On 2/27/19 3:07 PM, Diego Michelotto wrote:
GPFS is 'IBM General Parallel File System', also called
'IBM Spectrum Scale', stalls when a process issues a lot
of fsyncs in a short timeframe.
It is better if the VM I/O is not synchronous.
It is racommended to use writeback caching policy.
Using writeback may lead to a data loss if qemu is killed on destination
and underlying FS has not synced the data. Since I don't have GPFS I
can't confirm it's behaving safely on migration. And I wasn't fortunate
finding that on the internet either.
Michal