Similar existing Advisors are grouped together in Advisor category. To create a new Advisor, go to Configuration on top menu bar and choose the Advisors menu item, select the Create Advisor button on the General Advisors Control.

Graphs are defined using XML, and then imported into MEM. The new custom graph is displayed with the default graphs, sorted by name on the graphs page.

For an example of how to create a graph, see Section 26.1.9, “Creating a New Graph: An Example”.

The XML elements for creating a graph are as follows:

To create a new Advisor with all-new settings, click the Create Advisor button available on the Advisors page. To create an Advisor similar to an existing one, click the Advisor menu drop-down icon to the left of the Advisor title, and choose the Copy Advisor menu item . You can edit any Advisor element during the copying process, unlike editing an existing Advisor. You can also delete an existing Advisor created by you, click the Advisor menu drop-down icon to the left of the Advisor title, and choose click the Delete Advisor menu item.

You can change the Advisor name, change the Advisor category that an Advisor belongs to, set your own version number, and alter the threshold and frequency of an Advisor.

Most importantly, you can alter an Advisor's expression. Expressions are the core of a MySQL Enterprise Advisors and define the scenario being monitored. An expression can be as simple as a single server parameter or can be complex, combining multiple parameters with mathematical operations.

An expression has two main characteristics:

For example, if you decide that enabling binary logging is a best practice for a production server (as Oracle recommends), then this best practice is violated if log_bin is OFF. Consequently, the expression for the “Binary Logging Not Enabled” rule is “%log_bin% == OFF”. If this evaluates to 1, an event is raised because the best practice is not being followed.

An expression is made up of one or more variables and zero or more mathematical operators. The MySQL Enterprise Monitor product uses the Java Expression Parser. The operators and functions consist of:

For a complete list of the built-in variables used to create Advisors, see Server Option and Variable Reference.

Creating an expression is dependent on variables defined in the Variable Assignment frame. This frame links variables used in the expression field with data gathered from the target MySQL server instance: server status variables, operating system status information, and table information. Variable names are associated with elements in the Data Item drop-down menu. To define more than one variable, click the add row button.

The remaining fields determine the information that you receive in a notification email or the informational pop-up window associated with each advisor.

When MySQL Enterprise Monitor evaluates an expression, it replaces variables with values. For example, part of the expression for the “MyISAM Key Cache Has Sub-Optimal Hit Rate” rule calculates the hit rate as follows:

100-((%Key_reads% / %Key_read_requests%)*100)

If the current value of %Key_reads% is 4522 and the current value of %Key_read_requests% is 125989, the hit ratio is 96.4%:

100 -((4522 / 125989) * 100)

By convention, the Advisors supplied by MySQL use ‘%’ as the delimiter, for example, %Key_reads%. This makes variables more readily identifiable.

Variables can be used in the Description, Advice, Action, and Links attributes of a rule, as well as in expressions. This lets you report the current value of an expression. For instance, you can add the message, “The current value of Key_reads is %Key_reads%.” to the Advice text box. When this is displayed on the screen, the value of %Key_reads% is substituted into the text. If %Key_reads% has a value of 4522, the message becomes “The current value of Key_reads is 4522.”

Each expression has a threshold value that triggers an alert. The THRESHOLD keyword associates that value with an alert level: either an Notice, Warning, or Critical alert.

For example, the expression for the performance advisor, “Thread Cache Size May Not Be Optimal”, is:

100-((%Threads_created% / %Connections%) * 100) < THRESHOLD

The THRESHOLD is set at 95% for an Info level alert, 85% for a Warning alert, and 75% for a Critical alert, producing alerts of three different levels.

Expressions can be straightforward. The expression for “Binary Logging Not Enabled” (one of the Administration alerts) is:

%log_bin% == THRESHOLD

When the result is OFF, only one alert is triggered: a Warning level alert. You cannot just use the expression %log_bin% == "OFF", because this would not test binary logging against a threshold and so would not result in an alert.

Specify precise conditions when each expression should evaluated, to avoid false alarms. For example, the expression for the “MyISAM Key Cache Has Sub-Optimal Hit Rate” rule is:

(%Uptime% > 10800) && (%Key_read_requests% > 10000) && (100-((%Key_reads% / %Key_read_requests%) * 100) < THRESHOLD)

The first part of the expression, (%Uptime% > 10800), delays evaluating this expression until the system has been running for 10800 seconds (3 hours). When a server starts up, it might take a while to reach a state that is representative of normal operations. For example, the InnoDB buffer pool, MyISAM key cache, and the SQL query cache might require some time to fill up with application data, after which the cached data boosts performance.

In addition, if some part of the system is not heavily used, an alert might be triggered based on limited data. For example, if your application does not use the MyISAM storage engine, the “MyISAM Key Cache Has Sub-Optimal Hit Rate” rule could be triggered based on very limited use of other MyISAM tables such as the mysql.user table. For this reason, this advisor has a second part: (%Key_read_requests% > 10000). The rule is not evaluated unless there is plenty of activity associated with the key cache.

Enclose string values within double quotation marks in the Expression or the Thresholds text boxes. For example, the expression for the “Slave I/O Thread Not Running” rule is:

(%Slave_running% == "ON") && (%Slave_IO_Running% != THRESHOLD)

Similarly, the Critical Alerts threshold text box is set to a value of "Yes".

When the expression is evaluated, either "OFF" or "ON" is substituted for %Slave_running%, and "Yes" or "No" for %Slave_IO_Running%, depending on the state of your system. If the slave is running but the I/O thread is not, the expression becomes:

("ON" == "ON") && ("No" != "Yes")

Without quotation marks, this expression would not evaluate to TRUE as it should.

When editing or defining a rule, you can enter text in Wiki format in the Problem Description, Advice, Recommended Action, and Links and Further Reading text boxes. You can format and highlight text and add hyperlinks, using the notation listed in the following table.

So the following Wiki text:

Replication is a __very nice feature__ of MySQL.  Replication can be very
useful for solving problems in the following areas:
* Data Distribution
* Load Balancing
* Backup and Recovery
You can check replication status and start a slave using the following
commands: SHOW SLAVE STATUS \\\\G\\START SLAVE; '{'moreInfo:MySQL Manual: Replication
      FAQ|http://dev.mysql.com/doc/refman/5.6/en/faqs-replication.html'}'

Would be translated into the following HTML markup:

Replication is a <b>very nice feature</b> of MySQL.  Replication can be very
useful for solving problems in the following areas:
<ul>
  <li>Data distribution</li>
  <li>Load Balancing</li>
  <li>Backup and recovery</li>
</ul>You can check replication status and start a slave with the following
commands: SHOW SLAVE STATUS \G;<br/>START SLAVE;
<a href="http://dev.mysql.com/doc/refman/5.6/en/faqs-replication.html"
  target="_blank" >MySQL Manual: Replication FAQ</a>

This section documents the steps to create an Advisor.

To create an Advisor, select the Create Advisor button from the Advisors page. The new advisor page is displayed.

This example creates an Advisor that checks if connections have been killed using the KILL statement and generates an event.

Create your custom rule by following these steps:

After you create the Advisor, schedule it against the MySQL server you want to monitor. For instructions on Configure Advisor, see Table 19.3, “Advisor Edit Menu Controls”.

This section documents the steps to create a graph. Before creating a graph, review the preceding sections of this chapter as Graphs and Rules use similar components. And for an overview that's specific to graphs, see Section 26.1.2, “Overview of Graph Creation”

This example creates a graph that checks and compares disk usage, by displaying the usage and total available disk space over time.

Begin by navigating to the Configuration, Advisors page, and click the Import/Export link. Then note the Custom Rule/Graph/Data Items Import section. This is where the XML file is imported.

A definition to check disk space usage may look like the following:

<?xml version="1.0"?>
<com_mysql_merlin_server_graph_Design>
    <version>1.0</version> 
    <uuid>a57c2bba-ea9b-102b-b396-94aca32bee29</uuid>
    <name>my filename usage test</name>
    <rangeLabel>MB</rangeLabel>
    <series>
        <label>used</label>
        <expression>used_fs/1024/1024</expression>
    </series>
    <series>
        <label>total size</label>
        <expression>total_fs/1024/1024</expression>
    </series>
    <variables>
        <name>used_fs</name>
        <dcItem>
            <nameSpace>os</nameSpace>
            <className>fs</className>
            <attribName>fs_used</attribName>
        </dcItem>
        <instance>/</instance>
    </variables>
    <variables>
        <name>total_fs</name>
        <dcItem>
            <nameSpace>os</nameSpace>
            <className>fs</className>
            <attribName>fs_total</attribName>
        </dcItem>
        <instance>/</instance>
    </variables>
</com_mysql_merlin_server_graph_Design>

Upon successfully loading a graph, a popup notification may say "1 graph imported" in the MySQL Enterprise Monitor User Interface.

This also creates a new Advisor with the same title, which is unscheduled by default. Go to Configuration, Advisors, Graphing to locate and enable this new Advisor.

This graph is displayed on the appropriate graphs page (like every other graph) under the name defined within the definition, which is "my filename usage test" in the example above.

The following XML shows the structure of a custom data collection:

  
<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE classes SYSTEM "items-mysql-monitor.dtd">
<classes>

	<class>
		<namespace>NameSpace</namespace>
		<classname>ClassName</classname>
              <precondition><![CDATA[Add Precondition Query Here]]></precondition>
              <query><![CDATA[Add Main Query Here]]></query>
              <attributes
                  <attribute name="AttributeName1"/>
                  <attribute name="AttributeName2"/>
              </attributes>
	</class>

</classes>

The values in the namespace and classname elements are used as the first two elements of the name.

This section describes the precondition and main queries used to create custom data collections.

  <precondition>
      <![CDATA[SELECT @@version NOT LIKE '5.0%' AND @@version NOT LIKE '5.1%']]>
  </precondition>

SELECT X AS NAME FROM Y

<class>
	<namespace>mysql</namespace>
	<classname>anonymous_user</classname>
	<query><![CDATA[SELECT COUNT(*) AS user_count FROM mysql.user WHERE user='']]></query>
</class>

<query>
<![CDATA[SELECT GROUP_CONCAT('\\\\\n* ', '\'',user,'\'@\'',host,'\'' ORDER BY user, host) 
       as user FROM mysql.user WHERE password='' /*!50507 AND (plugin = '' OR plugin IS NULL 
       OR plugin = 'mysql_native_password') OR (plugin = 'sha256_password' 
       AND authentication_string  = '')*/]]>
</query>

To properly evaluate the data returned by the data collection, assign attributes to the returned values.

Attributes are defined using the following format:

<attributes>
            <attribute name="AttributeName1" counter="true" type="INTEGER"/>
            <attribute name="AttributeName2" counter="false" type="STRING"/>
</attributes>

<attributes>
  <default counter="true" type="INTEGER"/>
  <attribute name="bytes_read"/>
  <attribute name="bytes_written"/>
</attributes>

   
  <attributes>
	<default counter="true" type="INTEGER"/>
	<attribute name="total_wait_time_ms"/>
	<attribute name="total_statements"/>
	<attribute name="max_wait_time_ms" counter="false"/>
	<attribute name="total_errors"/>
	<attribute name="total_warnings"/>
	<attribute name="total_rows_returned"/>
	<attribute name="total_lock_time_ms"/>
</attributes>

   
  <class>
        <namespace>mysql</namespace>
        <classname>rpl_semi_sync_vars</classname>
        <query><![CDATA[
SHOW GLOBAL VARIABLES WHERE
Variable_name='rpl_semi_sync_master_timeout' OR
Variable_name='rpl_semi_sync_master_trace_level' OR
Variable_name='rpl_semi_sync_master_wait_no_slave' OR
Variable_name='rpl_semi_sync_master_enabled' OR
Variable_name='rpl_semi_sync_slave_enabled'
]]></query>
    <resulttype>CLASS_TYPE_1STCOL_ATTRIBUTES</resulttype>
     <attributes>
      <attribute name="rpl_semi_sync_master_timeout" counter="false" type="INTEGER"/>
      <attribute name="rpl_semi_sync_master_trace_level" counter="false" type="INTEGER"/>
      <attribute name="rpl_semi_sync_master_wait_no_slave" counter="false" type="STRING"/>
      <attribute name="rpl_semi_sync_master_enabled" counter="false" type="STRING"/>
      <attribute name="rpl_semi_sync_slave_enabled" counter="false" type="STRING"/>
     </attributes>
</class>

You can write a script to black out a server, rather than opening a web browser and typing entries into the address bar. This section documents a sample blackout script that can be run from the command line.

Create the following file and save it as blackout.pl.

  #!/usr/bin/perl

  use LWP 5.64;

  # USAGE: blackout.pl servicemanager:18443 admin password servername:3306 true

  # $ARGV[0] = management server hostname:port
  # $ARGV[1] = management server username
  # $ARGV[2] = management server password
  # $ARGV[3] = mysqld managed instance server name and port
  # $ARGV[4] = blackout state (true/false)


  my $browser = LWP::UserAgent->new;
  $browser->credentials(
    $ARGV[0],
    'MEM',
    $ARGV[1],
    $ARGV[2]
  );

  my $url = URI->new('https://'.$ARGV[0].'/rest');

  $url->query_form( # And here the form data pairs:
    'command' => 'blackout',
    'server_name' => $ARGV[3],
    'blackout_state' => $ARGV[4]
  );

  my $response = $browser->post( $url );

  if (!$response->is_success) {
    die $response->status_line . "\n";
  }

  if ($response->content =~ /UserUnauthorizedException/ ||
       $response->content =~ /ServerDoesNotExistException/) { 
  	die $response->content;
  }

On Unix systems, use the chmod +x blackout.pl command to make the file executable.

At the command line, enter blackout.pl servicemanager:18443 admin password servername:3306 true.

Check the configuration_report.txt file for the host name and port to use. Specify the correct port for the Tomcat server. Specify the server to black out using the name that appears in the Server Tree, including the colon and port number as shown in the preceding example. Specify the name of a user who is a "manager". A user with "dba" rights cannot black out a server, and the script does not display any error in this case.

To confirm that a server is blacked out, check that its name is greyed out in the Monitor UI. To end the blackout, run the same script, changing the final argument to false.