[Mailman-Users] invites for large user list fails

[Mark Sapiro]

Anne Ramey wrote:

>Is there a known problem with inviting relatively large groups of 
>addresses at once, ex. 6K at a time?  I usually just subscribe, so I 
>haven't run into this before, but I have a list owner who gets an error 
>when trying to subscribe a list of 6000 addresses, but I've been 
>breaking it up to smaller groups (<400) and they seem to work fine.  But 
>when I get up to 1000+ it fails.
>
>If we try to invite the whole list we get:
>
>
>  Server error!
>
>The server encountered an internal error and was unable to complete your 
>request.
>
>Error message:
>Premature end of script headers: admin
>
>If you think this is a server error, please contact the webmaster 
><mailto:anne.ramey at ncmail.net>.
>
>
>    Error 500
>
>on the screen and:
>
>admin(10796): [----- Mailman Version: 2.1.9rc1 -----]
>admin(10796): [----- Traceback ------]
>admin(10796): Traceback (most recent call last):
>admin(10796):   File "/usr/local/mailman/scripts/driver", line 101, in 
>run_main
>admin(10796):     main()
>admin(10796):   File "/usr/local/mailman/Mailman/Cgi/confirm.py", line 
>117, in main
>admin(10796):     subscription_confirm(mlist, doc, cookie, cgidata)
>admin(10796):   File "/usr/local/mailman/Mailman/Cgi/confirm.py", line 
>350, in subscription_confirm
>admin(10796):     userdesc = mlist.pend_confirm(cookie, expunge=False)[1]
>admin(10796): TypeError: unsubscriptable object
>admin(10796): [----- Python Information -----]
>admin(10796): sys.version     =   2.3.4 (#1, Jan  9 2007, 16:40:18)
>[GCC 3.4.6 20060404 (Red Hat 3.4.6-3)]
>admin(10796): sys.executable  =   /usr/bin/python
>admin(10796): sys.prefix      =   /usr
>admin(10796): sys.exec_prefix =   /usr
>admin(10796): sys.path        =   /usr
>admin(10796): sys.platform    =   linux2
>admin(10796): [----- Environment Variables -----]
>admin(10796):   HTTP_REFERER: 
>http://lists.domain.tld/mailman/confirm/listname/88534c254d1e0e5d4866814e
>7a4e3da15d75580a
>admin(10796):   SERVER_SOFTWARE: Apache/2.0.52 (Red Hat)
>admin(10796):   SCRIPT_NAME: /mailman/confirm
>admin(10796):   SERVER_SIGNATURE: <address>Apache/2.0.52 (Red Hat) 
>Server at lists.domain.tld Port 80</address>
>admin(10796):
>admin(10796):   REQUEST_METHOD: POST
>admin(10796):   PATH_INFO: /dma.npi.listserv
>admin(10796):   SERVER_PROTOCOL: HTTP/1.1
>admin(10796):   QUERY_STRING:
>admin(10796):   CONTENT_LENGTH: 122
>admin(10796):   HTTP_USER_AGENT: Mozilla/4.0 (compatible; MSIE 6.0; 
>Windows NT 5.1; SV1; .NET CLR 1.1.4322; Inf
>oPath.1; .NET CLR 2.0.50727; .NET CLR 3.0.04506.30)
>admin(10796):   HTTP_CONNECTION: Keep-Alive
>admin(10796):   SERVER_NAME: lists.domain.tld
>admin(10796):   REMOTE_ADDR: 11.22.12.123
>admin(10796):   PATH_TRANSLATED: /var/www/content/listname
>admin(10796):   SERVER_PORT: 80
>admin(10796):   SERVER_ADDR: 123.22.11.32
>admin(10796):   DOCUMENT_ROOT: /var/www/content
>admin(10796):   PYTHONPATH: /usr/local/mailman
>admin(10796):   SCRIPT_FILENAME: /usr/local/mailman/cgi-bin/confirm
>admin(10796):   SERVER_ADMIN: admin at domain.tld
>admin(10796):   HTTP_HOST: lists.domain.tld
>admin(10796):   HTTP_CACHE_CONTROL: no-cache
>admin(10796):   REQUEST_URI: /mailman/confirm/dma.npi.listserv
>admin(10796):   HTTP_ACCEPT: image/gif, image/x-xbitmap, image/jpeg, 
>image/pjpeg, application/x-shockwave-flash
>, application/vnd.ms-excel, application/vnd.ms-powerpoint, 
>application/msword, application/xaml+xml, applicatio
>n/vnd.ms-xpsdocument, application/x-ms-xbap, 
>application/x-ms-application, */*
>admin(10796):   GATEWAY_INTERFACE: CGI/1.1
>admin(10796):   REMOTE_PORT: 14506
>admin(10796):   HTTP_ACCEPT_LANGUAGE: en-us
>admin(10796):   CONTENT_TYPE: application/x-www-form-urlencoded
>admin(10796):   HTTP_ACCEPT_ENCODING: gzip, deflate


The above traceback is not from the error in inviting. It is from an
error that occurred in trying to accept (or possibly not accept) an
invitation or some other subscription confirmation. It was generated
by the confirm CGI, not the admin CGI.

I don't have time to look at it in more detail for a couple of days,
but I will. It could be due to some corruption in the list's pending
database caused by the invite error.


>This error didn't really tell me what was wrong, but I guess I'm just 
>not reading it right.  This was the only thing I could find that seemed 
>to be an error:
>TypeError: unsubscriptable object
>Does this mean there was just one address on the list that freaked it 
>out? f it is one email address, how do I tell which one?  They all look 
>OK to me when I scroll through.


I think the 'invite large list' error is the CGI running into an Apache
imposed CPU time limit and thus being killed and not given a chance to
log any error.

You may find more info in the apache error_log.

See
<http://www.python.org/cgi-bin/faqw-mm.py?req=show&file=faq06.018.htp>
which has to do with memory limits and MemoryError from
Mailman/Python, but in this case, you would be looking for an
RLimitCPU directive in your Apache config, not RLimitMEM.

-- 
Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan