4500 VSS. How to add a secondary standalone switch to a single VSS member without interruption?

[Vadym.BYELYAYEV]

Hello all,

Is it possible to convert a 2nd standalone switch to a single VSS (which does not have provisioned info about the second switch) without interrupting the service?

I have searched through all the documentation but it does not explain at a lower level regarding how the VSL sync occurs, so I am a bit confused..

Basically:

1. I have a customer that once had a VSS

2. The standby switch got into rommon mode and lost all the VSS config. The physical cables of all the access layer switches were moved to primary switch. (2nd does not have anything, this is why it is critical)

3. However, the Active switch got its VSS config rebuilt (it was not left as is), so it sees itself as an active switch, but does not have any information regarding the second chassis, so I could use the procedure which is called "replacing the supervisor on 6500", which simply takes the startup-config, loads it on the standby, changes the switch ID on standby and reloads the second chassis..

4. I cannot use it, because the Active switch in its config does not have the physical ports of the second chasis, so I can provision the second switch with this info so it auto-restores the VSL.

My question is, is it possible to somehow equal the config on both of the switches without rebooting the Active one?

And also, if I convert the second switch in VSS, will it go dual-active if for some reason the VSL does not come up? Will the Active reload itself if finally the VSL is rebuilt, maybe if I manually configure the VSL port-channel after I convert the second chassis in VSS or is it done automatically somehow?

Thanks a lot!!

[Mark Malone]

Hi
Its not possible without interruption during part of the sequence you have to convert the switches to VSS mode this causes a reload basically and puts them into 1 logical unit , even when you break VSS mode you go back through same process to get them into standalone mode , i have a good few sets of these and from experience always have a a window , we have had issu even fail multiple times with different issues and switches going into reload automatically , if the switch has been put back in standalone mode it will need to convert and for that to happen theres a hit unfortunately

[Vadym.BYELYAYEV]

Hello Mark,

Thanks a lot for your answer. Would this procedure be correct then:

1. Configure switch ID, switch domain with priority, Interfaces and port-channels on standby. And convert it to switch virtual even if the rest of config is different from the VSS Active. Per my understanding, the configs that really have to match include only those things mentioned above. The rest of the config will be overwritten by the active, so no need to match absolutely all the config?

2. Do nothing on the Active, it should reload itself?

Is this correct?

[Vadym.BYELYAYEV]

I forgot to mention that I do not perform convertion on both switches. I have an active VSS switch + a standalone one.

However it seems that the Active has some info missing about the ports, which prevents me from doing a simple provisioning to the standby (which according to the documentation is not service affecting, single it joing the standalone to an already functioning VSS system).

[balaji.bandi]

yes i agree with @Mark Malone  with out small downtime you can add in to VSS.

What is the Aim to make VSS ?  why not add new Switch as redundent and make Access/ device dual homed with new Switch ?

just thought ?

[Vadym.BYELYAYEV]

Balaji,

My project consists in restoing the VSS system which consists of an Active VSS single switch  + a standalone switch, which a long ago was part of the VSS system.

The Active switch seems like after the VSS got broken, was re-provisiones with the VSS configuration, but as a single-chassis VSS system.

Now it is time to add the standalone switch to this VSS, however, the procedure apparently cannot be done without affecting the service.

[balaji.bandi]

So you have working VSS, due to some reason the VSS disjoined, and you want to join back to new Device in to VSS.

is the new Device exactly same as working VSS device ? like identical ?

If possible can you post current config.

If it possible to join the new Device in to VSS, But VSS is bit sensitive and if you do mistake there is a chance of downtime.

i know we always like to do with out downtime, but technically always suggest to do this task in maintenance window.

[Vadym.BYELYAYEV]

Balaji,

This is the Active one.

SW1#sh switch virtual

Executing the command on VSS member switch role = VSS Active, id = 1

Switch mode : Virtual Switch
Virtual switch domain number : 1
Local switch number : 1
Local switch operational role: Virtual Switch Active
Cannot execute command on peer supervisor because it is not in SSO Standby state

SW1#sh switch virtual redundancy

Executing the command on VSS member switch role = VSS Active, id = 1

My Switch Id = 1
Peer Switch Id = 2
Last switchover reason = none
Configured Redundancy Mode = Stateful Switchover
Operating Redundancy Mode = Stateful Switchover

Switch 1 Slot 4 Processor Information :
-----------------------------------------------
Current Software state = ACTIVE
Image Version = Cisco IOS Software, Catalyst 4500 L3 Switch Software (cat4500e-UNIVERSALK9-M), Version 15.2(5)E1, RELEASE SOFTWARE (fc1)
Technical Support: http://www.cisco.com/techsupport
Copyright (c) 1986-2016 by Cisco Systems, Inc.
Compiled Tue 22-Nov-16 13:28 by prod_rel_team
BOOT = bootflash:cat4500e-universalk9.SPA.03.09.01.E.152-5.E1.bin,1;
Configuration register = 0x2102
Fabric State = ACTIVE
Control Plane State = ACTIVE
 Peer information is not available because
it is in 'DISABLED' state
Cannot execute command on peer supervisor because it is not in SSO Standby state

SW1#sh switch virtual redundancy           

Executing the command on VSS member switch role = VSS Active, id = 1

RRP information for Instance 1

--------------------------------------------------------------------
Valid Flags Peer Preferred Reserved
Count Peer Peer

--------------------------------------------------------------------
TRUE V 0 0 0

Switch Switch Status Preempt Priority Role Local Remote
Number Oper(Conf) Oper(Conf) SID SID
--------------------------------------------------------------------
LOCAL 1 UP FALSE(N ) 200(200) ACTIVE 0 0

Peer 0 represents the local switch

Flags : V - Valid
In dual-active recovery mode: No

The ports on this Active switch only are of this chassis

GigabitEthernet1/1/1 unassigned YES unset up up
GigabitEthernet1/1/2 unassigned YES unset up up
GigabitEthernet1/1/3 unassigned YES unset down down
GigabitEthernet1/1/4 unassigned YES unset up up
GigabitEthernet1/1/5 unassigned YES unset up up
GigabitEthernet1/1/6 unassigned YES unset down down
...

The standalone does not have the VSS info and it isolated from the network

[balaji.bandi]

I am more intrested current running config. and the one you going to add device proposed config.

[Vadym.BYELYAYEV]

Balaji, I do not think that the running-config can help you to determine what is going to happen if I plug the second switch. It is more about the behaviour and the sync process of the VSS in general.

[balaji.bandi]

Config really matter how you configured. this where if you have miss config you likely to have down time and switch reboot,

Example : 

If you have working VSS, one of the Partner Fails, and you raise a TAC case, they will arrange new kit

that where you join the kit to exiting partner, we assume that config already input as same as old box.

In your case we are not sure you have exiting config. as long as you have exiting config. Your VSS should work as expected.this is technically my point of view.

But again alway suggest to have maintenane window ( you may come up with an issue one time issue, which was i was expressed other post as per my real expereince - because of environment.)

[Mark Malone]

Hi have a look at this it may help , specifically about replacing 4500 failed VSS switch when there is still an active member

Catalyst 4500 Series Switch VSS Member Replacement Configuration Example
https://www.cisco.com/c/en/us/support/docs/switches/catalyst-4500-series-switches/117640-configure-vss-00.html

[Vadym.BYELYAYEV]

Hi Mark,

I restored the VSS last week. This is how it went:

1. I did not have to reboot the single-VSS after I converted the second standalone chassis into VSS. Once I converted the standalone, it rebooted. After this the Active VSS got up the LMP and other protocols and negotiated. I did not have to reboot the Active switch

2. After this I performed the following tests which resulted different from how it is explained in the documentation:

a) Switch Active has 200 and B has 100 as priority. I disconnect all the physical ports that participate in VSL. Standby becomes Active. I reconnect the cables back. The Standby sees the VSL restoration, says "%VSLP-5-RRP_MSG: Role change from Active to Standby and hence need to reload", so basically it does not keep the Active role, this role is "returned" to the previous Active switch..

b) If I do the same, bring the VSL back, but reboot the Active, it becomes the new Standby

c) If I add the fast-hello link and bring the VSL down to avoid the dual-active scenario, according to the documentation, the Standby becomes Active and the exActive enters into recovery until the VSL comes up. After this, the exActive in recovery reboots and comes up as a new standby.. In my case it ocurres the opposite way.. The standby enters into Recovery and the service in never interrupted in the Active.

d) I thought it was because of the priorities. I have set 200 on the standby, rebooted it, so both have 200, but the behavior is the same. The VSS does not have preempt.

[Mark Malone]

Hi
This is from one of the VSS design docs

https://www.cisco.com/c/en/us/td/docs/solutions/Enterprise/Campus/VSS30dg/campusVSS_DG/VSS-dg_ch2.html#wp1064580

Cisco recommends that you do not configure switch preemption for the following reasons:
•It causes multiple switch resets, leading to reduced forwarding capacity and unplanned network outages.
•The VSS is a single logical switch/router. Both switch members are equally capable of assuming the active role because it does not matter which is active—unless required by enterprise policy.