DOCSEARCHbyAlgolia
Submit Docs

Porter Documentation

website·Indexed 34 days ago

https://docs.porter.run

Documentation for deploying and managing applications on Porter. Covers deployment paths from GitHub or Docker registries, configuration as code using porter.yaml, autoscaling (including Temporal autoscaling), zero-downtime deployments, and add-ons such as managed Postgres, Redis, persistent storage (Amazon EFS), and custom Helm charts.

Tokens
73.7K
Snippets
288
Records
570
Agent score
64%

What's inside Porter

  1. Overview of Porter Sandboxes

    Porter Sandboxes are isolated container workloads designed for running untrusted code or workflows for agents. They provide a secure, fresh runtime environment suitable for code interpretation, agent tools, batch fan-out, or on-demand processing.

    Key features include:

    • Hosted within a user-owned cloud account (currently available on AWS).
    • Persistent volume support and S3 connection for object storage.
    • Snapshot and restore capabilities.
    • Sub-second spin-up times via a configurable warm pool.
    • Egress controls.

    Note: For long-running services requiring autoscaling and standard application lifecycle management, use Porter Applications instead of Sandboxes.

  2. Overview of Porter MCP Server

    Porter provides a remote Model Context Protocol (MCP) server at https://mcp.porter.run. This allows AI agents to perform tasks typically done in the Porter dashboard or CLI, such as reading logs, managing services, and deploying applications to AWS, GCP, or Azure. The server is hosted by Porter, meaning no local installation is required beyond configuring the MCP client and signing in via a browser.
    https://mcp.porter.run

  3. Overview of Porter Datastores

    Porter allows you to provision managed Postgres and Redis databases on AWS. It automates the networking setup by peering the datastore VPC to the cluster VPC, configuring subnets, routing tables, and security groups to ensure traffic flows exclusively through private subnets. Currently, datastores are supported on AWS only.

  4. Overview of Temporal Autoscaling in Porter

    Porter provides automatic scaling for worker services based on Temporal task queue depth. This ensures that worker replicas increase when there is a backlog of tasks and decrease when queues are empty, optimizing resource usage and processing latency.

  5. Overview of Porter service connection types

    Porter allows services to connect to external cloud resources via the connections array in porter.yaml. Supported connection types include:

    • awsRole: Attach an AWS IAM role for API access.
    • azureManagedIdentity: Bind an Azure User Assigned Managed Identity (UAMI) for API access.
    • gcpServiceAccount: Bind a GCP IAM service account for API access.
    • cloudSql: Connect to Google Cloud SQL instances via the Cloud SQL Auth Proxy.
    • disk: Attach persistent storage.

  6. Overview of Doppler integration in Porter

    Porter integrates with Doppler to sync secrets and environment variables as read-only managed environment groups. Porter installs the External Secrets Operator on the cluster and uses a Doppler service token to pull secrets directly from the Doppler API. Secret values are not persisted in Porter's infrastructure.

  7. Overview of Porter Sandbox Volumes

    Volumes provide persistent storage that lives independently of a sandbox's lifecycle. While a sandbox's own filesystem is discarded upon exit, volumes preserve contents across different sandboxes. They are ideal for input data, checkpoints, caches, and results. Volumes can be managed via the Python SDK, TypeScript SDK, CLI, or the Porter Dashboard.

  8. Introduction to Porter PaaS

    Porter is a platform as a service (PaaS) that runs directly within your own AWS, GCP, or Azure account. It automates the provisioning and management of underlying infrastructure, handling the process from building the application to autoscaling it, allowing developers to deploy applications with minimal manual configuration.

  9. Overview of Porter Sandbox Networking

    Porter allows sandboxes that open a port to be served over HTTPS on a custom domain. Networking is configured at the cluster level, and individual sandboxes are assigned URLs based on their name or a user-specified hostname. Porter supports two networking types:

    • Public: Uses an internet-facing load balancer; accessible by anyone on the internet.
    • Private: Uses an internal load balancer; accessible only by clients inside the VPC.

    Each networking type requires its own unique domain (e.g., sandbox.acme.com for public and internal.sandbox.acme.com for private). Porter manages wildcard HTTPS certificates for these domains via Let's Encrypt.

  10. Overview of Porter PaaS

    Porter is a Platform as a Service (PaaS) that provisions and manages underlying infrastructure within a user's own AWS, GCP, or Azure account. It allows developers to deploy applications by pointing to a repository; Porter then handles the build process and autoscaling while ensuring the user retains full control of their cloud environment.

  11. Overview of the Porter DevOps Agent

    The DevOps Agent is an AI-powered debugging assistant built into the Porter dashboard. It provides read-only access to your cluster in real time to investigate infrastructure issues without the ability to modify services. It analyzes data from the Porter API (deployment history, service config), live cluster state (logs, resource usage, events), and Porter documentation to provide actionable diagnoses and fix recommendations.

  12. Overview of available Porter add-ons

    Porter add-ons extend a cluster with additional infrastructure components. Available add-ons include:

    • Databases: Postgres (object-relational) and Redis (in-memory key-value).
    • Monitoring: Datadog, New Relic, Grafana, Langfuse (LLM engineering), and Helicone AI Gateway (AI gateway for LLM requests).
    • Logging: Mezmo.
    • Analytics: Metabase (BI tool), Quivr (generative AI second brain), and n8n (workflow engine).
    • Storage: Persistent Disk for application data storage.
    • Custom: Support for installing any Helm chart from a public repository.