[j-nsp] How to upgrade M20 RE's with least downtime

[Paul Goyette]

I can't say from actual experience what the hit will be, but
going from such diverse versions is probably going to cause
a lot pain!  I would not be at all surprised if upgrading one
RE to 6.3 starts sending "unexpected" stuff to the down-rev
software on the other RE, causing it to do unexpected stuff.

I would strongly suggest that you plan on a complete reboot,
possibly multiple reboots, as a way to estimate the down-time.

-----Original Message-----
From: juniper-nsp-bounces at puck.nether.net
[mailto:juniper-nsp-bounces at puck.nether.net]On Behalf Of Bill Petrisko
Sent: Friday, November 12, 2004 4:34 PM
To: Harry Reynolds
Cc: juniper-nsp at puck.nether.net
Subject: Re: [j-nsp] How to upgrade M20 RE's with least downtime


On Fri, Nov 12, 2004 at 03:27:33PM -0800, Harry Reynolds wrote:
> *If* this is within a major version, I have heard that you can get great
> results by:
> 
> 1. Enable Graceful restart to minimize/eliminate forwarding plane
> disruption
> 2. Upgrade backup RE
> 3. Manually switch (here is where graceful restart comes in)
> 4. Upgrade primary RE
> 5. If step 3 worked well, switch back
> 
> HTHs. Please note Juniper does not claim support for hitless upgrades,
> but this procedure has been known to result in pretty much that. All
> bets are off if you are crossing major JUNOS software revisions. You
> also need to enable graceful restart on peer nodes for it to buy you
> anything.

Ok, this is going from 5.6 (no graceful anything) to 6.3.

I assume I can still follow the above procedure, but will take
a hit when forcing a RE switchover?

What is the hit time?  (Not including the time for BGP/ISIS/etc
to reconverge.)

bill

_______________________________________________
juniper-nsp mailing list juniper-nsp at puck.nether.net
http://puck.nether.net/mailman/listinfo/juniper-nsp