20:00:07 #startmeeting octavia 20:00:08 Meeting started Wed May 13 20:00:07 2015 UTC and is due to finish in 60 minutes. The chair is xgerman. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:11 The meeting name has been set to 'octavia' 20:00:16 #chair blogan 20:00:16 Current chairs: blogan xgerman 20:00:18 o/ 20:00:19 o/ 20:00:43 It's possible blogan is running a bit late 20:00:50 getting towed? 20:00:58 :) 20:01:29 #topic Announcements 20:01:54 #link https://etherpad.openstack.org/p/LBaaS-FWaaS-VPNaaS_Summer_Midcycle_meetup 20:02:04 Mid cycle has a date and a location 20:02:27 i'm back 20:02:30 just got otu of meeting 20:02:31 o/ 20:02:41 o/ 20:02:42 blogan hi 20:02:47 o/ 20:02:53 o/ 20:03:13 please add yourself to the attendee list if you can come otherwise we will try to broadcast 20:03:26 * TrevorV|Home is late 20:03:30 o/ 20:03:45 #link https://etherpad.openstack.org/p/lbaas-vancouver-planning 20:03:59 xgerman and sballe, can you remove your parens at some point? 20:04:10 next week is OpenStack but we still need to collect more topics for the Octavia session 20:04:15 dougwig: i just add my own set of parens 20:04:28 the title of the session is "octavia as ref driver", fiy 20:04:50 dougwig: is octavia the working group or the fishbowl? 20:04:58 wg 20:04:58 and honestly what is the difference 20:05:15 working group gets more done i assume? 20:05:25 o/ 20:05:40 we rock either way 20:06:23 the other two links I copied from last time so will just skip 20:06:44 any other announcements? 20:07:06 #topic Joined neutron as an openstack project 20:07:19 so welcome our new overlord mestery 20:07:35 all hail overlord mestery 20:07:48 * dougwig genuflects 20:08:13 I shall hail NOTHING 20:08:30 I feel like a minion all of a sudden 20:08:33 banished him 20:08:37 Poof 20:09:26 we've all been pawns in his game of world domination 20:09:30 anyhow, not sure if we are now in Advanced Services under Lt. dougwig or not 20:09:44 lt danwig 20:10:27 some details, btw: 20:10:39 https://review.openstack.org/#/c/182748/ 20:10:49 our rename will likely happen after the summit. 20:10:59 ok 20:11:03 which means if you reference octavia by link, be sure to mention both, and the rename. 20:11:07 happens durign the talk adn everying breaks 20:11:10 i mean during the demos 20:11:24 haha, i already spoke to infra, it'll be after the summit. 20:11:26 During both. Happens multiple times, including roll-backs 20:11:41 yep, can’t we just have a redirect 20:11:57 L7 rukes 20:12:17 What happens to unmerged patches after the rename? Do they automatically get fixed? 20:12:31 is anyone concerned about governance? the spec that russel merged said that new repos have oversight from the neutron ptl (and lt's, if passed), but day-to-day is completely unchanged. 20:12:45 ajmiller: good question, i'll ask. 20:13:02 not at all I just like clarity 20:13:10 yeah i have no problem with the governance 20:13:11 in governance 20:13:20 should be transparent, except now your commits get atc creds. :) 20:13:34 right, i think in practice it'll be no different, personally 20:13:42 awesome! no more import reordering for me... 20:14:23 any questions here, or should we move on? 20:14:28 dougwig: i was wondering how to propose the rename ... thanks for that link 20:14:30 move on 20:14:38 #topic Brief progress reports 20:14:43 dougwig thanks for your help on making this happen 20:14:48 +100 20:14:50 russellb: also add yourself to the infra meeting agenda. there's a rename section. 20:16:05 dougwig: ack 20:16:19 I have been working on slides in prep for the presentation, should have those done today. At the same time I have been stacking and building up a script for the demo. Thanks for all the help on testing/bug fixes 20:16:33 octavia neutron driver is code complete, and has loaded successfully in neutron-server once. i'm expecting to test basic operations this evening, definitely before the weekend. 20:16:45 awesome 20:17:10 It would be super cool to use the nuetron client in the demo 20:17:16 REST driver/server get’s deployed; uses the SSL cert scheme we devised, and can boot a VM with it 20:17:21 dougwig: does it act asynchronously or synchronously? 20:17:28 blogan: synchronously 20:17:51 dougwig: okay, so create load balancer call will be slow, but the rest will be relatively fast 20:17:59 dougwig is the current patchset up on gerrit the code complete? 20:18:31 yes, but i don't recommend trying it yet. gimme today. 20:18:37 Ok 20:18:40 blogan: i can live with that for the demo. 20:18:40 Ok 20:19:51 ptoohill? 20:20:06 any progress? 20:20:07 I've been working on fixes, tweaks and tests in prep for summit/demo. 20:20:18 awesome!! 20:20:37 I have patches up that are working i believe, and im hoping to have this scenario test working soon to verify things throughly 20:20:41 I think blogan was going to add a section to ptoohill patchset. Any idea when we could consider merging that stuff? 20:21:09 Yea, think he has update for the postPlug method 20:21:16 johnsom: we can merge ptoohill's stuff now, i ahvet added my stuff to it, i can add it in a nother reveiw as its not modifying what he's done 20:21:27 i just thought i coudl get it out quickly enough, but ran into some issues 20:21:39 ok, sounds good 20:21:53 Ok, ptoohill good with merging after the meeting? 20:21:56 after that there's one bug i'd like to get fixed for the demo 20:22:03 Works for me 20:22:05 demo 1 20:22:14 https://bugs.launchpad.net/octavia/+bug/1453951 20:22:14 Launchpad bug 1453951 in octavia "Cannot send traffic to VIP on global namespace, but can on the VIP network's dhcp namespace" [Undecided,New] 20:22:36 aliens 20:23:37 oh and another thing, the defautl config value for base_path under the amphora driver section requires sudo to write the files to, so i've been just changing it to the ubuntu user's home, TrevorV has a patch otu to fix this, its just prepending the commands with sudo 20:24:03 ok 20:24:06 https://review.openstack.org/#/c/180336/ 20:24:11 I thought that merged already 20:24:17 nope 20:24:31 some comments left on it I believe 20:24:37 and a -1 isn’t addressed 20:24:47 yeah address it TrevorV 20:24:48 Oh, yeah, there is an ownership issue too 20:25:10 TrevorV|Home: 20:25:17 oh 20:25:51 Sorry, I can update those. I was certain I had tested it, but may have had faulty logic when I tested :( 20:25:52 My bad 20:26:05 I make a new patch set after meeting 20:26:15 Excellent, thanks! 20:26:49 well, I am still not thrilled we prepend things with sudo and not just have a method which does it programmatically (aka root wrap or so) but... 20:27:07 I already commented addressing that concern, xgerman. 20:27:17 yep, saw it... 20:27:31 well the method would still prepend it with sudo anyway because you can't just switch to root in one command and execute antoher command under root 20:27:46 but yeah wrapping it and cleaning it would be great 20:27:55 though i think most of these commands will run as root 20:27:56 brogan that;s true but it might look cleaner then prepend manually 20:28:08 yeah, what you said 20:28:23 brehgan 20:28:30 lol 20:28:30 :) 20:28:40 bruh bru 20:29:00 next topic? 20:29:25 #topic Release roadmap 20:29:39 johnsom wants to make a side with the release roadmap 20:29:45 side=slide 20:29:53 I would like to have an Octavia slide that talks about our plans for releases. 20:30:24 so basically 0.5 would be Liberty? or are we aiming for 1.0? 20:30:32 xgerman and I are thinking the old plan with 0.5 match reference, 1.0 active/standby, 2.0 horiz scale might not match where we are 20:31:14 We could call our liberty release 0.5, to give it a "beta" feel, but include the active/standby you guys have started. 20:31:17 Thoughts? 20:32:00 0.5 is still planned to have one controller, but it manages the containers by heartbeats and will act upon a stale heartbeat right? 20:32:12 if it's stable enough to be the ref in liberty, i'd be tempted to call it the 1.0. 20:32:20 Yeah, current plan, 0.5 has spares failover 20:32:49 so we skip 0.5 and go straight to 1.0? 20:32:54 dougwig Good point 20:34:56 I'm liking the sound of 1.0 in Liberty 20:35:05 ok, then let’s scarp 0.5 20:35:08 and 1.0 to you is? 20:35:13 and make Liberty 1.0 20:35:29 well whatever version we say it is, the roadmap should include what features are in that version 20:35:46 yeah, after we decided naming we can say features 20:35:53 so Liberty = 1.0 20:36:00 sounds like we need an octavia session to decide this 20:36:02 oh wait! 20:36:07 we have one! 20:36:11 :) 20:36:20 Hahaha, the problem is the session is after the presentation.... 20:36:26 lol good point 20:36:26 T_T 20:36:42 it's not an octavia meeting without a #startvote, that takes three tries to get the syntax right. 20:36:52 lol 20:37:05 #startvote should we all punch dougwig? 20:37:06 Begin voting on: should we all punch dougwig? Valid vote options are Yes, No. 20:37:07 Vote using '#vote OPTION'. Only your last vote counts. 20:37:14 #vote yes 20:37:16 #voyte Yes 20:37:16 #endvote 20:37:17 Voted on "should we all punch dougwig?" Results are 20:37:30 I currently have: 0.5 = Feature parity with existing ref, service vms, spares failover; 1.0 active/standby amphora, HA control plane; 2.0 Act/Act and amp horizontal scale 20:37:32 He stacked the votes in favor of his opinion. 20:37:32 * dougwig weeps 20:37:57 existing ref will be more scalable than what we would have 20:38:01 * TrevorV|Home noticed blogan didn't actually cast a valid vote... RIP 20:38:09 let’s move HA control plane into 2.0 20:38:28 scrpa 0.5 and we are good 20:38:36 ? 20:38:46 well it'd just be renaming 0.5 to 1.0 no? 20:38:57 and 1.0 to 2.0, etc 20:39:09 that’s fine with me, too 20:39:18 <_< 20:39:22 I just had the feel active/standby amphora was pretty far head already 20:39:32 rm_work, doesn't like renaming things, even if they make sense to rename 20:39:50 no, i don't like moving all of the features we had planned back a version T_T 20:39:58 its not far enough, just need to try to get it in genericly if that makes sense 20:40:11 i mean its not far off 20:40:13 k 20:40:25 but we definitely should discuss it and what values we need to store 20:40:33 multicast vs unicaste 20:40:33 we need codenames, clearly. and all roman things. vomitorium, crucifiction, ... 20:40:48 +1 20:40:49 dougwig +1 20:40:58 +1 20:41:06 i hate you dougwig 20:41:19 * dougwig basks in the love. 20:41:26 Can we agree to greco roman at a minimum? 20:41:44 Ok, so if we shift the numbers that means the specs will need updating too 20:41:51 http://en.wikipedia.org/wiki/List_of_Roman_emperors#Emperors_after_395 20:41:58 sorry for being late ;-( 20:41:59 johnsom just convinced me not to rename 20:42:22 Maybe renaming things is more difficult then it would be beneficial? 20:42:29 maybe 20:42:30 ah, yea that :P 20:42:45 if someone wants to tackle that and get it working, i'm all for it 20:43:33 i'd rather get the code working, and this isn't an external package, so we don't have any strict release requirements yet. 20:43:41 dougwig: What parens? 20:44:04 in the mid cycle meetup 20:44:17 oh yeah. I am coming ;-) 20:44:18 in the attendance section 20:44:35 so let’s sum up 20:44:58 the feature list stays but we might rename to some Roman codename 20:45:07 So, stick with what we have now for the presentation with caveats that this might change with the design session 20:45:22 +1 20:45:44 +11 20:45:59 Done. Thank you 20:46:30 #topic Open Discussion 20:47:01 so the stickers have been printed and will be delivered to barclay today 20:47:08 +1000 20:47:08 barclac 20:47:09 any issues anyone is having running the creates end to end? 20:47:16 tons 20:47:22 >< 20:48:03 I have tried to capture everything I have seen. I'm going to take another stack pass today. 20:48:48 https://s3.amazonaws.com/uploads.hipchat.com/8522%2F409386%2FLDXFN1XtBMau5R7%2FOctavia_sticker_die_cut_OS_09.png 20:49:23 johnsom: have you been able to make requests to the vip? 20:49:28 and have it load balance? 20:49:40 johnsom got it to loadbalance 20:49:47 woot 20:50:05 I am working the REST driver angle 20:50:20 If I put it all on one network. I think I was doing it from the amphora node last time, but got a balance 20:50:33 There is some wonky routing going on from devstack host 20:51:05 At least that was my guess 20:51:18 well, it would be great if we could give people a link after the demo they can download, stack.sh and are in business 20:51:58 well i'd like to be able to just do a curl from the single node devstack machine 20:52:10 makes the demo easier to follow 20:52:14 blogan Agreed 20:52:18 +1 20:52:24 Do you have thoughts on what might be wrong? 20:52:25 also we have 2 demos 20:52:32 right now you have to do sudo ip netns exec qdhcp- curl 20:53:09 johnsom: i think it might have to do with security groups, but not sure, it seems to happen after an interface is hot plugged 20:54:02 Yeah, that is a thought, those new secgroups that come in with the port 20:54:39 yeah but those sec groups would also block the traffic from the netns 20:54:42 but it doesn't 20:55:48 Like I said, I plan to restack with the lateset merged code (including the two we are about to do) and keep looking into it. 20:56:01 k - sounds good 20:56:22 any advice what we should cover in the vBrownbag demo — or I should just redo it with the other driver? 20:57:27 yeah this is soemthign we'llprobably have to consult some neutron experts with 20:57:55 +1 20:58:08 we are going g to place with a lot of them next week :-) 20:58:34 2 minutes left... 20:58:35 Demo is monday though 20:58:46 :-( 20:59:06 ok 20:59:13 guess we are done 20:59:16 yeah 20:59:17 #endmeeting