{"id":1845,"date":"2014-12-16T17:18:35","date_gmt":"2014-12-16T17:18:34","guid":{"rendered":"https:\/\/www.couchbase.com\/blog\/?p=1845"},"modified":"2025-06-13T23:52:04","modified_gmt":"2025-06-14T06:52:04","slug":"xdcr-aspnet-and-nancy","status":"publish","type":"post","link":"https:\/\/www.couchbase.com\/blog\/ko\/xdcr-aspnet-and-nancy\/","title":{"rendered":"ASP.NET \ubc0f Nancy\ub97c \uc0ac\uc6a9\ud55c XDCR"},"content":{"rendered":"

You’re probably aware of the new Cross Data Center Replication (XDCR<\/a>) feature of Couchbase Server 2.0.\u00a0 Its most obvious utility is to allow you to replicate data from one Couchbase cluster to another.\u00a0 However, there are more novel use cases for XDCR.\u00a0 One of the more notable examples is Couchbase and Elastic Search integration.<\/p>\n

Performing XDCR against a non-Couchbase cluster is possible because the communication endpoint is implemented using a simple RESTful API.\u00a0 By implementing a few simple methods, you can setup your own replication endpoint.\u00a0 There are a variety of reasons you might want to build your own XDCR endpoint.\u00a0 One simple example is to receive change notifications.<\/p>\n

Setting up an endpoint is pretty straightforward and may be done using any web framework suitable for building RESTful APIs.\u00a0 In .NET, one option is ASP.NET MVC and the new Web API.\u00a0 While a good option, a simpler solution would be to use the Sinatra<\/a> inspired Nancy<\/a>.<\/p>\n

Nancy is a lightweight, microframework for building HTTP based applications with .NET.\u00a0 In the most basic case, you create a module that extends from NancyModule<\/em> and create a route and handler.<\/p>\n

\n
public<\/span> class<\/span> SampleModule :<\/span> Nancy.<\/span>NancyModule<\/span>
\n{<\/span>
\npublic<\/span> SampleModule(<\/span>)<\/span>
\n{<\/span>
\nGet[<\/span>“\/”<\/span>]<\/span> =<\/span> _ =><\/span> “Hello World!”<\/span>;<\/span>
\n}<\/span>
\n}<\/span><\/div>\n<\/div>\n

To demonstrate how to create an XDCR endpoint using Nancy, I’ve published a Couchbase Labs project on GitHub named couchbase-xdcr-nancy<\/a>.\u00a0 This project is based heavily on Jasdeep’s couchbase-xdcr-sinatra<\/a> code.<\/p>\n

Inside the Visual Studio project, you’ll find a number of plumbing files.\u00a0 I won’t go over the code in detail here, but will show a couple of highlights.\u00a0 The most important code is inside of the XdcrModule<\/em> class.\u00a0 It’s there where you’ll find the handlers for the various resource requests made by the XDCR service.<\/p>\n

Within the constructor for this module, you’ll find a handfull of GET handlers that handle requests to URIs starting with \/pools.\u00a0 These are effectively handshake URIs used by XDCR to discover information about the cluster and its buckets (similar to the endpoints used by Couchbase SDKs to bootstrap client instances).\u00a0 Note that in this sample, the bucket is hardcoded to “default.”\u00a0 You could easily modify that value or make it a configuration setting.<\/p>\n

\n
\n

Get[<\/span>“\/pools”<\/span>]<\/span> =<\/span> x =><\/span>
\n{<\/span>
\nvar output =<\/span> new<\/span><\/a>
\n{<\/span>
\npools =<\/span> new<\/span><\/a> object<\/span>[<\/span>]<\/span>
\n{<\/span>
\nnew<\/span><\/a> {<\/span> name =<\/span> “default”<\/span>, uri =<\/span> “\/pools\/default?uuid=”<\/span> +<\/span> UUID_POOL }<\/span>
\n}<\/span>,
\nuuid =<\/span> UUID_POOL
\n}<\/span>;<\/span><\/p>\n

return<\/span> Response.<\/span>AsJson<\/span>(<\/span>output)<\/span>;<\/span>
\n}<\/span>;<\/span><\/p>\n<\/div>\n<\/div>\n

After the GET handlers, there are a couple of POST handlers that respond to the actual XDCR feed.\u00a0 The XDCR service will pass parameters as part of the path.\u00a0 Regex based routes are used for this purpose.<\/p>\n

The first POST handler receives a list of documents and their revisions from the XDCR service.<\/p>\n

\n
\n

Post[<\/span>REGEX_REVS_DIFF]<\/span> =<\/span> x =><\/span>
\n{<\/span>
\nvar body =<\/span> “”<\/span>;<\/span>
\nContext.<\/span>Request<\/span>.<\/span>Body<\/span>.<\/span>Position<\/span> =<\/span> 0<\/span>;<\/span>
\nusing<\/span> (<\/span>var sr =<\/span> new<\/span><\/a> StreamReader(<\/span>Context.<\/span>Request<\/span>.<\/span>Body<\/span>)<\/span>)<\/span>
\n{<\/span>
\nbody =<\/span> sr.<\/span>ReadToEnd<\/span>(<\/span>)<\/span>;<\/span>
\n}<\/span>
\nvar jobj =<\/span> JObject.<\/span>Parse<\/span>(<\/span>body)<\/span>;<\/span><\/p>\n

var outDict =<\/span> new<\/span><\/a> Dictionary<<\/span>string<\/span>, object<\/span>=<\/span>“”<\/span>><\/span>(<\/span>)<\/span>;<\/span>
\nforeach<\/span> (<\/span>var item in<\/span> jobj)<\/span>
\n{<\/span>
\nvar key =<\/span> item.<\/span>Key<\/span>;<\/span>
\nvar rev =<\/span> item.<\/span>Value<\/span>.<\/span>ToString<\/span>(<\/span>)<\/span>;<\/span>
\nif<\/span> (<\/span>handler.<\/span>IsMissing<\/span>(<\/span>key, rev)<\/span>)<\/span>
\n{<\/span>
\noutDict[<\/span>key]<\/span> =<\/span> new<\/span><\/a> {<\/span> missing =<\/span> rev }<\/span>;<\/span>
\n}<\/span>
\n}<\/span><\/p>\n

return<\/span> Response.<\/span>AsJson<\/span>(<\/span>outDict)<\/span>;<\/span>
\n}<\/span>;string<\/span>,><\/span><\/span><\/p>\n<\/div>\n<\/div>\n

Within this handler, the JSON is parsed to discover the key and revision.\u00a0 If that combination is not found to be replicated, then it is added to the list of “missing” keys included in the response to the request.\u00a0 XDCR will then send those documents to the another POST handler, where they’ll be created.<\/p>\n

\n
\n

Post[<\/span>REGEX_BULK_DOCS]<\/span> =<\/span> x =><\/span>
\n{<\/span>
\nvar body =<\/span> “”<\/span>;<\/span>
\nContext.<\/span>Request<\/span>.<\/span>Body<\/span>.<\/span>Position<\/span> =<\/span> 0<\/span>;<\/span>
\nusing<\/span> (<\/span>var sr =<\/span> new<\/span><\/a> StreamReader(<\/span>Context.<\/span>Request<\/span>.<\/span>Body<\/span>)<\/span>)<\/span>
\n{<\/span>
\nbody =<\/span> sr.<\/span>ReadToEnd<\/span>(<\/span>)<\/span>;<\/span>
\n}<\/span>
\nvar jobj =<\/span> JObject.<\/span>Parse<\/span>(<\/span>body)<\/span>;<\/span><\/p>\n

var newEdits =<\/span> jobj.<\/span>Value<\/span><<\/span>bool<\/span>><\/span>(<\/span>“new_edits”<\/span>)<\/span>;<\/span>
\nvar docs =<\/span> jobj.<\/span>Value<\/span><<\/span>JArray><\/span>(<\/span>“docs”<\/span>)<\/span>;<\/span>
\nforeach<\/span> (<\/span>var doc in<\/span> docs)<\/span>
\n{<\/span>
\nvar originalDoc =<\/span> Encoding.<\/span>UTF8<\/span>.<\/span>GetString<\/span>(<\/span>Convert.<\/span>FromBase64String<\/span>(<\/span>doc.<\/span>Value<\/span><<\/span>string<\/span>><\/span>(<\/span>“base64”<\/span>)<\/span>)<\/span>)<\/span>;<\/span>
\nvar meta =<\/span> doc[<\/span>“meta”<\/span>]<\/span> as<\/span> JObject;<\/span><\/p>\n

var document =<\/span> new<\/span><\/a> Document
\n{<\/span>
\nId =<\/span> meta.<\/span>Value<\/span><<\/span>string<\/span>><\/span>(<\/span>“id”<\/span>)<\/span>,
\nRevision =<\/span> meta.<\/span>Value<\/span><<\/span>string<\/span>><\/span>(<\/span>“rev”<\/span>)<\/span>,
\nExpiration =<\/span> meta.<\/span>Value<\/span><<\/span>int<\/span>><\/span>(<\/span>“expiration”<\/span>)<\/span>,
\nFlags =<\/span> meta.<\/span>Value<\/span><<\/span>int<\/span>><\/span>(<\/span>“flags”<\/span>)<\/span>,
\nValue =<\/span> originalDoc
\n}<\/span>;<\/span><\/p>\n

handler.<\/span>CreateDocument<\/span>(<\/span>document)<\/span>;<\/span>
\n}<\/span><\/p>\n

return<\/span> HttpStatusCode.<\/span>Created<\/span>;<\/span>
\n}<\/span>;<\/span><\/p>\n<\/div>\n<\/div>\n

Within the sample, I’ve included a very simple interface used to create a pluggable handler that will check the existence of the document and create it when necessary.<\/p>\n

\n
\n

public<\/span> interface<\/span> IReplicationHandler
\n{<\/span>
\nbool<\/span> IsMissing(<\/span>string<\/span> key, string<\/span> rev)<\/span>;<\/span><\/p>\n

void<\/span> CreateDocument(<\/span>Document document)<\/span>;<\/span>
\n}<\/span><\/p>\n<\/div>\n<\/div>\n

For demonstration purposes, I’ve also included an XmlReplicator<\/em> class that implements this interface.\u00a0 This class uses LINQ to XML to update and query an XML document that contains documents fed to it by the XDCR service.<\/p>\n

\n
\n

public<\/span> class<\/span> XmlReplicator :<\/span> IReplicationHandler
\n{<\/span>
\nprivate<\/span> readonly<\/span> string<\/span> _path;<\/span>
\nprivate<\/span> XDocument doc =<\/span> new<\/span><\/a> XDocument(<\/span>)<\/span>;<\/span><\/p>\n

public<\/span> XmlReplicator(<\/span>string<\/span> path =<\/span> @”C:tempreplication.xml”<\/span>)<\/span>
\n{<\/span>
\n_path =<\/span> path;<\/span>
\nif<\/span> (<\/span>!<\/span> File.<\/span>Exists<\/span>(<\/span>_path)<\/span>)<\/span>
\n{<\/span>
\n\/\/var xml = new XElement(“documents”);<\/span>
\nFile.<\/span>WriteAllText<\/span>(<\/span>path,““<\/span>, Encoding.<\/span>UTF8<\/span>)<\/span>;<\/span>
\n}<\/span>
\n}<\/span><\/p>\n

…<\/span>
\n}<\/span><\/p>\n<\/div>\n<\/div>\n

To check if the document exists, IsMissing <\/em>will query the XML for a document <\/em>element that has child id<\/em> and rev<\/em> elements matching the supplied key and revision.<\/p>\n

\n
public<\/span> bool<\/span> IsMissing(<\/span>string<\/span> key, string<\/span> rev)<\/span>
\n{<\/span>
\nvar xml =<\/span> XDocument.<\/span>Load<\/span>(<\/span>_path)<\/span>;<\/span>
\nvar documents =<\/span> xml.<\/span>Document<\/span>.<\/span>Root<\/span>.<\/span>Elements<\/span>(<\/span>“document”<\/span>)<\/span>;<\/span>
\nvar document =<\/span> documents.<\/span>Where<\/span>(<\/span>d =><\/span> d.<\/span>Element<\/span>(<\/span>“meta”<\/span>)<\/span>.<\/span>Element<\/span>(<\/span>“rev”<\/span>)<\/span>.<\/span>Value<\/span> ==<\/span> rev &&<\/span> d.<\/span>Element<\/span>(<\/span>“meta”<\/span>)<\/span>.<\/span>Element<\/span>(<\/span>“id”<\/span>)<\/span>.<\/span>Value<\/span> ==<\/span> key)<\/span>;<\/span>
\nreturn<\/span> document.<\/span>Count<\/span>(<\/span>)<\/span> ==<\/span> 0<\/span>;<\/span>
\n}<\/span><\/div>\n<\/div>\n

Documents are created simply by adding them to the existing XML file.<\/p>\n

\n
\n

public<\/span> void<\/span> CreateDocument(<\/span>Document document)<\/span>
\n{<\/span>
\nvar xml =<\/span> XDocument.<\/span>Load<\/span>(<\/span>_path)<\/span>;<\/span><\/p>\n

var docElement =<\/span> new<\/span><\/a> XElement(<\/span>“document”<\/span>,
\nnew<\/span><\/a> XElement(<\/span>“meta”<\/span>,
\nnew<\/span><\/a> XElement(<\/span>“id”<\/span>, document.<\/span>Id<\/span>)<\/span>,
\nnew<\/span><\/a> XElement(<\/span>“rev”<\/span>, document.<\/span>Revision<\/span>)<\/span>,
\nnew<\/span><\/a> XElement(<\/span>“expiration”<\/span>, document.<\/span>Expiration<\/span>)<\/span>,
\nnew<\/span><\/a> XElement(<\/span>“flags”<\/span>, document.<\/span>Flags<\/span>)<\/span>
\n)<\/span>,
\nnew<\/span><\/a> XElement(<\/span>“value”<\/span>, new<\/span><\/a> XCData(<\/span>document.<\/span>Value<\/span>)<\/span>)<\/span>
\n)<\/span>;<\/span><\/p>\n

xml.<\/span>Document<\/span>.<\/span>Root<\/span>.<\/span>Add<\/span>(<\/span>docElement)<\/span>;<\/span>
\nxml.<\/span>Save<\/span>(<\/span>_path)<\/span>;<\/span>
\n}<\/span><\/p>\n<\/div>\n<\/div>\n

By default, I’ve set up Nancy to use the ASP.NET development server.\u00a0 You can host Nancy however you wish (IIS, self hosted, etc.).\u00a0 Once you have the project running, you’ll need to setup XDCR in the Couchbase admin console.\u00a0 Some values to know:<\/p>\n