In Prometheus I want to calculate the total average of this specific power metric (dell_hw_chassis_power_reading) within a period of time. For example, the last 2 hours.
I currently am using this query below:
sum(avg_over_time(dell_hw_chassis_power_reading[2h])) by (node,instance)
This query seems to only give me the latest value within that 2-hour interval specified in the query. It does not give me the total average for all scrapes of that metric within the 2-hours.
The query avg_over_time(dell_hw_chassis_power_reading[2h]) returns the average values for raw samples stored in Prometheus over the last 2 hours. If Prometheus contains multiple time series with the dell_hw_chassis_power_reading name, then the query would return independent averages per each matching time series.
If you want calculating the average over all the time series with the name dell_hw_chassis_power_reading, then the following query must be used:
sum(sum_over_time(dell_hw_chassis_power_reading[2h]))
/
sum(count_over_time(dell_hw_chassis_power_reading[2h]))
If you need the average per some label or a set of labels, then the following query must be used:
sum(sum_over_time(dell_hw_chassis_power_reading[2h])) by (labels)
/
sum(count_over_time(dell_hw_chassis_power_reading[2h])) by (labels)
Related
I usually do it in Excel but it is not easy for me to do it in KIBANA as well
I have this table in Excel and every hour I want to average for all instancs in the fiels "detail" but excluding the lowest three values (nine details each hour, the average should be only for the the six highest of them). In Excel I use the LARGE function.
https://docs.google.com/spreadsheets/d/1LcKO8TGl49dz6usWNwxRx0oVgQb9s_h1/edit?usp=sharing&ouid=114168049607741321864&rtpof=true&sd=true
In your opinion is there any chance to do it directly in KIBANA?
No idea how to proceed
You can use lens table visualization and set the number of rows to 6 and order rows by descending order of your CPU load. Look at the sample data table here
The average here is calculated for the top 6 values of bytes only.
Here are the settings:
You can try replacing the clientIP here by details and bytes by CPU load
No, it is not possible to automatically remove the last N results from the equation in Kibana. You should be manually filtering out from the list in the visualization every time.
The only alternative I see is to add an extra step that deletes or flags the 3 results per hour you want to exclude, and then in Kibana you just add a regular filter.
The easiest way I can think of is creating a watcher that groups the results by hour, sort by CPU, and then ingest the first 6 results in a different index you can query using Kibana.
Docs: https://www.elastic.co/guide/en/elasticsearch/reference/current/xpack-alerting.html
If this acceptable for you I can edit this answer with more details about the Watcher I would create.
Be a subscription video streaming service. There are 3 subscription modalities, A, B, C. You want to calculate the average duration of the modality that has the highest number of subscribers per month.
The next chart show the amount of users by the type of suscription modality.
I need the expresion to generate a chart that shows the duration of the mode with the highest number of users per month.
I have the following fictitious data:
If you are using Month as a dimension on a chart, you can use the expression:
=FirstSortedValue(
Aggr(Avg(Duration), Msuscription, Month),
-Aggr(Count(user), Msuscription, Month)
)
This is sorting by the expression you used to create the chart in your question, and using - returns the highest value.
In a Graphana dashboard with several datapoints, how can I get the difference between the last value and the previouse one for the same metric?
Perhaps the tricky part is that the tiem between 2 datapoins for the same metric is not know.
so the desired result is the <metric>.$current_value - <metric>.$previouse_value for each point in the metricstring.
Edit:
The metrics are stored in graphite/Carbon DB.
thanks
You need to use the derivative function
This is the opposite of the integral function. This is useful for taking a running total metric and calculating the delta between subsequent data points.
This function does not normalize for periods of time, as a true derivative would. Instead see the perSecond() function to calculate a rate of change over time.
Together with the keepLastValue
Takes one metric or a wildcard seriesList, and optionally a limit to the number of ‘None’ values to skip over.
Continues the line with the last received value when gaps (‘None’ values) appear in your data, rather than breaking your line.
Like this
derivative(keepLastValue(your_mteric))
A good example can be found here http://www.perehospital.cat/blog/graphite-getting-derivative-to-work-with-empty-data-points
We have a counter metric in one our micro services which pushes data to DataDog. I want to display the total count for given time frame, and also the count per day (X axis would have the date and Y axis would have count). How do we achive this?
I tried using sum by and diff with Query value representation. It gives the total number of the count for given time frame. But I would like to get a bar graph with the X axis as the date and the Y axis as the count. Is this possible in DataDog?
It seems like there are 2 main questions here:
display the total count for a given time frame.
the count per day.
I think the rollup method is going to be your friend for both questions.
For #1 you need to pass in the time frame you want a total over: sum:<metric_name>.rollup(sum, <time_frame>) and the single value can be displayed using the Query Value visualization.
For #2 the datadog docs say you can get metrics per a day by
graphed using a day-long rollup with .rollup(avg,86400)
So this would look something like sum:<metric_name>.rollup(sum, 86400) and can be displayed a Timeseries with bars.
I am ingesting data to elasticsearch using flume, I want to create a time-series graph in kibana to show the events collected over time. BUT I also want to to the average per that time unit so the user knows if the current flow is around the average or not.
To create a timeline I am using line graph with #timestamp as X-axis and count as Y-axis.
The question is how to create the average line? and how to make this average dynamic e.g. as we zoom in average changes from average per day to average per hour.
While creating a visualization you can choose the type of y-axis metric. The default is "count". You can click on the icon to choose other type of metrics you want. It will have various options like average, sum, percentile etc.
As for the time range of average calculation, the the x-axis metrics, under buckets when you choose date histogram the default interval is auto.This means that the time range of average will chage automatically depending on overall time range selected.
You can change it to a fixed interval like per second, minute, hourly daily etc.
It's a bit odd, you would expect count to appear alongside field as something you can average. In reality you have to do it another way:
For the Y axis, instead of selecting count, select "Average Bucket"
then set up your bucket aggregation that you would like e.g. Date Histogram with second interval.
Below this you have another box for metric, e.g. the thing you're averaging, set this to count