Fedora 17 Beta status 2: open blockers, karma and blocker vote requests

Wed Mar 21 22:43:43 UTC 2012

Hi, folks - it's Beta blocker bugging time again!

We've had a giant buttload of bugs to fix, mainly with the 'noloader'
change in anaconda, but it's looking like Will has actually managed to
squish them all, amazingly. So we're looking fairly good right now. What
we do need is blocker status votes on quite a long list of proposed
blockers. Please, blocker voters, take a look through the list and cast
your votes. If there aren't sufficient votes later today, I'll propose
an off-schedule blocker review meeting tomorrow morning.

Action item checklist
---------------------

Anyone - provide karma for these updates:

* https://admin.fedoraproject.org/updates/dracut-017-59.git20120321.fc17
* https://admin.fedoraproject.org/updates/grubby-8.11-1.fc17
* https://admin.fedoraproject.org/updates/lxdm-0.4.1-1.fc17 (NTH)

Blocker status voters, please vote on the blocker status of these bugs:

* http://bugzilla.redhat.com/show_bug.cgi?id=805405
* http://bugzilla.redhat.com/show_bug.cgi?id=804309
* http://bugzilla.redhat.com/show_bug.cgi?id=804504
* http://bugzilla.redhat.com/show_bug.cgi?id=804506
* http://bugzilla.redhat.com/show_bug.cgi?id=804716
* http://bugzilla.redhat.com/show_bug.cgi?id=804813
* http://bugzilla.redhat.com/show_bug.cgi?id=805310
* http://bugzilla.redhat.com/show_bug.cgi?id=805166
* http://bugzilla.redhat.com/show_bug.cgi?id=804587
* http://bugzilla.redhat.com/show_bug.cgi?id=802069
* http://bugzilla.redhat.com/show_bug.cgi?id=805017
* http://bugzilla.redhat.com/show_bug.cgi?id=805718 (NTH) (we really
want to accept this one, it's the updated release notes)
* http://bugzilla.redhat.com/show_bug.cgi?id=805272 (NTH)

anaconda/kernel developers, please take a look at:

* http://bugzilla.redhat.com/show_bug.cgi?id=805166

Ben Skeggs and desktop team (or someone - me or spot could do it,
frankly), please set up a gnome-session blacklist for NV30 (and
optionally, NV10 and NV20):
http://bugzilla.redhat.com/show_bug.cgi?id=745202
http://bugzilla.redhat.com/show_bug.cgi?id=770197 (NTH)

Dan Williams, please push out a build to fix:

* http://bugzilla.redhat.com/show_bug.cgi?id=805405

abrt team could look at this bug, but I suspect it will be rejected:

* http://bugzilla.redhat.com/show_bug.cgi?id=804309

ditto ajax/ssp for this one, likely to be rejected:

* http://bugzilla.redhat.com/show_bug.cgi?id=805017

Detailed analysis of bugs follows. In the interests of brevity I've left
out a ton of bugs that are basically 'this will be fixed in the next
anaconda build, and we need a new compose to test and verify'. That
applies to all the modified anaconda bugs. There will soon be an
anaconda 17.14 with a giant pile of noloader fixes.

Accepted blockers
-----------------

1. http://bugzilla.redhat.com/show_bug.cgi?id=804454 -
"Fedora-17-Beta-TC2-i686-Live-KDE.iso is oversized (703 MiB)"

The dependent bug http://bugzilla.redhat.com/show_bug.cgi?id=804451 is
fixed, so we can be confident this is addressed. But we can only close
it off after doing the next compose and verifying the size is okay.

2. http://bugzilla.redhat.com/show_bug.cgi?id=736993 "error install
bootloader with serial interface install"

This one should be fixed in TC2, but we can't verify as there's a bug
which breaks serial console even earlier. We will have to wait for
TC3/RC1, with a fix for the other issue, to verify.

11. http://bugzilla.redhat.com/show_bug.cgi?id=800205 "kernel panic
after running preupgrade"

This one may be fixed with the new anaconda, but given how preupgrade
works, we can't test until anaconda >= 17.13 goes stable, because only
then will it show up on the mirrors where preupgrade gets its anaconda
image from, so no action possible for now.

13. http://bugzilla.redhat.com/show_bug.cgi?id=745202 "gnome-shell does
not display correctly with NV3x adapters - multicolor corruption of
panel, Shell-style menus and text [nvfx]"

Ben and I did some investigation and we think it's really only NV30
adapters that are definitely completely busted and need to be
blacklisted. NV40 adapters are not all busted, we think just a few of
them suffer the VRAM exhaustion bug (770197). So we just need to roll a
gnome-session that blacklists NV30 chips.

Proposed blockers
-----------------

1. http://bugzilla.redhat.com/show_bug.cgi?id=805405 "NM adds device to
firewall zone after successful activation, causing a DHCPv6 catch-22"

This needs votes on blocker status, from the people who usually show up
to blocker reviews. It seems like a clear blocker, as it's affirmed to
prevent OOTB IPv6. We're hoping it's the *last* bug that does so.

2. http://bugzilla.redhat.com/show_bug.cgi?id=804309 "[abrt]
abrt-addon-ccpp-2.0.7-4.fc17:
abrt-action-analyze-core:106:extract_info_from_core:IndexError: list
index out of range"

This also needs blocker status votes. It prevents abrt from filing a
report in some cases. It seems at first there was a worry that it
prevented abrt from filing *any* crash, but that no longer seems to be
the case. Given that, and the fact it can be well resolved with an
update, I'm inclined to think it won't be accepted as a blocker.

3. http://bugzilla.redhat.com/show_bug.cgi?id=804504 "vnc interface
doesn't work"

This also needs blocker status votes, and is a clear blocker - it
directly violates a simple criterion. Will believes he already has a fix
for it.

4. http://bugzilla.redhat.com/show_bug.cgi?id=804506 "anaconda does not
parse 'console=ttyS0'"

Another one that needs blocker status votes, and is a clear blocker: it
directly violates the requirement that serial console install work. Will
believes he has it fixed.

5. http://bugzilla.redhat.com/show_bug.cgi?id=804716 "after minimal 17
Beta TC2 install, network service doesn't work"

Another one that needs blocker status votes. This one has a couple of
+1s, but my personal opinion is it's not really a big enough issue to be
a blocker. You've always had to do a manual action to bring the network
up after a minimal install, and we've consistently accepted this as
okay. I don't really see that a change in what that action _is_ is a
huge deal that requires us to block the beta. Anyhow, please vote! Will
believes he has a fix for this in the pending anaconda build.

6. http://bugzilla.redhat.com/show_bug.cgi?id=804813 "repo type "nfsiso"
unsupported (mount: unknown filesystem type 'nfsiso')"

This also needs blocker votes, and seems a clear blocker, as it causes a
supported repo type to fail. Will believes this is fixed in a new dracut
build which needs karma:
https://admin.fedoraproject.org/updates/dracut-017-59.git20120321.fc17 .
It would be hard to verify the fix for this bug, but we can test that
dracut works for kernel updates and booting.

7. http://bugzilla.redhat.com/show_bug.cgi?id=805310 "Corrupted
grub2.cfg content: trailing { missing in menuentry lines"

Another one needs votes. This is the bug that's been under discussion
today involving the grub2 config file getting messed up by grubby when
you install updated kernels after installing F17 with grub2 1.99-19 or
higher.  I'm -1 blocker +1 NTH on this, but it's a close call and may be
accepted. We have a fix for it, which needs karma:
https://admin.fedoraproject.org/updates/grubby-8.11-1.fc17 . Please vote
and karma.

8. http://bugzilla.redhat.com/show_bug.cgi?id=805166 "anaconda doesn't
start with direct kernel boot and root=live:nfs://"

This also needs blocker status votes. This is probably another
'noloader' consequence but is assigned to kernel at least for now. It
seems quite borderline to me - the criteria don't explicitly require
remote delivery of the installer with direct kernel boot in this way to
work - but we need to think about it. Will is looking to see if it's
fixable anaconda side. He thinks it may just need a minor syntax change.

9. http://bugzilla.redhat.com/show_bug.cgi?id=802069 "Corrupted
rendering of GNOME Shell on NV44: looks like VRAM exhaustion, doesn't
seem to be"

Also needs votes. This is a funny one. It was merged into 745202
initially, but now Ben and I don't think that's correct. Kamil's
'confirmation' of this bug seems to clearly actually be a case of VRAM
exhaustion, hence #770197, but Paul's actually doesn't: he doesn't have
the log messages we'd expect to see in the VRAM exhaustion case, and his
card has 256MB of VRAM, which should be ample. On the other hand, thanks
to the Fedora Forum folks, we have a lot of confirmation that there are
many NV4x cards with sufficient VRAM which work just fine and don't see
this kind of corruption. So we don't think this is a general issue. All
I can think of at present is that there's some kind of issue specific to
Paul's config which happens to present very similarly to VRAM
exhaustion. It could, thinking about it, be overheating. Given the solid
data we have that most NV4x's aren't suffering from this, I suspect
it'll be rejected as a blocker.

10. http://bugzilla.redhat.com/show_bug.cgi?id=805017 "Segmentation
fault in anaconda when switching TTYs"

Also needs votes. This is a longstanding bug that probably anyone who's
run Fedora in a Spice/qxl KVM has seen: often, when you switch out to a
VT, X crashes. It's pretty annoying, but it doesn't really hit any of
the criteria (the VTs still *work*) and it's restricted to a specific VM
case. It was present in F16, and that didn't cause terrible plagues. So
I'm -1 on it and suspect it'll be rejected.

Approved NTH
------------

No point discussing the unmodified ones, but:

1. http://bugzilla.redhat.com/show_bug.cgi?id=794478 "[abrt] kernel:
BUG: soft lockup - CPU#0 stuck for 185s! [lxdm-binary:744]"

There's a fix for this that needs karma:
https://admin.fedoraproject.org/updates/lxdm-0.4.1-1.fc17 . It would be
nice to get this karma'ed up so we can pull it in and ensure the Beta
LXDE spin is fixed.

Proposed NTH
------------

1. http://bugzilla.redhat.com/show_bug.cgi?id=805718 "Fedora Release
Notes for F17 Beta need pulled into F17 Beta"

Pretty self-explanatory. We definitely want the updated release notes in
the Beta, so please +1 this, NTH status voters!

2. http://bugzilla.redhat.com/show_bug.cgi?id=805272 "failure to honor
valid legacy MBR when stale GPT is present"

This is the issue Chris Murphy identified where parted doesn't correctly
handle a disk that's been formatted as GPT and then formatted over as
MBR without all the GPT markers being removed. anaconda relies on
parted, so if you have such a disk and try to install to it, you can
only completely reformat it or abort. It seems like a clear Final
blocker, but bcl is of the opinion it's a bit obscure to be worth poking
parted for Beta at this stage. Please read the bug and vote, NTH status
voters!

Thanks for reading, everyone! Remember to help out with any actions you
can.
-- 
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net