Symptom

If following syslog is seen then customer might be hitting this issue. Below syslog is very high-level. To confirm whether asr9k is hitting this issue, please check the steps mentioned in "Conditions" section. In 4.2.x ========== RP/0/RSP0/CPU0:Nov 21 15:00:17.099 UTC: pfm_node_rp[358]: %PLATFORM-DIAGS-3-PUNT_FABRIC_DATA_PATH_FAILED : Set|online_diag_rsp[233591]|System Punt/Fabric/data Path Test(0x2000004)|failure threshold is 3, (slot, NP) failed: (0/1/CPU0, 0) (0/1/CPU0, 1) (0/1/CPU0, 2) (0/1/CPU0, 3) (0/1/CPU0, 4) (0/1/CPU0, 5) (0/1/CPU0, 6) 4.3.2 and greater: ========================== FABMGR[222]: %PLATFORM-FABMGR-5-FABRIC_TRANSIENT_FAULT : Fabric backplane crossbar link underwent link retraining to recover from a transient error: Physical slot 1

Conditions

Issue is due to fabric retrain in LC ==> RSP direction (RX to RSP). Collect ltraces from RSP's and LC to confirm whether there was fabric-retrain. All 3 lines of output should match. Example log: RP/0/RSP0/CPU0:asr9k#show controllers fabric ltrace crossbar location 0/RSP0/CPU0 | in retrain [or 0/RSP1/CPU0] Nov 21 14:57:17.494 crossbar 0/RSP0/CPU0 t1 init xbar_trigger_link_retrain: destslot:4 fmlgrp:3 rc:0 Nov 21 14:57:17.494 crossbar 0/RSP0/CPU0 t1 detail xbar_pfm_alarm_callback: xbar_trigger_link_retrain(): (4,1,0) initiated Nov 21 14:57:17.499 crossbar 0/RSP0/CPU0 t1 detail xbar_fmlc_handle_link_retrain: rcvd link_retrain for (4,1,0),(1,1,0),0 <=== Indicates RSP is receiving fabric retrain from LC in slot 1. See below for slot mapping info RP/0/RSP0/CPU0:asr9k# slot matching: [In the above output slot 1 is logical. Below data shows how to match logical-to-physical slot mapping ============== Chassis Type : 9010 Physical-Slot Logical-slot ========================== 0 0 1 1 2 2 3 3 4 RSP 5 RSP 6 4 7 5 8 6 9 7

Workaround

No workaround. This is not a hardware issue. No need for RMA.

Further Problem Description