Hi all again,
I rigged up a solution with PGPool (!) in master-slave mode, using the existing documentation and it works like a charm with a downtime of only a few seconds, which is acceptable in our case. However assume the following scenario:
1) Master goes down
2) slave gets promoted to master
owing to the usage of pgpool, this works fine. However, assuming that there is a certain amount of write traffic that goes to the new master. When the old master (sorry for the terminology) comes back up, how can I "sync" him with all the changes that have been performed? This is one of the two pieces of the puzzle missing for me (the other one is pgpool specific and I directed my query to their specific mailing list).
Thanks for any replies.