[distcc] Re: More debugging info for FIN_WAIT1 bug with RH 6

Martin Pool mbp at samba.org
Fri Sep 6 00:53:01 GMT 2002


On  5 Sep 2002, "Hien D. Ngo" <hien at moses.xp.com> wrote:
> 
> Turning off tcp_cork seems to make the problems under RH 6 go away; no more 
> FIN_WAIT1 connections, netstat is completely clean.  Turning tcp_cork off also works 
> for me on the RH 7 machines, so I can happily build with distcc now on all my 
> machines :)  

OK, that's good.  Things should work fine without corks.  They may
just give a performance boost under some circumstances.

> BTW, the FIN_WAIT1 connections never die off, they just keep
> accumulating as far as I can tell.

That's interesting.  In a way that makes sense: if there is no timer
on the socket to clean it up, then we should expect them to remain
forever.  It's additional evidence for the hypothesis that the kernel
really is failing to set the timer, rather than (say) netstat just
failing to report it.

Even though distcc is now working, I would really appreciate it if you
would try to get a tcpdump of reproducing the problem.  It really
looks like it is a kernel bug and I would like to characterize it and
pass it on to the TCP maintainers.

Thanks!
-- 
Martin 



More information about the distcc mailing list