Prometheus ships 2.11.0, delivers ratios for humans…


Prometheus 2.11.0 lit up the skies today, bringing four new features, 13 enhancements and a pile of bug fixes and changes.

The open source monitoring and alerting project’s updates are usually short on context, or indeed words in general.

However, according to the project’s GitHub page new features include the option to use the newer Alertmanager API v2. InitContainers is now included in Kubernetes Service Discovery – they were previously disregarded, which meant certain troubleshooting and other info was not being picked up. 

Also added is a humanizePercentage function template, which should express some alerts as percentages rather than ratios, making them easier for “humans” to grasp. The last new feature is that WAL records can now be compressed using the Snappy compression tool. In testing this has halved the size of WAL, apparently, with a minimal increase in CPU load.


WAL issues pop up in the 13 items strong list of “enhancements”. These include creating a new segment when starting the WAL, and reducing disk usage for WAL in small setups.

Other enhancements include: the addition of storage warnings to LabelValues and LabelNames API results; Query performance improvement, which had slashed the time taken by HashforLabels; and remote write allocs improvements.

Seven bug fixes are part of the update, including one which resolves a race condition in maxGague, while another fixes a Zookeeper connector leak. Other fixes include one for “unknown series references” after a clean shutdown, and another one for unsafe snapshots with head block.

- Advertisement -