can't recover VMs from downed node

  • Thread starter Thread starter Stanislavvv
  • Start date Start date
S

Stanislavvv

Guest
I'm install two nodes with shared storage.
Create VM, start it on master and node, migrate, etc...
One time i'm force poweroff for node with running VM (node hardware/power crash imitation).
After that i can't start VM on master, becouse master knows that VM is on the node and can't get it from here.

How i can force master to forget VMs on crashed node and start it?
 
Trying to migrate:

Code: 
/usr/bin/ssh -t -t -n -o BatchMode=yes 172.30.0.226 /usr/sbin/qmigrate 172.30.0.225 101
ssh: connect to host 172.30.0.226 port 22: No route to host 
VM 101 migration failed - 

500 read timeout

Stop, shutdown and reboot are same.
 
after some time i see in virtual machine list message "Unable to get data for Cluster Node 2" and even can't view downed vm's name.
 
Stanislavvv said:
I'm install two nodes with shared storage.
Create VM, start it on master and node, migrate, etc...
One time i'm force poweroff for node with running VM (node hardware/power crash imitation).
After that i can't start VM on master, becouse master knows that VM is on the node and can't get it from here.

How i can force master to forget VMs on crashed node and start it?
Click to expand...
Hi,
you need the config of the vm on a running node/master. If one node are down, the disks perhaps reachable (shared storage) but your config still missing.
If you backup the config before to the other machines, you only need to move the config to /etc/qemu-server/ and than you can start the VM.
Take a look at the config in /etc/qemu-server (for kvm-VMs).

Udo
 
dietmar said:
There is an obvious network problem - should be easy to track down.
Click to expand...

It is NOT network problem. Node is powered off (or crash).

udo said:
Take a look at the config in /etc/qemu-server (for kvm-VMs).
Click to expand...

Thank you.

Is it true, that no standard method for this recovering?
 
You must log in or register to reply here.
Top Bottom