Skip to content

Latest commit

 

History

History
168 lines (112 loc) · 8.16 KB

lab1.md

File metadata and controls

168 lines (112 loc) · 8.16 KB

Overview

This is an simple lab that builds a minimal dashboard showing data in a Table Format.

This labs assumes you are familiar with the Demo Environment

Access

The Demo environment is hosted in a DMZ and ca be accessed with the following credentials:

https://64.104.255.11/

user: guest password: guest

The guest user is able to modify the dashboards and run Explore queries however it can't save any of the configuration changes.

Recreate the ACI Faults Dashboard

This dashboard is a 1:1 copy of the faults that are present inside ACI. The main advantages compared to looking at the faults in the ACI UI are:

  • the ability to aggregating Faults from Multiple Fabrics in a single table
  • allowing advanced sorting and filtering

faults

By using the Fabric drop down menu you can select different Fabrics (or All) and you can use the Colum headers to filter/sort the data:

This is a good dashboard to understand how Grafana dashboards are built, so let's re-built the Fault By Last Transition table.

Note: In this example we are focusing on Grafana Dashboard, someone configured the aci-exporter and Prometheus to populate the aci_faults with data. If you wanna learn on how to configure the aci-exporter and Prometheus to work together you can check out the development guide.

Dashboard Editing

Warning: Since this is an environment open to the internet I have not allowed used to save any config changes so DO NOT close or reload the browser or you will loose your work !

  • Select Dashboards --> Tests --> Dashboard Test 1 --> Move your move over the empty dashboard --> press e on the keyboard. This should open up the editing mode.
  • Ensure that you have:
    • Prometheus selected as Data Source
    • Selected the Builder mode this is a good way to learn but we will also look a the code afterwards.

alt text

  • From the Metric Drop Down menu select aci_faults and click Run Query this will display a Graph, in the legend you can see that for each metric we have infos about the Fabric, cause, description etc... the Metric value itself (the 1.7Bil) is the Unix Time Stamp of the last transition time the fault.

alt text However this is not a very good visualization for this type of data, we can see interesting data in the legend but a time series is really not the right visualization as we are interested in a list of faults aka a table!

To switch to a Table view we need two steps:

  • Select the Table Format for our query: Go to Options --> Format --> Select Table

- Select the `Table` from the Visualization drop down Menu by clicking on `Time Series` and then picking `Table` (take a moment to see how many options there are here) 
  <img src=images/lab1/Visualization.png width="500">
  • With just these two simple changes the data should look much better already however:
    • The Time and created column are not the last transition time for the Fault but when the fault was first received in Prometheus and for our use case this is useless.
    • The table contains a few "useless" column that would be nice to hide
    • The Value (last column) that represent the last transition time for our Fault is a long number, not a date

To solve all these issue we need to manipulate our data, in this example we are going to use there 3 Grafana transformations:

  • Organize fields by name: This will allows us to rename, re-order and hide the table columns:
  • Convert field type: This will allow us to convert the Value from a Unix Time Stamp to an actual human readable data
  • Sort by: To sort our Events by Last Transition time i.e. Value
  • click on the Transform Data tab and select Add Transformation

Organize fields by name:

Select Transform Data --> Add Transformation --> Organize fields by name alt text Here you can

  • Change the ordering of the fields, by drag them by the vertical dots on the left
  • Hide them, by clicking on the eye symbol
  • Rename them by adding text in the empty box on the right of the field name

You are free to sort things as you please but I would recommend to at least:

  • Hide:
    • Time
    • aci
    • created
    • instance
    • job
  • Rename:
    • Values to Last Transition
    • Place Last Transition as first item in the table

Convert field type:

Select Add another transformation --> Convert field type

  • Field: Last Transition
  • Type: Time

If you have placed the Last Transition as the first colum you should now see dates but you probably also notice that are not quite right as they show 1970. This is due to the fact that the epoch is expected in milliseconds since 1970 but what we are getting is just seconds, we will fix this after for now ignore this.

Sort by:

Select Add another transformation --> Sort By

  • Field: Last Transition
  • Reverse: Enabled

Depending on how you have Organize the fields our table should look something like this:

alt text

Fix the Last Transition timestamp:

All we have to do is multiple the Last Transition (aka the Value of our Metric) by 1000

  • Click on Query --> + Operations
  • In the Search tab enter multiply and select Multiply by scalar alt text
  • Set the Value: to 1000
  • Click Run Queries

Now the time should be reflected correctly.

Switch to Code

The query Builder is a great tool to learn but as you star building more complex queries it will become too cumbersome to use and some advanced capabilities are also not available so is a good idea to also learn the PromQL syntax. Try to click on Code and you should see that the same expression can be written as:

aci_faults * 1000

Filter by Fabric

We will do this steps in the Code mode to learn a bit more. PromQL support filtering your queriers by the labels, this is super easy, just open a { after the metric name and you should see a dropdown menu with all our labels!

If you want for example to show Faults only from site1 you can type the following query aci_faults{fabric="site1"} * 1000 and now only faults from site1 should appear. If you want to filter by using Regular Expressions (RegEx) you can replace = with =~ we will use this syntax in the next task

Filter by Fabric with Dashboards Variables

Variables in Grafana allow you to create dynamic and interactive dashboards by enabling you to define placeholders that can be replaced with different values at runtime.

If you are still on the dashboard editing pane let's modify our query to look like this:

aci_faults{fabric=~"$fabric"} * 1000

The fabric=~"$fabric" part simply tells Grafana to use the variable $fabric in this filter and the =~ also allow us to treat this filtering expression as a RegEx so that we can select 1 or more fabrics at the same time.

Click apply, this will result in an empty dashboard, this is expected since the variable $fabric does not exists yet!

To create the $fabric variables to select our sites follow these steps:

  • Click on the gear icon (settings in the top right) and select "Variables."
  • Click "New variable"
    • Select variable type: Query
    • Name: fabric
    • Display Name: Fabric
    • Show on dashboard: Labels and Values
    • Data source: Prometheus
    • Query
      • Query type: Labels Values
      • Label: fabric Warning select the fabric label DO NOT select $fabric
    • Selection options: Enabled Multi-Values and Include All option
  • Click Apply
  • Click Close
  • If the Dashboard is still empty click the refresh button on the Top Right (the two spinning arrows)

Now your dashboard will have a new drop down menu where you can dynamically select the fabric to display!

The End

This concludes Lab1, feel free to play around more with this dashboard if you want or you can proceed to Lab2