llm

redhat/ocp4/4.13/4.13.gpu.llm.md

@@ -867,8 +867,8 @@ spec:
     preferredDuringSchedulingIgnoredDuringExecution: []
   persistentSpec:
     enabled: true
-    storageClassName: local-volume
-    volumeSize: 4500Gi
+    storageClassName: hostpath-csi
+    volumeSize: 20Gi
   createServiceAccount: true
   username: [email protected]
   clusterCredentialSecretRole: ''
@@ -907,7 +907,7 @@ oc get route -n llm-demo
 
 oc get secret rec -n llm-demo -o jsonpath="{.data}" | jq 'map_values(@base64d)'
 # {
-#   "password": "PP1BuaXT",
+#   "password": "gv7AZB8P",
 #   "username": "[email protected]"
 # }
 
@@ -955,6 +955,14 @@ oc create --save-config -n llm-demo -f ${BASE_DIR}/data/install/redis-db.yaml
 
 # oc delete -n llm-demo -f ${BASE_DIR}/data/install/redis-db.yaml
 
+oc get RedisEnterpriseDatabase/my-doc -n llm-demo -o json | jq .status.internalEndpoints
+# [
+#   {
+#     "host": "redis-13858.rec.llm-demo.svc.cluster.local",
+#     "port": 13858
+#   }
+# ]
+
 oc get secret -n llm-demo redb-my-doc -o jsonpath="{.data}" | jq 'map_values(@base64d)'
 # {
 #   "password": "ABmKLkR0",
@@ -963,13 +971,14 @@ oc get secret -n llm-demo redb-my-doc -o jsonpath="{.data}" | jq 'map_values(@ba
 #   "service_names": "my-doc, my-doc-headless"
 # }
 
+# VAR_HOST=`oc get RedisEnterpriseDatabase/my-doc -n llm-demo -o json | jq .status.internalEndpoints[0].host -r`
 VAR_PWD=`oc get secret -n llm-demo redb-my-doc -o jsonpath="{.data}" | jq 'map_values(@base64d)' | jq -r .password`
 VAR_PORT=`oc get secret -n llm-demo redb-my-doc -o jsonpath="{.data}" | jq 'map_values(@base64d)' | jq -r .port`
 
 VAR_REDIS="redis://default:${VAR_PWD}@my-doc-headless.llm-demo.svc.cluster.local:${VAR_PORT}"
 
 echo $VAR_REDIS
-# redis://default:ABmKLkR0@my-doc-headless.llm-demo.svc.cluster.local:13858
+# redis://default:Ygg8FgJc@my-doc-headless.llm-demo.svc.cluster.local:16432
 
 
 ```
@@ -1002,28 +1011,71 @@ gradio-hftgi-rag-redis
 
 - https://github.com/rh-aiservices-bu/llm-on-openshift/blob/main/examples/ui/gradio/gradio-hftgi-rag-redis/README.md
 
-```bash
-# https://github.com/wangzheng422/llm-on-openshift/tree/wzh/examples/ui/gradio/gradio-hftgi-rag-redis/deployment/wzh
 
-oc create -n llm-demo -f ${BASE_DIR}/wzh/
+### create index
 
-# oc delete -n llm-demo -f ${BASE_DIR}/wzh/
+Langchain-Redis-Ingest.ipynb
 
+![](imgs/2023-11-08-23-43-00.png)
 
-```
+![](imgs/2023-11-08-23-44-59.png)
 
+open this notebook
+- https://github.com/rh-aiservices-bu/llm-on-openshift/blob/main/examples/notebooks/langchain/Langchain-Redis-Ingest.ipynb
 
-### Langchain-Redis-Ingest.ipynb
+run it step by step, or all
 
-![](imgs/2023-11-08-23-43-00.png)
+![](imgs/2023-11-15-00-05-56.png)
 
-![](imgs/2023-11-08-23-44-59.png)
+### deploy 
 
 ```bash
 
+# we need to build image for module cache
+# on vultr
+systemctl enable --now docker.service
+
+mkdir -p /data/llm-build
+cd /data/llm-build
+
+cat << EOF > /data/llm-build/build.dockerfile
+FROM quay.io/rh-aiservices-bu/gradio-hftgi-rag-redis:latest
+
+RUN python app.py || true
+
+EOF
+
+docker build -t quay.io/wangzheng422/qimgs:2023-11-14-gradio-hftgi-rag-redis-01 -f build.dockerfile .
+
+docker push quay.io/wangzheng422/qimgs:2023-11-14-gradio-hftgi-rag-redis-01
+
+
+# docker run -it -d  \
+#   --name llm-build \
+#   --gpus all \
+#   --shm-size 1g \
+#   quay.io/rh-aiservices-bu/gradio-hftgi-rag-redis:latest
+
+# docker export llm-build | docker import - quay.io/wangzheng422/qimgs:2023-11-14-gradio-hftgi-rag-redis
+
+# docker push quay.io/wangzheng422/qimgs:2023-11-14-gradio-hftgi-rag-redis
+
+# go back to helper node
+# https://github.com/wangzheng422/llm-on-openshift/tree/wzh/examples/ui/gradio/gradio-hftgi-rag-redis/deployment/wzh
+
+oc create -n llm-demo -f ${BASE_DIR}/wzh/
+
+# oc delete -n llm-demo -f ${BASE_DIR}/wzh/
 
 ```
 
+then, you can see the webpage: 
+
+![](imgs/2023-11-15-00-07-26.png)
+
+![](imgs/2023-11-15-00-07-42.png)
+
+
 # end
 
 ![](imgs/2023-11-09-14-39-44.png)