Re: Plese help with Fatal error

To make sure everybody knows what I normally do:
- I compile linux + ac-patch
- I compile glibc 2.1.1 against the new kernel
- I change /usr/include/bits/types.h

I also run a 3000+ irc-server on a machine compiled like this (ircd is
compiled to use poll() ) and ircd never has a problem.

Dirk Moerenhout ///// System Administrator ///// Planet Internet NV

On Wed, 14 Jul 1999, Dirk Moerenhout wrote:

> My Squid 2.2 Stable 4 (and previous versions) on linux 2.2.10ac4 with
> (rebuild) glibc 2.1.1 has the problem too. Actually it does it several
> times a day. I haven't found a cause yet. I am trying to find the time to
> diagnose it. But as the thing is running in production I can't take it off
> line. For now we live with it as it brings only a few seconds of downtime.
>
> I also experience still a lot of leakage too (even with Stable 4) which I
> plan on looking into as soon as I can. One thing I might try is to try and
> use select() again instead of poll but I don't know if ac-patches allow
> this yet. In early 2.2.x-acx they did not.
>
> Dirk Moerenhout ///// System Administrator ///// Planet Internet NV
>
> On Tue, 13 Jul 1999, Clifton Royston wrote:
>
> > Flavio Pescuma writes:
> > > I tried without asynchronous I/O and the FATAL: Received Segment
> > > Violation...dying still exists. I think the problem began when I
> > > compiled with --enable-poll to be able to use more than 1024 fds.
> > >
> > > Do you have any other idea?
> >
> > Did you completely rebuild the kernel, *and* rebuild the C libraries,
> > after changing the kernel to allow more than 1024 fds?
> >
> > This sounds to me like the application user-space code may have some
> > data structure used with poll() defined too small for the modified OS.
> > At a guess, I'd suspect the C system libraries. In that case, the
> > problem would be replicateable by anyone else who did a Linux kernel
> > rebuild with this mod and failed to recompile the libraries.
> >
> > -- Clifton
> >
> > --
> > Clifton Royston -- LavaNet Systems Architect -- cliftonr@lava.net
> > "An absolute monarch would be absolutely wise and good.
> > But no man is strong enough to have no interest.
> > Therefore the best king would be Pure Chance.
> > It is Pure Chance that rules the Universe;
> > therefore, and only therefore, life is good." - AC
> >
>
Received on Wed Jul 14 1999 - 03:25:32 MDT