As far as I can tell the agent deployments are currently always done in parallel. It would be awesome if I could split the agent deployments up into batches of a size of my choosing (from size=1 to size=“as many as possible”).
What I mean is; if I have 8 agents/servers and I want to deploy to them in batches of 3, the process would be:
-
Batch 1
- Agent 1, Agent 2, Agent 3 (in parallel)
-
Batch 2 (after Batch 1 completes)
- Agent 4, Agent 5, Agent 6 (in parallel)
-
Batch 3 (after Batch 2 completes)
- Agent 7, Agent 8 (in parallel)
This process would let me handle load-balancers gracefully and ensure high availability, while still taking advantage of parallel deployments in some degree.