Why SNMP monitors are often not collecting data when using SNMP v3 credential?
When Data Collection stops-
- Check if data collection has stopped for all devices or particular devices,
- Check if any OOM error file is present in folder,
- Check the credential settings,
- Check mib response for that device's oids,
- Check if a monitor or a device is active in PolledData table,
- Check PolledData table whether polling is scheduled properly for that monitors (LASTTIMEVALUE,TIMEVAL columns),
- Execute the below query to check the load (If more monitor's are configured with polling interval is less than 3 minutes then polling is not scheduled as expected)
- SELECT PROTOCOL,PERIOD FROM PolledData GROUP BY PROTOCOL,PERIOD ORDER BY PERIOD
SNMP V3 Issues
- ENGINETABLE - (SNMP API Table - Contains SNMPv3 agent's engine information like engineId, engineTime and engineBoots, these are added while discovery/Query device SysOID/MibBrowser usage)
- USMTABLE - (SNMP API Table - Same as enginetable, this also contains SNMPv3 agent's engine information like engineId, engineTime and engineBoots along with that SNMPv3 user related information also stored as encrypted string)
- SNMPV3CREDENTIALDETAILS - (OpManager Table - Contains credential details credential information given in Credential Page)
- SNMPCREDENTIALMAP - (OpManager Table - Contains MOID and its corresponding associated credential ID of SNMPV3CREDENTIALDETAILS table)
- CREDENTIALMANAGER - (OpManager Table - Contains Credential Id of SNMPV3CREDENTIALDETAILS, along with that credential type)(WMI/SNMPv2/SNMPv3/CLI), credential name
Known Issues and its solution/work around Issue 1 : Device discovered as unknown, given SNMPv3 credentials
- Check whether engineID is duplicated in enginetable/usmtable for any other host
- Proper configuration is done in device end.
- Check with Standalone mib browser.
- Is it special type of device : Some device may not follow RFC standards which we strictly follow. So far we experienced this issue in Dell UPS, APC UPS
- Check if any context name is configured in device end. Some type of device strictly needs context name configuration, and some may not.
- Before discovery ensure that, there should be no entries in USMTABLE and ENGINETABLE for that device
- During discovery select only one working credential, not all credentials in SNMP
Issue 2 : SNMP Polling not happening only for SNMPv3 device
- Check whether the given credential is correct. Check with standalone Mib browser. Don't check with the Test Credential(till 123034)
- Check whether the agent respond for that OID, with SNMPv3 credential
- Try by associating the credential again in the device snapshot page
- Context Name or SNMP view may be configured in device end for that particular OID mib's sub tree, if so context name should be configured in the credential
- Check for timeout configuration
- Whether engineID is duplicated in enginetable/usmtable for any other host
Issue 3 : SNMP Polling is happen for the first time only after device delete and re-add
- Check enginetable/usmtable. If engineID is duplicated two or more devices this will happen,
- Delete that duplicated entries and add engineID in device end for which device's data collection is not happening
- Open snmp.conf file in device end (Linux -> snmpd.conf)
- Add mac address for engineID param -> (ex. engineID 00-09-0F-FE-00-01)
- Delete the device in OPM and discover again.