[OpenAFS-devel] release-team meeting minutes 2013-08-21

Stephan Wiesand stephan.wiesand@desy.de
Tue, 3 Sep 2013 17:11:14 +0200


Log: =
http://conference.openafs.org/release-team@conference.openafs.org/2013-08-=
21.txt

Attendance:

* Andrew Deason
* Ben Kaduk
* Derrick Brashear
* Jeffrey Altman
* Mike Meffie
* Stephen Quinney
* Stephan Wiesand

=3D=3D Changes merged or cleared for merging during the meeting =3D=3D

8910    afs: Avoid tracking file locks for RO volumes
9065    xstat: length check cm call info (for fs flushall)
9070    dafs: preattach should wait for exclusive states
9281    namei: Abstract out OGM functions a bit more
9282    salvager: Trust inode-based special data over OGM
9283    DAFS: VRS_r with VOL_SALVAGE_NO_OFFLINE in attach2
9286    afs: Make last_error always useful
9287    afs: Handle easy GetValidDSlot errors
9359    libafscp: NULL return value not pointer
9434    afs: Do not use separate array for srvAddrs
9435    auth: avoid excessive stat of cellservdb
9436    namei: force-close fd on read or write error
9973    libafscp: Actually return callback from FindCallback
10131   do not ship uncompressed tarballs

=3D=3D Changes needing more work =3D=3D

9571    bozo: retry start after error stops

A change to cap the retry delay is on master now and was pulled up as =
10148.

=3D=3D Server side NAT pings =3D=3D

Though it seems unclear whether they are the right solution, we still =
plan to have them in pre1 but back them out before the 1.6.6 release =
unless there's evidence that they are helpful.

=3D=3D Open bugs to fix in 1.6.6 =3D=3D

=3D=3D=3D RT #131709: Please apply fix to build dvolser on freebsd =3D=3D=3D=


Fix available in gerrit 10122.

=3D=3D=3D #131706: buserver segmentation fault in 1.6.5 =3D=3D=3D

Fix available in gerrit 10115 (pullup: 10166).

=3D=3D=3D Fileserver & salvager segfaults recently reported on =
openafs-info =3D=3D=3D

Reported by Harald Barth. Gerrit 10165 should fix the salvager problem. =
There's insufficient data on the fileserver problem.

=3D=3D=3D RT #131716: comerr checks in 1.6 don't do what aklog expects =
=3D=3D=3D

Recently submitted, not being worked on yet.

=3D=3D 1.6.x strategy =3D=3D

The last releases were rather conservative in terms of accepting =
features, non-critical fixes or code cleanups. Wen can't go o like this =
forever though. A new stable release series is not in sight, and the =
skew between 1.6.x and master is already problematic. The 1.6.6 release =
will thus have a few more changes of these kinds.

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