Bug 1498432

This is queued for the next update in March/2018 for fast-datapath.
If you need earlier, please state when and the business justification.
Thanks,
fbl

Hi Flavio, OVN is in tech preview for OSP12 and this showed up during the OSP12 validation tests, not from a customer request.

Would it be easier to have this patch [0] applied on top of current downstream version instead? As we want customers to move to OVN from OSP13 on, I think that it'd be great that this kind of bugs would be fixed whenever possible (especially as the fix is simple like this).


Thanks a lot!
Daniel

[0] https://github.com/openvswitch/ovs/commit/afaf4de07083863bccb1feaf01f37a59ca06a909

Please note that OVS 2.7.3 has an important fix [1] which significantly reduces the ovn-controller's CPU usage. This patch doesn't install the OF flows which are not required. More details in the commit message of the patch [1]

[1] - https://github.com/openvswitch/ovs/commit/6b5df63d9699efbee4701570f699ef043a3cff81

Having OVS 2.7.3 in OSP12 for OVN deployments would be really useful.

I ran neutron_tempest_plugin.scenario.test_trunk.TrunkTest.test_subport_connectivity
and it failed(attached log). During debugging I found few things:
1. VM doesnt get IP address to sub-port interface 
 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1442 qdisc pfifo_fast state UP 
 qlen 1000
    link/ether fa:16:3e:99:e7:4b brd ff:ff:ff:ff:ff:ff
    inet 10.100.0.6/28 brd 10.100.0.15 scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::f816:3eff:fe99:e74b/64 scope link 
       valid_lft forever preferred_lft forever
 3: eth0.10@eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1442 qdisc noqueue state 
 UP qlen 1000
    link/ether fa:16:3e:99:e7:4b brd ff:ff:ff:ff:ff:ff
    inet6 fe80::f816:3eff:fe99:e74b/64 scope link 
       valid_lft forever preferred_lft forever

2. there is no "br-trunk" which suppose to be connected to br-int so data with tag can be passed to the VM subinterface(this is how it work in ML2/ovs).


Adding Mojo for reference.
 https://mojo.redhat.com/docs/DOC-1088196-vlan-aware-vms

Created attachment 1387222 [details]
test results

(In reply to Eran Kuris from comment #10)
> I ran
> neutron_tempest_plugin.scenario.test_trunk.TrunkTest.
> test_subport_connectivity
> and it failed(attached log). During debugging I found few things:
> 1. VM doesnt get IP address to sub-port interface 
>  1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1
>     link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
>     inet 127.0.0.1/8 scope host lo
>        valid_lft forever preferred_lft forever
>     inet6 ::1/128 scope host 
>        valid_lft forever preferred_lft forever
>  2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1442 qdisc pfifo_fast state
> UP 
>  qlen 1000
>     link/ether fa:16:3e:99:e7:4b brd ff:ff:ff:ff:ff:ff
>     inet 10.100.0.6/28 brd 10.100.0.15 scope global eth0
>        valid_lft forever preferred_lft forever
>     inet6 fe80::f816:3eff:fe99:e74b/64 scope link 
>        valid_lft forever preferred_lft forever
>  3: eth0.10@eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1442 qdisc noqueue
> state 
>  UP qlen 1000
>     link/ether fa:16:3e:99:e7:4b brd ff:ff:ff:ff:ff:ff
>     inet6 fe80::f816:3eff:fe99:e74b/64 scope link 
>        valid_lft forever preferred_lft forever
> 
> 2. there is no "br-trunk" which suppose to be connected to br-int so data
> with tag can be passed to the VM subinterface(this is how it work in
> ML2/ovs).
> 
> 
> Adding Mojo for reference.
>  https://mojo.redhat.com/docs/DOC-1088196-vlan-aware-vms

Let's look into this.
BTW, in case of OVN there's no br-trunk and the tag/untag happens directly in br-int with OpenFlow rules.

When I originally opened this bug, it was because neutron.tests.tempest.scenario.test_trunk.TrunkTest.test_trunk_subport_lifecycle wasn't passing. This was due to the fact that there was a bug in core OVN where parent port was not properly set. This bug was fixed in openvswitch-2.7.3-2.git20171010.el7fdp and I could verify that the test passes with it. From our IRC conversation it looks like it also passed for you.

I'd agree that the title of this BZ is a bit too generic but the test you're seeing failing now it's a different one and it has a different root cause. Shall we rename this BZ and open a different one for this other issue?

according to Ofer .B. & Daniel I am setting this BZ as Verified.
Test "test_trunk_subport_lifecycle" passed on: 
cat /etc/yum.repos.d/latest-installed 
12   -p 2018-01-26.2
pm -qa |grep openvswitch-2.7.
openvswitch-2.7.3-3.git20180112.el7fdp.x86_64
python-openvswitch-2.7.3-3.git20180112.el7fdp.noarch


New BZ will be open about test_trunk.TrunkTest.test_subport_connectivity failure.

^
Adding the new BZ ID: https://bugzilla.redhat.com/show_bug.cgi?id=1539652

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2018:0248