1. Print current webmin screens for posgresql: particularly allowed hosts (don't need to do this,
can use old pg_hba.conf copied to /var/lib/pgsql-oldversion in step 2)
2. Backup in as many ways as possible.
terminal:
ps postgres
cd /home/postgresql
pg_dumpall > oldversion_dumpall
exit
webmin:
backup important databases individually both as a tar file and sql file
terminal:
cp -R /var/lib/pgsql /var/lig/pgsql-oldversion
3. Stop database server
terminal:
service postgresql stop
4. remove postgresql rpms
terminal:
rpm -e `rpm -qa|grep postgresql*` --nodeps
5. remove database cluster
terminal:
rm -rf /var/lib/pgsql
6. instal new rpms
terminal:
rpm -ivh postgresql-libs-8.2.7-1PGDG.rhel5.i686.rpm
rpm -ivh postgresql-8.2.7-1PGDG.rhel5.i686.rpm
rpm -ivh postgresql-devel-8.2.7-1PGDG.rhel5.i686.rpm
rpm -ivh postgresql-contrib-8.2.7-1PGDG.rhel5.i686.rpm
rpm -ivh postgresql-server-8.2.7-1PGDG.rhel5.i686.rpm
rpm -ivh compat-postgresql-libs-4-1PGDG.rhel5.i686.rpm
7. init and start server
terminal:
service postgresql initdb
service postgresql start
8. set postgresql password
terminal:
su postgres
psql
alter user postgres with password 'YOUR_PASSWORD';
\q
9. edit config files if needed
10. set up allowed host in webmin, or copy over previous pg_hba.conf (of course back up new one)
11. if Step 9 and Step 10 are done, restart posgresql server
terminal:
service postgresql restart
12. restore databases, we did it through webmin with tar files from Step 2, but could be done from the dumpall.
Note: This went succesfully for us on a test server, no problems restoring data, and moving on. If you do run into a problem, be prepared to restore the old rpms,
and use dumpall or copy back the /var/lib/pgsql-oldversion. Will update for further clarifications after performing on live server, if need be.
