Thread: -S is missing in postgresql.conf?
It seems -S option for postmaster (detaching ttys) does not exist in postgresql.conf. Is there any reason for this? -- Tatsuo Ishii
Tatsuo Ishii writes: > It seems -S option for postmaster (detaching ttys) does not exist in > postgresql.conf. Is there any reason for this? It has been declared evil, because it loses the log output. Using shell redirection, or better yet pg_ctl is the way to go. -- Peter Eisentraut peter_e@gmx.net http://yi.org/peter-e/
> > It seems -S option for postmaster (detaching ttys) does not exist in > > postgresql.conf. Is there any reason for this? > > It has been declared evil, because it loses the log output. Using shell > redirection, or better yet pg_ctl is the way to go. Really? I thought -S was evil because we had no working logging facilities other than logging to stdout/stderr. Now we have the fixed syslog functionarity, thus (at least part of ) the problem has gone. Moreover if -S is still evil, why don't we completely remove -S? It looks slightly inconsistent anyway. IMHO we should make -S be configurable in postgresql.conf and let users choose what they want. -- Tatsuo Ishii
Tatsuo Ishii <t-ishii@sra.co.jp> writes:
> IMHO we should make -S be configurable in postgresql.conf and let
> users choose what they want.
I agree. Mind you, I think we should discourage use of -S, because it
makes troubleshooting so much more difficult. But we shouldn't remove
the option; people who are running a stable application mix and not
having problems may well think they don't need to expend disk space on
a postmaster log. And if we have the option then it ought to be
supported uniformly via GUC.
regards, tom lane
> Tatsuo Ishii <t-ishii@sra.co.jp> writes: > > IMHO we should make -S be configurable in postgresql.conf and let > > users choose what they want. > > I agree. Mind you, I think we should discourage use of -S, because it > makes troubleshooting so much more difficult. But we shouldn't remove > the option; people who are running a stable application mix and not > having problems may well think they don't need to expend disk space on > a postmaster log. And if we have the option then it ought to be > supported uniformly via GUC. Ok. I have committed the changes. The name for the option in postgresql.conf is "silent_mode" taking a value true or false (default to false). If silent_mode is set to true, it would have a same effect as -S of postmaser. -- Tatsuo Ishii