[Data] "Task Completion Time Without Backpressure" Grafana panel is broken

[bveeramani]

The "Task Completion Time Without Backpressure" metrics chart is broken.

I think it's because #57788 renamed task_completion_time_without_backpressure without updating the associated code in data_dashboard_panels.py:

ray/python/ray/dashboard/modules/metrics/dashboards/data_dashboard_panels.py

Lines 680 to 694 in 395e60e

	TASK_COMPLETION_TIME_WITHOUT_BACKPRESSURE_PANEL = Panel(
	id=40,
	title="Task Completion Time Without Backpressure",
	description="Average time (in seconds) tasks spend executing their core logic, excluding backpressure, over a recent 5-minute window. This metric helps isolate actual computation time from delays caused by data flow bottlenecks, aiding in differentiating between computation-bound and data-flow-bound performance issues.",
	unit="s",
	targets=[
	Target(
	expr='increase(ray_data_task_completion_time_without_backpressure{{{global_filters}, operator=~"$Operator"}}[5m]) / increase(ray_data_num_tasks_finished{{{global_filters}, operator=~"$Operator"}}[5m])',
	legend="Task Completion Time w/o Backpressure: {{dataset}}, {{operator}}",
	),
	],
	fill=0,
	stack=False,
	)

What you need to do:

• Update the Grafana panel definition to use the correct name in the expr
• Run a simple Ray Data job and ensure the panel works correctly.
  • You'll need to setup Grafana and Prometheus to do this. Anecdotally, this can be difficult to set up for the first time. See https://docs.ray.io/en/latest/cluster/metrics.html for instructions.

[bveeramani]

(@kriyanshii will tentatively work on this)

[kriyanshii]

Will submit the pr soon.