[gentoo-user] Distcc borkage: Connection reset by peer

[gentoo-user] Distcc borkage: Connection reset by peer

[PaulNM]

Hi all,

 Distcc seems to not be working properly on one or all of my 3 systems.
I have :
Galactica 192.168.1.22 Pentium 2 (Problem system)
Robotech  192.168.1.40 Pentium 3
Optimus   192.168.1.80 AMD64 AthlonX2 (crossdev setup)

 All are Gentoo with distcc installed and configured. Optimus is solely
a helper system to the other two as far as distcc is concerned. If I
emerge something on Robotech, both Robotech and Optimus are used, but
Galactica seems to be ignored with no error messages. When I try on
Galactica, I get numerous
"distcc[910] (dcc_writex) ERROR: failed to write: Connection reset by peer
distcc[910] Warning: failed to distribute ckkv.c to 192.168.1.22,
running locally instead"
messages, also listing the other two ip's 40 and 80.

 "cat /etc/distcc/hosts" on all three results in "192.168.1.80
192.168.1.40 192.168.1.22".  They all have "sys-devel/distcc-2.18.3-r10"
and "/etc/conf.d/distccd" modified the same way :
DISTCCD_OPTS="${DISTCCD_OPTS} --allow 192.168.1.40 192.168.1.80
192.168.1.22"
Distccd was restarted after any config changes.

 All three have i686-pc-linux-gnu toolsets/chosts, and as far as I can
tell should work. No firewalls are in place for testing, distccd port is
open on all three (confirmed with nmap), and I'm out of ideas on where
to look for problems.

Any help or suggestions would be appreciated.

PaulNM
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Distcc borkage: Connection reset by peer

[Dan Farrell]

On Tue, 05 Jun 2007 06:15:25 -0400
PaulNM <gentoo@paulscrap.com> wrote:

> Hi all,
> 
>  Distcc seems to not be working properly on one or all of my 3
> systems. I have :
> Galactica 192.168.1.22 Pentium 2 (Problem system)
> Robotech  192.168.1.40 Pentium 3
> Optimus   192.168.1.80 AMD64 AthlonX2 (crossdev setup)
> 
>  All are Gentoo with distcc installed and configured. Optimus is
> solely a helper system to the other two as far as distcc is
> concerned. If I emerge something on Robotech, both Robotech and
> Optimus are used, but Galactica seems to be ignored with no error
> messages. When I try on Galactica, I get numerous
> "distcc[910] (dcc_writex) ERROR: failed to write: Connection reset by
> peer distcc[910] Warning: failed to distribute ckkv.c to 192.168.1.22,
> running locally instead"
> messages, also listing the other two ip's 40 and 80.
> 
>  "cat /etc/distcc/hosts" on all three results in "192.168.1.80
> 192.168.1.40 192.168.1.22".  They all have
> "sys-devel/distcc-2.18.3-r10" and "/etc/conf.d/distccd" modified the
> same way : DISTCCD_OPTS="${DISTCCD_OPTS} --allow 192.168.1.40
> 192.168.1.80 192.168.1.22"
> Distccd was restarted after any config changes.
> 
>  All three have i686-pc-linux-gnu toolsets/chosts, and as far as I can
> tell should work. No firewalls are in place for testing, distccd port
> is open on all three (confirmed with nmap), and I'm out of ideas on
> where to look for problems.
> 
> Any help or suggestions would be appreciated.
> 
> PaulNM
The example I see in etc/conf.d/distccd lists each allowed ip address
after its own --allow argument.  Furthermore, you might consider
whether you'd rather allow anyone on your local subnet to connect --
for example if you add new hosts -- and specify in cidr notation
something like "--allow 192.168.1.0/24"

As it is now, i think only Robotech will be allowed to distribute code
on any host.  I would think galactica would be included in that list,
but for some reason it doesn't seem to want to accept the connections.

I wonder, do you have MAKEOPTS set high enough on your systems to
compile on more than 2 computers at once?  D

id you use distcc-config?

Have you set the configs to accept the number of jobs you want each
computer to accept maximum?  This should be low for the pII and much
higher for the X2.  

Finally, I wanted you to entertain the notion that if robotech's
make.conf (wait, are these names from Transformers?) is set to compile
for a PIII, with sse instructions and such, the PII may not be able to
do it, or if it can, might be required to build a crossdev toolchain
too.  FWIW, pIIs on my network are no longer allowed to 'help' the
faster computers compile because they generally tend to slow them down
rather than speed them up. 

Good luck!!  
-- 
gentoo-user@gentoo.org mailing list