[newbie] Postmaster service on W2K

Hi,

I'm new to PostgreSQL and the list so please bear with me a bit.

I want to install PG as a service on my W2K system running the latest
Cygwin packages and have run up against some problems. I followed the
instructions in the /usr/share/doc/Cygwin/postgresql-7.4.5.README file
to install the service and adjusted a few things to suit my setup.
Here's the command I used to install the service:

cygrunsrv -I postmaster -p /usr/bin/postmaster -d "CYGWIN postmaster" \
-f "PostgreSQL server" -t manual -u postgres -y cygserver \
-a "-D /var/postgresql/data -i" -s INT -o -1 /var/log/postmaster.log \
-2 /var/log/postmaster.err

I'm not sure if I'm seeing the expected results: I have three 'postgres'
processes running under the 'postgres' account, an error in the Windows
Event Viewer console, and a 'postmaster' service that is not listed as
'started' in the Windows Services console even though I can connect with
the newly created Template1 database with PgAdmin III for Windows.

The 'postgres' processes appear like this:

$ ps -eaf | grep postgres
UID PID PPID TTY STIME COMMAND
postgres 2268 2296 ? 20:48:13 /usr/bin/postgres
postgres 2264 2268 ? 20:48:14 /usr/bin/postgres
postgres 1552 2264 ? 20:48:14 /usr/bin/postgres

The error logged in the Event Viewer is:
-----------------------------------------------------------------
System Log:

Event Type: Error
Event Source: Service Control Manager
Event Category: None
Event ID: 7031
Date: 02/09/2004
Time: 8:47:14 PM
User: N/A
Computer: KIRALY
Description:
The postmaster service terminated unexpectedly. It has done this 1
time(s). The following corrective action will be taken in 0
milliseconds: No action.
-----------------------------------------------------------------

In the /var/log/postmaster.log file, I see the following errors:

-----------------------------------------------------------------
WARNING: dup(0) failed after 3195 successes: Bad file descriptor
LOG: database system was shut down at 2004-09-04 18:58:21 PDT
LOG: checkpoint record is at 0/9E4758
LOG: redo record is at 0/9E4758; undo record is at 0/0; shutdown TRUE
LOG: next transaction ID: 898; next OID: 17148
LOG: database system is read
-----------------------------------------------------------------

>From other posts in this and other lists all the above points to
misconfigured file permissions, so I've adjusted a few things in an
attempt to 'fix' what might not be broken. Here's a summary of my file
and directory permissions as they stand now:

$ id Manager
uid=500(Manager) gid=544(Administrators) groups=544(Administrators)
$ id Users
uid=545(Users) gid=545(Users) groups=545(Users)
$ id postgres
uid=1007(postgres) gid=513(None) groups=513(None)

$ ls -ld /var/postgresql
drwxr-xr-x+ 3 Manager root 0 Sep 4 17:23 /var/postgresql/

$ ls -ld /var/postgresql/data (and all subdirectories and files)
drwxrwx---+ 6 postgres Users 4096 Sep 4 19:00
/var/postgresql/data/

$ ls -ld /var/log
drwxrwxrwx+ 4 Manager Users 4096 Sep 4 18:38 /var/log/

$ ls -ld /tmp
drwxrwxrwx+ 40 Manager Users 12288 Sep 4 23:31 /tmp/

The lock files appear as follows:
srwxrwxrwx 1 postgres Users 51 Sep 4 23:33 .s.PGSQL.5432=
-rw------- 1 postgres Users 26 Sep 4 23:33 .s.PGSQL.5432.lock

If someone has PG running as a service under W2K, it would be most
appreciated if you could look at these permissions and compare them with
those that work. Does anything appear that would suggest a reason for
the error I see in the Event Viewer and in the postmaster error log?

Thanks,

Alder