*** aojea has quit IRC | 00:20 | |
*** aojea has joined #openstack-kuryr | 00:21 | |
*** limao has joined #openstack-kuryr | 00:40 | |
*** yedongcan has joined #openstack-kuryr | 01:48 | |
*** hongbin has joined #openstack-kuryr | 02:03 | |
*** dougbtv__ has quit IRC | 02:22 | |
*** aojea has quit IRC | 02:25 | |
*** aojea has joined #openstack-kuryr | 02:26 | |
*** vikasc has joined #openstack-kuryr | 03:12 | |
*** hongbin has quit IRC | 03:55 | |
*** tonanhngo has joined #openstack-kuryr | 04:15 | |
*** salv-orlando has joined #openstack-kuryr | 04:16 | |
*** salv-orlando has quit IRC | 04:21 | |
*** gouthamr has quit IRC | 04:24 | |
*** salv-orlando has joined #openstack-kuryr | 04:28 | |
*** tonanhngo has quit IRC | 04:37 | |
*** tonanhngo has joined #openstack-kuryr | 04:41 | |
*** limao has quit IRC | 04:44 | |
*** tonanhngo_ has joined #openstack-kuryr | 04:56 | |
*** limao has joined #openstack-kuryr | 04:58 | |
*** tonanhngo has quit IRC | 04:59 | |
*** salv-orl_ has joined #openstack-kuryr | 05:05 | |
*** salv-orlando has quit IRC | 05:08 | |
*** tonanhngo_ has quit IRC | 05:17 | |
*** tonanhngo has joined #openstack-kuryr | 05:18 | |
*** tonanhngo has quit IRC | 05:23 | |
*** janki has joined #openstack-kuryr | 05:26 | |
*** salv-orl_ has quit IRC | 06:02 | |
*** ltomasbo|away is now known as ltomasbo | 06:13 | |
*** tonanhngo has joined #openstack-kuryr | 06:28 | |
*** dims has quit IRC | 06:44 | |
*** dims has joined #openstack-kuryr | 06:46 | |
*** rwallner has joined #openstack-kuryr | 06:51 | |
*** rwallner has quit IRC | 06:55 | |
*** pcaruana has joined #openstack-kuryr | 06:58 | |
*** roeyc has joined #openstack-kuryr | 07:06 | |
*** salv-orlando has joined #openstack-kuryr | 07:12 | |
irenab | ltomasbo, morning | 07:15 |
---|---|---|
ltomasbo | irenab, good morning! | 07:15 |
irenab | I have a question for running overcloud | 07:15 |
irenab | somehow it fails to stack | 07:15 |
irenab | in plugin.sh prepare_kuberentes_files | 07:16 |
ltomasbo | umm, that is one of the 'recent' additions, right? | 07:17 |
ltomasbo | not sure I've tried that yet | 07:18 |
dmellado | irenab: with which patch are you using it | 07:18 |
ltomasbo | what error do you get? | 07:18 |
dmellado | it failed to stack for me too, with the PS5 | 07:18 |
dmellado | I'll retry, as it could've been due to rdo cloud misbehaving yesterday | 07:18 |
dmellado | dns issues | 07:18 |
dmellado | irenab: if you experience such issues, please do change dns on your image | 07:19 |
ltomasbo | irenab, did you set up this: KURYR_CONFIGURE_BAREMETAL_KUBELET_IFACE=False | 07:19 |
ltomasbo | not sure if it is related, but for the overcloud some of the steps need to be skipped | 07:19 |
dmellado | ltomasbo: interesting, is that on the README? | 07:21 |
ltomasbo | dmellado, yes, it is in the local.conf example for the overcloud | 07:22 |
dmellado | ltomasbo: btw, how's life back in Madrid | 07:22 |
dmellado | did you melt? | 07:22 |
ltomasbo | waaarrrmmmm | 07:22 |
ltomasbo | almost! | 07:22 |
* dmellado goes back this afternoon and is scared | 07:22 | |
ltomasbo | it was hotter here when I arrived yesterday midnight, than the top temperature I got in Stockholm during the 3 days | 07:23 |
ltomasbo | yep, it will be +40 on Saturday... | 07:23 |
dmellado | ajjjj | 07:23 |
*** tonanhngo has quit IRC | 07:23 | |
apuimedo | irenab: I think I fixed that problem in the last patchset | 07:28 |
apuimedo | try to stack with --override-keystone | 07:28 |
dmellado | apuimedo: I'm going to -1 as there are some options in the local.conf that are not explained | 07:29 |
apuimedo | y=ー( ゚д゚)・∵. | 07:29 |
apuimedo | you're killing me | 07:29 |
apuimedo | xD | 07:29 |
*** neiljerram has joined #openstack-kuryr | 07:30 | |
dmellado | xD | 07:31 |
*** ltomasbo is now known as ltomasbo|away | 07:32 | |
irenab | apuimedo, I already have VM deployed. Do you have local.conf of the overcloud to share? | 07:33 |
dmellado | irenab: shouldn't that be in the lines of https://review.openstack.org/#/c/472763/7/devstack/local.conf.pod-in-vm.overcloud.sample | 07:34 |
dmellado | ? | 07:34 |
apuimedo | irenab: well... it is usually better to have ./devstack-overcloud-heat deploy the VM for you | 07:36 |
apuimedo | otherwise you have more manual steps to take | 07:36 |
irenab | apuimedo, I had to tweek DF to enable communication to VM, since we have some issue to fix with DNAT. | 07:37 |
irenab | in parallel I am checking your patch on reference ovs | 07:37 |
*** egonzalez has joined #openstack-kuryr | 07:37 | |
apuimedo | ok | 07:38 |
apuimedo | well, so then the steps are | 07:38 |
apuimedo | cloning devstack and kuryr-kubernetes on the VM | 07:38 |
apuimedo | get my latest patchset | 07:38 |
apuimedo | get a keystonerc from the undercloud | 07:39 |
apuimedo | source it | 07:39 |
apuimedo | and edit the local.conf that dmellado pointed you to | 07:39 |
apuimedo | (you should create some nets, subnets and LB resources in the undercloud as well | 07:39 |
apuimedo | ) | 07:39 |
apuimedo | There is a good reason I made ./devstack-overcloud-heat. It's a PITA to set this up by hand :P | 07:40 |
apuimedo | irenab: what kind of tweaks did you do? | 07:40 |
apuimedo | Maybe I can put them optional in the patch | 07:40 |
apuimedo | ltomasbo|away: ping | 07:44 |
apuimedo | dmellado: how are your viruses? | 07:45 |
dmellado | apuimedo: still so-so | 07:45 |
apuimedo | or rather... Who's winning? | 07:45 |
dmellado | better than yesterday, though | 07:45 |
dmellado | we're still in a stalemate | 07:45 |
apuimedo | :-) | 07:45 |
apuimedo | I am happy I managed to sleep with the window open | 07:45 |
dmellado | frenadol gets me better but gets me really, really sleepy | 07:45 |
apuimedo | the pollen counts must be lowering | 07:45 |
apuimedo | dmellado: did I tell you how bothering lazy consensus is? | 07:46 |
apuimedo | I thought we'd already have the tempest repo :/ | 07:46 |
dmellado | apuimedo: we do | 07:46 |
dmellado | I just need to add the gate to kuryr | 07:46 |
apuimedo | but the patch doesn't have w+1 | 07:46 |
dmellado | lol, what happened due to consensus | 07:47 |
apuimedo | does it? | 07:47 |
dmellado | ahhh you mean the governance one | 07:47 |
apuimedo | yeah | 07:47 |
dmellado | well, no worries about it | 07:47 |
dmellado | the repo itself it's already up 'n running on the openstack infra | 07:47 |
dmellado | the governance thing, well | 07:47 |
dmellado | tbh idk how long it'd take... | 07:47 |
dmellado | but the repo is now here | 07:48 |
dmellado | https://github.com/openstack/kuryr-tempest-plugin | 07:48 |
apuimedo | it takes a week | 07:48 |
apuimedo | dmellado: irenab: https://hub.docker.com/r/kuryr/controller/ | 07:49 |
apuimedo | I made a new container image (taking over vikasc containerization patch) | 07:50 |
dmellado | apuimedo: where are the descriptions? -1 xD | 07:50 |
apuimedo | dmellado: my parser doesn't understand | 07:50 |
apuimedo | desc... what? | 07:50 |
vikasc | apuimedo, i am just working on that patch to address review comments | 07:50 |
dmellado | apuimedo: I see that it was really vikasc patch | 07:50 |
dmellado | as I'm still waiting for the documentation xD | 07:51 |
apuimedo | vikasc: I thought you were not working on that one anymore | 07:51 |
dmellado | that said, I do also need to edit the docs for the new repo and didn't do that yet | 07:51 |
apuimedo | I'm working on splitting into two different images | 07:51 |
apuimedo | the controller one | 07:51 |
apuimedo | running as use | 07:51 |
vikasc | apuimedo, i was about to push | 07:51 |
apuimedo | *user | 07:51 |
apuimedo | and the cni one | 07:51 |
apuimedo | that I still didn't start | 07:51 |
apuimedo | I also took some measures to reduce the container size | 07:52 |
vikasc | apuimedo, sorry about the delay | 07:52 |
apuimedo | it would help if you keep us updated | 07:52 |
apuimedo | otherwise I assume that you're busy with other things and take over | 07:53 |
vikasc | apuimedo, IIRC i told you privately | 07:53 |
vikasc | apuimedo, that i will be busy a week or so | 07:53 |
dmellado | maybe we could do some kind of quick status on the weekly meeting | 07:53 |
vikasc | apuimedo, can i continue that patch now | 07:54 |
vikasc | apuimedo, or if you want to continue thats also fine with me | 07:54 |
*** aojea has joined #openstack-kuryr | 07:54 | |
apuimedo | vikasc: please, send what you were about to push | 07:55 |
apuimedo | we can converge with my changes later | 07:55 |
vikasc | apuimedo, ok, iwas testing again and addressing review comments such as templating variables in config map | 07:55 |
vikasc | apuimedo, you just pushed controller image, right? | 07:56 |
*** neiljerram has quit IRC | 07:57 | |
vikasc | apuimedo, built with some changes on top of dockerfile in my patch | 07:57 |
vikasc | apuimedo, i am thinking how it would be if i test with the container image that you pushed | 07:57 |
vikasc | apuimedo, rather building local docker image and adapt my patch to this image which you pushed. WDYT? | 07:58 |
vikasc | apuimedo, its just two week old patch man, you taking over without notifying :) | 07:59 |
apuimedo | vikasc: yes. It's in the kuryr organization on docker hub | 07:59 |
vikasc | apuimedo, cool | 07:59 |
apuimedo | vikasc: well. You'd need to make a change to the deployment | 07:59 |
apuimedo | so that the sections are mounted in /etc/kuryr/ | 08:00 |
apuimedo | I'm not using the bash file for startup | 08:00 |
apuimedo | I changed the configmap and mounts so that we have | 08:00 |
vikasc | apuimedo, where is the dockerfile for this image | 08:00 |
apuimedo | /etc/kuryr/neutron.conf /etc/kuryr/neutron_defaults.conf | 08:00 |
vikasc | apuimedo, which you created | 08:00 |
apuimedo | (from your sections) | 08:00 |
vikasc | apuimedo, so that i can understand | 08:00 |
*** neiljerram has joined #openstack-kuryr | 08:01 | |
apuimedo | I'll send the patchset with a different changeid so you can see | 08:01 |
vikasc | apuimedo, cool | 08:01 |
*** aojea has quit IRC | 08:01 | |
openstackgerrit | Antoni Segura Puimedon proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/474478 | 08:01 |
vikasc | apuimedo, i would like to continue my patch | 08:15 |
vikasc | apuimedo, i dont think i delayed that long | 08:15 |
apuimedo | vikasc: please continue | 08:16 |
vikasc | apuimedo, i worked a lot to figure out things on this patch | 08:16 |
*** ltomasbo|away is now known as ltomasbo | 08:17 | |
vikasc | apuimedo, thanks | 08:17 |
dmellado | vikasc: what about the documentation? xD | 08:22 |
vikasc | dmellado, documentation means steps to use it? | 08:23 |
apuimedo | vikasc: how to make the build | 08:23 |
apuimedo | it should probably have a description of what the kuryr.yml is for (the one you put is for pod-in-vm, for example) | 08:24 |
vikasc | apuimedo, dmellado will add more details to READMe | 08:24 |
apuimedo | vikasc: please, don't forget to take the '.dockerignore' from the patch I sent | 08:25 |
vikasc | apuimedo, sure | 08:25 |
apuimedo | otherwise the image is 1GiB instead of 327MiB | 08:25 |
apuimedo | (depending on the amount of tox in your system, of course) | 08:26 |
apuimedo | ltomasbo: I checked and in baremetal I didn't experience the failure to delete the ports | 08:26 |
irenab | apuimedo, I deployed the VM with heat patch, so trunk and all the subnets are defined. The issue was to ssh into this VM | 08:28 |
*** garyloug has joined #openstack-kuryr | 08:29 | |
apuimedo | irenab: a problem with the FIP DNAT? | 08:32 |
apuimedo | irenab: or did it also affect the user_data for the VM ? | 08:32 |
irenab | yes, with FIP. What do you mean by user_data? | 08:34 |
irenab | I created local conf like this: https://pastebin.com/1atBxeNy | 08:35 |
ltomasbo | apuimedo, then it could be due to a missing dependency on the kuryr-lib | 08:40 |
*** kzaitsev_ws has joined #openstack-kuryr | 08:40 | |
ltomasbo | apuimedo, perhaps the call failing is the one releasing the vlan_id | 08:40 |
ltomasbo | there is one fast thing you can try | 08:40 |
ltomasbo | comment out this line: https://github.com/openstack/kuryr-kubernetes/blob/master/kuryr_kubernetes/controller/drivers/nested_vlan_vif.py#L109 | 08:41 |
ltomasbo | and try again | 08:41 |
apuimedo | irenab: the thing is that you shouldn't need to create a local.conf | 08:42 |
apuimedo | if your cloud's metadata works | 08:42 |
apuimedo | it should already be done in the VM thanks to the user-data I put in the heat template | 08:42 |
apuimedo | ltomasbo: what do you plan on working next? | 08:43 |
ltomasbo | xD | 08:43 |
ltomasbo | two follow up patches I'll send later today | 08:43 |
apuimedo | 'bout? | 08:43 |
dmellado | about documentation? xD | 08:44 |
ltomasbo | readme documentation regarding vif_pool utilization | 08:44 |
ltomasbo | and another one to ensure no leftovers of ports | 08:44 |
apuimedo | dmellado: you keep bringing up these non-existing words | 08:44 |
ltomasbo | when kuryr-controller is rebooted | 08:44 |
dmellado | apuimedo: but I was right! | 08:44 |
dmellado | xD | 08:44 |
apuimedo | must be some invented word from Madrid | 08:44 |
dmellado | you all know that you can now propose patches to the awesome new repo | 08:44 |
dmellado | and yes, that another word is tests! | 08:45 |
dmellado | xD | 08:45 |
apuimedo | we know about tests | 08:45 |
apuimedo | unit tests | 08:45 |
dmellado | I'll be putting up the gate as non-voting first | 08:45 |
apuimedo | next you are going to tell me other kinds of tests exist | 08:45 |
dmellado | but be prepared | 08:45 |
dmellado | muahahaha | 08:45 |
dmellado | xd | 08:45 |
kzaitsev_ws | dmellado: apuimedo: speaking about docs — would appreciate if you would look at https://review.openstack.org/#/c/472653/ and tell me if it's the good direction =) | 08:45 |
apuimedo | ltomasbo: what is next after that? | 08:45 |
ltomasbo | apuimedo, good question | 08:45 |
apuimedo | :O | 08:46 |
dmellado | kzaitsev_ws: will totally do! thanks! | 08:46 |
apuimedo | kzaitsev_ws: you're doing God's work | 08:46 |
ltomasbo | multi nic support and/or the possibility of using different subnets | 08:46 |
dmellado | multi nic/multi cni | 08:46 |
apuimedo | ltomasbo: there's already kzaitsev_ws doing multivif | 08:46 |
dmellado | would be nice | 08:46 |
* dmellado wonders how rdo cloud dsvm can be sooo slow... | 08:47 | |
irenab | apuimedo, so no local.conf in VM? | 08:47 |
apuimedo | dmellado: it runs on Athlon64 | 08:47 |
apuimedo | irenab: it should already have been created by cloud-init | 08:47 |
dmellado | apuimedo: I bet it runs on 486SX2 | 08:47 |
dmellado | without math co-processor | 08:47 |
apuimedo | dmellado: just in case, ask them if they are using qemu or qemu-kvm | 08:48 |
apuimedo | xD | 08:48 |
dmellado | I'd say than more than 'qemu' | 08:48 |
dmellado | they're using 'quemal' | 08:48 |
dmellado | xD | 08:48 |
apuimedo | #dadjokes | 08:48 |
dmellado | lol | 08:48 |
dmellado | xD | 08:48 |
apuimedo | vikasc: did you read the comment I put to my patch? | 08:49 |
apuimedo | about using a debian container to build the binary | 08:49 |
vikasc | apuimedo, about cni? yes | 08:49 |
apuimedo | and then put that binary into the cni container that writes it into the host file system? | 08:49 |
apuimedo | (just like docker-compose does with their scripts/build/linux | 08:50 |
vikasc | apuimedo, i will get to it, once i have adressed existing review comments and incorporated changes you made in dockerfile | 08:50 |
apuimedo | ok | 08:50 |
apuimedo | of course, that wouldn't be that necessary if we were good folks and had already split CNI | 08:51 |
vikasc | apuimedo, yep | 08:51 |
vikasc | apuimedo because it is working with existing changes, will improve it in steps | 08:52 |
vikasc | apuimedo, i mean cni | 08:52 |
*** kzaitsev_ has joined #openstack-kuryr | 08:53 | |
janonymous | kzaitsev_ws: ping | 08:53 |
kzaitsev_ | sry my colleagues dragged me off to lunch | 08:53 |
vikasc | apuimedo, so you want kuryr.conf to get created directly via config map? | 08:54 |
kzaitsev_ | will be back in 15 min :)) | 08:54 |
dmellado | apuimedo: with your patch, on dsvm | 08:54 |
dmellado | the kuryr-kubernetes service is not working for me | 08:54 |
dmellado | -1 'al canto' | 08:54 |
dmellado | xD | 08:54 |
apuimedo | vikasc: no | 08:55 |
apuimedo | what my patch does is create a .conf file for each section | 08:55 |
apuimedo | and place those on /etc/kuryr/ | 08:55 |
vikasc | apuimedo, i see that you are passing --config-dir | 08:55 |
dmellado | looks like some kind of requirements issue | 08:55 |
dmellado | Jun 15 08:50:15 devstackrdo.novalocal python[6761]: 2017-06-15 08:50:15.065 6761 ERROR kuryr-kubernetes ImportError: No module named generic_vif | 08:55 |
apuimedo | then, instead of using --config-file /etc/kuryr/kuryr.conf | 08:55 |
apuimedo | we just pass --config-dir | 08:55 |
apuimedo | so all of them will be read by oslo | 08:55 |
vikasc | apuimedo, thats why my kryr.yml also does | 08:55 |
vikasc | s/what | 08:56 |
apuimedo | vikasc: but you do one extra step | 08:56 |
apuimedo | of putting it in /tmp | 08:56 |
apuimedo | and then on container start putting it all into a /etc/kuryr/kuryr.conf | 08:56 |
apuimedo | which is unnecessary | 08:56 |
vikasc | apuimedo, i was not aware that oslo will read directly | 08:56 |
apuimedo | vikasc: neutron uses --config-dir for some time | 08:56 |
vikasc | apuimedo, and was creating kuryr.conf myself | 08:56 |
apuimedo | dmellado: can I get more details? | 08:57 |
dmellado | apuimedo: I'll comment on the review, hold on | 08:57 |
vikasc | apuimedo, cool, so instead of tmp it those sections are to be copied in config-dir | 08:57 |
apuimedo | not copied | 08:57 |
vikasc | apuimedo, thanks | 08:57 |
vikasc | apuimedo, created | 08:57 |
apuimedo | they are mounted directly there | 08:57 |
vikasc | apuimedo, yep | 08:57 |
ltomasbo | dmellado, are you deploying with last master patches? | 08:58 |
dmellado | ltomasbo: yep, just fresh vm | 08:58 |
ltomasbo | that generic_vif seems to be due to the renaming to neutron_vif | 08:58 |
ltomasbo | perhaps kuryr.conf has the wrong driver then | 08:59 |
ltomasbo | can you check? if you have something there saying generic_vif? | 08:59 |
ltomasbo | otherwise we may have some leftovers... | 08:59 |
dmellado | ltomasbo: that could be | 09:00 |
dmellado | apuimedo: ltomasbo I put the trace on the commit comments | 09:01 |
ltomasbo | link? | 09:01 |
vikasc | apuimedo, i will also work on templating config parameters, that are used in config map | 09:01 |
dmellado | https://review.openstack.org/#/c/472763 | 09:02 |
apuimedo | dmellado: what command exactly did you use with ./devstack-overcloud-heat ? | 09:02 |
dmellado | apuimedo: I didn't even get to that | 09:03 |
dmellado | just stacking | 09:03 |
apuimedo | ah... YOu used my patch also for stacking the "undercloud"? | 09:03 |
* dmellado is stupid | 09:03 | |
dmellado | git pull --rebase origin master | 09:04 |
dmellado | over your patch | 09:04 |
dmellado | nvm | 09:04 |
dmellado | frenadol-- | 09:04 |
openstackgerrit | Daniel Mellado proposed openstack/kuryr-kubernetes master: contrib: devstack-overcloud-heat https://review.openstack.org/472763 | 09:04 |
*** kzaitsev_ has quit IRC | 09:05 | |
dmellado | apuimedo: yeah, I had that on the GITREF form kuryr-k8s dsvm plugin | 09:05 |
apuimedo | ah | 09:10 |
dmellado | I rebased it | 09:11 |
apuimedo | I saw | 09:17 |
*** dougbtv__ has joined #openstack-kuryr | 09:22 | |
openstackgerrit | Jaivish Kothari(janonymous) proposed openstack/kuryr-kubernetes master: Stopping Thread from oslo_service https://review.openstack.org/474519 | 09:32 |
kzaitsev_ws | ltomasbo: yeah, I have a small initial patch for the multi-vif. It adds just the ability to handle it, but not the mechanism to request it. | 09:36 |
kzaitsev_ws | janonymous: pong =) | 09:37 |
ltomasbo | kzaitsev_ws, nice! I'll take a look! | 09:39 |
ltomasbo | is it this one: https://review.openstack.org/#/c/471012/ | 09:39 |
kzaitsev_ws | ltomasbo: yep. I based my sriov patches on top of it | 09:41 |
ltomasbo | great! I'll take a look later today | 09:41 |
*** limao has quit IRC | 09:41 | |
ltomasbo | so, what is missing is the controller-driver side? | 09:41 |
ltomasbo | to requests the cni different vifs? | 09:41 |
kzaitsev_ws | thnks =) | 09:42 |
kzaitsev_ws | ltomasbo: the mechanism to actually request them | 09:42 |
ltomasbo | ok | 09:42 |
kzaitsev_ws | ltomasbo: I mean. for sriov the mechanism is to request opaque-int-resource in k8s | 09:42 |
ltomasbo | https://github.com/Huawei-PaaS/CNI-Genie | 09:42 |
ltomasbo | perhaps we can take a look at how they deal with that there, and do something similar | 09:43 |
ltomasbo | for requesting different vifs | 09:43 |
kzaitsev_ws | you say — I want 2 opir physnet3 and vif-handler says — take these 2 direct ports | 09:43 |
ltomasbo | that in the pod template, right? | 09:44 |
ltomasbo | kzaitsev_ws, ^^ | 09:44 |
ltomasbo | so perhaps we can use annotations to specify the pod will use 2 vifs, similarly how they decide to use 2 cnis | 09:45 |
kzaitsev_ws | ltomasbo: cni-genie is probably a dead end. it simply a proxy, that asks several cni drivers for ips. | 09:45 |
kzaitsev_ws | if we want all the IPs to be managed by kuryr — there's no point in using cni-genie (or multus-cni) | 09:45 |
ltomasbo | kzaitsev_ws, I didn't mean to use it, but to have a similar approach into how to request multiple vifs | 09:45 |
kzaitsev_ws | I guess they're very similar | 09:45 |
apuimedo | yes, it should be metadata | 09:45 |
apuimedo | probably customresources as well | 09:46 |
apuimedo | but that comes with 1.7.0 | 09:46 |
apuimedo | and there's no fscking documentation on them | 09:46 |
ltomasbo | :( | 09:46 |
kzaitsev_ws | ltomasbo: yeah. for sriov the pod asks for opaque int resources in the pod spec. | 09:46 |
kzaitsev_ws | one option would be to request ports by specifying port-id's | 09:46 |
ltomasbo | I was thinking in something like: metadata: annotations: vifs: 2 | 09:47 |
kzaitsev_ws | in the annotation | 09:47 |
kzaitsev_ws | ^^ is also an option, yeah. | 09:48 |
kzaitsev_ws | I specifically left the thing out of scope of https://review.openstack.org/#/c/471012/ patch, because there are several options to do it, while the mechanism to pass from controller to cni and for cni to handle multi-vif annotation doesn't really depend on the way vifs are requested | 09:49 |
ltomasbo | kzaitsev_ws, sounds good! | 09:50 |
*** rwallner has joined #openstack-kuryr | 09:51 | |
kzaitsev_ws | ltomasbo: the way to request multiple vifs would probably require some discussion and compromises anyway =) | 09:51 |
ltomasbo | kzaitsev_ws, sure! | 09:51 |
ltomasbo | something to discuss in a call or during the kuryr weekly IRC meeting | 09:52 |
*** rwallner has quit IRC | 09:56 | |
kzaitsev_ws | ltomasbo: apuimedo: even for mechanism there is one thing to agree. instead of a list of vifs I pass a dict {ifname:vif}, so it's responsibility of the handler to decide what would be the interface name | 09:56 |
kzaitsev_ws | that's not what's described in the bp, but we have discussed that at some point inside sriov spec with ivc_ (his suggestion =)) and I liked it and went with it | 09:57 |
kzaitsev_ws | anyway =) ltomasbo: would really appreciate if you would take a look at the patch and see if there' | 09:58 |
kzaitsev_ws | there's smth to improve there | 09:58 |
ltomasbo | kzaitsev_ws, sure, I'll do it later today | 10:05 |
dmellado | folks, I'm hopping to the train to hot city | 10:06 |
* dmellado sighs | 10:06 | |
dmellado | ttyl! | 10:06 |
openstackgerrit | vikas choudhary proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/466675 | 10:17 |
apuimedo | kzaitsev_ws: I'm a bit surprised by having a dict instead of list | 10:18 |
apuimedo | but it makes sense | 10:18 |
apuimedo | I haven't fully thought this through | 10:18 |
vikasc | apuimedo, irenab kzaitsev_pi ^ | 10:18 |
apuimedo | thanks vikasc | 10:18 |
vikasc | apuimedo, yw! i will now work to complete what you said about cni build | 10:19 |
apuimedo | thanks | 10:22 |
vikasc | irenab, kzaitsev_pi : i did not change configmap to template the variables because i am not sure exactly what is your expectation by that. is that for each kuryr.conf section, we should have a template/sample file in kuryr-kubernetes/ and then those should be read in configmap? | 10:23 |
apuimedo | vikasc: I don't think it should be templated | 10:23 |
vikasc | instead of directly changing configmap in yaml | 10:23 |
apuimedo | there should just be a README explaining which values should be changed | 10:23 |
apuimedo | and with what | 10:24 |
vikasc | apuimedo, i too felt so thats why asking again | 10:24 |
janonymous | kzaitsev_ws:I think this might be required for the bug : https://review.openstack.org/#/c/474519/ , but have to test locally... | 10:24 |
vikasc | apuimedo, may be we can add tox generated sample kuryr.conf | 10:24 |
apuimedo | vikasc: maybe instead of concrete example values, we can have something like | 10:24 |
apuimedo | my_subnet_id | 10:25 |
apuimedo | etc | 10:25 |
vikasc | apuimedo, how about adding tox generated | 10:25 |
apuimedo | vikasc: I'd rather not have the tox generated example | 10:25 |
apuimedo | the only option I consider is that we'll have a one-off pod that will use neutron to create the resources if there is no config map | 10:25 |
apuimedo | create the config map | 10:25 |
apuimedo | and that will retrigger the controller deployment | 10:26 |
mchiappero | hi guys | 10:26 |
apuimedo | but we can leave this improvement for later | 10:26 |
apuimedo | mchiappero: Hi! | 10:26 |
mchiappero | I'm afraid unstack.sh in many cases doesn't correctly clean up the load balancer | 10:26 |
mchiappero | apuimedo: hi! :) | 10:26 |
vikasc | apuimedo, cool | 10:27 |
apuimedo | mchiappero: interesting | 10:27 |
vikasc | apuimedo, let me figure out cni part, will get back if face confusion | 10:27 |
apuimedo | please. Report a bug and assign it to me | 10:27 |
mchiappero | I have to say I'm stacking and unstacking very often.. | 10:27 |
apuimedo | mchiappero: no excuse for not cleaning up | 10:27 |
mchiappero | I'll try to see if I find a consistent way to reproduce | 10:27 |
apuimedo | same I tell my son after lego | 10:27 |
mchiappero | lol | 10:28 |
mchiappero | today I'm busy but tomorrow I should be able to continue on macvlan and investigate this | 10:28 |
*** irena has joined #openstack-kuryr | 10:30 | |
apuimedo | perfect | 10:31 |
apuimedo | thanks mchiappero | 10:31 |
kzaitsev_ws | folks also a small heads up — I'll be travelling next week. I'll try my best to attend the weekly meeting, but I might have limited connectivity | 10:38 |
kzaitsev_ws | janonymous: yeah, I'll try to test it too ) just need to restore sonme of my devstacks =) | 10:39 |
apuimedo | thanks for the heads up kzaitsev_ws | 10:43 |
kzaitsev_ws | apuimedo: about the dict — it potentially gives us more power (i.e. user will be able to request vifs with specific names). And in the end either CNI or the handler would have to decide ifnames anyway. However I might be missing smth, so yeah this needs to be thought through | 10:45 |
irenab | apuimedo, let me know if you need me to host next weekly | 10:49 |
apuimedo | I can host Monday 19th, but I'll need you to host Monday 26th | 10:51 |
apuimedo | I'll be on a plane | 10:51 |
irenab | ok, just let me know | 10:52 |
*** salv-orlando has quit IRC | 10:56 | |
mchiappero | I'm facing a problem | 10:58 |
mchiappero | RequiredOptError: value required for option project in group [neutron_defaults] | 10:58 |
mchiappero | when trying to handle load balancer events | 10:59 |
mchiappero | what is the required config option? | 10:59 |
mchiappero | from File "/opt/stack/kuryr-kubernetes/kuryr_kubernetes/controller/drivers/default_project.py", line 51, in get_project | 10:59 |
mchiappero | from File "/opt/stack/kuryr-kubernetes/kuryr_kubernetes/controller/handlers/lbaas.py", line 74, in _generate_lbaas_spec | 11:00 |
janonymous | kzaitsev_ws: yup, i tried locally and seems to work. | 11:02 |
apuimedo | mchiappero: let me check | 11:02 |
apuimedo | mchiappero: of these, which are you missing | 11:03 |
apuimedo | project, pod_subnet, pod_security_groups, service_subnet, ovs_bridge | 11:04 |
apuimedo | the project one, right? | 11:04 |
mchiappero | let me check... | 11:05 |
irena | apuimedo, I hit this error in overcloud VM | 11:06 |
irena | /opt/stack/kuryr-kubernetes/devstack/plugin.sh:prepare_kubernetes_files:393 : openstack --os-cloud devstack-admin --os-region RegionOne subnet show k8s-service-subnet -c cidr -f value ResourceNotFound: No Subnet found for k8s-service-subnet | 11:06 |
openstackgerrit | Yuval Brik proposed openstack/kuryr-kubernetes master: devstack: fix dragonflow gate https://review.openstack.org/474544 | 11:07 |
mchiappero | oh my... sorry.. i just restacked and my config file has been replaced, it seems | 11:07 |
mchiappero | but IIRC it used to preserve pre-existing config files, am I wrong? | 11:08 |
apuimedo | irena: did you use the --override-keystone? | 11:11 |
mchiappero | BTW, another issue I have with devstack is that if docker is already installed and commented unstack will stop it | 11:11 |
apuimedo | mchiappero: which kuryr.conf or local.conf? | 11:11 |
mchiappero | kuryr.conf | 11:11 |
apuimedo | mchiappero: it probably did preserve before | 11:12 |
apuimedo | because it was not using the oslo-generator | 11:12 |
apuimedo | it was just doing iniset | 11:12 |
mchiappero | ok, wrong assumption on my side then, sorry | 11:12 |
apuimedo | we added oslo-generator there so that the stacking grabs any new options (and also so it tests that the config generation is not broken) | 11:12 |
mchiappero | I see, it's cool actually | 11:13 |
irena | apuimedo, I run the overcloud VM on devstack. As you suggested, no local.conf | 11:13 |
apuimedo | gotta go for lunch now | 11:20 |
apuimedo | will ping you later irena | 11:20 |
irenab | apuimedo, ok. Its in the poart that was added to enable k8s api service/router | 11:21 |
irenab | dmellado, does the overcloud VM works for you? | 11:22 |
mchiappero | apuimedo: sorted, thanks :) | 11:24 |
*** salv-orlando has joined #openstack-kuryr | 11:34 | |
openstackgerrit | vikas choudhary proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/466675 | 12:00 |
*** dougbtv__ has quit IRC | 12:03 | |
openstackgerrit | vikas choudhary proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/466675 | 12:06 |
openstackgerrit | vikas choudhary proposed openstack/kuryr-kubernetes master: Add support to install Kuryr as a network addon https://review.openstack.org/466675 | 12:07 |
apuimedo | irenab: the thing is. That part should not run if you called ./devstack-overcloud-heat with --override-keystone flag | 12:25 |
*** rwallner has joined #openstack-kuryr | 12:27 | |
*** rwallner has quit IRC | 12:28 | |
irenab | apuimedo, I didn't since its on devstack. | 12:35 |
irenab | currently I prefer to change settings inside the VM and not to re-run the devstack-overcloud-heat | 12:36 |
irenab | apuimedo, any advice on this? | 12:36 |
apuimedo | let's have a call in 30min and check it? | 12:36 |
apuimedo | or is it too late? | 12:36 |
irenab | I will need to leave in 20 mins ... | 12:37 |
apuimedo | :( | 12:37 |
apuimedo | I'm in a call now | 12:37 |
*** atoth has joined #openstack-kuryr | 12:40 | |
*** edisonxiang has quit IRC | 12:49 | |
*** rwallner has joined #openstack-kuryr | 12:53 | |
*** rwallner has quit IRC | 12:54 | |
*** rwallner has joined #openstack-kuryr | 12:55 | |
*** irena has quit IRC | 12:58 | |
*** gouthamr has joined #openstack-kuryr | 13:12 | |
openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Delete or recover precreated ports upon controller restart https://review.openstack.org/474576 | 13:27 |
*** yedongcan has quit IRC | 14:02 | |
*** yedongcan has joined #openstack-kuryr | 14:03 | |
*** aojea has joined #openstack-kuryr | 14:06 | |
openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add ports pool configuration information at README https://review.openstack.org/474589 | 14:09 |
*** tonanhngo has joined #openstack-kuryr | 14:11 | |
*** tonanhngo has quit IRC | 14:15 | |
*** limao has joined #openstack-kuryr | 14:51 | |
*** roeyc has quit IRC | 14:52 | |
*** limao_ has joined #openstack-kuryr | 14:56 | |
*** limao has quit IRC | 14:56 | |
*** janki has quit IRC | 15:12 | |
*** hongbin has joined #openstack-kuryr | 15:28 | |
*** limao_ has quit IRC | 15:31 | |
*** yedongcan has left #openstack-kuryr | 15:33 | |
*** aojea has quit IRC | 15:56 | |
*** egonzalez has quit IRC | 16:00 | |
*** garyloug has quit IRC | 16:37 | |
*** ltomasbo is now known as ltomasbo|away | 16:56 | |
*** neiljerram has quit IRC | 17:03 | |
*** roeyc has joined #openstack-kuryr | 17:04 | |
*** salv-orl_ has joined #openstack-kuryr | 17:14 | |
*** pcaruana has quit IRC | 17:16 | |
*** salv-orlando has quit IRC | 17:17 | |
*** kzaitsev_ws has quit IRC | 17:31 | |
*** roeyc has quit IRC | 18:59 | |
*** roeyc has joined #openstack-kuryr | 19:00 | |
*** zengchen has quit IRC | 19:04 | |
*** zengchen has joined #openstack-kuryr | 19:04 | |
*** egonzalez has joined #openstack-kuryr | 19:13 | |
*** roeyc has quit IRC | 19:16 | |
*** neiljerram has joined #openstack-kuryr | 20:23 | |
*** rwallner has quit IRC | 20:28 | |
*** rwallner has joined #openstack-kuryr | 20:32 | |
*** egonzalez has quit IRC | 20:35 | |
*** rwallner has quit IRC | 20:51 | |
*** oanson has quit IRC | 21:23 | |
*** lihi has quit IRC | 21:23 | |
*** gouthamr has quit IRC | 21:23 | |
*** lihi has joined #openstack-kuryr | 21:25 | |
*** oanson has joined #openstack-kuryr | 21:26 | |
*** rwallner has joined #openstack-kuryr | 21:37 | |
*** gouthamr has joined #openstack-kuryr | 21:42 | |
*** rwallner has quit IRC | 21:53 | |
*** oanson has quit IRC | 22:38 | |
*** salv-orl_ has quit IRC | 22:39 | |
*** oanson has joined #openstack-kuryr | 22:39 | |
*** neiljerram has quit IRC | 22:47 | |
*** neiljerram has joined #openstack-kuryr | 23:06 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!