u/s CI job failure on infra issue

[Doron Fediuck]

----- Original Message -----
> From: "Adam Litke" <alitke at redhat.com>
> To: "Dan Kenigsberg" <danken at redhat.com>
> Cc: kobi at redhat.com, "Vered Volansky" <vered at redhat.com>, "infra" <infra at ovirt.org>, "Doron Fediuck"
> <dfediuck at redhat.com>, "Martin Sivak" <msivak at redhat.com>
> Sent: Tuesday, June 17, 2014 5:20:34 PM
> Subject: Re: u/s CI job failure on infra issue
> 
> On 17/06/14 15:14 +0100, Dan Kenigsberg wrote:
> >On Tue, Jun 17, 2014 at 09:19:09AM -0400, Adam Litke wrote:
> >> On 17/06/14 13:35 +0100, Dan Kenigsberg wrote:
> >> >And this one looks like a mom bug:
> >> >   Thread-113::ERROR::2014-06-16
> >> >   15:22:34,115::BindingXMLRPC::1126::vds::(wrapper) unexpected error
> >> >   Traceback (most recent call last):
> >> >     File "/usr/share/vdsm/rpc/BindingXMLRPC.py", line 1110, in wrapper
> >> >       res = f(*args, **kwargs)
> >> >     File "/usr/share/vdsm/rpc/BindingXMLRPC.py", line 455, in getStats
> >> >       return api.getStats()
> >> >     File "/usr/share/vdsm/API.py", line 1268, in getStats
> >> >       stats.update(self._cif.mom.getKsmStats())
> >> >     File "/usr/share/vdsm/momIF.py", line 60, in getKsmStats
> >> >       stats = self._mom.getStatistics()['host']
> >> >     File "/usr/lib/python2.7/site-packages/mom/MOMFuncs.py", line 74, in
> >> >     getStatistics
> >> >       host_stats =
> >> >       self.threads['host_monitor'].interrogate().statistics[-1]
> >> >   AttributeError: 'NoneType' object has no attribute 'statistics'
> >>
> >> This is because 54d16950 was merged in vdsm before mom could support
> >> the named Collectors and Controllers.  Attempting to start MOM with
> >> missing plugins is a fatal error and MOM shuts down.  We can either
> >> revert 54d16950 or wait for the mom part to be merged.  In the later
> >> case, we'll need to ship our own mom rpm with the release since these
> >> changes will not be in the standard Fedora package for awhile.
> >
> >I believe that CPU tuning was deemed a required 3.5 feature, so
> >reverting the patch should be done as a last resort.
> >
> >In any case, 54d16950 should have included a strict dependency of a mom
> >version. What is the version that provides named Collectors? Would this
> >version be available for Jenkins (and humans) to test and install?
> 
> It's not even merged yet.  Kobi is still working on the patches.
> 

Actually CPU parts are merged into vdsm and partially in the engine as well.
Indeed as Dan mentioned this is a gating item which we need to conclude for 3.5
so reverting should be avoided if there's a way to handle it.
There was a patch initially merged to mom and reverted due to vdsm changes.

So I'm asking to stop the reverting sequence and make sure we have all current
patches in place to get cpu working. Martin Sivak is assisting Kobi so we should
have a resolution soon (if not already posted).

> >
> >Regrads,
> >Dan.
> 
> --
> Adam Litke
>