Re: NM refuses to manage a VLAN
- From: Dan Williams <dcbw redhat com>
- To: Pete Zaitcev <zaitcev redhat com>
- Cc: networkmanager-list gnome org
- Subject: Re: NM refuses to manage a VLAN
- Date: Tue, 15 Jul 2014 11:12:42 -0500
On Tue, 2014-07-15 at 09:08 -0600, Pete Zaitcev wrote:
On Fri, 11 Jul 2014 14:45:43 -0500
Dan Williams <dcbw redhat com> wrote:
[root elanor network-scripts]# nmcli c up id vlan-ethgray
Error: Connection activation failed: Device not managed by NetworkManager or unavailable
Could this be the issue?
I tinkered with it a bit more and figured it out. Apparently, I cannot
migrate piecemeal. The parent device must be managed by NM before
a VLAN on it can be managed. Once that was fixed, everything worked
pretty much as expected, with just a few small bugs.
Yeah, NM needs to know certain things about the parent interface before
it can be used with VLANs. Unfortunately, due to the nature of VLANs
(eg, they have a parent interface) that's necessary at this time.
I think the most super annoying part is that address cannot be set
from the shell, because connections are only created with the default
method, dhcp. The nmcli c mod cannot switch two properties at a
I believe we've fixed that earlier this year, so RHEL7 and F21 have the
ability:
nmcli con mod "my-vlan" ipv4.method manual ipv4.addresses
"192.168.1.2/24, 10.10.1.5/8"
time. Also, one cannot set method first and then ipv4.addresses,
ipv6.addresses (I forgot what happens if you try). So, one must
either do the "nmcli c edit xxx" thing, or hand-edit the connection
and then "nmcli c reload".
NM 0.9.9.1+ should allow any number of options with 'con mod':
nmcli con mod "my-wifi" ipv4.method manual ipv4.addresses
"192.168.1.2/24, 10.10.1.5/8" wifi.ssid "blahblah"
Other funny buglet that I remember is that if you create a VLAN
using e.g.
nmcli c add type Vlan ifname ethmain.3 con-name ethmain.3 autoconnect yes dev ethmain id 3
it still won't work, because HWADDR must be set.
I believe we've fixed that earlier this year too; all that should be
necessary these days is *either* the parent interface name (eg, "dev
ethmain") or the parent hardware address.
If you try out F21 or RHEL7, are these issues solved?
Dan
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]