Bug 987813

Created attachment 777647 [details]
host-deploy logs

after some investigation I've seen that the failing seems due to missing ifcfg-em1 file in network-scripts directory.
The interface was up thanks to networkmanager but the file for the legacy network script was not there.

creating the ifcfg-em1 file allow to add the host without failures.

On the host there is vdsm-4.12.0-0.1.rc3.fc19.x86_64 from ovirt-beta repo

Created attachment 777714 [details]
supervdsm logs from host

Created attachment 777715 [details]
vdsm logs from host

Created attachment 777716 [details]
engine log

Looking at the first getVdsCaps reveals that the cfg{} element is empty, while the engine uses to read the boot protocol information from it:

'nics': {'em1': {'netmask': '255.255.255.0', 'addr': '192.168.1.107', 'hwaddr': 'b8:ca:3a:76:9a:43', 'cfg': {}, 'ipv6addrs': ['fe80::baca:3aff:fe76:9a43/64'], 'speed': 1000, 'mtu': '1500'}},

Vdsm still requires ifcfg-based host configuration to function properly, and I do not see this changing for ovirt-3.3  :-(

I'm confused - Why is this an RFE?

Isn't this a bug? Does anyone know why the ifcfg file wasn't created during the initial setupNetworks, as it should have?

Some flavors of Fedora installation make without the legacy ifcfg files. We've never claimed to support something like that, hence the "RFE".

We should make without the ifcfg, too, and report the values from the system (that's easier for BONDING_OPTS, a bit harder for BOOTPROTO).

*** Bug 1057390 has been marked as a duplicate of this bug. ***

is this still targeted for 3.4?

No, it's not done yet.

bootproto4 should be reported for all devices. Gateway should also be reported just like 'addr' and 'netmask' are.

How should we verify it?

How this should be verified ?

(In reply to Michael Burman from comment #17)
> How this should be verified ?

Install Vdsm on a host with networking - but no ifcfg files, as suggested in comment 0, or even by removing ifcfg files and restarting vdsm.

Vdsm should keep reporting sane values in getVdsCaps, and Engine should be happy with the state of the host.

Dan, and what about reboots ? 
It should survive reboots? in such case..? because i'm getting this -->

Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/vdsm/netinfo.py", line 461, in _getNetInfo
    data.update({'ports': ports(iface),
  File "/usr/lib/python2.7/site-packages/vdsm/netinfo.py", line 187, in ports
    return os.listdir('/sys/class/net/' + bridge + '/brif')
OSError: [Errno 2] No such file or directory: '/sys/class/net/net-3/brif'
restore-net::INFO::2015-08-23 17:52:04,002::netinfo::482::root::(_getNetInfo) Obtaining info for net net-2.
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/vdsm/netinfo.py", line 461, in _getNetInfo
    data.update({'ports': ports(iface),
  File "/usr/lib/python2.7/site-packages/vdsm/netinfo.py", line 187, in ports
    return os.listdir('/sys/class/net/' + bridge + '/brif')
OSError: [Errno 2] No such file or directory: '/sys/class/net/net-2/brif'
restore-net::INFO::2015-08-23 17:52:04,003::netinfo::482::root::(_getNetInfo) Obtaining info for net net-1.
Traceback (most recent call last):
  File "/usr/lib/python2.7/site-packages/vdsm/netinfo.py", line 461, in _getNetInfo
    data.update({'ports': ports(iface),
  File "/usr/lib/python2.7/site-packages/vdsm/netinfo.py", line 187, in ports
    return os.listdir('/sys/class/net/' + bridge + '/brif')
OSError: [Errno 2] No such file or directory: '/sys/class/net/net-1/brif'


Traceback (most recent call last):
  File "/usr/share/vdsm/network/api.py", line 890, in setupNetworks
    configurator=configurator)
  File "/usr/share/vdsm/network/api.py", line 397, in _delBrokenNetwork
    implicitBonding=False, _netinfo=_netinfo)
  File "/usr/share/vdsm/network/api.py", line 219, in wrapped
    ret = func(**attrs)
  File "/usr/share/vdsm/network/api.py", line 486, in _delNetwork
    net_ent_to_remove.remove()
  File "/usr/share/vdsm/network/models.py", line 188, in remove
    self.configurator.removeBridge(self)
  File "/usr/share/vdsm/network/configurators/ifcfg.py", line 182, in removeBridge
    bridge.port.remove()
  File "/usr/share/vdsm/network/models.py", line 100, in remove
    self.configurator.removeNic(self)
  File "/usr/share/vdsm/network/configurators/ifcfg.py", line 231, in removeNic
    self.configApplier.removeNic(nic.name)
  File "/usr/share/vdsm/network/configurators/ifcfg.py", line 644, in removeNic
    with open(cf) as nicFile:
IOError: [Errno 2] No such file or directory: u'/etc/sysconfig/network-scripts/ifcfg-ens1f1'


- All networks without ifcfg-* are missing after reboot..host is down in rhev-M..
Isn't the fix for restore networks much earlier resolve such cases?
I'm a bit confused here..

(In reply to Michael Burman from comment #19)
> Dan, and what about reboots ? 

Reboots are unrealated to this bug. If 3.6's vdsm does not recover ifcfg, it's a material for another bug. In comment 18 I was only attempting to recreate a similar scenario to that of comment 0.

Verified on -  3.6.0-0.12.master.el6 with vdsm-4.17.3-1.el7ev.noarch

oVirt 3.6.0 has been released on November 4th, 2015 and should fix this issue.
If problems still persist, please open a new BZ and reference this one.