[ovirt-devel] oVirt 4.2.0 blockers review

[Dan Kenigsberg]

On Tue, Nov 28, 2017 at 10:29 PM, Michal Skrivanek
<michal.skrivanek at redhat.com> wrote:
>
>> On 28 Nov 2017, at 15:17, Dan Kenigsberg <danken at redhat.com> wrote:
>>
>>
>>
>> On Tue, Nov 28, 2017 at 9:54 PM, Michal Skrivanek <michal.skrivanek at redhat.com> wrote:
>>
>>> On 28 Nov 2017, at 06:36, Dan Kenigsberg <danken at redhat.com> wrote:
>>>
>>> On Tue, Nov 28, 2017 at 12:58 PM, Sandro Bonazzola <sbonazzo at redhat.com> wrote:
>>> Hi,
>>> I'm waiting for last blockers to be fixed for starting a 4.2.0 RC build.
>>> Assignee are in the TO list of this email.
>>> So far we are down to 7 bugs: https://bugzilla.redhat.com/buglist.cgi?quicksearch=flag%3Ablocker%2B%20target_milestone%3Aovirt-4.2.0%20status%3Anew%2Cassigned%2Cpost
>>>
>>> Please review them and provide an ETA for the fix. If the bug is marked as blocker by mistake, please remove the blocker flag and / or postpone the bug to a later release.
>>>
>>> Bug ID       Product Assignee        Status  Summary
>>> 1516113      cockpit-ovirt   phbailey at redhat.com     POST    Deploy the HostedEngine failed with the default CPU type
>>> 1509629      ovirt-engine    ahino at redhat.com        ASSIGNED        Cold merge failed to remove all volumes
>>> 1507277      ovirt-engine    eraviv at redhat.com       POST    [RFE][DR] - Vnic Profiles mapping in VMs register from data storage domain should be supported also for templates
>>>
>>> Patches are in initial stage of review. Yaniv Lavi is adamant that this is indeed a 4.2.0 blocker, so it would cause at least a day or two of delay.
>>>
>>> 1506677      ovirt-engine    dchaplyg at redhat.com     POST    Hotplug fail when attaching a disk with cow format on glusterfs
>>> 1488338      ovirt-engine    mlipchuk at redhat.com     NEW     SPM host is not moving to Non-Operational status when blocking its access to storage domain.
>>> 1512534      ovirt-hosted-engine-ha  pkliczew at redhat.com     ASSIGNED        SHE deployment takes too much time and looks like stuck.
>>> 1496719      vdsm    edwardh at redhat.com      POST    Port mirroring is not set after VM migration
>>>
>>> We're trying since morning to verify if this has been fixed as a side effect by Milan, currently blocked by environmental hurdles (storage server; odd SELinux problem in Engine). An answer is still expected today.
>>
>> different fix than the proper hotplug/unplug xml way?
>>
>> No, same one.
>
> ok. it is going to take some time, few days i suppose

Sorry, I misled you. Edy was trying francesco's "virt: metadata:
correctly store portMirroring"

>
>>
>> I believe we can also challenge the blocker status since the impact is on migrations with hotplugged NICs using mirroring
>>
>>
>> Correct. However, we don't even have an answer. We still did not manager to reach a clear verification.
>
> So what I’m saying is to not block on it even if it reproduces, so if it’s not even sure it reproduces we indeed should not block GA on it?
>
>> Last news I've heard was that Engine was sending both xml and conf on start VM. This sounds more disturbing.
>
> that’s unlikely. Who’s saying that and where? conf is not being sent since Nov 8th [1] in RunVM flow.  You still get conf on suspend and resume flow, 4.1 compatibility and such, there were some minor issues here and there, and the other patches in that bug are needed.

I still consider this only a romour, but it requires further scrutiny.