Symptom
Control-plane protocols are taken down on NCS5500 (e.g. LACP, ISIS, PIM, BGP, etc) running 7.1.2. The "show processes blocked location all" CLI command reports the mibd_infra process blocked on "mibd_infra ciscoNetsyncmib_mutex_lock:210" on the Route Processor as illustrated in the following example:
RP/0/RP0/CPU0:NCS-55A2#show proc block location all
Mon Apr 11 11:09:27.815 CEST
node: node0_RP0_CPU0
------------------------------------------
Jid Pid Tid ProcessName State TimeInState Blocked-on
0 29897 29897 sh_proc_ng_blocked Reply 0000:00:00.0072 3281 procfs_server
288 7498 31403 mibd_infra Mutex 0048:55:27.0556 0x7fe75fdff220 7625 mibd_infra ciscoNetsyncmib_mutex_lock:210
288 7498 31456 mibd_infra Mutex 0048:55:24.0446 0x7fe75fdff220 7625 mibd_infra ciscoNetsyncmib_mutex_lock:210
288 7498 31018 mibd_infra Mutex 0049:04:13.0762 0x7fe75fdff220 7625 mibd_infra ciscoNetsyncmib_mutex_lock:210
201 5117 5337 lpts_fm Reply 0000:00:01.0954 3833 lpts_pa
node: node0_0_CPU0
------------------------------------------
Jid Pid Tid ProcessName State TimeInState Blocked-on
RP/0/RP0/CPU0:NCS-55A2#
The following syslog message is also symptomatic of the potential problem:
mibd_infra[295]: %L2-FSYNC-4-MIB_DLL_ERROR : Frequency Synchronization MIB DLL error: Failed to lock mutex: Resource deadlock avoided
Conditions
The problem has been observed on NCS-55A2-MOD-S configured with PTP-1588 and Synchronous Ethernet (SyncE) protocols. The SNMP polling of the ciscoNetsyncmib MIB (OID 1.3.6.1.4.1.9.9.761) contributes to the problem.
Workaround
Configure an SNMP view to restrict/deny access to the ciscoNetsyncmib MIB.
Example:
snmp-server view SNMP-view1 1 included
snmp-server view SNMP-view1 1.3.6.1.4.1.9.9.761 excluded
