chore: update vllm and xinference version (#2278)

(cherry picked from commit 327db27)

Author: iziang

README.md

@@ -17,7 +17,7 @@ KubeBlocks add-ons.
 | hive | hive-3.1.2<br>hive-3.1.3<br>hive-4.0.1 | A Helm chart for Kubernetes |  |
 | influxdb | influxdb-2.7.11<br>influxdb-data-1.8.10<br>influxdb-meta-1.8.10 | InfluxDB is an open source time-series database. It is a core component of the TICK (Telegraf, InfluxDB(TM), Chronograf, Kapacitor) stack. |  |
 | kafka | kafka-broker-2.7.0<br>kafka-broker-2.8.2<br>kafka-broker-3.3.2<br>kafka-broker-3.7.1<br>kafka-broker-3.8.1<br>kafka-broker-3.9.0<br>kafka-combine-3.3.2<br>kafka-combine-3.7.1<br>kafka-combine-3.8.1<br>kafka-combine-3.9.0<br>kafka-controller-3.3.2<br>kafka-controller-3.7.1<br>kafka-controller-3.8.1<br>kafka-controller-3.9.0<br>kafka-exporter-1.6.0 | Apache Kafka is a distributed streaming platform designed to build real-time pipelines and can be used as a message broker or as a replacement for a log aggregation solution for big data applications. | caiq1nyu vipshop |
-| llm | ggml-0.1.0<br>llm-0.2.7 | Large language models. | ApeCloud |
+| llm | llm-ggml-20241115-latest<br>llm-vllm-0.11.2 | Large language models. | ApeCloud |
 | loki | loki-1.0.0 | Loki is a horizontally-scalable, highly-available, multi-tenant log aggregation system inspired by Prometheus. It is designed to be very cost effective and easy to operate. | Chen-speculation |
 | mariadb | mariadb-10.6.15 | MariaDB is a high performance open source relational database management system that is widely used for web and application servers | yinmin |
 | milvus | milvus-2.5.13<br>milvus-v2.3.2 | A cloud-native vector database, storage for next generation AI applications. | leon-inf |
@@ -46,6 +46,6 @@ KubeBlocks add-ons.
 | vanilla-postgresql | supabase-15.6.1-138<br>anilla-postgresql-12.15.0<br>anilla-postgresql-14.7.0<br>anilla-postgresql-15.7.0 | Vanilla-PostgreSQL is compatible with the native PostgreSQL kernel, enabling it to quickly provide HA solutions for various variants based on the native PostgreSQL kernel. | kizuna-lek |
 | victoria-metrics | ictoria-metrics-1.0.0 | VictoriaMetrics is a fast, cost-effective and scalable monitoring solution and time series database. | sophon-zt ButterBright |
 | weaviate | weaviate-1.19.6 | Weaviate is an open-source vector database. It allows you to store data objects and vector embeddings from your favorite ML-models, and scale seamlessly into billions of data objects. | iziang |
-| xinference | xinference-0.11.0<br>xinference-0.11.0-cpu | Xorbits Inference(Xinference) is a powerful and versatile library designed to serve language, speech recognition, and multimodal models. | ApeCloud |
+| xinference | xinference-0.11.0<br>xinference-0.11.0-cpu<br>xinference-1.13.0<br>xinference-1.13.0-cpu | Xorbits Inference(Xinference) is a powerful and versatile library designed to serve language, speech recognition, and multimodal models. | ApeCloud |
 | yashandb | yashandb-23.1.1-100 | YashanDB is a new database system completely independently designed and developed by SICS. Based on classical database theories, it incorporates original Bounded Evaluation theory, Approximation theory, Parallel Scalability theory and Cross-Modal Fusion Computation theory, supports multiple deployment methods such as stand-alone/primary-standby, shared cluster, and distributed ones, covers OLTP/HTAP/OLAP transactions and analyzes mixed load scenarios, and is fully compatible with privatization and cloud infrastructure, providing clients with one-stop enterprise-level converged data management solutions to meet the needs of key industries such as finance, government, telecommunications and energy for high performance, concurrency and security. | JesseAtSZ shanshanying |
 | zookeeper | zookeeper-3.4.14<br>zookeeper-3.6.4<br>zookeeper-3.7.2<br>zookeeper-3.8.4<br>zookeeper-3.9.2<br>zookeeper-3.9.4 | Apache ZooKeeper is a centralized service for maintaining configuration information, naming, providing distributed synchronization, and providing group services. | ApeCloud kissycn |

addons-cluster/xinference/templates/cluster.yaml

@@ -26,7 +26,7 @@ spec:
           cpu: {{ .Values.cpu | quote }}
           memory: {{ print .Values.memory "Gi" | quote }}
 {{ if .Values.cpuMode }}
-      serviceVersion: 0.11.0-cpu
+      serviceVersion: 1.13.0-cpu
 {{ else }}
-      serviceVersion: 0.11.0
+      serviceVersion: 1.13.0
 {{ end }}

addons/llm/templates/cmpd-ggml.yaml

@@ -5,11 +5,11 @@ metadata:
   labels:
     {{- include "llm.labels" . | nindent 4}}
   annotations:
+    apps.kubeblocks.io/skip-immutable-check: "true"
     {{- include "llm.annotations" . | nindent 4 }}
 spec:
   provider: kubeblocks
   serviceKind: ggml
-  serviceVersion: 0.1.0
   updateStrategy: BestEffortParallel
   scripts:
     - name: scripts
@@ -30,7 +30,6 @@ spec:
         emptyDir: {}
     containers:
       - name: ggml
-        image: {{ .Values.imageDev.registry | default ( .Values.image.registry | default "docker.io" ) }}/{{ .Values.imageDev.repository}}:{{ default .Chart.AppVersion .Values.imageDev.tag}}
         imagePullPolicy: {{default .Values.imageDev.pullPolicy "IfNotPresent"}}
         securityContext:
           runAsUser: 0

addons/llm/templates/cmpd-vllm.yaml

@@ -5,11 +5,11 @@ metadata:
   labels:
     {{- include "llm.labels" . | nindent 4}}
   annotations:
+    apps.kubeblocks.io/skip-immutable-check: "true"
     {{- include "llm.annotations" . | nindent 4 }}
 spec:
   provider: kubeblocks
   serviceKind: vllm
-  serviceVersion: 0.2.7
   updateStrategy: BestEffortParallel
   scripts:
     - name: scripts
@@ -42,7 +42,6 @@ spec:
         emptyDir: {}
     containers:
       - name: vllm
-        image: {{ .Values.image.registry | default "docker.io" }}/{{ .Values.image.repository}}:{{ default .Chart.AppVersion .Values.image.tag }}
         imagePullPolicy: {{ default "IfNotPresent" .Values.image.pullPolicy }}
         securityContext:
           runAsUser: 0

addons/llm/templates/cmpv-ggml.yaml

@@ -0,0 +1,20 @@
+apiVersion: apps.kubeblocks.io/v1
+kind: ComponentVersion
+metadata:
+  name: llm-ggml
+  labels:
+    {{- include "llm.labels" . | nindent 4 }}
+  annotations:
+    {{- include "llm.annotations" . | nindent 4 }}
+spec:
+  compatibilityRules:
+    - compDefs:
+        - {{ include "llm.cmpdNameGGML" . }}
+      releases:
+        - 20241115-latest
+  releases:
+    - name: 20241115-latest
+      serviceVersion: 20241115-latest
+      images:
+        ggml: {{ .Values.imageDev.registry | default ( .Values.image.registry | default "docker.io" ) }}/{{ .Values.imageDev.repository }}:20241115-latest
+

addons/llm/templates/cmpv-vllm.yaml

@@ -0,0 +1,20 @@
+apiVersion: apps.kubeblocks.io/v1
+kind: ComponentVersion
+metadata:
+  name: llm-vllm
+  labels:
+    {{- include "llm.labels" . | nindent 4 }}
+  annotations:
+    {{- include "llm.annotations" . | nindent 4 }}
+spec:
+  compatibilityRules:
+    - compDefs:
+        - {{ include "llm.cmpdNameVLLM" . }}
+      releases:
+        - 0.11.2
+  releases:
+    - name: 0.11.2
+      serviceVersion: 0.11.2
+      images:
+        vllm: {{ .Values.image.registry | default "docker.io" }}/{{ .Values.image.repository }}:v0.11.2
+

addons/llm/templates/scripts.yaml

@@ -19,16 +19,19 @@ data:
     if [[ "${domain,,}" == "baichuan-inc" ]]; then
       pip install transformers==4.33.1 tokenizers==0.13.3
     fi
+    apt-get update
+    apt install git-lfs
+    CLONE_MODEL_SCRIPT="git lfs install; git clone https://huggingface.co/${MODEL_NAME}"
     country=`curl https://ifconfig.io/country_code`
     if [ "$country" == "CN" ]; then
       CLONE_MODEL_SCRIPT="git lfs install; git clone https://www.modelscope.cn/${MODEL_NAME}.git"
-      export MODEL_NAME="../${MODEL_NAME##*/}"
+      export MODEL_NAME="./${MODEL_NAME##*/}"
     else
       curl --max-time 10 https://huggingface.co/${MODEL_NAME} > /dev/null 2>&1
       code=$?
       if [ "$code" -ne 0 ]; then 
         CLONE_MODEL_SCRIPT="git lfs install; git clone https://www.modelscope.cn/${MODEL_NAME}.git"
-        export MODEL_NAME="../${MODEL_NAME##*/}"
+        export MODEL_NAME="./${MODEL_NAME##*/}"
       fi
     fi 
     echo "model=${MODEL_NAME}"

addons/llm/values.yaml

@@ -3,7 +3,7 @@ image:
   pullPolicy: IfNotPresent
   registry: docker.io
   # Overrides the image tag whose default is the chart appVersion.
-  tag: v0.7.2
+  tag: v0.11.2
 imageDev:
   # refer: ghcr.io/abetlen/llama-cpp-python:latest
   repository: apecloud/llama-cpp-python

addons/xinference/Chart.yaml

@@ -21,7 +21,7 @@ version: 1.0.1
 # incremented each time you make changes to the application. Versions are not expected to
 # follow Semantic Versioning. They should reflect the version the application is using.
 # It is recommended to use it with quotes.
-appVersion: "0.11.0"
+appVersion: "1.13.0"
 
 maintainers:
   - name: ApeCloud

addons/xinference/templates/cmpd.yaml

@@ -10,7 +10,6 @@ spec:
   provider: kubeblocks
   description: Xinference is a powerful and versatile library designed to serve language, speech recognition, and multimodal models.
   serviceKind: xinference
-  serviceVersion: 0.11.0
   updateStrategy: BestEffortParallel
   scripts:
   - name: scripts