samba_3_0 disabled in build farm

[Gerald (Jerry) Carter]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Andrew Tridgell wrote:
> Jeremy and Jerry,
> 
> The current samba_3_0 test code is leaving behind huge 
> numbers of processes, with some of our build farm machines
> now at the stage where users can't login (for example,
> metze can't login to his Linux boxes). Something must have
> changed in the Samba3 code quite recently for this to
> be happening.
> 
> To prevent us having a general meltdown of the build farm, I have
> disabled samba_3_0 in the farm for now. 
> 
> Jerry, do you think there is any chance your recent rework 
> of the socket handling code could be contributing to this?
> I can't see anything wrong with it, but a fair bit changed
> so perhaps there is an unintended side effect.

That wasn't me.  That was Jeremy :-)

> The other likely cancidates would be:
> 
>   - anything that calls fork() or causes anything to call fork()
>   - signal handling

I'm checking the logs now, but the only things that I
have touched recently have been in the RPC code which couldn't
have caused this.

> 
> Here is an example of what is currently happening:
> 
>   http://samba.org/~tridge/metze01.txt

I'm looking at it now.





cheers, jerry
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCsB0BIR7qMdg1EfYRAiezAKC5c+orae9nOtn1BoGFubNFQamnVACgsOgz
q40/EYXt3ymFIWl5I9E5HCo=
=toGQ
-----END PGP SIGNATURE-----