# Start redpanda cluster (redpanda-2 depends on redpanda-1, which depends on redpanda-0) & Triton docker compose up -d redpanda-2 loki grafana prometheus scylla-0 # If you're using the HA scylla configuration in docker-compose, use this instead: # docker compose up -d redpanda-2 loki grafana prometheus scylla-2 # Wait a lil bit for redpanda to come up before executing config commands echo "Waiting 45 sec for Redpanda and ScyllaDB to start!" sleep 45s # Create topics & shrink segment sizes so they trigger automatic archiving ## Simple sensor - 250mb segments docker exec redpanda-0 rpk topic create -r 3 -p 5 raw.test.simple docker exec redpanda-0 rpk topic alter-config raw.test.simple --set segment.bytes=250000000 ## Simple model - 250mb segments docker exec redpanda-0 rpk topic create -r 3 -p 5 derived.model-simple.add-subtract docker exec redpanda-0 rpk topic alter-config derived.model-simple.add-subtract --set segment.bytes=250000000 # disable redpanda anonymous telemetry docker exec redpanda-0 rpk redpanda config set rpk.enable_usage_stats false echo "Disabled anonymous telemetry reporting" # Check status of the cluster docker exec redpanda-0 rpk cluster config status # let you read the topic creation and cluster status before continuing # Bring up the rest of the stack docker compose up -d