One NE5000E use GE interface connect with BRAS directly, we configure a lot of sub-interface that more 13 because the service in BRAS is mulriple, each sub-interface use OSPF protocol(configure silent-interface at the same time). The user which under the BRAS complain us that the speed he access the internet is too little. We telnet to our NE5000E from remote and found that our CPU usage more than 85%
The version of NE5000E is 5.10-02J.08
There're a lot of information that one GE port UP/DOWN in NE5000E log file.
1, Check the usage of CPU use follow command
Slot CPU Usage Memory Usage (Used/Total)
17 MPU(Master) 87% 19% 172MB/889MB
1 LPU 7% 32% 126MB/390MB
2 LPU 7% 32% 126MB/390MB
5 LPU 11% 30% 119MB/390MB
6 LPU 17% 37% 147MB/390MB
18 MPU(Slave) 2% 14% 128MB/889MB
2, Check the log file, found there's a lot of information show that one interface stage up/down again and again.
3, Shoudown the interface which UP/DOWN again and again, the CPU usage change to be normal.
4, We found the optical module between NE5000E and BRAS is not same, that should be the right reason that the interface stage UP/DOWN, when we change the optical module to be same, and then everything change to be ok.
Too much sub-interface configured in NE5000E GE sub-interface, and use OSPF in each sub-interface, when these 13 sub-interfaces surge again and again, lead to OSPF mode in MPU board change the stage of the interface again and again, so the OSPF mode in MPU should be too busy that the CPU usage change to be too high.