test: add an e2e case for a client-less CacheRuntime topology

.github/scripts/build-all-images.sh

@@ -11,6 +11,7 @@ get_image_tag() {
 
 build_images() {
     oss_emulator_img="${IMG_REPO}/oss-emulator:e2e"
+    mooncake_img="${IMG_REPO}/mooncake:e2e"
     images=(
         "${IMG_REPO}/dataset-controller:${IMAGE_TAG}"
         "${IMG_REPO}/application-controller:${IMAGE_TAG}"
@@ -23,10 +24,12 @@ build_images() {
         "${IMG_REPO}/fluid-webhook:${IMAGE_TAG}"
         "${IMG_REPO}/fluid-crd-upgrader:${IMAGE_TAG}"
         "${oss_emulator_img}"
+        "${mooncake_img}"
     )
 
     make docker-build-all
     docker build -t "${oss_emulator_img}" test/gha-e2e/jindo/oss-emulator
+    docker build -t "${mooncake_img}" test/gha-e2e/mooncake/image
 
     echo ">>> Cleaning docker build caches before loading images to free disk space..."
     docker builder prune -a -f

.github/scripts/gha-e2e.sh

@@ -105,8 +105,14 @@
     bash test/gha-e2e/curvine/test.sh
 }
 
+function mooncake_e2e() {
+    set -e
+    bash test/gha-e2e/mooncake/test.sh
+}
+
 check_control_plane_status
 alluxio_e2e
 jindo_e2e
 juicefs_e2e
 curvine_e2e
+mooncake_e2e

test/gha-e2e/mooncake/bad_mount_pod.yaml

@@ -0,0 +1,29 @@
+# Negative case: without a client component there is no FUSE mount point, so an
+# application pod that mounts this Dataset's PVC is bound to fail. Fluid still
+# creates the PVC/PV and reports them as Bound, which easily gives the
+# impression that they can be mounted, so the docs call this out in their FAQ
+# and this test pins the behaviour down.
+#
+# This pod is expected never to start; it is deleted as soon as the check is done.
+apiVersion: v1
+kind: Pod
+metadata:
+  name: mooncake-bad-mount
+  namespace: default
+spec:
+  # The mount keeps being retried anyway, no need to have kubelet restart the
+  # container over and over
+  restartPolicy: Never
+  automountServiceAccountToken: false
+  containers:
+    - name: app
+      image: busybox:1.36
+      imagePullPolicy: IfNotPresent
+      command: ["sleep", "infinity"]
+      volumeMounts:
+        - name: mc-vol
+          mountPath: /data
+  volumes:
+    - name: mc-vol
+      persistentVolumeClaim:
+        claimName: mooncake-demo

test/gha-e2e/mooncake/cacheruntime.yaml

@@ -0,0 +1,20 @@
+# One worker replica: the CI kind cluster is single-node, so extra replicas would
+# show no scheduling difference and only make the case slower. Replica scaling is
+# already covered by the curvine case and is not repeated here.
+apiVersion: data.fluid.io/v1alpha1
+kind: CacheRuntime
+metadata:
+  name: mooncake-demo
+  namespace: default
+spec:
+  runtimeClassName: mooncake-demo
+  master:
+    replicas: 1
+  worker:
+    replicas: 1
+    tieredStore:
+      levels:
+        - emptyDir:
+            quota: 1Gi
+          high: "0.8"
+          low: "0.5"

test/gha-e2e/mooncake/cacheruntimeclass.yaml

@@ -0,0 +1,92 @@
+# Mooncake is a cache system without POSIX mount semantics: applications talk to
+# the cache service directly through its own client, so this topology declares
+# only master and worker and omits the client component that would do the
+# mounting. That is exactly the scenario this case guards: the controller must
+# not panic when the client component is omitted.
+#
+# Note: do not declare resources on the containers here. When the
+# CacheRuntimeClass template sets resources but the CacheRuntime does not,
+# syncRuntimeSpec passes the zero value to updateResources, which overwrites the
+# template's resources with an empty value. That has two consequences:
+#   1. the resources declared in the template are silently dropped and the pod
+#      ends up with {};
+#   2. the patch triggers one ASTS rollout, during which the runtime is briefly
+#      not-ready, and once the Dataset flips to Failed it does not recover on its
+#      own (even after the component phase returns to Ready).
+apiVersion: data.fluid.io/v1alpha1
+kind: CacheRuntimeClass
+metadata:
+  name: mooncake-demo
+fileSystemType: mooncakefs
+topology:
+  master:
+    service:
+      headless: {}
+    executionEntries:
+      reportSummary:
+        command:
+          - bash
+          - -c
+          - /reportSummary.sh
+        timeout: 30
+    template:
+      spec:
+        restartPolicy: Always
+        containers:
+          - name: master
+            image: fluidcloudnative/mooncake:e2e
+            command:
+              - /custom-entrypoint.sh
+            args:
+              - master
+              - start
+            imagePullPolicy: IfNotPresent
+            readinessProbe:
+              tcpSocket:
+                port: 50051
+              initialDelaySeconds: 10
+              periodSeconds: 5
+              failureThreshold: 12
+            env:
+              - name: POD_NAME
+                valueFrom:
+                  fieldRef:
+                    fieldPath: metadata.name
+            ports:
+              - containerPort: 50051
+                name: rpc
+              - containerPort: 8080
+                name: metadata
+              - containerPort: 9003
+                name: metrics
+  worker:
+    service:
+      headless: {}
+    template:
+      spec:
+        restartPolicy: Always
+        containers:
+          - name: worker
+            image: fluidcloudnative/mooncake:e2e
+            command:
+              - /custom-entrypoint.sh
+            args:
+              - worker
+              - start
+            imagePullPolicy: IfNotPresent
+            readinessProbe:
+              tcpSocket:
+                port: 50052
+              initialDelaySeconds: 5
+              periodSeconds: 5
+              failureThreshold: 12
+            env:
+              - name: POD_NAME
+                valueFrom:
+                  fieldRef:
+                    fieldPath: metadata.name
+            ports:
+              - containerPort: 50052
+                name: data
+              - containerPort: 9300
+                name: http

test/gha-e2e/mooncake/dataset.yaml

@@ -0,0 +1,16 @@
+# Mooncake mounts no underlying storage (UFS); data is written directly into the
+# cache by the client. The mountPoint here is only a placeholder: the
+# CacheRuntimeClass declares no mountUfs execution entry, so Fluid never performs
+# an actual mount.
+apiVersion: data.fluid.io/v1alpha1
+kind: Dataset
+metadata:
+  name: mooncake-demo
+  namespace: default
+spec:
+  placement: Shared
+  accessModes:
+    - ReadWriteMany
+  mounts:
+    - name: mc
+      mountPoint: "mooncakefs:///"

test/gha-e2e/mooncake/image/Dockerfile

@@ -0,0 +1,41 @@
+# The Mooncake image used by the e2e case.
+#
+# It is built in-repo rather than pulled from an external registry, for the same
+# reason as test/gha-e2e/jindo/oss-emulator: e2e runs on every PR, an external
+# image going away turns the whole pipeline red, and the two scripts the image
+# provides for Fluid to invoke have to be reviewable.
+#
+# The only external dependency is Mooncake's official package on PyPI.
+# python:3.12.13-slim — slim is required: the full variant carries an extra
+# ~700MB build toolchain, growing the image from ~470MB to ~1.9GB, and e2e also
+# has to kind-load it onto the node and keep a second copy there.
+FROM python:3.12.13-slim@sha256:229a2c5bfa27522db7815ea81f9bed70af17ccb9de9fc7ad142b1877b5830d36
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+        libcurl4 \
+        libibverbs1 \
+        rdma-core \
+        librdmacm1 \
+        libnuma1 \
+        liburing2 \
+        curl \
+        jq \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN pip install --no-cache-dir \
+        mooncake-transfer-engine-non-cuda==0.3.12.post1 \
+        nvidia-cuda-runtime-cu12
+
+ENV LD_LIBRARY_PATH=/usr/local/lib/python3.12/site-packages/nvidia/cuda_runtime/lib:$LD_LIBRARY_PATH
+
+# 50051 master RPC / 50052 worker data / 8080 metadata service / 9003 master metrics / 9300 worker http
+EXPOSE 50051 50052 8080 9003 9300
+
+# The two scripts Fluid invokes:
+#   custom-entrypoint.sh  starts the right process based on the role (master/worker)
+#   reportSummary.sh      collects cache usage and emits it as JSON in the format Fluid expects
+COPY custom-entrypoint.sh /custom-entrypoint.sh
+COPY reportSummary.sh /reportSummary.sh
+RUN chmod +x /custom-entrypoint.sh /reportSummary.sh
+
+CMD ["python3"]

test/gha-e2e/mooncake/image/custom-entrypoint.sh

@@ -0,0 +1,71 @@
+#!/bin/sh
+# custom-entrypoint.sh
+# Usage: /custom-entrypoint.sh <master|worker|client> <start>
+
+set -e
+
+ROLE="$1"
+ACTION="$2"
+
+if [ "$ACTION" != "start" ]; then
+  echo "Error: unsupported action '$ACTION'"
+  exit 1
+fi
+
+case "$ROLE" in
+
+  master)
+    exec mooncake_master \
+      -v=1 \
+      --rpc_interface=eth0 \
+      --enable_http_metadata_server=true \
+      --http_metadata_server_host=0.0.0.0 \
+      --http_metadata_server_port=8080 \
+      --enable_metadata_cleanup_on_timeout=true \
+      --client_ttl=10
+    ;;
+
+  worker)
+    # Read the dynamic values from the RuntimeConfig JSON mounted by Fluid
+    if [ -z "$FLUID_RUNTIME_CONFIG_PATH" ] || [ ! -f "$FLUID_RUNTIME_CONFIG_PATH" ]; then
+      echo "Error: FLUID_RUNTIME_CONFIG_PATH not set or file not found"
+      exit 1
+    fi
+
+    CONFIG=$(cat "$FLUID_RUNTIME_CONFIG_PATH")
+
+    MASTER_SVC=$(echo "$CONFIG" | jq -r '.master.service.name')
+    WORKER_SVC=$(echo "$CONFIG" | jq -r '.worker.service.name')
+    QUOTA=$(echo "$CONFIG" | jq -r '.worker.tieredStoreLevels[0].quotas[0] // "1GiB"')
+
+    # Quota format conversion: Fluid hands over K8s-style "1Gi", Mooncake wants "1GB"
+    SEGMENT_SIZE=$(echo "$QUOTA" | sed 's/Gi$/GB/; s/Mi$/MB/')
+
+    NAMESPACE="${FLUID_DATASET_NAMESPACE:-default}"
+    MASTER_ADDR="${MASTER_SVC}.${NAMESPACE}.svc.cluster.local:50051"
+    METADATA_ADDR="http://${MASTER_SVC}.${NAMESPACE}.svc.cluster.local:8080/metadata"
+    WORKER_HOST="${POD_NAME}.${WORKER_SVC}.${NAMESPACE}.svc.cluster.local"
+
+    echo "Starting worker: master=$MASTER_ADDR, segment_size=$SEGMENT_SIZE, host=$WORKER_HOST"
+
+    exec mooncake_client \
+      --host="$WORKER_HOST" \
+      --port=50052 \
+      --global_segment_size="$SEGMENT_SIZE" \
+      --master_server_address="$MASTER_ADDR" \
+      --metadata_server="$METADATA_ADDR" \
+      --protocol=tcp \
+      --enable_http_server=true \
+      --http_port=9300
+    ;;
+
+  client)
+    echo "Error: client role not yet implemented for Mooncake"
+    exit 1
+    ;;
+
+  *)
+    echo "Error: unknown role '$ROLE'"
+    exit 1
+    ;;
+esac

test/gha-e2e/mooncake/image/reportSummary.sh

@@ -0,0 +1,47 @@
+#!/bin/bash
+set -euo pipefail
+
+RAW=$(curl -s http://localhost:9003/metrics/summary)
+
+if [ -z "$RAW" ]; then
+  echo "Error: empty response from metrics endpoint" >&2
+  exit 1
+fi
+
+# Extract the "Mem Storage: 0 B / 2.00 GB (0.0%)" part
+MEM_LINE=$(echo "$RAW" | grep -oE 'Mem Storage: [^|]+' || true)
+
+CACHED_RAW=$(echo "$MEM_LINE" | sed -E 's/Mem Storage: ([^/]+) \/.*/\1/' | xargs)
+CAPACITY_RAW=$(echo "$MEM_LINE" | sed -E 's/.*\/ ([^(]+) \(.*/\1/' | xargs)
+PERCENT_RAW=$(echo "$MEM_LINE" | grep -oE '\([0-9.]+%\)' | tr -d '()%')
+
+# Unit format conversion ("2.00 GB" -> "2.00GiB")
+normalize_unit() {
+  echo "$1" | sed -E 's/ ?GB$/GiB/; s/ ?MB$/MiB/; s/ ?B$/B/' | tr -d ' '
+}
+CACHED=$(normalize_unit "$CACHED_RAW")
+CACHE_CAPACITY=$(normalize_unit "$CAPACITY_RAW")
+
+# The number of keys is used as fileNum
+FILE_NUM=$(echo "$RAW" | grep -oE 'Keys: [0-9]+' | grep -oE '[0-9]+' || echo "0")
+
+# Approximate the hit ratio from the success rate of Get requests
+GET_STATS=$(echo "$RAW" | grep -oE 'Get=[0-9.]+/[0-9.]+' || echo "Get=0.00/0.00")
+GET_SUCCESS=$(echo "$GET_STATS" | cut -d= -f2 | cut -d/ -f1)
+GET_TOTAL=$(echo "$GET_STATS" | cut -d/ -f2)
+HIT_RATIO=$(awk -v s="$GET_SUCCESS" -v t="$GET_TOTAL" \
+  'BEGIN{ if (t>0) printf "%.0f", (s/t*100); else print "0" }')
+
+# Mooncake has no real underlying UFS total, so use the cache capacity instead
+UFS_TOTAL="$CACHE_CAPACITY"
+
+cat <<EOF
+{
+  "cached": "$CACHED",
+  "cachedPercentage": "$PERCENT_RAW",
+  "cacheCapacity": "$CACHE_CAPACITY",
+  "cacheHitRatio": "$HIT_RATIO",
+  "fileNum": "$FILE_NUM",
+  "ufsTotal": "$UFS_TOTAL"
+}
+EOF