[kuberay][autoscaler] Update KubeRay version to v1.0.0 #40918
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
kevin85421
commented
Nov 3, 2023
| @@ -79,7 +79,12 @@ def _get_ray_cr_config( | |||
| """ | |||
| with open(EXAMPLE_CLUSTER_PATH) as ray_cr_config_file: | |||
| ray_cr_config_str = ray_cr_config_file.read() | |||
| config = yaml.safe_load(ray_cr_config_str) | |||
|
|
|||
| kuberay_crd_sets = set(["RayCluster", "RayJob", "RayService"]) | |||
There was a problem hiding this comment.
A YAML file may have multiple K8s objects.
kevin85421
changed the title
kevin85421
force-pushed
the
update-kuberay-1.0.0-rc2
branch
from
3889a09
to
bec44b2
Compare
kevin85421
force-pushed
the
update-kuberay-1.0.0-rc2
branch
from
bec44b2
to
c8fbf90
Compare
kevin85421
requested review from
ericl,
architkulkarni and
a team
as code owners
architkulkarni
approved these changes
Nov 6, 2023
|
test_memory_pressure unrelated |
architkulkarni
added
the
tests-ok
ujjawal-khare
pushed a commit
to ujjawal-khare-27/ray
that referenced
this pull request
Nov 29, 2023
…0918) Create a YAML file ray-cluster.autoscaler-template.yaml for testing instead of using the YAML file ray-cluster.autoscaler.yaml in the KubeRay repository. Note that the tests assume that both head and worker Pods have exactly 1 CPU. Hence, if we set num-cpus: "0" in the head's rayStartParams, the current test logic would not work. Why do I remove the test for "Confirming that the operator and autoscaler ignore pods marked for termination"? KubeRay tries to ensure the number of runningPods, but the definition of runningPods is a bit different from different KubeRay versions. Definition 1: For KubeRay v0.6.0 and older, the definition of runningPods is the Pods that are running or pending and not terminating. Definition 2: For KubeRay v1.0.0, the definition of runningPods becomes that Pods that their Ray containers are not actually terminated. See [GCS FT] Consider the case of sidecar containers kuberay#1386 for more details. That is, in definition 1, KubeRay may create new Pods if some Pods are in the terminating process. Hence, it is possible to have more than maxReplicas Pods & Ray nodes from both Kubernetes and Ray perspectives. In definition 2, KubeRay only creates new Pods when the Ray nodes are actually terminated. --------- Signed-off-by: kaihsun <kaihsun@anyscale.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why are these changes needed?
Create a YAML file
ray-cluster.autoscaler-template.yamlfor testing instead of using the YAML fileray-cluster.autoscaler.yamlin the KubeRay repository.num-cpus: "0"in the head'srayStartParams, the current test logic would not work.Why do I remove the test for "Confirming that the operator and autoscaler ignore pods marked for termination"?
runningPods, but the definition ofrunningPodsis a bit different from different KubeRay versions.runningPodsis the Pods that are running or pending and not terminating.runningPodsbecomes that Pods that their Ray containers are not actually terminated. See [GCS FT] Consider the case of sidecar containers kuberay#1386 for more details.maxReplicasPods & Ray nodes from both Kubernetes and Ray perspectives. In definition 2, KubeRay only creates new Pods when the Ray nodes are actually terminated.Related issue number
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.