July 2017 Summaries

The document outlines various updates and improvements for the OneAgent software, including new support for Node.js on Node 8 and PHP on PHP 7.1. It highlights enhancements such as better detection of the OpenStack Heat service, reduced installation times on systems with SELinux policies, and full-stack monitoring capabilities for IBM AIX 7.x through an Early Access Program. Additionally, it mentions Unix domain sockets reporting, automatic naming for Cloud Foundry hosts managed by BOSH on Linux, and Docker injection for Linux PowerPC Little Endian. The update also features improvements in GO process discovery and grouping, the rollout of OneAgent on Google Container Engine-based clusters, network stability enhancements, and security measures like preserving monitored entity IDs for Security Gateway after upgrades on Windows systems.

Dynatrace has introduced a new custom alerting mechanism that enhances its anomaly-detection capabilities by allowing users to create tailored alerts without compromising the software's AI-driven detection of performance anomalies. This feature enables users to define specific metric-threshold breaches and set rules for the number of violations required before an alert is issued, thus preventing unnecessary alerts for known issues. Users can create custom alerts through the platform's settings, specify meaningful names, and view these alerts alongside standard problem notifications in their Problems feed. While custom alerts provide detailed insights into threshold breaches, they are not integrated into Dynatrace's problem correlation and causation analysis, meaning they serve as standalone notifications rather than factors in identifying root causes. Feedback on this new feature is encouraged through the Dynatrace Community, and a 15-day free trial is available for users to explore the platform's capabilities.

Node.js is an event-driven platform where the event loop, managed by the libuv library, orchestrates the execution of callbacks in a single thread, contrary to common misconceptions that involve separate threads or an extensive use of thread pools. The event loop consists of distinct phases—such as timers, I/O callbacks, polling, setImmediate, and close—that manage different tasks. Monitoring the event loop involves metrics like tick frequency and duration, work processed latency, and event loop latency, which provide insights into application performance and potential bottlenecks. These metrics are crucial for understanding application behavior and determining strategies for optimizing performance, such as utilizing all CPUs, tuning the thread pool size, or offloading CPU-intensive tasks to other services. The article highlights the development of a comprehensive event loop monitoring solution by Dynatrace, which plans to integrate event loop telemetry into its root cause detection for enhanced anomaly correlation.