XDCRTracer is a script for testing XDCR links between clusters and can handle varying topologies (Although is not a free in scope as Couchbase itself is yet).

XDCRTracer achieves this by using a list of source buckets and a list of destination buckets (Which includes all the source buckets). XDCRTracer will generate 1024 keys (One for each vBucket) into every source bucket and will check that they make it to every destination bucket.

XDCRTracer requires a JSON Configuration file, an example of the parameters is given in example.json. The path to the JSON file must be given as the first parameter to the script. e.g.

java -jar xdcr_tracer example.json

XDCRTracer will return the cluster name, bucket name and a list of vBuckets in the instance that the document was missing or an error was encountered whilst retrieving it.

 Missing documents in CHO/default: [100, 101, 102, ... , 199]

• Number interval How long to leave between upserting keys and checking for propagation.

• Number ttl How long should the tracing documents exist for. Minimum should be interval + time taken to insert keys + time taken to check keys. Recommend doubling interval.

• Boolean repeat Should the XDCR checker just run once or repeat

• Boolean debug Should debug messages be printed

• Object sources The associative array of source clusters to insert keys into and check keys on.

  • Array connection Array of connection strings to the cluster
  • Object buckets Associative array of bucket names to bucket passwords

• Object replicas The associative array of replica clusters to additionally check keys on.

It is recommended that you install maven, this can usually be handled by a package manager. For OSX Homebrew you can use

brew install maven

When using maven, building is simple:

mvn compile

If you require a JAR then:

mvn package