Microsoft Azure VM - Enabling Diagnostics extension for Windows & Linux VMs

Microsoft Azure VM - Enabling Diagnostics extension for Windows & Linux VMs

Steps to enable Diagnostics extension for Windows VMs

1.Log in to Azure Portal (https://portal.azure.com) using the credentials of 'Administrator'.
2.Navigate to your virtual machine.
3.Click on 'Diagnostics settings' on the vertical pane. Select 'Agent' tab and click on 'Remove' at the bottom, as shown in the below image:


4. Under 'Overview' tab, click on 'Enable guest level monitoring' and click Save as shown in the image below.

 
5. In Performance counters tab, all the Basic metrics are chosen by default, and the metrics monitored in Applications Manager are included in it. If the user wants to enable diagnostics only for the monitored metrics, select Custom and select the below-listed metrics and Save.

Metric Name in Azure Portal
Metric Name in AppManager
\\Processor Information(_Total)\\% User Time
User Time
\\Processor Information(_Total)\\% Privileged Time
Privileged Time
\\Processor Information(_Total)\\Processor Frequency
Processor Frequency
\\System\\Processes
Process Count
\\System\\Context Switches/sec
Context Switches/sec
\\Process(_Total)\\Thread Count
Thread Count
\\Process(_Total)\\Handle Count
Handle Count
\\Memory\\Committed Bytes
Committed Bytes
\\Memory\\Available Bytes
Available Bytes
\\Memory\\% Committed Bytes In Use
Committed Bytes In Use
\\Memory\\Page Faults/sec
Page Faults/sec
\\Memory\\Pool Paged Bytes
Pool Paged Bytes
\\Memory\\Pool Nonpaged Bytes
Pool Nonpaged Bytes

6. Restart the VM.

Steps to Enable Diagnostics Extension for Linux VMs

 1.  Log in to Azure Portal (https://portal.azure.com) using the credentials of 'Administrator'.
 2. Navigate to your virtual machine.
 3. Click on 'Diagnostics settings' on the vertical pane under 'Monitoring'.
 4. To uninstall the old diagnostic agent and reinstall again, go to 'Agent' and click 'Remove' to remove the current diagnostic agent. On the same page, now choose a storage account from the dropdown and then click 'Enable guest-level monitoring' to enable the VM diagnostic.



5. To reduce the amount of data stored in Storage Account tables, configure only the required diagnostics instead of all.

Once the diagnostic settings are updated successfully, in the same pane, click 'Metrics' and then click 'Custom' and remove the unwanted metrics. Refer to the below image to view the required metrics and their configuration. Also, set the aggregation intervals to 1 minute (Choose only 'PT1M' and remove other intervals like 'PT1H', 'PT5M', etc.) and finally click 'Save' to save the configuration.
Note:
* In Applications Manager, we retrieve the VM diagnostics only from the PT1M tables.
* Configuring only the required metrics and setting the aggregation to 1 minute is mainly to reduce your Azure Storage cost by avoiding unwanted storage. In addition, it removes the old tables from the configured Storage Account to minimize the Storage cost.


Metric Name in Azure Portal
Metric Name in AppManager
/builtin/disk/averagereadtimeAverage Read Time
/builtin/disk/averagewritetimeAverage Write Time
/builtin/memory/availablememoryAvailable Memory (MB)
/builtin/memory/percentusedswapUsed Swap memory (%)
/builtin/memory/usedmemoryUsed Memory (MB)
/builtin/memory/percentavailablememoryAvailable Memory (%)
/builtin/processor/percentidletimeIdle Time 
/builtin/processor/percentiowaittimeIO Wait Time 
/builtin/processor/percentprocessortimeProcessor Time(Linux)

6. To check whether the diagnostic agent is working properly by storing the diagnostic metrics in the configured Storage Account:
  1. In the Azure portal, go to Virtual Machine and click 'Metrics' under 'Monitoring' in the left pane.
  2. Make sure the 'Scope' is pointing to the respective Virtual Machine and choose 'Guest (Classic)' in 'Metric Namespace' dropdown.
  3. Choose any one metric, let us check by choosing 'CPU idle time' in 'Metric' dropdown and check the data points for the current timestamp as below:


How Applications Manager collects performance metrics of Azure VMs?

Once diagnostics are enabled for VMs in the Azure portal, the diagnostics metrics will be stored in the specified storage account every 1 minute. But the diagnostics of the current minute will be logged in the storage account only after some x minutes. The minimum delay in logging might be 3 minutes (from our observation). In some scenarios, the logging time delay may increase.

By default, Applications Manager collects 10 minutes of delayed data. In order to configure a lesser time delay in Applications Manager, go to Admin > Performance polling > Optimize data collection > Choose 'Azure Virtual Machine' monitor > Provide the delay (in minutes). Applications Manager will start collecting data accordingly.

In case, if data is not available for the delay provided, then this means that the logging delay of the Azure setup is higher than the mentioned value.