| *** tonanhngo has quit IRC | 00:02 | |
| *** tonanhngo has joined #openstack-kuryr | 00:08 | |
| *** vikasc has joined #openstack-kuryr | 00:11 | |
| *** yamamoto_ has joined #openstack-kuryr | 00:40 | |
| *** limao has joined #openstack-kuryr | 00:46 | |
| *** yamamoto_ has quit IRC | 00:48 | |
| *** shashank_hegde has quit IRC | 00:48 | |
| *** flaviof has quit IRC | 01:11 | |
| *** yamamoto_ has joined #openstack-kuryr | 01:17 | |
| *** salv-orlando has joined #openstack-kuryr | 01:17 | |
| *** yamamoto_ has quit IRC | 01:22 | |
| *** salv-orlando has quit IRC | 01:22 | |
| *** sdake_ has quit IRC | 01:25 | |
| *** flaviof has joined #openstack-kuryr | 01:26 | |
| *** sdake has joined #openstack-kuryr | 01:30 | |
| *** tonanhngo has quit IRC | 01:35 | |
| *** tonanhngo has joined #openstack-kuryr | 01:37 | |
| *** tonanhngo has quit IRC | 01:38 | |
| *** tonanhngo has joined #openstack-kuryr | 01:39 | |
| *** tonanhngo has quit IRC | 01:44 | |
| *** tonanhngo has joined #openstack-kuryr | 01:45 | |
| *** huikang has joined #openstack-kuryr | 01:58 | |
| *** yamamoto_ has joined #openstack-kuryr | 02:22 | |
| *** salv-orlando has joined #openstack-kuryr | 02:24 | |
| *** salv-orlando has quit IRC | 02:27 | |
| *** yamamoto_ has quit IRC | 02:29 | |
| *** tonanhngo has quit IRC | 02:32 | |
| *** yamamoto_ has joined #openstack-kuryr | 02:33 | |
| openstackgerrit | Liping Mao proposed openstack/kuryr-libnetwork: Fix no log in /var/log/kuryr/kuryr.log https://review.openstack.org/361795 | 02:33 |
|---|---|---|
| openstackgerrit | Liping Mao proposed openstack/kuryr-libnetwork: Fix no log in /var/log/kuryr/kuryr.log https://review.openstack.org/361795 | 02:35 |
| *** yamamoto_ has quit IRC | 02:42 | |
| *** yuanying has quit IRC | 02:51 | |
| *** yuanying has joined #openstack-kuryr | 02:52 | |
| *** huikang has quit IRC | 03:07 | |
| *** limao has quit IRC | 03:11 | |
| *** huikang has joined #openstack-kuryr | 03:18 | |
| *** huikang has quit IRC | 03:23 | |
| *** huikang has joined #openstack-kuryr | 03:31 | |
| *** yamamoto_ has joined #openstack-kuryr | 03:43 | |
| *** yuanying has quit IRC | 03:44 | |
| *** yamamoto_ has quit IRC | 03:47 | |
| *** shashank_hegde has joined #openstack-kuryr | 04:05 | |
| *** limao has joined #openstack-kuryr | 04:10 | |
| *** diga has joined #openstack-kuryr | 04:11 | |
| *** limao has quit IRC | 04:15 | |
| *** yamamoto_ has joined #openstack-kuryr | 04:17 | |
| *** salv-orlando has joined #openstack-kuryr | 04:18 | |
| *** yamamoto_ has quit IRC | 04:20 | |
| *** yamamoto_ has joined #openstack-kuryr | 04:20 | |
| *** yamamoto_ has quit IRC | 04:22 | |
| *** salv-orlando has quit IRC | 04:23 | |
| *** limao has joined #openstack-kuryr | 04:28 | |
| *** janki has joined #openstack-kuryr | 04:35 | |
| *** yuanying has joined #openstack-kuryr | 05:03 | |
| *** yamamoto_ has joined #openstack-kuryr | 05:07 | |
| *** sdake_ has joined #openstack-kuryr | 05:10 | |
| openstackgerrit | vikas choudhary proposed openstack/kuryr-libnetwork: Code restructuring: neutron client as rest driver from Kuryr lib https://review.openstack.org/342614 | 05:11 |
| *** sdake_ is now known as dake | 05:13 | |
| *** sdake has quit IRC | 05:13 | |
| *** huikang has quit IRC | 05:13 | |
| *** dake is now known as sdake | 05:13 | |
| *** yamamoto_ has quit IRC | 05:16 | |
| *** yamamoto_ has joined #openstack-kuryr | 05:20 | |
| *** sdake has quit IRC | 05:26 | |
| openstackgerrit | Liping Mao proposed openstack/kuryr-libnetwork: Fix oslo-config-generator failed in kuryr-libnetwork https://review.openstack.org/361841 | 05:33 |
| *** salv-orlando has joined #openstack-kuryr | 05:36 | |
| *** janonymous has quit IRC | 05:38 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/kuryr-libnetwork: Updated from global requirements https://review.openstack.org/351976 | 05:40 |
| *** yamamoto_ has quit IRC | 05:44 | |
| *** yamamoto_ has joined #openstack-kuryr | 05:45 | |
| *** yamamoto_ has quit IRC | 05:49 | |
| *** yamamoto_ has joined #openstack-kuryr | 05:49 | |
| *** yamamoto_ has quit IRC | 05:57 | |
| openstackgerrit | Liping Mao proposed openstack/kuryr-libnetwork: Fix oslo-config-generator failed in kuryr-libnetwork https://review.openstack.org/361841 | 06:04 |
| *** salv-orlando has quit IRC | 06:05 | |
| *** salv-orlando has joined #openstack-kuryr | 06:07 | |
| *** yuanying has quit IRC | 06:29 | |
| *** yamamoto_ has joined #openstack-kuryr | 06:58 | |
| *** gsagie has joined #openstack-kuryr | 06:59 | |
| *** yamamoto_ has quit IRC | 07:02 | |
| *** yamamoto_ has joined #openstack-kuryr | 07:04 | |
| apuimedo | Hi all | 07:04 |
| apuimedo | I'm back from holidays :-) | 07:04 |
| *** yamamoto_ has quit IRC | 07:13 | |
| gsagie | ohhh yeah! :) | 07:13 |
| *** shashank_hegde has quit IRC | 07:13 | |
| *** shashank_hegde has joined #openstack-kuryr | 07:15 | |
| *** yamamoto_ has joined #openstack-kuryr | 07:19 | |
| *** shashank_hegde has quit IRC | 07:23 | |
| *** yamamoto_ has quit IRC | 07:32 | |
| *** yamamoto_ has joined #openstack-kuryr | 07:34 | |
| *** yamamoto_ has quit IRC | 07:34 | |
| *** pablochacin has joined #openstack-kuryr | 07:41 | |
| apuimedo | limao: vikasc: In regards to https://review.openstack.org/#/c/361560/1/setup.cfg | 07:46 |
| apuimedo | it is against the infra guidelines to put stuff on places for which you need root access | 07:47 |
| apuimedo | we used to have the data_files already pointing at /usr/libexec/kuryr | 07:47 |
| apuimedo | and had to change it because of that | 07:47 |
| apuimedo | the current state is that packaging should be who installs the binding scripts on the location, not setup.py | 07:48 |
| *** yuanying has joined #openstack-kuryr | 07:51 | |
| limao | Hi apuimedo | 07:56 |
| apuimedo | hi | 07:56 |
| limao | I remove the +1, your comments make sense, currently devstack will use setup.py to install it, so we maybe need to update the devstack script to patch this? | 08:00 |
| *** yamamoto_ has joined #openstack-kuryr | 08:01 | |
| *** yamamoto_ has quit IRC | 08:03 | |
| *** icoughla has joined #openstack-kuryr | 08:04 | |
| vikasc | apuimedo, got it. | 08:12 |
| apuimedo | limao: right | 08:13 |
| vikasc | apuimedo, should we change default path of bindir in config.py then? | 08:13 |
| apuimedo | we have to update the plugin | 08:13 |
| apuimedo | *devstack plugin | 08:13 |
| apuimedo | vikasc: no, that is still the default location | 08:13 |
| apuimedo | just like neutron has a config place for /var/lib/neutron | 08:13 |
| apuimedo | but does not install it | 08:14 |
| vikasc | got it | 08:14 |
| apuimedo | (it creates it on runtime) | 08:14 |
| vikasc | apuimedo, thanks. will abandon | 08:14 |
| apuimedo | vikasc: thanks | 08:15 |
| vikasc | apuimedo, limao and irenab have some concerns on removing kuryr-libnetwork/common | 08:15 |
| vikasc | apuimedo, if we take neutron as reference then we should keep it but logically i would prefer not to have. | 08:17 |
| vikasc | https://review.openstack.org/#/c/361567/ | 08:17 |
| vikasc | may be we can discuss in meeting today | 08:19 |
| apuimedo | vikasc: yes. Best to discuss it in the meeting. I personally would drop 'common', since we don't have kuryr-libnetwork plugins | 08:21 |
| vikasc | apuimedo, and if we decide to keep, should be kept in kuryr-k8s as well | 08:21 |
| apuimedo | vikasc: right | 08:24 |
| *** vikasc has quit IRC | 08:26 | |
| *** vikasc has joined #openstack-kuryr | 08:27 | |
| *** gsagie has quit IRC | 08:36 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr: Drop unused requirements https://review.openstack.org/361937 | 08:38 |
| apuimedo | irenab: vikasc: limao: please, check the patch I just sent | 08:39 |
| apuimedo | I'm trying to make the packaging for kuryr | 08:39 |
| apuimedo | and having these extra deps is quite inconvenient | 08:39 |
| vikasc | apuimedo, thanks!! | 08:40 |
| apuimedo | ;-) | 08:40 |
| openstackgerrit | Dongcan Ye proposed openstack/kuryr: Remove pep8 exclusion for VMware in Kuryr https://review.openstack.org/361941 | 08:43 |
| *** devvesa has joined #openstack-kuryr | 08:46 | |
| *** salv-orlando has quit IRC | 08:49 | |
| *** yamamoto has joined #openstack-kuryr | 08:50 | |
| apuimedo | devvesa: did you see vikasc comment to your patch? | 08:52 |
| *** yamamoto has quit IRC | 08:56 | |
| vikasc | devvesa, config.py in kuryr-libnetwork is one level deeper than in kuryr-kubernetes because of "common". | 08:58 |
| vikasc | devvesa, thats why i said one .. | 08:58 |
| vikasc | devvesa, what do you think | 08:59 |
| vikasc | devvesa, base_dir should be site_packages or kuryr_libnetwork? | 09:00 |
| vikasc | i think kuryr_libnetwork | 09:00 |
| devvesa | apuimedo: vikasc: I've just answered | 09:00 |
| devvesa | I think the $pybasedir does not apply over the config.py module but over the root directory of the project | 09:01 |
| devvesa | I may be wrong, though | 09:01 |
| openstackgerrit | Jaume Devesa proposed openstack/kuryr-kubernetes: Introduce `kuryr-k8s` service https://review.openstack.org/359811 | 09:03 |
| vikasc | devvesa, Can you please confirm. I am also not sure. | 09:03 |
| apuimedo | devvesa: can you please check it, just to make sure we can disregard it? | 09:03 |
| devvesa | I can not say no now :D | 09:04 |
| devvesa | btw, apuimedo, what about python 3.5 on Red Hat? | 09:05 |
| apuimedo | devvesa: not good news yet. But I just got to RH two hours ago, I don't even have email account yet | 09:06 |
| apuimedo | :P | 09:06 |
| apuimedo | I'll try to update on today's meeting | 09:06 |
| apuimedo | limao: mind if I send a separate patch to add the missing oslo.i18n dep? | 09:06 |
| limao | sure, let me remove -1. thanks | 09:07 |
| devvesa | apuimedo: vikasc: "pybasedir = /home/devvesa/Code/kuryr-kubernetes/.tox/py35/lib/python3.5" | 09:08 |
| devvesa | with the two "../../". So I guess it is ok | 09:09 |
| vikasc | devvesa, thanks for confirming | 09:09 |
| devvesa | vikasc: thanks to you for being so meticulous on your reviews! | 09:10 |
| apuimedo | thanks | 09:11 |
| apuimedo | :-) | 09:11 |
| apuimedo | to all of you | 09:11 |
| vikasc | devvesa, :P | 09:11 |
| apuimedo | devvesa: I forgot to answer to your last comment, the reason why it does not work, is that in your current code you never get to the 'wait()' call :-) | 09:11 |
| openstackgerrit | Jaume Devesa proposed openstack/kuryr-kubernetes: Introduce asyncrhonous API requests https://review.openstack.org/360376 | 09:12 |
| devvesa | apuimedo: uhm... I would swear that I debug it | 09:13 |
| vikasc | devvesa, what are your plans for test cases | 09:14 |
| devvesa | vikasc: coming soon :) (maybe by the end of the week) | 09:15 |
| vikasc | devvesa, great :) | 09:15 |
| apuimedo | devvesa: I also run it through ipdb | 09:15 |
| apuimedo | :-) | 09:15 |
| devvesa | our current code downstream is based on mox, I want to refactor to mock and try to make it easier to develop new unit tests | 09:16 |
| devvesa | apuimedo: can you suggest me a change? | 09:16 |
| apuimedo | devvesa: that's doing god's work | 09:16 |
| apuimedo | (replacing mox) | 09:16 |
| devvesa | God or good? :) | 09:16 |
| devvesa | kind of divine justice work? | 09:17 |
| apuimedo | divine work | 09:17 |
| apuimedo | devvesa: once you replace the dummy start with one that just schedules with asyncio it will probably just work | 09:17 |
| *** limao has quit IRC | 09:17 | |
| *** salv-orlando has joined #openstack-kuryr | 09:20 | |
| *** limao has joined #openstack-kuryr | 09:21 | |
| devvesa | apuimedo: ok... I'll try on my last patch | 09:21 |
| devvesa | apuimedo: Who's the other +2 on kuryr_kubernetes? | 09:22 |
| apuimedo | devvesa: vikasc, irena, gal, taku, and banix | 09:24 |
| devvesa | irena hasn't | 09:26 |
| vikasc | apuimedo, devvesa i tried with lastest patch in chain day before yesterday. It was working | 09:27 |
| apuimedo | vikasc: so you can +2 ;-) | 09:27 |
| vikasc | apuimedo, cool | 09:28 |
| apuimedo | vikasc: https://review.openstack.org/#/c/361937/ too :-) | 09:29 |
| devvesa | vikasc: Cool! | 09:30 |
| vikasc | apuimedo, done! | 09:30 |
| apuimedo | thanks | 09:31 |
| vikasc | np | 09:31 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr: requirements: add missing oslo i18n dep https://review.openstack.org/361972 | 09:43 |
| limao | Hi apuimedo | 09:46 |
| apuimedo | limao: Hi! | 09:46 |
| apuimedo | I just sent the i18n patch | 09:46 |
| apuimedo | thanks a lot for pointing it out | 09:46 |
| limao | Is that patch will casue pep8 problem? maybe oslo.i18n should between oslo.concurrency and oslo.log | 09:46 |
| limao | I go through all the import in the code, it should be ok in requirement.txt, maybe we also need to clean up in test-requirement.txt | 09:48 |
| apuimedo | limao: that's a good point | 09:48 |
| apuimedo | I'll fix the order | 09:48 |
| limao | thanks, apuimedo | 09:48 |
| apuimedo | and we should definitely go through test-requirements.txt | 09:48 |
| apuimedo | :-) | 09:48 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr: requirements: add missing oslo i18n dep in order https://review.openstack.org/361972 | 09:50 |
| apuimedo | limao: done ^^ | 09:50 |
| limao | cool | 09:50 |
| *** janonymous has joined #openstack-kuryr | 09:51 | |
| apuimedo | :-) | 09:54 |
| *** limao has quit IRC | 09:55 | |
| *** yamamoto has joined #openstack-kuryr | 09:57 | |
| *** devvesa has quit IRC | 09:59 | |
| *** yamamoto has quit IRC | 10:03 | |
| *** devvesa has joined #openstack-kuryr | 10:19 | |
| *** garyloug has joined #openstack-kuryr | 10:27 | |
| openstackgerrit | vikas choudhary proposed openstack/kuryr: Nested-Containers: Add vlan management apis https://review.openstack.org/361993 | 10:27 |
| *** jerms has joined #openstack-kuryr | 10:52 | |
| jerms | vikasc: heya | 10:52 |
| *** lmdaly has joined #openstack-kuryr | 10:54 | |
| apuimedo | jerms: nice to see you here | 10:58 |
| *** yamamoto has joined #openstack-kuryr | 11:00 | |
| jerms | hey apuimedo :-) | 11:03 |
| *** lmdaly has quit IRC | 11:03 | |
| apuimedo | jerms: can I help you with anything? | 11:04 |
| jerms | apuimedo: hey -- not at the moment -- you're starting today? | 11:07 |
| apuimedo | jerms: yes, I do | 11:07 |
| jerms | ah cool. welcome! :-) | 11:07 |
| openstackgerrit | vikas choudhary proposed openstack/kuryr: Add 'deployment_type' configuration parameter https://review.openstack.org/362023 | 11:08 |
| apuimedo | thanks | 11:08 |
| jerms | apuimedo: i was just chatting with vikasc ... he sent me the bi-weekly call info | 11:08 |
| jerms | ill try to dial-in later | 11:08 |
| apuimedo | great | 11:09 |
| jerms | apuimedo: hey -- will you use the internal irc? | 11:12 |
| apuimedo | jerms: I will, once I remember the address :P | 11:12 |
| apuimedo | I must have it in my old weechat conf | 11:13 |
| openstackgerrit | vikas choudhary proposed openstack/kuryr: Nested-Containers: Add vlan management apis https://review.openstack.org/361993 | 11:14 |
| apuimedo | found it | 11:14 |
| openstackgerrit | vikas choudhary proposed openstack/kuryr: Nested-Containers: Add vlan management apis https://review.openstack.org/361993 | 11:17 |
| openstackgerrit | vikas choudhary proposed openstack/kuryr: Add 'deployment_type' configuration parameter https://review.openstack.org/362023 | 11:18 |
| *** yamamoto has quit IRC | 11:22 | |
| *** yamamoto has joined #openstack-kuryr | 11:27 | |
| *** janki has quit IRC | 11:28 | |
| *** yamamoto has quit IRC | 11:42 | |
| *** devvesa has quit IRC | 11:46 | |
| *** lmdaly has joined #openstack-kuryr | 11:51 | |
| *** limao has joined #openstack-kuryr | 11:58 | |
| *** devvesa has joined #openstack-kuryr | 12:02 | |
| *** limao_ has joined #openstack-kuryr | 12:03 | |
| *** limao has quit IRC | 12:06 | |
| *** salv-orl_ has joined #openstack-kuryr | 12:08 | |
| *** salv-orlando has quit IRC | 12:10 | |
| *** salv-orl_ has quit IRC | 12:16 | |
| openstackgerrit | Liping Mao proposed openstack/kuryr: Remove unused docker-py in test-requirements.txt https://review.openstack.org/362070 | 12:27 |
| *** salv-orlando has joined #openstack-kuryr | 12:29 | |
| *** yamamoto has joined #openstack-kuryr | 12:42 | |
| *** garyloug has quit IRC | 12:47 | |
| *** yamamoto has quit IRC | 12:49 | |
| openstackgerrit | Liping Mao proposed openstack/kuryr-libnetwork: Update requirements and test-requirements https://review.openstack.org/362094 | 12:53 |
| *** diga has quit IRC | 12:54 | |
| *** sdake has joined #openstack-kuryr | 13:03 | |
| *** tonanhngo has joined #openstack-kuryr | 13:16 | |
| *** sdake_ has joined #openstack-kuryr | 13:23 | |
| *** sdake has quit IRC | 13:25 | |
| *** huikang has joined #openstack-kuryr | 13:27 | |
| openstackgerrit | Dongcan Ye proposed openstack/kuryr: Duplicate utils code in kuryr-lib and kuryr-libnetwork https://review.openstack.org/362115 | 13:33 |
| *** garyloug has joined #openstack-kuryr | 13:34 | |
| apuimedo | devvesa: did you change much in the aio package? | 13:39 |
| devvesa | apuimedo: no, just adapt to async/await | 13:41 |
| *** banix has joined #openstack-kuryr | 13:42 | |
| openstackgerrit | Merged openstack/kuryr-kubernetes: Introduce `kuryr-k8s` service https://review.openstack.org/359811 | 13:43 |
| apuimedo | devvesa: very well | 13:44 |
| apuimedo | devvesa: congratulations on the skeleton patch getting merged | 13:44 |
| devvesa | \o/ | 13:44 |
| apuimedo | let's see if vikasc or banix can review the other patches | 13:44 |
| banix | will be getting back to reviews hopefully today | 13:45 |
| apuimedo | banix: thanks! | 13:46 |
| devvesa | apuimedo: apuimedo: did you see that I send you a patch from a project that does the same as the aio package? | 13:48 |
| devvesa | sorry, a link, not a patch | 13:48 |
| apuimedo | devvesa: yes. I knew about it before I started my own package | 13:48 |
| devvesa | oh | 13:48 |
| apuimedo | I would like to use it in the future | 13:48 |
| apuimedo | I have to see if it has packagers in distros | 13:48 |
| apuimedo | how they react to bug reports, etc | 13:48 |
| apuimedo | if the code is potable in case we need to fix stuff ourselves | 13:49 |
| apuimedo | but I'd much rather use aiohttp than my package | 13:49 |
| devvesa | I like this project too: https://github.com/python-hyper/hyper-h2 | 13:49 |
| devvesa | learned from this post: http://www.snarky.ca/network-protocols-sans-i-o | 13:50 |
| *** sdake_ has quit IRC | 13:50 | |
| *** sdake has joined #openstack-kuryr | 13:51 | |
| apuimedo | devvesa: I was studying it too :-) | 13:53 |
| devvesa | I like the idea of separate the protocol from I/O | 13:53 |
| apuimedo | I want to move to http2 or, if that is not possible, because h2 is not asyncio ready, at least use websockets | 13:53 |
| apuimedo | devvesa: yes, it's a good thing | 13:53 |
| apuimedo | but somebody will have to contribute the hyper-h2 support to aiohttp | 13:54 |
| apuimedo | there is an issue in aiohttp for it | 13:54 |
| apuimedo | but it didn't get much traction | 13:54 |
| devvesa | oh.. | 13:54 |
| apuimedo | devvesa: but it may not be that difficult to add it | 13:56 |
| apuimedo | either to our aio or to aiohttp | 13:56 |
| devvesa | I'll take a look and see if I can schedule time (depending on how effort it is) | 13:57 |
| devvesa | meanwhile we can live with the aiohttp package | 13:57 |
| devvesa | inside the repo | 13:57 |
| apuimedo | devvesa: I agree | 13:58 |
| openstackgerrit | Jaume Devesa proposed openstack/kuryr-kubernetes: Add asyncio eventloop. https://review.openstack.org/360629 | 13:59 |
| *** hongbin has joined #openstack-kuryr | 14:20 | |
| *** banix_ has joined #openstack-kuryr | 14:34 | |
| *** banix has quit IRC | 14:36 | |
| *** banix_ is now known as banix | 14:36 | |
| *** ivc_ has joined #openstack-kuryr | 14:52 | |
| *** huikang has quit IRC | 14:56 | |
| *** huikang has joined #openstack-kuryr | 14:56 | |
| devvesa | vikasc: I beg you modify your vote +1 to +2 here: https://review.openstack.org/#/c/360376/ | 15:04 |
| devvesa | :) | 15:04 |
| *** banix has quit IRC | 15:04 | |
| vikasc | devvesa, Done :) | 15:05 |
| *** banix has joined #openstack-kuryr | 15:05 | |
| devvesa | thanks | 15:05 |
| *** tonanhngo has quit IRC | 15:05 | |
| devvesa | apuimedo: I think that your idea about watcher to upload annotation on await it is much simpler than mine | 15:08 |
| devvesa | Let me come up with a patch that implements more or less this and you can express your ideas there | 15:08 |
| openstackgerrit | Merged openstack/kuryr-kubernetes: Introduce asyncrhonous API requests https://review.openstack.org/360376 | 15:09 |
| *** tonanhngo has joined #openstack-kuryr | 15:10 | |
| *** huikang has quit IRC | 15:12 | |
| *** huikang has joined #openstack-kuryr | 15:12 | |
| apuimedo | very well | 15:15 |
| apuimedo | thanks devvesa! | 15:15 |
| *** huikang has quit IRC | 15:17 | |
| *** sdake_ has joined #openstack-kuryr | 15:18 | |
| ivc_ | hello guys, Ilya Chukhnakov from Mirantis here. We've been doing some research on Neutron-K8s integration and Kuryr specifically and came up with this report (still a draft): https://docs.google.com/document/d/1AXyV4BkT1nqGxfVzW3USnCxKCEnP7hdz_550yjTEKPI | 15:18 |
| *** huikang has joined #openstack-kuryr | 15:19 | |
| apuimedo | ivc_: thanks for the heads up | 15:19 |
| ivc_ | we are also working on our PoC for Kuryr. originally it was based on Midokura's PoC, but right now we are working on Py2/eventlet version that uses LBaaSv2 | 15:19 |
| *** sdake has quit IRC | 15:20 | |
| apuimedo | ivc_: interesting | 15:20 |
| apuimedo | is it open? | 15:20 |
| ivc_ | it will be. i've just started working on a new iteration | 15:20 |
| devvesa | ivc_ do you think it is possible any kind of collaboration? | 15:22 |
| ivc_ | right now we got a working Midokura's prototype with OVS/Devstack and LBaasV2, but we decided to try a slightly different approach | 15:22 |
| devvesa | we are moving the midonet/kuryr https://github.com/midonet/kuryr/tree/k8s | 15:22 |
| ivc_ | ofc. right now i just want to share our findings and see what do you think about it :) | 15:22 |
| *** tonanhngo has quit IRC | 15:23 | |
| devvesa | to upstream in more 'collaborative' way | 15:23 |
| ivc_ | with midonet prototype there are several synchronisation issues (i think i've covered them in our report) | 15:23 |
| devvesa | ivc_ I'll take a look this afternoon. We can chat tomorrow about it | 15:23 |
| devvesa | ivc_ yes. It is not ready for production. For sure | 15:24 |
| ivc_ | right now my idea is to move port allocation to cni driver and leave raven with just service/endpoint watches | 15:24 |
| devvesa | ivc_ you mean call neutron to create ports from cni driver? | 15:24 |
| *** sdake has joined #openstack-kuryr | 15:24 | |
| ivc_ | yeah so basically cni driver will create port, plug veth, and wait for neutron port to become active | 15:25 |
| apuimedo | ivc_: I have to commend you on the document | 15:25 |
| apuimedo | it is really detailed and with very good and accurate diagrams | 15:25 |
| ivc_ | thnx :) | 15:26 |
| *** sdake_ has quit IRC | 15:26 | |
| devvesa | ivc_ : It can make sense on some cases but Neutron API use to live in a management Network. Many deployments don't have the management network attached to their worker nodes | 15:26 |
| devvesa | I even seem to recall that Fuel does not do it by default :) | 15:27 |
| ivc_ | devvesa, yeah thats right... | 15:27 |
| ivc_ | but still the problem is that cni driver has to wait for neutron to report port as active before passing it to k8s | 15:28 |
| ivc_ | otherwise k8s container will start while the network is not yet configured properly | 15:28 |
| apuimedo | ivc_: tehre's several optimizations we considered | 15:29 |
| apuimedo | ivc_: in kuryr-libnetwork we already have the waiting builtin | 15:29 |
| apuimedo | so that it doens't start before it is ready | 15:29 |
| devvesa | I would love to have Mirantis workforce to help us solve this issues :P | 15:29 |
| apuimedo | it should be easy to add the same fix the the cni/raven duo | 15:29 |
| *** devvesa has quit IRC | 15:29 | |
| *** devvesa has joined #openstack-kuryr | 15:30 | |
| ivc_ | devvesa, it's true that management net could be a problem, but the k8s api server is not much different from neutron server and we allow cni to talk to it | 15:31 |
| devvesa | ivc_ You are absolutely right on this | 15:32 |
| ivc_ | also i see you are going for py3/asyncio version. have you considered using requests with eventlet? i think it should work fine. at least it seems to work for ovn-kubernetes by shettyg | 15:32 |
| apuimedo | ivc_: we considered it. And it should be possible to make it work | 15:33 |
| ivc_ | https://github.com/shettyg/ovn-kubernetes | 15:33 |
| apuimedo | but starting a new project we thought it worthwhile to use the new and improved facilities python3 gives us | 15:34 |
| apuimedo | ivc_: is there some limitation on using python3 in Mirantis distro? | 15:34 |
| ivc_ | more specifically https://github.com/shettyg/ovn-kubernetes/blob/master/ovn_k8s/common/kubernetes.py#L23-L29 | 15:34 |
| apuimedo | ivc_: it is an option | 15:36 |
| ivc_ | apuimedo, we haven't discussed it yet, but to me it seemed a bit "inconsistent" (no offence :) ) with other parts of OSt that use py2/eventlet | 15:36 |
| apuimedo | ivc_: I understand perfectly | 15:36 |
| apuimedo | there was talk of python3 support accross OSt and we though a watcher was the perfect use case for spearheading the movement | 15:37 |
| apuimedo | but I am open (if a bit reluctant) to change if the kuryr community would prefer python2 with eventlet | 15:38 |
| ivc_ | apuimedo, also the yield self.delegate semantic does seem a bit weird | 15:39 |
| ivc_ | but thats just imho :) | 15:40 |
| apuimedo | ivc_: we are getting rid of the delegates | 15:40 |
| apuimedo | that was just a weird implementation | 15:40 |
| apuimedo | ivc_: and I agree that it was weird and ugly | 15:41 |
| ivc_ | imho py3 async is great but only starting with py3.5 syntax | 15:42 |
| apuimedo | ivc_: which is what devvesa is starting to use in the upstreaming | 15:42 |
| apuimedo | we agree that 3.4 syntax is less than ideal | 15:43 |
| apuimedo | ubuntu 16.04 already has 3.5 | 15:43 |
| apuimedo | EL7 is the problem | 15:43 |
| ivc_ | el7 = rhel/centos? | 15:43 |
| apuimedo | ivc_: yes. both | 15:45 |
| apuimedo | ivc_: what is the latest mirantis distro based on? | 15:45 |
| ivc_ | ubuntu 14.04 or centos i think | 15:46 |
| apuimedo | ok | 15:49 |
| ivc_ | ok i've checked with folks, right now its only ubuntu 14.04 for the nodes, but it is possible to make custom deployment with preinstalled os with some tweaks. | 15:50 |
| ivc_ | so, basically its 14.04 out of the box | 15:51 |
| *** limao_ has quit IRC | 15:51 | |
| apuimedo | ok | 15:52 |
| apuimedo | ivc_: well, the nodes don't need py3, since the cni part is py2 compatible | 15:52 |
| apuimedo | so the py35 requirement is only for the watchers | 15:52 |
| ivc_ | which can be deployed as pod as part of k8s right? | 15:53 |
| apuimedo | ivc_: that's exactly right ;-) | 15:53 |
| apuimedo | in fact I always run it as a container | 15:53 |
| apuimedo | and that's how I imagine it should always be | 15:53 |
| ivc_ | yeah that makes sense | 15:55 |
| apuimedo | gotta run | 15:55 |
| ivc_ | but right now raven is limited to a single instance per k8s, right? | 15:55 |
| apuimedo | I'll be back later in the night | 15:55 |
| apuimedo | ivc_: yes. We have already the ha design started | 15:55 |
| apuimedo | based on using etcd for leader election in active passive | 15:55 |
| *** apuimedo is now known as apuimedo|away | 15:56 | |
| ivc_ | i need to think about it. but if we let cni driver initiate the port creation (maybe not directly but by calling raven) then all the sync issues are handled by k8s for us | 15:58 |
| apuimedo|away | I'll think about it too | 15:59 |
| *** limao has joined #openstack-kuryr | 16:00 | |
| *** shashank_hegde has joined #openstack-kuryr | 16:03 | |
| *** huikang has quit IRC | 16:16 | |
| *** huikang has joined #openstack-kuryr | 16:17 | |
| *** shashank_hegde has quit IRC | 16:19 | |
| *** sdake has quit IRC | 16:19 | |
| *** huikang has quit IRC | 16:21 | |
| *** huikang has joined #openstack-kuryr | 16:24 | |
| *** devvesa has left #openstack-kuryr | 16:26 | |
| *** limao has quit IRC | 16:26 | |
| *** salv-orlando has quit IRC | 16:27 | |
| *** salv-orlando has joined #openstack-kuryr | 16:27 | |
| *** huikang has quit IRC | 16:28 | |
| *** sdake has joined #openstack-kuryr | 16:30 | |
| *** salv-orlando has quit IRC | 16:30 | |
| *** lmdaly has quit IRC | 16:52 | |
| *** shashank_hegde has joined #openstack-kuryr | 16:57 | |
| *** shashank_hegde has quit IRC | 17:04 | |
| *** huikang has joined #openstack-kuryr | 17:17 | |
| *** huikang has quit IRC | 17:23 | |
| *** garyloug has quit IRC | 17:27 | |
| *** shashank_hegde has joined #openstack-kuryr | 17:34 | |
| *** pablochacin has quit IRC | 17:35 | |
| *** huikang has joined #openstack-kuryr | 18:11 | |
| *** huikang has quit IRC | 18:16 | |
| *** salv-orlando has joined #openstack-kuryr | 18:34 | |
| *** salv-orlando has quit IRC | 18:42 | |
| *** huikang has joined #openstack-kuryr | 19:12 | |
| *** huikang has quit IRC | 19:17 | |
| *** huikang has joined #openstack-kuryr | 19:34 | |
| *** salv-orlando has joined #openstack-kuryr | 19:42 | |
| *** salv-orlando has quit IRC | 19:44 | |
| *** sdake has quit IRC | 19:48 | |
| *** huikang has quit IRC | 19:54 | |
| *** huikang has joined #openstack-kuryr | 20:08 | |
| *** huikang_ has joined #openstack-kuryr | 20:22 | |
| *** huikang has quit IRC | 20:26 | |
| *** salv-orlando has joined #openstack-kuryr | 20:51 | |
| *** sdake has joined #openstack-kuryr | 20:52 | |
| *** salv-orlando has quit IRC | 20:59 | |
| *** huikang_ has quit IRC | 21:11 | |
| *** huikang has joined #openstack-kuryr | 21:12 | |
| *** huikang_ has joined #openstack-kuryr | 21:13 | |
| *** salv-orlando has joined #openstack-kuryr | 21:16 | |
| *** huikang has quit IRC | 21:16 | |
| *** huikang_ has quit IRC | 21:17 | |
| *** huikang has joined #openstack-kuryr | 21:25 | |
| *** huikang has quit IRC | 21:30 | |
| *** sdake has quit IRC | 21:33 | |
| *** banix has quit IRC | 21:58 | |
| *** tonanhngo has joined #openstack-kuryr | 22:06 | |
| *** huikang has joined #openstack-kuryr | 22:33 | |
| *** huikang has quit IRC | 22:38 | |
| *** huikang has joined #openstack-kuryr | 22:39 | |
| *** huikang has quit IRC | 22:43 | |
| *** tonanhngo_ has joined #openstack-kuryr | 22:57 | |
| *** huikang has joined #openstack-kuryr | 22:59 | |
| *** tonanhngo has quit IRC | 23:00 | |
| *** hongbin has quit IRC | 23:03 | |
| *** huikang has quit IRC | 23:11 | |
| *** huikang has joined #openstack-kuryr | 23:12 | |
| *** huikang has quit IRC | 23:16 | |
| *** salv-orlando has quit IRC | 23:25 | |
| *** yamamoto has joined #openstack-kuryr | 23:30 | |
| *** huikang has joined #openstack-kuryr | 23:41 | |
| *** huikang has quit IRC | 23:44 | |
| *** huikang has joined #openstack-kuryr | 23:45 | |
| *** huikang has quit IRC | 23:49 | |
| *** huikang has joined #openstack-kuryr | 23:51 | |
| *** huikang has quit IRC | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!