https://community.databricks.com/t5/data-engineering/how-to-create-an-image-from-a-cluster-so-that-the-compute/m-p/165072#M55384 <P>Hi&nbsp;<a href="https://community.databricks.com/t5/user/viewprofilepage/user-id/182577">@NW1000</a>&nbsp;,</P><P>For Classic Compute, the best approach is using Databricks Container Services (DCS). This lets you package your exact OS, system libraries, and Python environment into a custom Docker image. You just push that image to a private registry and point your cluster to it. Your version-controlled Dockerfile then acts as the exact audit trail your regulators need. You can find the detailed guide for this here: <A class="" href="https://learn.microsoft.com/en-us/azure/databricks/compute/custom-containers" target="_blank" rel="noopener">https://learn.microsoft.com/en-us/azure/databricks/compute/custom-containers</A></P><P>For Serverless Compute, custom Docker images aren't supported because Databricks fully manages the underlying infrastructure. Instead, you replicate the environment by locking in a specific Databricks Runtime (DBR) version and using a strict requirements.txt file to pin your exact package versions. Storing those requirements and your cluster definitions (like JSON or Terraform) in Git gives you the verifiable history required for compliance.</P><P>Hope this points you in the right direction!</P> Fri, 07 Aug 2026 06:16:41 GMT ShamenParis 2026-08-07T06:16:41Z https://community.databricks.com/t5/data-engineering/how-to-create-an-image-from-a-cluster-so-that-the-compute/m-p/165049#M55383 <P>I am working in a regulated industry. It is necessary to document and replicate the compute environment. I guess to do this for a classic cluster vs a serverless one is different. Anyone knows how to do it? Thank you for your assistance.</P> Thu, 06 Aug 2026 19:15:49 GMT https://community.databricks.com/t5/data-engineering/how-to-create-an-image-from-a-cluster-so-that-the-compute/m-p/165049#M55383 NW1000 2026-08-06T19:15:49Z https://community.databricks.com/t5/data-engineering/how-to-create-an-image-from-a-cluster-so-that-the-compute/m-p/165072#M55384 <P>Hi&nbsp;<a href="https://community.databricks.com/t5/user/viewprofilepage/user-id/182577">@NW1000</a>&nbsp;,</P><P>For Classic Compute, the best approach is using Databricks Container Services (DCS). This lets you package your exact OS, system libraries, and Python environment into a custom Docker image. You just push that image to a private registry and point your cluster to it. Your version-controlled Dockerfile then acts as the exact audit trail your regulators need. You can find the detailed guide for this here: <A class="" href="https://learn.microsoft.com/en-us/azure/databricks/compute/custom-containers" target="_blank" rel="noopener">https://learn.microsoft.com/en-us/azure/databricks/compute/custom-containers</A></P><P>For Serverless Compute, custom Docker images aren't supported because Databricks fully manages the underlying infrastructure. Instead, you replicate the environment by locking in a specific Databricks Runtime (DBR) version and using a strict requirements.txt file to pin your exact package versions. Storing those requirements and your cluster definitions (like JSON or Terraform) in Git gives you the verifiable history required for compliance.</P><P>Hope this points you in the right direction!</P> Fri, 07 Aug 2026 06:16:41 GMT https://community.databricks.com/t5/data-engineering/how-to-create-an-image-from-a-cluster-so-that-the-compute/m-p/165072#M55384 ShamenParis 2026-08-07T06:16:41Z