Memory metrics missing from Azure dashboard
You can monitor memory and other "Guest" level metrics in Azure, here's how:
- in Azure, go to your virtual machine, scroll down the settings to Monitoring > Diagnostics Settings
- Click to enable Guest level monitoring, it can take a few minutes
- Then you can go into Metrics for the VM, or Monitor at the top level:
- choose the resource (the VM)
- choose Guest in the metric namespace, it will load all the new metrics
- choose Memory\Committed bytes or whatever ones you want.
- You can then pin to dashboard etc as you would normally
Cloud Services is not the same as Virtual Machines. When you use cloud services, Azure will provision VMs for you and Azure is able to install monitoring tools that see the amount of available memory. When you create your own VMs Azure can't and shouldn't do that. In other words, with VMs you are on your own. The metrics you do see in the portal are the ones that can be measured from outside the VM.
If you do deploy as a Cloud Service then initially you will only have the same metrics as for the VM. There are several ways you can change this.
The easiest is to go to the configuration for your cloud service in the Management Portal and change the logging level from Minimal to Verbose; That will enable a lot more metrics. Alternatively, you can specify which metrics you want collected in the cloud configuration in your project in Visual Studio. It is also possible to do this in code, though that is not the currently recommended practice, instead use the configuration tool in the cloud project in visual studio.
The key thing to understand about the metrics in Cloud Services is that, whichever way you elect to configure them, they are stored in a standard way in Table Storage and Blob Storage. That means using the Azure Management Portal or the tool in Visual Studio or code, the outcome is the same. This also means that a variety of tools including Cerebrata, Visual Studio and, indeed, the management portal can all read this data. It is also worth noting that because of the way this works, the configuration can be changed at runtime, usually through the portal but there are other tools and approaches in code.
In my experience, you normally only want to sample your performance metrics every two minutes, but do the log shipping every minute. Also note that you can configure trace logs and IIS logs etc to be available to tools like Visual Studio and Cerebrata. For Cloud Services, it is quite rich functionality but it takes some working with it before you start to "get" it all. Enjoy!