05:49:18 <soichi> #startmeeting taas
05:49:19 <openstack> Meeting started Wed Aug 17 05:49:18 2016 UTC and is due to finish in 60 minutes.  The chair is soichi. Information about MeetBot at http://wiki.debian.org/MeetBot.
05:49:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
05:49:22 <openstack> The meeting name has been set to 'taas'
05:49:27 <soichi> hi
05:49:32 <anil_rao> Hi
05:49:36 <kaz> hi
05:49:51 <anil_rao> kaz: Welcome back
05:50:02 <kaz> hi anil_rao
05:50:15 <soichi> #topic Open Discussion
05:50:43 <soichi> do you have any update?
05:50:56 <anil_rao> Yes, I have an update on the bug
05:51:29 <anil_rao> I ran several tests and it appears that the packets are untagged for only one VLAN on a host.
05:51:46 <anil_rao> This is typically the VLAN with the lowest numerical ID.
05:52:45 <anil_rao> I am inclined to belive that this is an internal optimization in OVS but I will try to confirm this with the folks on the OVS list.
05:53:06 <anil_rao> I have a fix that I am testing out.
05:53:23 <soichi> it sounds great
05:54:31 <anil_rao> It comprises of using two flows to capture unicast packets ingressing a VM's vNIC
05:54:54 <anil_rao> and two flows to capture multicast/broadcast packets ingressing that VM's vNIC.
05:55:10 <anil_rao> The multicast/broadcast flows are of course assocaited with the vNIC's VLAN.
05:55:42 <anil_rao> One of the flows will have the associated VLAN id and the destination MAC.
05:56:02 <anil_rao> the other flow will check for untagged packets and the destination MAC.
05:56:30 <anil_rao> The flow with the VLAN id has a slightly higher priority, so if the packets have VLAN tags it will be captured by this flow.
05:56:42 <anil_rao> Otherwise, it will be captured by the flow for untagged packets.
05:57:20 <soichi> i see
05:57:20 <anil_rao> I will conduct more tests over the next couple of days and if all looks good I'll send out a review request.
05:57:41 <soichi> thank you
05:58:06 <anil_rao> I have noticed that the flow that looks for untagged frames has matches only for one VLAN per host.
05:59:07 <anil_rao> For capturing packets egressing from a VM's vNIC we will continue to use the existing flow that is based on the in_port check.
06:00:01 <anil_rao> A similar logic, i.e. using two separate flows, will be employed to capture broadcast/multicast packets ingressing a VM's vNIC.
06:00:45 <soichi> Is the uniqueness of MAC related to matching rule mentioned above?
06:01:18 <anil_rao> Yes, as confirmed by Armando, MAC addresses are considered unique only within a single virtual network.
06:01:34 <soichi> i see
06:01:56 <anil_rao> This is why we have been using dst mac address + VLAN id for our ingress flows.
06:02:11 <soichi> yes
06:03:50 <soichi> we will review your patch and test on our site after your submittion
06:04:08 <anil_rao> soichi: Thanks!
06:04:15 <soichi> submittion -> submission
06:05:41 <soichi> Kaz is working for performance mesurement
06:06:10 <anil_rao> That is great. I am waiting for this bug fix to be completed before I set up the hardware based environment in my lab.
06:06:37 <anil_rao> Then I can do some perf tests too
06:06:49 <kaz> I am trying to measure the performance when the source VMs of iperf are increased.
06:07:21 <soichi> i think we can share results on next week IRC meeting
06:07:39 <anil_rao> kaz,soichi: Yes, that will be awesome. Thanks.
06:10:07 <soichi> any other topics?
06:10:12 <soichi> reedip?
06:10:30 <reedip> Hi
06:10:37 <anil_rao> Hi
06:10:51 <reedip> no topics, wanted to know though when are we starting the work on the presentation ??
06:11:52 <soichi> i guess we need to start preparation in the begining of Sep.
06:12:27 <anil_rao> We need to get selected first. ;-)
06:12:39 <soichi> sure
06:13:08 <soichi> i think we will get notification in the end of Aug
06:13:15 <reedip> Oops , I forgot!
06:13:17 <reedip> sorry!
06:13:18 <reedip> :)
06:14:24 <anil_rao> reedip: What was the issue that you had brought up last week but we kind of ran out of time.
06:14:47 <reedip> anil_rao devstack issue, I have not yet tried it again
06:14:54 <reedip> devstack is not working with the latest tag
06:15:37 <anil_rao> reedip: Are you referring to the native iface issue?
06:15:53 <reedip> anil_rao, I will share the error on the email
06:16:04 <anil_rao> reedip: sure.
06:18:29 <reedip> anil_rao : I habve redeployed it, so it would take some time but will give a result soon
06:18:34 <reedip> u can check it tomorrow morning
06:18:45 <reedip> amotoki and I both faced that issue with taas
06:19:05 <anil_rao> reedip: Thanks. I'll watch out for your email tomorrow
06:20:39 <reedip> anil_rao : if it fails:)
06:20:51 <anil_rao> :)
06:21:36 <soichi> how quick!
06:22:41 <soichi> no more topics from me
06:22:58 <anil_rao> No more topics from me too.
06:23:24 <soichi> it is a little bit earlier, but can we close today's IRC?
06:23:24 <kaz> i also have no topic
06:23:36 <anil_rao> soichi: Yes.
06:23:58 <soichi> #endmeeting