VM migration with ZFS disk causes VM's on destination node to slow down - SOLVED

[birdflewza]

Hi folks,

We have a bit of a strange issue. We have VM's on each of our nodes with their disks on ZFS pools. We have named the ZFS pools the same name, so we can perform VM migration between the nodes with ease. We set the bandwidth limits in the datacenter options to 60 MiB/s.

When we move a VM from node7 to node4, all of the VM's on node4 become unresponsive, and upon logging into node4 with SSH, I notice that the Disk write is maxed at 1 GiB/s which is the max write speed for that node. It seems that the datacenter bandwidth options are not being obeyed. Could anyone perhaps assist with this?

Many thanks

 

[birdflewza]

So it turns out I've solved my own problem. I had the "discard" option turned on for all of my hard disks under the hardware pane. As soon as I turned this off, shutdown the VM, and started the VM back up, transfers of the VM's between the nodes worked seamlessly and the bandwidth limits set in the datacenter options were obeyed as expected.