Forking thread "logical decoding : exceeded maxAllocatedDescs for .spill
files" for this side issue:
On Wed, Jan 08, 2020 at 09:37:04PM -0800, Noah Misch wrote:
> v10
> deletes PostgresNode base directories at the end of this test file, despite
> the failure[1].
> [1] It has the all_tests_passing() logic in an attempt to stop this. I'm
> guessing it didn't help because the file failed by calling die "connection
> error: ...", not by reporting a failure to Test::More via ok(0) or similar.
That is what happened. We should test the exit status to decide whether to
keep temporaries, as attached. PostgresNode does that, since commit 90627cf
(thread https://postgr.es/m/flat/6205.1492883490%40sss.pgh.pa.us). That
thread already discussed $SUBJECT[1] and the __DIE__ handler being
redundant[2]. I plan to back-patch, since it's most useful for v10 and v9.6.
[1] https://postgr.es/m/CAMsr+YFyFU=+MVFZqhthfMW22x5-h517e6ck6ET+DT+X4bUO7g@mail.gmail.com
[2] https://postgr.es/m/FEA925B2-C3AE-4BA9-9194-5F5616AD0794@yesql.se