Re: [squid-users] fake_auth exiting with signal 11

From: Robert Collins <robert.collins@dont-contact.us>
Date: Wed, 26 Dec 2001 23:44:40 +1100

----- Original Message -----
From: "George G. Zavyalov" <squid-mailing@altervita.ru>

> Hi!
>
> This is my first experiencing with writing to English-speaking
> mailing list, so, please excuse me for possible mistakes.

No problem. You've produced a very clear report of whats happening -
which is good.

> I have SQUID-2.5.DEVEL (squid-head-200112130001) installed as a
usual
> proxy at Freebsd 4.4-RELEASE. This proxy controls http at Windows
> 2000 environment (about 20 workstations and one Windows2000-based
> domain controller). We don't need "true" authentication, we just
> want to log user activities. Squid is running in daemon mode.

Are you building squid via the ports mechanism? If not, make sure that
you do not use -O2, as some buggy gcc versions have been released that
interacted badly with squid.

> Fake_auth parameters are:
>
> auth_param ntlm program /usr/local/squid/libexec/fakeauth_auth
> auth_param ntlm children 30

Unless you have a very large network you will not need 30 authentication
children. Start with 10, and if the 10th gets more than 1 request per
minute, then add 5 more.

> The problem is that fake_auth crashes with signal 11 PERIODICALLY,
> with this sort of diagnostics ( -X squid option):
>
>
>
> From one hand the tandem "squid - fake_auth" is all that we
> need now. This exeption situation is controlled by the parent
process, reloading
> the child, when it terminates abnormally. From the other hand, i
> must find network event, that caused fake_auth crash. We are
> waiting the ranging of our network to the number of about one
hundred
> http-clients. And i need the interpretation of that
"ntlmGetString: insane: l:0 o:64"
> diagnostics. What is this "insane" packet?

IIRC the code correctly you've recieved a NTLMSSP header with the length
set to 0. What Version of win2k are you running - what service pack,
what IE security settings.

fakeauth can be made more robust, and I'll do something for this bug
shortly. Can you tell me the info I've asked for above?

Rob
Received on Wed Dec 26 2001 - 05:44:36 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:05:29 MST