Stack Feature Limitations
Stack Setup Precautions
Optical or copper modules that are certified for Huawei data center switches must be used. If high-speed cables or AOC cables are used, you must purchase cables from Huawei.
Optical or copper modules that are not certified for Huawei data center switches or cables that are not purchased from Huawei cannot ensure transmission reliability and may affect service stability. Huawei is not liable for any problem caused by the use of optical or copper modules that are not certified for Huawei data center switches, or cables not purchased from Huawei, and will not fix such problems.
- When connecting stack members through stack cables, ensure that the stack members are directly connected.
- The stack protocol does not support authentication and encryption. If authentication and encryption are required, M-LAG is recommended.
- For security purpose, it is not recommended that member switches in a stack be deployed in different equipment rooms.
It is recommended that you configure DAD for a stack to minimize the impact of a stack split on services.
After a stack is set up, you need to run the save command to save the stack configuration.
- When a stack is running stably, do not shut down stack ports and DAD-configured ports. Otherwise, the stack will split and service traffic cannot be forwarded.
To ensure high network reliability, it is recommended that a stack connect to upstream and downstream devices using inter-device Eth-Trunks (in dual-homing networking).
Before setting up a stack, you are advised to load the same software version and patch version on member switches.
It is recommended that you use two or more links to set up a stack. If only one link is used to set up a stack (this method is not recommended), it is recommended that you run the carrier down-hold-time interval command on stack member ports to configure the delay after which the port Down state is reported. This configuration prevents the stack from becoming unstable because of intermittent port disconnection.
Before the master and standby switches complete batch backup, do not shut down or remove stack links. Otherwise, the standby switch may restart.
To check the batch backup status between the master and standby switches, run the display switchover state command.
Local stack member ports cannot connect to remote common service ports. Otherwise, traffic forwarding may fail or switches restart unexpectedly. Ports on both ends must be configured either as stack member ports or as service ports.
Do not run the flow-control command on a stack member interface to ensure that stack packets are properly forwarded.
When traffic control is configured in a stack and traffic is forwarded between stack member switches, traffic control does not take effect because stack ports do not support the flow-control command.
When flow control is configured in a stack and traffic is forwarded between member devices in the stack, flow control does not take effect because stack interfaces do not support the flow-control command.
- If the port crc-statistics trigger error-down command is not configured before service ports are configured as stack member ports, this command configuration is automatically cleared after stack member ports are restored as common service ports. If this command has been configured before service ports are configured as stack member ports, this command configuration is not automatically cleared after stack member ports are restored as service ports.
When the protocol of a stack member port is Down, the port will not enter the Error-Down state due to CRC error packets.
- Only known unicast and multicast packet statistics on stack member ports are collected, and unknown unicast packet statistics are collected as multicast packet statistics.
When both stacking and multiple other services need to be configured, some services may fail to be delivered because of insufficient ACL resources. For services that can be configured together with stacking, see "Using CSS/M-LAG with Other Services" in CloudEngine 12800, 12800E, 8800, 7800, 6800, and 5800 Series Switches ACL Technical Topics.
Stack Upgrade Precautions
Do not remove or reinstall cards, and optical modules or power cycle the switch during a software upgrade.
Ensure network stability and do not perform other service configurations on the network during a software upgrade.
- If there are MPUs, LPUs, or SFUs on the switch but these cards are not registered, the fast stack upgrade cannot be performed.
During a fast stack upgrade, to shorten traffic interruption, ensure that upstream and downstream devices are dual-homed to stacked switches using Eth-Trunk.
During the fast upgrade of a stack, the SFU reset due to heartbeat loss is a normal situation and does not affect services.
You are advised to configure a backup IP address for the stack management interface before a fast upgrade. This configuration prevents a failure to manage member switches when the stack fails the fast upgrade and splits.
Feature Support in a Stack
Feature |
Feature Support in a Stack |
|---|---|
Mirroring |
In a stack, inter-chassis mirroring is supported. In a stack, stack ports support only mirroring on incoming packets. |
Port split |
If a 100GE or 40GE optical port has been configured as a stack member port, it cannot be split. |
PFC |
The stacking and PFC functions are mutually exclusive. |