[OpenAFS-devel] release-team meeting minutes 2014-05-07

[Stephan Wiesand]

Log: =
http://conference.openafs.org/release-team@conference.openafs.org/2014-05-=
07.html

Participants:

* Andrew Deason
* Ben Kaduk
* Jeffrey Altman
* Marc Dionne
* Mike Meffie
* Simon Wilkinson
* Stephan Wiesand

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

RT #131855 turned out to be an old problem, not a recent regression. A =
fix is available in gerrit 11118 but seems late for 1.6.8 given the =
circumstances. Postponed to 1.6.9 unless the reporter absolutely wants =
it in 1.6.8.

Marc tested the current Linux 3.15rc, and we're still good.

A single kernel oops was observed while using the 1.6.8pre client, but =
there's no evidence that it was due to openafs.

Decision: Announce the additional pre2 binaries we have, and call for =
another week of testing. After that (and if no new issues are found), =
release pre2 as 1.6.8 without further changes. When announcing, mention =
the known issues RT #131855 (fakestat cache consistency issue) and RT =
#131852 (server crashes in libkrb5 on UNIX systems using Heimdal).

=3D=3D Other problem reports =3D=3D

RT #131859 and #131862 report two issues surfacing when trying to use =
our Red Hat packages with rpm-ostree:

The first one is that our packages require and use /sbin/depmod, which =
has moved to /usr/sbin as of Fedora 17 and RHEL 7. Not a problem in most =
cases. Fix available in gerrit 11128, to be considered for 1.6.9.

The second one is that in said environment /usr is immutable at run =
time, and our server packages have their configuration files in =
/usr/vice/etc and /usr/afs, and even modify some of them when the =
fileserver starts. A remedy for this is more intrusive. Changing from =
Transarc paths to FHS is not an option within the current stable series, =
and maybe not even for the next one. Moving configuration to /etc with =
symlinks pointing there from the /usr locations could be considered, but =
probably only for new platforms not yet released (F21, RHEL7). Stephan =
to look into this and find out whether it's feasible before the RHEL7 =
release. The exact locations should probably be those already used by =
other packaging. Simon also proposes considering to abandon our =
packaging altogether and simply use the RPMFusion one maintained by Ken.

=3D=3D Next stable branches =3D=3D

This topic was discussed in the openafs room after the meeting. Log: =
http://conference.openafs.org/openafs@conference.openafs.org/2014-05-07.ht=
ml . A summary by Andrew is here: =
https://lists.openafs.org/pipermail/openafs-info/2014-May/040687.html .

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