[Bug 19913] New: Recurring, hard to diagnose edit consistency problems

https://www.w3.org/Bugs/Public/show_bug.cgi?id=19913

          Priority: P1
            Bug ID: 19913
                CC: team-webplatform-admin@w3.org
          Assignee: schepers@w3.org
           Summary: Recurring, hard to diagnose edit consistency problems
        QA Contact: public-webplatform-bugs@w3.org
          Severity: normal
    Classification: Unclassified
                OS: All
          Reporter: komoroske@google.com
          Hardware: PC
            Status: NEW
           Version: unspecified
         Component: infrastructure
           Product: webplatform.org

For the past week or two there have been very odd edit consistency problems
that are difficult to debug and difficult to reproduce accurately. In sum,
though, they make editing templates and forms (and sometimes just normal
articles) extremely challenging to do correctly.

I've observed all of the following behavior in the past few weeks:
* Make an edit, it "sticks" (no session loss warning), but the page I'm taken
to after editing doesn't show the edit. Refreshing the browser multiple times
is sometimes necessary, and sometimes only a hard refresh (Edit > Refresh) will
fix it.
* Edits to templates don't show up on pages that use those templates for
minutes, even after hard-refreshing the pages that use them
* Changes that have been made, and show up when viewing the page, not appearing
in the changelog
* Changes in the changelog being out of order with times that are incorrect,
even though the final state of the page is correct. In one case I made an edit,
then a few seconds later made a second edit. Minutes later, the changelog was
showing that the second edit was made more than four minutes before the first!

This problem happens often. A recent page where oddities were observed:
http://docs.webplatform.org/wiki/Template:API_Object_Property

It seems like various backends are seeing an inconsistent picture of the status
of the database.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.

Received on Thursday, 8 November 2012 22:19:52 UTC