[Xorp-users] VLAN Interfaces

[Joe Coco]

Hi Ben,

Some responses inline.


> For this particular error, it might be using eth1's info (name, or perhaps if-index)
> instead of eth1.100

I was thinking that. Unfortunately, as I said before, I've not done much with netlink so I'm having
trouble understanding some of this code in the brief time I've looked at it. 


> It's also possible that there is a race and eth1.100 either isn't in the kernel yet, or
> hasn't been discovered by xorp.

I don't think that is it, see below, but I could be wrong.

> What version of xorp are you using?  Latest xorp.ct from git tree?

1.8.2b from Dec 8, 2010 https://github.com/greearb/xorp.ct/downloads

I uncommented some debugging print statements and added a few of my own so I could watch xorp 'step through it' when I hit commit:

(here is right after the commit)

end deltas
end deletions
[ 2011/02/11 14:34:36.857021  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth0  tree: system-config
[ 2011/02/11 14:34:36.857060  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth1  tree: system-config
[ 2011/02/11 14:34:36.857084  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth2  tree: system-config
[ 2011/02/11 14:34:36.857107  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth3  tree: system-config

-JC- Set name-type for VLAN

-JC- Created vlan device: eth1.100

[ 2011/02/11 14:34:36.860771  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth0  tree: system-config
[ 2011/02/11 14:34:36.860822  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth1  tree: system-config
[ 2011/02/11 14:34:36.860868  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth1.100  tree: system-config
[ 2011/02/11 14:34:36.860899  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth2  tree: system-config
[ 2011/02/11 14:34:36.860940  WARNING xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_parse_netlink_socket.cc:330 nlm_cond_newlink_to_fea_cfg ] newlink, interface: eth3  tree: system-config
[ 2011/02/11 14:34:36.861113  ERROR xorp_fea:7164 FEA fea/data_plane/ifconfig/ifconfig_set.cc:661 push_vif_address ] Failed to configure address because of device not found: IfConfigSetNetlinkSocket::add_addr: check_nl_req: $


So it appears it does in fact see the eth1.100 interface prior to trying to push the vif address. 



Thanks!

-- Joe