I have a similar issue. Interface discards on my ASA are incrementing by thousands per hour. After a call into Cisco I was advised that the total input discards as a percentage of total input packets was less than 1% and within normal ASA operating standards.
I'd like to see errors and discards represented as percentage instead of a flat number.
All those type of thresholds in NPM need changed to reflect context better. 1000 errored packets out of 10,000 is bad, out of 100,000 I don't care.
But there is a gotcha with discards. Packets discarded due to QoS on a CISCO Router are represented in in/out discards. So you also need to know that a QoS policy is applied to the interface.
Since the QOS-CBP-TARGET MIB isn't supported on most of the routers, you have to handle the MIBs with the QoS info in a special way to find out what Policies apply to what interface. Then, when the router is rebooted, you have to rediscover all of it because the Indexes are never reused after a reboot.