[j-nsp] proposed changes to "clear bgp neighbor"

[Chuck Anderson]

On Fri, Feb 28, 2014 at 01:17:38AM +1100, Julien Goodwin wrote:
> On 28/02/14 00:48, Phil Shafer wrote:
> > Sorry if I'm venturing toward shameless self promotion here, but
> > this really is an area we try to work at.  That's part of the
> > movation for asking if this one specific case is sufficiently
> > irritating to break our own rules.
> 
> But it's not "one specific case"
> 
> clear <foo> <thing>
> 
> Is a horrible outage-causing command for a bunch of things.
> 
> Unless it's been similarly fixed "clear rsvp session" is a great way to
> cause an outage[1] on many carrier networks.
> 
> What about "clear isis adjacency"?
> 
> I'd say review the lot of clear <foo> and fix them all.
> 
> It's *extremely* rare to actually want to reset all sessions on a real
> production router passing traffic in my experience, in all my time I can
> only think of one case where we deliberately used it.
> 
> Any automation relying on this I suspect has far worse problems.
> 
> 1: OK, *I'd* call this an outage, but "short term packet loss event" for
> those with lower standards.
> 

I agree...rather than just fixing this one thing "clear bgp neighbor",
fix them all to require the "all" option.