pgsql handler does not report failed database dumps

Debug: su - postgres -c "/usr/bin/pg_dumpall | /bin/gzip > /var/backups/postgres/myserver-all.sql.gz"
Debug: pg_dump: [archiver (db)] connection to database "mydatabase" failed: FATAL: database "mydatabase" does not exist pg_dumpall: pg_dump failed on database "mydatabase", exiting
Info: Successfully finished dump of pgsql cluster
Info: <<<< finished action /etc/backup.d/20.pgsql: SUCCESS

If I had not been running this pgsql backup handler in debug mode, I had not become aware of one of the databases not being dumped. It would be better if the whole process would exit with a non-zero status code if pd_dump failed to create just one of the dumps. Otherwise, you may end up with one database missing from your backups without the sysadmin ever becoming aware of it until she needs them.

I admit that this situation is somewhat constructed, though. I dropped the "mydatabase" by the time backupninja was creating the dumps. Nevertheless, there could be other resons for the database missing, or similar errors could occur and not get reported.

This is about backupninja 0.9.6-4 on Debian/GNU Linux 5.0.3 and postgresql 8.3.8-0lenny1.

(from redmine: created on 2009-10-18, closed on 2010-01-08)