Linuxbrige agent downs with tap device deletion timing issue

[Nachi Ueno]

after get the list of tap devices, the tap device may be deleted.
In this case, pyudev raisees an exception which will crash the linux bridge agent.

https://github.com/openstack/quantum/blob/master/quantum/plugins/linuxbridge/agent/linuxbridge_quantum_agent.py#L382
https://github.com/openstack/quantum/blob/master/quantum/plugins/linuxbridge/agent/linuxbridge_quantum_agent.py#L383

[Alex Xu (xuhj)]

reproduce this by added some fake code in LinuxBridgeManager.udev_get_tap_devices as below:

   def udev_get_tap_devices(self):
        devices = set()
        ip = ip_lib.IPWrapper(self.root_helper)
        root_veth, ns_veth = ip.add_veth('tap-test', 'tap-peer-test')
        deleted = False
        for device in self.udev.list_devices(subsystem='net'):
            if not deleted:
                root_veth.link.delete()
                deleted = True
            name = self.udev_get_name(device)
            if self.is_tap_device(name):
                devices.add(name)
        return devices

Then got exception as below:
2013-03-01 17:05:02.559 15417 TRACE quantum File "/home/soulxu/work-code/openstack/quantum/bin/quantum-linuxbridge-agent", line 24, in <module>
2013-03-01 17:05:02.559 15417 TRACE quantum main()
2013-03-01 17:05:02.559 15417 TRACE quantum File "/home/soulxu/work-code/openstack/quantum/quantum/plugins/linuxbridge/agent/linuxbridge_quantum_agent.py", line 654, in main
2013-03-01 17:05:02.559 15417 TRACE quantum plugin.daemon_loop()
2013-03-01 17:05:02.559 15417 TRACE quantum File "/home/soulxu/work-code/openstack/quantum/quantum/plugins/linuxbridge/agent/linuxbridge_quantum_agent.py", line 614, in daemon_loop
2013-03-01 17:05:02.559 15417 TRACE quantum device_info = self.br_mgr.update_devices(devices)
2013-03-01 17:05:02.559 15417 TRACE quantum File "/home/soulxu/work-code/openstack/quantum/quantum/plugins/linuxbridge/agent/linuxbridge_quantum_agent.py", line 371, in update_devices
2013-03-01 17:05:02.559 15417 TRACE quantum devices = self.udev_get_tap_devices()
2013-03-01 17:05:02.559 15417 TRACE quantum File "/home/soulxu/work-code/openstack/quantum/quantum/plugins/linuxbridge/agent/linuxbridge_quantum_agent.py", line 385, in udev_get_tap_devices
2013-03-01 17:05:02.559 15417 TRACE quantum for device in self.udev.list_devices(subsystem='net'):
2013-03-01 17:05:02.559 15417 TRACE quantum File "/usr/lib/python2.7/dist-packages/pyudev/core.py", line 404, in __iter__
2013-03-01 17:05:02.559 15417 TRACE quantum yield Device.from_sys_path(self.context, name)
2013-03-01 17:05:02.559 15417 TRACE quantum File "/usr/lib/python2.7/dist-packages/pyudev/device.py", line 224, in from_sys_path
2013-03-01 17:05:02.559 15417 TRACE quantum raise DeviceNotFoundAtPathError(sys_path)
2013-03-01 17:05:02.559 15417 TRACE quantum DeviceNotFoundAtPathError: No device at '/sys/devices/virtual/net/tap-peer-test'

Is this bug same as you said?

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/23259

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/23504

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/23259
Committed: http://github.com/openstack/quantum/commit/4a1a37ce9f8baf2078a1a44380895d144203a876
Submitter: Jenkins
Branch: master

commit 4a1a37ce9f8baf2078a1a44380895d144203a876
Author: He Jie Xu <email address hidden>
Date: Tue Mar 5 09:52:01 2013 +0800

    Fixes linuxbridge agent downs with tap device deletion timing issue

    Fixes bug 1136264

    Add try/except in deamon loop for preventing agent broken down by
    unhandled exception.

    Change-Id: I99a4f2b485f134b240630a895378d0f61d6382d8