We have replication setup already (bi-directional). How can I add JUST a "subscriber" to the replication? I want to do this with T-SQL. There does not seem to be any documentation that says "here are the steps to just add a subscriber in an existing replication environment". The steps are more geared towards "This is how you set up replication". So I am thinking ..... On each publisher, do I simply execute "sp_addsubscriber" ? Are there any other steps involved to add the subscriber? This seems too simple. Also, it is a good practice for a "subscriber only" to have a distributor database? is it even necessary if the subscriber does not publish? There is a lot of activity on the publishers, so I can build a distributor database if this is a good practice for a subscriber.. TIA
Its not well documented. There is a single kb article which talks about it and there is an error in it. Basically you have to adjust all identity seeds and increments for the new subscriber (the seed should be a multiple of 3 on node a, 3+1 on node b, and 3 +2 on node c). You need to adjust the next identity value to be assigned to be a value that would fall within the window of that publisher. For example consider the following server current identity value seed adjust to node a 6 3 No adjustment necessary node b 5 4 7 node c 34 5 35 Now you currently have nodes a and nodes b publishering and subscribing to each other. You have to have node c publish and subscrib to node a, and node c publish and subscribe to node b. wrote in message news:4f864065-1e89-4890-97c3-663583783ec1@k13g2000hse.googlegroups.com... > We have replication setup already (bi-directional). > > How can I add JUST a "subscriber" to the replication? > > I want to do this with T-SQL. > > There does not seem to be any documentation that says "here are the > steps to just add a subscriber in an existing replication > environment". The steps are more geared towards "This is how you set > up replication". > > So I am thinking ..... > On each publisher, do I simply execute "sp_addsubscriber" ? Are > there any other steps involved to add the subscriber? This seems too > simple. > > Also, it is a good practice for a "subscriber only" to have a > distributor database? is it even necessary if the subscriber does not > publish? There is a lot of activity on the publishers, so I can build > a distributor database if this is a good practice for a subscriber.. > > TIA > > >
Actually, I am not looking for node c to be a publisher, just a subscriber. This is where I am getting confused ... Right now "node a" and "node b" are publishers and subscribers. How do i add "node c" as a subscriber only? I guess I am looking for exact execution steps. 1. exec sp_addsubscriber (pointing to node c) on both "node a" and "node b" ? 2. should "node c" as a "subscriber only" have a distributor database? are there any other steps? On Jul 7, 6:22 am, "Hilary Cotter" wrote: > Its not well documented. There is a single kb article which talks about it > and there is an error in it. > > Basically you have to adjust all identity seeds and increments for the new > subscriber (the seed should be a multiple of 3 on node a, 3 on node b, and > 3 on node c). You need to adjust the next identity value to be assigned > to be a value that would fall within the window of that publisher. For > example consider the following > > server current identity value seed adjust to > node a 6 3 No adjustment > necessary > node b 5 4 7 > node c 34 5 35 > > Now you currently have nodes a and nodes b publishering and subscribing to > each other. You have to have node c publish and subscrib to node a, and node > c publish and subscribe to node b. >
Node c would be a subscriber to Node a or node b. Data entered at nodea will makes its way to nodeb and nodec, data entered at nodeb will make its way to node a and nodec. Data entered at node c will stay there. Note that in the initail subscription you will have to be careful that the snapshot does not what existing data. So select the keep existing object unchanged in the destination object settings. wrote in message news:1493cf8c-3a5e-44da-b955-d23997ed63f2@56g2000hsm.googlegroups.com... Actually, I am not looking for node c to be a publisher, just a subscriber. This is where I am getting confused ... Right now "node a" and "node b" are publishers and subscribers. How do i add "node c" as a subscriber only? I guess I am looking for exact execution steps. 1. exec sp_addsubscriber (pointing to node c) on both "node a" and "node b" ? 2. should "node c" as a "subscriber only" have a distributor database? are there any other steps? On Jul 7, 6:22 am, "Hilary Cotter" wrote: > Its not well documented. There is a single kb article which talks about it > and there is an error in it. > > Basically you have to adjust all identity seeds and increments for the new > subscriber (the seed should be a multiple of 3 on node a, 3+1 on node b, > and > 3 +2 on node c). You need to adjust the next identity value to be assigned > to be a value that would fall within the window of that publisher. For > example consider the following > > server current identity value seed adjust to > node a 6 3 No adjustment > necessary > node b 5 4 7 > node c 34 5 35 > > Now you currently have nodes a and nodes b publishering and subscribing to > each other. You have to have node c publish and subscrib to node a, and > node > c publish and subscribe to node b. >
