From 38cedb172e42cbbeb8c12225baab827583db6c3b Mon Sep 17 00:00:00 2001 From: Patryk Bundyra <73306396+PBundyra@users.noreply.github.com> Date: Tue, 30 Jan 2024 14:01:13 +0100 Subject: [PATCH] Improve links (#1672) --- keps/1136-provisioning-request-support/README.md | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/keps/1136-provisioning-request-support/README.md b/keps/1136-provisioning-request-support/README.md index cb36a412f4..4555149632 100644 --- a/keps/1136-provisioning-request-support/README.md +++ b/keps/1136-provisioning-request-support/README.md @@ -22,8 +22,8 @@ ## Summary -Introduce an (AdmissionCheck)[https://github.com/kubernetes-sigs/kueue/tree/main/keps/993-two-phase-admission] -that will use (`ProvisioningRequest`)[https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/proposals/provisioning-request.md] +Introduce an [AdmissionCheck](https://github.com/kubernetes-sigs/kueue/tree/main/keps/993-two-phase-admission) +that will use [`ProvisioningRequest`](https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/proposals/provisioning-request.md) to ensure that there is enough capacity in the cluster before admitting a workload. @@ -33,7 +33,7 @@ Currently Kueue admits workloads based on the quota check alone. This works reasonably well in most cases, but doesn't provide guarantee that an admitted workload will actually schedule in full in the cluster. With `ProvisioningRequest`, SIG-Autoscaling owned -(ClusterAutoscaler)[https://github.com/kubernetes/autoscaler/tree/master/cluster-autoscaler] +[ClusterAutoscaler](https://github.com/kubernetes/autoscaler/tree/master/cluster-autoscaler) opens a way for stronger (but still not hard-guaranteed) all-or-nothing scheduling in an autoscaled cloud environment. @@ -74,7 +74,7 @@ succeeds. ### Risks and Mitigations There doesn't seem to be much risks or mitigations. -(Two phase admission process)[https://github.com/kubernetes-sigs/kueue/tree/main/keps/993-two-phase-admission] +[Two phase admission process](https://github.com/kubernetes-sigs/kueue/tree/main/keps/993-two-phase-admission) was added specifically for use cases like this. ## Design Details @@ -98,7 +98,7 @@ will also need to watch `ProvisioningRequestConfigs`. * Watch all changes CA makes to `ProvisioningRequests`. If the `Provisioned` or `CapacityAvailable` condition is set to `True` then finish the `AdmissionCheck` with success (and propagate the information about `ProvisioningRequest` name to -workload pods - KEP #1145 under `"cluster-autoscaler.kubernetes.io/consume-provisioning-request"`. +workload pods - [KEP #1145](https://github.com/kubernetes-sigs/kueue/blob/main/keps/1145-additional-labels/kep.yaml) under `"cluster-autoscaler.kubernetes.io/consume-provisioning-request"`. If the `ProvisioningRequest` fails, fail the `AdmissionCheck`. * Watch the admission of the workload - if it is again suspended or finished,