Re: [libvirt] [PATCH] storage: Don't hold storage pool locked during wipeVol

On 08/13/2018 06:49 AM, Michal Privoznik wrote: Why not be more similar to storageVolCreateXMLFrom? That is handle the in_use incr/decr at the storage driver level. Seems we could create a helper that would follow the same steps... For volume wiping, RBD and iscsi-direct use the @pool (obj), but by incrementing not only vol->in_use, but the pool asyncjobs we can inhibit the undefine, destroy, or deletion of the pool that would cause issues for those uses AFAICT. Inhibiting refresh during these operations is a matter of perhaps opinion as to whether it's a good idea or not - although I suppose if a volume is open for write (locked), trying to open/read to get stat type information probably is going to wait anyway (although I'll admit I haven't put much time or research into that thought - just going by gut feel ;-)). BTW: Wouldn't uploadVol have the same issues? Seems we have only cared about build vol from. Since uploadVol checks vol->in_use it seems logical using the same argument as above that it too should use some new helper to change pool asyncjobs and vol in_use. The building setting could just be another parameter. John <sigh> Should be BackendISCSI instead of BackenISCSI and I see this whole new module used single blank line spacing between functions instead of 2 blank lines. Both would be trivial patches IMO. [...]