[Mageia-sysadm] Strange sympa problem

Michael Scherer misc at zarb.org
Wed Jan 12 03:53:09 CET 2011


Hi.

I just received a mail saying I was unsubscribed from a ml. After
reading the log, it seems that I was unsubscribed by the google bot :

Jan 12 02:56:25 alamut wwsympa[2665]: [robot ml.mageia.org] [session
27796412636479] [client 66.249.72.231] [user misc at zarb.org] [list
sysadmin-reports] main::do_signoff() do_signoff

# host 66.249.72.231
231.72.249.66.in-addr.arpa domain name pointer
crawl-66-249-72-231.googlebot.com.

Can someone think of a explanation that do not use the word paranormal ?

I doubt that google is reusing my cookie or something like that, as I do
not have a google account. I did surf using my gf laptop 3 days ago
( who was logged on her account ) but I doubt this would be a valid
explanation.

I did test with a different browser, and the url is not enough. 

I would suspect a cache problem, but there is no cache module or related
on alamut at apache level :

# rpm -qa | grep apache
apache-mod_wsgi-2.5-3mdv2010.1
apache-mpm-prefork-2.2.15-3.1mdv2010.1
apache-mod_fcgid-2.3.5-0.2mdv2010.1
apache-mod_perl-2.0.4-13mdv2010.1
apache-base-2.2.15-3.1mdv2010.1
apache-mod_fastcgi-2.4.6-10mdv2010.1
apache-conf-2.2.15-1mdv2010.1
apache-mod_ssl-2.2.15-3.1mdv2010.1
apache-modules-2.2.15-3.1mdv2010.1

There is however a cache in sympa, for various thing ( and for example
for user session ). 

And this could correspond to :
wwsympa/wwsympa.fcgi.in:  In some - difficult to precisely describe -
        configurations, sessions could be messed due to an inner cache of
        cookie environment variable, either in Apache of perl. Fixed by
        force-blanking this variable as soon as we affect it to a session.

So it seems that we should do the upgrade to the branch 6.1.X , and 6.1.2 seems to not 
be too buggy ( since there was 2 bugfixs minor releases ).

( yes, I notice the irony as this was proposed right from the start by Derek on 9/11 and 
that we didn't ask for a security upgrade on sympa ).

I open a bug ( and for the postgresql problem too ), and will work on a backports  
-- 
Michael Scherer



More information about the Mageia-sysadm mailing list