Hi everyone,
The sale of modular equipment in the Enterprise market has grown in recent months, mainly with the possibility of purchasing the NE8000-M8 for ISPs. Along with this increase in sales, we started to have more experiences with this type of equipment, we will share one of them in this post
Some advantages of modular equipment
The name of this type of equipment already indicates one of the advantages of the equipment, it is modular and we have several modules available for use, which can be purchased according to the customer's needs and also provide the possibility of expansion without the need to buy other equipment.
Another advantage of these models is that they usually have two or more processing units, so they work in redundancy and in case of failure of the active unit, the backup unit will take over almost instantaneously.
Multiple processing units
The image below shows a modular device with two IPUs (Integrated NetworK Processing Unit), one of which will work in active mode and the other in backup mode.
Both IPUs have the ability to keep all the equipment's resources running and need to carry out this exchange transparently. Then, when the second card is inserted, the MACs of the active IPU will be cloned to the backup IPU, so in a failure the backup IPU will assume the services and the protocols will continue to run automatically, as it will not be necessary to wait for any timeout or update protocol information, as the equipment's MAC has not been changed, for neighboring equipment nothing happened on the equipment that had the failure.
The problem
The cloning of MACs is important for this scenario, however, an ISP can have several devices on the network and, as necessary, can exchange IPUs between chassis, this process can cause two different devices to have the same MACs. Don't worry, it's not a firmware fault or a manufacturing fault.
It turns out that when a card that was operating as a slave is removed and placed as active in another chassis, its MACs remain the same as the active IPU of the previous chassis.
As these addresses are necessary for network communication and must be unique, if these two chassis are used, problems start to arise, because we have two different equipment with exactly the same MAC. If the equipment is directly connected, communication will simply not occur and if the two are connected to other services, these other services will start to have undesired behavior, which may degrade and even interrupt the services.
The solution
The ideal solution would be for the IPUs to do an ESN check of the chassis, if it is inserted into a new chassis that does not have another IPU, it would adjust its configuration for the factory MACs. However, at the time of writing this post this process does not occur automatically and it is necessary to follow these steps:
Check that the MACs are duplicated:
<HUAWEI> system-view
[-HUAWEI] diagnose
[-HUAWEI-diagnose]display mac-address mpu
2) If the duplication of MACs is confirmed, execute the commands below to reset the MACs of the IPU:
<HUAWEI> system-view
[-HUAWEI] diagnose
[-HUAWEI-diagnose] reset system-mac
Ready, now all MACs will return to the factory address of the product and communication problems will be solved.
Ps: If you are a Huawei engineer, it may be interesting to check the possibility of developing this ESN verification and leave this process automatic;)
Thank you all so much for your attention
