Spot Recovery

Spot Recovery is a Phase 3 feature, coming soon.

Automatically restart your workload when a spot instance is interrupted.

How It Works

Pulserun monitors your spot instances continuously
When a spot interruption is detected (within 30 seconds)
Pulserun immediately finds the next cheapest available GPU
A new instance is provisioned automatically
Your template and SSH keys are re-applied

Detection Time

Spot interruptions are detected within 30 seconds of the provider terminating the instance.

Enable Spot Recovery

curl -X POST https://api.pulserun.dev/v1/instances \
  -H "Authorization: Bearer pr_live_xxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{"gpu": "a100_80gb", "template": "pytorch", "spot": true, "auto_spot_recovery": true}'

Webhook Notifications

Subscribe to spot.preempted events to be notified of interruptions.

Auto-Migration Scheduled Jobs