Hello,

on a test oVirt environment I have export as ova task that, based on same disk size (20Gb) it could last 35 seconds or 11 minutes....

During the long lasting test I see on host that almost all and only I/O (in the measure of 6-7 MB/s) is done by a [loop0] process that has ppid 2 and apparently (with "fuser" command on /dev/loop0 file) is uses by noone.

Is it implicitly used by the "qemu-img convert " command of the export as ova?

Why so great difference in elapsed time? The disk is the same of the same VM.

The oVirt env has almost nothing running only 2 CentOS 8.1 VMs doing nothing.

Same host used for export as ova in the two runs in a 20 minutes timeframe. As the path for OVA I'm temporarily using a connected export domain subdirectory, using

directory='/rhev/data-center/mnt/10.4.192.69:_export_ovirt/dump'

Donna if it can have a role in the problem...

Thanks,

Gianluca