Functionality Added or Changed
Network file systems, such as NFS, SMB, and CIFS are now supported. They are displayed in the asset selector and the Advisors configuration page for filesystem-related advisors. It is also possible to override their configuration and filter them for reports, graphs and events. (Bug #18259946, Bug #71718)
It is now possible to edit the Agent CPU Thresholds in the Agent Health Advisor. The default value for the embedded agent is 40%. (Bug #18025021)
The Edit dialog for the following advisors was redesigned for improved usability:
Agent Health Advisor
MySQL Enterprise Backup
CPU Utilization Advisor
Filesystem Free Space
The contents of these advisors are now organised using collapsible sections.
The Database Availability tooltip for Month displayed the data for Week instead of Month. The graphs displayed the correct information. (Bug #18190436)
Under certain circumstances, NULL was returned if a non-contiguous list position was encountered, resulting in a NullPointerException. If this error occurred, the agent stopped monitoring the server. (Bug #18204794, Bug #18230493)
It was not possible to display My Oracle Support information in the Open Support Issues frame on the What's New page due to an issue with the MOS 6.8 rollout in February 2014. (Bug #18198557)
Under certain circumstances, for installations with more than
100 agents, the
thread blocked processing of all agent threads.
Agents which were active and available were reported as unavailable if the system was under heavy load. Emergency events were logged as a result. This issue often occurred after a system restart. (Bug #18205114)
Under certain circumstances, a deadlock occurred if a replication group was renamed. (Bug #18245772)
Stacked graphs were reset to line graphs if the page was refreshed. (Bug #18148597, Bug #71505)
Text was displayed incorrectly if the English (Canada) locale was selected. (Bug #18140344)
The Support Diagnostics files, generated from the MySQL
Instances dashboard, were very large due to logging of GTID
variables which were constantly changing. This information was
also written to the database, increasing the write load. As of
this release, GTID variables are no longer collected, with the
The upgrade process does not delete these entries from existing logs.
Under certain circumstances, on Linux systems, the CPU evaluation results were incorrect. This was caused by a Linux kernel issue, where the CPU tick counter reset, appearing to go backwards. Based on this inconsistent data, MySQL Enterprise Monitor inferred a very high CPU usage and generated false-positive alerts. As of this release, MySQL Enterprise Monitor does not record such tick resets. If resets occur, they are represented as gaps in the CPU graph, and the field and delta values are logged at debug level.
If an unmonitored master was detected, a group update was repeatedly performed. As of this release, if an unmonitored master is detected, the group update is not performed.
On Mac OS X, the agent reported Mac OS X version 10.9 as "Mac OS X Unknown".