james-server-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Noel J. Bergman" <n...@devtech.com>
Subject RE: Monitoring in James (was Monitoring)
Date Mon, 08 Dec 2003 19:52:59 GMT
Steve Short wrote:
> > I thought, from Chapter 8 of the JMX (JSR 77) specification
> > that SNMP is supported directly by JMX.

> There's a difference between JMX and JSR 77.  JMX is the basic
> management API and is covered by JSR 3.
> JSR 77 is for management of the J2EE platform.  It does include
> remote access via MEJB and SNMP.  Note that the MIBs defined in
> JSR 77 are for J2EE entities and are not oriented towards an
> MTA such as James.

JSR 77 Chapter 6 provides the basic structure for doing statistics, and
includes a (trivial) MIB for JavaMail and MIBs for JDBC, which could be
useful, although we'd want the JDBC related stats to come from DBCP.  I am
not sure what is happening with JSR 138.

> The JMX implementations that I know anything about (Jboss and MX4J)
> include Http and RMI adaptors but do not include an SNMP adaptor.

OK.  I don't know if one will come out of the Geronimo work, which is based
upon JMX with JSR 77 and 88.

> > > This makes the monitor service much simpler, but it puts a lot more
> > > burden on the individual components to set up the hierachy
> > > associations and the aggregation using listeners.
> > Is there a better way?

> I'd prefer to configure the metrics sets and associations in the monitor
> service and then the components simply do a lookup.  This is much less
> intrusive for the components and also means they do not need to know
> what relationship their metric set has to other metric sets in the
> system.

Can you layout the differences for those of us who haven't spent as much
time looking at this as you?

>From what you are saying, if a monitor has a metric that isn't filled in by
a service, it sits empty.  And if a service has a metric that wasn't setup
with the monitor, it can't be monitored.  The primary benefit to driving the
setup through the monitor service is associations between the metrics?

As I understand it, configuring a new metric would mean writing the code in
the service, and then configuring the service monitor, which is really
defining the registry.  Is that correct?

	--- Noel

To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
For additional commands, e-mail: server-dev-help@james.apache.org

View raw message