qemu-kvm-rhev packaging is being changed as described in BZ 1425820. We expect this to be a compatible change, but we also think this could give the RHV team an opportunity to improve their packaging as well. +++ This bug was initially created as a clone of Bug #1425820 +++ We now have modularization of the block layer in QEMU (bug #1378536) and libvirt (bug #1378540), but the QEMU packaging layout in RHEL might need changes for us to fully benefit from the changes. For example, it should be possible to install QEMU without Gluster or Ceph block drivers (each brings a large number of dependencies). BTW, I have not investigated this, but I have a report from Peter Krempa (libvirt) that "QEMU still can't be installed without gluster". I'm opening this BZ to request further investigation on what can be improved or changed in our packaging layout. +++ This bug was initially created as a clone of Bug #1378536 +++ The reason to modularize the block layer in QEMU is to allow smaller/minimal installations of QEMU, with two primary use-cases: 1. Allow the split of the QEMU package, bringing less dependencies to a minimal installation. This is particularly important for the case of remote storage drivers, such as Ceph and Gluster, which usually link to and require external libraries and tools. 2. Faster startup of QEMU, with less drivers to load. --- Additional comment from Ademar Reis on 2017-03-25 20:16:03 EDT --- Let's leave this for RHEL-7.5, so we have enough time to test it. We'll also need clones of this BZ for RHV and OSP.
we do not use "distribution" component, tracking under vdsm for now. Though perhaps it is an opportunity to move qemu dependencies to ovirt-host? Sandro, what do you think?
(In reply to Michal Skrivanek from comment #1) > we do not use "distribution" component, tracking under vdsm for now. Though > perhaps it is an opportunity to move qemu dependencies to ovirt-host? > Sandro, what do you think? VDSM uses qemu-kvm directly so the requirement should stay in VDSM. In oVirt Host we are requiring vdsm package so you can't depend on it from vdsm side or you'll get a circular dependency.
Re-targeting to 4.3.1 since it is missing a patch, an acked blocker flag, or both
Sandro, is this still relevant at all?
(In reply to Ryan Barry from comment #5) > Sandro, is this still relevant at all? I think that vdsm spec file has been already updated for consuming the change already in 4.3.5 or so. Since 4.3.6 had no issues using qemu-kvm-rhev / qemu-kvm-ev I'm closing this current release in 4.3.6
Sorry, correcting myself, this is for RHEL 8, not RHEL 7.6. So moving to QA for 4.4. vdsm has been already updated to consume virt module and new qemu-kvm build there.
WARN: Bug status (ON_QA) wasn't changed but the folowing should be fixed: [Found non-acked flags: '{}', ] For more info please contact: rhv-devops: Bug status (ON_QA) wasn't changed but the folowing should be fixed: [Found non-acked flags: '{}', ] For more info please contact: rhv-devops
Closing low severity bugs, based on QE capacity, if you would like to still verify this issue please reopen.
*** Bug 1588606 has been marked as a duplicate of this bug. ***