On 02/17/2018 02:00 PM, Rich Shepard wrote:
> Hi folks,
>
> Today I upgraded from -9.6.6 to -10.2 on my Slackware-14.2 desktop. The
> user and group IDs changed from before, but I have that all fixed now.
> Starting postgres (as user postgres) succeeded, but the role for me (as a
> use and owner of most databases) seems to have become lost during the
> transition.
How did you upgrade, dump/restore or pg_upgrade?
If dump/restore did you use pg_dumpall or pg_dump individual databases?
If pg_dump only did you use pg_dumpall -g to get the globals(of which
roles are one) to restore the global values to the new cluster?
>
> I try to open a database and see this:
>
> $ psql jerr2018-02-17 13:45:35.852 PST [5839] FATAL: password
> authentication failed for user "rshepard"
> 2018-02-17 13:45:35.852 PST [5839] DETAIL: Role "rshepard" does not exist.
> Connection matched pg_hba.conf line 80: "local all
> all md5"
> 2018-02-17 13:45:35.853 PST [5839] LOG: could not send data to client:
> Broken pipe
>
> So I edited pg_hba.conf to change the method to 'trust' as I'm the only
> user on this system. Ran /etc/rc.d/rc.postfix reload and see:
>
> # /etc/rc.postgresql reload
> Could not find 'postgres' binary. Maybe PostgreSQL is not installed
> properly?
What is in rc.postgresql?
Is there maybe another start script in etc/ for the new version?
>
> $ ps ax | grep postgres
> 5826 pts/0 S 0:00 postgres -D /var/lib/pgsql/10.2/data
> 5828 ? Ss 0:00 postgres: checkpointer process
> 5829 ? Ss 0:00 postgres: writer process
> 5830 ? Ss 0:00 postgres: wal writer process
> 5831 ? Ss 0:00 postgres: autovacuum launcher process
> 5832 ? Ss 0:00 postgres: stats collector process
> 5833 ? Ss 0:00 postgres: bgworker: logical replication
> launcher
>
> I would appreciate a pointer on what to check to determine why I cannot
> reload postgres to see the changed pg_hba.conf and let me access my
> databases.
>
> Regards,
>
> Rich
>
>
>
>
--
Adrian Klaver
adrian.klaver@aklaver.com