Last modified: 2014-04-01 01:05:41 UTC
I tried to set up a labs-vagrant instance on both clusters, with the labs::vagrant role, the uploadwizard vagrant role, and other than that, default configuration: uploadwizard-osm -- i-00000aa3.pmtpa.wmflabs -- http://uploadwizard-osm.wmflabs.org/ uploadwizard-osm-equiad -- i-00000154.eqiad.wmflabs -- http://uploadwizard-osm-equiad.wmflabs.org/ For both instances, MediaWiki page load time takes exactly 60s. Image and api.php loading times are normal. Local requests via CURL are equally slow. The page generation time reported by MediaWiki (on the debug toolbar, or as a HTML comment at the end of the page) is below 1 second. Apache itself (via the %T log parameter) reports the 60 sec loading time. I've played around with all the timeout settings in LocalSettings.php, but they did not have any effect. Disabling Redis caching by commenting out the related lines in /vagrant/LocalSettings.php did not have any effect either.
Seems to also affect non-labs-vagrant installations which follow https://wikitech.wikimedia.org/wiki/Help:Single_Node_MediaWiki
Exactly 60 seconds sounds like bug 63338. *** This bug has been marked as a duplicate of bug 63338 ***
