Wed Dec 13 07:41:04 PST 2006
- Previous message: [Slony1-general] Problem in Slony-I 1.2.1 with PostgreSQL 8.2
- Next message: [Slony1-general] Is Slony 1.2.stable compatible with postgreSQl 8.2?
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Ashish Karalkar wrote: > Yes, its true,but evn for 2/3 tables the replication > doesnt start and same massage keeps flashing for a > long time its more than 6 hours > The diagnosis remains true. Replication cannot start until the old transactions (which are apparently being held open for over six hours) complete. This tells me that you've got a fairly serious problem with your application and/or users, namely that they are holding transactions open for a very long time. If you cannot bring that condition under control, you will find that Slony-I behaves rather badly. Performance of replication won't be very good, as some of the internals tables will tend to bloat in size. A universal principle in OLTP applications is that long running transactions are a bad thing. In some applications, the locks held will tend to lead to lock escalation that leads to deadlocks (of the informal sort). With PostgreSQL, the trouble is that you'll see a lot of "crud" build up in any tables that are heavily updated. You should check to see why your applications/users are holding transactions open for so long. It should not normally be necessary to hold a transaction open that long.
- Previous message: [Slony1-general] Problem in Slony-I 1.2.1 with PostgreSQL 8.2
- Next message: [Slony1-general] Is Slony 1.2.stable compatible with postgreSQl 8.2?
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list