Functionality Added or Changed
Several new collections, rules, and graphs were added in support of the Thread Pool plugin.
The following new rules were added:
Thread Pooling Not Enabled: Recommends to enable thread pool when concurrent activity is high.
Thread Pool Stall Limit Too Low:
Recommends settings for the
thread_pool_stall_limit variable based on
the stall rate, which you can see in the new
Thread Pool Stall Rate graph).
The following new graphs were added:
Thread Pool Current Threads: Current thread counts by type.
Thread Pool Statement Activity: Statement activity by type for the past interval.
Thread Pool Stall Rate: The percentage of threads stalled in the last interval.
Thread Pool Queue Age: The current longest wait time in the thread queues.
Thread Pool Queues: The number of statements within each queue waiting to run.
The following rules were added or changed:
Updated the Too Many Concurrent Users rule, to kick in at the same time as recommending the Thread Pool plugin (now at 32).
Updated Security rules to take the new External Authentication plugins into account.
Added the Server Has No Locally Authenticated Root User rule. Each server should have at least one locally authenticated administrative user.
Added all of the new custom and variables items for the new plugins.
The Replication tab has been enhanced with columns, icons, and server tree that highlight more aspects of replication status and topology.
Bugs Fixed
When the Agent connected to an IPv6-enabled MySQL 5.5 server on the same machine, the local nature of the connection was not recognized. This issue could lead to problems collecting data items related to the operating system and replication. (Bug #13365142)
When using LDAP authentication with the Map LDAP Roles to Application Roles feature, after switching the user to not be authenticated via LDAP, their role could not be changed. (Bug #13253829)
MEM Incorrectly displays notification details for an alert the
first time an alert goes out for When notification details were
added to a rule after the rule was scheduled, the popup for the
alert details displayed No Notification Set,
even though the notifications were sent out. Now the popup for
the alert details shows the correct notification settings.
(Bug #13244784)
Added a new Agent configuration
option:--aggr-max-request-body-size, with a
default value of 1048576 (1 MB). This option limits the maximum
size of requests sent to the
Aggregator, to avoid
potential out-of-memory errors on the system running the
Agent (and its Aggregator
component).
(Bug #13112696)
The expression parser for rules has a new function,
IF: IF
( returns
either condition,
true_expression,
false_expression)true_expression or
false_expression, depending on
whether condition is true or false.
This function uses short-circuit evaluation, so only one of the
return expressions is evaluated.
(Bug #13102893)
SNMP closed traps are now sent correctly when
MySQL Enterprise Monitor events are auto-closed.
(Bug #13035347)
When doing an upgrade install of the Agent, the values of the
parameters agent-DNS-cachetimeout and
agent-IP-version are preserved in the
mysql-monitor-agent.ini configuration file,
rather than being reset to their defaults.
(Bug #13016321)
A new rule (Account Requires Unavailable Authentication
Plugins) checks whether some accounts on a MySQL
server require any of the authentication plugins introduced in
MySQL 5.5.16, when the applicable plugin is not loaded or not
enabled in the server.
(Bug #13013376)
A new rule (Cluster Has Stopped) and new
graph (Cluster Data Nodes in Ready State)
help to determine if a MySQL Cluster system is entirely
unavailable.
(Bug #13036986)
This fix addresses a possible hang when connecting via MySQL Enterprise Agent Proxy Service using a user authenticated with the Windows Native authentication plugin. (Bug #13019991)
The MySQL Enterprise Monitor upgrade installer now automatically handles the
case where the datadir
configuration option for the
repository database using
the bundled MySQL server was
switched to a non-default location, such as a disk partition
with more space. Previously, this configuration could not be
upgraded automatically, and the resulting error message could
erroneously report that your bundled MySQL server was a
“remote instance”.
(Bug #12999301, Bug #12860005)
The Dashboard could incorrectly list MySQL servers on 64-bit Windows systems as being 32-bit systems. (Bug #13007347)
A number of replication-related items were added to the Replication tab:
Slave_received_heartbeats.
rpl_% configuration options and
Rpl% status variables. (In particular, to
designate the asynchronous or semi-synchronous nature of
each replication server.)
slave% configuration options and
Slave%status variables.
This information is represented on the Replication tab as icons in the Server Tree and as fields in the replication info popup window for each server.
The following graphs were also added:
(Bug #12660632, Bug #61510)
The Dashboard was listing the operating system incorrectly for servers running Oracle Enterprise Linux 6. (Bug #12537844)
The agent would stop monitoring an instance after encountering a
timeout error trying to read the
mysql.inventory table. Now the agent retries
the operation rather than shutting down. The retry behavior is
controlled by the new agent-instance.ini
agent configuration options
connection-init-retry-attempts and
connection-init-retry-delay-ms. For more
information about these new options, see
???.
(Bug #59439, Bug #11766345)
