XGS2220-54 stacking question

otello1

We are testing a stack of 4 XGS2220-54.

The stack is created properly (apparently). So we have 4 switch every one have his stack ID.

1-2-3-4. The 4 is the Primary (the led) into the GUI it appear that is role is MASTER.

When we disconnect the 4 switch (the primary/master), the 3° become instantly the new primary. All the devices in the network continue to ping each other.
From now we see two strange behaviour and we have 2 questions:

1. The first strange thing is that the switch of the stack is not pingable anymore nor reachable via Graphical interfaces. The new primary does'nt have to be reachable? How we have to configure the switch from now if the 4° switch is dead or permanently removed from the stack (i.e. it is broken) ?
   
2. When we reconnect the old primary (the 4 switch) it become instantly the new primary and the led "primary" on the 3 switch turn off.
   
   In all the switches of the stack (1-2-3) after 2 seconds the pwr/sys leds start to blink for a good 60 seconds and for all this time ALL the devices of the network stop ping each other…So the netwotk stop work!! I don't think it is designed to work like this…It is normal? So what kind of convergence is that?
   
3. There is a way to set-up the master manually? Once the stack is created how set the switch priority for every member of the stack? In the menù there is the option "system priority" but for what if the value is only one for all the stack members?

otello1

Regarding the 1° above question we've made more test.

If we stack 4 switch they autoconfigure for example in the following schema:

4 - master (primary)

3 - Backup

2 - Linecard

1 - Linecard

If you are in ring topology there are no problems. You can cut connection at every level without issues.

When you are in chain you can lose connection at any level and the switches in theory have to continue functioning in group. Every group making his own primary switch.

For example if you broke the chain removing connection between 3° and 4° switch you will create 2 groups. The group 1+2+3 and the group with the 4° switch alone.

In this case everything work as expected: the 4° is alone and remain a master. The 1+2+3 group continue working alone promoting the switch in the following way:

switch 3 - From backup —> to master

switch 2 - from linecard —> to backup

switch 1 stay linecard.

In another example if you lose connection between 1+2 and 3+4 there are issues. The group with the 3°+4° switches continues to work because they remain master and backup.

The group with 1°+2° switch stop working. You cannot ping them and they are not reachable from the Graphical User Interfaces so they are unusable…That seems like a BUG

Zyxel_Melen

Hi @otello1,

May I know did you set slot 1 as a force master before you start testing?

The first strange thing is that the switch of the stack is not pingable anymore nor reachable via Graphical interfaces. The new primary does'nt have to be reachable? How we have to configure the switch from now if the 4° switch is dead or permanently removed from the stack (i.e. it is broken) ?

Could you help to check the running config on the new primary? Please access it via USB console. You can any type C cable to connect. The driver can be downloaded from Zyxel download library.

There is a way to set-up the master manually? Once the stack is created how set the switch priority for every member of the stack? In the menù there is the option "system priority" but for what if the value is only one for all the stack members?

About this question, please reference the User's Guide of XGS2220 series on page 208.

You can make one switch as a master by setting force master mode. The system priority is an individual priority for each stacking switch, and it can be changed after the stacking is created.

Zyxel_Melen

Hi @otello1,

1. The first strange thing is that the switch of the stack is not pingable anymore nor reachable via Graphical interfaces. The new primary does'nt have to be reachable? How we have to configure the switch from now if the 4° switch is dead or permanently removed from the stack (i.e. it is broken) ?

This could be because the backup switch (3° switch) hasn't saved the running configuration. When the stacking is disconnected, the backup switch will restore the backup configuration from the memory. If no backup configuration exists, the configuration will be problematic causing the new primary to be unreachable.

Please help to save the configuration before testing.

2. When we reconnect the old primary (the 4 switch) it become instantly the new primary and the led "primary" on the 3 switch turn off.

In all the switches of the stack (1-2-3) after 2 seconds the pwr/sys leds start to blink for a good 60 seconds and for all this time ALL the devices of the network stop ping each other…So the netwotk stop work!! I don't think it is designed to work like this…It is normal? So what kind of convergence is that?

This status is because the stacking system re-elects the master since a stacking switch was added. After the election, the non-master will synchronize with the master, and the service will temporarily go down as a normal recovery process.

The group with 1°+2° switch stop working. You cannot ping them and they are not reachable from the Graphical User Interfaces so they are unusable…That seems like a BUG

This is because their role was linecard. After the stacking cable between 2nd and 3rd switches from being unplugged, the linecard role will not take any action including changing role or remain running configuration, which is a normal behavior since it's not either Master role or Backup role.