ReqManagedAgents.Providers.BedrockAgentCore.WaitContext (ReqManagedAgents v0.13.2)

Copy Markdown View Source

What a provisioning wait observed when it gave up.

Wait errors previously carried bare atoms, so a timeout could not be tied to a harness, a last-seen status, or an elapsed time — the reason a real canary failure could not be diagnosed from its run log.

harness_id names the resource the wait was tracking: its harnessId during the :ready and :endpoint phases, and its name during the :deleted phase, because that wait polls ListHarnesses by name and the id of a harness that is going away is not what identifies it to the next create.

:endpoint is the wait for the harness endpoint an invoke will reach. endpoint_name names it, and is set only in that phase: the endpoint is chosen by the caller (:endpoint_name, defaulting to the one the data plane defaults to), so an endpoint wait that reports last_status: nil — every poll a 404 — says nothing at all unless it also says which endpoint it asked for.

Summary

Types

phase()
t()

Types

phase()

@type phase() :: :ready | :endpoint | :deleted

t()

@type t() :: %ReqManagedAgents.Providers.BedrockAgentCore.WaitContext{
  elapsed_ms: non_neg_integer() | nil,
  endpoint_name: String.t() | nil,
  harness_id: String.t(),
  last_status: String.t() | nil,
  phase: phase(),
  polls: non_neg_integer() | nil
}