Skip to main content
Monitoring, Sematext Cloud

Topping top! New Real-Time Process Monitoring

Stefan Thies Stefan Thies on

What are the essential things to monitor in your infrastructure? Sure, CPU utilization, memory usage, and IO throughput. However, once you notice a significant load somewhere in your infrastructure you want to know what is causing it, and that typically boils down to needing to find the process that’s using too much CPU or memory or that’s doing disk or network IO like there’s no tomorrow.
Once the server with performance issues is located the typical procedure is to run top via ssh (unless your infrastructure is containerized, in which case this is considered bad practice).  The top command shows processes with the highest resource utilization.

Screen Shot 2019 06 14 at 12.13.25

Linux top command

In the screenshot above you can instantly see which processes are consuming the most memory. That’s why this command is used at the beginning of most troubleshooting sessions. But what if you manage a large infrastructure with dozens, or even hundreds, of servers? Running top is helpful when you know which server has a performance problem. In many cases, troubleshooting starts with a ticket telling you “something is too slow” or simply “doesn’t work”. Wouldn’t it be awesome to see the top processes in your whole infrastructure at a glance?

Sematext has updated its Infrastructure Monitoring. Besides the existing Server Monitoring and Container Monitoring, we’ve introduced the new Real-Time Process Monitoring.

New Real-Time Process Monitoring

The new Process Monitoring displays the Top N processes in your entire infrastructure. No more ssh/top commands to identify hot processes. Regardless of the process type, services, cron jobs, command-line tools, containers, Sematext Agent detects the processes with the highest CPU, memory, and IO usage, and ships the metrics to Sematext Cloud.

Here you have a complete view of top processes across your whole infrastructure.

You can filter the processes by any tags connected to their metrics, such as hostname, command or container name. If you’re used to container monitoring you might have wondered about the difference between infrastructure load and total container load. Infrastructure load is often caused by processes like dockerd or other core services running directly on the hosts.

Getting Started with Process Monitoring

If you’re using Sematext’s Monitoring Integrations, you’ve already installed the Sematext Agent. To benefit from new features simply update the agent. Sematext Agent has a very low CPU and memory footprint. It collects server, process, and container metrics. Installation instructions are displayed when creating new Apps or when you open an existing App setup screen.

To provide support for Process Monitoring and other functionality we will be announcing shortly, we’ve introduced the new App type called Infra App. This Infra App holds all core infrastructure monitoring data, such as information about processes. A default Infra App is created automatically for you when you set up your first Monitoring App. You can create multiple Infra Apps, but you’d typically have only a few of them, often just one. If you run multiple environments, say on-premises, in the cloud, multi-cloud or hybrid-cloud you might want to create a separate Infra App for each environment. Or if you have your applications deployed in multiple data centers, or if you have production and staging environment you may want to have a separate Infra App for each such environment.

A few seconds after you’ve deployed the Sematext Agent, you will see metrics appearing in Sematext Cloud.

process monitoring 1

Once the Agent is deployed it continuously monitors all process stats and enriches the metrics with metadata by adding tags for hostname, process name, process.cmd, process.pid, container.id, container.name, container.image, and many more.

Finding a Needle in a Haystack

If you only look for the hottest processes you can find them in the Top N processes chart. The tile map visualization displays the processes with the highest resource usage, while the process view can be grouped by various tags such as the hostname. In this case, the tile map will show the top N processes for each group.

Beside grouping, the Process Monitoring view supports filters for all tags transmitted by the Sematext Agent.  This makes it very easy to find which processes are using the most resources in your entire infrastructure.

process monitoring 2

Summary

You can use the lightweight Sematext Agent for infrastructure monitoring in any computing environment:

  • On-Premises
  • Cloud and Multi-Cloud Deployments
  • Hybrid Cloud Environments

Sematext’s real-time Process Monitoring helps you to quickly identify the hottest processes in your entire infrastructure, in one place, via an intuitive user interface, without the hassle of connecting to servers.
Like the sound of our new real-time Process Monitoring? Give it a try! We’d love to hear your feedback!