Refinery, our tail-sampling proxy, has a new release! 1.20 has several features designed to help when operating Refinery at scale.
Some highlights include:
meta.dryrun.sample_rate instead of SampleRate so that queries are accurate in Usage ModeSee the full list of changes here!
We’ve updated our Time Comparison picker to show 28 days prior instead of 1 month, which was an imprecise duration and caused confusion. It now matches the query window time picker as well.
You can now see creation dates in your list of API keys! View your current list of API keys in your Environment settings.
We’ve improved the autocomplete experience for aggregates. Operators like COUNT, HEATMAP, and P95 are now prioritized first in the VISUALIZE autocomplete list above column names that include the same string.
You can now view additional details for each Honeycomb Metrics chart. Reference these details for more context about these queries when correlating your queries alongside your metrics. View query details by mousing over the icon in the upper left of your metrics chart.
Have you ever run a query with a Group By attribute and wanted to BubbleUp on each of the groups? Now you can! Navigate to the BubbleUp tab in your query results and you can click on each of the grouped chart lines to set that group as the selection. We’ve even added keyboard support, so you can tab through the choices and press enter to select.
We've made several improvements to the hover charts on our graphs in order to make them easier to use.
Metrics are now available to customers on our Free plan! Each metric event counts toward your monthly limit, so it’s a good idea to think about how you’d like to treat them, whether preaggregating or considering other ways to manage metrics volume, depending on how you’d like to query and correlate them alongside other non-metrics datasets. There are many options for getting started.
Log scales with fractional data now extrapolate missing values toward the bottom of the scale as the values approach 0, represented in 10e-n notation. They were previously extrapolated toward 1 (100) at the top of the scale. This new approach aligns missing data with the start of the scale, which makes for better readability.
We’ve updated when we show decimal notation (0.9) and scientific notation (9e-4) in the y-axis. Previously some charts would display SI prefixes (900m) which was commonly misread.
Additionally, graph lines no longer cross the y-axis in chart results!
Feel free to report any edge cases and questions about these updated in our Pollinators Slack workspace.
We released a new SDK distribution for OpenTelemetry! It simplifies configuration to send to Honeycomb, and adds multi-span attributes, deterministic head sampling, and a local visualization feature that gives you a link to a trace you created when you're developing with it.
To get started, check out our docs on Node.js. We also have several sample apps you can use, too.