On Mon, Apr 06, 2020 at 10:19:41PM -0400, Tom Lane wrote:
>James Coleman <jtc331@gmail.com> writes:
>> Fair enough. Unsure if Tomas is still online to comment and/or push,
>> but reverting SORT_TYPE_STILL_IN_PROGRESS back to 0 works for me as an
>> initial fix.
>
>I'm guessing he went to bed, so I'll push a fix in a moment.
>The patch has survived enough test cycles here now to make me
>moderately confident that it fixes the issue.
>
Nope, how I could I sleep with half of the buildfarm still red?
I came to the same conclusion (that the change in TuplesortMethod
definiton is the culprit) a while ago and was about to push a fix that
initialized it correctly in ExecSortInitializeDSM. But I agree reverting
it back to the old definition is probably better.
Thanks!
--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services