Metrics dashboards not working as it should

[Francisco_Gomez1 (Francisco Gomez)]

Hi there.
Im using elastic agent to load all my system metrics in the dashboards given by kibana, the problem is that when i use the default metrics recollection time (10 seconds) it works well, giving me a perfect view:

But when i change the recollection times for some metrics ( 1minute becouse 10 seconds is too low to be viable for a storage perspective)

i get this view :

Any idea on how i can fix this?

[stephenb (Stephen Brown)]

Hi @Francisco_Gomez1

Set your collection interval to 60s and collect 10 min of data and then time picker to Last 1hr <!-- Most important and let me know what you see. It should like below. Then I will explain what is happening and how to fix it.

[Francisco_Gomez1 (Francisco Gomez)]

Hi, it indeed helped

But check this out, the top process looks like this for some reason:

Also (i got 3 servers) in one of the servers my disk information looks like this

When i try to modify the interval i see this

This is the edit mode, looks like it should be looking like this

Also it dosent let me modify the interval for the disk info

Any ideas ?

[stephenb (Stephen Brown)]

Hi @Francisco_Gomez1

[Skipping long explanation of collection intervals and TSVB intervals etc ]

In short for TSVB visualization to work (well/best) when the TSVB interval is >= to the Metric Collection interval which you have set to 1m

First / Overall Issue:
You have chosen 1m collection interval which is totally fine.

When you you set the timepicker to 15mins, TSVB auto interval is set to 10s, therefor there is no data in the 2nd to last bucket (because we drop the last) for the gauge to work. When we set it to 1 hour the auto interval pick 60s / 1m

The solution is to set the TSVB interval to >=1m for the visualizations for a metrics collection rate of 1m. Today you will need to go in and fix them. If you do then they should work fine.

You should go in and fix that for all the visualizations, Including the Line / Graph and Top N.

With respect to the Top Processes (All Top N) set to interval to >=1m pretty sure there may still be an issue with that. Report Back

With respect to Disk Used. The Data Time Range more is set to Entire time range so that means that value represents the Average of the Entire time range of the Time Picker. 15min or 1 Hour etc. Because it is over the entire time range the interval is no longer relevant.

You can leave is this way if you want, Me I like to know the disk used recently so I set that back to Last value and set the interval to >=1m

Hope that helps!

[Francisco_Gomez1 (Francisco Gomez)]

Indeed disk is working well making those changes, about Top N i realized i set that one previusly on 5 minutes, so i change it back to 1 minute like the others to see if it is working and its looking like this

The timepicker is set in the last 5 minutes, and this is my panel configuration, even if change to 1 hour no differences

[stephenb (Stephen Brown)]

First when you try to squeeze down time when you have a low collection rate, the results are not going to be great... If you have 5 mins Timepicker and collect at 1 min and then throw out the last bucket you at max will have 4 data point (rare) at generally 3 ... Should still work but keep that in mind.

Yeah I think there is a bug with the Table I will look into it when I get a chance.... It works with entire time range but does not seems to be working with Last Value

[Francisco_Gomez1 (Francisco Gomez)]

@stephenb

t... If you have 5 mins Timepicker and collect at 1 min and then throw out the last bucket you at max will have 4 data point (rare) at generally 3 ... Shoul

Thanks in advance, i will be waiting for your response

[stephenb (Stephen Brown)]

No promises, as an open source company please feel free to replicate and write a bug here