Re: NM Bridging Support Status

[Thomas Goetz <thomas goetz citrix com>]

I merged the bridge patches into 0.9.4 and wrote some code for my daemon to make use of it and got the following result. Is this error you were expecting or something different?

I plan on continuing to debug this. Thanks for any help.


Aug 22 14:13:08 wolverine dmd[7514]: DEBUG: Adding bridge connection <----- adds bridge connection
Aug 22 14:13:08 wolverine dmd[7514]: DEBUG: Connecting bridge br-eth0 
Aug 22 14:13:08 wolverine dmd[7514]: DEBUG: Connecting wired eth0 <------- adds and activates a connection with eth0 as a slave of br-eth0
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): carrier is OFF
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): new Bridge device (driver: 'bridge' ifindex: 10)
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): exported as /org/freedesktop/NetworkManager/Devices/4
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): now managed
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): device state change: unmanaged -> unavailable (reason 'connection-assumed') [10 20 41]
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): bringing up device.
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): carrier now ON (device state 20)
Aug 22 14:13:08 wolverine NetworkManager[6506]: (nm-device.c:4396):nm_device_queue_state: runtime check failed: (priv->queued_state.id == 0)
Aug 22 14:13:08 wolverine NetworkManager[6506]: (nm-device.c:4396):nm_device_queue_state: runtime check failed: (priv->queued_state.id == 0)
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (br-eth0) starting connection 'br-eth0'
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): device state change: unavailable -> ip-config (reason 'none') [20 70 0]
Aug 22 14:13:08 wolverine NetworkManager[6506]: (nm-device.c:2343):nm_device_activate_schedule_stage3_ip_config_start: runtime check failed: (state >= NM_DEVICE_STATE_PREPARE && state <= NM_DEVICE_STATE_NEED_AUTH)
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (br-eth0) Stage 3 of 5 (IP Configure Start) scheduled.
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (eth0) starting connection 'eth0'
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (eth0): device state change: disconnected -> prepare (reason 'none') [30 40 0]
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (eth0) connection 'eth0' waiting on dependency 'br-eth0'
Aug 22 14:13:08 wolverine dmd[7514]: DEBUG: state changed for eth0 
Aug 22 14:13:08 wolverine dmd[7514]: eth0: state changing Offline (30) -> Connecting (40)
Aug 22 14:13:08 wolverine NxTopEngineUI: DEBUG: Event type = networkUpdate
Aug 22 14:13:08 wolverine kernel: [18381.056595] br-eth0: port 1(vif1.0) entering forwarding state
Aug 22 14:13:08 wolverine kernel: [18381.056615] br-eth0: port 1(vif1.0) entering forwarding state
Aug 22 14:13:08 wolverine dmd[7514]: DEBUG: Added bridge: br-eth0 
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (br-eth0) Stage 3 of 5 (IP Configure Start) started...
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (br-eth0) Stage 3 of 5 (IP Configure Start) complete.
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (br-eth0): device state change: ip-config -> activated (reason 'none') [70 100 0]
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (eth0) connection 'eth0' dependency ready, continuing activation
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (eth0) Stage 1 of 5 (Device Prepare) scheduled...
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (br-eth0) successful, device activated.
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (eth0) Stage 1 of 5 (Device Prepare) started...
Aug 22 14:13:08 wolverine NetworkManager[6506]: <error> [1345644788.885331] [nm-system.c:2231] nm_system_bridge_attach(): (br-eth0): interface is not a master  <------- is this the error you're expecting?
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (eth0): device state change: prepare -> failed (reason 'none') [40 120 0]
Aug 22 14:13:08 wolverine NetworkManager[6506]: <warn> Activation (eth0) failed.
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> Activation (eth0) Stage 1 of 5 (Device Prepare) complete.
Aug 22 14:13:08 wolverine dmd[7514]: DEBUG: state changed for eth0 
Aug 22 14:13:08 wolverine dmd[7514]: eth0: state changing Connecting (40) -> Failed (120)
Aug 22 14:13:08 wolverine dmd[7514]: DEBUG: disconnecting: eth0 (de697dae-4dd3-4759-827b-05b45dbbcb80) 
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (eth0): device state change: failed -> disconnected (reason 'none') [120 30 0]
Aug 22 14:13:08 wolverine NetworkManager[6506]: <info> (eth0): deactivating device (reason 'none') [0]

---

nmcli crashes after network manager gets in this state:

root@wolverine:~# nmcli nm

(process:6473): GLib-GObject-WARNING **: cannot retrieve class for invalid (unclassed) type `gchar'

(process:6473): GLib-GObject-WARNING **: g_object_new_valist: object class `gchar' has no property named `dbus-connection'

(process:6473): GLib-GObject-WARNING **: cannot retrieve class for invalid (unclassed) type `gchar'
Segmentation fault (core dumped)

---

here's what the connection file looks like:

[connection]
id=br-eth0
uuid=4fd3cfe8-200b-453f-bbb6-1ffa8fd88175
type=bridge

[ipv6]
method=ignore

[ipv4]
method=disabled

[bridge]
interface-name=br-eth0
stp=yes

On Aug 14, 2012, at 2:23 PM, Thomas Graf <tgraf redhat com> wrote:

> On Tue, 2012-08-14 at 12:07 -0400, Pavel Simerda wrote:
>>> On Tue, 2012-08-14 at 10:37 -0400, Thomas Goetz wrote:
>>>> What is the status of bridging support in Network Manager? Will it
>>>> be in an upcoming release? Is it available somewhere to try?
>>> 
>>> Patches to support bridging have been posted to the mailing list but
>>> have not been merged yet.
>> 
>> AFAIK they were not fully working. The configuration wouldn't work
>> on NetworkManager startup, which is crucial for most bridging use
>> cases. Have I missed some news?
> 
> It's a generic activation related bug that prevents auto activation.
> Dan mentioned that he would rework the activation code so I did not
> want to invest too much time before that happens.
> 
> ~Thomas
>