From 1724aead4d4182b233bef599d024c2d0daaf5122 Mon Sep 17 00:00:00 2001 From: rachfop <rachfop@users.noreply.github.com> Date: Thu, 20 Mar 2025 21:56:16 +0000 Subject: [PATCH] Documentation updates from Promptless --- docs/pods/instant-clusters.md | 70 +++++++++++++++++++++++++++++++++++ 1 file changed, 70 insertions(+) create mode 100644 docs/pods/instant-clusters.md diff --git a/docs/pods/instant-clusters.md b/docs/pods/instant-clusters.md new file mode 100644 index 00000000..0659e8ea --- /dev/null +++ b/docs/pods/instant-clusters.md @@ -0,0 +1,70 @@ +--- +title: "Instant Clusters" +description: "Learn how to deploy and manage multiple GPU machines together using RunPod's Instant Cluster feature." +id: instant-clusters +sidebar_position: 7 +--- + +Instant Clusters is a feature that enables you to quickly deploy and manage multiple GPU machines as a unified cluster. This feature simplifies the process of setting up distributed computing environments for tasks that require multiple GPUs working together. + +## Overview + +With Instant Clusters, you can: +- Deploy multiple GPU machines simultaneously +- Manage them as a single unit +- Connect to individual pods within the cluster +- Scale your computing resources efficiently + +## Creating an Instant Cluster + +1. From the RunPod dashboard, select **Instant Cluster**. + +2. Configure your cluster: + - Specify the number of servers you want in your cluster + - Choose your GPU type + - Select a template for your pods + - Review and verify your settings + +3. Click **Deploy Cluster** to create your cluster. + +## Managing Your Cluster + +After deployment, you'll be redirected to the Clusters page. From here, you can: + +1. Select your cluster to view details +2. Access the Pods page showing all pods in your cluster +3. Connect to individual pods as needed +4. Monitor the status and performance of your cluster + +## Connecting to Cluster Pods + +To connect to pods within your cluster: + +1. Navigate to the Pods page for your cluster +2. Select the **Connect** button for the desired pod +3. Choose your connection method based on your template type: + - SSH terminal + - Web terminal + - Jupyter notebook + - Other template-specific connections + +## Best Practices + +- Choose consistent templates across your cluster for better compatibility +- Monitor resource usage across all pods +- Use [Global Networking](networking.md) to enable secure communication between your cluster pods +- Consider using network volumes for shared storage across your cluster + +## Limitations and Considerations + +- All pods in a cluster must use the same template +- Resource availability may affect cluster deployment +- Pricing is based on the cumulative resources of all pods in the cluster + +## Related Resources + +- [Global Networking](networking.md) +- [Storage Options](storage/_volume.md) +- [Pod Management](manage-pods.md) + +For additional support or questions about Instant Clusters, please [contact our support team](https://contact.runpod.io/hc/en-us/requests/new). \ No newline at end of file