Tue Feb 15 19:16:12 PST 2005
- Previous message: [Slony1-general] Can you add a new cluster and have it copy all data from origin?
- Next message: [Slony1-general] Error de slony
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
I wanted to actually create the new namespace so an issue in one (like a badly scripted node addition) won't effect any other. I have two set up right now, both using the same master DB. I HAD made a mistake before and it only effectd the one set thankfully. Because I did the init and subscribtions BEFORE loading in data to my master, any new node in the namespace comes fully into sync. *** update** Turns out I had a bad line in my scripts, so the cluster did not get set correctly. Now that I fixed it, it is actually copying over the data with no issues. Thanks for you time all. > Tass Eh wrote: > > >I have a set up with 2 clusters currently. I did the slon inits, and > >subscribed all nodes, then imported the data to my master. This then got > >all the data to the nodes, and allowed me to add more nodes that would sync > >up with the master from the beginning. > > > >I now want to add a new cluster to the system, and am trying to find a way > >to add it, and have the nodes in the new cluster copy over all the data , > >just as I was able to do with my first 2. > > > >Is there a way to do this? I suspect there is some set of values in the > >slon bits that needs to be changed. > > > >The only other way I can think off is to stop updates to my master, > >pg_dump, set up slon for the new cluster and then import the data and start > >up slon, then allow writes to the master again. > > > >Any suggestions or help would be appreciated. > > > > > I suspect you're not talking about creating a new cluster. > > A Slony-I cluster is a set of database nodes that share common > configuration, stored in the same _Cluster_Namespace for each of them. > > Adding a cluster, therefore, would mean to create a new namespace, new > set of nodes, new set of origins, and such. > > It sounds instead as if you merely want to add a node to an existing > cluster. > > That certainly can be done. The steps: > > 0. Get the schema to be used on the new node. > > You might do a pg_dump, and take out the CLUSTER namespace. There's > apparently a script in CVS head that knows how to do this. > > 1. Create the new node, and set up connectivity. > > STORE NODE creates the node > STORE PATH is needed to get paths to/from other nodes > STORE LISTEN entries are needed (if you have N nodes now, then you > need N 'listen paths') to let events propagate > > 2. Start a slon for the new node > > 3. Subscribe the new node to the set... > > SUBSCRIBE SET > -- > <http://linuxdatabases.info/info/slony.html> > -- Tass eh.. ---------------------------------------------
- Previous message: [Slony1-general] Can you add a new cluster and have it copy all data from origin?
- Next message: [Slony1-general] Error de slony
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Slony1-general mailing list