As we start doing more and more in the public EC2 cloud, having a Go agent in the cloud to do builds and deployments is becoming an issue. However, the current design of the agent, at least as I know it, prevents this - we will not allow externally-initiated connections into our internal network where the Cruise/Go Server resides.
While the EC2 VPC would help here, it limits our options and flexibility especially for deployments to the public cloud.
It would be nice to have an option where the Cruise/Go Server could initiate all connections to a listening Agent. I don't know if the polling for work would also have to be under control of the Server.
Is this under consideration, or a possiblity as a new feature?
Please sign in to leave a comment.