Alethia Labs

Runners

Registering, monitoring, and managing Runner execution agents.

Runners

Runners are the execution agents that run Terraform on behalf of users. The Runners page manages their lifecycle. For how Runners work internally, see the Runner Agent documentation.

Runners Page

Located at /dashboard/runners, showing:

Status Filters

  • All, ONLINE, OFFLINE, DRAINING

Filter by Runner name.

Data Table

ColumnDescription
NameRunner identifier
Typecloud-hosted or self-hosted
RegionAWS region (cloud-hosted only)
StatusBadge: ONLINE (green), DRAINING (yellow), OFFLINE (red)
VersionRunner release version
CreatedRegistration timestamp
ActionsDeploy, update, remove buttons

Active Jobs Section

A separate table showing jobs currently assigned to Runners (QUEUED, CLAIMED, or PROCESSING).

Update Alert

If outdated Runners are detected (version < latest release), an alert appears with an "Update All" button that queues UPDATE_WORKER jobs for each outdated Runner.

Adding a Runner

Click "Add Runner" to open the registration sheet with two tabs:

Provisions a Runner automatically in ECS Fargate:

  1. Select a cloud identity and region
  2. Click Deploy → queues a DEPLOY_WORKER job
  3. The job provisions: ECS task definition, IAM roles, CloudWatch log group, security groups
  4. After ~2 minutes, the Runner appears as ONLINE

Cloud-hosted Runners support scale-to-zero — they shut down when no jobs are queued and restart automatically when needed.

For running a Runner on your own infrastructure:

  1. Enter a name for the Runner
  2. Click Register → generates a worker_id and worker_token
  3. Copy the credentials (shown once, never stored in plaintext)
  4. Set the environment variables on your machine and run the Runner binary

See Self-Hosted Mode for full setup instructions.

The worker token is shown only once at registration. If lost, the Runner must be re-registered. See Authentication Flows for how token verification works.

Status Indicators

Status is determined by heartbeat recency:

StatusMeaningHeartbeat
ONLINEReady to accept jobsWithin last 60 seconds
DRAININGFinishing current job, not accepting new onesSent with DRAINING flag
OFFLINEMissed heartbeat, jobs will be requeued> 60 seconds ago

Real-time Updates

The Runners page subscribes to workers table changes and provision_jobs changes via Supabase Realtime. Status changes, new registrations, and job assignments appear instantly.

Version Tracking

Each Runner reports its version on heartbeat. The page shows:

  • Current version per Runner
  • Latest available version (from worker_releases table)
  • Release Notes dialog showing the changelog for each version

When a new Runner release is published via the CI pipeline, existing cloud-hosted Runners are updated via UPDATE_WORKER jobs that trigger ECS force-new-deployment.

On this page