Feature description
- Improve workload fleet management API mechanics regarding when the device is deploying/updating an application along with status reporting.
Provide adequate technical acceptance criteria(s) associated with this feature below:
Details about deployment status semantics - Define when devices report (state transitions, periodic heartbeat, or both), how retries work (backoff, max age), how idempotency/dedup works, and what servers should expect during long disconnects.
Details about desired state semantics - Clarify what constitutes a "new revision" vs "same deployment updated" (does ID change on parameter change?), what failure handling is required (stop + report vs rollback), and whether devices must retry component apply.
Although not required, it is highly encouraged to provide feature use-cases below:
Improved workload management through the Management interface definition.
Additional information
Medium to High impact on GA release - if we do not specify when the WFM client reports workload state, or how a WFM client acts upon failed deployments (will it retry forever, give up after N attempts, backoff timers used, ...?), a WFM and WFM user would not know if and how to react upon failures - and might not be able to observe the actual state of deployments.
Feature description
Provide adequate technical acceptance criteria(s) associated with this feature below:
Details about deployment status semantics - Define when devices report (state transitions, periodic heartbeat, or both), how retries work (backoff, max age), how idempotency/dedup works, and what servers should expect during long disconnects.
Details about desired state semantics - Clarify what constitutes a "new revision" vs "same deployment updated" (does ID change on parameter change?), what failure handling is required (stop + report vs rollback), and whether devices must retry component apply.
Although not required, it is highly encouraged to provide feature use-cases below:
Improved workload management through the Management interface definition.
Additional information
Medium to High impact on GA release - if we do not specify when the WFM client reports workload state, or how a WFM client acts upon failed deployments (will it retry forever, give up after N attempts, backoff timers used, ...?), a WFM and WFM user would not know if and how to react upon failures - and might not be able to observe the actual state of deployments.