Cisco SD-WAN vSmart Controller

[RS19]

In my environment I have 2 vSmart Controllers.
Cisco is planning to have the maintenance of the vSmart controllers. 
They have planned for the maintenance 1 vSmart controller at a time ( 1 vSmart controller maintenance in 1 day & the 2nd vSmart controller in other day).

There will be reboot of the vSmart controllers during the maintenance. 
Cisco has mentioned that there will be interruption during this maintenance. 
Would like to understand why there will be interruption during this maintenance. I have 2 vSmart controllers. If 1 is rebooting there is 2nd vSmart controller & there should not be any impact.

Let me know will there be any impact in this scenario. When 2 vSmart controllers are there , my understanding is that there should not be any impact.

[RS19]

Any help ?

[MHM Cisco World]

No impact at All

Check show sdwan control connection

If vedge have control connection to both sdwan' and there is interconnect between vsmart then 

Vedge will get prefix from vsmart that is up

MHM

[RS19]

Thanks. you mentioned both sdwan : is it both vSmart ?

[MHM Cisco World]

Both vsmart 

MHM

[jadicc345]

In your setup with two vSmart controllers, typically, if one controller is rebooting, the other should continue to handle traffic, minimizing any impact. However, the interruption could be due to factors like the controllers being synchronized, which might cause brief disruption while the maintenance is performed. It’s always good to confirm the specifics with Cisco support, as each environment can have unique configurations.

For future reference, if you need any insights on managing this from a different perspective, like advice from infection control experts on minimizing downtime in critical systems, feel free to ask!

[RS19]

Thanks. I am not able to open the link due to some restriction. Will check it out later.
In the mean time wanted to check is there any way to check both the vSmart controllers are in sync ?

[Kanan Huseynli]

Hi,

check OMP table to understand which vSmart-distributed routes are preferred. Begin maintenance from the secondary vSmart.

After that, if you shutdown primary vSmart , secondary routes are re-installed to RIB/FIB and this may result very short outage (maybe even will not be noticed) - maybe this is the reason Cisco said for possible outage.

[RS19]

Thanks. Let me check it. But I have the below basic question.
The cEdge will be already having the routes learnt from the vSmart ( It can be either of the vSmart).
So the routing details are already available in the cEdge. In case if any new routes are learnt then it may have impact since vSmart can not advertise the routes to the cEgde.

But assuming there are no changes to the routing will there be still impact ?  The cEdges will be already having the routes.
Please help me in clarifying this.

[Kanan Huseynli]

Hi,

SD-WAN overlay FIB is based on OMP routing information. In OMP table, you have two routes (at least) for the same destination/remote TLOC. This is why because both vSmart advertise the same route. Normally, vSmart which has lowest systemIP is preferred and its routes are installed to RIB/FIB (in OMP table you see C,I,R). While for the secondary router has only C,R OMP routes (not installed). When you lose the primary (primary means preferred) vSmart , its routes are considered "stale" while the secondary vSmart's routes are "active". So, OMP table is re-populated. This will result re-programming in hardware i.e FIB. This action may result very short outage, however, normally are not recognized by users.

[RS19]

In addition there is Graceful restart timer right. My understanding is that by default it is 12 hours.
So even if both the vSmart controllers are down (say 2 hours) my understanding is that it should not have any impact since the Graceful restart timer value is 12 hours. The cEdge will have the routing informaiton for its data plane to work. I understand that in cas if any new routes or learnt or any change in the routing that will not be updated to cEdge.

But if there are no updates it should not have any impact. Let me know if my understanding is right.

[MHM Cisco World]

You are correct

Graceful restart make cedge/vedge keep prefix even if it CC to vsmart is lost

But we take bad scenario in which upgrade take time long than graceful time' in this case since you have two vsmart and upgrade one by one there is no problem' 

After graceful time is end the cedge will get prefix from other vsmart.

Again you need to check of vsmart are interconnect or not' common are interconnect but you must be sure.

MHM

[RS19]

in my scenario, Cisco is going to perform the vSmart controller upgrade 1 per day. On the scheduled day only 1 vSmart controller is getting upgraded. So my understanding is that there should be no impact.

[MHM Cisco World]

If vsmart's are interconnect 

Idea here 

Cedge1 and cedge2 and vsmart1 and vsmart2 

Cedge1 is connect only to vsmart1 

Cedge2 is connect only to vsmart2 

If you upgrade vsmart1 then cedge1 will lost prefix after 12 hr (graceful) 

But if vsmart1 and vsmart2 is interconnect then vsmart2 will learn prefix from vsmart1 and re-advertise to cedge1 here even if vsmart1 is upgraded the cedge1 not loss it prefix.

You need to take look how your network connect 

MHM

[RS19]

My cEdge is connected to both vSmart 1 & vSmart 2. (Cedge1 and cedge2 and vsmart1 and vsmart2 )
I am not sure how to check if vSmart1 and vSmart2 are interconnected ? Let me know how to check it.