Explanation of CPU Time graph in admin console

CPU Time graph is looking like below

Can some one explain me what does User CPU Time 628 ms denotes in the above image? Is it that in last 1 min cockroach needed 628 CPU seconds?

Anyone any thoughts on this one?

Hi tuk!

Your hunch is correct, that chart is displaying CPU seconds used per second. Depending on your selection at the top of the time series page, it will show either the aggregated (summed) CPU time used across the cluster or CPU time used on a specific node. Also, note that the chart shows a single data point for each 10 second increment. CPU time is averaged across this 10 second window for each data point. This scales with the number of cores on a machine so a 4 core machine at 100% utilization would report 4s of CPU time used per second.

For the 2.1 release of CRDB in October, we’ll be revisiting this chart to give a better sense of actual CPU utilization (#24205). Since you were asking, I’d be curious to know what your expectation for the chart was (did you expect it to report CPUs/s?) and what would be most useful to you to get a sense of cluster or node performance. I’d also be interested to hear whether you checked the tooltip on the chart (the small icon next to the chart title) and if the description was unclear. End user feedback is always appreciated!

Thanks @psingh it is clear now.

If you can make the INFO tooltip more explanatory like the way you have mentioned here that will be more helpful I feel.