Re: Mysql OperationalError on close() "fixed"

Hi Ralf:

In message <[email protected]>,
Ralf Schlatterbeck writes:
>On Sun, Mar 24, 2019 at 01:41:31PM -0400, John P. Rouillard wrote:
>> I have forced the version of the mysqlclient library installed via pip
>> to be 1.3.13. Pip was installing version 1.3.14. This "solves" the
>> issue with an exception being thrown during a close() operation.
>
>Is this the same problem we're seing in Travis?

Sorry, I wasn't clear. The change was to .travisci to fix the issue
there.

>> Joseph gave me a clue a couple of weeks back by noting that there were
>> no roundup code changes between a passing and a failing travis-ci run.
>> 
>> Note that the tests are still not passing.
>
>I had looked into the errors some time ago and it's still on my todo
>list to dig this up (I think I had written up what I found in some
>email, seems not to the roundup list(s)). The problems look as if they
>can be fixed by tweaking some mysql settings.

I tried changing the max_allowed_packet size to 500M. The rest of the
things I tried/discarded was discussed in:

  https://issues.roundup-tracker.org/msg6383

on issue: https://issues.roundup-tracker.org/issue2551025

The rollback in version was a last resort.

If we drop support for python 3.4, we can upgrade mysqlclient library
to 1.4.2post1 (newest release IIRC). I an not sure what distros still
ship python 3.4 (centos 7 maybe?).

>> Ralf, did you forget to check in the test code changes to match your
>> changes to rest.py? The changes that have been made to the rest code
>> was not reflected in test/rest_common. So the tests are still looking
>> for old style (without links, file content being reported as content)
>> output.
>
>No, I didn't look at the tests yet. I will do more this week and hope to
>have passing tests again soon.

Cool.

Have a great week.

--
				-- rouilj
John Rouillard
===========================================================================
My employers don't acknowledge my existence much less my opinions.