Use Grafana without a datasource

Grafana Plugin Development
1

Hi All,

I have a requirement to monitor Linux flavored servers without any standard monitoring tools and email the monitored values via email to any given point. Here I can run a script to gather data frequently in any format. For visualization purpose, I am thinking of using Grafana. Does anyone have experience in using Grafana without any standard data source or implement for a similar use case?

Thank you.

2

I can offer this solution

  1. Install any simple web-server or use self-made like this
  2. Collect metrics to files in web-server public folder. Store each metric as daily log-file e.g.
    CPU-201800512.log
    12:13 30
    12:14 55
    12:15 43
    …
  3. Next step (it’s can be hard to you): create own http-datasource plugin. This datasource must read current daily file by metric names, parse it, discard data outside the requested period and pass result to Grafana.

It’s a very bad solution. So I don’t want to help you and write this datasource :slight_smile:

Another way: collect all metrics to mysql/posgresql and read them native datasource.

3

Thank you for the valuable inputs.

Actually I was thinking of store data to a mysql database and read from there. But there is lack of documents in which format these data should store in a database in order to read from the tables it self.

4

You can put all metrics to one table stats with columns time, name and value.
Data can be formatted by sql query e.g. graph panel requires rows as [value, time] so the query will be
select value, time from stats where name = 'CPU' and ... (range condition)

5

Thank you aikon. I am able to implement a solution as you mentioned. Anyone else who has implemented the same?