CouchDB "replication" uses the HTTP API to replicate a database, whereas
the somewhat confusingly named "internal replication" is used within the
cluster itself as a way to synchronize state between the shard replicas.
The protocol for internal replication uses the Erlang distribution protocol
to communicate between nodes. So if your cluster is a connected set of
Erlang nodes then they will be able to synchronize database/shard state
amongst themselves.
As an example, consider a single node 1.x CouchDB server at
test1.foo.com:5984 where you want to replicate to a CouchDB 2.0 cluster at
test2.bar.com that has three db nodes and a load balancer, with the load
balancer actually having the test2.bar.com address. You communicate with
the cluster by way of the load balancer which will distribute requests to
all the nodes. The details of the cluster should be mostly transparent at
the HTTP layer, and so you can just do something like:
curl -i -X POST test1.foo.com/_replicator -d '{"source":"test1.foo.com/my_db",
"target":"test2.bar.com"}'
This will replicate "my_db" from test1 to test2 using the HTTP protocol
like you would expect from CouchDB. Within the test2 cluster the nodes will
communicate amongst themselves using internal replication (non HTTP) to
ensure the data is fully synchronized between the shard replicas.
Hopefully that clarifies things a bit.
-Russell
On Tuesday, September 6, 2016, Joey Samonte <[email protected]>
wrote:
> I think I understand it. Hoping to find more documentation on it.
>
> > From: [email protected] <javascript:;>
> > Date: Tue, 6 Sep 2016 15:55:00 -0700
> > Subject: Re: Adding a node to cluster
> > To: [email protected] <javascript:;>
> >
> > Nodes in the cluster do not replicate to one another.
> >
> > Replication takes place between databases. A single node isn't a
> > database; a clustered database spans multiple nodes.
> >
> > Each node has a black-box lump of data that happens to have a fraction
> > of a database inside, but that's an implementation detail. It's the
> > wrong level of abstraction to work with.
> >
> > On Tue, Sep 6, 2016 at 1:52 PM, Joey Samonte
> > <[email protected] <javascript:;>> wrote:
> > > If you have two nodes in the cluster, the replication should be
> two-way between the nodes?
> > > What if there are three nodes? How should the replication be setup
> between them?
> > >
> > >> Subject: Re: Adding a node to cluster
> > >> From: [email protected] <javascript:;>
> > >> Date: Mon, 5 Sep 2016 23:08:04 +0200
> > >> To: [email protected] <javascript:;>
> > >>
> > >> Hi,
> > >>
> > >> clustering and replication are indeed two (very) separate things -
> and you won't get a Cluster by setting up replication. Again: treat the two
> as separate. Clustering turns several shards (on several nodes) into one
> database (from an user/caller perspective) while replication happens
> _between_ databases.
> > >> Consequently, technical underpinnings differ as well, as Bob
> explained below.
> > >>
> > >> Hope that gets things in perspective a little...
> > >>
> > >> Best
> > >> Sebastian
> > >>
> > >> Von meinem iPhone gesendet
> > >>
> > >> > Am 05.09.2016 um 22:47 schrieb Joey Samonte <
> [email protected] <javascript:;>>:
> > >> >
> > >> > Does this mean that setting up replication is separate from setting
> up clustering?
> > >> >
> > >> > Does replication needs to be bi-directional between nodes?
> > >> >
> > >> >> From: [email protected] <javascript:;>
> > >> >> Subject: Re: Adding a node to cluster
> > >> >> Date: Thu, 25 Aug 2016 11:10:45 +0100
> > >> >> To: [email protected] <javascript:;>
> > >> >>
> > >> >> Ok, seems I've confused you.
> > >> >>
> > >> >> Couchdb replication occurs over http or https, as you know. The
> nodes in a couchdb 2.0 cluster do not communicate with each other over
> http. They use Erlang rpc. Erlang rpc can be configured for TLS
> encryption. It's in the Erlang faq and is fairly simple to set up in newer
> Erlang releases.
> > >> >>
> > >> >> I feel I owe an example of 2.0 cluster that exclusively uses TLS
> for all communications.
> > >> >>
> > >> >> Sent from my iPhone
> > >> >>
> > >> >>> On 24 Aug 2016, at 20:47, Joey Samonte <
> [email protected] <javascript:;>> wrote:
> > >> >>>
> > >> >>> What if we remove the reverse proxy and just set up the CouchDB
> nodes to allow only SSL connections, port 6984? https://wiki.apache.org/
> couchdb/How_to_enable_SSL
> > >> >>>
> > >> >>>> Subject: Re: Adding a node to cluster
> > >> >>>> From: [email protected] <javascript:;>
> > >> >>>> Date: Wed, 24 Aug 2016 19:43:51 +0100
> > >> >>>> To: [email protected] <javascript:;>
> > >> >>>>
> > >> >>>> Assuming you mean a 2.0 cluster, no, all those nodes need to be
> able to communicate with erlang rpc (service discovery over port 4369 and
> then whatever port the node is running ong).
> > >> >>>>
> > >> >>>>> On 24 Aug 2016, at 12:36, Joey Samonte <
> [email protected] <javascript:;>> wrote:
> > >> >>>>>
> > >> >>>>> Good day,
> > >> >>>>>
> > >> >>>>> Is it possible to add a node to a cluster from Fauxton if the
> remote host is behind a reverse proxy (nginx) configured as HTTPS?
> > >> >>>>>
> > >> >>>>> Regards,
> > >> >>>>> Joey
> > >> >
> > >
>
