mahara-contributors team mailing list archive

[Aaron Wells <1514608@xxxxxxxxxxxxxxxxxx>]

Okay, after further testing today I've confirmed that disabling Mahara's
usage of explicit database transactions (i.e. "start transaction;" and
"commit;") in MySQL resolves this problem. On the test server I was
using, I was getting deadlocks with as few as 4 concurrent users copying
a collection. After disabling transactions, I was able to run 50
concurrent users without any errors.

At that point it was taking 8.5 minutes for the copy operation to
finish, so I didn't test any higher than that. So, we still have some
performance & scaling work we could do, but it solves the problem for
deadlocks. :)

With further testing on the same server I also verified that the earlier
patch changing the transaction isolation level from "REPEATABLE READ" to
"READ COMMITTED" helps prevent deadlocks as well. When I disabled
explicit transactions, and also rolled back to the default "REPEATABLE
READ", I got deadlocks when I had 15 concurrent users copying the
collection.

So I think Mahara's immediate fix for the next minor release, has go to
to be:

1. Keeping the "READ COMMITTED" patch in place.
2. Disabling Mahara's usage of explicit transactions, in MySQL.

In the longer term we can look into all the places we use transactions,
and maybe remove some of them. We shouldn't be putting huge operations,
like the copying of a collection, inside of a database transaction. At
least, not for MySQL, unless we can figure out how to make InnoDB act a
lot more like Postgres in this regard.

Cheers,
Aaron

-- 
You received this bug notification because you are a member of Mahara
Contributors, which is subscribed to Mahara.
Matching subscriptions: Subscription for all Mahara Contributors -- please ask on #mahara-dev or mahara.org forum before editing or unsubscribing it!
https://bugs.launchpad.net/bugs/1514608

Title:
  MySQL deadlock issues with simultaneous users

Status in Mahara:
  In Progress

Bug description:
  The patch in Bug 1492919 resolved the deadlock issues for some users,
  but others are still reporting that they are getting deadlocks. This
  seems to be an issue unique to MySQL, and it can happen with as few as
  2 or 3 concurrent users.

  It has been reported across all stable versions. Strangely, it seems
  to be a relatively recent issue (i.e. we never heard this with Mahara
  1.8 or 1.7, etc), so this suggests it was triggered by something that
  changed fairly recently. One possible candidate is our switch to the
  mysqli driver (before that Mahara wasn't doing transactions at all, in
  mysql).

  See also:
  * https://mahara.org/interaction/forum/topic.php?id=7309
  * https://mahara.org/interaction/forum/topic.php?id=7432

To manage notifications about this bug go to:
https://bugs.launchpad.net/mahara/+bug/1514608/+subscriptions