What’s new in VividCortex? We’ll do these blog posts once a week to let you know all of the changes such as new features, bug fixes, performance improvements, and even a peek into the future.
If you haven’t seen the VividCortex application in action, please register for a live demo on production systems on May 14th.
Catch us at the following upcoming events:
- Agent management UI in the web application.
- More agent performance improvements.
- Events dashboard redesign and performance improvements.
- Replication dashboard, for real-time live updating view of your replication topology and status.
Improvements This Week
Most of our time this week has been invested in underlying improvements you won’t notice for a while.
- Improved resiliency of our metrics API. Note that an architectural change is underway to replace some of the backend systems and greatly improve the reliability in the face of infrastructure failures.
- Added per-database query metrics to the TCP sniffer agent. We will build additional UI features with these metrics. Note that this has a slight CPU cost (details below).
- Improved the design and layout of the graphs page; the new page is cleaner and makes it more obvious when graphs are filtered, which has been confusing (even to us) in the past.
- Changed the y-axis units of the replication delay graphs from microseconds to seconds.
- Added 1-second granularity to graph hover tooltips.
Agent bug fixes:
- vc-mysql-metrics sometimes measured replication delay inaccurately due to heartbeat row update failures.
- vc-agent-007 made unnecessary PID file checks while supervising, and could get confused.
- vc-os-metrics could fail to capture some metrics when /proc files were invalid.
Agent Performance This Week
Here’s a graph of agent performance on a typical server over the last week. Note that the increase in vc-mysql-query CPU consumption is due to additional per-database query metrics we’re gathering now.