Re: panic with heavy writing on md device

From: Brian Fundakowski Feldman <green_at_freebsd.org>
Date: Tue, 29 Jun 2004 19:52:21 -0400
On Tue, Jun 29, 2004 at 11:11:51PM +0200, Ronald Klop wrote:
> Hello,
> 
> I got a panic while doing a lot of IO on a md device.
> I attached a gdb backtrace of the kernelcore and a dmesg output.
> 
> It happened while untarring a 30 MB tar on the background (&) and
> traversing this new directory in the meantime.
> 
> uname -a
> FreeBSD guido.thuis.klop.ws 5.2-CURRENT FreeBSD 5.2-CURRENT #1: Fri Jun 25  
> 04:03:35 CEST 2004      
> root_at_guido.thuis.klop.ws:/usr/obj/usr/src/sys/GUIDO  i386
> 
> On the same time I was playing music with mpg123, but that doesn't make  
> use of the md device.
> 
> I have had this more and always while doing a lot of IO on the md device.
> 
> Is there more info you need?

It looks like there is a memory leak -- in general, you are not supposed
to run out of kernel virtual memory address space unless you have a
dire misconfiguration somewhere.

Unfortunately, I think the most useful pieces of information (zone and
malloc allocation) are not fetchable via kvm anymore.  Here's a fix for
vmstat so that you can provide vmstat -m output using the -M and -N
args again.  I think vmstat -c output would also be useful.

-- 
Brian Fundakowski Feldman                           \'[ FreeBSD ]''''''''''\
  <> green_at_FreeBSD.org                               \  The Power to Serve! \
 Opinions expressed are my own.                       \,,,,,,,,,,,,,,,,,,,,,,\

Received on Tue Jun 29 2004 - 21:52:23 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:59 UTC