Jujing v2.6 ships log keyword and rate alerting, full OTel histogram support, DM8 paging optimizations, and fixes for high-concurrency writes and timezone issues, plus upgrade steps and caveats.
New Features
Log alerting: keyword-based alerts on log content, such as firing immediately on an OutOfMemoryError line, plus rate-based alerts that trigger only when error lines per minute exceed a threshold — filtering out occasional noise.
OTel metric enhancements: full support for histograms and exponential histograms, with histogram query performance improved roughly 3x and smoother tail-latency analysis.
DM8 optimization: rewritten pagination queries, cutting a 10,000-row list page from 2.8 seconds to 0.4 seconds — a noticeable improvement in domestic environments.
Improvements
The Trace detail page now switches between flame graph and timeline views, with one-click drill-down from a slow span to method level.
Alert silence windows now support weekday and holiday configuration, so rotations don't page you on days off.
Dashboard variables support cascading: switching a cluster automatically loads its host list, reducing manual selection.
Bug Fixes
Fixed intermittent log-index write failures under high concurrency (#2841).
Fixed an 8-hour timestamp offset on some ARM collectors (#2857).
Fixed occasional session invalidation under TongWeb in domestic environments (#2863).
Upgrade Notes
Upgrading from v2.5 requires backing up DM8 or MySQL first; the upgrade script migrates the schema automatically with an expected 5–10 minute downtime, so run it in a low-traffic window.
Log alerting is new and disabled by default; enable it manually under Alert Rules and set your thresholds.
If you run ARM collectors, upgrade them to the v2.6-paired collector build to pick up the timezone fix.
See the download page for the full changelog and packages. If you hit issues during upgrade, file a ticket or post in the community with your version number and upgrade log — it speeds up diagnosis.