Skip to main content

Major new features

Autoscaling sandbox fleets with sticky disks

Pipeline sandbox runners can make builds up to 4× faster and Terraform runs up to 10× faster, depending on your workload. Behind those faster runs is a fully managed fleet: Ravion orchestrates isolated Firecracker microVMs across EC2 hosts in your AWS account, packing multiple sandboxes onto each host and scaling the fleet up and down with demand.
  • Fast starts and sticky disks. Steps start in roughly a second on warm hosts. Cached disks keep dependencies, build output and Docker layers between runs, and saved snapshots can follow your work to another host.
  • Automatic host selection and Spot fallback. Ravion chooses host types unless you specify them. New pools use Spot for eligible work with interruptible: true when more capacity is needed, trying compatible instance types and Availability Zones before falling back to On-Demand.
  • Disposable hosts without disposable caches. Hosts are ephemeral by default and deleted when idle; the fleet can scale to zero without losing saved sandbox disks. Keep a warm reserve for faster starts, or use persistent hosts that stop with their disks retained and wake for later work.
You configure what each step needs; Ravion manages the fleet underneath it. Read how sandbox fleets work for the architecture and cost trade-offs.To get started, update your AWS account to the latest permissions with Update AWS Permissions in AWS account settings, then follow the sandboxes getting started guide.

AWS quota management

When Ravion sees an AWS quota issue, we display an in-dashboard alert and let you request an increase in one click. We have limited coverage currently, but more will be coming.

Platform

Features

  • Execution environment usage: View estimated cost, Spot and On-demand usage, capacity allocation, startup timing, and pipeline attribution.
  • Pipeline stage descriptions: Add optional purpose text to pipeline steps and groups and view it when hovering their names. Learn more.

Improvements

  • Environment canvas loading now uses saved card positions and keeps metric cards at a consistent height while data loads.
  • Execution environment details now open on Usage with a clearer Cost and Usage summary and a pool overview on the Sandbox hosts tab.

Fixes

  • Personal GitLab repositories now load without a repository-stream error.
  • Cancel controls now offer only actions that the current pipeline run supports.
  • Successful Terraform applies no longer become failed runs because of a later refresh error.
  • Project configuration applies now skip deferred values and report the correct invalid value when a new module and its references are added together.
  • Sandbox launches now fail promptly with a useful error when AWS access, capacity, or host images prevent a host from starting.
  • Failed AWS Spot requests now move to another zone or instance type instead of repeating the same request.

AWS account IAM policy 1.0.50 → 1.0.54

  • Spot price history lookups: Allow spot price history lookups for usage capture.
  • Sandbox EC2 Fleets: Allow sandbox EC2 Fleets.
  • AWS Service Quotas: Allow reading and requesting increases for all AWS service quotas.
Update your AWS account connection in organization settings.

Modules

AWS Application Load Balancer 1.2.0 → 1.2.2

  • Add a web_acl_arn output for the associated WAF web ACL.

EC2 Service 2.1.0 → 2.2.0

  • Keep sandbox build disks between runs so Docker layers and dependencies stay cached.
  • Run sandbox builds on Spot hosts and restart them on another host if AWS reclaims the host; turn off Interruptible build for commands that change state outside the build.
  • New builds now use sandbox execution with 14 vCPU and 28 GiB by default.

ECS Cluster 2.1.0 → 2.1.2

  • Associate the public load balancer with the WAF web ACL when a web ACL ARN is set.

ECS Network Service 2.1.0 → 2.3.1

  • Keep sandbox build disks between runs so Docker layers and dependencies stay cached.
  • Run sandbox builds on Spot hosts and restart them on another host if AWS reclaims the host; turn off Interruptible build for commands that change state outside the build.
  • Let TCP and TLS listeners pass the original client IP to tasks through Advanced Terraform variables.
  • New builds now use sandbox execution with 14 vCPU and 28 GiB by default.

ECS Web Service 2.1.0 → 2.2.1

  • Keep sandbox build disks between runs so Docker layers and dependencies stay cached.
  • Run sandbox builds on Spot hosts and restart them on another host if AWS reclaims the host; turn off Interruptible build for commands that change state outside the build.
  • New builds now use sandbox execution with 14 vCPU and 28 GiB by default.

ECS Worker 2.1.0 → 2.2.1

  • Keep sandbox build disks between runs so Docker layers and dependencies stay cached.
  • Run sandbox builds on Spot hosts and restart them on another host if AWS reclaims the host; turn off Interruptible build for commands that change state outside the build.
  • New builds now use sandbox execution with 14 vCPU and 28 GiB by default.

Lambda Function 1.4.0 → 2.1.0

  • BREAKING: update callers, DNS targets, and IAM policies for the new function URL after applying the live-alias deployment change.
  • Keep sandbox build disks between runs so Docker layers and dependencies stay cached.
  • Run sandbox builds on Spot hosts and restart them on another host if AWS reclaims the host; turn off Interruptible build for commands that change state outside the build.
  • New builds now use sandbox execution with 14 vCPU and 28 GiB by default.

S3 Bucket 1.1.0 → 2.0.1

  • Add Requester Pays as an advanced Terraform variable so bucket readers pay for requests and data transfer.
  • BREAKING: apply the region change only after emptying a bucket that an earlier version created in another region, because it will be replaced there.

Static Hosting 1.3.0 → 1.4.0

  • Keep sandbox build disks between runs so Docker layers and dependencies stay cached.
  • Run sandbox builds on Spot hosts and restart them on another host if AWS reclaims the host; turn off Interruptible build for commands that change state outside the build.
  • New builds now use sandbox execution with 14 vCPU and 28 GiB by default.

CLI 1.10.1 → 1.14.12

Breaking changes

  • Sandbox interruptibility: Move interruptible into the sandbox infrastructure block before upgrading configurations.

Features

  • Multiple roles: Assign several roles to members and service accounts with ravion membership and ravion service-account commands.
  • API key permissions: Limit an API key to selected permissions and inspect the available permissions for its owner with ravion api-key.
  • Automatic sandbox host selection: Let sandbox pools choose suitable instance types, create a default pool on first use, and keep a reserve of idle ephemeral hosts.
  • Execution environment usage: View Spot and On-demand compute, busy-host allocation, and startup timing in usage data.
  • Sandbox capacity options: Configure persistent or ephemeral host storage and allow interruptible steps to use Spot capacity.

Improvements

  • Pipeline configuration validation: Send null values in pipeline version create requests without rejecting unknown or removed fields.

Fixes

  • Project configuration diffs: Preserve written number formatting in plan diffs and keep dry runs consistent with pull request comments.

Platform

Breaking changes

  • Removed top-level pipeline rollback: Remove the unsupported rollback key from pipeline configurations before applying them. Pipeline rollback is a planned future feature.

Features

  • Templated infrastructure types: Use templates that resolve to different infrastructure types when a pipeline step runs.

Improvements

  • Billing plan cards now make the whole card the change-plan action when a payment method is missing.
  • Trial notices now reserve urgent styling for the final three days.
  • Module reference fields now offer reference suggestions when values are entered manually.
  • Key-value fields now edit existing pairs in place, use the available width, and accept pasted JSON pairs.
  • Project breadcrumbs now link to the project from module pages without sending users back to the current page.
  • Default Terraform pipelines now pass execution environment overrides to their plan and apply steps.
  • Module status now reports STACK_DESTROYED after a successful stack destroy and excludes Terraform data sources from resource counts.

Fixes

  • Reapplying configuration with omitted immutable inputs now keeps their stored values.
  • Environment Logs now identifies AWS access denials and guides customers to update account permissions.
  • Applying project configuration now resolves system pipeline defaults for environments that do not yet exist.
  • Listing a stack’s current Terraform state now works when large state values would otherwise exhaust database sort memory.
  • ECS deployments now preserve the service’s existing tag propagation setting.
  • Transient binary download failures now retry with backoff before reporting an error.
  • EC2 deployments now use the repository’s default branch when no source branch is supplied.
  • Missing applied-module outputs now produce a specific validation error at the affected field.

AWS account IAM policy 1.0.48 → 1.0.50

  • AMI deployment permissions: The first AMI build in an account can create EC2 Image Builder’s service-linked role.
Update your AWS account connection in Settings / AWS accounts.

Modules

ACM Certificate 1.0.1 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

AMI 0.1.0 → 0.3.0

  • Configure whether AMIs remain private or can be published publicly across the build and additional regions.
  • Reference the module instance ID in templates and other module inputs.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

Aurora Database 1.3.0 → 1.4.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

AWS Application Load Balancer 1.1.0 → 1.2.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

AWS Compliance 0.1.0 → 0.2.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

AWS IAM Policy 1.0.1 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

AWS IAM Role 1.0.1 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

AWS KMS Key 0.1.0 → 0.2.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

AWS Secret 1.0.0 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

CloudFront CDN 1.3.2 → 1.4.0

  • Private origins now recover from in-use errors when moving to another load balancer or changing settings.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

EC2 Service 1.6.1 → 2.1.0

  • Run builds in sandbox microVMs on the execution environment’s host pool.
  • Configure runtime secret keys explicitly with name and value_from.
  • Avoid build environment variables being mistaken for runtime variables.
  • Avoid group name collisions for advanced EC2 services without changing existing groups on upgrade.
  • Reference supported Secrets Manager and Parameter Store keys for build environment variables.
  • BREAKING: replace the former scaling inputs with min_instances and max_instances before upgrading.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

ECS Cluster 1.1.0 → 2.1.0

  • BREAKING: rename EC2 capacity inputs to ec2_min_instances and ec2_max_instances before upgrading.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

ECS Network Service 1.6.0 → 2.1.0

  • Run builds in sandbox microVMs on the execution environment’s host pool.
  • Configure runtime and FireLens secret keys explicitly with name and value_from.
  • Avoid build environment variables being mistaken for runtime variables.
  • Reference supported Secrets Manager and Parameter Store keys for build environment variables.
  • Configure the application container’s stop timeout instead of using a fixed grace period.
  • BREAKING: rename scaling inputs to min_tasks, max_tasks, and desired_tasks, and re-enter saved CPU and memory values in task units before upgrading.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

ECS Web Service 1.6.0 → 2.1.0

  • Run builds in sandbox microVMs on the execution environment’s host pool.
  • Configure runtime and FireLens secret keys explicitly with name and value_from.
  • Avoid build environment variables being mistaken for runtime variables.
  • Reference supported Secrets Manager and Parameter Store keys for build environment variables.
  • Configure request-based autoscaling and the application container’s stop timeout.
  • BREAKING: rename scaling inputs to min_tasks, max_tasks, and desired_tasks, and re-enter saved CPU and memory values in task units before upgrading.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

ECS Worker 1.6.0 → 2.1.0

  • Run builds in sandbox microVMs on the execution environment’s host pool.
  • Configure runtime and FireLens secret keys explicitly with name and value_from.
  • Avoid build environment variables being mistaken for runtime variables.
  • Reference supported Secrets Manager and Parameter Store keys for build environment variables.
  • Configure the application container’s stop timeout instead of using a fixed grace period.
  • BREAKING: rename scaling inputs to min_tasks, max_tasks, and desired_tasks, and re-enter saved CPU and memory values in task units before upgrading.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

EFS File System 1.0.1 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

ElastiCache 1.0.1 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

Lambda Function 1.1.2 → 1.4.0

  • Run builds in sandbox microVMs on the execution environment’s host pool.
  • Choose Node.js 24.x for zip package functions.
  • Avoid build environment variables being mistaken for runtime variables.
  • Reference supported Secrets Manager and Parameter Store keys for build environment variables.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

RDS Database 1.3.0 → 1.4.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

RDS Proxy 0.1.0 → 0.2.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

Route 53 DNS 1.0.3 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

S3 Bucket 1.0.1 → 1.1.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

Static Hosting 1.1.3 → 1.3.0

  • Run builds in sandbox microVMs on the execution environment’s host pool.
  • Avoid build environment variables being mistaken for runtime variables.
  • Reference supported Secrets Manager and Parameter Store keys for build environment variables.
  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

Terraform Stack 1.3.1 → 1.4.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

VPC Network 1.1.0 → 1.2.0

  • Add Terraform plan and apply IAM policy inputs to scope stack runner permissions.

CLI 1.4.0 → 1.10.1

Features

  • Input reference status: See whether each module input reference is up to date, stale, or unknown, and review module output references.
  • Billing details: View past-due status and invoice collection method with ravion billing get.

Fixes

  • CloudWatch Logs queries: AWS access denials now return a typed permission error instead of a generic server error.
  • Code source creation: ravion code-source create now accepts a GitHub repository with the required type, installation, and repository flags.
  • Type-specific command flags: Commands now validate flags for the selected request type before calling the API.
  • Module status: Module status now reports STACK_DESTROYED after a successful stack destroy and excludes Terraform data sources from resource counts; see ravion module.

Platform

Features

  • Module output references: Set any module input from another module’s stack output with a << >> template in the dashboard, the project config file, or a pipeline, including across environments and projects. See output references.
  • AMI deployments: The new AMI module releases an AMI with every deploy. Each deploy builds the image with EC2 Image Builder, copies it to every region, publishes it when requested, and retires older images. Redeploys and rollbacks release earlier images again without rebuilding.
  • New project navigation: A new project sidebar keeps environments and pipelines visible and easy to navigate. When viewing a module, a module sidebar lets you switch to another module.

Improvements

  • Deployment logs now show that they are waiting when a running step has not produced logs yet.
  • The dashboard now uses a scaled desktop layout on phones so content fits the screen.
  • Projects pages now open the only project directly when you enter an organization with one project.
  • Route loading now shows page-specific skeletons instead of a blank dashboard shell.
  • instance_size arrays are now accepted for EC2 infrastructure configuration.

Fixes

  • Email verification now lets users request a fresh code when the original code expires or is rejected.
  • Published module versions now load correctly when opening a module version pane.
  • Failed resource loads now show an error with a retry option instead of an endless loading state.
  • GitLab subgroup repository URLs now work in code-source lookups and no longer appear as unexpected server errors when unconnected.
  • Authenticated requests for Terraform resources with encoded slashes now remain authenticated.
  • EKS deployments now release abandoned targets and report blocked deployments instead of waiting indefinitely.
  • Project configuration dry runs now fail clearly when a values input names no connected resource.
  • Failed requests now retry only transient network, timeout, gateway, and rate-limit errors, while deterministic errors are not resent.

AWS account IAM policy 1.0.47 → 1.0.48

  • AMI deployments: Ravion can build images with EC2 Image Builder, copy them to other regions, tag and publish them, retire older images, and manage its own image resources.
Update your AWS account connection in Settings / AWS accounts.

Modules

New modules

  • AMI 0.1.0: Provision EC2 Image Builder infrastructure that builds AMIs, copies them to additional regions, publishes them when requested, and retires older images.
  • AWS Compliance 0.1.0: Enable GuardDuty and registry-wide Basic ECR scan-on-push across selected AWS Regions. Applying replaces any existing ECR Enhanced scanning or custom filters, so review your current configuration first.
  • AWS Secret 1.0.0: Generate random keys and passwords in Parameter Store or Secrets Manager without storing them in Terraform state or Ravion, with optional multi-Region replication.

Aurora Database 1.2.1 → 1.3.0

  • Add storage capacity and read/write IOPS alarms for Aurora, including Serverless v2.

AWS Application Load Balancer 1.0.1 → 1.1.0

  • Add default CloudWatch alarms for load balancer 5xx responses, target 5xx responses, and target response time, and retain access logs for 365 days by default.

CloudFront CDN 1.3.1 → 1.3.2

  • Retain access logs for 365 days by default.

EC2 Service 1.6.0 → 1.6.1

  • Document that image scanning on push is enabled by default for SOC 2 compliance.

ECS Cluster 1.0.2 → 1.1.0

  • Add default log retention and CloudWatch alarms for public and private load balancers, and document that Container Insights is enabled by default for SOC 2 compliance.

ECS Network Service 1.5.1 → 1.6.0

  • Add inherited log retention and default CloudWatch alarms for CPU, memory, running tasks, and load balancer target health, and document default image scanning on push for SOC 2 compliance.

ECS Web Service 1.5.1 → 1.6.0

  • Add inherited log retention and default CloudWatch alarms for CPU, memory, running tasks, and load balancer target health, and document default image scanning on push for SOC 2 compliance.

ECS Worker 1.5.1 → 1.6.0

  • Add inherited log retention and default CloudWatch alarms for CPU, memory, running tasks, and load balancer target health, and document default image scanning on push for SOC 2 compliance.

Lambda Function 1.1.1 → 1.1.2

  • Enable configurable Lambda error-rate alarms by default, including Lambda@Edge region support, and document default image scanning.

RDS Database 1.2.1 → 1.3.0

  • Add configurable read/write IOPS alarms alongside existing free storage monitoring.

Static Hosting 1.1.2 → 1.1.3

  • Enable CloudFront 5xx error-rate alarms and retain access logs for 365 days by default.

Terraform Stack 1.2.4 → 1.3.1

  • Add independent pull request and primary branch run controls for plan-only or plan, approve, and apply workflows.
  • Clarify that pull request plan results are viewed in Ravion.

Platform

Features

  • GitHub access details: Git connections show organization member counts and repository users, including pending invitations.

Improvements

  • Project cards now show each environment’s module total and infrastructure-run counts.
  • The module canvas now remembers zoom separately for each environment.

Fixes

  • Billing usage periods now follow the contract anniversary and show progress through the current period.
  • Invoice totals now use dollar units and invoice lists show finalized invoices without draft entries.
  • Billing usage pages now handle days that span two billing periods without duplicate entries.
  • Advanced Terraform variables now override fields inside generated nested objects.
  • AMP metric bursts now queue and transient failures retry instead of failing immediately.
  • Releasing a stuck stack lock now checks the specific lock reviewed before unlocking it.
  • Creating a service with EC2 capacity through a referenced cluster now uses the cluster’s capacity provider strategy.
  • Nested module creation dialogs now show the form and can be closed.
  • Email verification no longer crashes when Chrome translation is enabled.

Modules

Aurora Database 1.2.0 → 1.2.1

  • Prevent selecting an older major engine version when updating a database.

EC2 Service 1.4.2 → 1.6.0

  • Add a managed ECR repository build source for services.
  • New services now default to one instance.

ECS Cluster 1.0.1 → 1.0.2

  • Clarify cluster labels and identify additional Network Load Balancer security groups in module forms.

ECS Network Service 1.3.0 → 1.5.1

  • Add a managed ECR repository build source for services.
  • Services using EC2 capacity now create successfully when the cluster is selected by reference.
  • Pre-deploy and post-deploy tasks now follow the capacity provider strategy in advanced Terraform variables.

ECS Web Service 1.3.0 → 1.5.1

  • Add a managed ECR repository build source for services.
  • Configure target group deregistration delays and use clearer cluster, routing, health check, and builder settings.
  • Services using EC2 capacity now create successfully when the cluster is selected by reference.
  • Pre-deploy and post-deploy tasks now follow the capacity provider strategy in advanced Terraform variables.

ECS Worker 1.3.0 → 1.5.1

  • Add a managed ECR repository build source for services.

Lambda Function 1.1.0 → 1.1.1

  • Use consistent Dockerfile and builder settings while keeping build policies in advanced settings.

RDS Database 1.2.0 → 1.2.1

  • Allow major engine upgrades while preventing selection of an older major version.

Static Hosting 1.1.1 → 1.1.2

  • Use consistent Dockerfile and builder settings while keeping build policies in advanced settings.

Platform

Improvements

  • ECS rolling deployments now succeed as soon as new tasks are healthy while old tasks drain in the background; deployment settings can choose a lower healthy-task threshold or wait for old tasks to drain. This applies to existing ECS modules, not just the new versions.

Fixes

  • GitLab code sources now refresh expiring credentials and recover access after successful reauthorization.
  • AWS account connection errors now explain that reconnection is needed and link directly to the account connection flow.
  • Orphaned EC2 runner instances are now reconciled and terminated after interrupted launches.
  • Successful EC2 deployments can now be redeployed or rolled back without failing during input resolution.
  • Organizations with billing data but no projects can now be deleted.

Modules

CloudFront CDN 1.3.0 → 1.3.1

  • CloudFront signed URLs now accept UUID-format key group IDs.

ECS Network Service 1.2.0 → 1.3.0

  • Rolling deployments now succeed as soon as new tasks are healthy while old tasks drain in the background; deployment settings can choose a lower healthy-task threshold or wait for old tasks to drain.

ECS Web Service 1.2.0 → 1.3.0

  • Rolling deployments now succeed as soon as new tasks are healthy while old tasks drain in the background; deployment settings can choose a lower healthy-task threshold or wait for old tasks to drain.

ECS Worker 1.2.0 → 1.3.0

  • Rolling deployments now succeed as soon as new tasks are healthy while old tasks drain in the background; deployment settings can choose a lower healthy-task threshold or wait for old tasks to drain.

Platform

Improvements

  • Module definition settings now offer category creation when an organization has no categories.
  • Deployment diffs now wrap long lines by default for easier reading.
  • ECS deployments now show when they are waiting behind their stack pipeline apply and identify the blocking work.

Fixes

  • The project creation wizard now stays open while moving between steps.
  • Cancelled EC2 deployments now target the current retry and remain within the original execution deadline.
  • GitLab webhook triggers now register project webhooks when configuration is applied.
  • Module references now resolve GitHub repositories that redirect to a new location.
  • You can now change the permitted identity providers for your organization; this setting was incorrectly locked before.

Modules

New modules

  • AWS KMS Key 0.1.0: Provision regional symmetric AWS KMS encryption keys with optional key-user access.

EC2 Service 1.4.1 → 1.4.2

  • Default new services to one instance so they scale out only when Maximum instances is increased.

ECS Network Service 1.1.1 → 1.2.0

  • Enable SOCI lazy loading automatically for Fargate builds while keeping it disabled for EC2 capacity.

ECS Web Service 1.1.0 → 1.2.0

  • Enable SOCI lazy loading automatically for Fargate builds while keeping it disabled for EC2 capacity.

ECS Worker 1.1.0 → 1.2.0

  • Enable SOCI lazy loading automatically for Fargate builds while keeping it disabled for EC2 capacity.

CLI 1.3.2 → 1.4.0

Features

  • AWS account references: AWS values commands accept a 12-digit AWS account number through --aws-account-id.
  • Feature flag listing: ravion feature-flag list evaluates flags for an organization and optional resource scope.
  • Billing cycle management: ravion billing cycle set switches between monthly invoicing and a prepaid annual year, with --yes for non-interactive use.
  • Deployment waiting phase: Deploy commands report when a deployment is queued behind its stack pipeline apply and identify the blocking step.

Platform

Features

  • SOCI lazy loading: Set soci_enabled on a build:image step to use lazy-loaded container images for Fargate. Standard modules will enable it by default after further testing.
  • Billing and usage settings: Manage organization plans, add-ons, payment methods, invoices, and usage from the settings pages.

Improvements

  • Build steps now cancel intermediate runs by default, while blocks apply concurrency only when configured directly.
  • Manual reference inputs no longer show default values as placeholders.
  • Image builds are faster now that cache exports use matching compression settings, avoiding unnecessary layer recompression.

Fixes

  • EC2 execution logs now include the user-facing AWS error when a launch fails.
  • Collapsed input tags now have a consistent height whether or not they include an icon.

AWS account IAM policy 1.0.45 → 1.0.47

  • EKS cluster access: Add eks:DescribeCluster to the deploy policy.
  • AMP metrics access: Add Amazon Managed Prometheus read actions to the broker policy for the EKS metrics dashboard.
Update your AWS account connection in Settings / AWS accounts.

CLI 1.2.2 → 1.3.2

Features

Platform

Features

  • Module canvas editor: Customize your module canvas with drag and drop, grouping, and configure which metrics show on each card
  • Module categories: Module definitions can now be organized into categories to make it easier for users to pick the correct module for their use case.
  • Module build status now on canvas: See build progress and failure details directly on module cards.
  • Destructive change approval: Destructive changes now require stack approval even when auto-approve is enabled.

Improvements

  • You can now create a module directly from the module catalog
  • Create-module dialogs now use more of the available height on short screens.
  • Creating a module now defaults to Create & Apply and uses clearer plan and apply actions.
  • The current-state resources view now uses stack-specific empty-state wording and tighter spacing.
  • API key empty states now use clearer language.
  • Terraform plan resources now use the same change colors as applied resources.
  • In-app AI chat can now set the log search query and time range in the logs view.
  • EC2 health-check status now separates release failures from target health and shows focused diagnostics.
  • Module canvas updates now avoid unnecessary re-rendering and layout work.
  • Modules can now reference producer modules before those producers have applied.
  • Creating an environment now offers Production, Staging, Development, and Test name presets.
  • Run tabs now show the same status icon as the corresponding stack tab, including timed-out runs.
  • Deploy-step drawers now show View deploy when the deployment is identified by the step output.

Fixes

  • Stack runs without current resources are now reachable from the stack page.
  • Pending-approval module cards now open their stack when clicked anywhere on the card.
  • The logs viewer now returns results reliably across supported time ranges.
  • Retrying a partially failed multi-module creation now reuses modules already created in the batch.
  • Pressing Enter inside an empty YAML flow collection now keeps the configuration valid.
  • Terraform resource changes now load correctly when the selected run is outside the pipeline bounds.
  • Cancelled queued runs now clear correctly instead of remaining stuck.
  • Failed partial applies no longer permanently lock module inputs when no live resources remain.
  • Static deployments now report genuine contention as queued and start uncontended work as in progress.

Modules

ECS Network Service 1.1.0 → 1.1.1

  • Clarify documentation that one service can expose up to five TCP, UDP, or TLS ports.

VPC Network 1.0.1 → 1.1.0

  • Add VPC endpoints, including a free S3 gateway endpoint enabled by default for new VPCs, an optional DynamoDB gateway endpoint, and selectable interface endpoints for AWS services.

CLI 1.0.0 → 1.2.2

Features

  • Project config apply: Apply configuration using a project given ID, create the project when it does not exist, and preview the same creation with --dry-run.
  • Project actions: See project creation actions and known-after-apply identifiers in configuration plans.
  • Saved module canvas: Read environment canvas arrangements, including card positions, groups, metrics, and connector shapes.
  • Module step filtering: List pipeline step executions for specific modules instead of every execution.
  • Module categories: Manage categories with generated API support and the ravion module category command group.

Improvements

  • Project config validation: Validation errors now appear as an indented tree with clear spacing around module details.

Platform

Features

  • Current stack state: View live infrastructure, outputs, and workspace information from a persistent current-state view. Learn more
  • 1-click shell into app commands: Copy AWS CLI commands for interactive ECS and EC2 sessions from resource details.

Improvements

  • Deployment resource drawers can expand to the full page width.
  • Stack view dividers can be dragged across a wider range.
  • Deployment definition diffs now work consistently across ECS, EC2, Lambda, and static deployments.
  • Planned Terraform resources are visually distinct from applied resources and show their run state.
  • Terraform resource tables now show failed and changed resources before unchanged resources.
  • List-valued stack outputs now show one value per row with individual copy actions.
  • Log controls move into a compact menu on narrow screens.
  • Module catalog and configuration pages load version lists without fetching full configuration data.

Fixes

  • Missing projects, environments, pipelines, modules, deployments, and resources now show a clear not-found page with a way back.
  • Declared custom-step outputs are preserved, and missing declared output references now fail the run instead of silently becoming empty values.
  • Finished pipeline runs now show unreached steps as skipped, and parallel steps without an explicit strategy let remaining branches finish.
  • Project configuration now resolves AWS account given IDs in nested inputs.
  • Pipeline runs now accept repository URLs without a scheme when starting runner work.
  • Stack configuration diffs now show “No configuration changes” instead of a blank pane when no fields can be displayed.
  • AWS default-VPC errors now explain the fallback and how to resolve the missing default VPC.
  • Deployment failure reasons now remain available as structured, user-facing errors.
  • Pipeline runs now recover more reliably after worker restarts and long-running lock contention.
  • Terraform reports still complete after a failed or canceled run, including when all resources were destroyed.
  • EC2 builds retry Git installation and fail early with a clear infrastructure error when installation cannot complete.
  • EC2 builds no longer fail on certain temporary-file permissions that differ between build environments.
  • Changes to a concurrency limit now take effect for the next run without requiring a new concurrency key.

AWS account IAM policy 1.0.44 → 1.0.45

  • Managed certificates: Request, describe, and delete ACM certificates and attach or detach ALB listener certificates for Ravion-managed domains.
Update your AWS account connection in Settings / AWS accounts.

Modules

CloudFront CDN 1.0.1 → 1.3.0

  • Add standalone ALB origins, including optional private VPC origins.
  • CloudFront cache policy changes and module-managed feature changes no longer fail because of in-use errors.

EC2 Service 1.0.1 → 1.4.1

  • Show an Endpoint link when HTTP traffic is enabled.
  • Mount the host Docker socket for container workloads when configured.
  • Configure extra Builder IAM policies or replace the default build policies for each build.
  • Rotate on-instance application logs with configurable Supervisor limits.
  • Document how to choose an instance type so services can use the newest generation available in their region.

ECS Network Service 1.0.1 → 1.1.0

  • Configure extra Builder IAM policies or replace the default build policies for each build.

ECS Web Service 1.0.1 → 1.1.0

  • Configure extra Builder IAM policies or replace the default build policies for each build.

ECS Worker 1.0.1 → 1.1.0

  • Configure extra Builder IAM policies or replace the default build policies for each build.

Lambda Function 1.0.1 → 1.1.0

  • Configure extra Builder IAM policies or replace the default build policies for each build.

Route 53 DNS 1.0.1 → 1.0.3

  • Quoted values in structured DNS records are now accepted.
  • Long TXT and SPF records are encoded correctly for the AWS provider.

Static Hosting 1.0.1 → 1.1.1

  • Configure extra Builder IAM policies or replace the default build policies for each build.
  • CloudFront cache policy changes and module-managed feature changes no longer fail because of in-use errors.

CLI 0.2.31 → 1.0.0

Features

  • Deployment waiting: Run ravion deploy wait with watch, timeout, until, and JSON output options. Deploy commands
  • Project configuration apply: See the pipeline run IDs started for each module and the matching wait commands.
  • Project file output: Use ravion project create --file to write a generated project configuration draft. Project commands

Platform

Features

  • Custom step outputs: Custom pipeline steps can publish declared string outputs for later steps to use. Learn more.
  • Email verification: Password signups now verify an email address before creating a session.

Improvements

  • Pending Terraform approvals no longer hold the shared stack lock or wait behind plans that have already been superseded.
  • Metric charts now show run status markers, support shared time ranges, and link directly to individual runs.
  • Timestamps across calendars, charts, logs, and timelines now use the viewer’s timezone.
  • Deployment resource logs now support resource-specific queries and incremental history loading.
  • Pipeline graphs now keep steps aligned after grouped sections and route group boundaries cleanly.

Fixes

  • Cleared module defaults now remain cleared instead of being restored automatically.
  • Redeploying an image with multiple tags no longer reports the image as missing.
  • Output-only Terraform changes now remain eligible for apply instead of being treated as no-ops.
  • ECS deployments no longer fail or revert when AWS reports a stale deployment during rollout.
  • Deployment links now remain valid when an environment identifier is available only through the module instance.
  • Deployment detail panels now keep unavailable task-definition messages aligned with their row labels.

AWS account IAM policy 1.0.42 → 1.0.44

  • Managed policy capacity: AWS account permissions are split across managed policies so future permission updates have room to grow without changing effective permissions.
  • Account connectivity: During this update, Ravion may briefly lose connectivity to your AWS account.
Update your AWS account connection in Settings / AWS accounts.

Modules

New modules

  • RDS Proxy 0.1.0: Provision an RDS Proxy for connection pooling in front of an RDS database or Aurora cluster.

Aurora Database 0.1.1 → 1.2.0

  • Add guarded password management controls for safe cluster imports.
  • Add optional RDS Proxy connection pooling.

ECS Cluster 0.5.0 → 1.0.1

  • Add security-group inputs for ALB ingress rules that reference source security groups.

ECS Web Service 0.8.4 → 1.0.1

  • ALB metric charts now include the required load balancer dimension.
  • Assign ALB listener rule priorities at apply time when no explicit priority is set.

RDS Database 0.4.3 → 1.2.0

  • Add guarded password management controls for safe database imports.
  • Add optional RDS Proxy connection pooling.

Route 53 DNS 0.2.1 → 1.0.1

  • Split TXT and SPF record values longer than 255 characters automatically and validate Route 53 record limits before apply.

CLI 0.2.23 → 0.2.31

Features

  • Terraform resource errors: ravion logs now reports the error code, operation, HTTP status, request ID, and the relevant configuration context in plain text.
  • Deployment resource log scope: CloudWatch Logs Insights queries can target the log streams belonging to one deployment resource.
  • Documentation access: CLI help now includes a link to the full Ravion documentation and MCP server.
  • Stack lock controls: Inspect stack locks and release abandoned locks with ravion stack get-lock and ravion stack unlock.

Platform

Features

  • Scheduled pipeline triggers: Schedule pipeline and module-stack runs with cron-based triggers. Learn more.
  • Deployment attribution: External ECS deployments show who initiated them and whether they came from the AWS Console, CLI, Terraform, SDK, or API.
  • Destructive-change confirmation: Stack approvals now identify planned destroys and replacements before allowing approval.
  • Terraform resource errors: Failed plan and apply resources now show structured provider errors and the configuration location that caused them.

Improvements

  • Deployment details now show normalized ECS task-definition changes consistently in cards and detail views.
  • Changed JSON and YAML diffs now start collapsed, while expanded content uses the enclosing panel’s scrolling.
  • Dialogs now stay within the viewport and scroll their bodies while keeping headers and actions visible.
  • Pipeline runs can use organization-level concurrency controls.
  • Module version lists can filter from a specified semantic version, with clearer upgrade indicators.
  • Config-apply stack runs now receive generated descriptions instead of action-only placeholders.
  • Metrics charts now load faster from persistent caching and show applied-run and deployment markers.
  • Module and deployment input values now ignore surrounding whitespace during validation.

Fixes

  • Retrying a failed Terraform apply now preserves its errored state snapshot in the customer AWS account.
  • Deauthorized or deleted EventBridge connections now recover automatically so ECS events continue to arrive.
  • Lambda and Static deployment failures now use the same structured error display as other deployment types.
  • Failed binary downloads now retry transient connection and HTTP failures.
  • Deployment canvases now route wire crossovers and card layouts more clearly.
  • Marketing pages now recover from missing deploy-time JavaScript chunks with a one-time reload.

AWS account IAM policy 1.0.39 → 1.0.42

  • External ECS deployment attribution: Grant CloudTrail event lookup access for identifying external ECS deployments.
  • Cross-account Lambda images: Allow Lambda container-image updates from ECR while keeping direct API calls restricted to Ravion’s IP range.
  • EventBridge connection recovery: Allow EventBridge API destinations to be rebound after a connection is recreated.
Update your AWS account connection in Settings / AWS accounts.

Modules

ACM Certificate 0.4.0 → 0.4.1

  • Allow existing certificates to upgrade to the domains input without being recreated.

CloudFront CDN 0.4.0 → 0.4.2

  • Add a configurable VPC origin.
  • Default allowed methods to read-only.

ECS Cluster 0.3.1 → 0.5.0

  • Default private load balancers to no IPv6 ingress and allow IPv6 ingress CIDR inputs.
  • BREAKING: rename ec2_managed_termination_protection to ec2_managed_termination_protection_enabled and ec2_managed_scaling_status to ec2_managed_scaling_enabled, and change both inputs to booleans.

ECS Network Service 0.3.1 → 0.3.2

  • Allow existing services to upgrade to the listeners input without being recreated.

ECS Web Service 0.8.3 → 0.8.4

  • Clarify build source, Railpack command, and builder instance guidance.

Lambda Function 0.3.2 → 0.4.0

  • Stop creating a module-owned ECR repository when pulling from an image registry and require an initial image tag or digest before the first deployment.

CLI 0.2.15 → 0.2.23

Features

  • Trigger-concurrency run status: The CLI now treats skipped trigger-concurrency runs as terminal and reports a failure while waiting for completion.
  • Module version validation: Use --dry-run with ravion module version create to validate configuration and return defaults without creating a version. Learn more.
  • Version filtering: Use --since-version with ravion module version list to show versions at or newer than a semantic version. Learn more.
  • Terraform execution pagination: Terraform execution summaries are now returned newest first with cursor pagination. Learn more.
  • Terraform resource errors: Terraform resources now expose structured errors for failed plan or apply changes, including provider details and configuration location.
  • Moved module inputs: Module inputs can declare moved_from values so upgrades that rename or merge immutable inputs do not fail the immutable-input check. Learn more.
  • Stack-run descriptions: Use --description with ravion project config apply to set the description on every stack run it creates. Learn more.

Platform

Features

  • Public module catalog: Browse globally published active modules without signing in. Module catalog
  • Nested module references: Configure module references inside object maps and arrays. Module definition schema
  • Task health graph: View task health relationships in a dedicated graph.

Improvements

  • Lists now show each page of results as it loads and continue displaying results already received when a later page fails.
  • Pipeline and stack resource lists now load additional results as you scroll near the bottom.
  • Stack run details now show configuration changes, inputs, outputs, and run context in a tabbed view.
  • Module and pipeline graphs now use clearer layouts for dependencies, groups, and cross-links.
  • New EC2 instances now wait for the latest successful release before joining service traffic.

Fixes

  • The Catalog breadcrumb now appears on module definition pages and the project breadcrumb now links to the pipelines list in pipeline views.
  • Invitation links now show the invitation details and let matching users accept invitations through email or OAuth sign-in.
  • Retrying failed pipeline steps now preserves successful steps and correctly handles steps skipped by dependencies.
  • Module reference value selectors now work during creation and retain mapped values when changing module versions.
  • Module configuration diffs no longer show changes caused only by server-resolved reference values.
  • Failed EC2 deployments now automatically restore the previous successful release when eligible.
  • Pipeline and stack pages now show the correct latest non-cancelled run status without flicker.

AWS account IAM policy 1.0.38 → 1.0.39

  • EC2 launch release gates: Auto Scaling launch release gates can send lifecycle heartbeats and complete lifecycle actions.
Update your AWS account connection in Settings / AWS accounts.

Modules

New modules

  • AWS Application Load Balancer 0.1.0: Provision an AWS Application Load Balancer with HTTP and HTTPS listeners.
  • AWS IAM Policy 0.1.0: Provision a reusable customer-managed AWS IAM policy from structured statements or a complete policy document.
  • EC2 Service 0.1.1: Run supervised workloads on an EC2 Auto Scaling Group with web and storage configuration.

CloudFront CDN 0.3.2 → 0.4.0

  • Add custom headers for primary and additional origins.

ECS Cluster 0.3.0 → 0.3.1

  • Add additional load balancer certificate ARN fields for SNI.

ECS Network Service 0.2.2 → 0.3.1

  • BREAKING: replace container_port, listener_port, and listener_protocol with one or more listeners items before upgrading, and move TLS settings into the relevant listener.
  • Improve build source, Railpack command, and builder instance guidance.

ECS Web Service 0.8.2 → 0.8.3

  • Improve build source, Railpack command, and builder instance guidance.

ECS Worker 0.3.2 → 0.3.3

  • Improve build source, Railpack command, and builder instance guidance.

Lambda Function 0.3.1 → 0.3.2

  • Improve build source and builder instance guidance.

Static Hosting 0.3.3 → 0.3.4

  • Improve build source, Railpack command, and builder instance guidance.

CLI 0.2.5 → 0.2.15

Features

  • Invitation signup: Use the invitation token from an invite link with ravion signup to accept the invitation during signup. Signup
  • Invitation lookup: Run ravion membership get-invitation invitationToken to view a pending invitation’s organization, invitee, and status. Membership
  • Bug and feedback reports: Run ravion report bug, ravion report feedback, ravion bug, or ravion feedback to submit Markdown reports. Report

Fixes

  • Session refresh: Active CLI sessions now roll refresh tokens before they expire and remain usable when a temporary refresh fails. Authentication
  • Expired sessions: ravion whoami now reports expired or revoked sessions as logged out and directs you to ravion login. Whoami