Some time ago I began supporting a busy ATG Dynamo server cluster and began having performance issues. In order to get a search for the root cause, I decided to extend Cricket to monitor Dynamo, since we were already using Cricket for other monitoring. In addition, Dynamo natively supports SNMP, and has it enabled by default.

So here is a Cricket Defaults file you can throw into your Cricket config tree. It has comments which should explain everything, but just to ensure you see the important comments:

• I have stripped the OIDs from my config file, until I decide whether ATG would sue me over publishing those. You can find the OIDs yourself by looking at your Dynamo installation in $DYNAMO_HOME/src/snmp/Dynamo3Mib.mib.
• Database connection pools can be monitored, but you may need to configure Dynamo to enable SNMP collection from your pools, if you use a pool other than the default "ConnectionPool." Furthermore, the JDBC data are presented in a table, which means that the slice number of your table must be appended to each OID. The Dynamo dbPoolServices array is just that, an array, so if you use the default ConnectionPool, it is automatically the first element in the array and accessed as index 1. If you add two more custom pools, they are indices 2 and 3. Hence append ".1" to the OID to access the JDBC pool data for the builtin pool, or append the appropriate index value corresponding to the position in the dbPoolServices array. The dbPoolServices property is configured in /atg/dynamo/server/snmp/DynamoMib.properties (see chapter 8 of the Dynamo Deployment Guide for further information).

This work is now quite old and could benefit from the techniques I have since learned adding support for other SNMP MIBs, particularly instance-mapping the connection pools. If I get back into a Dynamo environment, I probably will.

(C) 2000, Michael Han
$Id: dynamo.html,v 1.6 2003/06/28 22:01:43 mikehan Exp $