IssueRaiser

Parallelism control per-resource when count > 1

Feature/Enhancement request I'd like to see support for modifying the parallelism on a per-resource basis. Use cases: 1. Create a quantity of the same resource, but actions must be executed one at a time. An example of this would be chef-backend, and joining the cluster. A leader (the first) must be established, but then subsequent backend members (initially followers) must join one at a time due to the backend configuration steps going on in the background. This would avoid the need to craft a waiting mechanism for N objects (here's where `count` comes into play) 1. For load considerations, sometimes more or less than the current parallelisim can be handled. It would be nice to have control per resource on this. 2. In the case of a cluster initiator and cluster members joining (sequential) we have the possibility to control initialization versus joining, and do so in sequence in one resource versus two or more with complicated logic. 3. Sequential phases are more tightly controlled, and doesn't affect `depends_on` but reduces overall line count and appearances of resource duplication

Posted by mengesb on Jul 13, 2016, 9:02 PM

Tags

Related issues

Raise real-world problems, find people facing the same issue, and discuss solutions.5,000 issues0 comments

support (Other)

Parallelism control per-resource when count > 1

mengesbJul 13, 2016, 9:02 PM

Feature/Enhancement request I'd like to see support for modifying the parallelism on a per-resource basis. Use cases: 1. Create a quantity of the same resource, but actions must be executed one at a time. An example of this would be chef-backend, and joining the cluster. A leader (the first) must be established, but then subsequent backend members (initially followers) must join one at a time due to the backend configuration steps going on in the background. This would avoid the need to craft a waiting mechanism for N objects (here's where `count` comes into play) 1. For load considerations, sometimes more or less than the current parallelisim can be handled. It would be nice to have control per resource on this. 2. In the case of a cluster initiator and cluster members joining (sequential) we have the possibility to control initialization versus joining, and do so in sequence in one resource versus two or more with complicated logic. 3. Sequential phases are more tightly controlled, and doesn't affect `depends_on` but reduces overall line count and appearances of resource duplication

Comments (0)

No comments yet. Start the discussion.

Comment on this issue