Monitoring large file systems
By default, Resource Manager uses the Host Resources MIB to monitor file systems. A defect in the implementation of the Host Resources MIB in net-snmp causes file systems larger than 16TB to report incorrect utilization metrics, such that you might observe file system utilization values greater than 100%.
Note: Resource Manager uses the Host Resources MIB in the ethernetCsmacd template (rather than the UCD dskTable MIB) by default because most of the systems Resource Manager monitors do not have the UCD dskTable MIB enabled.
To work around this deficiency, Resource Manager can instead use the UCD dskTable MIB to monitor file system utilization.
Configuring the UCD dskTable MIB
To use the UCD dskTable MIB, you must modify the configuration of your data sources, thresholds, and graphs in your FileSystem template:
- Create an SNMP data source named dskPercent.
Set the OID of the new data source to:
Modify your thresholds to use the dskPercent data point. Remove any calculations that were associated with the Host Resource MIB data point. The dskPercent data point is reported as an integer from 0 to 100.
- Modify your graphs to use the dskPercent data point and thresholds.
Enable the UCD dskTable MIB on your managed hosts:
Add the following line to your /etc/snmp/snmpd.conffile:
Restart the snmpd daemon.
The UCD dskTable MIB associates different indexes with the file systems than the Host Resources MIB. As a result, you must remodel the devices to fetch the UCD dskTable indexes, and to begin plotting data on the dskPercent graph.