We'll upgrade this again along with pytest once pytest-dev/pytest#6925 is resolved.
Also, it looks like a flaky error in test class DB setup outside a transaction was mucking things up for the rest of the test shard. I didn't try to fix the flakiness, but this should at least limit the blast radius to just that test class. We really need to start using setUpTestData() more consistently.
We're seeing slow commits on production for courseware_studentmodule
updates. Based on the slow queries during those times, we think it
might be because multiple worker processes are trying to update the
same rows from within long-running transactions (since courseware
is relatively slow).
The risk with this is that since the whole view execution is no
longer wrapped in a big implicit transaction, it's possible that
XBlock state will update and things that key off of that (e.g.
completion progress information or pre-req milestones) will fail
in a way that will leave the database in an unplanned-for state,
though this is already the case for those actions that trigger
asynchronous tasks like grades recalculation.
The query counts for the index view test were adjusted down
because save points count towards the total and we're no longer
setting them at the top level around the view as a whole.
* Converting variable to string before assignment in tests
Django 2 is stricter of string typing, so it will no longer do implicit conversions in some cases.
While the release notes only mention dropping bytestring support in some places, they seem to have also dropped implicit conversion to str for whole host of things.
...and it turns out we don't need the switch anymore, anyway.
When we upgraded to Django 1.11, this flag was added in order to
allow for a database migration that would render the user table
unwriteable for up to half an hour:
https://github.com/edx/edx-platform/pull/17561
This involved swapping out the signal handler for logins via
`user_logged_in.disconnect(django_update_last_login)`, but with
Django 2.0, that disconnect is silently failing (returning
false). Likely the disconnect is now happening too soon.
(See edx-platform/common/djangoapps/student/apps.py line 21 in 61e1eda.)
The result is that by the time the waffle switch is consulted, the
normal handler has already run, and the user's last login date has
already been updated.
For now we're just removing the test, and have filed ARCHBOM-1084 for
followup (deleting the switch and related code).
