Matt Yule-Bennett

@davidszotten there isn't anything specific that already exists, but it's easy to imagine a request-response type entrypoint that might want to map remote errors to specific exceptions in the client....

Sorry for the slow reply here. I'm not very familiar with the websocket implementation, but I think at least the "client quit" case has come up before. Have you seen...

Yes, they are out of date I'm afraid. A PR would be great if you'd like to fix them ;) Note that we have a [v3 pre-release branch](https://github.com/nameko/nameko/tree/v3.0.0-rc) (where they...

Nice investigation. You're right that there's nothing that can be done at the Nameko level, but it does deserve calling out in the docs. This is a limitation of cooperatively...

> TL;DR - no guarantee that your task will complete if your service crashes. If you gently shut it down, I think the default behavior is to wait for the...

@laike9m `start()` does start the service, but you must `wait()` on the container or the runner to stop your process or thread from exiting. The best way to run a...

Sorry, I forgot about this thread. `stop()` waits for all running workers and managed threads to exit gracefully. You can't expose `container.stop()` in an entrypoint and expect to the container...

Yes indeed, the example above is not a demonstration of how you _should_ achieve this. It is only meant to illustrate an "out of band" thread. Generating a `SIGTERM` signal...

This definitely should be a topic in the docs, which is why this issue is here. Have you discovered [this forum post](https://groups.google.com/forum/#!searchin/nameko-dev/serializer%7Csort:relevance/nameko-dev/tzrWa9bV7Xc/2Y71SsqkBwAJ) on the topic? You should be able to...

@jayvdb is right that #703 is the underlying cause. Nameko's pytest plugin is applying the eventlet monkey patch (required for all tests that actually use Nameko) and it's interfering with...