This document describes deploying a local dss instance using an insecure multi-node CockroachDB cluster and HAProxy load balancers to distribute client traffic.
Multi-node local instance can be deployed by running ./build/dev/haproxy_local_setup.sh.
Once the setup is complete, the following endpoints will be available:
- Dummy OAuth Server: http://localhost:8085/token
- DSS Core Service: http://localhost:8082/healthy
- CockroachDB web UI: http://localhost:8080
Run ./check_dss.sh to ensure environment is up. The expected output is an empty list of ISAs (no ISAs have been announced).
Run ./build/dev/haproxy_local_setup.sh down to stop the system.
Setup includes three cockroachdb nodes: roacha, roachb and roachc , each on a separate container, an HAProxy server: dss-crdb-cluster-for-testing, and the dss services. All the services are connecting to HAProxy server, which would distribute the traffic across the cluster. HAProxy, acting as a load balancers decouples client health from the health of a single CockroachDB node. In cases where a node fails, the load balancer redirects client traffic to available nodes.
Setting up HAProxy requires generating a configuration file by running cockroach gen haproxy on one of the cluster nodes. that is preset to work with the running cluster. Generated haproxy.cfg file is then mounted to HAProxy container via local machine's ~/Download/haproxy/ folder.
To test the DB connection through HAProxy, run ./build/dev/check_scd_write.sh to perform some write operations on the database.
Stop one of the cluster nodes by running:
```docker rm -f roachc```
DSS services should still be up and running. Test it by running read operations:
```./build/dev/check_scd_read.sh```
Run following to bring up a new node in the cluster environment:
```docker run -d --name roachc \
--hostname roachc -p 8089:8080 \
--network dss_sandbox-default \
cockroachdb/cockroach:v21.2.3 start --insecure --join=roacha,roachb
```
Repeat calling check_*.sh scripts to test the cluster health.