The CPU overload control policy is not configured on the MA5600T. As a result, not all
packets necessary for filtering are filtered out during the running of the live network and
therefore the equipment CPU is overloaded in specific scenarios.
[Problem Description]
Trigger conditions:The problem may occur if batch automatic dial-up is enabled
for the users connected to the MA5600T on the live network in the following scenarios:
1. A reset or upgrade of the system. All users under the system will automatically dial up
after the system starts from a reset or upgrade.
2. Cutover or upgrade of the upper-layer switch, which will make all users go offline
and therefore the automatic dial-up function is triggered.
3. Malfunctioning of the upper layer switch. As a result, upstream services are interrupted and
all users are forced
to go offline, triggering the automatic dial-up function.
For PON services, if a large number of ONTs concurrently go online or offline, a lot of alarms
will be reported to the MA5600T and therefore the CPU of the MA5600T may be overloaded.
Symptom:
If packets flood the CPU of the MA5600T, the CPU usage may reach 100%, causing service
board reset and NMS unreachablility.
[Root Cause]
Generally, the CPU is able to process a limited number of service packets per second.
In case of packet flooding, the CPU usage increases. After the CPU usage reaches 100%,
the system failed to handle new packets from service boards and the communication
between the control board and service boards is interrupted. Also, the device fails to be
managed by the NMS and fails to be pinged. To ensure proper operation of the CPU,
packets must be controlled according to the CPU usage so that important packets can be
processed and the device can be properly managed.
[Impact and Risks]
The system is unreachable and services are interrupted.
[Measures and Solutions]
Currently, required patches for preventing CPU overload have been released to ensure
that the devices are functioning properly in case of concurrent massive dialup or packet
flooding:
V800R52SPC020: The number of PPPoE and DHCP dial-up packets is under
control according to the CPU usage.
V800R62SPC118 and V800R007C00SPC307: Packets are dynamically controlled if
the CPU usage keeps being higher than 80%, ensuring that the packets and tasks with
high priority can be scheduled and the CPU usage is not too high.
You are advised to upgrade the involved versions on the live network to versions later
than the versions above.
[Rectification Scope and Time Requirements]
None
[Notice Expiration]
This notice automatically expires after related patches are loaded.
More blog:
No comments:
Post a Comment