| *** gouthamr has joined #openstack-kuryr | 00:01 | |
| *** salv-orlando has joined #openstack-kuryr | 00:15 | |
| *** salv-orlando has quit IRC | 00:20 | |
| *** vikasc has quit IRC | 00:32 | |
| *** jgriffith has quit IRC | 01:00 | |
| *** vikasc has joined #openstack-kuryr | 01:02 | |
| *** salv-orlando has joined #openstack-kuryr | 01:16 | |
| *** jgriffith has joined #openstack-kuryr | 01:21 | |
| *** salv-orlando has quit IRC | 01:21 | |
| *** yamamoto has joined #openstack-kuryr | 02:16 | |
| *** janki has joined #openstack-kuryr | 03:44 | |
| *** hongbin has joined #openstack-kuryr | 03:45 | |
| *** hongbin has quit IRC | 04:21 | |
| *** aojea has joined #openstack-kuryr | 04:26 | |
| *** janonymous has joined #openstack-kuryr | 04:28 | |
| *** salv-orlando has joined #openstack-kuryr | 04:29 | |
| *** aojea has quit IRC | 04:30 | |
| *** yamamoto has quit IRC | 05:02 | |
| *** ltomasbo|away is now known as ltomasbo | 05:30 | |
| *** yamamoto has joined #openstack-kuryr | 05:45 | |
| *** xiangxinyong has quit IRC | 06:07 | |
| *** xiangxinyong has joined #openstack-kuryr | 06:08 | |
| *** kzaitsev_ws has joined #openstack-kuryr | 06:23 | |
| *** dimak_ has joined #openstack-kuryr | 06:25 | |
| kzaitsev_ws | apuimedo: ivc_: irenab: I've been thinking about single entity/single handler design and I think I understand where my confusion comes from. | 06:27 |
|---|---|---|
| kzaitsev_ws | the handlers are not named after the entities | 06:27 |
| kzaitsev_ws | this makes me feel like I can have multiple handlers per entity | 06:27 |
| kzaitsev_ws | if they were PodHandler, EndpointHandler, etc. it would be a lot clearer, that a PodHandler is designed to be the sole entry-point for all things Pod-related | 06:28 |
| kzaitsev_ws | instead it's a vif-handler and it feels like I can have multi-vif handler or something optional there. | 06:29 |
| kzaitsev_ws | anyway. thanks for reviews, would go through the comments soon | 06:30 |
| *** limao has joined #openstack-kuryr | 06:36 | |
| kzaitsev_ws | btw next week is p-2 I don't think it is anything kuryr team needs to do for it. just a heads up =) | 06:43 |
| irenab | kzaitsev_ws, based on past discussions with ivc_ , Handler names where based on the neutron resulting output | 06:47 |
| kzaitsev_ws | irenab: well, the naming does not hint that handlers actually map to k8s instances | 06:50 |
| *** limao_ has joined #openstack-kuryr | 06:50 | |
| kzaitsev_ws | and that was the cause of my confusion =) | 06:51 |
| *** limao has quit IRC | 06:53 | |
| kzaitsev_ws | I don't suggest that it should be renamed. But I hope to raise this question if someone else trips over those names in future =) | 06:58 |
| irenab | kzaitsev_ws, it was also my argument with ivc_ ... the mapping to k8s entity is inside the Handler | 06:59 |
| *** salv-orl_ has joined #openstack-kuryr | 07:04 | |
| *** gouthamr has quit IRC | 07:05 | |
| *** salv-orlando has quit IRC | 07:07 | |
| *** aojea has joined #openstack-kuryr | 07:20 | |
| *** aojea has quit IRC | 07:20 | |
| *** aojea has joined #openstack-kuryr | 07:20 | |
| *** limao_ has quit IRC | 07:37 | |
| *** egonzalez has joined #openstack-kuryr | 07:43 | |
| *** apuimedo has quit IRC | 07:54 | |
| *** apuimedo has joined #openstack-kuryr | 07:55 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: Add configuration for kubelet probes https://review.openstack.org/468100 | 08:03 |
| openstackgerrit | Marco Chiappero proposed openstack/kuryr-kubernetes master: Refactor the class hierarchy of controller drivers https://review.openstack.org/463519 | 09:46 |
| ivc_ | irenab, kzaitsev_ws, the idea was not to have a single handler per k8s entity (quite contrary) as it is possible we might need some other sort of handler for pod events | 10:44 |
| ivc_ | irenab, kzaitsev_ws, but in case of sr-iov it fits the vifhandler purpose since the outcome of sr-iov handler is the same as the current vif handler | 10:45 |
| ivc_ | irenab, kzaitsev_ws, so for example there could be a e.g. a VolumeHandler that also observes pod events, hence the handlers were named after the produced entity instead of the observed entity | 10:46 |
| ivc_ | irenab, kzaitsev_ws, and it is also possible to have a multi-k8s-entity handlers (e.g. for logging/monitoring/profiling a single handler can track pods and services), tho there is no use case for that yet, i tried to keep the design as flexible as possible :) | 10:50 |
| kzaitsev_ws | ivc_: the thing is. in a real-world scenario sriov vifs would be complementary to regular vif. So it's a bit confusing when you say that the outcome is the same | 10:53 |
| kzaitsev_ws | regular always returns a single vif. sriov might return 0 to N vifs | 10:53 |
| ivc_ | kzaitsev_ws i'm suggesting that sriov vifs should not be differentiated from regular vifs | 10:54 |
| ivc_ | kzaitsev_ws also note there are some plans on k8s side to add multi-network support | 10:55 |
| *** yamamoto has quit IRC | 10:55 | |
| *** ashishb_ has quit IRC | 10:55 | |
| ivc_ | kzaitsev_ws we discussed the need to support multi-vif pods earlier and sriov is just a subset of the multi-vif problem | 10:57 |
| kzaitsev_ws | So you basically suggest to store all the vifs in a list, and I think it should be possible to store physnet information somewhere inside network info object | 10:59 |
| ivc_ | kzaitsev_ws aye | 11:01 |
| ivc_ | kzaitsev_ws i'm on a meeting atm, we can discuss it in about 60-90 minutes | 11:01 |
| kzaitsev_ws | well, yep. it feels like it should be possible to do that. The only real problem I see here is that the handler code would become larger and more complex, but as long as everyone's ok with this design direction, I'm ok with it too | 11:02 |
| kzaitsev_ws | ivc_: sure, I'll ping you if I have more questions ) | 11:03 |
| kzaitsev_ws | just a bit of reasoning from my side | 11:03 |
| kzaitsev_ws | I initially went with a separate handler/separate cni approach because I was not sure if I would be able to upstream the whole thing and that you folks would want to accept it. | 11:05 |
| kzaitsev_ws | So I went with the design, that minimises changes to existing handlers/cni. | 11:06 |
| kzaitsev_ws | I have no strong attachment to it, as long as the final implementation would satisfy my usecases =) | 11:06 |
| *** aojea has quit IRC | 11:08 | |
| *** circ-user-TeTGj has joined #openstack-kuryr | 11:13 | |
| irenab | kzaitsev_ws, the way you started is the perfect way to PoC the feature, no changes to the current code and you have the needed functionality working. But as the final desgn I am aligned with ivc_ about SR-IOV and generic port being very similar, so should be served by the same Handler | 11:22 |
| irenab | Regarding multi-vif, I am not sure if its just adding the list or have separate Handler. Maybe this will result in different Vif Driver | 11:24 |
| ivc_ | irenab kzaitsev_ws that being said we can probably use the sr-iov handler as a base for experimenting with multi-vif handler | 11:24 |
| irenab | ivc_, +1 | 11:24 |
| ivc_ | irenab i expect that instead of calling a single vif driver the handler will call all configured drivers (and stevedore supports such 'broadcasts' too) | 11:28 |
| irenab | ivc_, this is what I meant by different Vif drivers. Seems we are aligned | 11:29 |
| ivc_ | irenab as always :) | 11:30 |
| kzaitsev_ws | irenab: ivc_: thanks, I'll work with the spec to reflect these ideas | 11:30 |
| ivc_ | kzaitsev_ws i think it would be fine to mention both your current approach and the multi-vif approach in the spec and state that for the initial (experimental) implementation you will follow the current approach of separate handler | 11:31 |
| kzaitsev_ws | at least as alternative, yes. | 11:34 |
| ivc_ | kzaitsev_ws i.e. i'm ok with having the current proposed poc implementation in tree if we will update it (e.g. merge with the vifhandler) later, unless others (irenab, apuimedo) have any objections | 11:34 |
| irenab | I am ok with this once spec states phase 1 (current) and phase 2 (single Handler) approach | 11:37 |
| kzaitsev_ws | oh, that's an interesting path forward indeed. It may be a good idea if multi-vif design discussion would start to take too much time. | 11:37 |
| irenab | but if the diff is not that big, prefer going directly to phase 2 | 11:38 |
| kzaitsev_ws | irenab: yeah. So far I feel optimistic about doing it right in the second attempt =) | 11:39 |
| *** salv-orl_ has quit IRC | 11:40 | |
| *** yamamoto has joined #openstack-kuryr | 11:44 | |
| *** yamamoto_ has joined #openstack-kuryr | 11:44 | |
| *** yamamoto has quit IRC | 11:48 | |
| *** janki has quit IRC | 12:05 | |
| irenab | apuimedo, ping | 12:12 |
| irenab | ltomasbo, ping | 12:18 |
| *** salv-orlando has joined #openstack-kuryr | 12:20 | |
| *** pcaruana has joined #openstack-kuryr | 12:20 | |
| ltomasbo | irenab, pong | 12:30 |
| irenab | ltomasbo, regarding the devstack-heat patch. I suceeded to deploy the nova server, but its seems to be no devstack on it. What result did you have? | 12:31 |
| ltomasbo | same here | 12:32 |
| ltomasbo | I think it just prepares everything to just do the .stack.sh | 12:32 |
| irenab | I also checked the port, and it is not the trunk one. I am not sure if it is supposed to be | 12:33 |
| irenab | so basically it is an alternative to vagrant | 12:34 |
| ltomasbo | irenab, yes, I think it is just to deploy a VM where you install your devstack with the desired patch | 12:35 |
| ltomasbo | so, no trunk ports or the like | 12:35 |
| ltomasbo | I believe it is mainly to make it easy to CI different patches | 12:36 |
| apuimedo | irenab: pong | 12:36 |
| irenab | for some reason I was expecting devstack to be running inside the VM. Will leave comment at the patch README | 12:36 |
| apuimedo | irenab: ltomasbo: it should be stacking it all | 12:36 |
| apuimedo | it takes about 30minutes | 12:36 |
| apuimedo | in my environment | 12:36 |
| apuimedo | but when it is done, it creates an empty file called "ready" in /opt/stack | 12:37 |
| irenab | apuimedo, I ssh-ed into the VM and only see /opt/stack | 12:37 |
| *** yamamoto_ has quit IRC | 12:37 | |
| irenab | any way to see the progress? | 12:37 |
| apuimedo | irenab: yes | 12:37 |
| apuimedo | journalctl -f | 12:37 |
| *** yamamoto has joined #openstack-kuryr | 12:37 | |
| apuimedo | you see everything the system is doing | 12:37 |
| apuimedo | cloud-init is the entries you should check for | 12:37 |
| ltomasbo | umm, I believe it did not stack for me either | 12:38 |
| apuimedo | (but strangely enough 'journalctl -u cloud-init' does not filter well | 12:38 |
| apuimedo | ltomasbo: please, check your journal for errors | 12:38 |
| ltomasbo | but maybe I miss something, but tried just the .stack.sh and worked | 12:38 |
| irenab | no journsl files were found .. | 12:39 |
| ltomasbo | I'll give it another try later then | 12:39 |
| apuimedo | irenab: sudo journalctl -b | 12:40 |
| apuimedo | vikasc: ping | 12:45 |
| vikasc | apuimedo, hi | 12:45 |
| apuimedo | vikasc: could you pleaes review and merge if appropriate https://review.openstack.org/#/c/468100/ ? | 12:46 |
| vikasc | sure apuimedo | 12:46 |
| apuimedo | thanks vikasc! | 12:47 |
| vikasc | yw! | 12:47 |
| apuimedo | same with https://review.openstack.org/#/c/436875/ vikasc | 12:47 |
| apuimedo | I want to start getting things in before p-2 | 12:48 |
| vikasc | sure, in mid of something, will do at night for sure. | 12:49 |
| vikasc | apuimedo, | 12:49 |
| apuimedo | thanks vikasc | 12:50 |
| *** aojea has joined #openstack-kuryr | 12:52 | |
| *** janki has joined #openstack-kuryr | 12:52 | |
| *** janki has quit IRC | 12:54 | |
| vikasc | yw! | 12:57 |
| *** salv-orl_ has joined #openstack-kuryr | 13:04 | |
| *** salv-orlando has quit IRC | 13:07 | |
| apuimedo | ltomasbo: where did you try it? | 13:08 |
| apuimedo | Is it also on a devstack installed openstack? | 13:08 |
| apuimedo | like irenab ? | 13:08 |
| ltomasbo | buff, plenty of tests lately | 13:08 |
| ltomasbo | no, it was an RDO installed openstack (I believe) | 13:09 |
| apuimedo | interesting | 13:09 |
| apuimedo | In irena's case the issue is that sudo fails | 13:09 |
| apuimedo | "sudo: sorry, you must have a tty to run sudo" | 13:09 |
| ltomasbo | I can try to test it later, but I need to complete some other tsted/demo configuration before | 13:11 |
| apuimedo | sure | 13:11 |
| openstackgerrit | Merged openstack/kuryr-libnetwork master: Switch to devstack for installing etcd https://review.openstack.org/467718 | 13:13 |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: Adding support for vif pool driver https://review.openstack.org/436875 | 13:27 |
| apuimedo | irenab: I fixed the issue with your setup | 13:28 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: contrib: Add devstack-heat https://review.openstack.org/466291 | 13:36 |
| dmellado | apuimedo: that sounds like a regular issue | 13:40 |
| apuimedo | ltomasbo: irenab: This should make it work for both of you ^^ | 13:40 |
| * dmellado recalls smthing abotu !requiretty | 13:40 | |
| dmellado | about | 13:40 |
| dmellado | for the sudo thing | 13:40 |
| dmellado | did you check that out? | 13:40 |
| apuimedo | dmellado: it doesn't happen in tonicloud though | 13:40 |
| apuimedo | dmellado: what should I check? | 13:41 |
| dmellado | sed -i -e 's/Defaults requiretty.*/ #Defaults requiretty/g' /etc/sudoers | 13:41 |
| dmellado | xD | 13:41 |
| dmellado | apuimedo: btw, before I forget, irenab I just realized that there are some days off in Israel this week | 13:43 |
| dmellado | I originally planned to do a brainstorming meeting this week re: functional testing and kuryr-k8s tempest plugin | 13:43 |
| dmellado | but I would've liked for you to attend | 13:43 |
| dmellado | maybe we can shift it to next week if it'd make it easier for you to be there | 13:44 |
| apuimedo | dmellado: make it wed next week maybe then | 13:44 |
| apuimedo | tuesday I'll be a zombie | 13:44 |
| dmellado | heh, just like always then? :D | 13:45 |
| dmellado | apuimedo: when are you coming back? | 13:45 |
| apuimedo | I fly tuesday at 5 am | 13:56 |
| apuimedo | so I suppose I'll be home at 11am | 13:56 |
| *** jgriffith has quit IRC | 14:00 | |
| *** jgriffith has joined #openstack-kuryr | 14:01 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-libnetwork master: Updated from global requirements https://review.openstack.org/461376 | 14:06 |
| *** jgriffith_ has joined #openstack-kuryr | 14:12 | |
| openstackgerrit | Marco Chiappero proposed openstack/kuryr-kubernetes master: Add MACVLAN based interfaces for nested containers https://review.openstack.org/440669 | 14:28 |
| *** jgriffith is now known as Guest55399 | 14:32 | |
| *** jgriffith_ is now known as jgriffith | 14:32 | |
| *** Guest55399 has joined #openstack-kuryr | 14:32 | |
| *** hongbin has joined #openstack-kuryr | 14:37 | |
| *** salv-orl_ has quit IRC | 14:54 | |
| *** salv-orlando has joined #openstack-kuryr | 14:55 | |
| *** salv-orlando has quit IRC | 14:59 | |
| *** dell has joined #openstack-kuryr | 15:04 | |
| *** salv-orlando has joined #openstack-kuryr | 15:16 | |
| *** dell is now known as zengchen | 15:21 | |
| *** gouthamr has joined #openstack-kuryr | 15:34 | |
| *** ltomasbo is now known as ltomasbo|away | 15:56 | |
| *** aojea has quit IRC | 16:01 | |
| *** aojea has joined #openstack-kuryr | 16:02 | |
| *** aojea has quit IRC | 16:06 | |
| *** dimak_ has quit IRC | 16:07 | |
| *** pcaruana has quit IRC | 16:17 | |
| *** egonzalez has quit IRC | 16:21 | |
| *** zengchen has quit IRC | 16:39 | |
| *** pcaruana has joined #openstack-kuryr | 16:45 | |
| *** gouthamr has quit IRC | 16:58 | |
| *** yamamoto has quit IRC | 17:06 | |
| *** xiangxinyong has quit IRC | 17:12 | |
| *** xiangxinyong has joined #openstack-kuryr | 17:12 | |
| *** aojea has joined #openstack-kuryr | 17:14 | |
| *** gouthamr has joined #openstack-kuryr | 17:15 | |
| *** salv-orlando has quit IRC | 17:18 | |
| *** aojea has quit IRC | 17:18 | |
| *** egonzalez has joined #openstack-kuryr | 18:04 | |
| *** yamamoto has joined #openstack-kuryr | 18:06 | |
| *** yamamoto has quit IRC | 18:19 | |
| *** aojea has joined #openstack-kuryr | 18:52 | |
| *** salv-orlando has joined #openstack-kuryr | 20:13 | |
| *** pcaruana has quit IRC | 20:24 | |
| *** aojea has quit IRC | 20:58 | |
| *** aojea has joined #openstack-kuryr | 20:59 | |
| *** aojea has quit IRC | 21:03 | |
| *** circ-user-TeTGj has quit IRC | 21:33 | |
| *** gouthamr has quit IRC | 21:37 | |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: Create infra for K8s API accessing pods https://review.openstack.org/467241 | 21:49 |
| openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: devstack: Retrieve column directly https://review.openstack.org/468962 | 21:49 |
| *** salv-orlando has quit IRC | 22:12 | |
| *** gouthamr has joined #openstack-kuryr | 22:42 | |
| *** egonzalez has quit IRC | 22:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!