[mythtv-users] OT. Howto diagnose machine dying/crashing

Mark mark at ripitup.ca
Fri Sep 15 17:50:13 UTC 2006


On Friday 15 September 2006 11:44, Tim Berger wrote:
> Let me take a wild guess - you're using xfs, aren't you?
>
> There's going to be a lot of people on this list with high praise for xfs,
> but on a 2.6 kernel with 4k stacks, there's a high likelihood of oopsing
> the kernel, especially when layered with raid and nfs.  ext3 works just
> fine with myth, and it's *stable*.
>
> http://oss.sgi.com/archives/xfs/2005-07/msg00041.html
>
> In any event, netdump is a great tool to capture an entire oops message
> with these sort of lockups, if it indeed it's oopsing.
>
> On 9/15/06, Mark <mark at ripitup.ca> wrote:
> > Combined NFS server / mythfront end box. Mythbackend on another machine
> > uses
> > the NFS.
> >
> > I have a 1TB software raid ( 4 x 320 IDE drives ) and a single sata drive
> > (
> > FC5 OS is on it ).
> >
> > About every week or 10 days, this box just up and dies.  No
> > /var/log/messages
> > entries ( you just see the reboot ), nothing on the screen etc...
> > I have a 410 watt power supply, a couple of fans, it seems cool, low load
> > etc....
> > I am not sure what could be causing this and I am getting no info from it
> > that
> > helps me diagnose it further.  SMART tests on the disks show all OK.  It
> > does
> > seem to be load related because I have reproduced it once by transfering
> > a hundred or so gigs of data into it and it died then.
> > FC5, mdadm raid..
> > The RAID is broken into one drive per bus ( onboard primary and secondary
> > and
> > an IDE PCI card with pri and sec as well ) I use to have all 4 drives on
> > the
> > master/slave of the built in IDE ports, but broke them up into their own
> > bus
> > as I thought that might be causing this.
> >
> > Any thoughts on what to try, look for, etc....?
> >
> > Thanks!
> >
> > Mark
> >
> > _______________________________________________
> > mythtv-users mailing list
> > mythtv-users at mythtv.org
> > http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
Interesting.  Actually you are right, I am using XFS.  I wonder if this could 
be the reason.  I am unable to switch now to JFS or EXT3 as I have 700Gigs of 
data on it.  Hmmmm.  I am trying to get a crashdump to no avail. I have been 
trying to setup netdump and I keep getting this error ( yes I have setup the 
server and client with NETDUMPADDR=192.168.0.2 ) but here is what I get one 
start of netdump client:
netdump: cannot arp  on eth0
netdump: cannot find 192.168.0.2 in arp cache
netdump: can't resolve 192.168.0.2 MAC address
netdump server address resolution                          [FAILED]

I have DEV=eth0 set as well which is correct.
Ideas?

Better ways to catch what is happening on a crash?




More information about the mythtv-users mailing list