[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Intermittent System Lockup (long)



On Wed, 17 Apr 2002 23:58:41 -0700
"Karsten M. Self" <kmself@ix.netcom.com> wrote:

> on Wed, Apr 17, 2002, Jamin W. Collins (jcollins@asgardsrealm.net) 
> > Created two copies of the 2.4.16 kernel source and have currently been
> > running two endless compiling loops in SSH sessions to the system. 
> > The loops have been running for 4+ hours now, planning to let them run
> > over night.
> 
> Sounds like a negative.

Yea, ran through all night long (~10 hours) until I stopped it and
remounted all the partitions sync'd for the strace on Mozilla.

> > > You might try mounting your drives 'sync' (synchronous mode), and
> > > launching Mozilla under strace, logging stderr.  This may be able to
> > > capture the final system calls of the program.
> > 
> > I'll give this a run tomorrow.  Hopefully I can readily get Mozilla to
> > drop the system.
> 
> You've largely eliminated memory and CPU.

Well, running the system with sync'd mounts and strace for Mozilla was
extremely slow, but I did manage to get the system to drop while using
Mozilla (during an attempted opening of freshmeat.net).  I ran strace with
'-ff -F -o moz.log' options and now have several trace output files.  It
appears to have managed to follow a good number of threads from Mozilla.

> Another possible HW problem might be a disk corruption in your swap
> partition.  I suggest this just because I now that Mozilla tends to
> grow, and stress swap.  Though I would tend toward a driver issue.  Not
> sure of a good swap tester, anyone have any suggestions?

This could be more or less ruled out by turning swap off during one of the
tests, right?  I know it wouldn't prove that the swap was good, but if the
system drops without the swap enabled, then we know something else is to
blame (swap could still be bad).  However, if the system won't drop with
the swap disabled, then I would tend to think the swap was the culprit,
right?

-- 
Jamin W. Collins


-- 
To UNSUBSCRIBE, email to debian-user-request@lists.debian.org 
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org



Reply to: