Problem Description

One day when the customer cleans the fiber operation of the STM-64MSP ring composed of 10OSN3500s in the network, MSP inversion occurs, and at this time, some branch boards in one site report V5_VCAIS and TU-AIS alarms. The customer detected no impact on user-side services, and the alarms gradually disappeared within 3 minutes. The next night, we did the MSP reversal test together with the customer, and the alarms were still there and the service was normal.

Alarm information

V5_VCAIS, TU-AIS

Processing

According to the conclusion, upgrade the master control and single board of the device to V1R7 version matching. There are no abnormal alarms after the inversion test.

Root Cause

Check the host version of this OSN3500 device: 5.21.13.47p01. This device is equipped with expansion subracks, and the total number of various types of single boards is 72 .

1.Viewing Information The V5_VCAIS alarm indicates that bits 5 through 7 of the V5 byte in the low-order channel VC-12 are all " 1s ". This alarm will affect the service.

2. Capture single-board black armor bb4.log and bb9.log After R&D analysis.

3. When the number of alarms reported to the host at the same time by the veneer through the veneer mode is too large (more than 1024 ), there will be alarm queue overflow phenomenon, and there is alarm queue ID overflow printing on the host Telnet, resulting in the discarding of some of the alarms of the END message (due to the reversal of reuse segments at that time, the veneer will have instantaneous business interruption and report a large number of alarms, but the actual business is only an instantaneous interruption). The actual service was only interrupted instantaneously, which did not affect the service and the reversal test. In addition, this network element is a very large number of single boards, alarm queue overflow possibility is very large). 4. This alarm must rely on the host of the 1-minute calibration event calibration, through the 3-minute alarm after the end of the end off (so see the reported alarms are within 3 minutes of the end of the host in the 3-minute calibration test, to filter out the filter).

5. Positioning analysis, the proposed single board with multiple packets to report alarms to the host, the single board will be packaged into 64 alarms for a packet reported to the host, and the host's message queue is the length of 1024, you can greatly improve the host's processing power, the phenomenon disappeared through the test, can be generated normally can be the end of the normal.

6. Conclusion: Because R1 ~ R6 did not modify the single-board software with multiple packets to report the version of the alarm, while the V1R7 software version has been resolved single-board multiple packets to report. Therefore, it is necessary to upgrade the single board to the V1R7 version and host matching.

Recommendations and Summary

No