[j-nsp] EX4200 virtual chassis

[Victor Sudakov]

Doug McIntyre wrote:
> On Mon, Nov 27, 2017 at 08:57:07AM +0700, Victor Sudakov wrote:
> > We are planning to add backup switches in a Virtual Chassis
> > configuration to our existing EX4200s.
> ...
> > If I have a standalone EX4200, should I power it off before plugging
> > in the Virtual Chassis Cable for the first time, and connecting
> > another (powered off) EX4200? 
> > 
> > The links above say only that the desired master should be powered on
> > first, but nothing is said about the possibility of hot plug
> > connection.
> 
> I don't know about "should", but I've never had a problem moving VC cabling
> around, unplugging, plugging. The cables aren't "smart"  (AFAIK), so its not
> like the box can detect if a VC cable gets attached, only if there is something
> on the other end.
> 
> I would expect changing from standalone to virtual chassis would be a
> service affecting change though, so be prepared to stop passing traffic if
> this isn't done in a maint window.

Encouraged by your reply, I hot-plugged the cable and another switch
(in a lab environment) and all went well. The cable is indeed not
smart, there was nothing in /var/log/messages when I attached the
cable and the powered off switch. When I powered the second switch on,
a new configuration was generated automagically (with new interfaces
alongside the old ones) and the VC started working as a single unit.

I was also pleasantly surprised by the fact that the RS232 management
console follows the master chassis no matter which is the current
master (so you don't have to disconnect and reconnect it). Don't know
how they do it but I love Juniper!

-- 
Victor Sudakov,  VAS4-RIPE, VAS47-RIPN
AS43859