Re: bug report: 7.4beta3 pg_ctl log handling

Simplest case to test is:

pg_ctl start -D /valid/db/dir -l /invalid/dir
pg_ctl: /invalid/dir: No such file or directory
postmaster successfully started

Postmaster hasn't started and the error could probably use some work.

Perhaps some more tests after line 340 of pg_ctl would be appropriate.
Something along the lines of (perlish pseudocode):

if ( -w $logfile || ( -w `dirname $logfile` )

Should it also test -O to ensure ownership if the file does exist? I'll
brush up on my sh and submit a patch if requested.

Arguile <arguile(at)lucentstudios(dot)com> writes:
> Simplest case to test is:
> pg_ctl start -D /valid/db/dir -l /invalid/dir
> pg_ctl: /invalid/dir: No such file or directory
> postmaster successfully started

> Postmaster hasn't started and the error could probably use some work.

If you don't specify -w then pg_ctl only launches the postmaster,
it does not wait around to see if the postmaster really fired up
sucessfully. If there's a bug here, it's only that -w isn't the
default behavior.

regards, tom lane

On Thu, 2003-10-02 at 00:54, Tom Lane wrote:
> Arguile <arguile(at)lucentstudios(dot)com> writes:
> > Simplest case to test is:
> > pg_ctl start -D /valid/db/dir -l /invalid/dir
> > pg_ctl: /invalid/dir: No such file or directory
> > postmaster successfully started
>
> > Postmaster hasn't started and the error could probably use some work.
>
> If you don't specify -w then pg_ctl only launches the postmaster,
> it does not wait around to see if the postmaster really fired up
> sucessfully. If there's a bug here, it's only that -w isn't the
> default behavior.

Oops. You can tell how often I start Pg by hand ;). Sorry for the
bother.