open-athena/lambda-gha

GitHub Action for creating a self-hosted runner on Lambda Labs GPU cloud.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
action_reflambda-gha Git ref (branch/tag/SHA) to use for fetching scriptsnomain
debugDebug mode: false=off, true/trace=set -x only, number=set -x + sleep N minutes before shutdownno
extra_gh_labelsExtra GitHub labels for the runner (comma-separated, no spaces)no
instance_countNumber of instances to create (default: 1)no1
check_availabilityPre-check capacity before attempting launches (default: auto, enabled when multiple types/regions)no
instance_typeLambda instance type(s), comma-separated for fallback (e.g., gpu_1x_a10,gpu_1x_a100)no
lambda_api_keyLambda Labs API key (falls back to secrets.LAMBDA_API_KEY)no
max_instance_lifetimeMaximum instance lifetime in minutes before shutdown (default: 360)no
regionLambda region(s), comma-separated for fallback (e.g., us-east-1,us-west-1)no
retry_countNumber of retries per instance type/region combination (default: 1)no1
retry_delayInitial delay between retries in seconds, uses exponential backoff (default: 5)no5
repoGitHub repo to register runner with (default: current repo)no
runner_grace_periodSeconds before terminating after last job completes (default: 60)no
runner_initial_grace_periodSeconds before terminating if no jobs start (default: 180)no
runner_poll_intervalSeconds between termination condition checks (default: 10)no
runner_registration_timeoutMax seconds to wait for runner registration (default: 300)no
ssh_key_namesSSH key names registered in Lambda Labs (comma-separated)no
ssh_private_keySSH private key for connecting to Lambda instancesno
userdataAdditional script to run before runner setupno
namedescription
mtxJSON array for matrix strategies: [{idx, id (label), instance_id}, ...]
labelRunner label (for single instance use)
instance-idLambda instance ID (for single instance use)