[OpenAFS-devel] release-team meeting minutes 2014-04-23

[Stephan Wiesand]

Log: =
http://conference.openafs.org/release-team@conference.openafs.org/2014-04-=
23.txt

Participants:

* Andrew Deason
* Ben Kaduk
* Daria Brashear
* Mike Meffie
* Simon Wilkinson
* Stephan Wiesand

=3D=3D Linux news =3D=3D

3.15-rc2 was released. We don't know yet whether we're still good. =
[Update: we are - thanks Marc]

=3D=3D Problem reports =3D=3D

Linux stack overflows (RT #131831):

=
http://gerrit.openafs.org/#q,project:openafs+branch:master+topic:stack-red=
uction,n,z are making good progress, but are still being worked on on =
the master branch. No blocker for 1.6.8pre2.


Fileserver crash in libkrb5 (RT #131852):

Seems to affect at least not all Heimdal versions, and it's unknown =
which ones. Gerrit 11075 was discussed and updated during the meeting. =
We won't block 1.6.8pre2 on it. If evidence comes up that the problem is =
common, we can issue a pre3. [Update: probably more related to platform, =
environment and Heimdal build rather than Heimdal version. See the RT =
ticket for details]


Transfers stalling for files > 1328 bytes:

See =
https://lists.openafs.org/pipermail/openafs-info/2014-April/040643.html =
. Looked like an MTU problem, but that seems uncertain now. Single =
report (against 1.6.1), not enough data yet. No blocker for 1.6.8pre2.


Reading unixusers from /proc on Linux broken (gerrit 11093):

Problem introduced a while ago, breaks a debugging aid. Deferred to =
1.6.9.


cmd.c possible buffer overflow reported on -info:

See =
https://lists.openafs.org/pipermail/openafs-info/2014-April/040649.html =
. This is already fixed in 1.6.8pre1 by gerrit 10731, and not believed =
to have a real security impact.

=3D=3D Pulling up fixes from master =3D=3D

The cmd.c issue was fixed on master by gerrit 7092 (which is probably =
not applicable to 1.6.x), which is part of a series of clang related =
fixes (see =
http://gerrit.openafs.org/#q,status:merged+project:openafs+branch:master+t=
opic:clang,n,z ). Stephan brings up the question whether it makes sense =
to generally pull up such fixes to 1.6.x (as done in =
http://gerrit.openafs.org/#q,project:openafs+branch:openafs-stable-1_6_x+t=
opic:coverity,n,z ). There seems to be limited enthusiasm for this, but =
it's probably the right thing to do because it reduces the risk of =
missing an important fix, reduces the skew between master and stable, =
and makes it less likely that a new compiler on some platform requires =
extra work later.

=3D=3D 1.6.8 release =3D=3D

Preliminary pre2 tarballs were uploaded during the meeting. After some =
positive results from smoke testing by release-team, we'll tag and =
announce. [Update: done] If it turns out that we have a problem with =
Linux 3.15, or anything else urgent comes up, we can issue a pre3. =
Otherwise, this is likely to be the final prerelease.

=3D=3D Testing =3D=3D

Some 10 sites indicated to Andrew that they're willing to run =
prereleases in a test environment or on a test server in their =
production environment. That's just what we're asking the community to =
do every time we (pre-)release, but individual contacts may just work =
better. Andrew will ask those sites to test 1.6.8pre2 in this sense once =
it's out.

Some sites would appreciate readily usable repositories for prereleases. =
Stephan will provide those for EL/Fedora [for SUSE, those should already =
exist too]. Andrew will approach Russ to see what can be done for =
debian.

=3D=3D Next stable branch(es) =3D=3D

No news. Andrew still volunteers for posting to -info, describing the =
options and asking for feedback, and he proposes reserving a meeting =
(maybe the one right after the 1.6.8 release) for this topic because =
starting a discussion on it after an hour of discussing the daily issues =
seems not to be working out.

--=20
Stephan Wiesand
DESY -DV-
Platanenenallee 6
15738 Zeuthen, Germany