Обсуждение: Re: pgsql: Use data directory inode number, not port, to select SysVresour
On 9/6/19 3:51 PM, Andrew Dunstan wrote: > On 9/6/19 2:42 PM, Tom Lane wrote: >> Andrew Dunstan <andrew.dunstan@2ndquadrant.com> writes: >>> Given your stated intention, I think the simplest way to get it is just >>> this, without worrying about what the perl modules might do: >>> -if ($@) >>> +if ($@ || $windows_os) >> WFM, do you want to push that? >> >> > > done. > > [redirected to -hackers] I'm going to disable this test (src/test/recovery/t/017_shm.pl) on Windows on the back branches too unless there's a violent objection. The reason is that the script runs "postgres --single" and that fails on Windows when run by an administrative account. We've carefully enabled postgres and its tests to run safely under an admin account. I discovered this as part of my myss2 testing. cheers andrew
Andrew Dunstan <andrew.dunstan@2ndquadrant.com> writes:
> I'm going to disable this test (src/test/recovery/t/017_shm.pl) on
> Windows on the back branches too unless there's a violent objection.
As I said before, I think that test does nothing useful unless SysV
shmem is in use, so I see no reason not to disable it on Windows.
regards, tom lane
On Sun, Sep 08, 2019 at 05:54:12PM -0400, Andrew Dunstan wrote: > I'm going to disable this test (src/test/recovery/t/017_shm.pl) on > Windows on the back branches too unless there's a violent objection. The > reason is that the script runs "postgres --single" and that fails on > Windows when run by an administrative account. We've carefully enabled > postgres and its tests to run safely under an admin account. I > discovered this as part of my myss2 testing. I'm reading that the test falsified this assertion that we've enabled postgres to run safely under an admin account. Enabling safe use of admin accounts entails fixing single-user mode. (Alternately, one could replace the "vacuum that database in single-user mode" errhint with a reference to some not-yet-built alternative. That sounds harder.)