mirror of
https://github.com/NVIDIA/OpenShell.git
synced 2026-10-31 07:44:57 +08:00
DiveshandDrew Newberry
44ef984b58
* fix(supervisor): guard shared supervisor pool configuration Raise the gateway's minimum supervisor pool recovery timeout from 10 to 30 seconds. A failed pool member takes up to about 24 seconds to detect, so a shorter deadline ended workloads another member would have adopted. The Sandbox Protocol range stays 10 to 300 seconds for compatibility. Refuse an [openshell.gateway.supervisor_pool] table during config preflight, and at startup before the compute driver is built, unless the selected driver hosts sandboxes on the shared supervisor pool. A new ComputeDriverFactory::uses_shared_supervision hook reports this; the Kubernetes factory returns true with supervisor_topology = "gateway". Widen the readiness deadline of a shared-startup supervisor test that timed out under full-workspace load. Signed-off-by: divesh <dgude@nvidia.com> * fix(gateway): require driver selection for pool preflight Signed-off-by: Drew Newberry <anewberry@nvidia.com> * docs(gateway): remove pool preflight note Signed-off-by: Drew Newberry <anewberry@nvidia.com> --------- Signed-off-by: divesh <dgude@nvidia.com> Signed-off-by: Drew Newberry <anewberry@nvidia.com> Co-authored-by: Drew Newberry <anewberry@nvidia.com>