Qube2 Freezing

yarrumretep · n00b Joined: 06 May 2006 Posts: 8

Greetings!

I have successfully installed Gentoo on my mips based Qube2. Everything works pretty well - I have it controlling several X10 devices via a CM11a via the SerialPort. However, occasionally (well, lately, daily) the box has been freezing - no network, no switching my X10 devices. Reboot is required. I can't find any logs which indicate any problems. Can someone give me some tips on debugging this kind of situation? The serial port is taken over by the X10 stuff - so I can't connect a terminal to see if that would work - but I assume the box is locked if the scripts doing the X10 switching aren't running (confirmed by their logs) and the network interface isn't responding to pings or anything.

Any tips, thoughts, help?

Cheers,

pete

yarrumretep · n00b Joined: 06 May 2006 Posts: 8

uname -a
Linux cobalt 2.6.15.7-mipsgit-20060109 #4 Sun May 7 01:52:32 EDT 2006 mips Nevada V10.0 FPU V10.0 Cobalt Qube2 GNU/Linux
c

Redhatter · Posted: Wed May 24, 2006 2:17 am Post subject:

This is unusual... Have you tried updating the kernel? There's a 2.6.16.4 image in my devspace (see the website link below) you can try, I know it's stable -- not sure if it'll have the drivers needed for your X10 equipment though. It's also a 64-bit kernel.

A possibility, perhaps a PCI serial card could be used for your X10 equipment? That would allow for capture of oopses via the serial console. I haven't tested this however, I'm not sure what PCI RS232 controllers would work on the Qube2.

For what it's worth though:

yarrumretep · n00b Joined: 06 May 2006 Posts: 8

Thanks, Redhatter, unfortunately, I've already got a wifi pci card in the pci slot. The X10 stuff doesn't use any drivers - just talks through ttyS0. All else I have running is

lighttpd
php (fastcgi)
sshd
dhcpd
heyu (X10 stuff)
syslogd
atd
ntpd
udevd

The wifi card is using the madwifi driver.

There is 256 MB RAM in the box (could it be a bad stick of ram?) / 512MB swap - should be enuf RAM. Normal operation indicates only 60MB RAM used and no swap - so it doesn't seem like an out of memory problem (I've seen that lock up linux as it tries to kill off processes to make room).

And a brand new 100GB disk.

Is there any way to tell the kernel to log to the disk if it has problems? Point the console to a disk file?

Thanks!

-pete

Sanguinary · n00b Joined: 17 Oct 2004 Posts: 16

Redhatter · Posted: Wed May 24, 2006 11:32 pm Post subject:

yarrumretep: Well, 256MB RAM should definately be enough... I run 128MB in mine... and with a PCI SCSI card, it runs quite well.

syslogd can do remote logging... perhaps it's worth trying that? That, and perhaps netconsole? I'm no authority on setting this up though... Peter Horton would be your best point of reference, he'd know how to go about setting up netconsole. Also have a look at the netconsole docs.

If you do figure out how to get it going, please, let me know... I'm looking at setting something up for the Cobalt netboot images. :-)

Sanguinary: Heh, mine's on its second PSU, having blown up the original. 12v PSUs are pretty easy to come by... if you're stuck, grab an AT PSU, hack one of the molex connectors off and use the Yellow and Black wires to power the Qube2. This is what I did before getting a replacement PSU. :-)

_________________
Stuart Longland (a.k.a Redhatter, VK4MSL)
I haven't lost my mind - it's backed up on a tape somewhere...

Gentoo/MIPS Cobalt developer, Mozilla herd member.

yarrumretep · n00b Joined: 06 May 2006 Posts: 8

Thanks for the thoughts. Assuming the problem is power related - can I buy/make a beefier power supply? Will that help matters? I assume if the device is drawing too much power, the voltage drops below tolerances for the cpu and it stops? If that's the case (seems reasonably likely as the hard-drive is full-sized and much bigger than the one that was in there and wi-fi probably takes some juice), I probably wouldn't see any kernel messages anyway - the thing would just stop (which seems like what is happening).

It's now happening a couple of times a day, so this is untenable as it stands.

Cheers,

pete

Redhatter · Posted: Fri May 26, 2006 1:20 am Post subject:

I found the power supply would start cutting out unexpectedly, causing the whole machine to reboot. This was the bare machine with no peripherals. Basically, the power supply was being overloaded, and so the protection circuit cut in, killed the power.

If the machine looses power momentarily, this could cause data in CPU registers to become scrambled, and thus causes a hard lock.
_________________
Stuart Longland (a.k.a Redhatter, VK4MSL)
I haven't lost my mind - it's backed up on a tape somewhere...

Gentoo/MIPS Cobalt developer, Mozilla herd member.

yarrumretep · n00b Joined: 06 May 2006 Posts: 8

I've spliced in a beefier 12v power supply from an old laptop that is now defunct. I'm hoping that does the trick. I'll report back with my findings. If that doesn't work, I'll try getting the extra logging set up.

Rock on,

pete

yarrumretep · n00b Joined: 06 May 2006 Posts: 8

Update: