Mailinglist Archive: opensuse-bugs (2150 mails)

< Previous Next >
[Bug 871704] every second boot fails during mount of a /etc/fstab entry

--- Comment #59 from Andrey Borzenkov <arvidjaar@xxxxxxxxx> 2014-04-19 17:07:08
UTC ---
(In reply to comment #55)
today the mounts are missing ... the old bug is back ...

Apr 18 18:02:58 linux02 systemd-udevd[756]: timeout: killing '/sbin/lvm pvscan
--cache --activate ay --major 8 --minor 16' [841]
Apr 18 18:02:58 linux02 systemd-udevd[756]: '/sbin/lvm pvscan --cache
--activate ay --major 8 --minor 16' [841] terminated by signal 9 (Killed)
Apr 18 18:02:58 linux02 systemd-udevd[751]: timeout: killing '/sbin/lvm pvscan
--cache --activate ay --major 8 --minor 32' [840]
Apr 18 18:02:58 linux02 systemd-udevd[751]: '/sbin/lvm pvscan --cache
--activate ay --major 8 --minor 32' [840] terminated by signal 9 (Killed)

So we are back at square one.

What we are facing here is apparent lvmetad deadlock. There are two upstream
patches that fix possible deadlock case there. I added them to my package and
updated it. You could test it.

What would be really helpful - before updating try to trigger the problem and
provide lvmetad backtrace. This would show where it is stuck.

Find out lvmetad PID (e.g. using "pgrep -l lvmetad") and then use gdb to obtain
backtrace with "thread apply all backtrace", like

bor@opensuse:~> pgrep -l lvmetad
557 lvmetad
bor@opensuse:~> sudo gdb -p 557
root's password:
GNU gdb (GDB; openSUSE 13.1)
Copyright (C) 2013 Free Software Foundation, Inc.
(gdb) thread apply all backtrace

Thread 1 (Thread 0x7f23a2ff3800 (LWP 557)):
#0 0x00007f23a20b9913 in select () at ../sysdeps/unix/syscall-template.S:81
#1 0x0000000000406332 in daemon_start ()
#2 0x0000000000403112 in main ()

Note that gdb will most likely warn you that debugging information is missing
and will suggest zypper commands to install it. You will need to have debug
repos configured for it. Please install all requested debugging packages before
generating backtrace.

Configure bugmail:
------- You are receiving this mail because: -------
You are on the CC list for the bug.

< Previous Next >