A node's worker label is worker, so a device is only a paired one

- NodeDef `device`/`device_policy` become `worker`/`worker_policy`, the
  issue `ignored_device` becomes `ignored_worker`, and the SDK's `@node`
  takes `worker=` and `worker_policy=`
- FlowStore._migrate renames the keys in published and draft documents
  once, bumping their version; a queued run's `needs` reads the old key
- placement passes the label as `label`, client regenerated
- docs: worker placement wording, MCP issuer without DOMAIN, device rows in
  the CLI data directory and the API table, device links in compose comments
This commit is contained in:
Melvin Strobl
2026-10-16 16:56:05 +02:00
parent 29d3ec467f
commit 1dc891aa1b
40 changed files with 215 additions and 178 deletions
+2 -2
View File
@@ -108,7 +108,7 @@ def test_a_required_label_waits_for_its_machine_rather_than_running_here(loop):
placer = placer_over(hub)
worker, _ = attach(hub, loop, labels=("gpu",))
with placer.claim(Resources(cpus=2), device="gpu", policy="require") as (
with placer.claim(Resources(cpus=2), label="gpu", policy="require") as (
target,
_,
):
@@ -120,7 +120,7 @@ def test_a_preferred_label_falls_back_here_and_is_still_accounted(loop):
hub = RemoteWorkerHub()
placer = placer_over(hub)
with placer.claim(Resources(cpus=3), device="gpu", policy="prefer") as (
with placer.claim(Resources(cpus=3), label="gpu", policy="prefer") as (
target,
allocation,
):