Actually pg_ctl -D $PGDATA is redundant....
-D datadir
Specifies the file system location of the database
files. If
this is omitted, the environment variable PGDATA is used.
In fact if $PGDATA is for some reasong not loaded, you'll paint yourself
to a bad spot
where -D -o will be regarded as -o is the datadir location.
scott.marlowe wrote:
>On Mon, 18 Nov 2002, Johnson, Shaunn wrote:
>
>
>
>>su -l postgres -s /bin/sh -c "/usr/bin/pg_ctl -D $PGDATA -o '-i -B 128 -N 64
>>-d 2' \
>>-p /usr/bin/postmaster start >/dev/null "
>>
>>
>
>Doing it that way, you should be able to leave out the parts from -o on
>and get it to use the postgresql.conf file settings.
>
>
>
>>--is this true? perhaps someone can verify this.
>>
>>
>
>Generally speaking, while it still considered bad practice (or at least
>somewhat "rude" :-) to kill -9 the postmaster, it isn't particularly
>dangerous. I've done it while heavily testing a database (pgbench -c 100
>-t 5000) and it never once corrupted my database.
>
>
>
>>--this is what the /proc/mdstat says
>>
>>--[snip from /proc/mdstats]
>>
>>Personalities :
>>read_ahead not set
>>unused devices: <none>
>>
>>--[/snip from /proc/mdstats]
>>
>>That looks like a dead drive in your RAID array. What does 'cat
>>/proc/mdstat' say about the drive
>>
>>--i'm doing a linux software raid
>>
>>
>
>Is there anymore to the /proc/mdstat entry? normally you should have a
>line that has a bit that names the drive partitions in the software raid
>and has a bit showing which drives are online that looks something like
>this: [UUUU] for all drives being UP. And like this: [UU_U] where the
>underscore shows a missing drive. If you don't have a line like that
>then Linux isn't doing the software RAID array. Could it be that you
>aren't actually running RAID but think you are? Just wondering.
>
>
>---------------------------(end of broadcast)---------------------------
>TIP 4: Don't 'kill -9' the postmaster
>
>