We're using the AMQPUrlReceiver and in three of four of our crawlers, H3 has stopped picking up messages. In at least two of the three 'hung' crawler jobs, there was an error like this:
and then, a little later, an error like this:
I believe an uncaught exception in a thread will at least take out that thread, and possibly a whole threadpool. In which case, the error we are seeing will presumably take out the StarterRestarter and prevent any closed RabbitMQ connection from being re-opened.
Firstly, does this sound about right? Secondly, should the BDB error we are seeing ever happen? Thirdly, should we switch to catching RuntimeExceptions in the UrlConsumer.