#3517 ambassadors membership problems updating inactive users
Closed: Fixed Opened by kevin.

The daily script that generates ambassadors membership is outputting an error when trying to update the inactives page.

{{{
Subject: Cron root@bapp02 /usr/local/bin/membership-mwclient.py > /dev/null
Date: Wed, 10 Oct 2012 08:38:02 +0000 (UTC)

Traceback (most recent call last):
File "/usr/local/bin/membership-mwclient.py", line 296, in
calc_list()
File "/usr/local/bin/membership-mwclient.py", line 293, in calc_list
page.save(inactive_list, summary = time)
File "/usr/lib/python2.6/site-packages/mwclient/page.py", line 142, in save
result = do_edit()
File "/usr/lib/python2.6/site-packages/mwclient/page.py", line 137, in do_edit
data)
File "/usr/lib/python2.6/site-packages/mwclient/client.py", line 161, in api
info = self.raw_api(action,
kwargs)
File "/usr/lib/python2.6/site-packages/mwclient/client.py", line 244, in raw_api
json_data = self.raw_call('api', data).read()
File "/usr/lib/python2.6/site-packages/mwclient/client.py", line 232, in raw_call
self.wait(token)
File "/usr/lib/python2.6/site-packages/mwclient/client.py", line 266, in wait
raise errors.MaximumRetriesExceeded(self, token, args)
mwclient.errors.MaximumRetriesExceeded: (, , ('api', 'format=json&bot=1&starttimestamp=20121010055444&title=Ambassadors%2FMembershipService%2FInactives&basetimestamp=20120916055414&summary=2012-10-10+05%3A54%3A23.421097&token=a4c9ff0024455d38c3666184b1140c1d%2B%5C&notminor=1&text=%5B%27aab%27%2C+%27aaroncas%27%2C+%27acardenas%27%2C+%27acedip%27%2C+%27ador%27%2C+%27aguskov%27%2C+%27ahmedelgamil%27%2C+%27ajamison%27%2C+%27akashmkj%27%2C+%27alphonse%27%2C+%27alukin%27%2C+%27ananthgs%27%2C+%27bhaslinux%27%2C+%27bm90fox%27%2C+%27byone%27%2C+%27cheekyboinc%27%2C+%27cleitonlima%27%2C+%27cwtucker%27%2C+%27dahlan%27%2C+%27daoud%27%2C+%27dbpatankar%27%2C+%27dejavu87%27%2C+%27dpaulo%27%2C+%27dvation%27%2C+%27edugeek76%27%2C+%27fabriziolapiello%27%2C+%27fbijlsma%27%2C+%27fdagnino%27%2C+%27foxhaund%27%2C+%27frankice%27%2C+%27frankiemangoa%27%2C+%27fusion94%27%2C+%27galdax%27%2C+%27gamaray1%27%2C+%27gantu%27
%2C+%27glommer%27%2C+%27gonz0%27%2C+%27gregoryfenton%27%2C+%27gsorian%27%2C+%27hans%27%2C+%27hussamaismail%27%2C+%27jaymeayres%27%2C+%27jcsniper%27%2C+%27jfautley%27%2C+%27kevin79%27%2C+%27kingheart%27%2C+%27kristho%27%2C+%27laurent%27%2C+%27leroylim%27%2C+%27linuxsamurai%27%2C+%27mabs%27%2C+%27macmmc%27%2C+%27masudsp%27%2C+%27mcascio%27%2C+%27mcleanj%27%2C+%27megenius%27%2C+%27mgob%27%2C+%27mimage%27%2C+%27mj0vy%27%2C+%27mohammedsafwat%27%2C+%27n4is3n%27%2C+%27neverho0d%27%2C+%27nthomas%27%2C+%27paceh%27%2C+%27parsah%27%2C+%27pratyk%27%2C+%27quentindumont%27%2C+%27rafaelconrado%27%2C+%27rafaelliu%27%2C+%27rmenezes%27%2C+%27rsdomowicz%27%2C+%27shanix%27%2C+%27sigmoun%27%2C+%27soothink%27%2C+%27taylon%27%2C+%27theamoeba%27%2C+%27theandruu%27%2C+%27thiagopeixoto%27%2C+%27ush%27%2C+%27wcervini%27%5D&action=edit'))
}}}

This may be related to a wiki upgrade, but the page actually hasn't been updated in a long time.


http://sourceforge.net/apps/mediawiki/mwclient/index.php?title=MaximumRetriesExceeded

Running the script against stg succeeds while prod still tries to update. Whatever causes failing is the difference between stg and prod likely :-(

The difference is that prod returns this:

{{{
HTTP/1.1 500 Internal Server Error\r\n
Date: Wed, 10 Oct 2012 23:10:38 GMT\r\n
Server: Apache/2.2.15 (Red Hat)\r\n
X-Powered-By: PHP/5.3.3\r\n
X-Content-Type-Options: nosniff\r\n
Vary: Accept-Encoding,User-Agent\r\n
AppTime: D=897424\r\n
AppServer: app01.phx2.fedoraproject.org\r\n
Content-Type: text/html; charset=utf-8\r\n
X-Varnish: 1279643385\r\n
Age: 0\r\n
Via: 1.1 varnish\r\n
Content-Encoding: gzip\r\n
ProxyTime: D=1080364\r\n
ProxyServer: proxy1.phx2.fedoraproject.org\r\n
Content-Length: 20\r\n
Connection: close\r\n
\r\n
}}}

This moves the cause from membership-mwclient.py to the infrastructure IMHO. Does above help you for debugging? I hope, I did not make any typos from the broken output that I received locally.

Interestingly, this is no longer happening in production. Since 2012-10-17.

However, it's now happening in staging. Perhaps that will make it easier to track down.

This now seems to be working fine in stg too. No idea what the issue was, but it's cleared up.

Metadata