Feature Focus: March 2023By Nick Rycar | Last modified on April 6, 2023
Spring has sprung, and the bees have been busy. Let’s have a look at what’s new in Honeycomb at the close of March.
In the spirit of spring cleaning, we’ve made some minor tweaks to how articles are organized to make things easier to find. We’ve also fleshed out a few use cases. For example, we’ve updated the OpenTelemetry (OTel) Collector Deployment Patterns documentation to provide guidance on when to deploy the collector in Gateway Mode vs. Agent Mode.
We’ve also added a new section to our integration docs for CI/CD, and specifically, buildevents. Buildevents is a utility that creates distributed traces out of your CI/CD pipelines, where spans represent build tasks and testing. You can see where a particular build is spending the majority of its time, and visualize the impact of parallelizing and streamlining steps. It is a small binary that’s easy to integrate with most CI/CD solutions. In addition to the docs page, we have detailed guides available for CircleCI, GitLab, and GitHub Actions.
What’s more, traces generated by buildevents can be treated just like any other trace sent to Honeycomb. That means you can run queries, generate heatmaps, and use BubbleUp to surface anomalous behavior.
One month is now 28 days
When you query data with Honeycomb, you have the option to compare the results against that same data from two hours, one day, one week, or one month prior. The only problem is that “one month” can be anywhere from 28 to 31 days. That’s perfectly acceptable precision for casual conversation, but not so much for reading graphs.
We’ve updated the dropdown to read “28 days” so you know exactly what you’re looking at. 🗓️
We’re released Refinery 1.20
Refinery is a trace-aware tail-based sampling proxy. In short, it provides a programmatic way to reduce the volume of traces you send to Honeycomb, with intelligence to apply sampling decisions such that important troubleshooting data is not discarded in the process.
1.20 is a sizable update, notably introducing Stress Relief, a mechanism that sheds load when Refinery is under duress, preventing crash loops. Other improvements include configurable trace & span IDs, longer trace caching, and improved dry-run functionality.
For a full list of updates, be sure to check out the release notes.
Python OpenTelemetry SDK
We have a new auto-instrumentation option available: OpenTelemetry distribution for Python.
In addition to simplified OpenTelemetry initialization, the Honeycomb distribution makes it easier to add multi-span attributes and configure sampling, as well as link directly to your traces in Honeycomb with local visualizations. Full details are available in our docs, or in the GitHub repo.
Full disclosure: this is technically an April update. In my defense, you can hardly expect me to keep news this exciting under my hat for another month. If it’s any consolation, I promise to crow about it in the next Feature Focus as well.
Speaking of which…
Wrapping up our Feature Focus March 2023 edition
See you next month!
In our continuous journey to support teams grappling with the complexities of Kubernetes environments, we’re thrilled to announce the launch of Honeycomb for Kubernetes, a...