Functionality Added or Changed
The performance of the Data Purge Behavior has been improved.
In previous releases, the data purge executed all DELETE statements before committing, which could result in a very large, final transaction. As of this release, each DELETE is committed individually. (Bug #18482800)
It is now possible to configure the LDAP connection timeout values. The following were added to the LDAP Authentication frame of the Global Settings page:
Connect timeout: time elapsed without establishing a connection to the LDAP server. If a connection is not established within the defined number of seconds, an error is returned.
Read timeout: time elapsed without a response to a request for data from the LDAP server. If no response is received within the defined number of seconds, an error is returned.
If the value is set to 0, the timeout value defaults to the operating system's default value.
The Group Overview Configuration toolbar enables you to define the period for which data is displayed in the timeseries graphs on the Overview dashboard. It also enables you to specify a default group of MySQL instances.
Under certain circumstances, the defined event handlers were not displayed in the MySQL Enterprise Monitor User Interface. This occurred if one or more event handlers were used to monitor an automatically-created replication group which was deleted. If the instances which made up the replication group could not be detected, the replication group was deleted in MySQL Enterprise Service Manager also.
As of this release, if all instances which make up a replication
group are deleted, the replication group monitored by
MySQL Enterprise Service Manager is not deleted, but is displayed with the
following naming convention:
Group (id=NNNN), where NNNN is the unique ID of the
Non-existent replication groups must be removed manually
Under certain circumstances, the MySQL Enterprise Monitor Agent stopped collecting data and was reported as down if it was monitoring multiple mysqld instances. (Bug #18435199)
The MySQL Enterprise Monitor Agent started as root instead of starting as the mysql user if the mysql user was defined in LDAP/NIS.
As of this release, the initialization script checks for non-local users. (Bug #18392067)
It was not possible for MySQL Enterprise Service Manager 3.0.0 to use a MySQL Enterprise Monitor Agent of version 3.0.4 or higher. The agent returned a Null Pointer Exception on start-up.
It is recommended that you upgrade both MySQL Enterprise Service Manager and MySQL Enterprise Monitor Agent to their most recent versions to avail of all fixes and enhancements.
The upgrade process changed the properties of the files backed up to the backup directory. As a result, if the installation was reverted to a previous version, the local repository would not start.
As of this release, the properties of these files are not altered during the upgrade's backup process. (Bug #18341449)
The full and upgrade installers did not correctly check for available disk space.
As of this release the installers use the following formula:
Full installation: Free Space - Installation Size - Extra Space > 0
Upgrade: Free Space - Backup Space - Installation Size - Extra Space > 0
Free Space = the amount of space available on the target disk.
Installation Size = the amount of space required by the product installation
Extra Space = a hard-coded buffer of 500MB.
Backup Space = the amount of space required to backup the previous installation. This value can be very large.
The y-axis labels for the following graphs did not display correctly for large negative values, more than one hour, or very small negative values, less than one second:
Replication Slave Delay
(Bug #18312245, Bug #71685)
The tooltips for several advisors contained broken links, or links to outdated content. (Bug #18288215)
If a time difference existed between the MySQL Enterprise Service Manager and the monitoring agent servers, the agent was reported as SHUTDOWN or TIMEDOUT, although the agent was still running. As of this release, if a time difference exists, but the agent is still collecting and transmitting data to the MySQL Enterprise Service Manager, the agent is reported as active.
It is recommended that you synchronise the MySQL Enterprise Service Manager server, and all monitored servers, with the same time server.
In previous releases, the upgrade process did not correctly stop
the running Tomcat process. This could result in multiple
running Tomcat processes after the upgrade completed. As of this
release, the upgrade process stops the Tomcat process and the
jsvc process which hosted it.
This fix does not affect the 3.0.7 to 3.0.8 upgrade process. It corrects the issue for future upgrades. Before upgrading to 3.0.8, it is recommended that you stop the running Tomcat instance manually.
This issue also affected the upgrade's backup process.
Under certain circumstances, when monitoring a large number of servers in the MySQL Enterprise Service Manager browser interface, a memory management issue occurred. As a result, the browser consumed a large amount of system resources and performance was reduced. (Bug #18182101)
The following enhancements have been made to the MySQL Enterprise Monitor Agent to improve the MySQL Enterprise Service Manager start-up performance:
Delta Quantizer logging has been moved to the debug log-level.
The number of default backlog reporting threads has been
reduced to 4. This option can be configured in the
data-reporting-threads parameter of the
In previous releases, if the monitored server was under heavy load, the monitoring agent was unable to collect data at the defined rate. As a result, gaps were displayed in the Query Response Time Index graph.
As of this release, while it is still possible for such gaps to occur, their occurrence is greatly reduced by improvements to the agent's performance. (Bug #17733081)
A Null Pointer Exception was generated if the Query Analyzer received EXPLAIN data which was either incomplete or invalid. As of this release, the incomplete data is logged.
The units used in the Filesystem Usage Change Rates graph were incorrectly defined.
Null values were displayed in the Filesystems section of the MySQL Instance Details dashboard. This occurred for certain types of pseudo-filesystem, such as /proc, and for network filesystems. As of this release, these values are not displayed.
CPU graph data was not properly stored in the Service Manager repository after the MySQL Enterprise Service Manager was upgraded. As a result, some graphs, such as the CPU Utilization graph, displayed gaps. It was necessary to restart the monitoring agent to restore the graphs to normal. As of this release, the CPU-related data is not dropped after an upgrade.
Under certain circumstances, it was not possible to generate Support Diagnostics for a remotely monitored instance. A Null Pointer Exception was returned.
The URL displayed at the end of the MySQL Enterprise Service Manager installation was incorrect.