On Thu, Nov 15, 2018 at 01:18:25PM +0100, Wolfgang Bumiller wrote: > pve-storage >= 5.0-31 (on pvetest currently) received new methods to > request a path to a block device node explicitly for storages which can > distinguish between "URLs" (eg. rbd:pool/name) where qemu handles the > connection and protocol and local devices (eg. using the kernel's `krbd` > module). This is currently mostly to avoid the need to have separate > storages for KVM and lxc for eg. ceph. > > There's currently no "default" fallback to just calling path(), therefore > the ABI version was bumped. > > Depending on whether other storage operations need a device available > and mapped into /dev a storage plugin may now choose to defer doing this > from activate_storage() to map_volume(). Otherwise, it *should* be > sufficient to just call path() from map_volume() and do nothing in > unmap_volume() for many storages. > > https://git.proxmox.com/?p=pve-storage.git;a=commit;h=40d698932e27ee94b5fa06330dbf07bc5d2815cc
Hi Wolfgang, thanks for the detailed description, makes sense, we will adapt the plugin accordingly. May I ask how packages from your "pvetest" transition into stable? Will this hit us very soon, or on the next major release (PVE 6.0). Regards, rck _______________________________________________ drbd-user mailing list [email protected] http://lists.linbit.com/mailman/listinfo/drbd-user
