Hi,
Today i was watching the dashboard for checking HTTP requests and i found out that the graphs of the Hours interval was different with the ones of Days. I put three images below showing the discrepancy. Can someone tell me why i have different datas?
The first one is the graph of Hours that shows a peak of requests. In the second and third screenshot that peak disappeared in the days graph. The third one is a zoom in view of the second.
Edit: Maybe the data is spread over a wider time and therefore doesn’t add up? In this way, however, possible peaks are lost. Is there a way to have the detailed graph for a longer period?
Edit 2: However, the data shows me an average of 10/12 requests in that period of time and the two peaks of 500 and 50 in the hour graph. However, what doesn’t add up is that in the graph of days the value remains an average of 12.
The graph is requests on the management port (usually 8091).
If you have an unexpected spike, look for a client that is attempting to connect, but is failing. The SDK will try hard, long and quickly to connect. Even if it is due to a bad username/password.
The first one is the graph of Hours that shows a peak of requests. In the second and third screenshot that peak disappeared in the days graph.
That’s expected, right? 500 requests in one minute would be 500/60 = 8.3 request/hour. (simliar with requests/second vs requests/minute)
I believe the stats come from here. There is an rest_requests section.
Thanks for the reply.
Yes, in fact, looking back on it, the values ​​spread over a longer period could be correct. At this point, however, I wonder if, for example, one really wanted to see these sporadic peaks over a longer period where they could be seen.
However, the peak was linked to a node where some indexes reside so I don’t think it’s an access problem.
