Troubleshooting

Sometimes things do not work as they should. Action could be timing out due to a non-obvious error, docker containers could fail to start due to an error in your code, or the underlying instance might misbehave.

Easiest things you could do first:

• start fresh: remove the pullpreview label, wait until the environment is destroyed, then re-apply the label. This will make sure you start from a fresh instance.
• increase instance size: sometimes the deployment seems to hang forever, especially when building anything related to node / webpack, since it uses enormous amounts of memory to compile bundles. A 2GB swap is enabled on the instances (which is meant so that you don't need to switch to a larger instance just for some compilation), but sometimes it's not enough or gets too slow and your deployment can time out.
• if related to timeout issues: increase the timeout by 10min to see if it makes any difference.

If need to debug further:

• log into the EC2 instance while the deployment is occurring, and have a look around with htop, docker logs, docker ps etc. see if you are running out of memory or see misbehaving containers.
• log into the EC2 instance, and manually start the compose stack to see if any errors are output interactively:

ssh ec2-user@IP
cd /app
cat $(env /etc/pullpreview/env) docker-compose -f docker-compose.yml,docker-compose.preview.yml up