Hung VM on VMWare stops creating new VM's with OpenNebula

Community Support
1

Hello all,

We are using Opennebula 4.10.2 in combination with a VCenter/VMWare ESXi 5.1 cluster.

When we tried to restore a VM created in OpenNebula with Veeam, somehow the VM ended in limbo, not only in OpenNebula but also in VMWare. Probably due to a hung of failed datastore migration. It took some effort to remove this hung machine on the VMWare platform.

Now the real problem: during the time that the hung VM existed newly created VM’s in OpenNebula were not created in VMWare. Only after the errorneous VM in VMWare was deleted new VM’s could be created again.

Also the VM’s that were created during the time the errorneous VM existed were never created on VMWare. We ended up deleting those VM’s in OpenNebula and inform the users.

So it seems that one “hung” VM on the VMWare platform (originally created by OpenNebula) can stop the creation of new VM’s with OpenNebula.

Question: what can be done to prevent this ? Is this by design ? Or caused due to having specified only one single thread (-t 1) for VM_MAD ?

Kind Regards,
Rolf

2

Hi,

One possible explanation for this can be that the Veeam created VM breaks somehow the monitoring from OpenNebula, it would be helpful to take a look at the /var/log/one/oned.log file to ascertain if this is indeed the case.

When the monitoring breaks, the host goes into error state and no new VMs are deployed to it.