Re: A test for replay of regression tests

Hi,

On 2022-01-27 14:36:32 -0800, Andres Freund wrote:
> > On my windows test instance where I noticed this (w10,
> > msys2/ucrt), check took 516s and this test took 685s.
>
> Hm. That's both excruciatingly slow. Way way slower than what I see here, also
> w10, msys2/ucrt. Any chance the test instance has windows defender running,
> without a directory exclusion? I saw that trash performance to a near
> standstill.

Could you post the regression test output with the timings? Unless it's AV, I
don't see why a windows VM with a moderate amount of memory should take that
long.

Do the test times get less bad if you use PG_TEST_USE_UNIX_SOCKETS=1
PG_REGRESS_SOCK_DIR: "c:/some-dir/"?

I see there's reports that the connection-timeout problem can be a lot worse
on windows, because several applications, e.g. docker, add additional names
for localhost. Are there any non-commented entries in
C:\Windows\System32\drivers\etc\hosts

> Does it get better with the attached patch?

I pushed something like it now - seemed to be no reason to wait, given it
makes think less slow on my VM.

Greetings,

Andres Freund