Add metric for data availability wait time #13534
Conversation
| dataAvabilityWaitedTime = promauto.NewSummary(prometheus.SummaryOpts{ | ||
| Name: "da_waited_time_milliseconds", | ||
| Help: "Total time to wait for a data availability check in ReceiveBlock(), a node is blocked during this time", | ||
| }) |
There was a problem hiding this comment.
suggest “total time spent waiting”. also do we need the comment about the node blocking in the metric description? i’m sure that fact is true about a number of other metrics.
There was a problem hiding this comment.
"total time spent waiting" is a bit vague. It could be waiting for a lot of things. How about "total time spent waiting da"?
There was a problem hiding this comment.
That's basically what I meant - replace "Total time to wait for a data availability check", with "Total time spent waiting for a data availability check".
| @@ -209,6 +209,7 @@ func (s *Service) ReceiveBlock(ctx context.Context, block interfaces.ReadOnlySig | |||
|
|
|||
| timeWithoutDaWait := time.Since(receivedTime) - daWaitedTime | |||
| chainServiceProcessingTime.Observe(float64(timeWithoutDaWait.Milliseconds())) | |||
| dataAvabilityWaitedTime.Observe(float64(daWaitedTime.Milliseconds())) | |||
There was a problem hiding this comment.
There are some error returns between this metrics update and where daWaitedTime is set. I think we should call Observe right after line 124.
This PR introduces a valuable metric: data availability wait time. This metric measures the duration a node is blocked at
ReceiveBlock, making it crucial to monitor.