Re: Help tracking down cause of Hunchentoot crashes

The only logs I am aware of that might be relevant are the message and
access logs that Hunchentoot writes, and kern and syslog in
/var/log/. The latter two show nothing interesting over the interval
in which it most recently died. The last access log entry was days
after the last message log entry. Is there another log I don't know
about?

I had an emacs instance connected to the running server through
slime-connect since forever. In the past, when something bad happened,
it would have entered the debugger. In this case the slime connection
was simply closed. No debugger. No *Messages*. Nothing.

I turned on gc logging today, in case that might shed some light. 



On Thu, 10 Apr 2025 19:49:30 +0300
Stas Boukarev <[email protected]> wrote:

> It can't die silently, there has to be something in the logs.
> 
> On Thu, Apr 10, 2025 at 7:10 PM Jeff Cunningham
> <[email protected]> wrote:
> >
> > I'm trying to debug an infrequently occurring crash of a Hunchentoot
> > server. The logs show nothing and there are no error messages. It's
> > died three times in the last week after running continuously for
> > months before that.
> >
> > I suspect it might be a heap-stack collision and have increased the
> > dynamic-space-size of the core image accordingly. It might be helpful
> > if I could periodically log the current status of the stack and heap
> > in some way. Is this possible? It might help me rule out this as the
> > problem or otherwise provide some hints as to what's causing it.
> >
> > I'd appreciate any suggestions.
> >
> > -- Jeff
> >
> >
> >
> >
> >
> >
> >
> >
> > _______________________________________________
> > Sbcl-help mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/sbcl-help  



_______________________________________________
Sbcl-help mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sbcl-help