I'm planning on setting up a master database and multiple hot standby slaves using streaming replication.
If I use a large(*) value on wal_keep_segments do I really need archive_mode = on then?
Any potential problems with this strategy I should be aware about?
(*) With large value I mean a value such that a failed slave will be up and running again (with a few days margin) before the master server starts to clean out old WAL segments.
I believe you are correct. At least that's what it says here...