On a fully up2date'd RHL 8 machine, the new postgresql-server-7.2.3-5.80 RPM installs an /etc/init.d/postgresql with incorrect permissions. It's now not world-readable like all the other init scripts. [kaboom@verdande lib]$ ls -l /etc/init.d/postgresql -rwx------ 1 root root 6157 Dec 20 10:57 /etc/init.d/postgresql [kaboom@verdande lib]$ If I remember right, this bug is introduced by the errata -- I think it was correctly 755 before the update
This was an oversight on my part and I apologize. However, we have discussed the issue and have decided not to reissue the erratum with a modified initscript. Hopefully this will not cause any real problems. Should a non-root user wish to run the initscript for status purposes, they can get the same output by executing the following commands: $ /sbin/pidof postmaster (this will give you the process ID's of any running postmasters) $ cat /var/run/postmaster.pid (if the first command output nothing, but this command indicates that the file exists, then the service is most likely not running but crashed somehow so an old process ID is stored there to prevent it from restarting and potentially harming data) $ cat /var/lock/subsys/postgresql (if the above two commands output nothing, but this command indicates that the lockfile exists, then something happened to cause the postgresql system to lock. Data integrity should be verified and backups made prior to attempting to delete this file and/or restart the postmaster in this case) If none of these commands indicated the situations described above, then the database system is not currently running and it *should* be safe to start it back up (obviously as root if you're using the initscript). HTH.
Changing resolution to RAWHIDE since this is will be fixed in future releases of PostgreSQL.