At the point of the hang the following messages are on the console:
Losing too many ticks!
TSC cannot be used as a timesource. Possible reasons for this are:
You're running with Speedstep,
You don't have DMA enabled for your hard disk (see hdparm),
Incorrect TSC synchronization on an SMP system (see dmesg).
Falling back to a sane timesource now.
Running 'date' repeatedly on the machine shows that the system clock is totally broken at this point.
I've tried with kernel-image-2.6.8-1-686, kernel-image-2.6.8-1-386 (which does not afaik use TSC), and my own compiled version against kernel-source-2.6.8 - same problem with all, with the exception that the 386 kernel did not bring up the console message - it just hung.
Kernel 2.4 works fine, but unfortunately I have a seperate issue with VMware that means I can't run with it in that config. Hey ho.
There was a thread on the fedora 2 mailing list on the redhat site regarding this same issue which suggested running without SMP support or specifying 'acpi=noht' (to disable hyperthreading) - neither of these fixes worked for me.
I'm going to try 2.4.10rc2 to see if that helps - more later.
Regards,
mike
====
Re: Computer goes unacceptably slow
Kent West Mon, 22 Nov 2004 14:08:03 -0800
Tapio Lehtonen wrote:
I have a computer class with 13 IBM ThinkCentre computers, running Debian GNU/Linux Sarge since August.
Do you mean you're the admin for these 13, or that you're a student and are just trying to diagnose the problem?
Right from the start they go to a strange state where computer is extremely sluggish. Looks like any command takes about a minute to execute. This happens randomly, at least it looks random to me. Sometimes it happens in less than half an hour. It looks to me that if the computer is up, it eventually goes to this state before 8 hours.
Did you install these machines individually, or did you clone them? If you cloned them, do they all have their own hostname? Is /etc/hosts properly configured?
What if you kill the network on one of the machines? Does the problem still occur?
Any cron jobs running?
How much swap space do you have? Is it turned on?
Are the perms correct on /tmp?
Any NFS sharing going on?
Google found on other report of what seems the same problem: http://groups.google.fi/groups?hl=fi&lr=&threadm=2K35l-3D9-27%40gated-at.bofh.it&rnum=1&prev=/groups%3Fq%3Dacpi%2Bibm%2Bthinkcen%26hl%3Dfi%26btnG%3DGoogle-haku <http://groups.google.fi/groups?hl=fi&lr=&threadm=2K35l-3D9-27%40gated-at.bofh.it&rnum=1&prev=/groups%3Fq%3Dacpi%2Bibm%2Bthinkcen%26hl%3Dfi%26btnG%3DGoogle-haku> I found this searching for "acpi ibm thincen" from newsgroups. That report says the problem is not present with 2.6.6 kernel, but is there in 2.6.7 and later kernels. I have tried 2.6.7, and 2.6.8 debian version -1 and -5. Now I have Sarge on the hosts, updated on 2004-11-21 to the current Sarge.
Wait; you say the problem is not present with 2.6.6, but is with 2.6.7; then you say you've tried 2.6.7. Did you expect the problem to go away when you tried the known-to-have-problems kernel?
Can you disable ACPI in the BIOS, or in Debian itself (I'm ACPI ignorant)?
Have you tried Knoppix (or similar LiveCD/USB Flashdrive) in one of the machines to see if it also exhibits the problem?
-- Kent
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
