Kubernetes support for Scanner

Files needed:

• .scanner.toml with your configuration
• google-key.json service account key (Credentials > Create credentials > Service account key > JSON) Disposition-Notification-To: Will Crichton wcrichto@halfmoon.i-did-not-set--mail-host-address--so-tickle-me

Environment variables needed:

• AWS_SECRET_ACCESS_KEY and AWS_ACCESS_KEY_ID (see GCS Settings > Interoperability)

First, set up your local Docker containers:

python cluster.py build
docker-compose build
docker-compose up -d

After this, all commands (except python cluster.py build) will run a local Docker container, which you can enter by doing ./run.sh.

Start a Kubernetes cluster:

python cluster.py create

Run an example computation on the cluster:

python client.py

Commands

Update existing containers:

kubectl delete deploy --all
python cluster.py create

Resize the cluster (e.g. to 5 workers):

python cluster.py resize 5

Delete the Kubernetes cluster:

python cluster.py delete

Dashboard

First, if your local Docker container is not on your laptop, you'll need to setup an SSH tunnel:

ssh -L 8001:localhost:8001 <your server>

Then visit http://localhost:8001/ui.

Note: this is broken until I can figure out how to fix the permissions issue changed in k8s 1.8 (see here and here).