feat: cluster deployment — build both stemedb-api and stemedb-node binaries

- Dockerfile: build stemedb-api + stemedb-node, add entrypoint.sh
- entrypoint.sh: runs both binaries when STEMEDB_CLUSTER_MODE=true
- Woodpecker: deploy to StatefulSet instead of Deployment

.woodpecker.yml

@@ -26,6 +26,6 @@ steps:
         from_secret: kubeconfig
     commands:
       - mkdir -p /root/.kube && echo "$KUBECONFIG_DATA" | base64 -d > /root/.kube/config
-      - kubectl set image deployment/stemedb-api stemedb-api=registry.threesix.ai/stemedb-api:${CI_COMMIT_SHA:0:8} -n stemedb
+      - kubectl set image statefulset/stemedb stemedb=registry.threesix.ai/stemedb-api:${CI_COMMIT_SHA:0:8} -n stemedb
-      - kubectl rollout status deployment/stemedb-api -n stemedb --timeout=300s
+      - kubectl rollout status statefulset/stemedb -n stemedb --timeout=300s
     depends_on: [build]

Dockerfile

@@ -42,9 +42,9 @@ RUN cargo chef cook --release --recipe-path recipe.json
 # Inherits compiled deps from cacher; only workspace source is compiled here.
 FROM cacher AS builder
 COPY . .
-RUN cargo build --release -p stemedb-api
+RUN cargo build --release -p stemedb-api -p stemedb-cluster
 # Strip debug symbols before copying to runtime image
-RUN strip target/release/stemedb-api
+RUN strip target/release/stemedb-api target/release/stemedb-node
 
 # Stage 4: Runtime — minimal production image
 FROM debian:bookworm-slim AS runtime
@@ -59,19 +59,22 @@ RUN apt-get update && \
 RUN useradd --system --no-create-home --shell /bin/false stemedb
 
 COPY --from=builder /app/target/release/stemedb-api /usr/local/bin/stemedb-api
+COPY --from=builder /app/target/release/stemedb-node /usr/local/bin/stemedb-node
+COPY scripts/entrypoint.sh /usr/local/bin/entrypoint.sh
 
-RUN mkdir -p /data/wal /data/db && chown -R stemedb:stemedb /data
+RUN chmod +x /usr/local/bin/entrypoint.sh && \
+    mkdir -p /data/wal /data/db && chown -R stemedb:stemedb /data
 
 USER stemedb
 
 ENV STEMEDB_WAL_DIR=/data/wal \
     STEMEDB_DB_DIR=/data/db \
     STEMEDB_BIND_ADDR=0.0.0.0:18180 \
-    RUST_LOG=stemedb_api=info
+    RUST_LOG=stemedb_api=info,stemedb_cluster=info
 
-EXPOSE 18180
+EXPOSE 18180 18181 18182 18183
 
-HEALTHCHECK --interval=5s --timeout=3s --start-period=10s --retries=3 \
+HEALTHCHECK --interval=5s --timeout=3s --start-period=15s --retries=3 \
     CMD curl -f http://localhost:18180/v1/health || exit 1
 
-CMD ["stemedb-api"]
+ENTRYPOINT ["entrypoint.sh"]

scripts/entrypoint.sh

@@ -0,0 +1,37 @@
+#!/bin/sh
+# StemeDB cluster entrypoint — runs both stemedb-api (storage) and stemedb-node (gateway/SWIM).
+#
+# In single-node mode (STEMEDB_CLUSTER_MODE unset or "false"), only stemedb-api runs.
+# In cluster mode (STEMEDB_CLUSTER_MODE=true), both binaries run side-by-side.
+
+set -e
+
+CLUSTER_MODE="${STEMEDB_CLUSTER_MODE:-false}"
+
+if [ "$CLUSTER_MODE" = "true" ] || [ "$CLUSTER_MODE" = "1" ]; then
+    echo "Starting StemeDB in cluster mode"
+
+    # Start stemedb-api in background (storage engine on :18180)
+    stemedb-api &
+    API_PID=$!
+
+    # Wait briefly for API to bind before starting the gateway
+    sleep 1
+
+    # Start stemedb-node in foreground (gateway :18181, gRPC :18182, SWIM :18183)
+    stemedb-node &
+    NODE_PID=$!
+
+    # Trap signals to shut down both processes
+    trap 'kill $API_PID $NODE_PID 2>/dev/null; wait' TERM INT
+
+    # Wait for either process to exit — if one dies, kill both
+    wait -n $API_PID $NODE_PID 2>/dev/null || true
+    EXIT_CODE=$?
+    kill $API_PID $NODE_PID 2>/dev/null || true
+    wait
+    exit $EXIT_CODE
+else
+    echo "Starting StemeDB in single-node mode"
+    exec stemedb-api "$@"
+fi