When upgrading to MySQL Enterprise Monitor 3.0.22, you must upgrade your Agents also. The change in TLS support makes it impossible for MySQL Enterprise Service Manager 3.0.22 to communicate with earlier versions of MySQL Enterprise Monitor Agent
MySQL Enterprise Service Manager 3.0.22 can communicate with MySQL Enterprise Monitor Agent 3.0.22, only.
The Data Purge process now removes the historical data in order, from oldest to newest. (Bug #18251338)
References: See also: Bug #20554946.
The bundled JRE was updated to 1.8.0_45 for both the Agent and Service Manager.
The Rickshaw graph library bundled with MySQL Enterprise Monitor is upgraded to version 1.5.1.
The monitored instance disconnected the Agent if the Agent's connection was idle between data collections. This forced repeated reconnection between the Agent and the monitored instance.
As of this release, the Agent provides the
wait_timeoutvalue for connections to the monitored server. The new timeout value is 1800 seconds, which is sufficiently long to prevent such reconnection.
As of this release, the Agent sets
sql_mode=NO_AUTO_CREATE_USERin addition to
NO_ENGINE_SUBSTITUTION. This is set to prevent repeated occurrences of the following error from MySQL 5.7.x:
Warning | 3129 | Unsetting sql mode 'NO_AUTO_CREATE_USER' is deprecated. It will be made read-only in a future release.
As of this release, TLSv1 support is removed. Only TLSv1.1 and TLSv1.2 are supported. TLSv1.2 is preferred. For a full list of the supported ciphers, see OpenSSL Ciphers.Important
The change in TLS support makes it impossible for MySQL Enterprise Service Manager 3.0.22 to communicate with earlier versions of MySQL Enterprise Monitor Agent
MySQL Enterprise Service Manager 3.0.22 can communicate with MySQL Enterprise Monitor Agent 3.0.22, only.Important
As a result of this change, the supported browser list has changed. For more information, see Supported Browsers.
The following MySQL Enterprise Firewall system and variables are available for use in custom graphs and advisors:
For more information on these variables, see MySQL Enterprise Firewall Reference
Default sorting was incorrect for the following:
Events: now sorted by Current and Time.
Query Analyzer: now sorted by Latency:Total.
Also, the column names were standardized in the Query Analyzer's Show/hide columns selector drop-down list. As of this release, the column names listed in the selector now match those used in the Query Analyzer. (Bug #21216799)
If MySQL Enterprise Service Manager was upgraded to 3.0.21, and there were active Event Handlers, a
NullPointerExceptionwas logged for each SMTP Notification Action. This error did not interrupt the notifications, but could quickly fill the logs. (Bug #21196480)
A Unix/Linux script,
setenv.sh, was called during the update process on Windows platforms. The installation process was unaffected, but an error was logged in the installation log file. (Bug #21073969)
The event text for MySQL Availability and Agent Not Reachable events contained a reference to non-existent documentation. These events have been updated with links to relevant documentation. (Bug #21066744)
Query Analyzer EXPLAINs failed for views. A connection error was displayed. The
SHOW VIEWSprivilege was not automatically granted to the Limited and General users. As of this release, the
SHOW VIEWSis automatically granted to the auto-created users, General and Limited.
If you are manually managing your users, and want to run EXPLAINs on views, you must grant this privilege to your users. (Bug #20974495)
The tooltip was empty for the MySQL Instance is Down icon on the Replication dashboard. (Bug #20916529)
Text was displayed incorrectly if the English (Canada) locale was selected. (Bug #20896159)
The time displayed on the X-axis of some graphs was incorrect. A period of 0 to 45 seconds was repeated instead of displaying the time of day. (Bug #20744237)
If an event handler had more than one group associated with it, and one of those groups was removed automatically, by a replication topology change for example, MySQL Enterprise Service Manager failed to start. A
ConcurrentModificationExceptionerror was listed in the logs. (Bug #20603989)
If MySQL Enterprise Service Manager was configured to use a reverse proxy, Inventory links were not correctly resolved. As of this release, all links are server-relative. (Bug #20519137)
MySQL Enterprise Backup events were not auto-closed even if the associated Event Handler was set to auto-close. If the issue was corrected, and the backup was successful, no events were generated.
As of this release, an event is generated for each backup run. Each event is identified by the MySQL Enterprise Backup backup ID. Successful backup events are auto-closed.
If a backup fails, generating a failure event, but a subsequent backup using the same command line is successful, both the success and failure event are auto-closed. (Bug #19776693)
In previous releases, many of the graphs, and some expression-based Advisors, used a hard-coded frequency of 60 seconds instead of the defined schedule.
As of this release, the variable
%defaultFreq%is added to the expressions of all affected graphs and Advisors. This variable ensures the graphs and Advisors use the defined schedule.Note
This new variable can also be used in custom graphs and Advisors.
Under certain circumstances, it was not possible to retrieve filesystem details for some types of NFS disk. As a result, the Filesystem Advisor could not generate Free Space warnings.
install.logwas not removed from the MySQL Enterprise Service Manager root directory after a complete uninstallation was performed. This occurred if the
install.logwas created by a product update.
The event text for Binary Log Space Exceeds Specified Limit events reported the disk space used in bytes. As of this release, the disk space used is reported in the more readable gigabytes (GB) and megabytes (MB).
On older, slower systems, the agent startup process reported that the agent had not started, even though the agent was running. This was caused by the agent taking longer to start than the defined timeout on the agent start process.
As of this release, the agent startup timeout is changed from 60 to 90 seconds, thereby allowing for old or under-resourced systems.