[ENH] Compaction service setup (#1945)

## Description of changes

*Summarize the changes made by this PR.*
 - Improvements & Bug fixes
	 - ...
 - New functionality
	 - This PR adds the necessary artifacts to setup compaction service.

## Test plan
*How are these changes tested?*

- [ ] Tests pass locally with `pytest` for python, `yarn test` for js,
`cargo test` for rust

## Documentation Changes
*Are all docstrings for user-facing APIs updated if required? Do we need
to make documentation changes in the [docs
repository](https://github.com/chroma-core/docs)?*

Tiltfile

@@ -25,7 +25,16 @@ docker_build(
   'local:query-service',
   '.',
   only=["rust/", "idl/", "Cargo.toml", "Cargo.lock"],
-  dockerfile='./rust/worker/Dockerfile'
+  dockerfile='./rust/worker/Dockerfile',
+  target='query_service'
+)
+
+docker_build(
+  'local:compaction-service',
+  '.',
+  only=["rust/", "idl/", "Cargo.toml", "Cargo.lock"],
+  dockerfile='./rust/worker/Dockerfile',
+  target='compaction_service'
 )
 
 k8s_yaml(
@@ -82,6 +91,8 @@ k8s_resource(
     'compaction-service-memberlist-readerwriter:ClusterRole',
     'compaction-service-compaction-service-memberlist-binding:clusterrolebinding',
     'compaction-service-memberlist-readerwriter-binding:clusterrolebinding',
+    'compaction-service-serviceaccount:serviceaccount',
+    'compaction-service-serviceaccount-rolebinding:RoleBinding',
 
     'test-memberlist:MemberList',
     'test-memberlist-reader:ClusterRole',
@@ -99,7 +110,8 @@ k8s_resource('sysdb-migration', resource_deps=['postgres', 'namespace'], labels=
 k8s_resource('logservice', resource_deps=['sysdb-migration'], labels=["chroma"], port_forwards='50052:50051')
 k8s_resource('sysdb', resource_deps=['pulsar', 'sysdb-migration'], labels=["chroma"], port_forwards='50051:50051')
 k8s_resource('frontend-service', resource_deps=['pulsar', 'sysdb', 'logservice'],labels=["chroma"], port_forwards='8000:8000')
-k8s_resource('query-service', resource_deps=['sysdb', 'pulsar'], labels=["chroma"])
+k8s_resource('query-service', resource_deps=['sysdb'], labels=["chroma"])
+k8s_resource('compaction-service', resource_deps=['sysdb'], labels=["chroma"])
 
 # I have no idea why these need their own lines but the others don't.
 k8s_resource(objects=['query-service:service'], new_name='query-service-service', resource_deps=['query-service'], labels=["chroma"])

k8s/distributed-chroma/templates/compaction-service.yaml

@@ -0,0 +1,64 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: compaction-service
+  namespace: {{ .Values.namespace }}
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: compaction-service
+  template:
+    metadata:
+      labels:
+        app: compaction-service
+        member-type: compaction-service
+    spec:
+      serviceAccountName: compaction-service-serviceaccount
+      containers:
+        - name: compaction-service
+          image: "{{ .Values.compactionService.image.repository }}:{{ .Values.compactionService.image.tag }}"
+          imagePullPolicy: IfNotPresent
+          ports:
+            - containerPort: 50051
+          env:
+            - name: CHROMA_compaction-service__MY_IP
+              valueFrom:
+                fieldRef:
+                  fieldPath: status.podIP
+      topologySpreadConstraints:
+        - maxSkew: 1
+          topologyKey: "kubernetes.io/hostname"
+          whenUnsatisfiable: ScheduleAnyway
+          labelSelector:
+            matchLabels:
+              member-type: compaction-service
+      volumes:
+        - name: chroma
+          emptyDir: {}
+
+---
+
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: compaction-service-serviceaccount
+  namespace: {{ .Values.namespace }}
+
+---
+
+apiVersion: rbac.authorization.k8s.io/v1
+kind: RoleBinding
+metadata:
+  name: compaction-service-serviceaccount-rolebinding
+  namespace: {{ .Values.namespace }}
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: Role
+  name: pod-watcher
+subjects:
+- kind: ServiceAccount
+  name: compaction-service-serviceaccount
+  namespace: {{ .Values.namespace }}
+
+---

k8s/distributed-chroma/values.yaml

@@ -55,6 +55,11 @@ queryService:
     repository: 'local'
     tag: 'query-service'
 
+compactionService:
+  image:
+    repository: 'local'
+    tag: 'compaction-service'
+
 sysdbMigration:
   image:
     repository: 'local'

rust/worker/Cargo.toml

@@ -7,6 +7,10 @@ edition = "2021"
 name = "query_service"
 path = "src/bin/query_service.rs"
 
+[[bin]]
+name = "compaction_service"
+path = "src/bin/compaction_service.rs"
+
 [dependencies]
 tonic = "0.10"
 prost = "0.12"

rust/worker/Dockerfile

@@ -21,20 +21,41 @@ RUN curl -OL https://github.com/protocolbuffers/protobuf/releases/download/v25.1
     && unzip -o $PROTOC_ZIP -d /usr/local 'include/*' \
     && rm -f $PROTOC_ZIP
 
+FROM builder as query_service_builder
 # We need to replace the query node's real main() with a dummy at the expected location.
 RUN mkdir -p rust/worker/src/bin/ && echo "fn main() {}" > rust/worker/src/bin/query_service.rs
-RUN if [ "$RELEASE_MODE" = "1" ]; then cargo build --release; else cargo build; fi
+RUN if [ "$RELEASE_MODE" = "1" ]; then cargo build --bin query_service --release; else cargo build --bin query_service; fi
 RUN rm -rf rust/
 
 COPY rust/ rust/
 RUN touch rust/worker/src/bin/query_service.rs
-RUN if [ "$RELEASE_MODE" = "1" ]; then cargo build --release; else cargo build; fi
+RUN if [ "$RELEASE_MODE" = "1" ]; then cargo build --bin query_service --release; else cargo build --bin query_service; fi
 RUN if [ "$RELEASE_MODE" = "1" ]; then mv target/release/query_service .; else mv target/debug/query_service .; fi
 
-FROM debian:bookworm-slim
+FROM debian:bookworm-slim as query_service
 
-COPY --from=builder /chroma/query_service .
-COPY --from=builder /chroma/rust/worker/chroma_config.yaml .
+COPY --from=query_service_builder /chroma/query_service .
+COPY --from=query_service_builder /chroma/rust/worker/chroma_config.yaml .
 RUN apt-get update && apt-get install -y libssl-dev
 
-ENTRYPOINT [ "./query_service" ]
+ENTRYPOINT [ "./query_service" ]
+
+FROM builder as compaction_service_builder
+
+# We need to replace the compaction node's real main() with a dummy at the expected location.
+RUN mkdir -p rust/worker/src/bin/ && echo "fn main() {}" > rust/worker/src/bin/compaction_service.rs
+RUN if [ "$RELEASE_MODE" = "1" ]; then cargo build --bin compaction_service --release; else cargo build --bin compaction_service; fi
+RUN rm -rf rust/
+
+COPY rust/ rust/
+RUN touch rust/worker/src/bin/compaction_service.rs
+RUN if [ "$RELEASE_MODE" = "1" ]; then cargo build --bin compaction_service --release; else cargo build --bin compaction_service; fi
+RUN if [ "$RELEASE_MODE" = "1" ]; then mv target/release/compaction_service .; else mv target/debug/compaction_service .; fi
+
+FROM debian:bookworm-slim as compaction_service
+
+COPY --from=compaction_service_builder /chroma/compaction_service .
+COPY --from=compaction_service_builder /chroma/rust/worker/chroma_config.yaml .
+RUN apt-get update && apt-get install -y libssl-dev
+
+ENTRYPOINT [ "./compaction_service" ]

rust/worker/README.md

@@ -1,12 +1,15 @@
 # Readme
+
 This folder houses the Rust code for the query and compactor nodes. It is a standard rust crate managed using cargo.
 
 ### Testing
+
 `cargo test`
 
 ### Building
-`cargo build`
 
+`cargo build`
 
 ### Rust version
-Use rust 1.74.0 or greater. 
+
+Use rust 1.74.0 or greater.