github.com-cloudflare-containers

Initial pass at Container class

Refactoring

refactor container class 1. fix type issues 2. fix alarm not triggering issue 3. simplifying some paths

Add interceptHttps option to allow intercepting HTTPS via Workers and add allowed/denied Hosts option

Temporarily throwing errors based on retry count and not time or abort signal - Surfaces error better - Does not autostart a container in constructor, so we can throw error nicely Generally working though hacky Update error response Tweaks

experimental outbound interception

containers: tighten up all catch() paths in main monitor tracking of the container There would be cases that the customer would see a 'running' container on an unknown monitor error when in reality it was stopped already.

Add basic e2e tests (#80) * Version Packages * add basic local fixture tests * add github action to run tests on PRs * delete redundant node example --------- Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

add params to set handlers, and remember last setOutbound call

adjustements

track container state in 'ctx.storage' so we can trigger hooks that guarantee atleast once delivery We are also introducing a monitor() call when the container is running and there is not an existing one.

fix: Make sure an alarm can't call onStop in the middle of a start() call We are also making sure onStop can't set 'stopped' if the method calls start().

improve port checking (#77) * fix: check ports in startAndWaitForPorts even if the container has already started * feat: add waitForPort() and refactor startAndWaitForPorts() accordingly. also make sure user-provided waitInterval is actually being used * and tests

tls: improve docs, dont trust certs on main.go (instead only entrypoint trust), and make sure removing hostnames work

fix: Do not blockConcurrencyWhile on startAndWaitForPorts It's completely okay to have multiple healthchecks to the container with concurrent requests to the DO. This is also a useful fix because it fixes the issues we were seeing with waiting for ports to be healthy if you were to call start() beforehand.

containers: Add `onActivityExpired` so users can configured how to act on activity expiration Activity expiration now is measured since the last time a HTTP request has arrived to the container class. We can't reliably measure when a response gets closed to renew the activity right now, so let's rely for now on the user to decide if they want to renew activity by optionally contacting with the container.

containers: Do not run alarm all the time and do setTimeout

change readme to showcase the new changes, remove manualStart, document the new behaviours

fixes: Network Connection Lost handling and set again monitor callbacks when the container restarts

cleanup the code