[Bug 666494] New: umount failures during halt or reboot
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c0 Summary: umount failures during halt or reboot Classification: openSUSE Product: openSUSE 11.2 Version: Final Platform: x86-64 OS/Version: openSUSE 11.2 Status: NEW Severity: Critical Priority: P5 - None Component: Basesystem AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: andihartmann@freenet.de QAContact: qa@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:2.0b10pre) Gecko/20110123 Firefox/4.0b10pre SeaMonkey/2.1b2pre If the system is rebooted or halted, it is possible, that there are errors during umounting the active filesystems. The errors occure on the filesystems with the following mountpoints: /var /home I put some debugcode (lsof) to the start / stopscripts (here to /lib/cryptsetup/boot.crypto.functions) to see, why the error occures (/var and /home are crypted). This is what I found: /var-filesystem: libvirtd blocks because of access to /var/run/nscd/passwd: libvirtd 6448 user4 mem REG 253,20 217016 164197 /var/run/nscd/passwd (nscd has been already shutdown) (I'm running virsh to manage kvm vm's. I don't use the global libvirtd, but a libvirtd for each user, which runs a vm) Another problem was kblankscr (KDE 4.4.4 release 3) kblankscr 15671 user mem REG 253,20 16240640 524984 /var/tmp/kdecache-gemeinsam/kpc/kde-icon-cache.data kblankscr 15671 user mem REG 253,20 4296704 524841 /var/tmp/kdecache-gemeinsam/kpc/kde-icon-cache.index kblankscr 15671 user mem REG 253,20 39056 377083 /var/cache/fontconfig/77e41c5059666d75f92e318d4be8c21e-x86-64.cache-2 kblankscr 15671 user mem REG 253,20 22792 376997 /var/cache/fontconfig/1890317d8a3fded7854d26fdd2e67c05-x86-64.cache-2 kblankscr 15671 user mem REG 253,20 12152 376999 /var/cache/fontconfig/c952848b32a83a3db68a27ab871288f2-x86-64.cache-2 kblankscr 15671 user mem REG 253,20 377544 376889 /var/cache/fontconfig/df311e82a1a24c41a75c2c930223552e-x86-64.cache-2 kblankscr 15671 user mem REG 253,20 681416 377089 /var/cache/fontconfig/7ef2298fde41cc6eeb7af42e48b7d293-x86-64.cache-2 This problem occured during halt started on the console while a KDE-session has still been running. /home-filesystem The home-filesystem was blocked by kdesud: kdesud 13272 user1 cwd DIR 253,23 3464 44891 /raid/home/user1/Documents kdesud 13272 user1 1w REG 253,23 108554 71 /raid/home/andreas/.xsession-errors This problem also occured during halt started on the console while a KDE-session has still been running. It seems, that the kill and term-script at the beginning of the reboot / halt procedure doesn't kill the processes described above. As consequence, the filesystems can't be unmounted. This should be fixed! These are the rpm-packages which are installed: sysvinit-2.86-215.3.1.x86_64 fuse-2.7.4-3.4.1.x86_64 device-mapper-1.02.31-11.3.1.x86_64 lvm2-2.02.45-11.3.1.x86_64 cryptsetup-1.0.7-10.1.x86_64 kdebase4-runtime-4.4.4-3.7.3.x86_64 Reproducible: Always Steps to Reproduce: 1. Take care, that kblankscr, libvirtd and kdesud are running. The init-script of libvirtd is disabled, because it isn't used. 2. Do a reboot / halt on the console. Actual Results: The umount of some of the filesystems wasn't possible, because kdesud, kblancscr and libvirtd are still running and have not been killed. Expected Results: I expect that the term and kill script, which has been running before umounting the filesystem does kill all the processes which could block the umounting of the filesystems. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c wei wang <wewang@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |wewang@novell.com AssignedTo|bnc-team-screening@forge.pr |coolo@novell.com |ovo.novell.com | -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c Stephan Kulow <coolo@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|coolo@novell.com |werner@novell.com -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c1 Dr. Werner Fink <werner@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |andihartmann@freenet.de --- Comment #1 from Dr. Werner Fink <werner@novell.com> 2011-01-25 10:27:47 UTC --- Does this also happen with openSUSE 11.4 /Factory (AFAIK I've added a fix for this at least to openSUSE 11.3) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c2 Dr. Werner Fink <werner@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |coolo@novell.com, | |dmueller@novell.com, | |jfehlig@novell.com, | |kde-maintainers@suse.de --- Comment #2 from Dr. Werner Fink <werner@novell.com> 2011-01-25 10:45:53 UTC --- The question rises: why does kdesud blocking HOME at shutdown? As a sidemark: In cases of a user space process is providing a file system (like fuse) the scripts can not term of kill such processes otherwise the parent file system can not unmounted and will be checked at next (re)boot. And also what happens if the libvirtd init script is *not* disabled. That is that it will stopped at shutdown/reboot. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c3 --- Comment #3 from Andreas Hartmann <andihartmann@freenet.de> 2011-01-25 15:12:28 UTC --- Created an attachment (id=410181) --> (http://bugzilla.novell.com/attachment.cgi?id=410181) umount error during reboot I set up a VM with openSUSE-DVD-Build0957-x86_64.iso. I installed a default KDE environment with the following individual partition-layout: /dev/mapper/system-root 1,2G 254M 838M 24% / devtmpfs 998M 196K 997M 1% /dev tmpfs 1004M 4,0K 1004M 1% /dev/shm /dev/vda1 152M 30M 115M 21% /boot /dev/mapper/system-home 2,0G 259M 1,7G 14% /home /dev/mapper/system-usr 16G 3,2G 12G 22% /usr /dev/mapper/system-var 1,2G 132M 1,1G 12% /var I switched off the firewall, switched on sshd and logged in as root. After KDE has been running, I issued halt or reboot on the ssh-session. Then, I could take the attached snapshot. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c4 --- Comment #4 from Dr. Werner Fink <werner@novell.com> 2011-01-25 16:59:13 UTC --- What exactly does this mean: Do have enabled libvirtd service boiot script, is this newer than 11.2? Please note that I'll not fix 11.2 as this may simply be fixed by an update. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c5 --- Comment #5 from Dr. Werner Fink <werner@novell.com> 2011-01-25 17:28:45 UTC --- The image of attachment is for 11.4 as I see, nevertheless enable libvirtd to see if /var remains busy. And if this is true, then please report which process holds /var open rw. The question rises why I've to be the baby sitter for KDE even if I do not use KDE nor GNOME ;) -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c6 --- Comment #6 from Andreas Hartmann <andihartmann@freenet.de> 2011-01-25 17:40:32 UTC --- To comment 5: The image belongs to a system which runs in a VM. Libvirtd wasn't enabled / running at all. It was another process, which had been access to the fs. Unfortunately I hadn't enabled lsof at this point and wasn't able to reproduce it until now. But the problem is not libvirtd or kdesud or kblancscr or whatever could be running during shutdown / umount. The problem is, that a filesystem can't be unmounted which is in use. There should be a solution to prevent a shutdown / reboot if the filesystem can't be unmounted. Minimally, there should a lsof be integrated, if the umount fails in order to provide the user the possibility to see the reason why the umount fails. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c7 --- Comment #7 from Dr. Werner Fink <werner@novell.com> 2011-01-25 17:52:27 UTC --- Just do a cp /usr/bin/lsof /bin/lsof and add /bin/lsof -x +d /var before umount call -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c8 --- Comment #8 from Andreas Hartmann <andihartmann@freenet.de> 2011-01-26 09:59:25 UTC --- To Comment 1: I tried to apply aaa_base-11.4-228.1.x86_64.rpm, cryptsetup-1.2.0-60.3.x86_64.rpm, libcryptsetup1-1.2.0-60.3.x86_64.rpm to my openSuSE 11.2 - system, but this doesn't work because of a relocation error of cryptsetup. Could you please tell me, which files (from which rpm-packet) are necessary to be changed to check if your fixes resolve the problem? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c9 Dr. Werner Fink <werner@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |lnussel@novell.com --- Comment #9 from Dr. Werner Fink <werner@novell.com> 2011-01-26 10:17:47 UTC --- The mkill(8) is part of aaa_base -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c10 --- Comment #10 from Andreas Hartmann <andihartmann@freenet.de> 2011-01-26 11:55:59 UTC --- I took mkill from sysvinit-tools-2.88-113.1.x86_64.rpm (it wasn't in aaa_base) and tested with OpenSuSE 11.2. md5sum /sbin/mkill a7af18fdca70e2f5a706c62943e5fa4f /sbin/mkill I started yast in kde as user with administration authorization (-> switch to root with the way KDE does it). The behaviour during reboot is unchanged (here the home-directory /raid/home): kdesud 15230 andreas cwd DIR 253,23 3496 44891 /raid/home/andreas/Documents kdesud 15230 andreas 1w REG 253,23 67155 71 /raid/home/andreas/.xsession-errors kdesud 15230 andreas 2w REG 253,23 67155 71 /raid/home/andreas/.xsession-errors mount | grep home /dev/mapper/cr_homeRaid on /raid/home type reiserfs (rw,acl,user_xattr) mkill sends term and kill-signal to processes, which have open files on filesystems which should be unmounted. What happens if they don't stop after they got the signal? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c11 --- Comment #11 from Dr. Werner Fink <werner@novell.com> 2011-01-26 12:53:29 UTC --- If SIGKILL does not help, no user space program is able to do anything. This may happen if a process is locked in D state due not happened file system sync. e.g. for NFS shares. The question is;: which process is in D state. Please debug the problem as described herein with lsof otherwise I need a crystal ball. First of all I'm interested which process does make /var busy in 11.4 then we can debug 11.2. For 11.2 the mkill has to be used in boot.localfs as it is done in 11.4 -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c12 Dr. Werner Fink <werner@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW CC| |bnc-team-gnome@forge.provo. | |novell.com, | |dimstar@opensuse.org, | |werner@novell.com InfoProvider|andihartmann@freenet.de | AssignedTo|werner@novell.com |lnussel@novell.com --- Comment #12 from Dr. Werner Fink <werner@novell.com> 2011-01-27 11:34:00 UTC --- As determined from a private mail of Andreas I guess that the umount calls done by /etc/init.d/boot.localfs do not chekc for busy partitions and therefore do not terminate and later on kill the processes making the encrypted partitions busy. AFAICS from the provate mail the /var partition is locked by packagekitd and the /home partition is locked by kdesud -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c13 --- Comment #13 from Andreas Hartmann <andihartmann@freenet.de> 2011-01-28 10:09:43 UTC --- To Comment 11: Unfortunately I was not able to reproduce the umount problem in OpenSuSE 11.4 (running in a VM) until now. I changed kvm to 0.12.5 because 0.13 has a big problem with the SDL-interface, which slows down one cpu completely. But: Even when running in normal speed, I couldn't reproduce it. Sorry. But I reported this bug for OpenSuSE 11.2. In 11.2, the problem can be most of the time reproduced e.g. by manually searching for new updates with the kdeupdateapplet while the configured DNS in resolv.conf is not responding (and there is no TCP/RST from the server, because the server can't be reached at all (like a TCP DROP)). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c14 --- Comment #14 from Ludwig Nussel <lnussel@novell.com> 2011-01-31 16:15:03 CET --- so this turned out to be a problem with PackageKit which doesn't terminate in reasonable time, right? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c15 --- Comment #15 from Dr. Werner Fink <werner@novell.com> 2011-01-31 16:20:39 UTC --- Even not with SIGKILL -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c Ludwig Nussel <lnussel@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|lnussel@novell.com |vuntz@novell.com -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c16 --- Comment #16 from Andreas Hartmann <andihartmann@freenet.de> 2011-03-21 07:16:13 UTC --- I installed openSuSE 11.4 on a laptop (32 bit). I created the partitions /var, /usr, /opt, /, swap and /home on a crypted LVM-volume. /boot is uncrypted. All filesystems where formated with reiserfs. On reboot, no umount got executed - the whole umount-section wasn't executed with RUN_PARALLEL enabled. If it was disabled, it has been working as expected. Strange: After I changed the filesystems from reiserfs -> ext4, the umount section after the init-scripts have been called again. I tried to reproduce this with a VM - the error can't be seen there. Very strange. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c17 Vincent Untz <vuntz@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |andihartmann@freenet.de --- Comment #17 from Vincent Untz <vuntz@suse.com> 2012-07-09 15:47:12 UTC --- I'm confused about the last comment: does it mean things are working fine now? Also, if you can still reproduce, can you try to kill packagekitd while it's running and see if it stops (with SIGQUIT, and then SIGKILL)? If it doesn't stop, can you run it in gdb, and provide a stack trace to know where it's blocked? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c18 --- Comment #18 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 06:31:08 UTC --- Created an attachment (id=497894) --> (http://bugzilla.novell.com/attachment.cgi?id=497894) Layout of reiser system I set up a openSUSE 11.4 system (VM) with actual patch level (KDE standard installation). You can see the different partitions on the attached picture. The filesystems used are reiserfs. The physical lvm device is crypted. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c19 --- Comment #19 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 06:44:44 UTC --- Created an attachment (id=497895) --> (http://bugzilla.novell.com/attachment.cgi?id=497895) OpenSUSE 11.4 reiserfs system shut down with original configuration The system was shut down with original configuration grep PARALLEL /etc/sysconfig/boot RUN_PARALLEL="yes" It looks fine, but the second look shows: the filesystems aren't unmounted at all. You can see it later on when starting up the system again: you're getting a lot of replays during mount of filesystems. There is a big chance to loose data! -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c20 --- Comment #20 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 06:48:21 UTC --- Created an attachment (id=497896) --> (http://bugzilla.novell.com/attachment.cgi?id=497896) OpenSUSE 11.4 reiserfs system shut down with RUN_PARALLEL=no The shutdown works fine, if you set RUN_PARALLEL to no like this grep PARALLEL /etc/sysconfig/boot RUN_PARALLEL="no" -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c21 --- Comment #21 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 06:54:26 UTC --- Created an attachment (id=497898) --> (http://bugzilla.novell.com/attachment.cgi?id=497898) openSUSE 11.4 ext4 system layout That's the second system with ext4 filesystem and 3 crypted physical LVM devices and one vggroup (all 3 physical LVM devices are in the same vggroup). Here, you can find a /etc/crypttab, which contains all of the physical LVM devices (remember: it was not there on the reiser test system with just one physical LVM device). -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c22 --- Comment #22 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 06:58:41 UTC --- Created an attachment (id=497899) --> (http://bugzilla.novell.com/attachment.cgi?id=497899) OpenSUSE 11.4 ext4 system shut down with original configuration The picture shows the shutdown of the ext4 system without any changes. You can see the errors for cr_vda3 and cr_vda2. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c23 --- Comment #23 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 07:04:14 UTC --- Created an attachment (id=497900) --> (http://bugzilla.novell.com/attachment.cgi?id=497900) OpenSUSE 11.4 ext4 system shut down with modified /etc/crypttab This is the shutdown of the ext4 system with modified /etc/crypttab. The lines for cr_vda2 and cr_vda3 have been disabled. cr_vda4 could be disabled without any problem during shutdown. Remember: cr_vda4 is part of the same volume group as cr_vda2 and 3. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c24 --- Comment #24 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 07:06:17 UTC --- Created an attachment (id=497902) --> (http://bugzilla.novell.com/attachment.cgi?id=497902) openSUSE 11.4 ext4 system layout pvdisplay This shows, that each physical LVM device is part of the same volume group. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c25 Andreas Hartmann <andihartmann@freenet.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|andihartmann@freenet.de | --- Comment #25 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 07:17:28 UTC --- Summary: - Using reiserfs with an original openSUSE 11.4 crypted installation means high risk of data loss. Maybe too, if it isn't crypted at all. Switching off parallel execution of the scripts in /etc/init.d resolves the problem. - Using ext4 and more than one crypted physical LVM devices brings up errors during shutdown (I think the same could be seen with any other fs). Editing /etc/crypttab makes the errors disappear. I'm wondering why there is a /etc/crypttab in the ext4 system, but there isn't any in the reiser system. So, if you leave the systems untouched, you have great chance to hit some data loss (reiserfs). The problems seen on the ext4 systems seem to be more optical nature, but uncomely. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c26 Dr. Werner Fink <werner@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |andihartmann@freenet.de --- Comment #26 from Dr. Werner Fink <werner@suse.com> 2012-07-10 11:40:29 UTC --- Does this also happen if you have boot loggin disabled? That is e.g. using the keyword `quiet' on the kernels command line but doing parallel boot? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c27 --- Comment #27 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 13:18:40 UTC --- Created an attachment (id=497994) --> (http://bugzilla.novell.com/attachment.cgi?id=497994) OpenSUSE 11.4 reiserfs system shut down with original configuration but without quiet and splash=silent kernel commandline options Hmm, quiet is set by default, means: all you can see here is with quiet set on kernels command line. reiserfs problem: Now, I removed "quiet" and couldn't see any difference but the sequence of stopped daemons (maybe chance). Next I additionally removed splash=silent. Now, I could see one try (of 3), which showed a fine umount of the filesystems. I attached this one :-). Seems to be chance if the umount is executed or not. If you compare the screenshot from comment 19 with this screenshot, you most probably will see, that there are some other jobs missing too during shutdown. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c28 Andreas Hartmann <andihartmann@freenet.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|andihartmann@freenet.de | --- Comment #28 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-10 13:19:19 UTC --- Forgot to remove NEEDINFO status. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c29 Vincent Untz <vuntz@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |vuntz@suse.com AssignedTo|vuntz@suse.com |werner@suse.com --- Comment #29 from Vincent Untz <vuntz@suse.com> 2012-07-11 12:43:38 UTC --- Unless I misunderstand, the last few comments seem to indicate this is not just an issue with packagekitd not stopping, but something more complex in sysvinit. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c30 Dr. Werner Fink <werner@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |andihartmann@freenet.de --- Comment #30 from Dr. Werner Fink <werner@suse.com> 2012-07-20 11:51:25 UTC --- (In reply to comment #28) Do you know *which* volume stays open even if the system tries to umount *what* volume/partition? Beside this: is this umount down within /etc/init.d/boot.crypto or in /etc/init.d/boot.localfs -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c31 Andreas Hartmann <andihartmann@freenet.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|andihartmann@freenet.de | --- Comment #31 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-20 12:59:17 UTC --- Created an attachment (id=499385) --> (http://bugzilla.novell.com/attachment.cgi?id=499385) slightly patched shutdown ext4: according to the attached screen shots, it's the / filesystem. I can see a wrong order during shutdown: At first, boot.crypto tries to deactivate the physical volumes of the LVM group. This can't succeed as long as the LV devices /, /usr, ... are mounted and the LV group (system) isn't deactivated. Next is boot.localfs. Here, the filesystems /dev/mapper/system/var, usr, opt, home are umounted. Volume group system can't be deactivated because of the mounted root fs. As long as the root fs is mounted (in the system-group), the system group can't be deactivated and therefore, the underlying physical devices cr_vda2 and cr_vda3 can't be deactivated. I'm surprised that cr_vda4 has been deactivated although it's part of the system group, too. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c32 Dr. Werner Fink <werner@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO InfoProvider| |andihartmann@freenet.de --- Comment #32 from Dr. Werner Fink <werner@suse.com> 2012-07-20 13:42:00 UTC --- First of all: do we really support crypted root file system? Second: do we support crypted root file system on an LVM? I do not think that there is a wrong order chain as the order if simply reverse in comparision to the boot sequence, simply compare ls -l /etc/init.d/boot.d/*{localfs,crypto,lvm,device-mapper} to see this. Only boot.crypto could try to disentangle the incoming file system. Beside this the root file system will be never unmounted but mounted ro only. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c33 Andreas Hartmann <andihartmann@freenet.de> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDINFO |NEW InfoProvider|andihartmann@freenet.de | --- Comment #33 from Andreas Hartmann <andihartmann@freenet.de> 2012-07-23 06:54:24 UTC --- Created an attachment (id=499536) --> (http://bugzilla.novell.com/attachment.cgi?id=499536) crypted lvm during installation Answer to your first two questions: Well, take a look at your installer. This is normal standard with linux and nothing special with openSUSE. Please: Do an installation yourself with the released images and all of your questions will be answered. It's easy. Do it in a VM! It's enough! -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c34 Dr. Werner Fink <werner@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|werner@suse.com |lnussel@suse.com --- Comment #34 from Dr. Werner Fink <werner@suse.com> 2012-07-23 13:02:01 UTC --- I'm speaking from both crypted root file system on an LVM. If we really support this then boot.crypto has to support it. Ludwig? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c35 --- Comment #35 from Ludwig Nussel <lnussel@suse.com> 2012-07-30 10:32:02 CEST --- yes, root on lvm in encrypted container is supported. This layering of devices is complicated to handle in shell scripts though. So yes, there may be errors displayed but it shouldn't harm. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
https://bugzilla.novell.com/show_bug.cgi?id=666494 https://bugzilla.novell.com/show_bug.cgi?id=666494#c36 Ludwig Nussel <lnussel@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Resolution| |WONTFIX --- Comment #36 from Ludwig Nussel <lnussel@suse.com> 2013-07-05 09:53:30 CEST --- Finally WONTFIX in this old distro. Shouldn't happen with systemd nowadays anymore. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
participants (1)
-
bugzilla_noreply＠novell.com