postmaster and postgres options assimilation
I've looked at the issue of assimilating the options of postmaster and
postgres, which has been mentioned now and then over the years. Basically,
we have five conflict cases that need to be resolved by breaking one or the
other, namely:
-N
postmaster: max_connections
postgres: do not end command at newline
-o
postmaster: postgres options
postgres: output file
-p
postmaster: port number
postgres: internal use
-S
postmaster: silent mode
postgres: work_mem
-s
postmaster: use SIGSTOP on backend crash
postgres: statistics
Renaming the postgres side of -N, -o, -p, and -s might not really do any harm,
but the -S option used to be very popular on the postgres command-line via -o
from the postmaster, so I'm afraid that this might break existing start
scripts.
Does anyone have suggestions about resolving these conflicts, or should we
just forget about this issue altogether?
Peter Eisentraut <peter_e@gmx.net> writes:
-S
postmaster: silent mode
postgres: work_mem
Renaming the postgres side of -N, -o, -p, and -s might not really do
any harm, but the -S option used to be very popular on the postgres
command-line via -o from the postmaster, so I'm afraid that this might
break existing start scripts.
Actually I'd vote for renaming the silent-mode switch. That's pretty
well deprecated anyway isn't it? I've certainly not seen it used in
start scripts.
Similarly, I think it would be less likely to cause problems if you
renamed postmaster -s (SIGSTOP) than postgres -s. I can't imagine
that anyone is using postmaster -s in the field.
Does anyone have suggestions about resolving these conflicts, or should we
just forget about this issue altogether?
I don't think we should let the risk of breaking a few start scripts
dissuade us from cleaning up this mess.
regards, tom lane