Fedora 22 Beta blocker bug status #1

Thu Mar 26 05:57:31 UTC 2015

Hi folks! It's everyone's favourite time of the year - time for QA to 
make you feel bad about blocker bugs!

We're still six days out from Beta freeze, but I wanted to try and get 
out ahead of the blocker list and try to keep up the 'shipping on 
time' trend we started with Alpha...

If you're directly CCed on this email, it's for one of these reasons:

1) You maintain one of the packages that needs maintainer attention
2) You don't actually maintain one of the packages that needs 
maintainer attention but I think you do because I can be an idiot 
sometimes
3) You're a Great Leader of some sort (hi, Paul, Jaroslav) and thus 
get to read lots of mail from idiot monkeys

If you're a case of 2), please feel free to yell at me at your leisure!

tl;dr
=====

Testers, please test and karma these updates:

https://admin.fedoraproject.org/updates/libedit-3.1-11.20141030cvs.fc22
https://admin.fedoraproject.org/updates/FEDORA-2015-4398 (you can up-
karma that one if networking works in the TC4 installer)
https://admin.fedoraproject.org/updates/fedora-release-22-0.14

Developers, the following bugs need fixes. If you believe any of them 
should not be a blocker, please follow up here or in the bug report.

anaconda-n-friends
------------------

https://bugzilla.redhat.com/show_bug.cgi?id=1166598 "going back to 
installation destination picker swaps partitions on disks"
https://bugzilla.redhat.com/show_bug.cgi?id=1185117 "Custom 
partitioning does not allow convenient removal of volume including 
snapshots (btrfs, LVM)"

others
------

https://bugzilla.redhat.com/show_bug.cgi?id=1197894 "growpart: sfdisk 
dropped --show-pt-geometry option" (NEEDS ATTENTION)
https://bugzilla.redhat.com/show_bug.cgi?id=1204612 "eth0 is going 
missing in the cloud images"
https://bugzilla.redhat.com/show_bug.cgi?id=1204739 "Installing Fedora 
Server netinst ends up with blocked Cockpit port"
https://bugzilla.redhat.com/show_bug.cgi?id=1183807 "network spoke 
listed as not connected despite having assigned IP and hostname"
https://bugzilla.redhat.com/show_bug.cgi?id=1185604 "fedup upgrade 
fails within the initramfs-fedup env"

Detailed bug discussion
=======================

1.  https://bugzilla.redhat.com/show_bug.cgi?id=1166598- anaconda - 
"going back to installation destination picker swaps partitions on 
disks"

AIUI, someone (vpodzime?) is tasked to work on this and there's a 
plan, but it's moderately complex and there's some suggestion from 
anaconda team that this is more appropriately a Final blocker than 
Beta. The issue is fairly well understood and reproducible.

2. https://bugzilla.redhat.com/show_bug.cgi?id=1185117 - anaconda - 
"Custom partitioning does not allow convenient removal of volume 
including snapshots (btrfs, LVM)"

Again there's a question as to whether this is 'really' a blocker, 
though personally I think cmurf made a strong argument; it's clearly 
painful to remove a btrfs/LVM volume with more than a few snapshots in 
custom partitioning, and we can't really pretend snapshotted volumes 
aren't a thing. There might be an argument for pushing it to Final. I 
do not know what the status of work on this bug is.

3. https://bugzilla.redhat.com/show_bug.cgi?id=1191842 - anaconda - 
"unable to continue the installation after type into a weak password"

This is the good ol' password strength bikeshed. anaconda implemented a
system that allows the password strength checking to be customized, 
but FESCo has some concerns about whether it's sufficient to their 
requirements. I think I'm gonna leave this one to FESCo to herd, they 
seem to be on top of what they want from it.

4. https://bugzilla.redhat.com/show_bug.cgi?id=1184173 - bind - 
"dhclient fails to renew lease, results in dropped IPv4 network 
connection"

This one is basically done, the update is being pushed stable. The fix 
for this bug did cause the showstopper bug in TC3 whose fix caused the 
showstopper bug in TC4, though. I hate software. Anyone want to bet 
what the fix for the TC4 showstopper will break in TC5?!

5. https://bugzilla.redhat.com/show_bug.cgi?id=1197894 - cloud-utils - 
"growpart: sfdisk dropped --show-pt-geometry option"

This one rather seems to have been sitting around since Alpha; we 
could do with some urgency on it. Is anyone from Cloud WG able to step 
up and ensure this gets resolved?

6. https://bugzilla.redhat.com/show_bug.cgi?id=1204612 - distribution -
"eth0 is going missing in the cloud images"

Kushal Das seems to have taken ownership of this one today, so it 
looks like it's being taken care of.

7. https://bugzilla.redhat.com/show_bug.cgi?id=1204739 - fedora-
release - "Installing Fedora Server netinst ends up with blocked 
Cockpit port"

I believe 
https://admin.fedoraproject.org/updates/fedora-release-22-0.14 is 
intended to fix this, though it's not marked as such. Probably needs 
testing.

8. https://bugzilla.redhat.com/show_bug.cgi?id=1201120 - libblockdev - 
"DeviceTreeError: could not find parent for subvol"

We have a potential fix for this as a scratch build awaiting testing 
by the reporter who submitted the bug. I don't think anyone else has a 
reproducer at present, though it should be fairly easy to work one up 
if need be.

9. https://bugzilla.redhat.com/show_bug.cgi?id=1201897 - libedit - 
"SIGSEGV in libedit call in installer"

This is the showstopper bug in TC4 (in non-live images, anaconda 
crashes shortly after reaching the hub) that was triggered by the fix 
for the showstopper bug in TC3 (see next bug!) We traced this one out 
over the last couple of days and it looks like 
https://admin.fedoraproject.org/updates/libedit-3.1-11.20141030cvs.fc22 
addresses the problem; TC5 is currently underway.

10. https://bugzilla.redhat.com/show_bug.cgi?id=1204031 - lorax - "22 
Beta TC3 install images do not bring up network (unless updates image 
or kickstart used)"

So the bind update which we pulled into TC3 to fix bug #4 moved its 
libraries around for $REASONS, and requires a file in 
/etc/ld.so.conf.d to be respected when building the linker cache in 
order for dhclient to work. Turned out /etc/ld.so.conf was left out of 
the anaconda environment, so files in /etc/ld.so.conf.d weren't 
respected when ldconfig.service rebuilt the linker cache on boot, and 
dhclient stopped working. So we fixed lorax to include /etc/ld.so.conf 
in the anaconda environment...which fixed this bug, but caused bug #9, 
approximately because llvm. Oh software, how we hate you. Anyway, the 
lorax change does seem to fix *this* bug (right before anaconda 
crashes in TC4, you can see the network's working...), so unless 
anyone sees any other problems with it, we should up-karma the update -
that's https://admin.fedoraproject.org/updates/FEDORA-2015-4398 - and 
push it stable.

11. https://bugzilla.redhat.com/show_bug.cgi?id=1183807 - 
NetworkManager - "network spoke listed as not connected despite having 
assigned IP and hostname"

The status of this one is a bit mysterious. pwhalen stopped seeing it 
in TC3 and TC4, but nothing went into those builds which would be 
*expected* to fix it so far as anyone knows, and bcl mentioned seeing 
something like it today in IRC. Probably needs testing with TC5 and 
more investigation.

12. https://bugzilla.redhat.com/show_bug.cgi?id=1185604 - systemd - 
"fedup upgrade fails within the initramfs-fedup env"

Both the cause and the fix for this seem to be known at this point, 
and we're waiting on the systemd maintainers finding time to do an F21 
build. We would really appreciate it if they'd do that soon, as we 
want to be able to test upgrades properly. Thanks, systemd maintainers.

Extra credit: proposed blockers
===============================

We also have a couple of proposed blockers which look like fairly 
strong candidates and should be treated with priority.

1. https://bugzilla.redhat.com/show_bug.cgi?id=1205534 - gnome-initial-
setup - "gnome-initial-setup crashes upon selecting language"

This one just came onto the radar today, but it looks fairly 
significant, and is more or less described by the title.

2. https://bugzilla.redhat.com/show_bug.cgi?id=864198 - grubby - 
"grubby fatal error updating grub.cfg when /boot is btrfs"

This is something of a greatest hit; grubby has had trouble handling 
/boot on btrfs for a long time. For the last couple of releases 
anaconda disallowed /boot-on-btrfs which meant it was effectively a 
non-issue, but as of now, F22 anaconda allows it again; we either need 
to ban it again, or fix grubby.
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net
http://www.happyassassin.net