Why SNMP monitors are often not collecting data when using SNMP v3 credential?

Check if data collection has stopped for all devices or particular devices.
Check if any OOM error file is present infolder.
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

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.

Check whether engineID is duplicated in enginetable/usmtable for any other host
Proper configuration is done in device end.
Check with Standalone mib browser.
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 the 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.

Check whether the given credential is correct. Check with standalone Mib browser. Don't check with the Test Credential (Applicable till build 123034)
Check whether the agent responds for that OID, with SNMPv3 credential.
Try associating the credential again in the device snapshot page.
Context Name or SNMP view can 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 configurations.
Check whether engineID is duplicated in enginetable/usmtable for any other host.

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.