| sdake | morphed into corosync | 00:00 |
|---|---|---|
| Mech422 | ahh yeah | 00:00 |
| *** ravig has joined #openstack-kolla | 00:00 | |
| Mech422 | I never got into the 'old' linux clustering stuff... | 00:00 |
| *** ravig has quit IRC | 00:04 | |
| dcwangmit01 | Hi All. I'm thinking about filing a blueprint. There may be an network configuration issue when running kolla containers on kubernetes. https://gist.github.com/dcwangmit01/f27eae3e7634861aedcb0a2315f56d1a I'll be back reading IRC tomorrow 10AM PST. If it's valid, I'll file the blueprint | 00:08 |
| openstackgerrit | Joshua Harlow proposed openstack/kolla: Add non-intrusive ctrl-c handling https://review.openstack.org/321913 | 00:12 |
| wirehead_ | Rackspace tried to get me to move to Texas multiple times. | 00:16 |
| *** sacharya has joined #openstack-kolla | 00:17 | |
| wirehead_ | There are advantages to living in the bay area. Mostly having to do with physically being able to meet with people and chat. | 00:18 |
| *** dwalsh has quit IRC | 00:18 | |
| Mech422 | wirehead_: yeah - there is no tech community around like the bay area | 00:19 |
| wirehead_ | Seems pretty darn silly to require people in teech to move to a particular not-bay-area location (or maybe not-Austin, not-Portland, not-Seattle) sort of location. | 00:19 |
| Mech422 | wirehead_: but as I get older, I like computers more and people less....soo..... | 00:19 |
| openstackgerrit | Vikas Jain proposed openstack/kolla: Changed namespace from kolla to kollaglue in oslo https://review.openstack.org/327433 | 00:19 |
| wirehead_ | Heh. I actually like people more and computers less as I’ve aged. | 00:20 |
| Mech422 | wirehead_: really? wow...what kind of computers do you hang around with :-P | 00:20 |
| wirehead_ | Don’t think of it as a problem with the computers. Think about it as meta-programming. Or, when dealing with lisp programmers, meta-meta-programming. | 00:21 |
| Mech422 | wirehead_: when I was younger I used to pugs/lugs/meetups - google used to host some cool stuff at the HQ... | 00:21 |
| Mech422 | wirehead_: now, I resent anything that makes me put on pants | 00:21 |
| Mech422 | :-P | 00:21 |
| Mech422 | wirehead_: LOL | 00:21 |
| Mech422 | meta^2 ? :-) | 00:21 |
| wirehead_ | Well, it also kinda goes back to the loss of the vibrant artist community that was in the bay area, fed by all sorts of broke frustrated ex-techies, in 2002-2005. | 00:22 |
| wirehead_ | I mourn the crazies who were priced out of the bay. | 00:22 |
| *** sacharya has quit IRC | 00:23 | |
| Mech422 | wirehead_: I don't get that whole bay area 'housing outrage' - cities like New York have had gentrification issues for decades and don't make class warfare out of it ? | 00:23 |
| sdake | wirehead_ i dont know how places like mcdoandls stay in business | 00:23 |
| *** ssurana1 has quit IRC | 00:23 | |
| wirehead_ | Yeah, so I’m on my neighborhood mailing list. And some of the people clutching at their pearls about various subjects is quite…. special. | 00:24 |
| Mech422 | wirehead_: hehe | 00:24 |
| wirehead_ | Mostly it’s that Manhattan Manhattanified. Whereas here, they see that as a dirty word. | 00:24 |
| Mech422 | wirehead_: also, if SFO would build some dam apartments, it would ease the housing crunch - but everything is a 'historic preserve' | 00:24 |
| Mech422 | wirehead_: eh - I get wanting to preserve the 'flavor' of the city...but by definition, thats going to drive costs up... | 00:25 |
| wirehead_ | Well, they preserve the ‘flavor’ of the city by ruining the flavor of the city. | 00:26 |
| Mech422 | wirehead_: oh? I haven't been there in like 6 years - has it changed much ? | 00:26 |
| Mech422 | wirehead_: ( better rapid transit would help too - barts nice, but too limited) | 00:27 |
| wirehead_ | So, I had a convo last year with a professional ballerina in one of the bay area’s second-tier dance companies. | 00:28 |
| wirehead_ | And, note that first-tier is one of the top-5 ballet companies, so second tier is pretty damn good. | 00:29 |
| wirehead_ | She has no car and the best she could do is a shared room in an apartment that’s not too hard to get to the studio via transit. | 00:29 |
| Mech422 | wirehead_: yeah - thats what I mean about transit - New York, no one drove - you could subway/bus/taxi anywhere... | 00:31 |
| wirehead_ | Well, I’m WFH today, which means I’m carless. | 00:31 |
| Mech422 | wirehead_: of course - PHX is 10x worse for transit - no car, no life :-P | 00:31 |
| wirehead_ | The car is really my wife’s car. It’s actually annoying to adjust it for my settings. And I can get away without the car for most of daily life. | 00:32 |
| wirehead_ | But I have to bike, because otherwise I’d spend all my time on the transit system that, frankly, is slower than biking. | 00:32 |
| Mech422 | wirehead_: I just don't get the arguement "we have this really cool place where everyone wants to live, and now its too expensive" ? | 00:32 |
| Mech422 | wirehead_: yeah - buses are the worst for travel time .. stopping every block ugh | 00:33 |
| wirehead_ | For me, I’m just concerned that it’s going to get really lame and borring because I rather like ballet and art and stuff. | 00:33 |
| Mech422 | wirehead_: I hear ballet, opera, syphmonies, etc are having a hard time all over the country though... | 00:34 |
| Mech422 | wirehead_: seems like the 'old culture' isn't as appreciated by the younger generations ? | 00:35 |
| wirehead_ | Well, the band scene is kinda lame too | 00:35 |
| Mech422 | wirehead_: really? that was one of the big draws to sfo | 00:36 |
| wirehead_ | Very very very few live music venues compared to other cities | 00:37 |
| Mech422 | ahh | 00:37 |
| wirehead_ | There’s still some large concert halls for touring acts. But cost per square foot is too much for there to be space wasted for a band. | 00:37 |
| Mech422 | yeah - I remember the 'cow palace' or something ? | 00:38 |
| wirehead_ | Yeah, so I bemoan the change that is the bay area, but my values and concerns seem to be totally different than 90% of the locals. | 00:39 |
| Mech422 | wirehead_: heh - being a geek, you should be used to that :-) | 00:39 |
| wirehead_ | True, but that 90% includes a lot of the tech community. :D | 00:40 |
| Mech422 | wirehead_: I was amazed when I moved to cupertino how BAD the schools were | 00:40 |
| Mech422 | wirehead_: mesa az had a median income of $35K/year at the time, and spent $300/yr PER STUDENT more then cupertino did | 00:41 |
| Mech422 | wirehead_: I thought engineers would value good schools, but it appears they value lower taxes more | 00:41 |
| wirehead_ | Well, there are a lot of engineers who didn’t reproduce. | 00:41 |
| wirehead_ | Or they send their children to a private school. | 00:41 |
| Mech422 | wirehead_: yeah - and a lot tha ...exactly | 00:41 |
| wirehead_ | Or they can afford to live in Palo Alto and send their kids to Gunn. | 00:42 |
| Mech422 | wirehead_: the 2 big things I miss from silly valley, are Fry's electronics and LOTS of takeout options | 00:42 |
| Mech422 | I can literally only order delivery food here from like 6 places | 00:42 |
| *** sdake has quit IRC | 00:42 | |
| wirehead_ | Of course, Gunn has this problem where the children of overachieving bay area wealthy people tend to snap and commit suicide on the CalTrain line I take to and from work. | 00:43 |
| wirehead_ | Oh, Fry’s is downhill. | 00:43 |
| wirehead_ | Food is still pretty good. Delivery is great, built on the backs of a lot of the ‘normal’ people. | 00:43 |
| Mech422 | wirehead_: No...Not frys! that was my mecca! | 00:44 |
| Mech422 | the sunnyvale HQ was AWESOME! | 00:44 |
| wirehead_ | Well, you can still get a soldering iron and some resistors on Sunday while screwing with stuff. | 00:44 |
| Mech422 | wirehead_: yeah - and you could get and RC toy, new cpu and a cup of coffee to go with it.... | 00:45 |
| wirehead_ | Same. | 00:45 |
| wirehead_ | But Powell Books’s in Portland has done better surviving Amazon’s onslaught than Fry’s has about Amazon’s weird-shit-and-electronics. | 00:45 |
| Mech422 | wirehead_: yeah - I get my fix from NewEgg now... | 00:46 |
| Mech422 | wirehead_: for some reason, about the only thing I get from Amazon is books...dunno why | 00:46 |
| Mech422 | I go on amazon ebook binges every so often - though, why their ebooks are only $10 less then phsyical I don't know | 00:47 |
| openstackgerrit | Joshua Harlow proposed openstack/kolla: Stop using a global logger for all the things https://review.openstack.org/321884 | 00:47 |
| wirehead_ | I’m not sure if I’d miss Fry’s. I miss the Indian businesses in Sunnyvale if they went away. | 00:48 |
| wirehead_ | Well, there’s Overdrive for ebooks and Amazon Unlimited. | 00:49 |
| wirehead_ | My wife is 10 x the bibliophile I am, and I’m pretty bad. Last year when we were in LA, we had to stop at a LA City Public Library to get a card, because they’ve got a great collection on Overdrive. | 00:50 |
| openstackgerrit | Joshua Harlow proposed openstack/kolla: Use 'image' not 'item' https://review.openstack.org/324756 | 00:51 |
| Mech422 | wirehead_: I haven't used overdrive in like 2 years... Mesa library had it, but I never bothered to renew my card when it expired | 00:51 |
| wirehead_ | Yeah, all of my local library cards have expired, probably with late fees on each one. But the LA collection is nice. :D | 00:52 |
| *** salv-orlando has quit IRC | 00:53 | |
| Mech422 | wirehead_: I should prolly look at Oreilly Safari again... | 00:53 |
| *** salv-orlando has joined #openstack-kolla | 00:53 | |
| *** daneyon_ has joined #openstack-kolla | 00:54 | |
| wirehead_ | Yah. | 00:54 |
| *** ravig has joined #openstack-kolla | 00:55 | |
| *** phuongnh has joined #openstack-kolla | 00:55 | |
| *** vhosakot has quit IRC | 00:56 | |
| *** vhosakot has joined #openstack-kolla | 00:57 | |
| *** daneyon_ has quit IRC | 00:58 | |
| *** vhosakot has quit IRC | 01:02 | |
| *** ravig has quit IRC | 01:02 | |
| *** ravig has joined #openstack-kolla | 01:04 | |
| *** murphyslawbbs has quit IRC | 01:04 | |
| *** vhosakot has joined #openstack-kolla | 01:10 | |
| *** rhallisey has quit IRC | 01:18 | |
| *** ravig has quit IRC | 01:32 | |
| *** tfukushima has joined #openstack-kolla | 01:48 | |
| *** vhosakot has quit IRC | 01:53 | |
| *** vhosakot has joined #openstack-kolla | 01:54 | |
| *** sacharya has joined #openstack-kolla | 01:59 | |
| *** sacharya has quit IRC | 02:07 | |
| *** vhosakot has quit IRC | 02:07 | |
| *** vhosakot has joined #openstack-kolla | 02:07 | |
| *** sacharya has joined #openstack-kolla | 02:08 | |
| kklimonda | hmmm, is using ssd for ceph osd journal documented somewhere? | 02:08 |
| *** ayoung has quit IRC | 02:17 | |
| *** ayoung has joined #openstack-kolla | 02:25 | |
| *** ayoung has quit IRC | 02:29 | |
| *** mdorman has joined #openstack-kolla | 02:31 | |
| *** ravig has joined #openstack-kolla | 02:32 | |
| *** jgriffith_away is now known as jgriffith | 02:42 | |
| *** yuanying has quit IRC | 02:50 | |
| *** mdorman_ has joined #openstack-kolla | 02:53 | |
| *** mdorman has quit IRC | 02:53 | |
| *** mdorman_ is now known as mdorman | 02:53 | |
| *** mdorman has quit IRC | 02:58 | |
| *** mdorman has joined #openstack-kolla | 03:02 | |
| *** mdorman has quit IRC | 03:03 | |
| *** coolsvap has joined #openstack-kolla | 03:04 | |
| *** mdorman has joined #openstack-kolla | 03:07 | |
| *** sdake has joined #openstack-kolla | 03:10 | |
| sdake | we have fry's electronics in arizona Mech422 | 03:10 |
| openstackgerrit | Merged openstack/kolla: Changed namespace from kolla to kollaglue in oslo https://review.openstack.org/327433 | 03:15 |
| *** sdake has quit IRC | 03:26 | |
| mdorman | i added some notes to our etherpad at https://etherpad.openstack.org/p/kolla-gd-thoughts-concerns-remarks , hopefully that’s helpful in understanding what we’re trying to do (or at least my version of what i think we’re trying to do, hah) | 03:29 |
| *** iceyao has joined #openstack-kolla | 03:38 | |
| *** d_code has quit IRC | 03:40 | |
| *** iceyao has quit IRC | 03:43 | |
| *** yuanying has joined #openstack-kolla | 03:48 | |
| *** d_code has joined #openstack-kolla | 03:50 | |
| kklimonda | ok, found ceph external journal documentation: https://review.openstack.org/#/c/324539/ | 03:59 |
| patchbot | kklimonda: patch 324539 - kolla - Document the use of external Ceph journal drives | 03:59 |
| *** tfukushima has quit IRC | 04:00 | |
| *** vhosakot has quit IRC | 04:00 | |
| Mech422 | sdake: the fry's electronics here are pretty bad compared to the one in sunnyvale... | 04:04 |
| *** d_code has quit IRC | 04:05 | |
| wirehead_ | It's just not the same. | 04:05 |
| Mech422 | kklimonda: I have some patches in review to make that work with partitions as well as full devices | 04:05 |
| wirehead_ | Of course, you might come back to SF for some reason, Mech422, and realize that ours have gone downhill too | 04:05 |
| *** d_code has joined #openstack-kolla | 04:06 | |
| Mech422 | wirehead_: true - nothing ever lives up to the memory... | 04:06 |
| wirehead_ | Perk of living in Sunnyvale: After finishing up today's hacking, went on a lovely bike ride in the mountains. | 04:06 |
| wirehead_ | Mech422, do you remember Weird Stuff Warehouse in Sunnyvale? | 04:09 |
| Mech422 | Oh god yes! | 04:09 |
| Mech422 | weird stuff was awesome!! | 04:09 |
| wirehead_ | It doesn't have as many discarded prototypes as it did in the 90s, but it's still pretty damn weird. | 04:10 |
| Mech422 | just trying to figure out what have the stuff was, was a trip :-) | 04:10 |
| wirehead_ | When I worked at Y! I would blow off steam by browsing the aisles of crap. | 04:10 |
| Mech422 | what was the other one - over by fry's - it was more of a ham shack... | 04:10 |
| Mech422 | wirehead_: oh yeah - walking distance! I'd be broke all the time! | 04:10 |
| wirehead_ | HSC | 04:10 |
| *** tfukushima has joined #openstack-kolla | 04:11 | |
| wirehead_ | Mech422: In 2008, I managed to use my willpower to avoid coming home with this: https://www.flickr.com/photos/cyberspace/2423245919/ | 04:12 |
| Mech422 | Indies! | 04:13 |
| Mech422 | oh yeah - Halted! | 04:13 |
| Mech422 | they had an electron microscope on sale there for like $300 the first time I walked in.... | 04:14 |
| Mech422 | wirehead_: my neighbor ran a test lab for Hitachi Data Systems - he gave me like 6 SGI O2s then they updated the lab | 04:14 |
| wirehead_ | o2's were the beginning of the end, but those were the ones I learned Maya and Renderman on before I learned that the video game and computer graphics industry is a horrible industry to work in. | 04:15 |
| wirehead_ | I've managed to not acquire too large of a stack of useless vintage computer gear, actually. | 04:16 |
| Mech422 | it was funny to watch O2's run win3.1 programs in emulation faster then PC's did native... | 04:16 |
| Mech422 | wirehead_: I'm a huge MMORPG geek.... it sad how bad the game industry has gotten | 04:17 |
| Mech422 | wirehead_: they really milk the 'I wanna make video games' dreams of the RCGs | 04:18 |
| wirehead_ | Yeah, basically I left college a bit unhappy because I was doing finance instead of games or graphics. Then I learned how to make art neon, and then I met people who were chewed up and spit out by the games industry and realized I'd dodged a bullet. | 04:20 |
| Mech422 | wirehead_: yeah - they really seem to treat people like crap | 04:21 |
| Mech422 | ok - gonna watch Zootopia with the girlfriend...back in an hour or two | 04:22 |
| *** salv-orl_ has joined #openstack-kolla | 04:25 | |
| *** salv-orlando has quit IRC | 04:28 | |
| *** d_code has quit IRC | 04:28 | |
| *** salv-orl_ has quit IRC | 04:51 | |
| *** salv-orlando has joined #openstack-kolla | 04:51 | |
| *** harlowja_at_home has joined #openstack-kolla | 04:56 | |
| *** coolsvap has quit IRC | 04:56 | |
| *** salv-orlando has quit IRC | 05:00 | |
| openstackgerrit | Joshua Harlow proposed openstack/kolla: Add non-intrusive ctrl-c handling https://review.openstack.org/321913 | 05:00 |
| openstackgerrit | Joshua Harlow proposed openstack/kolla: Stop using a global logger for all the things https://review.openstack.org/321884 | 05:06 |
| *** harlowja_at_home has quit IRC | 05:08 | |
| *** coolsvap has joined #openstack-kolla | 05:22 | |
| *** mgkwill has quit IRC | 05:23 | |
| *** godleon has quit IRC | 05:23 | |
| *** macsz has quit IRC | 05:23 | |
| *** tpeoples has quit IRC | 05:24 | |
| *** DuncanT has quit IRC | 05:24 | |
| *** godleon has joined #openstack-kolla | 05:24 | |
| *** elemoine_ has quit IRC | 05:24 | |
| *** serverascode has quit IRC | 05:25 | |
| *** NachoDuck has quit IRC | 05:25 | |
| *** ediardo has quit IRC | 05:25 | |
| *** NobodyCam has quit IRC | 05:25 | |
| *** smekel_ has quit IRC | 05:25 | |
| *** akwasnie has quit IRC | 05:25 | |
| *** akwasnie_ has quit IRC | 05:25 | |
| *** DuncanT has joined #openstack-kolla | 05:27 | |
| *** NobodyCam has joined #openstack-kolla | 05:28 | |
| *** ediardo has joined #openstack-kolla | 05:28 | |
| *** elemoine_ has joined #openstack-kolla | 05:28 | |
| *** akwasnie_ has joined #openstack-kolla | 05:30 | |
| *** ravig has quit IRC | 05:30 | |
| *** tpeoples has joined #openstack-kolla | 05:32 | |
| *** akwasnie has joined #openstack-kolla | 05:33 | |
| *** elemoine_ has quit IRC | 05:33 | |
| *** NachoDuck has joined #openstack-kolla | 05:33 | |
| *** smekel_ has joined #openstack-kolla | 05:33 | |
| *** serverascode has joined #openstack-kolla | 05:34 | |
| *** mgkwill has joined #openstack-kolla | 05:34 | |
| *** elemoine_ has joined #openstack-kolla | 05:35 | |
| *** macsz has joined #openstack-kolla | 05:37 | |
| Mech422 | that was pretty good :-) | 06:12 |
| *** daneyon_ has joined #openstack-kolla | 06:19 | |
| *** daneyon_ has quit IRC | 06:23 | |
| *** Mr_Broken has joined #openstack-kolla | 06:27 | |
| *** sacharya has quit IRC | 06:31 | |
| *** zhiwei has joined #openstack-kolla | 06:37 | |
| *** belmoreira has joined #openstack-kolla | 06:38 | |
| *** zhiwei has quit IRC | 06:44 | |
| *** zhiwei has joined #openstack-kolla | 06:45 | |
| *** muhammadshahzeb has joined #openstack-kolla | 06:46 | |
| *** muhammadshahzeb has left #openstack-kolla | 06:48 | |
| *** belmoreira has quit IRC | 06:48 | |
| *** muhammadshahzeb has joined #openstack-kolla | 06:51 | |
| *** muhammadshahzeb has quit IRC | 06:52 | |
| *** muhammadshahzeb has joined #openstack-kolla | 07:01 | |
| *** Serlex has joined #openstack-kolla | 07:02 | |
| muhammadshahzeb | Hello guys I am from PLUMgrid, I have recently been looking to integrate PLUMgrid ONS http://www.plumgrid.com/product/overview/ with kolla. | 07:03 |
| muhammadshahzeb | I heard about the OSIC cluster testing project and wanted to volunteer and help, talked to Michał Jastrzębski, he told me to join the IRC channel for more discussion. | 07:04 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/kolla: Revert "Changed namespace from kolla to kollaglue in oslo" https://review.openstack.org/327490 | 07:06 |
| coolsvap | Jeffrey4l, ^^ | 07:07 |
| coolsvap | please approve the revert | 07:07 |
| coolsvap | mandre, if you are available ^^ we need to get the revert in due to this we are getting empty kolla-build.conf | 07:07 |
| *** belmoreira has joined #openstack-kolla | 07:07 | |
| *** matrohon has joined #openstack-kolla | 07:08 | |
| *** mikelk has joined #openstack-kolla | 07:25 | |
| *** sacharya has joined #openstack-kolla | 07:31 | |
| *** jmccarthy has left #openstack-kolla | 07:33 | |
| *** jmccarthy has quit IRC | 07:33 | |
| *** zhiwei has quit IRC | 07:34 | |
| *** sacharya has quit IRC | 07:37 | |
| *** salv-orlando has joined #openstack-kolla | 07:41 | |
| *** salv-orl_ has joined #openstack-kolla | 07:42 | |
| *** tfukushima has quit IRC | 07:44 | |
| *** tfukushima has joined #openstack-kolla | 07:45 | |
| *** salv-orlando has quit IRC | 07:46 | |
| *** zhiwei has joined #openstack-kolla | 07:46 | |
| *** muhammadshahzeb has quit IRC | 07:48 | |
| *** tzn has joined #openstack-kolla | 07:48 | |
| *** jmccarthy has joined #openstack-kolla | 07:51 | |
| *** openstackgerrit has quit IRC | 08:03 | |
| *** openstackgerrit has joined #openstack-kolla | 08:03 | |
| *** daneyon_ has joined #openstack-kolla | 08:07 | |
| *** daneyon_ has quit IRC | 08:11 | |
| *** muhammadshahzeb has joined #openstack-kolla | 08:19 | |
| *** muhammadshahzeb has quit IRC | 08:23 | |
| *** salv-orl_ has quit IRC | 08:25 | |
| *** salv-orlando has joined #openstack-kolla | 08:39 | |
| *** phuongnh has quit IRC | 08:56 | |
| *** salv-orlando has quit IRC | 08:56 | |
| *** pbourke_ has joined #openstack-kolla | 09:08 | |
| *** coolsvap has quit IRC | 09:09 | |
| *** athomas has joined #openstack-kolla | 09:10 | |
| *** SiRiuS__ has joined #openstack-kolla | 09:10 | |
| *** tfukushima has quit IRC | 09:13 | |
| *** jmccarthy has left #openstack-kolla | 09:14 | |
| *** tfukushi_ has joined #openstack-kolla | 09:16 | |
| *** SiRiuS__ has quit IRC | 09:21 | |
| *** SiRiuS__ has joined #openstack-kolla | 09:22 | |
| *** jmccarthy has joined #openstack-kolla | 09:22 | |
| *** mbound has joined #openstack-kolla | 09:27 | |
| *** zhiwei has quit IRC | 09:28 | |
| *** dmk0202 has joined #openstack-kolla | 09:28 | |
| *** pbourke_ has quit IRC | 09:30 | |
| *** sacharya has joined #openstack-kolla | 09:33 | |
| *** sacharya has quit IRC | 09:37 | |
| *** pbourke_ has joined #openstack-kolla | 09:43 | |
| *** pbourke_ has left #openstack-kolla | 09:44 | |
| *** pbourke_ has joined #openstack-kolla | 09:44 | |
| *** jmccarthy has left #openstack-kolla | 09:47 | |
| *** salv-orlando has joined #openstack-kolla | 09:48 | |
| *** tzn has quit IRC | 09:50 | |
| *** daneyon_ has joined #openstack-kolla | 09:55 | |
| openstackgerrit | Steve Hindle proposed openstack/kolla: Read partition tables directly, bypassing stale kernel data for ceph. https://review.openstack.org/326609 | 09:59 |
| *** daneyon_ has quit IRC | 09:59 | |
| *** tzn has joined #openstack-kolla | 10:10 | |
| *** pbourke has quit IRC | 10:13 | |
| *** pbourke has joined #openstack-kolla | 10:13 | |
| *** tzn has quit IRC | 10:14 | |
| *** tfukushi_ has quit IRC | 10:19 | |
| *** jmccarthy has joined #openstack-kolla | 10:19 | |
| *** tfukushima has joined #openstack-kolla | 10:22 | |
| *** prithiv has joined #openstack-kolla | 10:28 | |
| *** belmoreira has quit IRC | 10:31 | |
| *** muhammadshahzeb has joined #openstack-kolla | 10:35 | |
| *** salv-orlando has quit IRC | 10:41 | |
| *** salv-orlando has joined #openstack-kolla | 10:49 | |
| *** prithiv has quit IRC | 10:50 | |
| *** iceyao has joined #openstack-kolla | 10:51 | |
| *** muhammadshahzeb has quit IRC | 10:53 | |
| *** iceyao has quit IRC | 10:55 | |
| *** ppowell has joined #openstack-kolla | 10:59 | |
| *** salv-orlando has quit IRC | 11:03 | |
| openstackgerrit | Merged openstack/kolla: Use 'image' not 'item' https://review.openstack.org/324756 | 11:18 |
| ccesario | morning :) | 11:18 |
| openstackgerrit | Merged openstack/kolla: Revert "Changed namespace from kolla to kollaglue in oslo" https://review.openstack.org/327490 | 11:18 |
| openstackgerrit | Merged openstack/kolla: Fix the prechecks for the ansible version https://review.openstack.org/327410 | 11:18 |
| cu5 | Hi. I'm trying to build images based on Centos but my Keystone and Horizon images fail due to an error unpacking httpd. I'm using Kolla's stable/mitaka branch. Any suggestions? I know it's not really an issue with Kolla but I'm hoping someone here tackled this issue and may be able to save me some time by sharing the solution ;) | 11:19 |
| openstackgerrit | Merged openstack/kolla: Bump the ansible version to >2 in the kolla-toolbox https://review.openstack.org/311625 | 11:19 |
| *** alyson_ has joined #openstack-kolla | 11:19 | |
| pbourke_ | cu5: can you pastebin the error? | 11:24 |
| *** tfukushima has quit IRC | 11:25 | |
| *** mliima has joined #openstack-kolla | 11:31 | |
| *** dwalsh has joined #openstack-kolla | 11:32 | |
| *** salv-orlando has joined #openstack-kolla | 11:35 | |
| *** coolsvap has joined #openstack-kolla | 11:35 | |
| mandre | cu5: try changing the docker graph driver, http://docs.openstack.org/developer/kolla/image-building.html#known-issues | 11:38 |
| openstackgerrit | Md Nadeem proposed openstack/kolla: Added unit test for start container https://review.openstack.org/327247 | 11:39 |
| *** rhallisey has joined #openstack-kolla | 11:43 | |
| *** daneyon_ has joined #openstack-kolla | 11:43 | |
| *** daneyon_ has quit IRC | 11:48 | |
| *** ppowell has quit IRC | 11:48 | |
| *** dwalsh has quit IRC | 11:51 | |
| cu5 | pbourke_: here's the pastebin link: http://pastebin.com/48vg8NGa | 11:54 |
| *** sdake has joined #openstack-kolla | 11:56 | |
| cu5 | A test image from centos:latest with just a yum update && yum install httpd builds without error. | 11:56 |
| pbourke_ | i think the issue is with docker build though | 11:57 |
| pbourke_ | see mandre's link | 11:57 |
| cu5 | Yeah I just clicked it. Going to try that now. | 11:58 |
| *** salv-orlando has quit IRC | 12:06 | |
| *** ppowell has joined #openstack-kolla | 12:07 | |
| *** prithiv has joined #openstack-kolla | 12:07 | |
| *** d_code has joined #openstack-kolla | 12:14 | |
| *** coolsvap has quit IRC | 12:16 | |
| *** belmoreira has joined #openstack-kolla | 12:20 | |
| sdake | morning | 12:28 |
| *** Mr_Broken has quit IRC | 12:33 | |
| *** Mr_Broken has joined #openstack-kolla | 12:33 | |
| *** Mr_Broken has quit IRC | 12:38 | |
| *** prithiv has quit IRC | 12:38 | |
| *** zhiwei has joined #openstack-kolla | 12:39 | |
| *** Mr_Broken has joined #openstack-kolla | 12:44 | |
| *** zhiwei has quit IRC | 12:46 | |
| mliima | morning all | 12:47 |
| *** tzn has joined #openstack-kolla | 12:47 | |
| *** sdake has quit IRC | 12:48 | |
| *** JoseMello has joined #openstack-kolla | 12:48 | |
| *** salv-orlando has joined #openstack-kolla | 12:49 | |
| openstackgerrit | Merged openstack/kolla: Fix the neutron reconfigure issue https://review.openstack.org/326280 | 12:54 |
| *** stvnoyes has joined #openstack-kolla | 12:57 | |
| *** salv-orlando has quit IRC | 13:03 | |
| *** salv-orlando has joined #openstack-kolla | 13:04 | |
| *** absubram has quit IRC | 13:05 | |
| *** jtriley has joined #openstack-kolla | 13:08 | |
| *** ayoung has joined #openstack-kolla | 13:11 | |
| apuimedo | mandre: ping | 13:14 |
| mandre | apuimedo: hello | 13:15 |
| apuimedo | I saw https://review.openstack.org/#/c/315027/4/docker/plugins/neutron-server/networking-midonet.j2 | 13:15 |
| patchbot | apuimedo: patch 315027 - kolla - [WIP] Add support for third party plugins | 13:15 |
| apuimedo | in the binary part | 13:15 |
| apuimedo | shouldn't it also use network-midonet from packages? | 13:15 |
| *** salv-orlando has quit IRC | 13:15 | |
| mandre | apuimedo: yeah, it should | 13:17 |
| mandre | but as far as I know, midonet only provides packages for debian based systems | 13:17 |
| mandre | maybe even just for ubuntu | 13:17 |
| apuimedo | nope | 13:18 |
| apuimedo | wrong ;-) | 13:18 |
| apuimedo | mandre: https://builds.midonet.org/midonet-5/stable/el7/noarch/ | 13:18 |
| apuimedo | ubuntu and el7 | 13:19 |
| mandre | apuimedo: looks like you're right, https://builds.midonet.org/openstack-mitaka/unstable/el7/noarch/ | 13:20 |
| *** dwalsh has joined #openstack-kolla | 13:21 | |
| mandre | apuimedo: I should tell you that I didn't test if the patch you referenced produces a functional image | 13:21 |
| apuimedo | mandre: I work at midokura ;-) | 13:22 |
| mandre | it was to demonstrate a new plugin mechanism | 13:22 |
| apuimedo | I know | 13:22 |
| apuimedo | thanks for trying it with MidoNet! | 13:22 |
| apuimedo | It is much appreciated | 13:22 |
| apuimedo | I'm putting comments to help with it | 13:22 |
| mandre | apuimedo: :) | 13:22 |
| mandre | apuimedo: so we can count on you to contribute the midonet-api image? ;-) | 13:23 |
| apuimedo | mandre: heh | 13:23 |
| apuimedo | I already have it done | 13:23 |
| apuimedo | mandre: https://hub.docker.com/r/midonet/cluster/ | 13:23 |
| apuimedo | and also | 13:23 |
| apuimedo | https://hub.docker.com/r/midonet/agent/ | 13:24 |
| apuimedo | but I guess I should send them to Kolla | 13:24 |
| apuimedo | or is it okay to spawn third party containers? | 13:24 |
| mandre | apuimedo: excellent! would be great to have it integrated with kolla | 13:24 |
| apuimedo | indeed | 13:24 |
| mandre | apuimedo: yeah, it would make more sense to have it in kolla tree | 13:25 |
| apuimedo | mandre: so you think they should be submitted as kolla images in docker/midonet/{agent,cluster} | 13:25 |
| apuimedo | ? | 13:25 |
| mandre | apuimedo: exactly | 13:25 |
| apuimedo | ok | 13:26 |
| apuimedo | I'll try to send it next week then | 13:26 |
| apuimedo | in the new plugin approach for neutron | 13:26 |
| apuimedo | where do the neutron.conf and plugin.conf modifications go? | 13:26 |
| *** Mr_Broken has quit IRC | 13:29 | |
| *** inc0 has joined #openstack-kolla | 13:29 | |
| *** Mr_Broken has joined #openstack-kolla | 13:30 | |
| inc0 | good morning | 13:30 |
| apuimedo | good morning | 13:32 |
| *** Mr_Broken has quit IRC | 13:34 | |
| apuimedo | inc0: I took a look at the bp/third-party-plugin-support | 13:34 |
| apuimedo | good job ;-) | 13:35 |
| inc0 | thanks | 13:35 |
| inc0 | feel free to put a review on it;) | 13:36 |
| inc0 | or test it out | 13:36 |
| apuimedo | thanks, I'll try | 13:38 |
| apuimedo | inc0: do I get it right that after this https://review.openstack.org/#/c/326688/5/docker/base/Dockerfile.j2 | 13:38 |
| patchbot | apuimedo: patch 326688 - kolla - Add customizations for apt keys and packages | 13:38 |
| apuimedo | in the override file for a plugin | 13:38 |
| inc0 | so when you pull next one it will pull all 4 | 13:39 |
| inc0 | make sure to use checkout | 13:39 |
| inc0 | and you need all 4 as they are dependant on each other | 13:39 |
| apuimedo | you will just be able to, for example, declare base_apt_keys | 13:39 |
| *** sdake has joined #openstack-kolla | 13:39 | |
| apuimedo | blcok | 13:39 |
| apuimedo | *block | 13:39 |
| apuimedo | and it will replace completely the content in the base? | 13:39 |
| sdake | huh | 13:40 |
| sdake | morning mliima | 13:40 |
| inc0 | not yet, to re-declare packages we need something like this https://review.openstack.org/#/c/327271/2 | 13:40 |
| patchbot | inc0: patch 327271 - kolla - Customizations continued | 13:40 |
| apuimedo | if that is the case, can it still use what's defined in base_apt_keys? | 13:40 |
| sdake | morning folks | 13:40 |
| inc0 | re-declare keys* | 13:40 |
| mliima | morning sdake | 13:40 |
| sdake | apuimedo we are working on complete cusotmization of images | 13:40 |
| sdake | apuimedo it will alnd in newton definately | 13:40 |
| sdake | apuimedo if yu want something for an earlier rev of kolla you will either have to backport the changes or roll your own | 13:41 |
| inc0 | apuimedo, so issue is, this is declared and used in same dockerfile | 13:41 |
| mliima | https://bugs.launchpad.net/kolla/+bug/1590603 | 13:41 |
| openstack | Launchpad bug 1590603 in kolla "Cinder + LVM/iSCSI does not work on Ubuntu 16.04" [Undecided,Confirmed] | 13:41 |
| inc0 | so you can't inject modification between declaration and usage | 13:41 |
| apuimedo | sdake: I'm fine with master | 13:41 |
| inc0 | even block unfortunately doesn't work | 13:41 |
| apuimedo | I just want to understand the progress | 13:41 |
| sdake | apuimedo cool | 13:41 |
| inc0 | however what I did | 13:41 |
| apuimedo | and put some comments like I did with mandre's patch :P | 13:42 |
| sdake | progress is we have a rough prototype of the implementation | 13:42 |
| sdake | apuimedo community seems to agree on approach | 13:42 |
| sdake | apuimedo we need to finish the job on the prototype | 13:42 |
| inc0 | is that if you declare base_apt_keys_append=[append_this_key] | 13:42 |
| sdake | and by we, I mean i do personally | 13:42 |
| inc0 | it will well..append | 13:42 |
| sdake | apuimedo ocne the prototype is done, I plan to distribute the work to our team | 13:42 |
| sdake | we have 40 container sets | 13:42 |
| apuimedo | yes | 13:42 |
| sdake | which means 40 patches | 13:42 |
| sdake | i think the work takes about 20-30 minutes per container set | 13:42 |
| apuimedo | and mandre just asked me to add midonet/cluster and midonet/agent :P | 13:43 |
| sdake | once the base macros are implemented | 13:43 |
| sdake | apuimedo I think realistically we are 3-4 weeks out from customization in master | 13:43 |
| inc0 | apuimedo, yeah, Mech422 is also playing with Mido | 13:43 |
| apuimedo | inc0: as I put above | 13:43 |
| inc0 | you might want to join forces;) would be super cool to have midonet, calico and such | 13:43 |
| apuimedo | I have containers already | 13:43 |
| sdake | apuimedo if you do deploy amster, I'd highly recommend deplooying mitaka version of openstack | 13:43 |
| apuimedo | I just need to kollaize them | 13:44 |
| sdake | apuimedo because mater of openstack is in various states of brokenness at various times | 13:44 |
| inc0 | cool | 13:44 |
| apuimedo | https://hub.docker.com/r/midonet/agent/ | 13:44 |
| apuimedo | and https://hub.docker.com/r/midonet/cluster | 13:44 |
| apuimedo | they are very simple | 13:44 |
| apuimedo | sdake: master kolla pointing to mitaka for OSt services? | 13:45 |
| *** vhosakot has joined #openstack-kolla | 13:45 | |
| sdake | apuimedo right | 13:45 |
| apuimedo | I was planning on being a bit more conservative and point to liberty :P | 13:45 |
| sdake | apuimedo that wont work | 13:45 |
| apuimedo | good to know then :P | 13:45 |
| sdake | apuimedo you will need to cherrypick a patch in liberty to get that to work | 13:45 |
| *** ravig has joined #openstack-kolla | 13:45 | |
| sdake | apuimedo its likely as simple as a cherrypick of one patch | 13:45 |
| sdake | with possibly some merge conflicts | 13:46 |
| apuimedo | sdake: I should be able to use mitaka | 13:46 |
| apuimedo | with MidoNet's master plugin | 13:46 |
| sdake | apuimedo if you are planning to implement midonet for kolla, that should go in master | 13:46 |
| sdake | and it would be greatly appreciated ;) | 13:46 |
| sdake | why was Revert "Changed namespace from kolla to kollaglue in oslo" | 13:48 |
| sdake | merged? | 13:48 |
| inc0 | sdake, wrong | 13:48 |
| inc0 | master kolla == master openstack | 13:48 |
| sdake | sbezverk_ found that if kollaglue was used, the images don't build correctly | 13:48 |
| sdake | rather if kolla is used | 13:48 |
| inc0 | mitaka kolla == mitaka openstack | 13:48 |
| sdake | inc0 yes I know | 13:48 |
| sdake | pbourke mandre ping | 13:50 |
| sdake | re https://review.openstack.org/#/c/327490/ | 13:50 |
| patchbot | sdake: patch 327490 - kolla - Revert "Changed namespace from kolla to kollaglue ... (MERGED) | 13:50 |
| mandre | sdake: coolsvap said this change was causing kolla-build.conf to be empty | 13:51 |
| sdake | mandre sbezverk_ determined this chane was needed during lbaas dev | 13:51 |
| vhosakot | sdake: yes, it is needed for magnum PS as well | 13:52 |
| vhosakot | I used kollaglue namespace for magnum | 13:52 |
| sdake | vhosakot are ou confirming this revert is wrong then? | 13:52 |
| sdake | https://review.openstack.org/#/c/327490/ | 13:53 |
| patchbot | sdake: patch 327490 - kolla - Revert "Changed namespace from kolla to kollaglue ... (MERGED) | 13:53 |
| vhosakot | yes, I saw issues when I changed namespace kollaglue to kolla when testing magnum | 13:53 |
| sdake | originally when lbaas work started, the namespace was set to kolla | 13:53 |
| mandre | sdake: then I guess it needs a bit more work if it generates emopty config files | 13:53 |
| sdake | sbezverk_ determined it needs to be set to kollaglue | 13:53 |
| vhosakot | yep, lbaas needed kollaglue | 13:53 |
| sdake | mandre it didn't when I watched sbezverk_ use it over webex :) | 13:53 |
| inc0 | sooo...how the hell oslo namespace affects docker namespace? | 13:53 |
| sdake | inc0 our dockerhub registry is kollaglue | 13:54 |
| sdake | inc0 the playbooks defult namespace is kollaglue | 13:54 |
| inc0 | ok, but this is completely irrelevant to what is in oslo | 13:54 |
| sdake | as a result o th docker hub registry namespae | 13:54 |
| apuimedo | sdake: wasn't kollaglue abandoned? I didn't see any new image | 13:54 |
| sdake | apuimedo no I push new imagess | 13:54 |
| apuimedo | link? | 13:54 |
| inc0 | apuimedo, we do this on per-release manner | 13:54 |
| apuimedo | I didn't see mitaka stuff | 13:55 |
| sdake | apuimedo I pushed source and build for centos and ubuntu of liberty and mitaka | 13:55 |
| vhosakot | Has anyone see this bug in master ? | 13:55 |
| vhosakot | https://bugs.launchpad.net/kolla/+bug/1590619 | 13:55 |
| openstack | Launchpad bug 1590619 in kolla "oslo config generator defaults to kolla instead of kollaglue" [High,Fix released] - Assigned to Vikas Jain (vikjain) | 13:55 |
| sdake | vhosakot yes, I saw it over webex on sbezverk_ 'z screen | 13:55 |
| sdake | inc0 there is the bug in question | 13:55 |
| inc0 | yes, nothing there suggest what's the real issue | 13:55 |
| sdake | inc0 the real issue is as described | 13:55 |
| sdake | the playbooks defualt to kollaglue | 13:55 |
| sdake | with oslo-config-generator, the default images are kolla | 13:56 |
| sdake | therefore the images don't work | 13:56 |
| vhosakot | sdake: ah..... I saw lbaas errors when namespace is kolla | 13:56 |
| vhosakot | sdake: let me find the error... | 13:56 |
| vhosakot | it does not affect magnum... | 13:56 |
| sdake | it affects custom plugins at thee moment | 13:56 |
| inc0 | it seems we ran into naming problem | 13:56 |
| inc0 | but changing namespace is not a fix | 13:56 |
| apuimedo | I was checking for example keystone https://hub.docker.com/r/kollaglue/centos-rdo-keystone/tags/ | 13:56 |
| apuimedo | and found it old | 13:56 |
| inc0 | we need to change variable nanme | 13:56 |
| sdake | and pretty much anytime anyone runs custom builds | 13:56 |
| sdake | inc0 - our playbooks use kollaglue | 13:57 |
| sdake | inc0 kolla was taken on the docker hub | 13:57 |
| sdake | inc0 hence, we use kollaglue | 13:57 |
| inc0 | sdake, think of this problem for a moment please | 13:57 |
| inc0 | we have 2 variables with same name in our conf | 13:57 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: Add the KOLLA_KUBERNETES flag to containers https://review.openstack.org/325675 | 13:57 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: The Keystone bootstrap job need to run a db sync https://review.openstack.org/325665 | 13:57 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: Use the Kube endpoint to dictate state instead of etcd https://review.openstack.org/325503 | 13:57 |
| inc0 | one should be kolla another kollaglue | 13:57 |
| mandre | sdake: btw, looks like kolla is available on the dockerhub | 13:58 |
| inc0 | fix to problem of 2 variables with same name is to change name of variable | 13:58 |
| inc0 | not value | 13:58 |
| sdake | inc0 the values o the variables must be synched | 13:58 |
| sdake | atm the values are out of sync | 13:58 |
| inc0 | why? these are 2 different variables | 13:58 |
| sdake | inc0 one is involved in BULDING one is involved in DEPLOYING | 13:58 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Added kolla-kubernetes specific instructions for Docs https://review.openstack.org/326769 | 13:59 |
| sdake | we want to use the same images that we build as we deploy | 13:59 |
| sdake | here is the bottom line | 13:59 |
| sdake | if you use coolsvan's revert, and you build an image with oslo-config-generator's build.conf | 14:00 |
| inc0 | false sdake | 14:00 |
| sdake | it builds images called kolla/whatever | 14:00 |
| sdake | our deployment tools deploy kollaglue/whatever | 14:00 |
| inc0 | one is about generation of /etc/kolla/build-conf and another is to both deploy and build | 14:00 |
| sdake | inc0 obviously you haven't tried it first hand :) | 14:00 |
| sdake | inc0 if you generate a kola-build.conf | 14:01 |
| sdake | it sets the default namespace to kolla | 14:01 |
| sdake | this builds kolla/centos-source-base:400 | 14:01 |
| sdake | the deploy tooks deploy kollaglue/centos-source-base | 14:01 |
| sdake | one of them needs to change | 14:01 |
| sdake | mandre i just tried to register kolla, and it says "username taken" | 14:03 |
| sdake | what would be ideal is to get kolla back from docker hub | 14:03 |
| *** wmiller has quit IRC | 14:03 | |
| sdake | what happened is I registered kolla aroudn the time I had the job change | 14:03 |
| *** wmiller has joined #openstack-kolla | 14:03 | |
| sdake | and it has my redhat email associated with it | 14:03 |
| sdake | which is a black hole | 14:03 |
| mandre | sdake: I was talking about https://hub.docker.com/u/kolla/ returning a 404 | 14:04 |
| sdake | mandre yeah - but the login is taken | 14:04 |
| mandre | would be nice to get the kolla namespace back | 14:04 |
| vhosakot | sbezverk_: ping | 14:05 |
| inc0 | https://review.openstack.org/#/c/327690/1 sdake vhosakot try this | 14:08 |
| patchbot | inc0: patch 327690 - kolla - Change name of config to docker-namespace | 14:08 |
| sdake | mandre i am submitting a support request now to docker.com - hopefully they will address it | 14:08 |
| sdake | https://support.docker.com/hc/en-us/requests/16132 | 14:09 |
| vhosakot | inc0: I see you changed namespace to docker-namespace in the PS | 14:10 |
| sdake | inc0 i see - you think namespace has to do with oslo.config internals | 14:10 |
| inc0 | yes. | 14:10 |
| inc0 | http://docs.openstack.org/developer/oslo.config/generator.html | 14:10 |
| sdake | inc0 ok - so you understand the problem I had pointed out | 14:11 |
| inc0 | changing it to kollaglue messes up imports | 14:11 |
| inc0 | ofc I do, just fix wasn't fixing it | 14:11 |
| sdake | inc0 we had jus tanalyzed it incorrectly, the problem is that namespace is overloaded | 14:11 |
| sdake | inc0 got it, thought you didn't get it ;) | 14:12 |
| sdake | besides the fact that we are violating the deprecation period | 14:12 |
| sdake | which I don't see any solution to | 14:12 |
| sdake | lgtm ;) | 14:12 |
| inc0 | I just demoed behaviour, hence -W | 14:12 |
| inc0 | we can, and should really, put these configs out of [DEFAULT] | 14:13 |
| inc0 | we deprecated this group anyway so we can safely change group rather than config | 14:13 |
| sdake | kuzma 4 point - yay :) | 14:13 |
| *** ravig has quit IRC | 14:18 | |
| *** liyi has joined #openstack-kolla | 14:18 | |
| *** liyi has left #openstack-kolla | 14:19 | |
| *** ccesario has quit IRC | 14:29 | |
| *** ccesario has joined #openstack-kolla | 14:34 | |
| *** vhosakot has quit IRC | 14:52 | |
| apuimedo | sdake: it shouldn't be a black hole | 14:52 |
| apuimedo | your previous manager usually gets the emails | 14:52 |
| apuimedo | at least that's what I saw when I left RH | 14:53 |
| sdake | apuimedo depends on manager | 14:53 |
| apuimedo | I guess os | 14:53 |
| apuimedo | *so | 14:53 |
| sdake | directors generally black hole their emails | 14:53 |
| apuimedo | in any case | 14:53 |
| apuimedo | if you request control back from Docker | 14:53 |
| apuimedo | and they can't contact anybody at that email | 14:53 |
| apuimedo | you'll probably get it back | 14:53 |
| apuimedo | it worked for us with the 'midokura' namespace | 14:53 |
| d_code | I’m running kolla/mitaka… I have intermittent timeouts w/ cloud-init… I’m curious, I’ve got 2 compute nodes and one server that’s all-in-one. should `nova-compute/nova.conf` point to the local hypervisor or the AIO? | 14:55 |
| d_code | I also see that neutron-metadata-agent points to the IP served by haproxy | 14:56 |
| d_code | is that what’s actually served to the guest? | 14:56 |
| *** vhosakot has joined #openstack-kolla | 14:58 | |
| mag009_ | hey | 15:05 |
| mag009_ | how do I tag blueprint | 15:05 |
| mag009_ | i just type blueprint calico ? | 15:05 |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla this is a blueprint of the changes that I made to integrate calico into kolla. blueprint calico https://review.openstack.org/327759 | 15:06 |
| inc0 | mag009_, just type Implenets: blueprint calico | 15:06 |
| *** mliima has quit IRC | 15:06 | |
| sdake | i its only partial | 15:06 |
| sdake | use "Partially-Implements: blueprint calico" | 15:07 |
| inc0 | sdake, why partially? | 15:07 |
| sdake | IF its only partial :) | 15:07 |
| sdake | my keyboard is a bit worn | 15:07 |
| sdake | yay just found 2k out of thin air | 15:08 |
| *** coolsvap has joined #openstack-kolla | 15:08 | |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla adding calico driver into kolla. adding all dependencies of calico etcd and felix. blueprint calico https://review.openstack.org/327759 | 15:08 |
| * sdake hates expense reports | 15:08 | |
| mag009_ | oh | 15:08 |
| mag009_ | ok I'll change it | 15:08 |
| inc0 | mag009_, sooo...you still need quagga tho right? | 15:08 |
| mag009_ | the client is opt to the user it shouldn't be part of the kolla deployement | 15:09 |
| mag009_ | we use quagga but some are using bird or else | 15:09 |
| mag009_ | it's more custom config when it come to the client.. | 15:09 |
| inc0 | I disagree, we should be able to enable_calico: yes and it will do everything at the end imho | 15:09 |
| mag009_ | hard to have a default template that fit for everyone anyway | 15:09 |
| inc0 | it's not about fitting everyone, it's about having batteries included that people can override later | 15:10 |
| inc0 | also mag009_ if you'd write short howto deploy it so it will be fully functional, I'd appreciate it | 15:11 |
| mag009_ | the doc is not done i know | 15:11 |
| Lyncos | did you guys talk about lsblk problem we had ? | 15:11 |
| mag009_ | but if you can review... I just push it out there so you can review it | 15:11 |
| mag009_ | i will add the doc later on | 15:11 |
| inc0 | ok | 15:12 |
| mag009_ | also like lyncos mentionned | 15:12 |
| inc0 | I just really wanted to test it | 15:12 |
| Lyncos | Sorry i just came in :-) | 15:13 |
| mag009_ | hm you'll need a L3 setup | 15:13 |
| inc0 | Lyncos, I'm not aware of this issue, care to remind me? | 15:13 |
| mag009_ | switch that talk bgp or ospf | 15:13 |
| Lyncos | inc0 we started trying to use cache pool .. and we are unable to set a LABEL on the GPT partition | 15:13 |
| sdake | hey folks good news | 15:13 |
| Lyncos | because of the device | 15:13 |
| sdake | docker gave us back kolla ;) | 15:13 |
| mag009_ | they are finally integrate your patch ? | 15:14 |
| mag009_ | no more library ? | 15:14 |
| inc0 | nice, we can move then | 15:14 |
| inc0 | Lyncos, soo...it really boils down to "how do we mark disks for ceph" | 15:15 |
| Lyncos | Yep | 15:15 |
| inc0 | and I'm more and more inclined to do this in inventory | 15:15 |
| mag009_ | inc0: how do you wanna do that ? | 15:15 |
| Lyncos | I think we'll have no problems with that | 15:15 |
| mag009_ | oh i see | 15:16 |
| inc0 | host1 ceph_osd_disks=/dev/vda,/dev/vdb ceph_cache_disks=/dev/vdc | 15:16 |
| mag009_ | a flag true|false | 15:16 |
| *** mliima has joined #openstack-kolla | 15:16 | |
| inc0 | will require a lot of writing tho | 15:16 |
| Lyncos | We are thinking of using dynamic inventories anyway | 15:17 |
| *** mbound has quit IRC | 15:17 | |
| Lyncos | with our in house inventory software | 15:17 |
| coolsvap | sdake, i didn't get | 15:17 |
| mag009_ | we are preping the device anyway with ansible I wrote a small playbook for that | 15:17 |
| mag009_ | which we pass an array defined in our inventory already | 15:18 |
| mag009_ | inc0: if you want to test all you need is to install quagga or bird | 15:18 |
| mag009_ | thats about it | 15:18 |
| inc0 | yeah, I'd really love to have quagga contianer tho | 15:19 |
| mag009_ | i can do one ... | 15:19 |
| Lyncos | We'll eventually need one for coreos | 15:19 |
| mag009_ | its just a matter of installing quagga thats it | 15:19 |
| inc0 | if somebody wants to deploy bird, sure | 15:19 |
| inc0 | just I wanted to deploy full fledged calico easily | 15:19 |
| mag009_ | sure i'll add the container for quagga* | 15:20 |
| Lyncos | Quagga is required for openstack but also for the baremetal part of our deployments | 15:20 |
| *** daneyon_ has joined #openstack-kolla | 15:20 | |
| inc0 | man, this stuff seems so cool | 15:20 |
| Lyncos | It is :-) | 15:20 |
| *** mbound has joined #openstack-kolla | 15:22 | |
| inc0 | mag009_, can you set it up on vms over openstack? | 15:23 |
| inc0 | I have access to cloud, baremetal will be harder | 15:23 |
| mag009_ | I know there's cumulus that you can simulate in vms | 15:25 |
| Lyncos | On cloud you maybe can use cumulus vm | 15:25 |
| mag009_ | which we haven't tried yet | 15:25 |
| *** daneyon_ has quit IRC | 15:25 | |
| Lyncos | Cumulus VX I think | 15:25 |
| *** mikelk has quit IRC | 15:25 | |
| Lyncos | but setting this another story | 15:25 |
| Lyncos | not that it is hard.... but more steps :-) | 15:25 |
| mag009_ | we are going to deploy our devs environment with that actually we aren't sure yet how :) | 15:25 |
| mag009_ | sounds like many devs** i meant dev* | 15:26 |
| inc0 | so let's work together and make it available for public, how about that? | 15:26 |
| inc0 | I mean let's produce total howto deploy devenv with vms with calico | 15:26 |
| mag009_ | our time is short here we have to deliver | 15:26 |
| mag009_ | prod | 15:26 |
| inc0 | sure, this is priority | 15:26 |
| mag009_ | thats why i'm working like a horse on getting calico stuff integrated | 15:27 |
| inc0 | hopefully I can take bulk of it on myself | 15:27 |
| Lyncos | and future is unsure | 15:27 |
| mag009_ | so far so good with kolla | 15:28 |
| inc0 | question | 15:28 |
| inc0 | this code will not deploy ovs, is that planned? | 15:28 |
| mag009_ | you may | 15:29 |
| mag009_ | you may wanna mix your compute | 15:29 |
| inc0 | what I mean | 15:29 |
| inc0 | is neutron_pluing_agent | 15:29 |
| inc0 | if you setup calico | 15:29 |
| inc0 | what will deal with L2 inside compute node? | 15:29 |
| Lyncos | there is no L2 with calico | 15:29 |
| *** belmoreira has quit IRC | 15:29 | |
| Lyncos | Only L3 | 15:29 |
| mag009_ | if a compute is setup to use calico it's only L3 | 15:30 |
| inc0 | hold on now, vms needs to be bridged with dhcp agent somehow | 15:30 |
| mag009_ | yes calico provides | 15:30 |
| Lyncos | it's natted to the VIP of the controller | 15:30 |
| inc0 | ahh | 15:30 |
| mag009_ | calico-dhcp-agent | 15:30 |
| Lyncos | sorry | 15:30 |
| mag009_ | is installed on each compute | 15:30 |
| Lyncos | I'm wrong | 15:30 |
| inc0 | so calico also works with L2 in this manner? | 15:30 |
| inc0 | cool | 15:30 |
| inc0 | that's what I missed | 15:31 |
| Lyncos | for the metadata it is a nat to the vip | 15:31 |
| Lyncos | created by Felix (calico) | 15:31 |
| *** mbound has quit IRC | 15:31 | |
| inc0 | so every vm will get IP from global pool and calico will give them IPs per compute right? | 15:31 |
| Lyncos | Ip's can be assigned anywhere in the cluster | 15:31 |
| Lyncos | it dosen't matter routes will be propagated | 15:32 |
| mag009_ | each compute have a dnsmasq/calico which will assign an ip from the pool correct | 15:32 |
| inc0 | they have to be unique right? | 15:32 |
| Lyncos | no | 15:32 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Convert RabbitMQ to work without HostNetwork=True https://review.openstack.org/326763 | 15:32 |
| Lyncos | if you have 2 time the same ip it will be load balanced via ICMP | 15:32 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Convert MariaDB to work without HostNetwork=True https://review.openstack.org/325613 | 15:32 |
| Lyncos | traffic will reach both IP | 15:32 |
| mag009_ | ECMP* | 15:32 |
| openstackgerrit | Merged openstack/kolla-kubernetes: Convert Memcached to work without HostNetwork=True https://review.openstack.org/326757 | 15:32 |
| Lyncos | yeah ECMP :-) | 15:32 |
| inc0 | ahh, IP load balancing | 15:32 |
| inc0 | cool | 15:32 |
| Lyncos | but Openstack dosen't support that yet | 15:32 |
| inc0 | we have good neutron team | 15:33 |
| Lyncos | there is a small bug in the Kernel in ECMP that prevent it to work properly | 15:33 |
| inc0 | ahh | 15:33 |
| inc0 | will it be fixed? | 15:33 |
| inc0 | anytime soon? | 15:33 |
| Lyncos | ECMP persistance is per packet instead of streams .... | 15:33 |
| Lyncos | Calico already asked Kernel team to re-add this feature (was already there in old kernels) | 15:34 |
| Lyncos | I don't know the status of it... Calico team seems quite small | 15:34 |
| Lyncos | so the features are taking some time to get out | 15:34 |
| Lyncos | But I hope it will be fixed | 15:34 |
| Lyncos | Depends on Linux | 15:34 |
| Lyncos | ;-) | 15:34 |
| Lyncos | I mean Linus | 15:35 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: Add the KOLLA_KUBERNETES flag to containers https://review.openstack.org/325675 | 15:35 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: The Keystone bootstrap job need to run a db sync https://review.openstack.org/325665 | 15:35 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla-kubernetes: Use the Kube endpoint to dictate state instead of etcd https://review.openstack.org/325503 | 15:35 |
| *** mark-casey has joined #openstack-kolla | 15:36 | |
| mark-casey | cu5: was reading scrollback in irc. you get that httpd thing figured out? | 15:36 |
| mark-casey | ^^ me thinks you may be building centos containers on ubuntu | 15:37 |
| *** sacharya has joined #openstack-kolla | 15:37 | |
| *** esharao has joined #openstack-kolla | 15:38 | |
| esharao | hello - i'm running into an issue when i run kolla-ansible build .. i get the following error - 404 Client Error: Not Found (\"Container command \\'kolla_start\\' not found or does not exist.\" | 15:40 |
| esharao | have i missed some step? | 15:40 |
| inc0 | esharao, it seems like build failed somehow, how did you build containers? | 15:40 |
| *** dmk0202 has quit IRC | 15:41 | |
| inc0 | and which container raises this error | 15:41 |
| inc0 | ? | 15:41 |
| esharao | TASK [common : Starting heka container] **************************************** | 15:41 |
| inc0 | which version of kolla? | 15:41 |
| esharao | Mitaka | 15:42 |
| inc0 | ansible 1.9.4 I presume? | 15:42 |
| inc0 | docker version 1.10? | 15:42 |
| inc0 | or 1.11? | 15:42 |
| esharao | ansible --version | 15:43 |
| esharao | ansible 2.1.0.0 | 15:43 |
| inc0 | for stable/mitaka you need 1.9.4 | 15:43 |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla https://review.openstack.org/327759 | 15:43 |
| inc0 | we moved to 2.1+ in master | 15:43 |
| esharao | oh i just followed the instructions.. | 15:44 |
| inc0 | probably for master;) | 15:44 |
| esharao | and docker is Docker version 1.11.2, | 15:44 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Remove liberty warning https://review.openstack.org/327782 | 15:46 |
| inc0 | esharao, you need to downgrade ansible to 1.9.4 | 15:46 |
| esharao | Ok.. | 15:46 |
| esharao | let me do that | 15:46 |
| inc0 | just in case- try to rebuild containers | 15:47 |
| inc0 | because this error seems dockeris | 15:47 |
| inc0 | h | 15:47 |
| esharao | ok.. also.. quick question.. i followed the vagrant kolla env.. so is it the pip ansible i need to downgrade or is it via apt | 15:47 |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla https://review.openstack.org/327759 | 15:48 |
| inc0 | I'm not sure how it works in vagrant | 15:48 |
| inc0 | but if stable/mitaka deploys ansible 1.9.4 | 15:49 |
| inc0 | seems like vagrant bug to me | 15:49 |
| inc0 | I mean our Vagrantfile bug | 15:49 |
| *** godleon has quit IRC | 15:49 | |
| esharao | ok.. let me try to manually downgrade to 1.9.4 and see what happens | 15:49 |
| esharao | will also do a rebuild with 1.9.4 if that does not work | 15:49 |
| esharao | thanks | 15:49 |
| *** Serlex has quit IRC | 15:50 | |
| mag009_ | i've opened the blueprint in launchpad | 15:53 |
| esharao | @inc0 - after downgrading to 1.9.4 i'm running into errors - ImportError: No module named plugins.action | 15:56 |
| esharao | :(.. will just delete the whole file and retry the install again | 15:56 |
| inc0 | esharao, are you sure you're running stable/mitaka? | 16:02 |
| inc0 | not master? | 16:03 |
| mag009_ | mauricio | 16:03 |
| esharao | hmm was checking that as well.. i think im pulling in the latest.. so its master | 16:03 |
| mag009_ | you asked me to put each package on a single line then you asked me to put them back ? | 16:03 |
| inc0 | then 2.1 is ansible version you need;) | 16:03 |
| inc0 | and it seems to fail on container build | 16:03 |
| *** jmccarthy has quit IRC | 16:03 | |
| esharao | container start.. i guesss.. | 16:03 |
| inc0 | which distro/install tyep you use? | 16:04 |
| inc0 | centos/binary? | 16:04 |
| esharao | i was able to do the container build after some changes to vagrantfile (in terms of provider and vcpu) | 16:04 |
| esharao | centos | 16:04 |
| mag009_ | nvm | 16:04 |
| *** severion has quit IRC | 16:04 | |
| *** v1k0d3n has joined #openstack-kolla | 16:05 | |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla https://review.openstack.org/327759 | 16:07 |
| *** ppowell has quit IRC | 16:08 | |
| inc0 | mag009_, and quagga role? would that be possible? | 16:09 |
| inc0 | for ansible | 16:10 |
| inc0 | so you really can deploy full stuff | 16:10 |
| inc0 | full thing | 16:10 |
| mag009_ | yes ok but like i said for the config of quagga itself won't be possible it cannot be generic | 16:11 |
| mag009_ | config will have to be managed with the merge I'll just put the default config | 16:11 |
| inc0 | ok | 16:13 |
| inc0 | once I'll understand what's involved, I'll try to help | 16:13 |
| *** murphyslawbbs has joined #openstack-kolla | 16:13 | |
| *** vhosakot has quit IRC | 16:15 | |
| Mech422 | seriosly, if we're going to enforce grammar in commit messages... | 16:16 |
| Mech422 | people should realize you indent the first line of a paragraph | 16:16 |
| Mech422 | oh! and Good Morning! | 16:17 |
| Mech422 | :0( | 16:17 |
| Mech422 | :-) | 16:17 |
| *** jeh has quit IRC | 16:22 | |
| *** murphyslawbbs has quit IRC | 16:22 | |
| Mech422 | everyone is quiet this morning ? | 16:24 |
| *** sdake has quit IRC | 16:25 | |
| Lyncos | We are all at lunch time :-) | 16:27 |
| Mech422 | Lyncos: oohhh! what we having ? Anything good ? | 16:28 |
| inc0 | Lyncos, moose with maple syrup? | 16:32 |
| Mech422 | Lyncos: oh - VT or CA ? | 16:34 |
| Lyncos | Canada | 16:34 |
| Lyncos | lol | 16:34 |
| *** klindgren has quit IRC | 16:34 | |
| Lyncos | inc0: no today I have a beaver soup | 16:34 |
| Lyncos | moose was yesterday | 16:34 |
| inc0 | :D | 16:34 |
| Mech422 | Lyncos: err...ya..I just released 'ca' looks more like 'california' then 'canada' | 16:34 |
| * Mech422 needs more coffee | 16:35 | |
| Lyncos | yeah | 16:35 |
| Lyncos | Take a redbull | 16:35 |
| Mech422 | Lyncos: When I'm awake, I grew up in NH, up by the canadian border | 16:35 |
| Mech422 | but right now, I'm not really sure what I am :-P | 16:35 |
| Lyncos | I leave by the VT border | 16:36 |
| Mech422 | Lyncos: ahh cool :-) There was an awesome Anthony Bordain 'Parts Unknown' about quebec.... | 16:36 |
| Mech422 | complete with beaver stew | 16:37 |
| Lyncos | it seems Beaver is good | 16:37 |
| Lyncos | I never tried | 16:37 |
| Mech422 | thats what Bordain said...but he eats everything :-P | 16:37 |
| inc0 | you're about to according to what you've said;) | 16:37 |
| Lyncos | ahah | 16:37 |
| inc0 | so east Canada would cool for me if there would be any mountains:/ | 16:38 |
| Lyncos | the montains are in the US | 16:38 |
| *** sdake has joined #openstack-kolla | 16:38 | |
| Lyncos | in NY state | 16:38 |
| mag009_ | inc0: done | 16:38 |
| Lyncos | and a bit in VT | 16:38 |
| mag009_ | for quagga | 16:38 |
| inc0 | well, I need them 20mins of drive;) | 16:38 |
| inc0 | I'm miserable in TX because of lack of mountains | 16:38 |
| Lyncos | what about 1h drive ? | 16:39 |
| Lyncos | not that far | 16:39 |
| Mech422 | inc0: I thought it was cuz you were in TX... | 16:39 |
| inc0 | 1h is ok | 16:39 |
| inc0 | Mech422, well, that too | 16:39 |
| inc0 | not my favorite state | 16:39 |
| sdake | esharao Mech422 lol | 16:39 |
| *** ppowell has joined #openstack-kolla | 16:39 | |
| inc0 | not that I've seen lots of them | 16:39 |
| Mech422 | inc0: hehe - Erricsson seems to be having a hard time convincing openstack people that they want to live in TX | 16:39 |
| Mech422 | sdake: Morning :-) | 16:40 |
| inc0 | Ericsson is in TX? I didn't know | 16:40 |
| *** ntpttr has quit IRC | 16:40 | |
| inc0 | it's just too damn hot here for my liking | 16:40 |
| Mech422 | yeah - in a smallish town - plano ? | 16:40 |
| inc0 | and too flat | 16:40 |
| Mech422 | inc0: Come to AZ - can have nice hot phx, or cool mountain flagstaff :-) | 16:41 |
| inc0 | ahh Dallas | 16:41 |
| Mech422 | inc0: we even have moose upstate | 16:41 |
| inc0 | Mech422, I'm a northerner | 16:41 |
| inc0 | I used to live north of Canadian border as latitude goes | 16:41 |
| Mech422 | inc0: oh? I grew up in some very cold weather with lots of snow - the heat really does make a nice change of pace.... | 16:43 |
| mag009_ | inc0: you can look at my last commit | 16:43 |
| Mech422 | inc0: the heat heat isn't like most places - since its so dry, it can go up to about 105 before its too bad | 16:43 |
| mag009_ | you have the quagga role | 16:43 |
| inc0 | mag009_, thanks a lot! | 16:43 |
| *** ntpttr has joined #openstack-kolla | 16:44 | |
| inc0 | now I need to figure out how to deploy it;) | 16:44 |
| mag009_ | enable_calico , enable_quagga | 16:44 |
| mag009_ | thats about it and put the compute to use calico | 16:44 |
| inc0 | mag009_, but I also need to deploy this cumulus stuff | 16:44 |
| mag009_ | or if you have any bgp setup it would work | 16:45 |
| *** sdake_ has joined #openstack-kolla | 16:45 | |
| mag009_ | bgp or ospf your choice | 16:45 |
| *** sdake has quit IRC | 16:45 | |
| inc0 | well...in a single network L2 this should work anyway right? | 16:45 |
| inc0 | it's when I want to go multi-switch I need to do stuff | 16:45 |
| inc0 | obviously I need multi-switch to properly test it, but for simple use case it should work, right? | 16:46 |
| *** ssurana has joined #openstack-kolla | 16:46 | |
| *** klindgren has joined #openstack-kolla | 16:46 | |
| mag009_ | i'm thinking... lol | 16:46 |
| inc0 | so quagga would announce just simple routes | 16:47 |
| mag009_ | technically yes | 16:47 |
| inc0 | where next hop is goal | 16:47 |
| mag009_ | you dont even need quagga in that case... | 16:47 |
| mag009_ | you'll see that calico assigned an ip and you'll only be able to ping it from your compute node | 16:47 |
| mag009_ | but yes you can test it that way | 16:48 |
| mag009_ | or from your l2 it should also work | 16:48 |
| mag009_ | so I did quagga for nothin ;) | 16:48 |
| mag009_ | well | 16:48 |
| inc0 | nah | 16:48 |
| inc0 | I'll go multi-rack:P | 16:49 |
| inc0 | with it | 16:49 |
| inc0 | it's super simple test we're talking about | 16:49 |
| mag009_ | alright now you make me happy :P | 16:49 |
| inc0 | so question | 16:49 |
| inc0 | each compute node will be separate broadcast domain right? | 16:49 |
| inc0 | and compute node NATs IPs to vms it hosts? | 16:50 |
| Lyncos | broadcast domain is limited to switch port and server that's it | 16:51 |
| Lyncos | there is no nat | 16:51 |
| mag009_ | you allocate a specific range in openstack and it will allocate X ip to your vm | 16:52 |
| mag009_ | thats it | 16:52 |
| Lyncos | we did a /22 subnet for vm's | 16:52 |
| mag009_ | than normally quagga advertise this ip | 16:52 |
| Lyncos | sorry /16 | 16:52 |
| *** sdake_ has quit IRC | 16:52 | |
| inc0 | soo | 16:53 |
| inc0 | I could do 3* non ovelapping /24 instead right? | 16:53 |
| Lyncos | yes | 16:53 |
| inc0 | so if I'd want to setup test env on cloud | 16:53 |
| Lyncos | but everything will be reachable (according to openstack security groups) | 16:53 |
| Lyncos | if you want to test on single machine... no need for quagga | 16:53 |
| inc0 | I can create 3 /24 nets, connect them by cumulus xv vms | 16:53 |
| Lyncos | it will just ping locally | 16:53 |
| inc0 | setup quagga in each pool | 16:54 |
| inc0 | and it would work? | 16:54 |
| Lyncos | 1 quagga for all | 16:54 |
| Lyncos | just need to tell quagga what ranges to advertise and where | 16:55 |
| Lyncos | we are using 192.168.0.0/29 just to make quagga talk to the top of rack switch just to advertise | 16:55 |
| *** diogogmt has joined #openstack-kolla | 16:57 | |
| *** shardy has quit IRC | 16:58 | |
| Mech422 | pbourke: re: stale partition data stuff - I can see a difference in /dev/disk/by-partlabel/ vs sgdisk -p | 17:00 |
| sean-k-mooney | mag009_ maybe since you seem to know how calico works can you answer one question for me. how much work would it be to use calico with ovs. | 17:00 |
| sean-k-mooney | mag009_: i know it currently does not use it at all but in theory a /32 kernel route is not that different then an openflow rule that matches on the ip address | 17:01 |
| mag009_ | you want ovs + calico on the same compute ? | 17:02 |
| sean-k-mooney | mag009_: i was actully thinking of calico programing ovs | 17:03 |
| Lyncos | sean-k-mooney: I don't think this make any sense | 17:03 |
| Lyncos | Calico is replacing OVS | 17:04 |
| sean-k-mooney | Lyncos: well if calico could convert it kernel routes into openflow rule and iptables filters into contrack rules it might | 17:04 |
| *** vhosakot has joined #openstack-kolla | 17:04 | |
| Lyncos | in that case you would have to ask calico guys.. but i'm pretty sure it's not currently supported | 17:04 |
| sean-k-mooney | Lyncos: yes i know i just like the idea of using bgp routing with say dpdk ovs | 17:04 |
| Mech422 | sean-k-mooney: you mean you want ovs for bonding/vlans/etc and Calico to do the routing ? | 17:04 |
| Lyncos | sean-k-mooney: in that case you may be better to use midonet or something like that | 17:05 |
| sean-k-mooney | Mech422: ya basically. | 17:05 |
| Mech422 | sean-k-mooney: that sounds err.. ^^^ what Lyncos said | 17:05 |
| sean-k-mooney | hehe it was just a taught | 17:05 |
| Mech422 | sean-k-mooney: thats how midonet works - it highjacks the OVS kernel module and programs it directly :-) | 17:05 |
| sean-k-mooney | Mech422: but i dont want to use the ovs kernel module at all | 17:06 |
| sean-k-mooney | i want to use the dpdk to bypass the kernel entirely | 17:06 |
| sean-k-mooney | Mech422: anyway just a taught i had not someting im activly looking at | 17:07 |
| Mech422 | sean-k-mooney: ahh - that part, I dunno - my networking know-how isn't deep enough to get a snake's belly wet | 17:07 |
| sean-k-mooney | Mech422: no worries | 17:07 |
| sean-k-mooney | Mech422: calico is interesting | 17:07 |
| Lyncos | We decided to go with calico to not have the ovs overhead | 17:08 |
| sean-k-mooney | after spending 2 years enableing ovs with dpdk and vhost-user in openstack i was just interested if they could be made work together | 17:08 |
| *** daneyon_ has joined #openstack-kolla | 17:09 | |
| *** coolsvap has quit IRC | 17:09 | |
| sean-k-mooney | Lyncos: ya we use dpdk to avoid the kernel over head. | 17:09 |
| sean-k-mooney | its too different approches to the same problem | 17:09 |
| sean-k-mooney | if you keep the dataplan entirely in the kernel or entirly in userspace it can be fast with low overhead | 17:10 |
| Lyncos | why not using OVS hardware offload ? | 17:11 |
| sean-k-mooney | when you mix the two the kernel<->userspace context switchs and packet copies kill your performance | 17:11 |
| sean-k-mooney | Lyncos: that is still slower then dpdk | 17:11 |
| Lyncos | Really ? | 17:11 |
| sean-k-mooney | Lyncos: dpdk can use the hadware offloads too | 17:11 |
| Lyncos | ok | 17:12 |
| Lyncos | I see | 17:12 |
| sean-k-mooney | sorry didnt want to distract form the conversation ye were having about enabling calico/quagga | 17:12 |
| Lyncos | nah it's ok | 17:13 |
| Lyncos | it's nice to see what others think :-) | 17:13 |
| *** mbound has joined #openstack-kolla | 17:13 | |
| *** daneyon_ has quit IRC | 17:13 | |
| inc0 | haha calico+dpdk | 17:14 |
| inc0 | let's compete in "esoteric network tech actually deployed? | 17:14 |
| inc0 | " | 17:14 |
| sean-k-mooney | inc0: stranger things could happen. though i would not like to have to debug it | 17:15 |
| Lyncos | I already hate debugging ovs | 17:16 |
| sean-k-mooney | Lyncos: funny i hate debugging the kernel network stack alot more | 17:16 |
| inc0 | mag009_, so I won't block your PS on that, but I'll need to code up non-ini config merges to make quagga work it seems | 17:16 |
| mag009_ | inc0: thats what I was trying to explain :P | 17:17 |
| mag009_ | the way i proceed is i deploy my quagga during the baremetal deployement | 17:17 |
| inc0 | yeah and when I said "I just need to see what's involved" that's what I meant;) | 17:17 |
| mag009_ | :) | 17:17 |
| inc0 | well you want quagga in container | 17:17 |
| mag009_ | yes eventually | 17:18 |
| inc0 | I'll get right to container configs then | 17:18 |
| mag009_ | we are thinking about switching to coreos for baremetal | 17:18 |
| inc0 | we need to solve it anyway | 17:18 |
| *** mbound has quit IRC | 17:18 | |
| inc0 | you'll be able to use etcd there | 17:18 |
| sean-k-mooney | has anyone deployed kolla on coreos here? | 17:18 |
| mag009_ | yes but I'll still need one for the compute | 17:19 |
| mag009_ | compute are acting as proxy/client | 17:19 |
| mag009_ | i'm working on the precheck right now remember the precheck for ceph | 17:20 |
| mag009_ | /var/lib/ceph | 17:20 |
| mag009_ | i'm going to submit that in a bit | 17:20 |
| sean-k-mooney | mag009_: out of interest are you planning on using coreos to host your kolla container or for your "tenant" baremetal image | 17:21 |
| mag009_ | to host everything | 17:22 |
| sean-k-mooney | ah ok cool | 17:23 |
| mag009_ | even in the cloud I'm thinking about doing coreos + docker | 17:23 |
| sean-k-mooney | i tried to deploy kolla on alpine a while ago and hit some issues. none tha i probably could not have fixed but i wanted to try coreos at some point too | 17:24 |
| sean-k-mooney | i also tried rancher os. i think there is a good case to be made for using supper thin host os with kolla to miniumise the maintence footprint for the host os | 17:26 |
| inc0 | sean-k-mooney, yeah, issue is that we need packages | 17:27 |
| inc0 | things like ceph dont exist in alpine repos | 17:27 |
| inc0 | package management is hard on it's own and we have a lot of dependencies across the board | 17:27 |
| sean-k-mooney | inc0: im not refering to useing alpine to build the container | 17:27 |
| inc0 | ok | 17:28 |
| sean-k-mooney | inc0: i was refering to using it to run them on | 17:28 |
| inc0 | well, lightweight OS on hsot is good idea and I'd go for it | 17:28 |
| sean-k-mooney | the issue i hit with alpine is by default its security polices are too strict to allow docker to run | 17:29 |
| sean-k-mooney | it is relitly easy to fix but at the time i did not want to spent the weekend debuging it | 17:29 |
| *** lmiccini has quit IRC | 17:31 | |
| *** sdake has joined #openstack-kolla | 17:31 | |
| *** salv-orlando has joined #openstack-kolla | 17:32 | |
| *** sdake_ has joined #openstack-kolla | 17:35 | |
| *** salv-orlando has quit IRC | 17:35 | |
| mark-casey | sean-k-mooney: I've gotten prechecks to mostly pass on coreos and gotten the first few containers to deploy | 17:35 |
| mark-casey | I believe andrewwebber has gotten it running fully https://gist.github.com/andrewwebber/f788790772cef47ed5bff39aae2212cf | 17:35 |
| *** aginwala has joined #openstack-kolla | 17:35 | |
| *** salv-orlando has joined #openstack-kolla | 17:36 | |
| sean-k-mooney | mark-casey: is that using coreos as the base os of the contaienr or as the host os on the server? | 17:36 |
| mark-casey | host os on server | 17:36 |
| sean-k-mooney | ok cool | 17:37 |
| *** sdake has quit IRC | 17:37 | |
| Mech422 | Is there a standard config file for the pep8/flake8 stuff? | 17:37 |
| *** lmiccini has joined #openstack-kolla | 17:38 | |
| * Mech422 is housecleaning today | 17:38 | |
| sean-k-mooney | mark-casey: it would be cool to document this in the kolla docs if there was a know good way to deploy it. | 17:39 |
| sean-k-mooney | by the way im not sure if people have seen this but i started work on a playbook to install kolla dependcies on baremetal hosts | 17:40 |
| sean-k-mooney | https://review.openstack.org/#/c/325631/ | 17:40 |
| patchbot | sean-k-mooney: patch 325631 - kolla - [WIP] add baremetal role to install kolla deps. | 17:40 |
| sean-k-mooney | i basically just automate the steps in the quick start | 17:41 |
| *** athomas has quit IRC | 17:43 | |
| sean-k-mooney | it needs some work but the idea is that you can install a basic ubunut or centos image and it will get you to the state where most of the precheck will pass | 17:44 |
| Mech422 | sean-k-mooney: yeah - looks good... | 17:45 |
| Mech422 | sean-k-mooney: my bootstrap role also creates an ansible user, and distributes the ssh keys around | 17:46 |
| Mech422 | sean-k-mooney: and adds all the nodes to /etc/hosts | 17:46 |
| sean-k-mooney | Mech422: ah ok do you have a review up for that? | 17:47 |
| sean-k-mooney | or did i miss it in the codebase? | 17:47 |
| Mech422 | sean-k-mooney: nah - I figured it was pretty site specific | 17:47 |
| sean-k-mooney | ah ok | 17:47 |
| Mech422 | sean-k-mooney: also, on ubuntu 14.04 - looks like i had to ditch pip from apt, and manually install a more recent one ? | 17:47 |
| mark-casey | Mech422: yes. me too. | 17:48 |
| Mech422 | and I pip install ansible==2.1.0.0 | 17:48 |
| Mech422 | mark-casey: oh good - so its not just me then :-) | 17:48 |
| sean-k-mooney | Mech422: i have to double check but i think i install python-setuptools then uses easy_install to install pip | 17:48 |
| mark-casey | sean-k-mooney: I'll add a review on that in the next couple days | 17:49 |
| sean-k-mooney | mark-casey: cool there are alot of white space things in need to fix but any input is welcome | 17:49 |
| Mech422 | sean-k-mooney: I'm just doing get_url: url=https://bootstrap.pypa.io/get-pip.py cuz I'm a slacker :-P | 17:50 |
| Mech422 | sean-k-mooney: oh! if you install the wily kernel - should probably reboot in there somewhere? | 17:51 |
| sean-k-mooney | Mech422: i was tempted to do 'curl -sSL https://get.docker.com/ | sh' but i taught it would be better to do it my self | 17:51 |
| sean-k-mooney | Mech422: yes | 17:52 |
| sean-k-mooney | Mech422: i had planed to register reboot_required and reboot at the end | 17:52 |
| sean-k-mooney | i was also thinking of making the reboot and install of the kernel optional | 17:52 |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla https://review.openstack.org/327759 | 17:52 |
| sean-k-mooney | just incase people did not want to do it but i know there are aufs issue if you dont | 17:52 |
| Mech422 | sean-k-mooney: ahh - cool :-) | 17:53 |
| *** murphyslawbbs has joined #openstack-kolla | 17:53 | |
| sean-k-mooney | part of the reason i am doing this is beace of the bifrost poc. | 17:53 |
| sean-k-mooney | it was annoying to have to keep installing this stuff every time i redeployed the os | 17:54 |
| Mech422 | sean-k-mooney: not sure if you want to do this in bootstrap/baremetal - but you have to add the '--insecure-registry' stuff to docker and restart it at some point if your doing multi-node | 17:54 |
| sean-k-mooney | i need to get back to the bifrost poc too | 17:55 |
| *** ravig has joined #openstack-kolla | 17:55 | |
| Mech422 | sean-k-mooney: LOL - I hear ya :-) Same for me | 17:55 |
| sean-k-mooney | Mech422: oh yes i missed that | 17:55 |
| sdake_ | Mech422 we are making a second set of playbooks to handlethe bare metal initialization | 17:55 |
| *** mummer has joined #openstack-kolla | 17:55 | |
| cinerama | hi sean-k-mooney | 17:55 |
| sean-k-mooney | cinerama: hi | 17:55 |
| sdake_ | sean-k-mooney isn't that correct? | 17:55 |
| sean-k-mooney | sdake_: yes https://review.openstack.org/#/c/325631/1 | 17:56 |
| patchbot | sean-k-mooney: patch 325631 - kolla - [WIP] add baremetal role to install kolla deps. | 17:56 |
| Mech422 | sdake: I actually setup a lil ansible 'preprocessor' - its basically a glorified dictionary merge...but it lets me spec clusters of machines with a 'default' entry so I only have to specify whats different for each machine | 17:56 |
| sean-k-mooney | sdake_: i was not sure if it should be a seperate playbook or just a seperate role | 17:56 |
| *** ravig has quit IRC | 17:57 | |
| Mech422 | really nice feature when you have lots of nics to configure | 17:57 |
| mag009_ | inc0: im splitting the etcd config as asked | 17:57 |
| *** tpeoples has quit IRC | 17:57 | |
| mag009_ | hopefully everything is in order.... afterward | 17:57 |
| cinerama | sean-k-mooney, did you want us to help rebase your change? | 17:57 |
| *** serverascode has quit IRC | 17:58 | |
| sean-k-mooney | cinerama: i was hoping to get back to it tommorow anyway so i should have a rebase done mid friday | 17:59 |
| *** JoseMello has quit IRC | 17:59 | |
| Mech422 | sean-k-mooney: you install 'tox' right ? looks like I had to do it via 'pip' ... | 17:59 |
| *** serverascode has joined #openstack-kolla | 17:59 | |
| *** mgkwill has quit IRC | 18:00 | |
| sean-k-mooney | cinerama: did ye merge the other confliting changes? | 18:00 |
| *** tpeoples has joined #openstack-kolla | 18:00 | |
| sean-k-mooney | Mech422: actully i dont think i do. that was one of the things i wanted to ask about | 18:00 |
| sean-k-mooney | Mech422: its not listed as a dependecy but it is required to generate the configs | 18:00 |
| sean-k-mooney | Mech422: so i was debating if i should install it but its simple to add. | 18:01 |
| *** mgkwill has joined #openstack-kolla | 18:01 | |
| *** JoseMello has joined #openstack-kolla | 18:01 | |
| Mech422 | sean-k-mooney: ansible:latest is gonna blow up the pre-checks I think - they check version numbers on ansible, docker and a couple of others | 18:01 |
| *** aginwala has quit IRC | 18:01 | |
| Mech422 | sean-k-mooney: yeah - its an easy add - I just think the ubuntu 14.04 is old, since I did it via pip ? | 18:01 |
| *** aginwala has joined #openstack-kolla | 18:02 | |
| sean-k-mooney | Mech422: yes that was the other thing i wanted to ask about. i did not cap anisble because i was not sure what version we currently needed | 18:02 |
| mark-casey | remind me what configs tox generates? I've never needed it to get a deploy done | 18:02 |
| sean-k-mooney | mark-casey: it is used to generate the kolla build config | 18:03 |
| Mech422 | sean-k-mooney: actually - on master, I'm having to run from the .tox dirs... | 18:03 |
| sean-k-mooney | mark-casey: so its not needed for deploy just building i think | 18:03 |
| cinerama | sean-k-mooney, yeah i think the controversial things have already been merged - we had a cleanup of handling of booleans in the playbook | 18:03 |
| mark-casey | sean-k-mooney: oh ok. I just pass options to build.py | 18:03 |
| Mech422 | build.py craps out trying to import kolla.image if I use the one in ./tools | 18:03 |
| sean-k-mooney | cinerama: ok cool thats the chage to useing bools and checking for true instead of if not defiend? | 18:04 |
| sean-k-mooney | mark-casey: i could make tox install condtional? | 18:05 |
| sean-k-mooney | Mech422: really i only ever us the build.py in tools | 18:05 |
| sean-k-mooney | Mech422: i never install kolla system wide | 18:05 |
| Mech422 | sean-k-mooney: on stable/mitaka I could use the one in tools... | 18:06 |
| Mech422 | sean-k-mooney: on master, it barfs on me | 18:06 |
| sean-k-mooney | hum ok i have not build master in a little bit. has anyone else had that issue? | 18:06 |
| *** pbourke_ has quit IRC | 18:06 | |
| sdake_ | sean-k-mooney installing kolla system wide is mean for operators that want to use prepackaged containers | 18:06 |
| *** pbourke_ has joined #openstack-kolla | 18:07 | |
| cinerama | sean-k-mooney: yep. so you basically will want to be sure you don't reintroduce that stuff, also we usually will do "value | bool == true" vs "value | bool" | 18:07 |
| sean-k-mooney | sdake_: sure that makes sense. i just normally pip instal -r requirements.txt and then use tools | 18:07 |
| sdake_ | sean-k-mooney that is a developer workflow | 18:07 |
| sdake_ | this i swhat i do as well | 18:08 |
| *** ppowell has quit IRC | 18:08 | |
| inc0 | mag009_, reviewing | 18:08 |
| sean-k-mooney | cinerama: cool that is no problem im happy to use bool== true instead its cleaner i think anyway | 18:08 |
| sean-k-mooney | sdake_: so currently my install role clones kolla into /opt/stack/kolla and does the developer style install | 18:09 |
| inc0 | mag009_, https://review.openstack.org/#/c/327759/7/ansible/roles/etcd/templates/etcd.j2 | 18:10 |
| patchbot | inc0: patch 327759 - kolla - merge calico integration to kolla | 18:10 |
| sean-k-mooney | sdake_: should i have an option to do both so eiter from source or install directly form pip | 18:10 |
| inc0 | so, this should be done differently | 18:10 |
| inc0 | you have 2 configs there | 18:11 |
| mag009_ | yes I know I'm spliting it as we speak :) | 18:11 |
| inc0 | in reality | 18:11 |
| inc0 | ahh | 18:11 |
| mag009_ | i've fixed the rest | 18:11 |
| inc0 | cool, thanks | 18:11 |
| *** ravig has joined #openstack-kolla | 18:12 | |
| *** peterjenkins has quit IRC | 18:13 | |
| *** peterjenkins has joined #openstack-kolla | 18:13 | |
| sean-k-mooney | just realised its after 7 and im ment to be in a pub having goodbye drink with a colleague so got to run. | 18:14 |
| sean-k-mooney | cinerama: ill rebase my change and submit it tommorow | 18:15 |
| Mech422 | sean-k-mooney: have one...errr...two for me! | 18:15 |
| inc0 | drink one for glory of kolla sean-k-mooney | 18:15 |
| inc0 | and have fun;) | 18:15 |
| cinerama | sean-k-mooney, that'd be awesome. but you should go have fun now :) | 18:15 |
| sean-k-mooney | haha i will. catch people tommrow | 18:15 |
| *** ravig has quit IRC | 18:15 | |
| *** akwasnie has quit IRC | 18:15 | |
| *** clayton has quit IRC | 18:17 | |
| *** clayton has joined #openstack-kolla | 18:17 | |
| *** akwasnie has joined #openstack-kolla | 18:18 | |
| *** salv-orlando has quit IRC | 18:21 | |
| *** salv-orlando has joined #openstack-kolla | 18:22 | |
| *** dwalsh has quit IRC | 18:22 | |
| *** mark-casey has quit IRC | 18:25 | |
| *** salv-orlando has quit IRC | 18:26 | |
| *** salv-orlando has joined #openstack-kolla | 18:26 | |
| *** ravig has joined #openstack-kolla | 18:29 | |
| *** pbourke_ has quit IRC | 18:29 | |
| *** Lyncos has quit IRC | 18:30 | |
| *** ravig has quit IRC | 18:33 | |
| *** ayoung has quit IRC | 18:33 | |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla https://review.openstack.org/327759 | 18:33 |
| *** dwalsh has joined #openstack-kolla | 18:37 | |
| *** sdake_ has quit IRC | 18:38 | |
| mag009_ | hm | 18:47 |
| mag009_ | why theres a guy working on etcd ? | 18:47 |
| mag009_ | https://review.openstack.org/#/c/298451 | 18:47 |
| *** ayoung has joined #openstack-kolla | 18:47 | |
| *** NachoDuck has quit IRC | 18:49 | |
| *** NachoDuck has joined #openstack-kolla | 18:51 | |
| inc0 | rhallisey, you are bing summoned | 18:55 |
| rhallisey | eh? | 18:55 |
| inc0 | mag009_, rhallisey talk about etcd | 18:56 |
| rhallisey | that's the ectd container | 18:56 |
| rhallisey | hui wrote that | 18:56 |
| *** sdake has joined #openstack-kolla | 18:59 | |
| mag009_ | yes sure but what is used for ? | 18:59 |
| mag009_ | cause its going to clash with mine.. | 18:59 |
| inc0 | mag009_, k8s | 18:59 |
| inc0 | don't clash, reuse each others work | 19:00 |
| mag009_ | his does not create a cluster... you have to export the config | 19:00 |
| inc0 | mag009_, make it a review:) | 19:01 |
| inc0 | at the end we all want to have etcd cluster | 19:01 |
| inc0 | we need to cooperate towards the end | 19:01 |
| mag009_ | well, mine does it | 19:01 |
| mag009_ | i'm just saying all I need to do is create flag to not use the client | 19:02 |
| mag009_ | mine I require the compute to be client | 19:02 |
| *** NachoDuck has quit IRC | 19:02 | |
| mag009_ | i'll just adjust it if that is the cluster that is only needed for k8s | 19:02 |
| *** NachoDuck has joined #openstack-kolla | 19:03 | |
| *** tzn has quit IRC | 19:06 | |
| *** peterjenkins has quit IRC | 19:10 | |
| *** peterjenkins has joined #openstack-kolla | 19:11 | |
| *** aginwala has quit IRC | 19:11 | |
| *** ravig has joined #openstack-kolla | 19:12 | |
| *** harlowja has quit IRC | 19:12 | |
| wirehead_ | So, the etcd stuff in k8s is interesting. | 19:13 |
| wirehead_ | With Kube 1.2, it’s probably a really bad idea to try to connect containers running on the k8s clusters to the k8s etcd. | 19:14 |
| wirehead_ | mag009_ / inc0 ^ | 19:14 |
| *** aginwala has joined #openstack-kolla | 19:15 | |
| *** aginwala has quit IRC | 19:16 | |
| wirehead_ | Given that there’s already great progress in kubernetes/contrib to have a ansible playbook to create the etcd cluster as part of the HA-kubernetes config, it’s unlikely we’d want to create for kolla-kubernetes a etcd cluster. | 19:16 |
| *** aginwala has joined #openstack-kolla | 19:16 | |
| wirehead_ | Conversely, installing a second instance of etcd inside of the k8s cluster might be necessary. | 19:16 |
| wirehead_ | Kolla-kubernetes workflow can store data in the k8s etcd using Kubernetes Annotations and tags, so anything specific there won’t require that. | 19:17 |
| *** sdake has quit IRC | 19:19 | |
| *** pbourke_ has joined #openstack-kolla | 19:28 | |
| *** murphyslawbbs has quit IRC | 19:29 | |
| *** aginwala has quit IRC | 19:30 | |
| *** pbourke_ has quit IRC | 19:30 | |
| *** Jeffrey4l_ has joined #openstack-kolla | 19:31 | |
| *** Jeffrey4l has quit IRC | 19:34 | |
| *** ravig has quit IRC | 19:36 | |
| *** ravig has joined #openstack-kolla | 19:36 | |
| rhallisey | dcwangmit01, so we need to change bind_hosts | 19:40 |
| rhallisey | everywhere | 19:40 |
| rhallisey | dcwangmit01, let's go over the options one more time | 19:40 |
| rhallisey | change in ansible so we can set it to 0.0.0.0 | 19:41 |
| dcwangmit01 | So, let me describe the problem we are trying to solve for everyone | 19:41 |
| rhallisey | extend dockerfiles to auto discover | 19:42 |
| rhallisey | sure | 19:42 |
| rhallisey | inc0, ^ you may have some input | 19:42 |
| inc0 | I don't understand, can you please describe it more? | 19:43 |
| inc0 | what's the issue | 19:43 |
| inc0 | ? | 19:43 |
| *** dwalsh has quit IRC | 19:43 | |
| *** ravig has quit IRC | 19:43 | |
| dcwangmit01 | Kolla genconfig sets configs of all containers to bind to the IP assigned to api_interface. This works fine for kolla, where targeted hosts are known upfront, facts are gathered, and configs written out properly. However, when running kolla containers in kubernetes, the interface IPs are essentially random. Thus, we need kolla ansible scripts to configure kolla processes to bind to 0.0.0.0 | 19:43 |
| dcwangmit01 | (all intefaces) | 19:44 |
| inc0 | hold on | 19:44 |
| inc0 | we do IP binding for purpose | 19:44 |
| inc0 | why do we need to change it in ansible? | 19:44 |
| rhallisey | dcwangmit01, so we may not need to change it in ansible since this is a kube specific case | 19:44 |
| *** tyrola has joined #openstack-kolla | 19:44 | |
| *** tyrola has quit IRC | 19:45 | |
| rhallisey | I does hurt our ability to consume kolla's config | 19:45 |
| rhallisey | :/ | 19:45 |
| inc0 | only slightly | 19:45 |
| inc0 | you can use same templates | 19:45 |
| rhallisey | inc0, well the way kolla has it does not give kolla-kube the ability to consume it | 19:45 |
| inc0 | just fill if with different variables | 19:45 |
| inc0 | au conraire | 19:46 |
| rhallisey | hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] | 19:46 |
| inc0 | you do need to generate configs per container anyway | 19:46 |
| dcwangmit01 | If we can change the kolla ansible scripts to have the option to override api_interface and instead set bind_host to 0.0.0.0, then we can avoid modifying extend_start.sh for containers to detect if we are in kubernetes, and thus somehown force the process to bind to the right host | 19:46 |
| inc0 | ahh right | 19:46 |
| inc0 | ansible specific stuff | 19:46 |
| inc0 | but we can't rid ourselves of it | 19:46 |
| rhallisey | bind_address will be specific to the host ad run time | 19:46 |
| dcwangmit01 | There are many instances of this: | 19:46 |
| rhallisey | sorry, config time | 19:46 |
| dcwangmit01 | vagrant@vagrant-ubuntu-trusty-64:~/Dev/kolla/ansible/roles$ grep -rni -E '(listen|bind)'| grep -E '(host|ip)' |grep -v haproxy | 19:46 |
| dcwangmit01 | cinder/templates/cinder.conf.j2:37:osapi_volume_listen = {{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }} | 19:46 |
| dcwangmit01 | glance/templates/glance-api.conf.j2:8:bind_host = {{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }} | 19:46 |
| dcwangmit01 | glance/templates/glance-registry.conf.j2:7:bind_host = {{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }} | 19:46 |
| dcwangmit01 | heat/templates/heat.conf.j2:26:bind_host = {{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }} | 19:47 |
| dcwangmit01 | heat/templates/heat.conf.j2:32:bind_host = {{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }} | 19:47 |
| inc0 | we bind to specific address for security reasons | 19:47 |
| inc0 | and we won't change that to 0.0.0.0 | 19:47 |
| inc0 | in ansible | 19:47 |
| rhallisey | inc0, the point is to allow for something to say they want '0.0.0.0' | 19:48 |
| inc0 | but we can figure out way to enable you to override this | 19:48 |
| rhallisey | instead of checking the inventory file | 19:48 |
| inc0 | with ansible we can do this with config overrides and merge_conf | 19:48 |
| *** murphyslawbbs has joined #openstack-kolla | 19:48 | |
| inc0 | we can do something like this | 19:49 |
| *** harlowja has joined #openstack-kolla | 19:49 | |
| wirehead_ | Okay, but the overrides are a bit odd. For mariadb, you want it listening to 0.0.0.0, but you want it to connect to ‘mariadb.openstack’ | 19:49 |
| inc0 | wirehead_, you don't want to listen on 0.0.0.0 | 19:49 |
| inc0 | you want it to be accessible ONLY from network it's meant to use | 19:50 |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla https://review.openstack.org/327759 | 19:51 |
| wirehead_ | Except for HostNetwork=True containers (e.g. Neutron), things are passing through the configured SDN and the right network to use is 0.0.0.0. | 19:51 |
| rhallisey | inc0, but how would you know the network ahead of time? | 19:51 |
| rhallisey | since the service could be on any machine | 19:51 |
| inc0 | witn ansible you know it | 19:51 |
| rhallisey | but with kube your service could be anywhere | 19:52 |
| inc0 | with k8s you can pass it via ENV | 19:52 |
| inc0 | and override it in config | 19:52 |
| rhallisey | but what if the service gets re scheduled | 19:52 |
| rhallisey | ansible can't help that case | 19:52 |
| inc0 | so most naive way to do it | 19:52 |
| inc0 | ifconfig -> get IP from mgmt network and sed it in config | 19:52 |
| dcwangmit01 | inc0: Yes, here's what the pass through ENV option looks like... not pretty, and still requires modifying extend_start.sh for every single kolla container. https://gist.github.com/dcwangmit01/f27eae3e7634861aedcb0a2315f56d1a | 19:52 |
| inc0 | in extended_start | 19:52 |
| *** pbourke_ has joined #openstack-kolla | 19:53 | |
| inc0 | dcwangmit01, we won't have 0.0.0.0 binds because it will open new vector of attack | 19:53 |
| inc0 | it does compromise security | 19:53 |
| wirehead_ | Not in Kubernetes. | 19:53 |
| inc0 | I'm sure we can figure out how to lay configs | 19:53 |
| inc0 | without doing that | 19:53 |
| *** pbourke_ has quit IRC | 19:54 | |
| inc0 | wirehead_, for k8s maybe not if you don't use host networking | 19:54 |
| wirehead_ | If you want to add protection to make sure there’s no possible way to set the listen address in non-k8s to 0.0.0.0, that’s one thing, but a container with HostNetwork=false has no ports exposed ever unless you tell k8s to do so in the pod config. | 19:54 |
| inc0 | if you use host networking, you're screwed just the same | 19:54 |
| dcwangmit01 | inc0: I get what you are saying. There are other options as in not 0.0.0.0. They just require more code modifications, more risk. I'm game for anything | 19:54 |
| wirehead_ | I agree. | 19:54 |
| wirehead_ | We’re not concerned with hostnetwork=true for this. | 19:55 |
| *** dwalsh has joined #openstack-kolla | 19:55 | |
| wirehead_ | For hostnetwork=true we’re going to have to do a bunch of extra work. | 19:55 |
| inc0 | so without host networking, I'm ok with 0.0.0.0 | 19:55 |
| rhallisey | yes without host networking | 19:55 |
| inc0 | you need to figure out configs anyway | 19:55 |
| inc0 | our configs from ansible are jinja2 templates | 19:55 |
| rhallisey | wirehead_, dcwangmit01, ok let's go with configs then | 19:56 |
| inc0 | are you sure you want to use them as they are today? | 19:56 |
| rhallisey | wirehead_, we don't need to spend too much time on a bp | 19:56 |
| wirehead_ | So, we’re proceeding forward with the idea that we’re going to try and preserve the ansible jinja2 templates. | 19:56 |
| rhallisey | wirehead_, though the detail you have is excellent | 19:56 |
| wirehead_ | rhallisey: that’s dcwangmit01’s detail. | 19:56 |
| inc0 | so one thing we can do | 19:56 |
| inc0 | is to move all ansible-specific stuff out of templates | 19:57 |
| rhallisey | dcwangmit01, nice :) | 19:57 |
| inc0 | so somewhere in plays we set up "api_interface_address = hostvars..." | 19:57 |
| rhallisey | I'd rather do this on the config side so we can avoid any automagic | 19:57 |
| inc0 | and in tpls we provide var api_interface_address | 19:57 |
| inc0 | which you can override | 19:57 |
| inc0 | or you can to simulate ansible's datastructures | 19:57 |
| inc0 | so you will render templates with dict hostwar... | 19:58 |
| rhallisey | dcwangmit01, https://blueprints.launchpad.net/kolla/ | 19:58 |
| rhallisey | copy paste the details into a new BP | 19:58 |
| rhallisey | and tag the patches with the BP name | 19:58 |
| *** ppowell has joined #openstack-kolla | 20:00 | |
| dcwangmit01 | inc0: Need to clarify. api_interface_address = hostvars or "0.0.0.0" or {any-ip}. Then if api_interface_address == hostvars, then do the existing resolving of interface to IP. Otherwise directly copy the setting into any bind_ip or bind_host setting. This will still require modifications to lots of templates | 20:00 |
| dcwangmit01 | ^ Is that what you want? | 20:00 |
| inc0 | not really | 20:01 |
| inc0 | currently we call {{ hostvars ... }} | 20:01 |
| inc0 | in templates | 20:01 |
| dcwangmit01 | ^ yes | 20:01 |
| inc0 | if we create new variable | 20:01 |
| inc0 | which will host value of ipv4 from hostvars | 20:02 |
| inc0 | and use it in templates | 20:02 |
| inc0 | logic stays the same | 20:02 |
| inc0 | but you'll have shorted value to override | 20:02 |
| wirehead_ | So, right now, MariaDB is {{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }} | 20:04 |
| wirehead_ | We’re changing that to {{ iface[ … ] }} | 20:04 |
| wirehead_ | where iface is the new variable? | 20:04 |
| inc0 | yeah | 20:04 |
| wirehead_ | Okay, and then we’re going to have something in globals.yml that sets a key for iface[mariadb] = 0.0.0.0 | 20:06 |
| inc0 | pretty much | 20:07 |
| inc0 | of actual ip if you want to go fancy | 20:07 |
| *** ravig has joined #openstack-kolla | 20:07 | |
| wirehead_ | and then there’s a step in `kolla-ansible genconfig` that copies by default the hostvars over to iface, but does an override otherwise. | 20:08 |
| inc0 | nope | 20:08 |
| *** ravig has quit IRC | 20:08 | |
| inc0 | it won't be separate step | 20:08 |
| inc0 | it will be task | 20:08 |
| inc0 | - set: mgmt_iface_addr = hostvars... | 20:09 |
| *** ravig has joined #openstack-kolla | 20:10 | |
| *** huikang has joined #openstack-kolla | 20:10 | |
| wirehead_ | inc0: ok so something in ansible/roles/mariadb/tasks/config.yml | 20:11 |
| inc0 | yeah, or even common | 20:11 |
| *** ravig has quit IRC | 20:11 | |
| inc0 | since we use this var multiple times | 20:11 |
| wirehead_ | So, in terms of patch order, first we do a patch that adds that task to copy vars. And then a series of patches against various containers to actually use the new var | 20:13 |
| inc0 | correct | 20:13 |
| dcwangmit01 | inc0: More clarification... So mgmt_interface_addr is a new variable defined by a task. It defaults to hostvars[inventory_hostname... unless iface[mariadb] is set. Then we change all of the templates from bind_host = {{ hostvars[inventory_hostname]['ansible_' + api_interface]['ipv4']['address'] }} to: bind_host = mgmt_interface_addr | 20:16 |
| inc0 | no, no unless | 20:17 |
| inc0 | you're not using ansible playbooks so setup config, so you don't need to change anything | 20:17 |
| inc0 | just make sure {{ mgmt_iface_addr }} is set to whatever you want | 20:17 |
| wirehead_ | I’m not quite following the ‘no unless’ part, but that might be my lack of ansible-fu which is charatably at the level of manager for the most part (e.g. “It’s a simple matter of ansible, right?”) | 20:21 |
| inc0 | no, that was my language-fu failing | 20:21 |
| inc0 | I was commenting "unless iface[mariadb] is set" | 20:22 |
| inc0 | we don't do logic there | 20:22 |
| inc0 | we only fill variable | 20:22 |
| wirehead_ | mgmt_interface_addr = iface[mariadb] or hostvars[… | 20:22 |
| dcwangmit01 | If iface[mariadb] is set, then mgmt_interface_addr = the value of iface[mariadb]. Otherwise mgmt_inteface_addr = hostvars[... ? | 20:23 |
| inc0 | wirehead_, no | 20:23 |
| inc0 | you're not using ansible unless I'm mistaken | 20:23 |
| inc0 | hold on, I'll make example patchset | 20:23 |
| dcwangmit01 | You've given up on us :) | 20:23 |
| wirehead_ | Like I said “It’s just a simple matter of ansible, right” level of ansible. :D | 20:23 |
| inc0 | yeah I've given up.:P j/k | 20:24 |
| inc0 | just demo will be quick | 20:24 |
| *** ppowell has quit IRC | 20:24 | |
| inc0 | and you'll see what I mean instantly | 20:24 |
| *** mbound has joined #openstack-kolla | 20:27 | |
| *** aginwala has joined #openstack-kolla | 20:28 | |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: WIP: enable kolla k8s to modify configs https://review.openstack.org/327925 | 20:31 |
| inc0 | see? | 20:31 |
| inc0 | now do this for everything else and we're golden:P | 20:31 |
| mag009_ | inc0: got a sec | 20:31 |
| inc0 | you might want to drop this galera stuff too | 20:31 |
| inc0 | mag009_, shoot | 20:31 |
| mag009_ | remember that : https://review.openstack.org/#/c/325999/ | 20:32 |
| patchbot | mag009_: patch 325999 - kolla - add become to mount osds getting a permission deni... (ABANDONED) | 20:32 |
| mag009_ | I've added a check to make sure the /var/lib/ceph is writable | 20:32 |
| inc0 | yeah | 20:32 |
| mag009_ | but now i'm getting a perm denied to /etc/fstab... | 20:32 |
| inc0 | uhh | 20:32 |
| mag009_ | mount definitely need become... | 20:32 |
| mag009_ | it need to write the mount to fstab | 20:33 |
| inc0 | yeah that's true | 20:33 |
| inc0 | well, meh I guess | 20:33 |
| inc0 | I could argue that fstab can be writable to underprivileged uset | 20:33 |
| inc0 | I won't. | 20:33 |
| mag009_ | well yes if the user is in the group | 20:34 |
| mag009_ | hm not even | 20:34 |
| *** dwalsh has quit IRC | 20:34 | |
| mag009_ | fstab is root writable only | 20:34 |
| mag009_ | how would you write as an unprivileged user ? | 20:34 |
| mag009_ | via sudoers ? | 20:35 |
| inc0 | anyway mag009_ I don't want to argue that | 20:35 |
| inc0 | sudo would work | 20:35 |
| inc0 | but become will do sudo | 20:35 |
| mag009_ | i know but it's a bug... | 20:35 |
| mag009_ | you cannot deploy a ceph osd because of that.. I don't understand how user have been able to use ceph osd out of the box.. | 20:36 |
| mag009_ | i'm not arguing I want to fix it | 20:36 |
| mag009_ | :) | 20:36 |
| *** ppowell has joined #openstack-kolla | 20:37 | |
| *** murphyslawbbs has quit IRC | 20:37 | |
| mag009_ | so ? | 20:39 |
| mag009_ | what is your recommendation? | 20:39 |
| inc0 | sure, just become it | 20:39 |
| *** jtriley has quit IRC | 20:39 | |
| inc0 | and make note in docs that for ceph you really need to deploy as root | 20:39 |
| mag009_ | well I'm using sudo not root | 20:40 |
| inc0 | ansible become can use sude too | 20:41 |
| inc0 | all I'm saying, just add this become: True and be done with it | 20:41 |
| wirehead_ | Okay, I think I’m fine with that as long as the larger Kolla community is fine with that. :D | 20:41 |
| dcwangmit01 | inc0: Okay, we get it now. After running the code. | 20:42 |
| dcwangmit01 | +2, if I could | 20:42 |
| dcwangmit01 | So I should write a blueprint? | 20:42 |
| inc0 | well, needs more testing | 20:42 |
| dcwangmit01 | I was joking about, "hey every +2 real quick" | 20:43 |
| inc0 | and you guys need to figure out what to do with wsrep vars and vars that requires sets of addresses | 20:43 |
| inc0 | like haproxy | 20:43 |
| dcwangmit01 | So, do I write a doc in kolla/spec, and then add a blueprint to link to that? | 20:43 |
| inc0 | most of it will be about clusters, so not relevant to k8s | 20:43 |
| inc0 | no, just figure it out and take over this patchset | 20:43 |
| wirehead_ | In the immortal words of every developer ever: “It works on my machine!” :D | 20:44 |
| inc0 | SOA1: Standard operators answer 1: works for me | 20:44 |
| rhallisey | dcwangmit01, did you write a spec or a blueprint | 20:46 |
| rhallisey | write a blueprint. A spec is meant for major changes | 20:46 |
| wirehead_ | rhallisey: I think we’ve diverged very far from the blueprint dcwangmit01 wrote out already. | 20:46 |
| dcwangmit01 | I did write something up. THe problem statement is still correct, and I can modify with the details of what was agreed here | 20:47 |
| rhallisey | wirehead_, it's ok. Just write a bp that say something like 'we need to allow for bind_host to be more configurable for kolla-kubernetes' | 20:47 |
| inc0 | rhallisey, https://review.openstack.org/#/c/327925/ | 20:48 |
| patchbot | inc0: patch 327925 - kolla - WIP: enable kolla k8s to modify configs | 20:48 |
| rhallisey | the details of why should be in the dropping net=host in kolla-kube | 20:48 |
| rhallisey | https://blueprints.launchpad.net/kolla-kubernetes/+spec/remove-net-host | 20:48 |
| inc0 | well, issue is why we use net=host in the firs tplace | 20:48 |
| inc0 | but you guys need to determine if we won't die on net performance | 20:49 |
| inc0 | I don't trust docker networking | 20:49 |
| rhallisey | I think wirehead_ has been playing around with it a bit | 20:49 |
| wirehead_ | Yeah. | 20:49 |
| wirehead_ | Modern Kubernetes doesn’t have a proxy, a la docker networking. | 20:49 |
| wirehead_ | AFAICT it does most of the work by orchestrating iptables. | 20:50 |
| *** d_code_ has joined #openstack-kolla | 20:51 | |
| wirehead_ | I still have some suspicions that Neutron is going to conflict hilariously with kubernetes networking. | 20:51 |
| inc0 | so forget about dropping net=host for neutron agents | 20:52 |
| inc0 | neutron needs full access to namespaces, iptables and such | 20:52 |
| *** d_code has quit IRC | 20:52 | |
| *** d_code_ is now known as d_code | 20:52 | |
| *** neophy has joined #openstack-kolla | 20:52 | |
| inc0 | you don't want ot to float around too | 20:52 |
| wirehead_ | Yeah. My perception, as a bowtie-wearing-Kubernetes partisan, is that we should default to kube networking and then re-assess if necessary. | 20:53 |
| inc0 | yeah, that'd be my approach | 20:53 |
| inc0 | *if* k8s networking is as performant as host | 20:53 |
| inc0 | or at least close enough | 20:53 |
| *** Lyncos has joined #openstack-kolla | 20:54 | |
| wirehead_ | Then again, this is a giant break with the Way Things Have Always Been. | 20:54 |
| wirehead_ | So the fear of everything blowing up in our face is a rational fear. | 20:54 |
| *** d_code_ has joined #openstack-kolla | 20:55 | |
| *** iceyao has joined #openstack-kolla | 20:55 | |
| mag009_ | The requested image does not exist: kollaglue/oraclelinux-source-heka:2.0.0 | 20:55 |
| mag009_ | in my CI run | 20:56 |
| *** aginwala has quit IRC | 20:56 | |
| inc0 | mag009_, we enabled oraclelinux | 20:56 |
| inc0 | if you need help with that, pbourke would be person to talk to | 20:56 |
| inc0 | but he's asleep now probably | 20:56 |
| mag009_ | alright well it seem all my deploy run failed | 20:56 |
| mag009_ | not just oracle | 20:56 |
| *** d_code has quit IRC | 20:56 | |
| *** d_code_ is now known as d_code | 20:56 | |
| wirehead_ | rhallisey / dcwangmit01: Are we going to more carefully specify api_interface_address? I don’t think we need to. | 20:57 |
| wirehead_ | rhallisey: Either it’s a neutron agent or it’s hostnetwork=false. | 20:57 |
| mag009_ | i'll check with him tomorrow | 20:57 |
| *** iceyao has quit IRC | 21:00 | |
| *** aginwala has joined #openstack-kolla | 21:00 | |
| *** aginwala has quit IRC | 21:01 | |
| openstackgerrit | Marc-Andre Gatien proposed openstack/kolla: merge calico integration to kolla https://review.openstack.org/327759 | 21:03 |
| *** mliima has quit IRC | 21:05 | |
| Mech422 | mag009_: I need image tag 2.0.2 for stable/mitaka and 3.0.0 for master | 21:06 |
| mag009_ | ? | 21:07 |
| mag009_ | what image | 21:07 |
| Mech422 | umm - all of them | 21:07 |
| Mech422 | sec | 21:07 |
| *** aginwala has joined #openstack-kolla | 21:07 | |
| mag009_ | how do i do that? | 21:07 |
| Mech422 | # Valid option is Docker repository tag | 21:07 |
| Mech422 | openstack_release: "3.0.0" | 21:07 |
| Mech422 | #openstack_release: "2.0.1" | 21:07 |
| Mech422 | in globals | 21:08 |
| mag009_ | im not getting it | 21:09 |
| mag009_ | what you want me to do? | 21:09 |
| Mech422 | in /etc/kolla/globals - set openstack release to "3.0.0" if your using master | 21:09 |
| Mech422 | and 2.0.1 if your using stable/mitaka | 21:09 |
| Mech422 | that seems to control the version tag used in looking up images | 21:10 |
| Mech422 | your trying to find: oraclelinux-source-heka:2.0.0 | 21:10 |
| Mech422 | but my build runs seem to create oraclelinux-source-heka:3.0.0 on master | 21:10 |
| *** murphyslawbbs has joined #openstack-kolla | 21:11 | |
| mag009_ | I'm talking about the CI when I push my code | 21:11 |
| Mech422 | mag009_: oh sorry | 21:11 |
| dcwangmit01 | wirehead_: I don't think we need to more carefully specify api_inteface_address. The way it's organized by inc0's patch it should be easy to add in the future if we need that resolution. Right now, the only need for the setting is for kubernetes or not | 21:11 |
| Mech422 | mag009_: Oracle linux blew up on my CI run too - I think it might be being 'tempermental' | 21:12 |
| openstackgerrit | Ken Wronkiewicz proposed openstack/kolla-kubernetes: Add documentation for api_interface_address https://review.openstack.org/327961 | 21:13 |
| wirehead_ | Did I mention I’m obsessed with docs? :D | 21:14 |
| mag009_ | mine failed on all | 21:14 |
| mag009_ | for the other i didnt check yet | 21:14 |
| mag009_ | might be my fault | 21:14 |
| Mech422 | I get the same heka failure on oracle....maybe someone needs to update the globals file used for CI ? | 21:17 |
| wirehead_ | Sounds like we’ve got a heka lot of failures going on. | 21:23 |
| openstackgerrit | Ken Wronkiewicz proposed openstack/kolla-kubernetes: Add documentation for api_interface_address https://review.openstack.org/327961 | 21:24 |
| *** ssurana has quit IRC | 21:25 | |
| *** ssurana has joined #openstack-kolla | 21:25 | |
| Mech422 | wirehead_: *Groan* :-p | 21:32 |
| *** ayoung has quit IRC | 21:34 | |
| *** ayoung has joined #openstack-kolla | 21:35 | |
| *** murphyslawbbs has quit IRC | 21:37 | |
| *** aginwala has quit IRC | 21:38 | |
| *** ppowell has quit IRC | 21:40 | |
| *** murphyslawbbs has joined #openstack-kolla | 21:46 | |
| *** mbound has quit IRC | 21:47 | |
| *** ravig has joined #openstack-kolla | 21:49 | |
| *** inc0 has quit IRC | 21:51 | |
| *** huikang has quit IRC | 21:56 | |
| *** esharao has quit IRC | 21:57 | |
| mag009_ | i passed the CI!!! | 21:57 |
| Mech422 | mag009_: Sweet - how'd you make Oracle happy ? | 21:58 |
| Lyncos | rm -rf oracle | 21:58 |
| Lyncos | ;-) | 21:58 |
| Mech422 | Lyncos: no, that makes ME happy :-P | 21:58 |
| mag009_ | oracle fail | 21:58 |
| mag009_ | but i got +1 from the CI | 21:58 |
| mag009_ | so they probably know its failing | 21:59 |
| mag009_ | it's new for oracle | 21:59 |
| *** jtriley has joined #openstack-kolla | 21:59 | |
| Mech422 | mag009_: ahh - cool | 21:59 |
| Mech422 | mag009_: heh - that sounded like a pretty big patch too - your ending the week STRONG! :-) | 22:01 |
| *** murphyslawbbs has quit IRC | 22:02 | |
| *** ayoung has quit IRC | 22:05 | |
| wirehead_ | rhallisey / dcwangmit01: OK. So, I applied inc0’s patch and, outside of rhallisey’s 3 patches, mainline should correctly provision mariadb. | 22:06 |
| dcwangmit01 | wirehead_: I'm still testing, I mean re-building my vagrant env. | 22:07 |
| wirehead_ | dcwangmit01: https://xkcd.com/303/ | 22:08 |
| wirehead_ | Where are the swords? | 22:08 |
| *** murphyslawbbs has joined #openstack-kolla | 22:16 | |
| *** murphyslawbbs has quit IRC | 22:18 | |
| *** murphyslawbbs has joined #openstack-kolla | 22:18 | |
| openstackgerrit | Ken Wronkiewicz proposed openstack/kolla: Fix interface address for memcached atop kubernetes https://review.openstack.org/327987 | 22:19 |
| *** rhallisey has quit IRC | 22:22 | |
| *** murphyslawbbs has quit IRC | 22:23 | |
| *** vhosakot has quit IRC | 22:25 | |
| *** SiRiuS__ has quit IRC | 22:25 | |
| *** jtriley has quit IRC | 22:33 | |
| *** neophy has quit IRC | 22:37 | |
| *** ravig has quit IRC | 22:40 | |
| *** murphyslawbbs has joined #openstack-kolla | 22:40 | |
| *** ravig has joined #openstack-kolla | 22:40 | |
| *** mbound has joined #openstack-kolla | 22:48 | |
| *** mbound has quit IRC | 22:53 | |
| openstackgerrit | David Wang proposed openstack/kolla: Enable kolla k8s to override bind api bind address in genconfig https://review.openstack.org/327925 | 23:08 |
| *** ravig has quit IRC | 23:09 | |
| *** ravig has joined #openstack-kolla | 23:09 | |
| *** jtriley has joined #openstack-kolla | 23:16 | |
| *** ravig has quit IRC | 23:19 | |
| *** salv-orl_ has joined #openstack-kolla | 23:27 | |
| *** salv-orl_ has quit IRC | 23:28 | |
| *** salv-orl_ has joined #openstack-kolla | 23:29 | |
| *** salv-orlando has quit IRC | 23:31 | |
| *** jtriley has quit IRC | 23:46 | |
| *** jtriley has joined #openstack-kolla | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!