william-wang comments

Results 100 comments of


                                            william-wang

"pvc not found" message converted to "gang unschedulable" message

@hwdef 👍

quick termination of jobs with hdrf queues

@Robert-Christensen-visa Thanks for your reporting. I will try to reproduce the issue with your step.

quick termination of jobs with hdrf queues

@Thor-wl Have you reproduced this issue in your environment?

Tune default weights for nodeOrder plugin

Thanks for your pr. please make the ci happy.

podgroup not created if OwnerReferencesPermissionEnforcement enabled in Kubernetes

@Robert-Christensen-visa Totally agree with you. We need consider crd like spark-operator/flink-operator with all basic resource as well. Volcano is supposed to support arbitrary pod scheduling in the cluster.

Add the allocation policy for deserved in scenario guarantee

@jinzhejz please help to review this plugin, thanks :)

Increase the default weights of several arguments in nodeorder plugin

Thanks for your report. It seems reasonable to do this change. > The risk is it's a break change which will result in the different scheduling results, but I think...

volcano support checkpoint

May I know which kind of computing engine e.g spark are you using? Usually the chickpoint function is implemented in upper layer framework, e.g pytorch.

The pods were deleted but the resource of job and podgroup still there

@YuxiangJohn Thanks for your reporting. I did not reproduce your issue based on the description. Would you mind providing the repreduce steps?

Is it possible to support application mode in this operator?

I am interested with this too. Per-job mode is not supported in flink1.12 any more. The existing operator users want to use the task-manager pod scaling up and down feature...