Kiln - VOD transcoding in your AWS account

Kiln usage (v1.0.0)

Image: 709825985650.dkr.ecr.us-east-1.amazonaws.com/r-erto/klin:1.0.0 Support: https://robertokiln.com/#support  Email: support@robertokiln.com  (reply within 24 hours)

Need: Marketplace subscription, Docker Compose or Kubernetes+Helm, Postgres, shared object volume. Kiln runs in your account. Media stays on your storage.

1. DOCKER COMPOSE (fastest)

Save as docker-compose.yml (replace IMAGE with the URI above):

services: postgres: image: postgres:16-alpine environment: POSTGRES_USER: kiln POSTGRES_PASSWORD: kiln POSTGRES_DB: kiln ports: ["55432:5432"] healthcheck: test: ["CMD-SHELL", "pg_isready -U kiln -d kiln"] interval: 2s timeout: 3s retries: 20 api: image: IMAGE command: ["api"] environment: &env KILN_DSN: postgres://kiln:kiln@postgres:5432/kiln?sslmode=disable KILN_LISTEN: ":8080" KILN_API_URL: http://127.0.0.1:8080  KILN_DATA_DIR: /var/kiln/objects KILN_TIER: starter KILN_ENCODER: fake ports: ["8080:8080"] volumes: [objects:/var/kiln/objects] depends_on: postgres: { condition: service_healthy } scheduler: image: IMAGE command: ["scheduler"] environment: { <<: *env } volumes: [objects:/var/kiln/objects] depends_on: postgres: { condition: service_healthy } api: { condition: service_healthy } worker: image: IMAGE command: ["worker"] environment: <<: *env KILN_NODE_ID: worker-1 KILN_GPU: "1" KILN_WORKER_SLOTS: "2" volumes: [objects:/var/kiln/objects] depends_on: postgres: { condition: service_healthy } api: { condition: service_healthy } volumes: objects:

Login and run: aws ecr get-login-password --region us-east-1 | docker login --username AWS --password-stdin 709825985650.dkr.ecr.us-east-1.amazonaws.com docker compose up -d docker compose exec api kiln version docker compose exec api kiln health docker compose exec api kiln ladders docker compose exec api kiln submit --wait --new-id /etc/kiln/job.json docker compose exec api kiln jobs docker compose exec api kiln cluster docker compose exec api kiln status JOB_ID docker compose exec api kiln invoice JOB_ID docker compose exec api kiln translate /etc/kiln/mediaconvert-job.json docker compose exec api kiln support-bundle /tmp/kiln.tgz Stop: docker compose down -v Samples in image: /etc/kiln/job.json and /etc/kiln/mediaconvert-job.json

2. ONE CONTAINER (Postgres required) docker compose up -d postgres docker run --rm -p 8080:8080
   -e KILN_DSN='postgres://kiln:kiln@host.docker.internal :55432/kiln?sslmode=disable'
   -e KILN_LISTEN=:8080 -e KILN_DATA_DIR=/var/kiln/objects
   -e KILN_TIER=starter -e KILN_ENCODER=fake
   IMAGE run

3. KUBERNETES helm upgrade --install kiln ./deploy/helm/kiln
   --set image.repository=709825985650.dkr.ecr.us-east-1.amazonaws.com/r-erto/klin
   --set image.tag=1.0.0 --set tier=starter
   --set data.storageClassName=EFS_OR_NFS_RWX_CLASS Object PVC must be ReadWriteMany. Set tier to starter, growth, or scale. kubectl port-forward svc/kiln-api 8080:8080 kubectl exec -it deploy/kiln-api -- kiln submit --wait --new-id /etc/kiln/job.json

4. COMMANDS api | scheduler | worker | run submit [--wait] [--new-id] SPEC.json status | cancel | health | ladders | jobs | chunks | cluster invoice | translate | support-bundle | version HTTP API on KILN_LISTEN (default :8080).

5. ENV VARS KILN_DSN (required), KILN_LISTEN, KILN_API_URL, KILN_DATA_DIR, KILN_TIER (starter|growth|scale), KILN_ENCODER (fake for eval), KILN_NODE_ID, KILN_WORKER_SLOTS, KILN_GPU

6. IF STUCK kiln health kiln cluster kiln status JOB_ID (leased=0 + high pending usually means workers cannot poll or bad DSN) kiln support-bundle /tmp/kiln.tgz then email support@robertokiln.com 

NOTE: Fake encoder is for install checks. For production, give workers ffmpeg/codecs and set KILN_ENCODER for production. More: https://robertokiln.com/