pg_ctl changes
I've implemented the following changes in pg_ctl:
* Make -w the default for shut down, add -W option to specify no wait.
* Add -l option to name log file. If the option is omitted, then the
postmaster log goes to stdout, so you can use 'pg_ctl ... > log' and still
get pg_ctl's stderr on the terminal.
* Proper file descriptor redirection to allow postmaster to detach from
shell's process group.
* Add -s option to turn off informational messages, so as to not mess up
the carefully crafted Red Hat-style startup screens. Error messages still
show up.
This should make pg_ctl a lot more suitable for startup scripts and
overall. If this is okay then I'll also try to merge the contrib/linux
and contrib/init.d scripts and reflect these changes there.
--
Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/
Peter Eisentraut <peter_e@gmx.net> writes:
I've implemented the following changes in pg_ctl:
Sounds good to me.
* Add -l option to name log file. If the option is omitted, then the
postmaster log goes to stdout, so you can use 'pg_ctl ... > log' and still
get pg_ctl's stderr on the terminal.
Let's see if I've got this straight --- the postmaster's stdout and
stderr go to the -l file (by default pg_ctl's stdout), while pg_ctl's
own stderr is kept separate for failure notices? That seems like it
will work ... in particular one could imagine doing
pg_ctl start | log_splitter_process
to handle log rotation stuff.
regards, tom lane
Peter Eisentraut wrote:
I've implemented the following changes in pg_ctl:
[snip]
This should make pg_ctl a lot more suitable for startup scripts and
overall. If this is okay then I'll also try to merge the contrib/linux
and contrib/init.d scripts and reflect these changes there.
Sounds good to me. It appears that you haven't checked in your changes
to CVS as of a few minutes ago, but I like the looks of what you've
posted.
--
Lamar Owen
WGCR Internet Radio
1 Peter 4:11