[Bug 832537] New: Home partition freezes in kernel?
https://bugzilla.novell.com/show_bug.cgi?id=832537 https://bugzilla.novell.com/show_bug.cgi?id=832537#c0 Summary: Home partition freezes in kernel? Classification: openSUSE Product: openSUSE 12.2 Version: Final Platform: x86-64 OS/Version: openSUSE 12.2 Status: NEW Severity: Critical Priority: P5 - None Component: Kernel AssignedTo: kernel-maintainers@forge.provo.novell.com ReportedBy: dfreeman@ieee.org QAContact: qa-bugs@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:22.0) Gecko/20100101 Firefox/22.0 This has been happening to me for a while, and I've recently narrowed down the number of possible causes. Still I recognise that it's probably too vague to fix without further investigation, and I request some possible things to try. The typical scenario goes like this. Either my KDE desktop freezes completely, and I can't click on other tasks in the task bar, or I'm saving a file with Firefox, having saved many other files, and this time it freezes on the file dialogue. I then switch to a virtual console and log in. Things seem fine, but some user applications are listed as defunct when I run 'top'. But the shell will freeze when I run certain commands, and I'll have to switch to another virtual console and log in again. For example, 'lsof' will freeze. I have narrowed this to anything that tries to modify my /home partition - 'echo >/home/foo.bar' won't return. So my defunct applications are dead and unable to update their state on disc - just as if I'd done a hard reset on them. I'm a little worried about my mail client and PIM losing information this way. I should note that very lightweight applications that don't need to write to /home will continue working, although the Plasma desktop is dead so I can't interact with them. I just have to walk away and ignore my dead computer until those applications are finished and I'm ready to reboot. I try to shut-down cleanly with 'shutdown', but after long delays, the system reboots and then has to replay a lot of transactions from the journals. My configuration is roughly as follows. openSUSE 12.2 on Core 2 Duo, 64 bit version. /home is formatted ReiserFS v3, on LVM2, on LUKS, on a SSD. The other partition on the same volume group, root, is not affected, although it is also ReiserFS v3. I also have swap on this vg, and it does get used. I once thought that this issue crept in slowly over time with usage - typically after a few days of uptime. But I've seen it happen five minutes after I've had to reboot. This was due to me concurrently saving a lot of files in Firefox. It might be that Firefox is writing multiple files in a funny way, with lots of incremental small writes, or it might be the way that Firefox keeps a consistent on-disc state that can be recovered from. Maybe every time I save a file it overwrites the same sectors on the disc to update the details of the downloading files. All I know is that I'm having trouble diagnosing this. I don't see any errors in the logs or from 'dmesg'. It's only been happening for six months or so, which may correspond to the last time that I updated openSUSE (from 12.1) as I migrated onto an SSD. A couple of days ago I did a low level format on the logical volume, verified that I can read back all zeroes (after decryption), and reformatted ReiserFS v3 and restored all the files from backup. It's already happening again. So please suggest something that I can try from console when it happens again. Please focus on anything that might have been changed in the kernel that could cause this - e.g. the new file-system freezing features. Or, if you know of problems with SSDs that don't show up in the logs, please let me know. It looks to me like a bug in ReiserFS.. but as development has been mostly dead for some time, I'm amazed that something that worked flawlessly for so many years should now be unstable. Reproducible: Sometimes Steps to Reproduce: 1. Save lots of files in Firefox 2. KDE environment freezes 3. Go to root console and verify that writing to /home freezes the process -- 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=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c1
Jeff Mahoney
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c2
--- Comment #2 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c3
--- Comment #3 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c4
Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c5
Jeff Mahoney
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c6
--- Comment #6 from Jeff Mahoney
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c7
Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c8
Jeff Mahoney
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c9
Marcus Meissner
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c10
--- Comment #10 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c11
--- Comment #11 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c12
--- Comment #12 from Marcus Meissner
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c13
James Knott
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c14
Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c15
--- Comment #15 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c16
--- Comment #16 from James Knott
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c17
--- Comment #17 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c18
--- Comment #18 from James Knott
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c19
--- Comment #19 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c21
--- Comment #21 from James Knott
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c22
--- Comment #22 from Darren Freeman
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c23
--- Comment #23 from James Knott
https://bugzilla.novell.com/show_bug.cgi?id=832537
https://bugzilla.novell.com/show_bug.cgi?id=832537#c24
--- Comment #24 from James Knott
http://bugzilla.novell.com/show_bug.cgi?id=832537
Takashi Iwai
I haven't had a lock-up for some time now, and I'm not using ReiserFS on /home any more. But I/O performance is still pretty terrible.
If I copy a large file from an NFS mount to a USB stick, for example, my system is unusable. KDE is just dead until it finishes, with large regions of the screen waiting to redraw, but all my command-line stuff works fine if I switch to a virtual console.
I would love an explanation for this, but I'm not sure that it's the same bug. It never used to be this way.
The I/O stall by a bulk copy to USB stick is infamous problem. You might be able to work around it by tweaking the cache size, though. In anyway, this should have been already addressed in the recent kernels. Please try the recent upstream kernel found in OBS Kernel:stable. It should work even with openSUSE 13.1 user-space. If the very same problem is still seen with the recent kernels, it's still worth to track. -- You are receiving this mail because: You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=832537
--- Comment #27 from Darren Freeman
http://bugzilla.novell.com/show_bug.cgi?id=832537
http://bugzilla.novell.com/show_bug.cgi?id=832537#c28
--- Comment #28 from Darren Freeman
http://bugzilla.novell.com/show_bug.cgi?id=832537
Darren Freeman
participants (1)
-
bugzilla_noreply@novell.com