Overview of Cloud Foundry Application Container Metrics

Application metrics can generally be divided into two main categories

• Metrics for the container itself
• Metrics generated by your application (usually unique to the application and they are known as custom app metrics)

Every container managed by Cloud Foundry emits a set of metrics that can be viewed and analyzed to understand what is going on within the container; these are known as the container metrics.

Remember: a container is where an application runs once you’ve deployed it with cf push.

Container metrics are generally very simple, and mostly provide information about resource utilization. They are averaged and emitted about every 15 seconds. Container metrics include, but may not be limited to, the following by default:

To see the complete list and more detailed information on these metrics you can check out the official page on Diego Container Metrics.

What if you need more detailed information, like what parts of your app are using how much memory each, for example? That’s where custom app metrics come in. Custom metrics are beyond the scope of this tutorial, but you can check the tutorial Overview of Custom Metrics for more information.

The memory metric is about the overall used memory in the app container and doesn’t provide any details about the used memory for the different processes running inside the container. You will most often see memory metrics displayed as memory of memory_quota, or in other words, the amount of memory used out of the total available memory for an instance. For example your app could be using 30MB of memory (memory) out of 256MB of memory (memory_quota), and this would be represented as “30M of 256M”. Disk metrics are displayed similarly to memory metrics except that they are represented as disk of disk_quota (e.g. “100M of 1G”.). The disk measurements are how much space is used by the application on disk, not a measure of disk I/O.

The total memory and disk space can be set with the application manifest, or through the command line via arguments to cf push. SAP allows a maximum memory per instance of 16GB, starting with 1024MB of memory by default. SAP also allows a maximum disk quota of 10GB per instance, and by default each app instance starts with 512MB.

When it comes to the CPU metrics it’s important to know the difference between the cpu metric and the cpu_entitlement metric. The cpu metric gives you the percentage of CPU used by an application instance in relation to the total CPU available on the Diego Cell. The cpu_entitlement metric on the other hand gives the percentage of CPU used by the application instance in comparison to what the application instance is actually entitled to use on the Diego Cell.

Given the above, if an application instance is entitled to use 30% of the system CPU, and it is using all of that CPU, then the cpu_entitlement metric will show that the instance is using 100% of its CPU. If you were to look at the cpu metric for the same instance at the same point in time, however, it would show that the instance is only using 30% of maximum available CPU. The cpu metric is not so much of interest for the application but it is the original CPU metric emitted by Cloud Foundry that is why it is mentioned here.

The cpu_entitlement metric will always be the most useful CPU measurement as it shows the percentage of CPU usage in relation to the amount of CPU the instance can use.

Now that you know about the CPU, disk, and memory metrics, you probably also want to know how to actually see them. To view these metrics for your cf-nodejs application using the Cloud Foundry Command Line Interface (cf cli) run the following command in your terminal (after your application has started):

cf app cf-nodejs

The log cache, besides storing logs, also stores container metrics. Viewing metrics contained in the log cache is very similar to viewing logs with the log cache plugin as seen in the Overview of Cloud Foundry Application Logs tutorial. Using the cf tail command you just need to change the value of the -c flag to metrics:

cf tail -c metrics cf-nodejs