Symptom

A Cisco Switch in either an HA or SVL setup can experience a CPUHOG or memory leak in the "cpf_msg_rcvq_process" of the standby supervisor / switch. Either of the following syslogs might be seen: %SYS-3-CPUHOG: Task is running for (XXXX)msecs, more than (2000)msecs (Y/Y),process = cpf_msg_rcvq_process. %SYS-2-MALLOCFAIL: Memory allocation of XXX bytes failed from 0xXXXXXXXXXXXXXXXX, alignment X Pool: Processor Free: XXXXXXXX Cause: Memory fragmentation Alternate Pool: None Free: 0 Cause: No Alternate pool -Process= "cpf_msg_rcvq_process", ipl= 0, pid= XX In addition, on the standby switch console, memory under "STP HA HELPER" may constantly increase: SWITCH# request platform software console attach switch standby R0 SWITCH-stby# show memory all totals ... Allocator PC Summary for: Processor Total Count Name PC XXXXXXXX XXXXX STP HA HELPER :XXXXXXXXXXXX+XXXXXXX

Conditions

Issue seen in IOS-XE 17.2.x and older releases, including 16.12.5b or 16.12.61k VLANs are configured and a port-channel was configured as dot1q-tunnel.Other triggers may be possible if they cause enough STP-related data to be synced down to the standby switch via the CPF (Checkpoint Facility). CPUHOG / memory leak is a result of an internal STP message queue getting congested.

Workaround

Upgrade the device to 17.3.1 or any other later release if possible. Otherwise, downgrade to 16.12.4 or lower

Further Problem Description