ContextQMD
Back to Cadence

Setup Cadence Locally With Replication

docs/howtos/setup-cadence-locally-with-replication.md

1.4.02.3 KB
Original Source

How to run Cadence locally with replication

This is a guide for using a local laptop instance and running three Cadence clusters, all cross-replicted. They will (for the purposes of demonstration) be entirely independentent, but poll one another for workflows for replication and write to their own keyspaces independently.

It's a good way to work through how Cadence's domains are configured with replication, how failovers work and so forth.

This is a somewhat more advanced tutorial, and so it's recommended to ensure that you have a basic familiarity with Cadence and it's concepts first before trying this.

Prerequisites

  • Docker
  • The ability to perform a local go build

Getting started

All commands are expected to be run from the repo's base path.

  1. Ensure there's no existing containers running or preexisting state. You'll be creating a new keyspace and a new docker-container only for Cassandra, and so this can conflict in a few ways with other cassandra instances for fully-containerized cadence. Be sure to stop and/or remove them before running this.

  2. Ensure you build the cadence-server binaries by using make bins. This will build the main application and produce a binary build of the local instance of cadence in the repo's base path.

  3. Start docker with the XDC configuration: docker compose -f docker/dev/cassandra.yml up -d

  4. setup the schema for each of the instances with until make install-schema-xdc; do sleep 1; done (since Cassandra can take a while to setup, loop this check until it completes).

  5. start (in tmux or terminal windows) each of the Cadence instances

  • ./cadence-server --config config --zone xdc --env development_xdc_cluster0 start
  • ./cadence-server --config config --zone xdc --env development_xdc_cluster1 start
  • ./cadence-server --config config --zone xdc --env development_xdc_cluster2 start

  1. Register a global domain that has multiple clusters enabled using the cadence CLI, so that it's worklows will be replicated across clusters: ./cadence --domain test-domain domain register --gd true --clusters cluster0,cluster1

  2. Start and test workflows as needed and failover the domain to make another cluster active with a command such as this: ./cadence --domain test-domain domain update --ac cluster1