koajs / cluster Goto Github PK

Is there anything I can do to prevent koa-cluster from auto-restarting like 100 times a second if the server crashes instantly?

Even if it requires only running one, non-clustered process, allowing --debug and --debug-brk on koa-cluster itself and passing that through to node will allow people to debug without changing their code to make exceptions for running in non-clustered mode.

Does this work with koa 2.0?

if you do something async before an error is thrown, and another connection is created while the async stuff is waiting, the error will be caught in the wrong domain (the domain created in the second request), resulting in a 500 being sent to the wrong client. This is why we need domain support in co for domains to be used with koa.

Noticed in the Node v0.12 changelog:

http://blog.nodejs.org/2015/02/06/node-v0-12-0-stable/

• Cluster
  • Now has two modes of operation, the new default is a round robin distribution mechanism where the master accepts new connections and distributes them to your workers. If you want you can still opt back into the old method where your workers are responsible for acception connections.

Does that mean anything for koa-cluster?