Web browser: ---
Bug ID: 59642
Summary: Flow: browser test got 'failed insert' exception
adding a topic
Product: MediaWiki extensions
CC: ebernhard...@wikimedia.org, oke...@wikimedia.org,
Mobile Platform: ---
Similar to bug 58996, a browser test failed at line 498 of
includes/Data/ObjectManager.php: with failed insert.
The difference is this is adding a new topic (in $storage->put( $firstPost )),
not editing an existing post. Again there were overlapping tests hitting the
same Flow QA page at the time, maybe another browser test was trying to edit
this post just as the API call was returning; or could two browsers hitting the
same Flow_QA page cause problems even if they're adding unrelated topics?
If you look at the Sauce Labs report (scroll down and click the camera), you
can see the [Add topic] quickly resulted in a pink errorbox with "Exception
Caught: failed insert". Yet, you can find the topic in RecentChanges (search
for its random string "0.6552491559"), and you can view it and its first post,
seemingly no worse for wear, at
. So if storing the post failed, why is it there? It was only two hours ago,
am I seeing cached success hiding a DB failure?
It's unfortunate there isn't more information about what went wrong; the
insert() only returns true/false, but maybe lower levels could log something.
Here's the output in fluorine:/a/mw-log/archive/exception.log-20140104.gz , I
couldn't find anything related in the compressed (huge) api.log, dberror.log,
external.log, or anything at exactly that time in test2wiki.log. Maybe DB
failures on flowdb are written somewhere else, see
2014-01-04 06:34:46 mw1202 test2wiki: [e4273fd5] /w/api.php Exception from
line 498 of
#8 /usr/local/apache/common-local/php-1.23wmf9/api.php(76): ApiMain->execute()
#9 /usr/local/apache/common-local/w/api.php(3): require('/usr/local/apac...')
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
Wikibugs-l mailing list