Open zhiyi57 opened 1 year ago
@Jeffwan FYI
@zhiyi57
The immediate fix is pretty trivial
What fix is this? Would you be willing to open a PR with the fix?
@zhiyi57
The immediate fix is pretty trivial
What fix is this? Would you be willing to open a PR with the fix?
The basic fix is to check whether this filled is set or not before use it. A better fix would be set default value on initialization. Not sure whether we would want a pr down on this road though.
Any updates of this issue?
Search before asking
KubeRay Component
ray-operator
What happened + What you expected to happen
When label RayJob with volcano scheduler and queue,ray operator crashes. Raycluster object would be created with no status. None of relevant cluster pod would be created. Rayjob is the sample job provided in examples. This happens in both v.0.6.0 and v.1.0.0 Tried same setup for RayCluster, everything works as expected.
The log from ray operator is the following:
Reproduction script
apiVersion: ray.io/v1alpha1 kind: RayJob metadata: name: rayjob-sample labels: ray.io/scheduler-name: volcano volcano.sh/queue-name: kuberay-test-queue spec: entrypoint: python /home/ray/samples/sample_code.py rayClusterSpec: rayVersion: '2.7.0' # should match the Ray version in the image of the containers
Ray head pod template
...
Anything else
The immediate fix is pretty trivial. Not sure whether further refactor is desired. No response
Are you willing to submit a PR?