Using S3 with Spinnaker instead of Cassandra Backend

Cassandra is used by Spinnaker to persist pipelines configuration and cluster information. While Cassandra may suit your needs it might be overkill for POCs or early development phases where the number of configurations and continuous deployments are small. While Spinnaker can be configured to have an Amazon S3 backend, it’s not always trivial to setup.

You might find an error in the front50 log that looks like the following:

org.springframework.beans.factory.NoUniqueBeanDefinitionException: No qualifying bean of type [com.netflix.spinnaker.front50.model.pipeline.PipelineDAO] is defined: expected single matching bean but found 2: pipelineRepository,s3PipelineDAO

Setting an Active Spring Profile

Spinnaker heavily relies on the Spring framework. The Spring framework expects there only one object that matches the interface for a backend repository.

You can change the default options passed into spinnaker by editing the defaults file for spinnaker /etc/default/spinnaker

in there you can change the default active profile by setting the java front50 options to the following:

FRONT50_OPTS="-Dspring.profiles.active=s3"

Another common mistake is that folks forget to disable Cassandra in their front50.yml which you can find at /opt/spinnaker/config. Make the following change:

    enabled: false```
Hope these changes help!
If you haven't heard of Spinnaker, its a next generation cloud deployment tool originally built by Netflix and subsequently open-sourced.  It's best know for continuous delivery at Netflix and helping them achieve more than 4,000 deployments/day.  It's currently being contributed to by over 30 developers from Google, Pivotal, Netflix, Veritas and Microsoft.