Tomas Tormo

Results 36 comments of Tomas Tormo

> Do we know if Kueue requires to have this field mutable for suspended TrainJobs? Yes, before resuming a job, [Kueue injects its scheduling info via `PodTemplateOverrides`](https://github.com/kubernetes-sigs/kueue/blob/e93103571c0fe37d09ccae39a36f8212fa2cdd35/pkg/controller/jobs/trainjob/trainjob_controller.go#L290). With this change,...

> [@tenzen-y](https://github.com/tenzen-y) [@astefanutti](https://github.com/astefanutti) [@kaisoz](https://github.com/kaisoz) [@mimowo](https://github.com/mimowo) Do we have any Kueue E2Es test that verifies TrainJob preemption and usage of `suspend` API? Unfortunately, all Kueue E2E tests start from an already...

We found that the TrainJob admission webhook rejects TrainJobs with more than one `PodSpecOverride` targeting the same job. This effectively breaks the integration if the user supplied TrainJob already contains...

> We found that the TrainJob admission webhook rejects TrainJobs with more than one `PodSpecOverride` targeting the same job. This effectively breaks the integration if the user supplied TrainJob already...

> Actually, now I wonder maybe this is not necessarily needed if we could introspect the field manager by the TrainJob .`ManagedFields`. This should generally keep the actor already, but...

> As I synced with @kaisoz he was looking into this already and apparently when using +listType=map it requires specifying +listMapKey, which uniquely identifies each element of the list. However,...