Lockups with blinking keyboard leds

[Guest blaze]

My new Mandrake 9.1 installation crashed twice today, leaving me staring at frozen display and blinking caps and scroll lock leds. Nothing aside from reset button seemed to help, no Ctrl+Alt+Fn, no Ctrl+Alt+Backspace, no Ctrl+Alt+Del. First time I was when I was happily IRCing and surfing with Mozilla, the second time when I was as happily watching TV in the other room. Any ideas on what might be the cause? What are the blinking leds trying to tell me?

 

An IRC buddy hinted me it might be the hardware. Surely my Win2k installation would be crashing too, if my hardware was faulty, so perhaps it's a (in)compatibility issue? I have AthlonXP 2400+, MSI KT400 Ultra mobo (which has some C-Media sound chip integrated), Radeon 9500 Pro, 3Com NIC and a TeleWell ISDN card. Am I stuck with using Windows with this configuration?

 

All suggestions welcome.

[Steve Scrimpshire]

The blinking LEDs just mean you had a kernel oops, probably. See if you can find any mention of a kernel oops in /var/log/messages. It should give you a hint of what process caused it. You'll see a buncha memory address numbers and stuff, but somewhere in there you should se something like 'process BLAH'.

[kmack]

You might take a look at the 9.1 errata as there is a ref there to this problem. Not sure it applies to you, but worth a look.

 

http://www.mandrakelinux.com/en/errata.php3

 

Also, 9.1 has LOTS of updates so you need to run Mandrake Update and get the bug fixes and patches which may cure your problem. The new kernel may be the ticket too.

 

I think your hw is OK, it is probably a bug that needs fixed.

[Steve Scrimpshire]

The only thing I see in the errata related to this looks like install problems.

[SoulSe]

Follow Steve's directions for finding the process which caused the crash. If, however, you do not find the problem, try using a different WM and see if it still occurs.

 

My machine used to freeze in Gnome, couldn't find the miserable little process that caused it, but once I switched to Fluxbox, it dissapeared.

[kmack]

Sorry if my pointer to errata wasn't too helpful since it deals with install issues. Steve and SoulSe have lots more experience at troubleshooting than I do. My thought was the memory issue in 9.1 with large amts of RAM.

 

In the past, I've had a few crashes like you described too, and the var/log/messages log showed nothing at all since it crashed so hard the log didn't save any indication of the problem. That is irritating when it happens and makes it hard to find the problem. If that is the case with your crash I suggest this: check your memory

 

I had an intermittent problem with kernel panics due to a bad memory card. My computer worked great in Windows, but had kernel panics like you describe at intermittent times. It was hard to troubleshoot since it did not correlate to particular activity and the logs showed nothing wrong. I ran memtest86 and other software checks and things showed fine. But when I finally removed all the memory cards and only put one memory card in at a time. it narrowed it down to one memory card that Linux did not like. The system would not boot using only that card, but did fine on all the others.

 

You might try removing your memory and reseating the cards just to make sure they are ok. If you continue to have problems, it might be worth doing the memory swap to see if that isolates a card that is the problem. Crucial replaced my card under warranty by the way!

 

Hopefully your log will show something and your memory is OK and you won't need to do this.

:)

[Steve Scrimpshire]

Well, don't rule out the 1+ Gig of RAM issue totally. Did it set you up with the Enterprise kernel? Doing

 

uname -r

 

should tell you.

[Guest blaze]

See if you can find any mention of a kernel oops in /var/log/messages. It should give you a hint of what process caused it. You'll see a buncha memory address numbers and stuff, but somewhere in there you should se something like 'process BLAH'.

 

Can't see anything like that, before the second crash however, there's a longish list of stuff like lines below:

 

Jun 16 21:12:18 sakura2 smbd[4198]: [2003/06/16 21:12:18, 0] lib/fault.c:fault_report(38)

Jun 16 21:12:18 sakura2 smbd[4198]: ===============================================================

Jun 16 21:12:18 sakura2 smbd[4198]: [2003/06/16 21:12:18, 0] lib/fault.c:fault_report(39)

Jun 16 21:12:18 sakura2 smbd[4198]: INTERNAL ERROR: Signal 11 in pid 4198 (2.2.7a)

Jun 16 21:12:18 sakura2 smbd[4198]: Please read the file BUGS.txt in the distribution

Jun 16 21:12:18 sakura2 smbd[4198]: [2003/06/16 21:12:18, 0] lib/fault.c:fault_report(41)

Jun 16 21:12:18 sakura2 smbd[4198]: ===============================================================

Jun 16 21:12:18 sakura2 smbd[4198]: [2003/06/16 21:12:18, 0] lib/util.c:smb_panic(1094)

Jun 16 21:12:18 sakura2 smbd[4198]: PANIC: internal error

 

So samba's not behaving? I'll try disabling it. I don't think I need right now anyway.

 

Nothing like that before the first crash though, just some stuff about modem hangups (courtesy of bandwidth-on-demand perhaps?)

 

Sorry if my pointer to errata wasn't too helpful since it deals with install issues.

 

No problem, I had already been there anyway. That page turned up in a Google search.

 

You might try removing your memory and reseating the cards just to make sure they are ok.  If you continue to have problems, it might be worth doing the memory swap to see if that isolates a card that is the problem.

 

Good idea, somewhat difficult to do though, since I have only one memory module that fits in this 'puter.

I'll keep that in mind in case nothing else seems to help.

 

I think I'll try to crash it once more and see if it manages to leave any more hints in the log. I'll need to try some other WM too at some point (using KDE currently).

[Steve Scrimpshire]

Well, it may not be Samba that is the problem. It could just be a symptom. It wouldn't hurt to disable it, though, if you're not using it. You can always re-enable it later when you need it.