[Rack] Our managed switches

Jonathan Lassoff jof at thejof.com
Thu Apr 22 06:27:28 UTC 2010


Excerpts from Rubin Abdi's message of Wed Apr 21 20:18:14 -0700 2010:
> I'm noticing on the Dell switch that it's issuing DHCP leases much
> slower then over wifi. I've got a feeling that it's not allowing traffic
> over the port until it "activates" it or some such.
> 
> The Cicso switches seem to be in the same boat unfortunately.
> 
> So one of the nice things about having that shitty not managed switch
> was it didn't take ages to activate a port.

Which port is being problematic?

<sidenote>
Weird... dnsmasq on r00ter seemed locked up somehow. I couldn't resolve
switch1.noise, switch2.noise, or switch3.noise (the managed switches).
A quick kill -HUP, and restart, and DNS queries to 172.30.0.1 (r00ter)
are resolving again.
</sidenote>

I checked the configuration on switch2.noise and switch3.noise (the two
managed Ciscos), and portfast (the feature that puts ports into the
forwarding state right away) seems to be enabled on all the ports I
setup and know are going to end hosts.
Portfast is potentially dangerous since if you cable up a loop between
switches, it'll just forward packets in an infinite loop, so I wanted to
be cautious and only enable it on ports that I knew ran to end hosts.

If you're sure you're cabling up to an end host and hate the default
behavior that takes too long:

 - telnet to the switch
 - guess the username and password
 - run these commands (you can abbreviate as long as the command is
   unambiguous):
  configure terminal
   interface FastEthernet0/{the port number you want}
    spanning-tree portfast


Seems like the process is similar for the Dell:

 - telnet to the switch
 - guess the username and password
 - run these commands

  conf
   interface ether 1/e{port number}
    spanning-tree portfast



More information about the Rack mailing list