Hi Ken,
----- Original Message -----
> On 12/05/13 09:01, Nathan Scott wrote:
> > Changes committed to git://oss.sgi.com/pcp/pcp.git dev
> > ...
> > Date: Sat May 11 17:44:59 2013 +1000
> >
> > Set a planned release date for early next week, update changelogs
>
> This seems a bit premature to me.
Possibly, and I'd prefer a bit more time too. On the other hand, we
have reached some internal Red Hat milestone points and we're after a
feature release to rebase to by those dates I've set. While this might
not be of interest for community PCP folks, it is very important to us
to have a base to move forward from, and we have set early-next-week as
a deadline target we really must hit.
So, at this point we could do a Red Hat specific release, if that'd
help ease those concerns, but I'd rather go for a normal release and
try to maximise exposure for everyone with the new features and all
the changed code. My current plan is to continue my ongoing testing
efforts through tomorrow, and release on Tuesday with whatever seems
right and ready by then. And then continue onward with an extended
period focussing on more bug-fixing, QA, and just small feature work
(like new tools, Pauls GFS2 work, other client tools that other folk
are working on, and so on) into a minor point release phase.
> pmcollectl and/or qa/709 are hanging QA with pmconfirm and pmquery and
> pmlogger processes left running, so I cannot do auto QA across all hosts.
Hmm, some of those are new to me - I have seen that pmconfirm window
hanging around like a bad smell, but it doesn't hang QA for me. And
I've not ever come across any unwanted loggers at all, FWIW.
So the pmcollectl issues are not good but they don't affect many people
(not even many pmcollectl users) - they seem like a good minor point
release fix candidate and not something that should hold up the next
major release for.
> I am also seeing higher than expected QA failure rates when I can run QA.
>
> I'd like to see a functional code freeze, bug fix only changes and some
> concerted effort to get more QA passing before we commit to a release.
When I've gone through the failures from others (incl. the QA farm list
earlier) they have *not* indicated any deep lurking issues that I'd be
concerned about. For my local testing I'm consistently on two failures
now, which are 024 & 722, which I've thrown over the wall to yourself
and Stan (but both look like test infrastructure failures, and not code
failures).
cheers.
--
Nathan
|