Created attachment 1032854 [details]
engine log from 3.5
Description of problem:
Install VDSM on PPC64LE machine, add it as host to engine (3.5, 3.6).
In engine choose ppc64 arch.
There is an error:
Host ibm8 moved to Non-Operational state as host CPU type is not supported in this cluster compatibility version or is not supported at all
Please make possible support of ppc64le hosts.
the filtering by architecture in vdsm needs to be changed similarly as for x86 vs x86_64 when parsing libvirt's cpu_map.xml
note you currently have to change the list of models in engine db to power8e instead of power8.
Not yet clear if we need to have both or just one of them
VDSM/libvirt identifies the CPU as power8e. We believe it's libvirt bug and won't fix it at the moment. Engine expects "power8"
WA for engine: use engine-config to get & set the value of 'ServerCPUList' from model_power8 to model_power8e
is it possible for QE to apply this WA to automation in order to test ppc64le?
We might be able to workaround, but we must make sure we have a libvirt BZ open and an ETA for a fix.
Michal, do we have such a BZ yet?
(In reply to Michal Skrivanek from comment #4)
> VDSM/libvirt identifies the CPU as power8e. We believe it's libvirt bug and
> won't fix it at the moment. Engine expects "power8"
> WA for engine: use engine-config to get & set the value of 'ServerCPUList'
> from model_power8 to model_power8e
We're getting this fix in as well (decided to treat both CPUs separately)
michal, i see both patches are merged, that means the bug should be on modified or we're waiting for another patch?
we need an engine build to run it right?
it means we'll need to also provide to QE rhevm 3.6.0 build early in process.
can this fix backported to 3.5? if we plan to support ppc64le host from 3.5 engine
there is still a possible change to a solution, pending libvirt's updates
# virsh -c qemu:///system 'hostname ; capabilities' | egrep "^i|model>p"
added find into engine with power8e cluster setup. adding the same host into power8 (BE) cluster failed as expected (different cpu model).
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.