Stacktape Console executes GitOps deployments on a dedicated EC2 runner in your connected AWS account. Each Git-backed project has its own runner, AWS region, and instance type.
The runner keeps dependency and build caches on its EBS volume between deployments. It automatically stops after 15 minutes without deployment or GitHub Actions work, so compute charges stop while the project is idle. The next job starts the existing instance and reuses its caches.
When the runner is used
The EC2 runner executes operations started by Stacktape Console, including:
- deployments triggered by pushes or pull requests
- deployments started manually from the Console
- stage deletions started from the Console
- self-hosted GitHub Actions jobs when that feature is enabled for the project
Local CLI commands and custom CI/CD pipelines normally run on the machine or CI runner that invokes Stacktape. Their compute and lifecycle are independent from the Console-managed runner.
Runner configuration
Runner settings are configured per project in the Stacktape Console. Open Runners, select the project's runner and open its Settings tab to choose the instance size, the build cache limit, what happens when the runner is idle and the project's setup commands. The runner region is selected when you create a Git-backed project.
Instance types
The instance type controls the CPU and memory available during builds. Larger instances can shorten CPU-intensive builds but cost more while running.
| Instance type | vCPU | Memory |
|---|---|---|
m6a.large | 2 | 8 GB |
m6a.xlarge | 4 | 16 GB |
c7a.xlarge | 4 | 8 GB |
c7a.2xlarge | 8 | 16 GB |
c7a.4xlarge | 16 | 32 GB |
c7a.8xlarge | 32 | 64 GB |
The default is c7a.2xlarge with 8 vCPU and 16 GB of memory. The available families suit different workloads:
c7ainstances favor CPU-intensive builds, such as compiling applications and building container images.m6ainstances provide more memory per vCPU for memory-heavy builds.
Start with the default and change the size if runner metrics or build duration show that the project needs a different balance. A new size takes effect before the next operation executes.
Runner region
The runner region determines where its EC2 instance, EBS volume, and supporting resources live. Choose a region supported by Stacktape's runner image and suitable for the AWS account that will own the infrastructure.
Choose a runner region close to the regions where you usually deploy. This can reduce latency and cross-region data transfer during builds and uploads.
Setup commands
Setup commands are shell commands that each deployment on a runner runs before stacktape deploy, in a fresh checkout
of your code. Use them to install tools your build needs, point a package manager at a private registry, or export
environment variables for the build (export NAME=value).
- Every project: owners and admins edit them with the Setup commands button at the top of the Runners page.
- One project: edit them on the Settings tab of the project's runner. They run after the commands for every project.
The commands run in order in Bash as a user without root, so install tools into the home directory (for example
pip install --user or npm install -g). If a command fails, the deployment stops and its log shows the output.
Deletes, rollbacks and GitHub Actions jobs do not run setup commands. A line starting with # names the commands under
it, and a blank line starts a new group. Keep the commands deterministic, pin tool versions and do not print secrets.
Lifecycle and cost
The runner is provisioned when first needed and reused for later work. When a job arrives, Stacktape makes sure the instance is running and ready before dispatching the operation. After 15 minutes without active work, Stacktape stops the instance and attempts to hibernate it where AWS supports that lifecycle.
You pay AWS directly for the runner resources in your account:
- EC2 compute while the instance is running
- EBS storage while the runner exists, including while stopped or hibernated
- ordinary AWS data transfer and logging charges generated by runner activity
Rates vary by region and change over time. Use the AWS Pricing Calculator or your AWS billing tools for current estimates.
Build caches
The runner stores package-manager, source, and build caches on its persistent volume. This avoids downloading and rebuilding unchanged inputs on every deployment. Stacktape isolates individual operation work directories while reusing the caches that make repeated jobs faster.
Use --no-cache for a CLI deployment when you need Stacktape to rebuild artifacts without its normal build cache. That flag does not delete the runner or all operating-system-level caches.
Build Runners FAQ
Do I need a Console runner for custom CI/CD?
No. GitHub Actions, GitLab CI, CircleCI, and other custom pipelines can run the Stacktape CLI on their own workers. The Console-managed runner is for operations dispatched through Stacktape Console and for the optional self-hosted GitHub Actions integration.
What instance type should I choose?
Start with c7a.2xlarge. Choose a smaller type for lightweight or infrequent builds, a larger c7a type for CPU-bound compilation and container builds, or an m6a type when the build needs more memory per CPU. Re-evaluate based on real build duration and AWS cost rather than choosing the largest instance by default.
What happens when the runner is idle?
Stacktape stops the EC2 instance after 15 minutes without active deployment or GitHub Actions work. The EBS volume remains so caches survive, and AWS continues billing for that storage. The next job starts the runner before executing.
Can deployment and GitHub Actions jobs run at the same time?
They use the same project runner. Stacktape coordinates runner work so jobs do not corrupt shared state. If the runner is busy, new work waits or reports that the runner is occupied rather than starting an unsafe overlapping operation.