We are using technical data coming from a transfer monitor for stream performance analysis.
We had developed a kind of ETL to extract and transform data, then load it in several influxDB databases and display in Grafana.
Unfortunately, we have a lot of evening and night tasks such as backups, server reboots, … which are fooling the results.
Indeed, I am using mean calculation in my panels and it gets fooled because of wrong data performance between 8:00 pm and 6:00 am.
I had already considered to get rid of this range of data in my transform scripts, because I have all the necessary stuff to do it. But I do think it is a dirty workaround, and looking for way to do it properly in Grafana.
I will have a look to your links. I don’t know if I will be able to use this in Grafana queries keeping my now-18d to now-1d/d range for end users.
I assume the best way to fix it is to add a flag for the 2 ranges in my transform scripts. This way, it will be easy to get rid of the 8:00PM to 6:00AM range just with a where clause.
Sounds much better than cloning datasources filtering the proper time range.
We have calculated those stats for months. I might be able to handle my existing data, adding this new field in my export format, and create a new datasource to import it. If it works flawlessly, changing the datasource in my panels and adding a where clause might do the job.
Thanks again for your help. Didn’t figure it out before