Open jage opened 9 years ago
Note: closed issues above can be closed because the issues are covered by other issues (duplicates), not because they are solved
RabbitMQ server shutdown
We have HA in our datacenter, but twingly-amqp can't switch server without causing problems.
Found an interesting example which might be useful for testing, i.e. compare this with a twingly-amqp-implementation.
Upgraded RabbitMQ, which caused restart. Two projects did not like it still.
Bunny >=2.9 tries to reconnect on timeout errors: https://github.com/ruby-amqp/bunny/blob/master/ChangeLog.md#connection-recovery-now-retries-on-timeouts
Related PR: https://github.com/ruby-amqp/bunny/pull/537
Could be some help in the article at https://www.cloudamqp.com/blog/2018-01-19-part4-rabbitmq-13-common-errors.html
One thing I wondered if we are following was:
Use separate connections for publish and consume.
Use separate connections for publish and consume.
Without looking, I'd say no. I think we share the same connection.
I think it's important that we try to figure out how to handle different Bunny (and OpenSSL) exceptions in this gem. In most of our applications it's mostly annoying (Sentry notifications), but in others like Mako and Ganges it could take the application down if exceptions are not handled carefully.
Basically we need to handle: