On Thu, Jan 12, 2017 at 19:45:52 -0500, John Ferlan wrote: > Here's the difference between the check and removing the check (not > withstanding the no PARTED and no BLKID available)... > > Create a pool using ext4, start it - life is happy. Destroy the pool. > Define a pool using xfs. Start it > > w/ my change: > > # virsh pool-start fs > error: Failed to start pool fs > error: Storage pool already built: Device '/dev/sde' formatted cannot > overwrite using 'xfs', requires build --overwrite
This error message doesn't make any sense to me. You are starting a pool, but libvirt is complaining that it is already built (well, it must be, otherwise you cannot start it) and suggesting to use --overwrite, which is specific to virsh (API users would be confused) and gives the feeling the command is trying to format the device while it only tries to mount it. In other words, the check makes sense when a user tries to built (i.e., format) the pool, but it is pretty misleading here. > w/o my change (e.g. if the change is reverted): > > # virsh pool-start fs > error: Failed to start pool fs > error: internal error: Child process (/usr/bin/mount -t xfs /dev/sde > /home/vm-images/fs) unexpected exit status 32: mount: wrong fs type, bad > option, bad superblock on /dev/sde, On the other hand, this error is pretty clear. Jirka -- libvir-list mailing list [email protected] https://www.redhat.com/mailman/listinfo/libvir-list
