16:02:09 <frankbrockners> #startmeeting FDS synch
16:02:09 <collabot`> Meeting started Thu Sep  8 16:02:09 2016 UTC.  The chair is frankbrockners. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:09 <collabot`> Useful Commands: #action #agreed #help #info #idea #link #topic.
16:02:09 <collabot`> The meeting name has been set to 'fds_synch'
16:02:33 <frankbrockners> Hey folks - quick status check in on FDS...
16:02:50 <jlinkes_> #info Juraj Linkes
16:03:04 <raymaika> #info Raymond Maika
16:03:22 <frankbrockners> michal-cmarada jlinkes_  and I talked earlier today
16:03:51 <michal-cmarada> #info Michal Cmarada
16:04:01 <frankbrockners> #info key open issues are per a discussion earlier today: https://wiki.opnfv.org/display/fds/FDS+Open+Issues#FDSOpenIssues-OpenIssues
16:05:06 <frankbrockners> #info also trozet mentioned to me that he'll work on resolving https://jira.opnfv.org/browse/APEX-258 today (detect if the dataplane is fdio or ovs dpdk and configure images appropriately post install)
16:05:47 <frankbrockners> michal-cmarada - any updates from your end on testing "deleting ports"
16:05:50 <jlinkes_> flavors, not images :-)
16:06:06 <frankbrockners> #undo
16:06:06 <collabot`> Removing item from minutes: <MeetBot.ircmeeting.items.Info object at 0x2c88410>
16:06:25 <frankbrockners> #info also trozet mentioned to me that he'll work on resolving https://jira.opnfv.org/browse/APEX-258 today (detect if the dataplane is fdio or ovs dpdk and configure flavors appropriately post install)
16:06:34 <frankbrockners> thanks jlinkes_
16:07:26 <frankbrockners> michal-cmarada?
16:07:43 <trozet> frankbrockners: we also need the other fix in puppet-fdio to not strip the vlan tag right
16:07:46 <michal-cmarada> we have found out that when we are deleting BD from HC, sometimes ODL doesn`t get a response in 10seconds end this will end up in unknown state for the deletion of BD
16:08:11 <frankbrockners> trozet - yes - do we need to create a Jira ticket for that?
16:08:29 <trozet> frankbrockners: no we can just do that, but have a question
16:08:49 <trozet> frankbrockners: cant we disable the hw vlan strip always, i think even e1000 driver supports it
16:09:03 <trozet> frankbrockners: like if we just enable it always, will there be a problem with non-ucs servers?
16:09:09 <frankbrockners> michal-cmarada - can you drop maros a note on that issue (and also cc edwarnicke)
16:10:16 <michal-cmarada> yes, i think we already had these issues and thats why we introduced the 10 sec timeout. I will check with them and see what we can do about that
16:11:14 <frankbrockners> trozet - that is a good question that I don't have an immediate answer for - obviously this would only apply to non-trunk ports
16:11:35 <trozet> frankbrockners: right
16:11:41 <frankbrockners> but for scenarios where we don't expect a VLAN tag, that might be fine
16:11:53 <frankbrockners> let me check with folks on potential side effects
16:12:00 <frankbrockners> will come back via email
16:12:45 <frankbrockners> trozet - can we make things configurable for now? then we can still make this "always on" if we have to
16:13:03 <trozet> frankbrockners: how do we decide to configure it or not?
16:13:23 <frankbrockners> ideally part of the scenario config
16:13:42 <frankbrockners> you run on a UCS-B - switch it on - otherwise - don't worry
16:13:53 <trozet> frankbrockners: hmm ok I need to think about this one
16:14:54 <frankbrockners> jlinkes_ - any updates / news from your end?
16:15:55 <jlinkes_> I let functest run on a setup without any stale configuration
16:16:18 <jlinkes_> I introduced a 15 sleep after creating a subnet and after attaching network to router
16:17:01 <jlinkes_> That didn't help, the vms still didn't receive a dhcp address
16:17:39 <michal-cmarada> frankbrockners: hw vlan strip always, is probably needed only for VLAN case for VXLAN case it should have the default value, so I think it would be best to have it configurable
16:18:13 <jlinkes_> I observed that there were no tx packets on the tunnel between computes, that might be what caused the issues
16:18:15 <frankbrockners> michal-cmarada - that's a good point
16:18:33 <frankbrockners> trozet - so making VLAN strip configurable makes sense
16:19:08 <jlinkes_> One additional thing that functest vping does is create a security group for ssh/icmp traffic
16:19:30 <frankbrockners> jlinkes_ are the vxlan tunnels operational?
16:19:41 <frankbrockners> where does it create a sec group?
16:19:57 <frankbrockners> we don't do sec groups on vpp yet
16:20:42 <jlinkes_> Yeah, I don't see how that would affect anything, but it is a difference, so I wanted to mention it
16:21:44 <frankbrockners> do we know whether the vxlan tunnels are operational - i.e. can i pass probe traffic?
16:22:23 <jlinkes_> You mean were they operational during the vping test?
16:22:40 <michal-cmarada> jlinkes_: frankbrockners: I found old bug reported by me few months ago, this is probably the same issue as we are seeing with BD not deleted. https://bugs.opendaylight.org/show_bug.cgi?id=6105
16:22:56 <frankbrockners> no - prior - you said that it looks as if you don't get a dhcp response across the vxlan tunnels - right?
16:23:23 <jlinkes_> that was during the vping test
16:23:44 <jlinkes_> When I tried it on fresh setup manually it worked
16:25:12 <frankbrockners> ok
16:25:46 <frankbrockners> michal-cmarada - https://bugs.opendaylight.org/show_bug.cgi?id=6105 indeed sounds like a bug
16:25:51 <frankbrockners> that we have to get fixed
16:26:02 <frankbrockners> will email tyler and jan
16:26:46 <frankbrockners> ok - anything else?
16:29:05 <frankbrockners> ok... looks like we're done for today. I've also updated: https://wiki.opnfv.org/display/fds/FDS+Open+Issues#FDSOpenIssues-OpenIssues
16:29:08 <frankbrockners> #endmeeting