+++ This bug was initially created as a clone of Bug #1657713 +++ Description of problem: oVirt imageio internally uses NBD. However from virt-v2v and v2v warm conversions we still have to use the HTTPS API which is slow. Using NBD natively makes a lot more sense, and so I think we should expose the NBD socket to clients, at least locally. For full details see this email: https://www.redhat.com/archives/libguestfs/2018-December/msg00111.html See also: https://bugzilla.redhat.com/show_bug.cgi?id=1657713
This bug didn't get any attention for a while, we didn't have the capacity to make any progress. If you deeply care about it or want to work on it please assign/target accordingly
This is important for virt-v2v, and we expect that backup vendors will be interested in this for performance reasons. We already have issues with some backup vendors not using ovirt-imageio APIs because of performance issues. Having access to local NBD socket will allow the vendor to perform backups using qemu-img. Supporting this requires small changes in engine side, described in https://www.redhat.com/archives/libguestfs/2018-December/msg00111.html. More work is required to expose the NBD socket remotely. This requires adding TLS-PSK support in vdsm to configure qemu-nbd and qemu. Targeting to 4.4.z for now.
Proposing for 4.4.2. This needed for virt-v2v. May speed up virt-v2v imports make the process more robust by simplifying the stack.
Fixed dependency, this bug blocks the virt-v2v bug.
Fixing component, this must be implemented in engine.
Continuing Nir's comment (comment 8) this bug would be verified by scale team.
We are past 4.5.0 feature freeze, please re-target.
Since this might be useful for migration from RHV to CNV (part of MTV) we may want to do it although we didn't get reports on that so far Moving to ovirt-imageio on GitHub: https://github.com/oVirt/ovirt-imageio/issues/70