You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@@ -595,7 +595,7 @@ _This section must be completed when targeting alpha to a release._
595
595
596
596
###### How can this feature be enabled / disabled in a live cluster?
597
597
598
-
-[] Feature gate (also fill in values in `kep.yaml`)
598
+
-[X] Feature gate (also fill in values in `kep.yaml`)
599
599
- Feature gate name: MemoryManager
600
600
- Components depending on the feature gate: kubelet
601
601
- Will enabling / disabling the feature require downtime of the control
@@ -653,8 +653,9 @@ fields of API types, flags, etc.?**
653
653
_This section must be completed when targeting beta graduation to a release._
654
654
655
655
###### How can an operator determine if the feature is in use by workloads?
656
-
657
-
The memory manager data will be available under pod resources API. When it configured with the static policy
656
+
In order for workloads to request exclusive memory allocation, the pod QoS must be "guaranteed".
657
+
The memory manager data will be available under pod resources API.
658
+
When it is configured with the static policy,
658
659
you will see memory related data during call to the pod resources API List method under the container.
659
660
660
661
###### How can someone using this feature know that it is working for their instance?
@@ -666,14 +667,13 @@ _This section must be completed when targeting beta graduation to a release._
666
667
667
668
*For a pod author:*
668
669
669
-
* Pod succeeded to start. You have two options to verify if containers pinned to the NUMA node
670
+
* Pod succeeded to start. You have two options to verify if containers are pinned to the NUMA node
670
671
- Via pod resources API, you will need to connect to grpc socket and get information from it, see [pod resource API doc page](https://kubernetes.io/docs/concepts/extend-kubernetes/compute-storage-net/device-plugins/#monitoring-device-plugin-resources) for more information.
671
672
- Checking the relevant container CGroup under the node.
672
673
673
674
* Pod failed to start because of the admission error.
674
675
675
-
To understand the reason you will need to check via pod resources API
676
-
the amount of allocatable memory and memory reserved by containers.
676
+
To understand the reason, you will need to check via pod resources API the amount of allocatable memory and memory reserved by containers.
677
677
678
678
###### What are the reasonable SLOs (Service Level Objectives) for the enhancement?
679
679
This does not seem relevant to this feature.
@@ -750,6 +750,8 @@ Are there any tests that were run/should be run to understand performance charac
750
750
and validate the declared limits?
751
751
-->
752
752
753
+
No.
754
+
753
755
### Troubleshooting
754
756
755
757
The Troubleshooting section currently serves the `Playbook` role. We may consider
0 commit comments