On Tue, 5 May 1998, Brian Behlendorf wrote:
> At 03:33 PM 5/5/98 +0200, you wrote:
> >On Mon, May 04, 1998 at 03:34:02PM -0700, Brian Behlendorf wrote:
> >> What architecture is this on?
> > i386-sni-sysv4
> >
> >> How much traffic are you seeing?
> > +/- 500 ... 1500 reqs/day
> > +/- 1 ... 20 MB/day
>
> Do you think you could try replaying the requests which led to the core
> dump? Ideally there'd only be a couple hundred between the time you
> restarted and when you got a core dump. Of course this discounts the
> possibility of some other request headers causing the stack corruption; if
> you don't get a core dump when you replay the requests then maybe capture
> the IP traffic is necessary... eek. This is bugly.
We have some code to replay 'logfiles' including even some limited IP
source spoofing; i.e. mimic a handfull of clients from a single box with
some limited reverse dns support. Would that help in any way to recontruct
?
Dw
|