I recently got rancid set up and running, and after a few hurdles it
seems to generally be working fine.
Issues I'm experiencing/have experienced/lessons learned/things I've noticed:
**clogin is great for adding commands to every switch. Our network
has the switches set up between 10.0.0.1 and 10.0.0.80, so a script
with a simple loop (10.0.0.i for 1 to 80) and reference to a command
list textfile makes it easy to apply commands globally, such as
changing from uptime to datetime for logging, or for issuing a show
interface and then writing the results to a file that can then be
examined (for a port description, to check if a VLAN is in use, or
other uses)
**we are logging messages locally, so the flash directory was always
changing... so I changed rancid to skip the flash directory (by
uncommenting the code to return(1) if $type matched the things it was
supposed to skip and put in 35 so it would skip the 3500s and 3550s
that it was having problems with
**one thing I noticed while running a show interface on all of the
switches was that there was no way to do anything with cluster members
-- sending an "rcommand 1" does not wait for the new prompt to come
up, does anyone know of a way to use clogin while addressing cluster
members, or indeed to add support for rancid's hourly run? my initial
thought on this would be to add another piece of information to the
router.db signifying how many cluster members there are and then
running through the rancid script once for each one
Overall, Rancid has shown itself to be very nice, and being able to
send commands to all the switches automatically is a big bonus!
Chris Stave
Computing and Network Services
Drew University
seems to generally be working fine.
Issues I'm experiencing/have experienced/lessons learned/things I've noticed:
**clogin is great for adding commands to every switch. Our network
has the switches set up between 10.0.0.1 and 10.0.0.80, so a script
with a simple loop (10.0.0.i for 1 to 80) and reference to a command
list textfile makes it easy to apply commands globally, such as
changing from uptime to datetime for logging, or for issuing a show
interface and then writing the results to a file that can then be
examined (for a port description, to check if a VLAN is in use, or
other uses)
**we are logging messages locally, so the flash directory was always
changing... so I changed rancid to skip the flash directory (by
uncommenting the code to return(1) if $type matched the things it was
supposed to skip and put in 35 so it would skip the 3500s and 3550s
that it was having problems with
**one thing I noticed while running a show interface on all of the
switches was that there was no way to do anything with cluster members
-- sending an "rcommand 1" does not wait for the new prompt to come
up, does anyone know of a way to use clogin while addressing cluster
members, or indeed to add support for rancid's hourly run? my initial
thought on this would be to add another piece of information to the
router.db signifying how many cluster members there are and then
running through the rancid script once for each one
Overall, Rancid has shown itself to be very nice, and being able to
send commands to all the switches automatically is a big bonus!
Chris Stave
Computing and Network Services
Drew University