MA5600T connects lots of ONTs(HG850,HG850A,HG8245). After the uplink network interrupted and restored, lots of Voice and Internet services is not normal, and the version of MA5600T is V800R005C33B118.
When you log in the device remotely and input the command, it will take several minute to run it .
Take off all of 14 GPBC boards and insert GPBC board one by on. The method is like this: after the services in one GPBC board are normal and CPU is low, you can insert another one.
So many Voice subscribers sending the arp request to the gateway, together lots of PPPOE subscrbers dialing, result in CPU high. Normally, these subscribers dial not in the same time. Because of the uplink network interruption, it caused it's in the same time. However if the rate of arp is 600pps for R52 version, the CPU will be higher than 88%; if the rate of PPPOE is 400pps, CPU will be higher than 90%. And the version of R52 does not have the protection to control this scenario so that it will cause the CPU always high and lots of services is not normal.
The version of R52 does not have any protection to prevent the problem of CPU always high , but the version of R7 can support it. The principle is that when it happens, the host will discard dial packets to protect the CPU. The result is some of the subscriber is not normal at the beginning but it will be OK later(If CPU is low, it can dial).