Symptom

The affected device shows high memory usage through the 'MallocLite' process: ------------------ show process memory sorted ------------------ Processor Pool Total: 1445417856 Used: 1302083920 Free: 143333936 PID TTY Allocated Freed Holding Getbufs Retbufs Process 0 0 0 0 991438776 0 0 *MallocLite* In this example the holding value is 991MB of memory. The top allocator is a single function with almos all of the memory assigned, use 'show process memory mlite' and then look into the size value, this example shows a size of 973MB for the PC :556FEDA58000+3642F1F Router#show process memory mlite Process ID: 0 Process Name: *MallocLite* Total Memory Held: 993079576 bytes Processor memory Holding = 993079576 bytes size = 973913248, count = 14839, pc = :556FEDA58000+3642F1F size = 6431936, count = 98, pc = iosd_shr_m_uk9_SNMP_Core_crb:7F78C3FB5000+CB478 size = 4922400, count = 75, pc = iosd_shr_m_uk9_SNMP_Core_crb:7F78C3FB5000+8DD32 size = 4003552, count = 61, pc = iosd_shr_m_uk9_SNMP_Core_crb:7F78C3FB5000+C9172 TAC assistance is needed to determine what process is related to that PC value, the decode of TAC has to point to the SNMP OID cswDistrStackPhyPortInfo (MIB: 1.3.6.1.4.1.9.9.500)

Conditions

TAC assistance is needed to determine what process is related to that PC value, the decode of TAC has to point to the SNMP OID cswDistrStackPhyPortInfo (MIB: 1.3.6.1.4.1.9.9.500) The device is configured as StackWise: Switch#show stackwise-virtual Stackwise Virtual Configuration: -------------------------------- Stackwise Virtual : Enabled And the MIB CISCO-STACKWISE-MIB is polled via SNMP. Use 'show snmp stats oid' to confirm: Switch#show snmp stats oid time-stamp #of times requested OID 09:51:39 UTC Mar 12 2020 1604 cswDistrStackPhyPortInfoEntry.4 09:51:38 UTC Mar 12 2020 1602 cswDistrStackPhyPortInfoEntry.3 09:51:36 UTC Mar 12 2020 1603 cswDistrStackPhyPortInfoEntry.2 09:51:34 UTC Mar 12 2020 1609 cswDistrStackPhyPortInfoEntry.1 09:51:32 UTC Mar 12 2020 400 cswDistrStackLinkInfoEntry.2 09:51:32 UTC Mar 12 2020 796 cswStackPortInfoEntry.2 09:51:31 UTC Mar 12 2020 797 cswStackPortInfoEntry.1 09:51:30 UTC Mar 12 2020 398 cswSwitchInfoEntry.14 09:51:29 UTC Mar 12 2020 399 cswSwitchInfoEntry.13

Workaround

Instead of using snmpwalk or snmp bulkget, use snmp get on a specific OID to avoid memory leak.

Further Problem Description