Title: Partial Redeployment of App — UCS Tools Documentation URL Source: https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html Published Time: Thu, 30 Oct 2025 07:23:04 GMT Markdown Content: Partial Redeployment of App[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#partial-redeployment-of-app "Link to this heading") --------------------------------------------------------------------------------------------------------------------------------------------------------------------- Housekeeping[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#housekeeping "Link to this heading") --------------------------------------------------------------------------------------------------------------------------------------- See the Getting Started documentation for installing and setting up: * microk8s * UCS Tools * Accessing NGC * Setting up repositories Introduction[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#introduction "Link to this heading") --------------------------------------------------------------------------------------------------------------------------------------- In this tutorial, you will learn how to partially redeploy only the updated resources in the application using the `helm upgrade` command. Some other points of interest in this tutorial are: * Adding parameters to a microservice and setting them from the application * Using the `configs` directory This tutorial continues from [Sharing Storage between Pod Replicas](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/04_Sharing_Storage_Replicas.html#tutorial-4). Updating the microservices and application[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#updating-the-microservices-and-application "Link to this heading") --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Create a file called `config.json` in the `configs` directory of `http-server` with the following contents: { "workingDir": "$params.workingDir" } Files added under the `configs` directory get automatically added as a configmap and mounted to the `/opt/configs` directory in the microservice containers. Here you have added a field `workingDir` set to a param `workingDir`. The `$params` placeholder will automatically be replaced with the actual value during application build. Update the microservice manifest of the HTTP server microservice `http-server` to the following: type: msapplication specVersion: 2.5.0 name: ucf.svc.http-server chartName: http-server description: http server version: 0.0.4 tags: [] keywords: [] publish: false ingress-endpoints: - name: http description: REST API endpoint protocol: TCP scheme: http mandatory: False data-flow: in-out params: workingDir: "/localvol" #> type: string #> description: Working directory for listing contents #> flags: mandatory fileToCreate: somefile.txt #> type: string #> description: Name of the file to create in the working directory #> flags: mandatory --- spec: - name: http-server-deployment type: ucf.k8s.app.deployment parameters: apptype: stateless - name: http-server-container type: ucf.k8s.container parameters: image: repository: nvcr.io/nvidia/pytorch tag: 22.04-py3 command: [sh, -c] args: [ "WORKING_DIR=$(cat /opt/configs/config.json | jq -r '.workingDir') && cd $WORKING_DIR && echo $PWD && touch $(FILE_TO_CREATE) && ls && python -m http.server 8080 "] env: - name: POD_NAME valueFrom: fieldRef: fieldPath: metadata.name - name: FILE_TO_CREATE value: $params.fileToCreate ports: - containerPort: 8080 name: http volumeMounts: - name: localvol mountPath: /localvol - name: svc type: ucf.k8s.service parameters: ports: - port: 8080 protocol: TCP name: http - name: localvol type: ucf.k8s.volume parameters: persistentVolumeClaim: claimName: local-path-pvc You have three changes to the manifest file: * Two parameters named `workingDir` and `fileToCreate` have been added along with some annotations (`type`, `description`, `flags`.) * A new environment variable `FILE_TO_CREATE` has been added with value set to `$params.fileToCreate` placeholder. * The container arguments have been updated to read `workingDir` from the `config.json` file, change into the directory, create a file with the name passed as microservice parameter `fileToCreate` and list the contents of the directory. Update the `app.yaml` file to the following: specVersion: 2.5.0 version: 0.0.4 doc: README.md name: server-client-app description: Server Client Application dependencies: - ucf.svc.curl-client:0.0.2 - ucf.svc.http-server:0.0.4 components: - name: client type: ucf.svc.curl-client - name: http-server type: ucf.svc.http-server parameters: workingDir: /usr/local fileToCreate: myfile.txt connections: client/http: http-server/http The parameters `workingDir` and `fileToCreate` have been set on the `http-server` and the version of the `http-server` and the app has been updated. Building Microservices and Applications and Deploying them[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#building-microservices-and-applications-and-deploying-them "Link to this heading") ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Follow the steps mentioned below but remember to update the version of the `http-server` and `curl-client` under the `dependencies` section in app.yaml: * [Connecting Microservices - Building Microservices and Applications](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/02_Connecting_Microservices.html#build-ms-app-cli-02) to build the services and app using CLI * [Connecting Microservices - Creating and Building Application using Studio](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/02_Connecting_Microservices.html#create-app-studio-02) to create and build the app visually using Studio * [Connecting Microservices - Deploying and Running Microservices and Application](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/02_Connecting_Microservices.html#deploy-app-02) - to deploy the app Inspecting and Debugging Microservices and Application[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#inspecting-and-debugging-microservices-and-application "Link to this heading") --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Verify that the application was deployed successfully: $ microk8s kubectl get all NAME READY STATUS RESTARTS AGE pod/curl-client-curl-client-deployment-569849964-shcc2 1/1 Running 0 17s pod/http-server-http-server-deployment-5cfdc4bfc9-cmfbq 1/1 Running 0 17s NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE service/kubernetes ClusterIP 10.152.183.1 443/TCP 49d service/http-server-http-server-deployment-svc ClusterIP 10.152.183.245 8080/TCP 21s NAME READY UP-TO-DATE AVAILABLE AGE deployment.apps/curl-client-curl-client-deployment 1/1 1 1 19s deployment.apps/http-server-http-server-deployment 1/1 1 1 19s NAME DESIRED CURRENT READY AGE replicaset.apps/curl-client-curl-client-deployment-569849964 1 1 1 18s replicaset.apps/http-server-http-server-deployment-5cfdc4bfc9 1 1 1 18s You can check the `http-server` container logs to see that the current working directory was `/usr/local` and a file named `myfile.txt` was created as specified in the application. $ microk8s kubectl logs --tail -1 -l "app=http-server-http-server-deployment" /usr/local bin cuda cuda-11 cuda-11.4 cuda-11.6 etc games include lib man mpi myfile.txt nvm sbin share src ucx You can also get logs to verify if the file was created in the mounted volume: $ microk8s kubectl logs --tail -1 -l "app=curl-client-curl-client-deployment" % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 936 100 936 0 0 457k 0 --:--:-- --:--:-- --:--:-- 457k Directory listing for /

Directory listing for /

As you can see, `myfile.txt` and some other files inside the `/usr/local` directory show up in the client, meaning that both the parameter values set by the application took effect. Redeploying the application[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#redeploying-the-application "Link to this heading") --------------------------------------------------------------------------------------------------------------------------------------------------------------------- Update the parameter value of `fileToCreate` in the application: specVersion: 2.5.0 version: 0.0.4 doc: README.md name: server-client-app description: Server Client Application dependencies: - ucf.svc.curl-client:0.0.2 - ucf.svc.http-server:0.0.4 components: - name: client type: ucf.svc.curl-client - name: http-server type: ucf.svc.http-server parameters: workingDir: /usr/local fileToCreate: someotherfile.txt connections: client/http: http-server/http Rebuild the app using: $ cd $HOME/ucf_tutorial/apps $ ucf_app_builder_cli app build server-client-app/ -f To redeploy only the updated resources of the application (partial redeployment), use the `helm upgrade` command: $ microk8s helm upgrade server-client server-client-app/server-client-app-0.0.4 Reinspect the kubernetes resources and pods: $ microk8s kubectl get all NAME READY STATUS RESTARTS AGE pod/curl-client-curl-client-deployment-569849964-shcc2 1/1 Running 0 3m40s pod/http-server-http-server-deployment-7df656c66f-mbl2z 1/1 Running 0 9s pod/http-server-http-server-deployment-5cfdc4bfc9-cmfbq 1/1 Terminating 0 3m40s NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE service/kubernetes ClusterIP 10.152.183.1 443/TCP 49d service/http-server-http-server-deployment-svc ClusterIP 10.152.183.245 8080/TCP 3m44s NAME READY UP-TO-DATE AVAILABLE AGE deployment.apps/curl-client-curl-client-deployment 1/1 1 1 3m42s deployment.apps/http-server-http-server-deployment 1/1 1 1 3m42s NAME DESIRED CURRENT READY AGE replicaset.apps/curl-client-curl-client-deployment-569849964 1 1 1 3m41s replicaset.apps/http-server-http-server-deployment-7df656c66f 1 1 1 9s replicaset.apps/http-server-http-server-deployment-5cfdc4bfc9 0 0 0 3m41s You can see that the change in the parameter value of the `http-server` microservice caused its existing pod to terminate and launch a new pod with the updated value, while other resources which were not changed (e.g., `curl-client` microservice pods) are not affected. You can check the container logs of the new pod of `http-server` to see that a file named `someotherfile.txt` was created as specified in the updated application. $ microk8s kubectl logs --tail -1 -l "app=http-server-http-server-deployment,microservice_version=0.0.4" /usr/local bin cuda cuda-11 cuda-11.4 cuda-11.6 etc games include lib man mpi nvm sbin share someotherfile.txt src ucx Check the latest logs of the `curl-client` pod: $ microk8s kubectl logs --tail -1 -l "app=curl-client-curl-client-deployment" ... % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 922 100 922 0 0 450k 0 --:--:-- --:--:-- --:--:-- 450k Directory listing for /

Directory listing for /

As you can see, the output lists the updated file name `someotherfile.txt`. Stopping and Cleaning up Microservices and Applications[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#stopping-and-cleaning-up-microservices-and-applications "Link to this heading") ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Finally, to stop and clean up the application run: $ microk8s helm3 uninstall server-client Some Notes[#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#some-notes "Link to this heading") ----------------------------------------------------------------------------------------------------------------------------------- * Partial redeployment works even when microservice version is updated in the application without changing any other parameters. Resources belonging to only those microservices whose version is changed should be redeployed, others are not affected. * It also works when some microservices are added / removed / replaced with other microservices. To try this: * Build and deploy the [Text Chatbot](https://docs.nvidia.com/ucf/2.10.0/text/UCS_Reference_Apps.html#text-chatbot). * Wait for the **Text Chatbot** application to run fully and interact with it. * Build the [Speech Chatbot](https://docs.nvidia.com/ucf/2.10.0/text/UCS_Reference_Apps.html#speech-chatbot). * Deploy the **Speech Chatbot** application using `helm upgrade` command. * Unchanged microservices **Riva Speech Skills** and **BotMaker Dialog Manager** should remain unaffected and pods continue to run without interruption. * Resources related to **BotMaker Text Web App** should be terminated while resources related to **BotMaker Speech Controller** and **BotMaker Speech Web App** be created. * Wait for the **Speech Chatbot** application to run fully and interact with it. * Changing parameters that cause only configmap contents to be updated do not cause the microservice pods to restart. So even though the `helm upgrade` command succeeded, the existing pods can continue to run and use stale config files. This is addressed in the next tutorial [Restarting Pods on Config Changes](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/06_Restarting_Pods_Config_Changes.html#tutorial-6). Links/Buttons: - [#](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/05_Partial_Redeployment_App.html#some-notes) - [Sharing Storage between Pod Replicas](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/04_Sharing_Storage_Replicas.html#tutorial-4) - [Connecting Microservices - Building Microservices and Applications](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/02_Connecting_Microservices.html#build-ms-app-cli-02) - [Connecting Microservices - Creating and Building Application using Studio](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/02_Connecting_Microservices.html#create-app-studio-02) - [Connecting Microservices - Deploying and Running Microservices and Application](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/02_Connecting_Microservices.html#deploy-app-02) - [Restarting Pods on Config Changes](https://docs.nvidia.com/ucf/2.10.0/text/tutorials/06_Restarting_Pods_Config_Changes.html#tutorial-6) - [Text Chatbot](https://docs.nvidia.com/ucf/2.10.0/text/UCS_Reference_Apps.html#text-chatbot) - [Speech Chatbot](https://docs.nvidia.com/ucf/2.10.0/text/UCS_Reference_Apps.html#speech-chatbot)