVS/Repositories/0004fb000003000028bea91df1332359/VirtualMachines/0004fb00000600007ad64af712addb0f/vm.cfg
to start two VM's after a reboot/kernel panic of the XEN server.
8 and 9 are the order.
3 is the init
the bold is the subfolder for the VM
libxml2.x86_64 2.6.26-126.96.36.199.0.2 installed libxml2-python.x86_64 2.6.26-188.8.131.52.0.2 installed