*** maysams has joined #openstack-kuryr | 00:37 | |
*** maysams has quit IRC | 00:38 | |
openstackgerrit | Yash Gupta proposed openstack/kuryr-kubernetes master: Refactor DriverBase.get_instance() https://review.openstack.org/596672 | 01:28 |
---|---|---|
openstackgerrit | Yash Gupta proposed openstack/kuryr-kubernetes master: Use same pool_driver for different pod_vif_drivers https://review.openstack.org/596673 | 01:31 |
*** maysams has joined #openstack-kuryr | 02:33 | |
*** maysams has quit IRC | 04:31 | |
*** yboaron_ has joined #openstack-kuryr | 05:27 | |
*** itzikb has joined #openstack-kuryr | 06:03 | |
*** threestrands has joined #openstack-kuryr | 06:13 | |
*** pcaruana has joined #openstack-kuryr | 06:23 | |
*** gkadam has joined #openstack-kuryr | 06:25 | |
*** ccamposr has joined #openstack-kuryr | 06:29 | |
openstackgerrit | Itzik Brown proposed openstack/kuryr-kubernetes master: Add kuryr_daemon_enabled when kuryr-daemon exists https://review.openstack.org/596690 | 06:33 |
openstackgerrit | Merged openstack/kuryr-tempest-plugin master: Refactor options definitons https://review.openstack.org/590415 | 06:41 |
aperevalov | hi, I see kuryr-kubernetes replaces netns in convert_netns, replaces by CONF.cni_daemon.netns_proc_dir, I didn't find where is generates, but in has in http://logs.openstack.org/52/601552/1/check/kuryr-kubernetes-tempest-daemon-containerized-octavia/0d5ba85/controller/logs/kubernetes/configmaps.txt.gz. | 06:46 |
aperevalov | Ok, I see configmap generates by provided kuryr.conf, here no problem, but I think it was better just pass CNI_NETNS env to container in docker exec already modified instead of checking it every time. But it's my IMHO. | 06:49 |
openstackgerrit | Itzik Brown proposed openstack/kuryr-tempest-plugin master: Add test_pod_pod_ping_different_host https://review.openstack.org/599314 | 07:20 |
*** yboaron_ has quit IRC | 07:20 | |
*** celebdor has joined #openstack-kuryr | 07:26 | |
*** pmannidi has quit IRC | 07:50 | |
*** jistr|off is now known as jistr | 07:55 | |
*** yboaron_ has joined #openstack-kuryr | 08:12 | |
*** garyloug has joined #openstack-kuryr | 08:20 | |
*** itzikb is now known as itzikb|afk | 08:24 | |
*** celebdor has quit IRC | 08:30 | |
openstackgerrit | Genadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Function for updating configMap https://review.openstack.org/599576 | 09:11 |
openstackgerrit | Itzik Brown proposed openstack/kuryr-tempest-plugin master: Add test_pod_pod_ping_same_node test https://review.openstack.org/600443 | 09:23 |
openstackgerrit | Danil Golov proposed openstack/kuryr-kubernetes master: Add SR-IOV pod vif driver https://review.openstack.org/512280 | 09:23 |
openstackgerrit | Danil Golov proposed openstack/kuryr-kubernetes master: Add SR-IOV binding driver to CNI https://review.openstack.org/512281 | 09:23 |
openstackgerrit | Danil Golov proposed openstack/kuryr-kubernetes master: Add HOWTO for SRIOV use case https://review.openstack.org/594125 | 09:23 |
openstackgerrit | Danil Golov proposed openstack/kuryr-kubernetes master: Introduce test case document for SRIOV functionality https://review.openstack.org/600022 | 09:23 |
*** danil has joined #openstack-kuryr | 09:25 | |
*** itzikb|afk is now known as itzikb | 09:29 | |
openstackgerrit | Genadi Chereshnya proposed openstack/kuryr-tempest-plugin master: Function for updating configMap https://review.openstack.org/599576 | 09:34 |
*** itzikb is now known as itzikb|afk | 09:37 | |
openstackgerrit | Merged openstack/kuryr-kubernetes master: Health Manager details update https://review.openstack.org/600720 | 09:51 |
*** itzikb|afk is now known as itzikb | 10:06 | |
*** s1061123 has quit IRC | 10:07 | |
*** celebdor has joined #openstack-kuryr | 10:14 | |
*** celebdor has quit IRC | 10:15 | |
*** celebdor has joined #openstack-kuryr | 10:15 | |
openstackgerrit | Itzik Brown proposed openstack/kuryr-tempest-plugin master: Check that kuryr-cni runs as a daemon https://review.openstack.org/596564 | 10:24 |
*** maysams has joined #openstack-kuryr | 11:58 | |
openstackgerrit | Ashish Billore proposed openstack/kuryr-kubernetes master: Add missing key check while lbaas member add https://review.openstack.org/602017 | 12:30 |
dulek | aperevalov: Either way works really. Your approach is probably a bit more clean as all the container quirks would be packed into container script. | 12:39 |
dulek | aperevalov: So I'm not against it if you'll make it work. :) | 12:40 |
aperevalov | dulek: I decided to abandon it becase there is use cases where convert_netns is calling with hard coded string, like in kuryr_kubernetes/cni/binding/base.py | 12:44 |
aperevalov | dulek: path = utils.convert_netns('/proc/self/ns/net') | 12:45 |
aperevalov | so convert_netns will exist and will be necessary, even CNI_NETNS will be passed with correct /host_proc/ path. Or we have to provide special pathes for /proc in container and keep it somewhere in config | 12:46 |
dulek | aperevalov: Ah, you're right. Thanks for looking at this! | 12:48 |
aperevalov | dulek: thanks for review ;) | 12:49 |
*** gkadam has quit IRC | 12:49 | |
*** garyloug_ has joined #openstack-kuryr | 13:01 | |
*** garyloug has quit IRC | 13:05 | |
*** itzikb is now known as itzikb|afk | 13:13 | |
*** threestrands has quit IRC | 13:24 | |
*** itzikb|afk is now known as itzikb | 13:36 | |
aperevalov | we'll have meeting tomorrow, I prepared plan for https://etherpad.openstack.org/p/KuryrSyncSRIOV-DPDK currently it's just a list of issues | 13:46 |
*** itzikb is now known as itzikb|afk | 13:54 | |
*** garyloug_ has quit IRC | 14:00 | |
*** irenab has quit IRC | 14:02 | |
*** garyloug has joined #openstack-kuryr | 14:03 | |
*** gcheresh_ has joined #openstack-kuryr | 14:04 | |
*** irenab has joined #openstack-kuryr | 14:04 | |
*** gcheresh_ has quit IRC | 14:07 | |
*** hongbin has quit IRC | 14:07 | |
*** itzikb|afk is now known as itzikb | 14:14 | |
*** rossella_s has joined #openstack-kuryr | 14:20 | |
dmellado | dulek: I'll add you to the meeting | 14:22 |
dmellado | just in case you can attend | 14:22 |
rossella_s | mrostecki, I was talking to celebdor , he's very keen about seeing the cilium integration upstream | 14:23 |
dulek | dmellado: Sure, thanks! | 14:23 |
celebdor | I'd really love it | 14:36 |
dmellado | hey rossella_s, sure! | 14:38 |
dmellado | that'd be awesome | 14:39 |
rossella_s | cool now it's a bit too early but we will keep you in the loop once we have something celebdor dmellado | 14:43 |
dmellado | rossella_s: sure, thanks! | 14:44 |
dmellado | rossella_s: will you be at the summit? maybe we can get to chat about this there f2f | 14:46 |
celebdor | thanks rossella_s | 14:54 |
* celebdor -> picking up kids from school | 14:54 | |
dulek | ltomasbo: Hey, can you elaborate briefly why we cannot achieve namespace isolation by editing VIP SG? | 14:59 |
ltomasbo | dulek, we cannot edit VIP sg | 15:01 |
ltomasbo | dulek, problem is that even if the loadbalancer is created by tenant 'kuryr', the SG belongs to admin tenant (or service tenant on tripleO deployments) | 15:02 |
rossella_s | dmellado, yes I will be there | 15:02 |
dmellado | rossella_s: cool, let's save some time to have a chat re: $topic on the forum | 15:03 |
rossella_s | sure | 15:03 |
dulek | ltomasbo: So why don't remove the Octavia SG from VIP and add our own one? | 15:07 |
ltomasbo | dulek, because lb resources don't belong to the tenant | 15:07 |
ltomasbo | dulek, so, for instance, we cannot even find the SG attached, as it is not visible to the tenant | 15:08 |
dulek | ltomasbo: VIP does, from what Octavia folks told me. | 15:08 |
ltomasbo | VIP port? | 15:08 |
ltomasbo | or SG? | 15:08 |
*** irenab has quit IRC | 15:09 | |
dulek | ltomasbo: VIP port. | 15:10 |
dulek | ltomasbo: If it's tenant's (because otherwise it would be impossible to attach FIP), then we can clean the SG's on it and add our own. | 15:10 |
dulek | I'm definitely missing something, don't I? | 15:10 |
*** irenab has joined #openstack-kuryr | 15:10 | |
ltomasbo | well, VIP has a security group attached, that belongs to admin | 15:11 |
ltomasbo | I guess we could remove it (let me check) | 15:11 |
dulek | ltomasbo: It should be possible, yes. | 15:11 |
ltomasbo | ok, seems like we may have a possible workaround until they implement a proper way | 15:12 |
*** itzikb is now known as itzikb|afk | 15:13 | |
ltomasbo | dulek, nop | 15:14 |
ltomasbo | it is not working... | 15:14 |
ltomasbo | openstack port unset --security-group 4e9f280f-a1fa-46ef-b9fc-9cf8d84b23f7 fd340d37-c362-4901-8e43-d77157ecc938 | 15:14 |
ltomasbo | No SecurityGroup found for 4e9f280f-a1fa-46ef-b9fc-9cf8d84b23f7 | 15:14 |
ltomasbo | dulek, as the attached security group does not belong to the tenant, we cannot remove it | 15:15 |
ltomasbo | dulek, even though we can see the id by listing the sg associated to the port... | 15:15 |
*** irenab has quit IRC | 15:15 | |
dulek | ltomasbo: Hm, interesting. | 15:15 |
dulek | ltomasbo: What happens if you do --no-security-group? | 15:16 |
ltomasbo | let me see... | 15:16 |
*** irenab has joined #openstack-kuryr | 15:16 | |
ltomasbo | dulek, is tehre such an option? | 15:17 |
dulek | ltomasbo: Aw, it's for port create… | 15:17 |
dulek | ltomasbo: port set --no-security-group. | 15:17 |
ltomasbo | seems to be on the set, yes | 15:17 |
dulek | https://docs.openstack.org/python-openstackclient/pike/cli/command-objects/port.html#port-set | 15:17 |
ltomasbo | seems to have worked... | 15:18 |
dulek | ltomasbo: Does that help at all? :D | 15:18 |
ltomasbo | dulek, I guess then we can create our own | 15:19 |
ltomasbo | dulek, so, I guess at least it gave us some more options... | 15:20 |
dulek | ltomasbo: Does it solve https://storyboard.openstack.org/#!/story/2003686 ? | 15:20 |
ltomasbo | dulek, well, I see it more as a workaround, rather than the actual solution | 15:20 |
ltomasbo | to me, it will help to skip the problem on the kuryr side, but I still think octavia should add the option to better define the listeners | 15:21 |
dulek | ltomasbo: Sure, but given that Octavia folks aren't able to dedicate time to that story, we might need to follow that. | 15:21 |
ltomasbo | sure, I'll try to see if I can modify kuryr to do that | 15:22 |
ltomasbo | dulek, thanks for the tip! | 15:22 |
dulek | ltomasbo: Awesome! All credit goes to johnsom and cgoncalves who proposed that when we were sipping beer yesterday. :) | 15:23 |
dulek | ltomasbo: Maybe we can even backport that. | 15:23 |
ltomasbo | :D | 15:23 |
* dulek moves to the conference rooms. | 15:24 | |
*** irenab has quit IRC | 15:25 | |
*** irenab has joined #openstack-kuryr | 15:26 | |
*** irenab has quit IRC | 15:34 | |
*** irenab has joined #openstack-kuryr | 15:36 | |
*** ccamposr has quit IRC | 15:42 | |
*** celebdor has quit IRC | 15:45 | |
*** pcaruana has quit IRC | 15:55 | |
*** yboaron_ has quit IRC | 16:01 | |
*** garyloug_ has joined #openstack-kuryr | 16:37 | |
*** itzikb|afk has quit IRC | 16:38 | |
*** garyloug has quit IRC | 16:39 | |
*** garyloug_ has quit IRC | 16:45 | |
*** pcaruana has joined #openstack-kuryr | 16:51 | |
*** celebdor has joined #openstack-kuryr | 17:04 | |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Add support for python3 containerized kuryr-kubernetes https://review.openstack.org/599457 | 17:09 |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Fix Python 3 compatiblity in K8sClient.watch() https://review.openstack.org/600116 | 17:09 |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Pass sys.stdin.buffer to json.load() in Python 3 https://review.openstack.org/602099 | 17:09 |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Cleanup cni.main file https://review.openstack.org/602102 | 17:20 |
openstackgerrit | Merged openstack/kuryr-kubernetes master: Add kuryr_daemon_enabled when kuryr-daemon exists https://review.openstack.org/596690 | 17:20 |
*** maysams has quit IRC | 17:47 | |
*** celebdor has quit IRC | 17:47 | |
*** hongbin has joined #openstack-kuryr | 17:56 | |
*** celebdor has joined #openstack-kuryr | 18:10 | |
*** hongbin has quit IRC | 18:17 | |
*** pcaruana has quit IRC | 18:47 | |
celebdor | dulek: how does https://review.openstack.org/#/c/600116/2..3/kuryr_kubernetes/k8s_client.py fix the new issue you found in http://logs.openstack.org/16/600116/2/experimental/kuryr-kubernetes-tempest-daemon-containerized-octavia-py36/ef8f370/controller/logs/screen-kubelet.txt.gz#_Sep_07_13_07_44_125471 | 18:49 |
celebdor | also, why did you remove the depends https://review.openstack.org/#/c/600116/2..3//COMMIT_MSG ? | 18:50 |
*** hongbin has joined #openstack-kuryr | 19:12 | |
*** hongbin has quit IRC | 19:29 | |
dulek | celebdor: It's now depending on your patch directly, like using branches. | 19:42 |
dulek | celebdor: The change you look at wasn't introduced by me. That patch haven't changed, I've put a fix into another commit: https://review.openstack.org/602099 | 19:43 |
dulek | celebdor: Aaaaand… The py36 job is green there! | 19:43 |
dulek | \o/ | 19:44 |
*** pcaruana has joined #openstack-kuryr | 19:58 | |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Add support for python3 containerized kuryr-kubernetes https://review.openstack.org/599457 | 20:10 |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Fix Python 3 compatiblity in K8sClient.watch() https://review.openstack.org/600116 | 20:10 |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Pass sys.stdin.buffer to json.load() in Python 3 https://review.openstack.org/602099 | 20:10 |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Add non-containerized Python 3.6 gate https://review.openstack.org/602150 | 20:10 |
*** hongbin has joined #openstack-kuryr | 20:12 | |
*** pcaruana has quit IRC | 20:20 | |
celebdor | yay! | 20:21 |
celebdor | dulek: this is really great! | 20:23 |
dulek | celebdor: I've added a regular tempest job with Python 3.6 and switched a knob on your to make sure rest of services run in 3.6 (let's see if that works for all of our dependencies). | 20:24 |
celebdor | dulek: which rest of services? | 20:25 |
celebdor | you mean neutron and the like? | 20:25 |
dulek | celebdor: I think so. Such job works fine on Nova, so hopefully it'll work for us as well. | 20:25 |
celebdor | ok | 20:26 |
*** janki has joined #openstack-kuryr | 20:42 | |
*** maysams has joined #openstack-kuryr | 20:43 | |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: Add non-containerized Python 3.6 gate https://review.openstack.org/602150 | 20:44 |
*** hongbin has quit IRC | 20:49 | |
*** hongbin has joined #openstack-kuryr | 20:55 | |
celebdor | dulek: I added some more comments to your gdoc | 21:12 |
dulek | celebdor: I've seen that. Will review it once Octavia folk finish talking about containerized amphorae. | 21:13 |
dulek | celebdor: There's a WIP patch - https://review.openstack.org/#/c/579405/ ! | 21:13 |
celebdor | dulek: the mythical containerized amphora? | 21:13 |
celebdor | still using nova-lxd? | 21:13 |
dulek | celebdor: Nah, it's Zun this time. | 21:14 |
celebdor | oh | 21:14 |
celebdor | that's quite interesting | 21:14 |
celebdor | zun + kuryr-libnetwork? | 21:14 |
*** hongbin has quit IRC | 21:16 | |
dulek | celebdor: Not sure how networking is done here, will ask. | 21:16 |
celebdor | dulek: I know zun can use kuryr-libnetwork, not sure if this is the case here | 21:16 |
dulek | celebdor: Ah yes, it's Kuryr. | 21:16 |
celebdor | dulek: full circle! | 21:17 |
dulek | celebdor: Oh come on, kuryr-libnetwork != kuryr-kubernetes. xD | 21:17 |
celebdor | thankfully | 21:17 |
celebdor | there was a time when we wanted to do it monorepo | 21:18 |
*** celebdor has quit IRC | 21:32 | |
*** hongbin has joined #openstack-kuryr | 21:51 | |
openstackgerrit | Michał Dulko proposed openstack/kuryr-kubernetes master: WIP: Add HA gate https://review.openstack.org/588223 | 21:57 |
*** hongbin has quit IRC | 22:43 | |
*** hongbin has joined #openstack-kuryr | 23:07 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!