20:00:02 #startmeeting Octavia 20:00:03 Meeting started Wed Oct 21 20:00:02 2015 UTC and is due to finish in 60 minutes. The chair is xgerman. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:05 Greetings, y'all! 20:00:06 The meeting name has been set to 'octavia' 20:00:09 o/ 20:00:10 #chair blogan, johnsom 20:00:11 Current chairs: blogan johnsom xgerman 20:00:18 hi 20:00:22 #topic Announcements 20:00:37 o/ 20:00:46 Our design session: https://etherpad.openstack.org/p/mitaka-neutron-next-adv-services 20:01:07 hello! 20:01:13 hi 20:01:15 o/ 20:01:57 also next week is no meeting 20:02:19 #info skipping 10/28 meeting due to summit 20:02:23 well, next week is one big meeting. ;) 20:02:45 yeah, IBM throwing a party? 20:02:56 lol 20:03:00 Are we going to do a groupme again? 20:03:04 I assume so 20:03:06 I hope so 20:03:08 i usually just rename the old one 20:03:09 +1 20:03:13 +2 20:03:13 so everyone is already on it 20:03:21 i'll do that now 20:03:27 xgerman: I find out tonight. I've not been following IBM's stuff on the summit that closely 20:03:41 rm_work: Excellent! 20:04:14 done 20:04:17 #topic Design session planning/recap 20:04:34 remember to install / unmute groupme 20:04:35 so dougwig told us from an airport that we totally need to have this topic 20:04:49 so... a planning meeting for our planning meeting? 20:04:54 * rm_work is down 20:05:08 apparently 20:05:16 Er... 20:05:17 Ok. 20:05:31 Stealth illegal cellular 20:05:38 lol dougwig 20:05:38 Hahaa 20:05:42 lol 20:05:46 are you AIRBORN? 20:05:48 Haha 20:05:52 Heh, I just wanted a head up for folks to provide input. 20:06:00 No, waiting for runway 20:06:01 that actually gives me hope for my flight ;-) 20:06:01 because I thought cell signal just *wouldn't work* on airplanes.... due to like.... PHYSICS 20:06:03 lol k 20:06:20 It works fine, you just cycle towers too fast. 20:06:23 right 20:06:25 that's the issue 20:06:26 yep 20:06:31 you are moving too fast :P 20:06:36 handoffs can't happen quickly enough 20:06:43 This flight has international gogo anyway. 20:06:55 yep, how much is that for international? 20:07:00 I usually pay like $15 20:07:02 Yay for satellite internets. 20:07:03 im pretty sure my flight does not have wifi 20:07:03 So I'll disappear for 20 mins. 20:07:07 k 20:07:11 k 20:07:12 Ok. 20:07:16 I am assuming it'll be more for a 12 hour flight 20:07:20 $4 for a smartphone pass 20:07:31 dougwig: not bad 20:07:33 Score 20:07:35 can’t you expense it? 20:07:38 ... i wonder if i can share that via USB 20:07:46 o/ 20:08:15 anyway, are all the topic we want to talk on that pad? 20:08:16 So, back on topic, is there another etherpad of topics we need to update or something? 20:08:26 rm_work ssl passphrase? 20:08:38 Or, just the one LBaas/FWaas 20:08:39 The Lbaas one, linked above. 20:08:56 yeah we'll need to discuss that passphrase issue 20:09:01 https://review.openstack.org/#/c/237807/1/specs/mitaka/neutron-lbaas-store-tls-passphrase.rst,cm 20:09:12 yep, so it should be a topic? 20:09:21 Yes 20:10:09 done + done 20:10:24 though I could just -2 the proposal ;-) 20:10:36 and then i can -4 it 20:10:45 Haha! 20:10:58 I'll scrub that into the template format tonight, so please add ideas today if you can. 20:11:19 cool! Are you doing FWaaS, too? 20:12:05 I'll prep for both. I don't care who moderates. Can be anyone. 20:12:15 ok, cool 20:12:41 so any other topic? 20:12:48 Imo, if the moderator does all the talking, the session was a bust. :) 20:13:08 Unless that moderator is me. :) 20:13:12 Just sayin' 20:13:15 Lol 20:13:19 or the moderator got 100% consensus 20:13:21 and wsa amazing 20:13:37 well, as a co-chair I will not give you the talking token 20:13:44 Aaw! 20:13:49 sbalukoff will get the hot seat for active active. 20:14:01 yep, with a trap door 20:14:06 Which I'm sure we will all agree on immediately. 20:14:10 Haha! 20:14:12 I want the shocker 20:14:42 Heat = Buzz 20:14:45 Rough crowd. :) 20:14:56 Ok, on the runway. 20:14:59 yeah, I guess we need gavels to call the room to order 20:15:05 johnsom: You might discover I agree with you on some of this, eh. ;) 20:15:12 Seeya, dougwig! 20:15:16 I know, but it still sounds like fun 20:15:20 have a safe flight!! 20:15:36 johnsom: What's a little electrocution between friends, eh. ;) 20:16:02 And there go the engines spinning up... 20:16:10 See y'all in a few. 20:16:14 adios 20:16:21 bye 20:16:45 so any other topics we need to talk about in that session or topics we can remove ? 20:17:12 oh i have one, let me add it 20:17:18 multiple controller workers 20:17:26 that’s good one 20:17:34 Oh, right! 20:17:54 How long is the session??? 20:18:00 Let's call that: Fully redundant / horizontally scalable controllers. 20:18:04 40 minutes or so 20:18:07 Because it'll be more than just the worker. 20:18:12 Wow... 20:18:12 yeah 20:18:23 We need to scope out space we can crash 20:18:23 Ok, so... yeah. We're going to be lucky to get through all this. 20:18:29 johnsom: +1 20:18:43 Well, there are two topics there, which or both? 20:18:44 I propose we coordinate that on groupme. 20:18:44 doesn't hurt to get a list and then moderator prioritizes 20:19:01 Controller worker drivers and/or control plane HA 20:19:49 (Coordinate finding crash space on groupme, I mean-- since none of us know the layout well enough to say where to meet.) 20:20:08 +1 20:20:10 I am there Sunday night… so can scout it out until johnsom gets there 20:20:22 xgerman: Awesome! Let us know what you find. 20:20:24 Yep. I was thinking we just keep our eyes open for good spots 20:20:27 I'm there... I think on Tuesday. 20:20:33 Or something. 20:20:37 you get there on tuesday? 20:20:37 * xgerman wonders if I should crash WoOS 20:20:40 the day of teh talk? 20:20:57 blogan: Wait, no! I get there on Monday. 20:20:57 sounds like sbalukoff 20:20:58 Durr. 20:21:02 The talk moved to Thursday remember they asked us about it 20:21:03 sbalukoff: whew 20:21:06 xgerman: I know mestery will be at WoOS 20:21:08 Sorry, off by one error. 20:21:09 johnsom: ohhh yeah 20:21:12 and so will I 20:21:13 i forgot about that 20:21:15 I get there on Monday too 20:21:18 johnsom: thansk for reminder 20:21:42 Actually-- 20:21:42 first time this talk hasn't been on the first day 20:21:49 :-) 20:21:50 could we announce the times of the talk and the lab here? 20:21:53 Hands on is Wednesday, talk is Thursday 20:21:56 what about tradition... 20:21:56 yeah when is the lab 20:21:58 ah k 20:21:58 I've not yet heard final scheduling on that. 20:22:06 we are looking at imaging the USB drives 20:22:10 Sure, let me look it up again so I don't get it wrong 20:22:10 did you guys link an image for them? 20:22:13 I seem to recall 20:22:16 yeah, we did 20:22:18 Thanks, johnsom! 20:22:22 can you re-link that as a #link 20:22:41 #link http://tinyurl.com/octavia-workshop 20:23:15 #topic Brief progress reports 20:23:22 let’s move on 20:23:24 main conference lbaas talk is thursday at 240 20:23:38 Hands on lab - Wednesday @2:50pm 20:23:40 #link https://mitakadesignsummit.sched.org/mysessions?iframe=yes&w=&sidebar=yes&bg=no 20:23:45 Nice! 20:23:51 Oh, that didn't work 20:24:06 Well, we've got the dates and times now. 20:24:07 #link http://sched.co/49HH 20:24:13 That works too. 20:24:34 wow, slow DL :/ 20:24:35 Load Balancing as a Service - Thursday @ 2:40pm 20:24:37 is this HP's CDN? 20:24:37 #link http://sched.co/49HH 20:24:41 <_< 20:25:13 maybe local network is bad 20:25:33 so back on topic 20:25:35 lol 20:25:49 For the USB, I think we want the image, the trial VMware workstation, trial VMware fusion, and putty on the stick right? 20:26:00 yep, it’s all on that link 20:26:07 Oh cool 20:26:14 download, unpack or not and put on the drive 20:26:16 i assumed this would be a single iso to flash with dd 20:26:21 but that works too 20:26:22 i've just been workign as much as i can on the containers flows, also adding in bug fixes 20:26:28 just rsync to USB drive 20:26:45 yep, or drag & drop if you are on Mac OS X 20:27:02 still rsync <_< 20:27:12 checksum verification FTW 20:27:16 Haha! 20:27:22 I did some TLS verification and then ended up roping in johnsom 20:28:06 so we need some reviews for his patch 20:28:06 One of my other team at my company took over my work for heat support for LBaaS v2. So I am available for other tasks if we have any. 20:28:08 Yep. I have been working on VRRP. killing bugs. 20:28:29 ...are we onto the progress reports now? 20:28:33 yep 20:28:34 #link https://review.openstack.org/#/c/237707/ 20:28:36 Ok! 20:28:38 ^^ TLS bug 20:28:38 Yay! 20:28:43 So! There's this: 20:29:01 #link https://review.openstack.org/#/c/234639/ 20:29:04 o/ 20:29:22 #link https://review.openstack.org/215359 20:29:22 Active-active topology. Y'all were told about it last week. Please definitely review before the design session... 20:29:31 cert rotation still needs some reviews 20:29:56 But if you can also get some comments added prior to that that would be great: The IBM team is already putting together POC code, and it's annoying to have to back-track if we're not going the right direction with this, eh. 20:30:09 Also! 20:30:14 #link https://review.openstack.org/#/c/232173/ 20:30:15 #link https://review.openstack.org/#/c/215359/ ——cert rotation 20:30:16 Optimistic aren't they 20:30:22 That's documentation updates. 20:30:46 I have to add diagrams still (and hate all our options there)... but most importantly I've got the quick start guide first draft done. 20:30:56 also the doc about neutron-lbaas 20:30:58 PLEASE have a look at that and tell me where I'm lying and where I've forgotten stuff. 20:30:58 #link https://review.openstack.org/#/c/224929/ 20:31:26 johnsom: I'm willing to workflow 237707 right now 20:31:36 (I tried to balance between specificity when deploying Octavia in production, and having a "quick start" guide that was way too long.) 20:31:51 Also, I need the slides done before noon pacific on Saturday so I can fix formating, etc. on the plnae 20:32:03 johnsom: For the talk? the lab? both? 20:32:08 both 20:32:11 Ok. 20:32:22 I will fix the lab on the plane as well ;-) 20:32:36 augh i haven't even been able to spin octavia in devstack yet, much less look at troubleshooting it or updating the slides for that section 20:32:46 did someone say that section would be sort of a group effort to create anyway? 20:33:07 rm_work: Use your script! 20:33:14 right 20:33:23 i have only run it to debug that it works for other people :P 20:33:27 rm_work: That's what I did in the quickstart guide. ;) 20:33:32 haven't actually *used* the resulting devstack 20:33:32 lol 20:33:35 rm_work for the devstack portion of it anyway. ;) 20:33:40 yeah 20:34:07 ok, let’s make it a group effort then 20:34:24 I can at least show them how to change the debug level an d where the logs are 20:34:44 looking forward to the planning session tonight just so i can see how to set it all up :P 20:34:53 Maybe some pointers around that fire hose of a debug log 20:35:01 I can do presentation and actual hands-on helping no problem 20:35:12 if people are better at making slides than presenting them 20:35:15 rm_work: Ditto! 20:35:31 presenting is the prize for making slides ;-) 20:35:35 heh 20:35:40 or the curse (blogan) 20:35:53 lol 20:36:03 this is very true 20:36:16 ok, guess we are at 20:36:20 #topic Open Discussion 20:36:41 If you don't help with slides we will just put you on stage with a white title screen. Your name in big letters 20:37:07 lol 20:37:11 Maybe even a "email this person with questions" 20:37:11 yep, and some lorem ispum 20:37:20 Just to reiterate: I really do want y'all's input on that quick start guide. I think it's going to be important for getting Octavia adopted... and if it's wrong (which parts of it probably are right now), that doesn't help us at all! 20:37:40 yeah i'll review that too 20:37:45 +1 20:37:50 Thanks! I very, very much appreciate it! 20:38:23 I will try. I'm getting into crunch time with the demo. I have round_robin up and running, but I want to demo session persistence which has at least one bug. 20:38:29 also if you have trouble installing it there are vendors who provide commercial support ahem 20:38:44 or fanatical support ;-) 20:38:55 johnsom: I totally understand. We're down to the wire and getting the demo right is really important. 20:38:56 oh i wanted to mention the horizon dashboard 20:39:03 johnsom: yeah, demo useful for actually finding and squashing bugs, eh? :P 20:39:19 im thinkign that we shouldn't show what we plan to do over what has already been done 20:39:25 rm_work: Haha! Too true! 20:39:29 rm_work See above testing topic for the design session.... 20:39:39 blogan: Oh? 20:39:56 well, designs are also a work product... 20:40:15 sbalukoff: yeah people worked on the current dashboard, i dont think we shoudl mitigate what they've done by showing something else 20:40:39 ok, so does the current thing work? 20:40:54 blogan I am confused. So, no octavia Active/Standby demo? 20:41:06 Ok... but I suspect that astute observer will realize that LBaaS v1 dashboards are mostly going to be incompatible with LBaaS v2 dashboards. 20:41:11 yes, Active-Standby demo... 20:41:34 I think it is a good demo and will hopefully land in M1 20:41:42 im talking about the horizon dashboard images 20:41:43 yes, demo it 20:42:09 but we discussed demo'ing what has been done versus what we realy want in the end 20:42:14 which isn't done, just some mock up images 20:42:17 blogan: I'm still trying to understand: Is that so we don' t piss of people who worked on the old dashboard? 20:42:19 multiple listeners 20:42:31 Oh! 20:42:33 Right. 20:43:06 ok just to be clear, the neutron-lbaas-dashboard repo is for v2, and the dashboard that has in it does not support multiple listeners on a load balancer? 20:43:08 xgerman: ^^ 20:43:40 it supports them in a convoluted way + but leaves barbican out, etc. 20:43:53 I'm confused, but I don't think I'm helping to clarify anything, so I'll shut up for once. 20:43:58 xgerman: yeah so we discussed in our hangout not showing that but just shwoing what we wanted the end product to look like 20:44:05 yep 20:44:05 i am also confused T_T 20:44:12 sbalukoff: i'm trying to clear it up here too 20:44:14 but we'll hash it out in our pre-meeting meeting? 20:44:24 Yeah-- let's hash it out then. 20:44:35 well, we have mocks of what we think it should look like 20:44:51 and we have some reality which works well for eBay... 20:44:53 Well, the next meeting is for the lab. So, if it's going in the slides we need to decide now 20:44:56 well my point is, i think it woudl be a bit slap in the face to not show what the people who worked on the current v2 dashboard, and show mock ups of something we havent even worked on 20:45:25 we can talk about this afetr the meeting to move everyting else along 20:45:39 but would like a decision today 20:45:40 Actually, what else do people want to discuss now? 20:46:05 yeah, i think we had nothing else anyway 20:46:09 I've said my two bits... so if we can hash this out now, it sounds like it might be worthwhile in preparation for the meeting later today. 20:46:41 well, I am fine with showing the eBay stuff — wouldn’t be the first time we demo that ;-) 20:46:50 xgerman: +1 20:47:04 blogan I hear you on that. Is there a way we can show what's in the repo and then some WIP content? Do a dance? 20:47:20 +! 20:47:26 johnsom: +1 20:47:39 okay 20:47:43 If ultimately we plan on scrapping the UI work that ebay has done we should be honest about that. 20:48:00 well, it needs to be angularJS one day 20:48:11 that’s just the way Horizon rolls 20:48:30 But it also goes to show-- any operator / vendor can develop their own UI for this stuff if they want. In theory, the UI should only be manipulating APIs on the back-end anyway. 20:48:31 lol 20:48:33 well the ebay work is v1 20:48:35 so.... 20:48:41 I know that IBM built their own UI against lbaasV1. 20:48:42 right? 20:48:52 no its v2 20:48:54 nope, they gave us a v2 20:48:56 and I'm now putting the screws on them to update it to v2. 20:48:58 I mean, Rackspace uses our own UI entirely 20:48:59 which looks like their v1 20:49:07 it's pretty obvious it is possible to roll your own 20:49:12 https://github.com/openstack/neutron-lbaas-dashboard 20:49:29 So, framed that way: "Operators or Vendors may develop their own UI. Here are a couple examples." 20:49:36 It doesn't seem like a slap in the face to ebay. 20:49:36 yep, and does that actually work? 20:49:39 sbalukoff: that works 20:49:50 yeah 20:49:56 ^^ are there screenshots for that UI 20:49:59 their docs link is broken 20:50:05 and I don't see any in the README 20:50:16 do I have to install it to see examples? :/ 20:50:27 well, last time we check that UI creating a load balancer didn’t work but I have been told it has now been fixed 20:50:41 Ultimately, I'm guessing there will probably be a 'reference UI' of some kind. But I kind of doubt that anyone here ultimately is going to be using it primarily. 20:50:44 the mock-ups are just vaporware at that point 20:51:31 and the eBay, IBM people approached us for design guidance... 20:51:37 After all, when it comes to UI, our marketing people start to care. ;) 20:51:51 xgerman: Haha! Awesome! 20:52:21 well my concern is just ignoring any work people put into the current v2 dashboard 20:52:42 agreed, we should try to show both 20:52:52 blogan: I don't think we have to ignore it. We can showcase it and the one we also plan to build in a positive light. 20:52:56 WIP is to be expected in OpenStack, so I think we give credit to work done and talk about WIP ideas floating around. 20:53:00 i appreciate any work we get from anyone, especially when its done with horizon which we've all kind of stayed away from like the plague 20:53:11 Agreed. 20:53:15 agreed. 20:53:20 yeah, I think we agree 20:53:31 okay i like sbalukoff's ideas 20:53:40 Ok, everyone, you heard it here first! 20:53:43 I'm so quoting that. 20:53:52 * xgerman had to read that twice 20:53:56 blogan is ill 20:54:21 bad lunch truck day? 20:54:34 truck towed? 20:55:04 * blogan is tired 20:55:54 Ok, just a reminder, summit lab hangout in an hour and a half 3:30 pacific 20:56:10 johnsom: Thanks! 20:56:17 What's the hangout link? 20:56:34 #link https://plus.google.com/hangouts/_/hpe.com/summit 20:56:38 thx 20:56:39 Sweet! 20:56:46 +1 20:56:50 Also, what is the groupme info? 20:57:13 groupme is an app you need to install on your phone 20:57:18 rm_work: Can you add ajmiller? 20:57:18 will send you an invite 20:57:21 I've got the app 20:57:23 Awesome. 20:57:26 ok 20:57:28 yeah anyone can invite 20:57:33 Oh, handy! 20:57:39 The more you know. 20:57:54 (The less annoying you might be.) 20:57:57 ;) 20:58:03 https://media3.giphy.com/media/Y2nbrJyAR6RiM/200_s.gif 20:58:36 Anything else to discuss? 20:59:04 #endmeeting