Hi folks! Time for an update on the Fedora 27 Beta status.
tl;dr action summary
ACTION: QA to test and karma update:
then check whether backgrounds are fixed for all blocking desktops
after compose run with update included.
ACTION: QA to test and karma update:
ACTION: kernel team to provide a suitable update, ARM QA to test it
ACTION: anaconda and LVM folks to co-ordinate and provide a fix
ACTION: QA to confirm this is fixed already
ACTION: lvrabec to fix remaining SELinux denials during FreeIPA
ACTION: QA to test and see how commonly encountered, desktop team to fix
ACTION: QA to answer outstanding questions in bug
QA to cover several missing tests (see below for details), stand by to
test with new validation compose soon.
- distribution - NEW
Fedora 27 Beta backgrounds must be different from Fedora 26
This is kind of a tracker for all the work that's needed to get the
backgrounds updated in all blocking desktops. We now have the package
reviewed and submitted as an update:
with the desktop-backgrounds changes included, but there may be
other changes needed also. So at the least we need karma for that
update, then we need to figure out what else needs to change to ensure
at least Xfce, GNOME and KDE have the updated backgrounds.
- grub2 - ON_QA
Wrong version on legacy variant (e.g. grub2-pc) Obsoletes:
This is the bug that causes upgrades to choke on grub2 package
dependencies. An update is available and just needs testing and karma -
you should be able to test just by verifying that the issues no longer
appear if you try upgrading with the updates-testing repo enabled.
- kernel - MODIFIED
Raspberry Pi 3: run-initial-setup hangs
Peter's done a build that ought to fix this, but not yet submitted an
update - so we either need a new update created, or the existing update
for kernel-4.13.0-1.fc27 edited to include the new build. Ideally we'd
prefer a build with just what's in the frozen repos now plus the fix
for this, but the 4.13.0-1.fc27 update has been in testing for a while
and has +2 karma, so it probably works OK. However, there's another
issue: both 4.13.0-1 and 4.13.1-301 were built with debugging enabled,
and I think we usually ship Beta with debugging disabled. So we kinda
need another build with debugging disabled, I think.
- lvm2 - NEW
Rescue mode fails while trying to access LVM volumes from existing install
This is a failure of the installer's rescue mode when trying to mount
LVM volumes from an existing Fedora install. There's no fix built yet,
but there's some suggestion of a 'quick fix' that could be done on the
anaconda side in the bug:
We need someone to take charge of deciding how to fix this, and...fix
- python-blivet - ON_QA
calls e2fsck on all ext volumes, provides no status indicator, and hangs indefinitely
if e2fsck doesn't exit
This is almost certainly fixed, we just need someone to test and
confirm with a system where the effect is obvious. The fix has actually
broken filesystem resizing in most cases, but that technically blocks
Final rather than Beta.
- selinux-policy-targeted -
'map' denial for comm 'ns-slapd' path
'/run/dirsrv/slapd-DOMAIN-LOCAL.stats' (breaks FreeIPA deployment)
The specific denial initially reported here is fixed, but there are
still denials preventing FreeIPA server deployment working with the
latest selinux-policy that has reached stable (-280), so we need
Lukas to fix those. I have listed the remaining denials in this bug
and in #1488404.
- gnome-shell - ASSIGNED
Program terminated with signal SIGSEGV, Segmentation fault. #0 0x00007f16279aa68f in
_cogl_boxed_value_set_x () ...
This is a crash in mutter which seems quite easy to trigger by running
VMs in virt-manager and Boxes. There's an Ubuntu bug which reports
the same crash triggered by some sort of Ubuntu update tool as well.
I have reproduced this on two boxes and there are at least three other
reporters across this bug, an upstream GNOME bug and a Launchpad bug,
so it seems fairly easy to hit. It would be useful if other folks
running GNOME on Wayland in F27 could test launching VMs in Boxes and/or
virt-manager and confirm whether doing this sometimes crashes their
GNOME session. Of course, we could also do with the GNOME folks
figuring out a fix.
- selinux-policy - ON_QA
There is FW Raid set, but there is no /dev/md* device
There are some questions in the bug that need answering: does this work
OK if using a network or DVD install image? Does it work OK if booting
a live image with enforcing=0 ? You need an Intel firmware RAID set
(that's known to basically work with previous Fedora releases) to test
Looking at https://www.happyassassin.net/testcase_stats/27/
some things jump out...
* The Cloud tests appear never to have been run during this cycle.
* Most of the Beta desktop tests have never been run on KDE during this cycle.
has never been
run during this cycle.
* Our old friend https://fedoraproject.org/wiki/QA:Testcase_install_to_SAS
has never been run during this cycle.
* Neither has https://fedoraproject.org/wiki/QA:Testcase_Kickstart_File_Path_Ks_Cfg
* Many upgrade tests have no results, this is because they constantly fail in
openQA, but not necessarily for release blocking reasons; I will look into
that and update.
* The tests for reporting crashes from anaconda haven't been run at all:
* Server test coverage looks bad, but this is mainly because openQA doesn't
report failures and FreeIPA has been broken for the entire F27 cycle.
With a couple more SELinux fixes most of those tests should show up
as passing. We do still need the AD tests run, though.
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net