From: Oliver G. <oli...@ou...> - 2007-09-04 11:48:56
|
Hi, I've been working on CVS NetDisco, and in particular the Vlan support. This also strays into SNMP::Info land of course, with CiscoVTP.pm. There are two patches initially, which I'd like comment on (one for NetDisco and one for SNMP::Info). The NetDisco patch fixes two things: * Actually scrubs out Vlan data from the DB (device_port_vlan table) before re-generating, which wasn't happening. * Shows trunk Vlans in the web GUI as well as non-trunk (which was already in CVS). This follows my own style of placing them in parentheses, like so: "1, (7,120,634,720,990,991,1000,2999)" where the native Vlan is not in parentheses. There is a new config var "trunk_threshold", but read the patch for explanation of what that does. http://tinyurl.com/373ufk The SNMP::Info patch also has two features and one caveat: * i_vlan_membership() returns data about extended Vlans (i.e. up to 4095 rather than just 1023). * The OID munging code has been factored out into a subroutine so that the i_vlan_membership() and set_*() subs can all share it and operate on extended Vlans. * However, the set_*() subs have NOT been patched to use it. They have had the minimal patch to just continue using Vlans 1..1023. http://tinyurl.com/38ru37 I'd be grateful if folks could take a look and say whether they like the way this is being done, or would prefer a different way. To my mind this is just completing on the excellent start which is already in CVS, so there's not much to it. I'd appreciate guidance on what to do about the set_*() subs as well. I don't quite understand what the difference between vtp_trunk_vlans($partial) and vtp_trunk_vlans($ifindex) and vtp_trunk_vlans($new_list, $ifindex) really is. FYI - I am also working on patching the Device Search to search for devices with ports carrying a Vlan, but this isn't finished yet (because the SQL query isn't a trivial patch). regards, oliver. -- Oliver Gorwits, Network and Telecommunications Group, Oxford University Computing Services |