Miguel wrote: >Alan Hodgson wrote: > > > >>On March 20, 2006 08:46 am, Miguel <[EMAIL PROTECTED]> wrote: >> >> >> >> >>>Hi, im wondering if the subscriber nodes has to start without any data?, >>>i have a 25Gb database and i want to replicate it with another server, >>>same hardware specs, can i do i pg_dump/restore and slony will do only >>>incremental updates from there? >>> >>> >>> >>> >>Slony will delete the existing data and start over. You are better off >>starting with an empty database. You do need to copy the schema before >>starting, though. >> >> >> >> >> >does slony need exclusive lock for this initial transfer? > > It will ultimately need exclusive locks on the subscriber in order to complete the copy, as it will be modifying the table schemas a bit.
In version 1.2, the plan is for those exclusive locks to be requested up front, so that there is no opportunity for deadlock. >also, if i dont start the slon process, the master node operate in >normal "mode" , i mean, slony-i agnostic? > > I'm not quite sure what you mean by that. If you don't start a slon for the node that is the origin for a set, then: - tuples will build up in sl_log_1 as tables that were added to replication sets are modified, and will never be cleaned out. - no SYNC events will get issued, thereby preventing event activity from propagating. Maybe you could explain what you think a "normal mode" is, or what "Slony-I agnostic" means; we might be able to correct some misunderstandings, there... _______________________________________________ Slony1-general mailing list [email protected] http://gborg.postgresql.org/mailman/listinfo/slony1-general
