Feisty crash possibly related to mdadm/raid5

Could you please give some more information on the Crash? Did your computer while booting? Or was your system up and running? After fixing Bug 107080 on my System, it has run normally aside from bug 103603 which is not connected to my RAID problems.

Sorry. This means: "Did your computer crash while booting?"

It crashed after having been up for atleast some hours. (crashed while I was sleeping) The raid array was mounted, but as I said I don't really know if it's related or not.

The Output of mdadm --query --detail shows that your RAID is currently rebuilding (Rebuild Status : 4% complete). The UUID and the Status of the drives seems to be OK. It's hard to guess what the reason for the crash was.

Yes I know everything seems fine now, but at boot it refused to start, I had to manually re-add one of the drives and mdadm --run it.

The crash has happened twice. I'm currently checking if it also occurs with the old kernel (2.6.17-11)

The reason I upgraded in the first place was to prevent data corruption on raid5 with cryptsetup, as reported for kernels < 2.6.19.

Same problem here. RAID-5 array made from 3 partitions (not whole disks) sda3 + sdb2 + sdc1 randomly sets one of the devices "removed". After rebuild everything's fine. Takes about one day to strike again. Not figured out yet, if it corelates with disk stress.

I can confirm that everything works as it should on 2.6.17-11

I haven't come any closer to figuring out what happens on 2.6.20-15 and 2.6.20-16. Sometimes it crashes on mount, sometimes it can work for hours.

Is this a kernel crash? (i.e. the system is hung?) If so, can you capture the crash output and attach it to this bug? Without more information, it's not clear how to proceed with reproducing or diagnosing this bug. Thanks!

Beeing somewhat new to Linux I'm not sure what constitutes a kernel crash. Everything locks up completely, sysrq does not work. No output to the monitor.

G.Koehler. are you able to catch any portion of the kernel crash? It will be hard to move forward with this bug without some way to reproduce it or see specifically what areas are failing. Thanks!

Well, the crazy thing is, that it did not happen again. Not even under heavy stress, which led to crashes before.
The only explanation I can find is that maybe it's gone since I downloaded and installed all available updates, as ubuntu suggested.
I had to take this chance to get going with a project and so from that perspective I am glad it worked.
On the other hand this is not very usefull for tracing back a problem - sorry.
Would be interesting if I am the only one who "solved" the problem that way...
So in this situation I think we should leave it as it is, should'nt we ?
Being in the troubleshooting business for quite a while, this hurts me a little, as it is NOT the professional way to solve such problems.
If I can find ANY useful hint, I'll let you know for sure.

Thanks and best regards from Austria.
Gerald

-----Ursprüngliche Nachricht-----
Von: <email address hidden> [mailto:<email address hidden>] Im Auftrag von Kees Cook
Gesendet: Montag, 18. Juni 2007 18:15
An: SOFTMILL - Ing. Gerald Köhler
Betreff: [Spam Gefunden!] [Bug 110304] Re: Feisty crash possibly related to mdadm/raid5
Wichtigkeit: Niedrig

G.Koehler. are you able to catch any portion of the kernel crash? It will be hard to move forward with this bug without some way to reproduce it or see specifically what areas are failing. Thanks!

--
Feisty crash possibly related to mdadm/raid5
https://bugs.launchpad.net/bugs/110304
You received this bug notification because you are a direct subscriber of the bug.

I'm going to close this bug for now, since there doesn't seem to be a good way to diagnose (or reproduce) this problem at the moment. Thanks for everyone's input. If a backtrace from the kernel or a series of steps to reproduce this problem become available, please feel free to reopen the bug. Thanks!