This seems to work well, for both destination types 'stderr' (with
redirect_stderr set), and 'eventlog'. The only minor gripe is that the
'eventlog' case is a bit laiden with extraneous DLL warnings.
The logger seems to shutdown cleanly, for both the case where pg_ctl is used,
and where the service manager is.
Here are some sample outputs:
i) Eventlog
Logging information gets into the event log ok: (including errors and shutdown
notification)
An error (select from non-existent table):
The description for Event ID ( 0 ) in Source ( PostgreSQL ) cannot be found.
The local computer may not have the necessary registry information or message
DLL files to display messages from a remote computer. The following information
is part of the event: ERROR: relation "pg_classes" does not exist
Final shutdown:
The description for Event ID ( 0 ) in Source ( PostgreSQL ) cannot be found.
The local computer may not have the necessary registry information or message
DLL files to display messages from a remote computer. The following information
is part of the event: LOG: database system is shut down
ii) Stderr
new files are created for seach startup, and file contents includes both error
and notification messages. Note that this case was shutdown with the service
manager (while I had an idle connection still open)
LOG: database system was shut down at 2004-08-07 15:04:08 New Zealand Standard
Time
LOG: checkpoint record is at 0/6A3FB998
LOG: redo record is at 0/6A3FB998; undo record is at 0/0; shutdown TRUE
LOG: next transaction ID: 545; next OID: 10047239
LOG: database system is ready
ERROR: relation "pg_classes" does not exist
LOG: received fast shutdown request
LOG: aborting any active transactions
FATAL: terminating connection due to administrator command
LOG: shutting down
LOG: database system is shut down
LOG: logger shutting down
regards
Mark
Quoting Tom Lane <tgl@sss.pgh.pa.us>:
> I have just committed a somewhat trimmed-down version of Andreas Pflug's
> syslogger patch. It seems to work okay on Unix (with or without
> EXEC_BACKEND) but I'm not in a position to test it on Windows. Would
> someone give it a try and report back? Please check in particular that
> the logger shuts down cleanly after the postmaster is gone.