kernel panics at google

[yoann en mandrakesoft.com]

David desJardins <desj en google.com> writes:

> Update: We (Google) set /proc/sys/net/ipv4/tcp_retrans_collapse to "0"
> on our webservers.  Now, instead of a kernel panic, they seem to
> spontaneously reboot without any errors or explanation in
> /var/log/messages.  For us, this is a significant improvement over the
> previous sitation: at least we don't have to manually reboot them.  But
> it still leaves unanswered the question of what is causing it.
> 
> It also seems that the reboots are now happening significantly less
> often than the crashes did.
> 
> Any suggestion on how we can collect more information about why the
> machines are rebooting?
> 

Maybe giving a trace of the kind of packet received before the crash
could help understanding / tracing the problem...


tcpdump -s2000 -vvv -w outfile
( make sure no private data are passing on the network ).

Also, you should modify tcpdump call fflush() after each
line it append to the file.

-- 
		-- Yoann http://www.security-addict.org
 It is well known that M$ product don't make a free() after a malloc(),
the unix community wish them good luck for their future developement.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo en vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/