Re: pg_regress cleans up tablespace twice. - Mailing list pgsql-hackers

From Thomas Munro
Subject Re: pg_regress cleans up tablespace twice.
Date
Msg-id CA+hUKG+nLPEJZtTpQ70A7BnRdweOsyyWQ4nWyTAoWCMBk7wgkA@mail.gmail.com
Whole thread Raw
In response to Re: pg_regress cleans up tablespace twice.  (Michael Paquier <michael@paquier.xyz>)
Responses Re: pg_regress cleans up tablespace twice.  (Michael Paquier <michael@paquier.xyz>)
List pgsql-hackers
On Sat, Jun 20, 2020 at 6:46 PM Michael Paquier <michael@paquier.xyz> wrote:
> On Sat, Jun 20, 2020 at 03:01:36PM +1200, Thomas Munro wrote:
> > Thanks for the clue.  Appveyor runs your build script as a privileged
> > user (unlike, I assume, the build farm animals), and that has caused a
> > problem with this test in the past, though I don't know the details.
> > I might go and teach it to skip that test until a fix can be found.
>
> Thanks, I was not aware of that.  Is it a fix that involves your code
> or something else?  How long do you think it would take to address
> that?  Another strategy that we could do is also a revert of 2b2a070
> for now to allow the cfbot to go through and then register this thread
> in the CF app to allow the bot to pick it up and test it, so as there
> is more room to get a fix.  The next CF is in ten days, so it would be
> annoying to reduce the automatic test coverage the cfbot provides :/

I'm not sure what needs to change, but in the meantime I told it to
comment out the offending test from the schedule files:

+before_test:
+  - 'perl -p -i.bak -e "s/^test: tablespace/#test: tablespace/"
src/test/regress/serial_schedule'
+  - 'perl -p -i.bak -e "s/^test: tablespace/#test: tablespace/"
src/test/regress/parallel_schedule'

Now the results are slowly turning green again.



pgsql-hackers by date:

Previous
From: Tomas Vondra
Date:
Subject: Re: Improve planner cost estimations for alternative subplans
Next
From: Tom Lane
Date:
Subject: Re: Improve planner cost estimations for alternative subplans