Closed gdw2 closed 6 years ago
Looking through other issues, I see that this type of question is preferred to be discussed in the mailing list. https://github.com/rabbitmq/rabbitmq-peer-discovery-aws/issues/8#issuecomment-355834363
If so, feel free to close, and I apologize.
The Race Conditions During Initial Cluster Formation section in the docs answers this exact question.
I tried clarifying what peer discovery backend registration is. Let me know if it's good enough in your opinion.
Looks good, thx.
I've been deploying RMQ repeatidly and have noticed that the nodes don't always cluster. I suspect that its simply due to the race condition described in the docs:
Interestingly, it doesn't mention what strategy the AWS plugin uses.
When I look at my logs, I see this:
I don't know exactly what is meant by "registration", but I would think the lack of support for it would warrant a startup delay? I can of course inject my own startup delay, but am mostly writing this ticket to ask if the AWS plugin makes any attempts to mitigate startup race conditions?