Plugins 〉Azure Data Explorer Datasource

Data Source
grafana

Azure Data Explorer Datasource

  • Overview
  • Installation
  • Change log
  • Related content

Azure Data Explorer data source for Grafana

Build Status

Azure Data Explorer is a log analytics cloud platform optimized for ad-hoc big data queries.

Installation

This plugin requires Grafana 8.0.0 or newer as of version 4.0.0 , it requires Grafana 7.1.0 or newer as of version 3.0.0. Plugin versions prior to 3.0.0 require Grafana 6.3.6.

Grafana Cloud

If you do not have a Grafana Cloud account, you can sign up for one here.

  1. Click on the Install plugin button on the Azure Data Explorer page on Grafana.com. This will automatically add the plugin to your Grafana instance. It might take up to 30 seconds to install.

  2. Login to your Hosted Grafana instance (go to your instances page in your profile): https://grafana.com/orgs/<yourUserName>/instances/ and the Azure Data Explorer datasource will be installed.

Installation Instructions on the Grafana Docs Site

Docker

  1. Fetch the latest version of grafana from Docker Hub: docker pull grafana/grafana:latest

  2. Run Grafana and install the Azure Data Explorer plugin with this command:

    docker run -d --name=grafana -p 3000:3000 -e "GF_INSTALL_PLUGINS=grafana-azure-data-explorer-datasource" grafana/grafana:latest
    
  3. Open the browser at: http://localhost:3000 or http://your-domain-name:3000

  4. Login in with username: admin and password: admin

  5. To make sure the plugin was installed, check the list of installed datasources. Click the Plugins item in the main menu. Both core datasources and installed datasources will appear.

This ia an alternative command if you want to run Grafana on a different port than the default 3000 port:

docker run -d --name=grafana -p 8081:8081 -e "GF_SERVER_HTTP_PORT=8081" -e "GF_INSTALL_PLUGINS=grafana-azure-data-explorer-datasource" grafana/grafana:master

It is recommended that you use a volume to save the Grafana data in. Otherwise if you remove the docker container, you will lose all your Grafana data (dashboards, users etc.). You can create a volume with the Docker Volume Driver for Azure File Storage.

Installing the Plugin on an Existing Grafana with the CLI

Grafana comes with a command line tool that can be used to install plugins.

  1. Upgrade Grafana to the latest version. Get that here.
  2. Run this command: grafana-cli plugins install grafana-azure-data-explorer-datasource
  3. Restart the Grafana server.
  4. Open the browser at: http://localhost:3000 or http://your-domain-name:3000
  5. Login in with a user that has admin rights. This is needed to create datasources.
  6. To make sure the plugin was installed, check the list of installed datasources. Click the Plugins item in the main menu. Both core datasources and installed datasources will appear.

Installing the Plugin Manually on an Existing Grafana

If the server where Grafana is installed has no access to the Grafana.com server, then the plugin can be downloaded and manually copied to the server.

  1. Upgrade Grafana to the latest version. Get that here.
  2. Get the zip file from Grafana.com: https://grafana.com/plugins/grafana-azure-data-explorer-datasource/installation and click on the link in step 1 (with this text: "Alternatively, you can manually download the .zip file")
  3. Extract the zip file into the data/plugins subdirectory for Grafana.
  4. Restart the Grafana server
  5. To make sure the plugin was installed, check the list of installed datasources. Click the Plugins item in the main menu. Both core datasources and installed datasources will appear.

Configuring the datasource in Grafana

The steps for configuring the integration between the Azure Data Explorer service and Grafana are:

  1. Create an Azure Active Directory (AAD) Application and AAD Service Principle.
  2. Log into the Azure Data Explorer WebExplorer and connect the AAD Application to an Azure Data Explorer database user.
  3. Use the AAD Application to configure the datasource connection in Grafana.

Creating an Azure Active Directory Service Principle

Follow the instructions in the guide to setting up an Azure Active Directory Application.

An alternative way to create an AAD application is with the Azure CLI:

az ad sp create-for-rbac -n "http://url.to.your.grafana:3000"

This should return the following:

{
  "appId": "XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX",
  "displayName": "azure-cli-2018-09-20-13-42-58",
  "name": "http://url.to.your.grafana:3000",
  "password": "XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX",
  "tenant": "XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX"
}

Assign the Reader role to the Service Principal and remove the Contributor role:

az role assignment create --assignee <your appId> --role Reader
az role assignment delete --assignee <your appId> --role Contributor

Connecting AAD with an Azure Data Explorer User

Navigate to the Azure Web UI for Azure Data Explorer: https://dataexplorer.azure.com/clusters/nameofyourcluster/databases/yourdatabasename

You can find the link to the Web UI in the Azure Portal by navigating to:

  1. All services-> Azure Data Explorer Clusters option
  2. Choose your cluster
  3. Databases -> click on your database
  4. Choose the Query option -> then click on the "Open in web UI" link

To create a cluster and database, follow the instructions here.

The AAD application that you created above needs to be given viewer access to your Azure Data Explorer database (in this example the database is called Grafana). This is done using the dot command add. The argument for .add contains both the client and tenant id separated by a semicolon:

.add database Grafana viewers ('aadapp=<your client id>;<your tenantid>')

A real example with a client/app id and tenant id:

.add database Grafana viewers ('aadapp=377a87d4-2cd3-44c0-b35a-8887a12fxxx;e7f3f661-a933-4b3f-8176-51c4f982exxx')

If the command succeeds you should get a result like this:

Azure Data Web Explorer Add result

Configuring Grafana

  1. Accessed from the Grafana main menu, newly installed datasources can be added immediately within the Data Sources section. Next, click the "Add datasource" button in the upper right.

  2. Select Azure Data Explorer Datasource from the datasource list:

    Data Source Type

  3. In the name field, a default name is filled in automatically but it can be changed to anything.

  4. You need 3 pieces of information from the Azure portal (see link above for detailed instructions):

    • Tenant Id (Azure Active Directory -> Properties -> Directory ID)
    • Client Id (Azure Active Directory -> App Registrations -> Choose your app -> Application ID)
    • Client Secret ( Azure Active Directory -> App Registrations -> Choose your app -> Keys)
  5. Paste these three items into the fields in the Azure Data Explorer API Details section: Azure Data Explorer API Details

  6. Click the Save & Test button. After a few seconds once Grafana has successfully connected then choose the default database and save again.

Configuring On-Behalf-Of authentication (Beta)

⚠️ This feature is in Beta and subject to breaking changes

For information about setting up and using the OBO flow: on-behalf-of documentation

Writing Queries

Queries are written in the new Kusto Query Language.

Queries can be formatted as Table, Time Series, or ADX Time Series data.

Table Queries

Table queries are mainly used in the Table panel and row a list of columns and rows. This example query returns rows with the 6 specified columns:

AzureActivity
| where $__timeFilter()
| project TimeGenerated, ResourceGroup, Category, OperationName, ActivityStatus, Caller
| order by TimeGenerated desc

Time Series Queries

Time Series queries are for the Graph Panel (and other panels like the Single Stat panel). The query must contain exactly one datetime column, one or more number valued columns, and optionally one more more string columns as labels. Here is an example query that returns the aggregated count grouped by the Category column and grouped by hour:

AzureActivity
| where $__timeFilter(TimeGenerated)
| summarize count() by Category, bin(TimeGenerated, 1h)
| order by TimeGenerated asc

The number valued columns are considered metrics and the optional string columns are treated as tags. A time series is returned for each value column + unique set of string column values. Each series has name of valueColumnName {stringColumnName=columnValue, ... }.

For example, the following query will produce series like AvgDirectDeaths {EventType=Excessive Heat, State=DELAWARE}``EventCount {EventType=Excessive Heat, State=NEW JERSEY}:

StormEvents
| where $__timeFilter(StartTime)
| summarize EventCount=count(), AvgDirectDeaths=avg(DeathsDirect) by EventType, State, bin(StartTime, $__timeInterval)
| order by StartTime asc

ADX Time Series Queries

ADX Time Series are for queries that use the Kusto make-series operator. The query must have exactly one datetime column named Timestamp and at least one value column. There may also optionally be string columns that will be labels.

Example:

let T = range Timestamp from $__timeFrom to ($__timeTo + -30m) step 1m
  | extend   Person = dynamic(["Torkel", "Daniel", "Kyle", "Sofia"])
  | extend   Place  = dynamic(["EU",     "EU",     "US",   "EU"])
  | mvexpand Person, Place
  | extend   HatInventory = rand(5)
  | project  Timestamp, tostring(Person), tostring(Place), HatInventory;

T | make-series AvgHatInventory=avg(HatInventory) default=double(null) on Timestamp from $__timeFrom to $__timeTo step 1m by Person, Place | extend series_decompose_forecast(AvgHatInventory, 30) | project-away *residual, *baseline, *seasonal

Time Macros

To make writing queries easier there are some Grafana macros that can be used in the where clause of a query:

  • $__timeFilter() - Expands to TimeGenerated ≥ datetime(2018-06-05T18:09:58.907Z) and TimeGenerated ≤ datetime(2018-06-05T20:09:58.907Z) where the from and to datetimes are taken from the Grafana time picker.
  • $__timeFilter(datetimeColumn) - Expands to datetimeColumn ≥ datetime(2018-06-05T18:09:58.907Z) and datetimeColumn ≤ datetime(2018-06-05T20:09:58.907Z) where the from and to datetimes are taken from the Grafana time picker.
  • $__timeFrom - Expands to datetime(2018-06-05T18:09:58.907Z), the start time of the query.
  • $__timeTo - expands to datetime(2018-06-05T20:09:58.907Z), the end time of the query.
  • $__timeInterval - expands to 5000ms, Grafana's recommended bin size based on the timespan of the query, in milliseconds. In alerting this will always be 1000ms, it is recommended not to use this macro in alert queries.

Templating Macros

  • $__escapeMulti($myVar) - is to be used with multi-value template variables that contains illegal characters. If $myVar has the value '\\grafana-vm\Network(eth0)\Total','\\hello!', it expands to: @'\\grafana-vm\Network(eth0)\Total', @'\\hello!'. If using single value variables there no need for this macro, simply escape the variable inline instead - @'\$myVar'

  • $__contains(colName, $myVar) - is to be used with multi-value template variables. If $myVar has the value 'value1','value2', it expands to: colName in ('value1','value2').

    If using the All option, then check the Include All Option checkbox and in the Custom all value field type in the following value: all. If $myVar has value all then the macro will instead expand to 1 == 1. For template variables with a lot of options, this will increase the query performance by not building a large where..in clause.

Templating with Variables

Instead of hard-coding things like server, application and sensor name in your metric queries you can use variables in their place. Variables are shown as dropdown select boxes at the top of the dashboard. These dropdowns make it easy to change the data being displayed in your dashboard.

Create the variable in the dashboard settings. Usually you will need to write a query in the Kusto Query Language to get a list of values for the dropdown. It is however also possible to have a list of hard-coded values.

  1. Fill in a name for your variable. The Name field is the name of the variable. There is also a Label field for the friendly name.

  2. In the Query Options section, choose the Azure Data Explorer datasource in the Data source dropdown.

  3. Write the query in the Query field. Use project to specify one column - the result should be a list of string values.

    Template Query

  4. At the bottom, you will see a preview of the values returned from the query:

    Template Query Preview

  5. Use the variable in your query (in this case the variable is named level):

    MyLogs | where Level == '$level'
    

    For variables where multiple values are allowed then use the in operator instead:

    MyLogs | where Level in ($level)
    

Read more about templating and variables in the Grafana documentation.

Note: Usage of template variables in the Builder is currently not supported.

Databases Variable

There is no way to fetch a list of databases with the Kusto query language. When creating a template variable as described in the Templating with variables section, use the following function in the Query field to return a list of databases:

databases()

This variable can be used in the databases dropdown. This gives you the ability to switch databases without editing the query in the panel.

To use the variable, type the name of your variable into the dropdown. For example, if the name of your variable is database, then type $database.

Annotations

An annotation is an event that is overlaid on top of graphs. The query can have up to three columns per row, the datetime column is mandatory. Annotation rendering is expensive so it is important to limit the number of rows returned.

  • column with the datetime type.
  • column with alias: Text or text for the annotation text
  • column with alias: Tags or tags for annotation tags. This should return a comma separated string of tags e.g. 'tag1,tag2'

Example query:

MyLogs
| where $__timeFilter(Timestamp)
| project Timestamp, Text=Message , Tags="tag1,tag2"

Query Builder - Data Types

<!-- TODO: Update the paragraph below once #353 is fixed -->

The query builder provides an easy to use interface to query Azure Data Explorer. However, there are limitations on the supported data types that a column can possess. Currently, if a column is typed as dynamic it is fully not included as an option for the following operations: Where, Aggregate, Group by. The reason for this is that columns of type dynamic can potentially contain values that have any of the primitive data types, but also arrays (where the array can then have values of any type) and JSON objects. The query builder does not currently support querying values that are either arrays or JSON objects.

Note that only the 50.000 first rows of a table are evaluated in order to obtain possible values to show as options in the query builder. Additional values can be manually written in the different selectors if they don't appear by default.

See the below documentation for further details on how to handle dynamic columns appropriately via the KQL editor.

Kusto Data Types - Documentation on data types supported by Kusto.

Dynamic Data Type - Detailed documentation on the dynamic data type.

CHANGELOG

See the Changelog.

Installing Azure Data Explorer Datasource on Grafana Cloud:

For more information, visit the docs on plugin installation.

Change Log

[4.1.10]

  • Fix: Invalid code editor loaded for Grafana versions that don't follow semantic versioning by @aangelisc in https://github.com/grafana/azure-data-explorer-datasource/pull/506
  • Fix error unmounting ADX query editor by @aangelisc in https://github.com/grafana/azure-data-explorer-datasource/pull/519
  • Security: Upgrade Go in build process to 1.19.3

[4.1.9]

  • Security: Upgrade Go in build process to 1.19.2
  • Fix: Schema mapping displaying macro functions

[4.1.8]

  • Report interaction on dashboard load for feature tracking

[4.1.7]

  • Fix crash when creating Alerts
  • Autocomplete now works with dynamic values
  • Fix template variables for values containing parentheses

[4.1.6]

  • Change default logic for dynamic columns: Cast to double if type exists in schema

[4.1.5]

  • Fix: Update table in the KQL expression when changing the database.

[4.1.4]

  • Change the default format to table data to avoid accidental high consumption of memory.
  • Fix: Quote columns with spaces in the query builder.

[4.1.3]

  • Fix: Correctly cast dynamically typed columns in the query builder.

[4.1.2]

This release include several bug fixes:

  • Fix reload schema button in configuration.
  • Fix dynamic resolution for simple types in the query builder.
  • Fix "Aggregate" and "Group by" removal logic also for the builder.
  • Return configured default database instead of the first one.

[4.1.1]

Several bug fixes for the visual query builder:

  • Add materialized views as tables.
  • Fix template variable quoting.
  • Fix syntax dynamic fields with multiple types.

[4.1.0]

  • New Feature: The visual query editor now supports dynamic columns. This includes columns with one or more arrays of dynamic values.

[4.0.2]

  • Breaking Change on Beta feature: On-Behalf-Of flow is now disabled by default

[4.0.1]

  • Bugfix: Remove custom token cache used for On-Behalf-Of flow (Beta) and rely on Microsoft Authentication Library to keep a local cache.

[4.0.0]

  • Breaking Change: Azure Data Explorer plugin now requires Grafana 8.0+ to run.
  • Breaking Change: obo_latency_seconds metric was removed.
  • Bugfix: Included new Kusto query editor. NOTE: This new editor will be only available if used with Grafana 8.5 or later. Fixes #325.
  • Bugfix: Filter dynamic columns from Where/Aggregate/Group by clauses to prevent syntax errors.
  • Bugfix: Add logical operators for timespan types in the query builder.
  • Internal: Client secret authentication via Grafana Azure SDK.
  • Internal: OBO authentication via MSAL for Go.

[3.7.1]

  • Bugfix: Fix scope for national clouds

[3.7.0]

  • Chore: Added test coverage script

[3.7.0-beta1]

  • Feature: Add On-Behalf-Of Token Authorization
  • Bugfix: Eliminate Client ID Panic
  • Bugfix: Append azure error to query unsuccessful message
  • Bugfix: Fix macro regex on columns with hyphens
  • Internal: Update plugin dependencies

[3.6.1]

  • Reverted change made in 3.6.0 and reuse the previous code editor until we fix the related issues.

[3.6.0]

  • Replaced custom query editor with @grafana/ui common editor, with support for Kusto.

[3.5.1]

  • Bugfix: Fixed issue where HTTP timeout setting was not being applied
  • Bugfix: Fixed issue when typing vs copy/pasting client secret in configuration
  • Bugfix: Fixed issue where annotation queries were not being displayed

[3.5.0]

  • Add support for national clouds
  • Replace plugin proxy routes with call resource handler
  • Add instance manager, shared http client and use new token provider

[3.4.1]

  • Bugfix: Fix an error loading schemas on the configuration page.

[3.4.0]

Note: The minimum required version of Grafana is now 7.4

  • Bugfix: Fixed issue where query builder did not handle table names that contained special characters
  • Bugfix: Fixed empty WHERE lines staying in the query builder when cleared
  • Template variables can now be used in the queries of other template variables.

[3.3.2]

  • Bugfix: Fixed an issue where the KQL Monaco editor wouldn't load when Grafana is served from a sub path
  • Bugfix: Fixed template query variables not working

[3.3.1]

  • Bugfix: nil check plugincontext user before setting tracking header

[3.3.0]

  • Add tracking capabilities by making it possible to pass the logged in Grafana user's username as a header to ADX
  • Use jsoniter instead of encoding/json in order to improve performance
  • Bugfix: Expand query template variables before building query
  • Bugfix: Fix minor typo in confirmation dialog

All notable changes to this project will be documented in this file.

[3.2.1]

  • Locked grafana-packages version and upgrade toolkit.

[3.2.0]

  • Added support for decimal data type.
  • Removed global query limit to prevent data being truncated.
  • Improved the visual query builder to make it easier to add aggregations to a query.
  • Added support for handling schema mappings to filter out parts of the database schema being available in the visual query builder.
  • Bugfix: prevent empty queries from triggering when creating dashboard.
  • Bugfix: fixing so we properly select ADX time series option when editor is in raw mode.
  • Bugfix: added a timeout setting and will make sure the plugin is handling timeout of long running queries properly.

[3.1.0]

  • Global query limit is now configurable in datasource settings.
  • Auto complete will include other filters when doing the search for possible values.
  • Added !has and has_any operators.
  • Adding datasource setting to set default view when creating new queries.
  • Dynamic caching added to enable caching setting on a per query basis.
  • Column names will not sort exact match on top.
  • Columns with auto complete support will now pre-populate options prior to typing.
  • Added support to perform time shift queries.
  • Added dcount() operator for aggregations.
  • Bugfix: operator descriptions are now wider and readable.
  • Bugfix: display template variables as options.
  • Bugfix: excluding empty/missing operators from query.
  • Bugfix: aggregations without group-by are not working.
  • Bugfix: time interval off by 1000ms.

[3.0.5]

  • Bugfix: when selecting template variables in the visual editor for table or database the values wasn't properly set. This should now be fixed.

[3.0.4]

  • Bugfix: visual editor now includes template variables in the database selector.

[3.0.3]

  • Bugfix: displays proper error message when credentials for datasource is invalid.
  • Bugfix: visual editor now supports time fields in dynamic columns.

[3.0.2]

  • Bugfix: Fixed issue with schema not updating when changing datasource.
  • Improved performance when loading table schema.
  • Improved performance when doing auto complete searching.

[3.0.1]

  • Support for value autocomplete in the visual query editor.
  • Support for dynamic columns in the visual query editor. Dynamic fields are automatically read from the table schema and are selectable when building up a query. Value autocomplete also works for dynamic columns.
  • Migration script for existing dashboards.
  • Performance improvements for the autocomplete and dynamic column features.
  • Performance improvements when loading the table schema.

[3.0.0]

  • Adds support for a new visual query editor.
  • Ports the existing query editor to React.

[2.1.0]

  • Adds support for the databases() macro for template variable queries and the database variable can then be used in the databases dropdown in the query editor. This allows the user to switch databases for a query without editing it.

[2.0.6]

  • Signed Plugin for v7

[2.0.5]

  • Bugfix for issue #61. This is a temp fix, as a proper fix requires refactoring some of the backend.

[2.0.4]

  • Bugfix for issue #73

[2.0.3]

  • Bugfix for monaco loader

[2.0.2]

  • Bugfix for issue #60
  • Updated packages

[2.0.1]

  • Add key value support to plugin (based off of mysql plugin)
  • New feature for metric naming and aliasing

[2.0.0]

  • Time series queries now support alerting.
  • Time series queries now support multiple value and multiple string columns.
  • The Kusto "time series" type created with the Kusto make-series operator is now supported.
  • Macros have been added so as not to conflict with Grafana's built-in query Macros: $__timeFrom, $__timeTo, and $__timeInterval.
  • Caching of Table and Time Series queries has been removed until backend plugins support caching.
  • Queries no longer have an ORDER by clause appended when there is not one, however if time series is unsorted there will be a warning in the query editor.

[1.3.2] - 2019-06-19

  • Bugfix for issue #8
  • Updated packages
  • Added circleci

v1.3.0

  • Adds an order by clause to the query if there is none specified. It uses the datetime field from the where clause or summarize...bin().
  • Removes the Subscription Id field from the config page as is no longer needed.

v1.2.0

  • Adds a config option for caching. The default in-memory cache period is 30 seconds, the new Minimal Cache Period option allows you to change that.

v1.1.0

  • Adds $__escapeMulti macro

v1.0.0

  • First version of the Azure Data Explorer Datasource.