Hi guys,

I'm running slackware 10 and it seems that the call to rc.pgsql in rc.M causes rc.M to exit. The machine boots completely but nothing after Postgres starts, Postgres itself starts.

I don't know if it's related but a point to note is that the output from MySQL starting comes after 'echo "Finished starting MySQL"' (this just means MySQL forked right?)

Apache and everything after it doesn't start.

There doesn't seem to be anything abnormal in /var/log that I can see.

Any ideas?

TIA, Karl.

----------------
rc.M snippet
----------------
# Start the MySQL database:
if [ -x /etc/rc.d/rc.mysqld ]; then
echo "Started starting MySQL"
. /etc/rc.d/rc.mysqld start
echo "Finished starting MySQL"
fi

# Start the PostgreSQL database:
if [ -x /etc/rc.d/rc.pgsql ]; then
echo "Started starting PostgreSQL"
. /etc/rc.d/rc.pgsql start
echo "Finished starting PostgreSQL"
fi

# Start Apache web server:
if [ -x /etc/rc.d/rc.httpd ]; then
echo "Started starting Apache"
. /etc/rc.d/rc.httpd start
echo "Finished starting Apache"
fi

----------------
Bootup output snippet
----------------
Started starting MySQL
Finished starting MySQL
Started starting PostgreSQL
Starting PostgreSQL: Starting mysqld daemon with databases from /var/lib/mysql
ok

Welcome to Linux 2.4.26 (tty1)

linux login:
-----------------
rc.pgsql snippet
----------------
$ECHO_N "Starting PostgreSQL: "$ECHO_C
su - $PGUSER -c "$DAEMON start -D '$PGDATA' -s -l $PGLOG -o -i"
echo "ok"
----------------

If you don't even see the ok (but do get the message
about starting PostgreSQL) I'd suggest to chuck a few
extra echo's into your rc.mysqld, rc.pgsql (where did
that one come from, btw, Postgres doesn't supply slack
startup-scripts to the best of my knowledge) and maybe
your rc.M as well ... a first attempt of mine would be to
include an & at the end of
like so


Cheers,
Tink

Thanks for the reply Tinkster,

the ampersand didn't help, but looking further into pc.pgsql I found the last line was:
exit 0
after commenting this, startup runs as expected.

This prompts another question. It was my belief that programs called from within bash (rc.pgsql is a sh script) were started within their own shell. Is this not the case for sh scripts and the cause of my problems? (insofar as an exit call on a child process will kill the parent too).

WRT the rc.pgsql's origin, I believe that the only installs of postgres that have been on this machine are those that have been built from posgresql.org tarballs. So I'd guess it's come from a 'make install' of postgres.

-Karl.

Glad it's fixed, even though the solution kind of
surprises me :)

The NORMAL action is that the parent spawns a sub-shell,
and waits for its completion (unless you start the sub-process
with an ampersand to detach it) ... which is why I find the "solution"
to remove the exit surprising.

Has never been supplied by Postgres, at least not in
versions 6.x to 8.x ...


Cheers,
Tink

There are two things :
and
The first execute the script in a sub-shell ->'exit 0' ends the sub-shell
The second include the script -> 'exit 0' ends the main program