Hi,
we are using <mountby config:type="symbol">device</mountby> for every partitionin our profiles. After installing a bunch of hosts it turned out that this option is ignored on many hosts. We couldn't find a rule for it. It happened with creating and with reusing disks, on hosts with lvm and without, etc.
Uwe, can I send one y2log from a host where the option was ignored? Do you have time to look into it?
cu, Frank
On Tuesday 19 February 2008, Frank Steiner wrote:
we are using <mountby config:type="symbol">device</mountby> for every partitionin our profiles. After installing a bunch of hosts it turned out that this option is ignored on many hosts. We couldn't find a rule for it. It happened with creating and with reusing disks, on hosts with lvm and without, etc.
I was able to reproduce it with reusing a disk. It's a bug in autoyast and yast2-storage code. Will be fixed in SP2 and 11.0
Uwe Gansert wrote
I was able to reproduce it with reusing a disk. It's a bug in autoyast and yast2-storage code. Will be fixed in SP2 and 11.0
Is it possible to get a driverupdate for SLES10 or is it too complicated to backport? Or maybe a least a diff file so that I can try to hack it into AY myself?
cu, Frank
On Tuesday 26 February 2008, Frank Steiner wrote:
Is it possible to get a driverupdate for SLES10 or is it too complicated to backport? Or maybe a least a diff file so that I can try to hack it into AY myself?
diffs attached. AutoinstStorage has to be recompiled. Storage.ycp too
Currently the ata_piix modules gives me on older - standard ide disk - hardware problems thus i disabled it with brokenmodules=ata_piix but now machine with sata drives will fail.
The latest solution was to remove the broken_module from the command line and set insmod=piix instead. Thus the piix modules gets loaded before the ata_piix but in case of sata drives the ata_piix will succeed
cheers Hajo
Uwe Gansert wrote
On Tuesday 26 February 2008, Frank Steiner wrote:
Is it possible to get a driverupdate for SLES10 or is it too complicated to backport? Or maybe a least a diff file so that I can try to hack it into AY myself?
diffs attached. AutoinstStorage has to be recompiled. Storage.ycp too
Hmm, I added the patches to autoyast2.rpm and yast2-storage.rpm (the versions that came with SLES10 SP1) and recompiled Storage.ycp and used the AutoinstStorage.ycp from the sles10sp1_root_fix_v0.2 driverupdate when recompiling autoyast2.rpm.
Then I added the resulting Storage.ybc and AutoinstStorage.ybc to the driverupdate (also the autopart.ycp to y2update/include/ because I wasn't sure if that's needed, too) and restarted the installation.
As result, AY hangs forever working on the partitioning. I aborted it after y2log had grown to 2.1GB.
What can I do? Could it be the patch does not work against SLES10 SP1?
cu, Frank
On Thursday 28 February 2008, Frank Steiner wrote:
What can I do? Could it be the patch does not work against SLES10 SP1?
the patch should work against SP1. I did not try it against the root_fix driverupdate but it should work there too. I tried multiple partitioning schemes and had no problem with the patch.
Uwe Gansert wrote
On Thursday 28 February 2008, Frank Steiner wrote:
What can I do? Could it be the patch does not work against SLES10 SP1?
the patch should work against SP1. I did not try it against the root_fix driverupdate but it should work there too. I tried multiple partitioning schemes and had no problem with the patch.
I will do another careful recompile and try different schemes and let you know...
Ok, it definitely works on some schemes, on some not. It seems to work always when creating the partitions and hangs on some profile trying to keep existing partitions.
I have two quite similar schemes that might help you to catch the bug. Logfiles are quite big so I put them on my web page.
A profile created (by our script) from this template works with keeping: /dev/sda1 /dos vfat defaults 8gb /dev/sda2 swap swap defaults 4gb /dev/sda3 / ext3 defaults 25gb /dev/sda5 /local ext3 defaults 50gb /dev/sda6 /usr/local/storage ext3 defaults max
A profile based on this template works with creating, but fails with keeping partitions (when it ran through with creating before so that the partitions indeed exist). It works with keeping partitions without the patch. /dev/sda1 /dos vfat defaults 8gb /dev/sda2 swap swap defaults 4gb /dev/sda3 / ext3 defaults 25gb /dev/sda5 /local ext3 defaults 50gb /dev/sda6 /tmp ext3 defaults 4gb /dev/sda7 /usr/local/storage ext3 defaults max
Little difference, but seems to be an imporant one... Failing here means it loops endlessly and I killed it when y2log was larger than 1GB.
The logs and profiles:
Working with keep: http://www.bio.ifi.lmu.de/~steiner/ay/part_riemann_keep.xml http://www.bio.ifi.lmu.de/~steiner/ay/y2log.riemann.keep.works.bz2
Works with create: http://www.bio.ifi.lmu.de/~steiner/ay/part_cantor_create.xml http://www.bio.ifi.lmu.de/~steiner/ay/y2log.cantor.create.works.bz2
Fails with keep (woks without patch): http://www.bio.ifi.lmu.de/~steiner/ay/part_cantor_keep.xml http://www.bio.ifi.lmu.de/~steiner/ay/y2log.cantor.keep.failed.bz2
Hope this helps! I don't mind to act as beta tester for any try of a patch.
cu, Frank
Frank Steiner wrote
Ok, it definitely works on some schemes, on some not. It seems to work always when creating the partitions and hangs on some profile trying to keep existing partitions.
However, I can at least confirm that the patch for mountBy does indeed work in those cases where the partitioning algorithm does not hang :-)
autoinstall@lists.opensuse.org
-
Frank Steiner -
Hans-Joachim Ehlers -
Uwe Gansert