| *** salv-orlando has quit IRC | 00:05 | |
| *** david-lyle has joined #openstack-kuryr | 00:08 | |
| *** hongbin has quit IRC | 00:14 | |
| *** david-lyle has quit IRC | 00:15 | |
| *** tianquan has joined #openstack-kuryr | 00:16 | |
| *** tianquan has quit IRC | 00:20 | |
| *** saneax is now known as saneax-_-|AFK | 00:22 | |
| *** limao has joined #openstack-kuryr | 00:23 | |
| *** yedongcan has joined #openstack-kuryr | 00:45 | |
| *** david-lyle has joined #openstack-kuryr | 00:54 | |
| *** david-lyle has quit IRC | 00:54 | |
| *** pmannidi has quit IRC | 01:30 | |
| *** salv-orlando has joined #openstack-kuryr | 01:37 | |
| *** salv-orlando has quit IRC | 01:41 | |
| *** david-lyle has joined #openstack-kuryr | 01:52 | |
| *** tianquan has joined #openstack-kuryr | 02:18 | |
| *** tianquan has quit IRC | 02:22 | |
| *** tianquan has joined #openstack-kuryr | 02:31 | |
| *** salv-orlando has joined #openstack-kuryr | 02:38 | |
| *** salv-orlando has quit IRC | 02:42 | |
| *** hongbin has joined #openstack-kuryr | 03:00 | |
| *** salv-orlando has joined #openstack-kuryr | 03:39 | |
| *** salv-orlando has quit IRC | 03:43 | |
| *** tianquan has quit IRC | 03:48 | |
| *** hongbin has quit IRC | 03:59 | |
| *** salv-orlando has joined #openstack-kuryr | 04:39 | |
| *** salv-orlando has quit IRC | 04:44 | |
| *** tianquan has joined #openstack-kuryr | 04:48 | |
| *** tianquan has quit IRC | 04:53 | |
| *** salv-orlando has joined #openstack-kuryr | 05:40 | |
| *** salv-orlando has quit IRC | 05:45 | |
| *** tianquan has joined #openstack-kuryr | 05:54 | |
| *** saneax-_-|AFK is now known as saneax | 06:03 | |
| *** salv-orlando has joined #openstack-kuryr | 06:33 | |
| *** janki has joined #openstack-kuryr | 07:00 | |
| *** yamamoto has quit IRC | 07:14 | |
| irenab | vikasc: hi | 07:16 |
|---|---|---|
| vikasc | hi irenab | 07:17 |
| vikasc | i pinged you regarding https patch, https://review.openstack.org/#/c/440979/ | 07:17 |
| vikasc | irenab, in config options , ssl_ca_crt_file is being taken as String type because this will be file path | 07:18 |
| irenab | vikasc: I just went though the doc you shared, added few comments. | 07:19 |
| vikasc | irenab, thanks, i will take a look and respond. | 07:22 |
| *** salv-orl_ has joined #openstack-kuryr | 07:22 | |
| irenab | vikasc: checking the patch. My responces today maybe slow, I got sick and not sure how much time will be able to work | 07:24 |
| vikasc | irenab, oh, np .. please take proper rest. | 07:24 |
| irenab | vikasc: thanks | 07:24 |
| *** salv-orlando has quit IRC | 07:25 | |
| irenab | vikasc: regardign the string/False config option. Maybe it should default to None? | 07:25 |
| vikasc | irenab, anyways i am converting string type "false" to boolean type internally. so i can use "None" also. Or would you prefer a seperate config option like "verify_server_crt" True/False? | 07:30 |
| *** tianquan has quit IRC | 07:32 | |
| vikasc | irenab, then to enable https k8s client user will have to set this new parameter also to "True" along with other three file paths, crt,key and ca | 07:32 |
| irenab | vikasc: with None, I think | 07:32 |
| irenab | you don’t even need to mention default value | 07:32 |
| vikasc | irenab, cool | 07:32 |
| irenab | vikasc: maybe additional setting is better, it makes clear what is expected | 07:33 |
| vikasc | irenab, yeah.. i was not sure that should i add one more config parameter or not | 07:33 |
| vikasc | irenab, i too feel it be clearer, will update readme as well | 07:34 |
| irenab | vikasc: check how it is with SSL for kuryr-libnetwork, https://github.com/openstack/kuryr-libnetwork/blob/master/kuryr_libnetwork/config.py#L48 | 07:34 |
| irenab | it has dedicated enable setting | 07:34 |
| vikasc | irenab, but that enable has different meaning than here' | 07:35 |
| *** danil has joined #openstack-kuryr | 07:35 | |
| vikasc | irenab, in kuryr-libnetwork that say "enable ssl secured kuryr server" | 07:35 |
| irenab | vikasc: it has separation between cert settings and flag is apply it or not | 07:35 |
| vikasc | irenab, in kuryr-k8s, it will mean "verify the server certs with CA" | 07:36 |
| irenab | vikasc: the analogy I used is for the separation, not the meaning of the settings. But it can be implicit as well | 07:37 |
| irenab | or even if fail to connect via https, try to fallback | 07:38 |
| vikasc | irenab, sounds good | 07:38 |
| irenab | vikasc: another question. Hoe it works for k8s components? How kubelet is configured? Maybe we need same settings | 07:39 |
| vikasc | irenab, no, actually.. may be user will not want to compromise on security for the sake of connecting | 07:39 |
| irenab | vikasc: here is a matter of cluster deployment. I think kuryr should follow the same as other k8s Controllers/kubelet | 07:40 |
| vikasc | irenab, so i think if user has configured ca file path and it fails.. we should not do failover with "verfiy=False" | 07:40 |
| vikasc | irenab, how will kuryr know kubelet https failover behavior? | 07:43 |
| irenab | vikasc: I think its a deployment issue, once admin setups a cluster, he provides each component with required settings. | 07:44 |
| vikasc | irenab, so lets provide a config option rather doing failover with http implicitly. | 07:44 |
| vikasc | irenab, so that if https fail, admin can change the configuration to "False" which is equivalent to "--insecure" | 07:45 |
| irenab | vikasc: I just suggest to align kuryr settings with what is done for the k8s components. Probably failover is not a good idea, we may need to have some sort of disabling it for the development | 07:46 |
| vikasc | irenab, it can be disabled by setting cert and key file path parameters to None in config file. am i still missing your point? | 07:49 |
| irenab | vikasc: this should be enough | 07:49 |
| irenab | Did you check this: https://kubernetes.io/docs/admin/master-node-communication/#cluster---master | 07:50 |
| vikasc | irenab, just read it. is there a specific point that you want me to pay attention towards? seems nothing is stated about failover explicitly there. | 07:55 |
| irenab | vikasc: I had more generic question, not related to failover. Since secure API connectivity is cluster wide, I asked if you checked for alignment with other k8s components, such as kubelet/kube-proxy that probabably need similar settings, we better have the same. I think eventually we would like to deploy kuryr components as k8s Pods | 07:59 |
| *** tianquan has joined #openstack-kuryr | 07:59 | |
| irenab | its ok if not yet, we can deal with it later | 07:59 |
| *** yedongcan has left #openstack-kuryr | 08:00 | |
| vikasc | irenab, not yet. | 08:00 |
| vikasc | irenab, i will take a look | 08:01 |
| irenab | vikasc: sorry for so many questions, I didn’t check these details too. But since kuryr is just an additional component to the k8s deployment, I think we better understand how the whole system works | 08:03 |
| vikasc | irenab, your questions are all valid and worth asking. I really appreciate! | 08:03 |
| *** janonymous has quit IRC | 08:04 | |
| *** ltomasbo|away is now known as ltomasbo | 08:11 | |
| *** devvesa has joined #openstack-kuryr | 08:12 | |
| *** janonymous has joined #openstack-kuryr | 08:30 | |
| *** irenab has quit IRC | 08:33 | |
| *** irenab_ is now known as irenab | 08:33 | |
| openstackgerrit | Jaivish Kothari(janonymous) proposed openstack/kuryr-libnetwork master: Checking Tests Run https://review.openstack.org/432863 | 08:37 |
| *** irenab_ has joined #openstack-kuryr | 08:42 | |
| *** garyloug has joined #openstack-kuryr | 09:35 | |
| *** tianquan has quit IRC | 09:37 | |
| *** yedongcan1 has joined #openstack-kuryr | 09:43 | |
| *** tianquan has joined #openstack-kuryr | 10:21 | |
| irenab_ | apuimedo: ping | 10:56 |
| irenab_ | the new feature policy is to be added into generic kuryr or kuryr-k8s and kuryr-libnetwork projects? | 10:57 |
| apuimedo | irenab_: I'm tempted to just put it in the wiki and link it from the README of all the repos | 10:58 |
| irenab_ | apuimedo: sounds reasonable | 10:59 |
| *** salv-orl_ has quit IRC | 11:00 | |
| *** salv-orlando has joined #openstack-kuryr | 11:01 | |
| *** salv-orlando has quit IRC | 11:01 | |
| *** salv-orlando has joined #openstack-kuryr | 11:02 | |
| apuimedo | ivc_: please, get https://review.openstack.org/#/c/442866/ in | 11:17 |
| *** yedongcan1 has left #openstack-kuryr | 11:25 | |
| *** yamamoto has joined #openstack-kuryr | 11:29 | |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: requirements: add pyroute2 to the runtime reqs https://review.openstack.org/442867 | 11:33 |
| irenab_ | apuimedo: https://wiki.openstack.org/wiki/Kuryr#Kuryr_Policies | 11:34 |
| apuimedo | irenab_: nice! Thanks! | 12:04 |
| irenab_ | apuimedo: I will add links from the README | 12:04 |
| apuimedo | awesome | 12:05 |
| openstackgerrit | Berezovsky Irena proposed openstack/kuryr-kubernetes master: Add Kuryr development policy link to README file. https://review.openstack.org/443598 | 12:16 |
| *** yamamoto has quit IRC | 12:21 | |
| *** yamamoto has joined #openstack-kuryr | 12:21 | |
| *** limao has quit IRC | 12:22 | |
| *** limao has joined #openstack-kuryr | 12:22 | |
| *** limao has quit IRC | 12:27 | |
| *** salv-orlando has quit IRC | 12:38 | |
| *** garyloug has quit IRC | 12:43 | |
| *** garyloug has joined #openstack-kuryr | 13:05 | |
| *** tianquan has quit IRC | 13:06 | |
| *** yamamoto has quit IRC | 13:07 | |
| *** yamamoto has joined #openstack-kuryr | 13:13 | |
| *** yamamoto has quit IRC | 13:16 | |
| *** yamamoto has joined #openstack-kuryr | 13:17 | |
| *** yamamoto has quit IRC | 13:21 | |
| *** yamamoto has joined #openstack-kuryr | 13:25 | |
| *** salv-orlando has joined #openstack-kuryr | 13:28 | |
| *** danil has quit IRC | 13:30 | |
| *** tianquan has joined #openstack-kuryr | 13:41 | |
| *** limao has joined #openstack-kuryr | 13:44 | |
| *** tianquan has quit IRC | 13:46 | |
| *** limao has quit IRC | 13:53 | |
| *** limao_ has joined #openstack-kuryr | 13:53 | |
| *** limao_ has quit IRC | 13:55 | |
| *** limao has joined #openstack-kuryr | 13:56 | |
| *** janki has quit IRC | 14:01 | |
| *** limao has quit IRC | 14:07 | |
| *** limao has joined #openstack-kuryr | 14:07 | |
| *** limao has quit IRC | 14:18 | |
| *** limao has joined #openstack-kuryr | 14:19 | |
| *** yamamoto has quit IRC | 14:23 | |
| *** limao has quit IRC | 14:29 | |
| *** limao has joined #openstack-kuryr | 14:30 | |
| *** saneax is now known as saneax-_-|AFK | 14:35 | |
| *** limao has quit IRC | 14:41 | |
| *** limao has joined #openstack-kuryr | 14:41 | |
| mchiappero | hey folks | 14:42 |
| mchiappero | plenty of: | 14:42 |
| mchiappero | 2017-03-09 14:34:10.067 16727 DEBUG kuryr_kubernetes.cni.api [-] CNI output: {'msg': 'timeout', 'cniVersion': '0.3.0', 'code': 200} _write_dict /opt/stack/kuryr-kubernetes/kuryr_kubernetes/cni/api.py:101 | 14:42 |
| mchiappero | 2017-03-09 14:34:10.068 16727 DEBUG kuryr_kubernetes.cni.main [-] timed out _timeout /opt/stack/kuryr-kubernetes/kuryr_kubernetes/cni/main.py:90 | 14:42 |
| mchiappero | 2017-03-09 14:34:10.068 16727 INFO kuryr_kubernetes.watcher [-] Stopped watching '/api/v1/namespaces/default/pods?fieldSelector=metadata.name=test-4-server-2' | 14:42 |
| mchiappero | any suggestion on how to debug this? | 14:42 |
| mchiappero | the controller side seems to be clean, no output | 14:43 |
| mchiappero | either with VLAN or MACVLAN it's proving impossible to run more than a handful of containers | 14:43 |
| mchiappero | something is wrong with either the controller or neutron, in neutron I see many orphan ports | 14:45 |
| *** hongbin has joined #openstack-kuryr | 14:52 | |
| *** limao has quit IRC | 14:52 | |
| *** limao has joined #openstack-kuryr | 14:53 | |
| *** limao_ has joined #openstack-kuryr | 15:04 | |
| *** limao has quit IRC | 15:04 | |
| *** tianquan has joined #openstack-kuryr | 15:06 | |
| *** tianquan has quit IRC | 15:12 | |
| *** salv-orlando has quit IRC | 15:14 | |
| *** salv-orlando has joined #openstack-kuryr | 15:15 | |
| apuimedo | mchiappero: ltomasbo is debugging the slowness on the neutron side | 15:19 |
| apuimedo | he found already some interesting things in the flow management | 15:20 |
| apuimedo | (when creating 50 subports it takes several tens of seconds | 15:20 |
| apuimedo | to set up the flows) | 15:20 |
| apuimedo | hongbin: ping | 15:21 |
| hongbin | apuimedo: hey | 15:21 |
| apuimedo | hongbin: hey! | 15:22 |
| apuimedo | hongbin: which distro do you use for kuryr-libnetwork development? | 15:22 |
| hongbin | hongbin: ubuntu | 15:22 |
| apuimedo | xenial? | 15:22 |
| hongbin | yes | 15:22 |
| apuimedo | hongbin: can you check what's the latest available pyroute2 package in the repos? | 15:23 |
| hongbin | apt-get install pyroute2? | 15:23 |
| *** yamamoto has joined #openstack-kuryr | 15:23 | |
| hongbin | apuimedo: i can help for that, just let me know the steps to check | 15:24 |
| apuimedo | hongbin: apt install python-pyroute2 | 15:25 |
| apuimedo | don't confirm | 15:26 |
| apuimedo | just check which version it proposes | 15:26 |
| apuimedo | if it's 0.3.5-1 it's bad | 15:26 |
| hongbin | apuimedo: http://paste.openstack.org/show/602083/ | 15:26 |
| apuimedo | bleh | 15:27 |
| apuimedo | this is gonna be a problem for debian/ubuntu packaging | 15:27 |
| apuimedo | I'll see what can be done | 15:27 |
| apuimedo | thanks hongbin | 15:27 |
| hongbin | apuimedo: np | 15:27 |
| *** limao_ has quit IRC | 15:28 | |
| *** yamamoto has quit IRC | 15:29 | |
| dmellado | apuimedo: I see that you're still lovin' packaging xD | 15:39 |
| apuimedo | dmellado: please, put me out of my misery | 15:39 |
| dmellado | already wanting a quick death? | 15:40 |
| apuimedo | preferable in a non-painful way | 15:40 |
| dmellado | you're youg and have kids! can't do that! | 15:40 |
| dmellado | xD | 15:40 |
| dmellado | young | 15:40 |
| apuimedo | I'm checking the rdo-packaging octavia and neutron examples | 15:40 |
| dmellado | xD | 15:40 |
| apuimedo | dmellado: they'll take vengeance and rid the world of packages | 15:40 |
| dmellado | apuimedo: I guess at least by now you'd be over with the packages IRL | 15:41 |
| dmellado | are you done with on your new house? xD | 15:41 |
| apuimedo | still need to replace the light switches | 15:41 |
| apuimedo | dmellado: do you know ansible? | 15:41 |
| dmellado | yeah, I know it | 15:42 |
| dmellado | it has some stickers with an stylish 'A' on it, isn't it? | 15:42 |
| dmellado | xD | 15:42 |
| apuimedo | yes, that's the one | 15:43 |
| dmellado | so what would you like to do with ansible | 15:43 |
| dmellado | if it gets to the part of some tripleO integration, I'd politely refuse | 15:43 |
| dmellado | xD | 15:43 |
| apuimedo | just knowing how much I can bother you tomorrow | 15:44 |
| apuimedo | dmellado: I know your price | 15:44 |
| dmellado | be aware that I have to wake at a 5 am | 15:44 |
| dmellado | xD | 15:44 |
| apuimedo | you can sleep on the train | 15:45 |
| apuimedo | or better, take the switch with you | 15:45 |
| apuimedo | play | 15:45 |
| apuimedo | and then show it to me | 15:45 |
| apuimedo | :P | 15:45 |
| dmellado | heh, I was thinking actually about bringing it | 15:45 |
| dmellado | xD | 15:45 |
| dmellado | but let's see xD | 15:46 |
| dmellado | I might just bring my kindle so I start reading some stuff | 15:47 |
| dmellado | only to fell asleep at page 3 | 15:47 |
| apuimedo | dmellado: I have a book on discrete mathematics that guarantees deep sleep in three pages | 15:49 |
| dmellado | apuimedo: my girlfriend has decided to get another degree | 15:49 |
| apuimedo | mchiappero: do the vifs get patched on the pods? | 15:49 |
| dmellado | and I'm having to coach her on statistics | 15:49 |
| dmellado | guaranteed sleep too | 15:49 |
| dmellado | xD | 15:49 |
| dmellado | but anyway, I have a few reviews to do too | 15:50 |
| dmellado | including some blueprints | 15:50 |
| dmellado | so I might try gertty | 15:50 |
| apuimedo | never tried it before | 15:51 |
| dmellado | I'll show you tomorrow, in case you like it | 15:51 |
| apuimedo | cool | 15:52 |
| mchiappero | apuimedo: patched? | 15:52 |
| apuimedo | I had my own project | 15:52 |
| apuimedo | mchiappero: yeah, if the metadata for the vif was added | 15:52 |
| mchiappero | as annotation? | 15:52 |
| apuimedo | dmellado: https://github.com/celebdor/perryt | 15:52 |
| apuimedo | mchiappero: right | 15:52 |
| mchiappero | I'll check, I'm trying to run new tests now so I cleaned up | 15:53 |
| apuimedo | mchiappero: very well | 15:56 |
| apuimedo | mchiappero: my suspicion is that it got added, but it never got modified to active | 15:57 |
| apuimedo | due to the ovs problem | 15:57 |
| apuimedo | and as such, CNI timed out | 15:57 |
| mchiappero | apuimedo: I'm forcing the active state in requrest_vif | 16:00 |
| mchiappero | still no joy | 16:00 |
| apuimedo | mchiappero: so in the annotation it shows as active? | 16:01 |
| mchiappero | do you have the command at hand? | 16:02 |
| mchiappero | is it just describe? | 16:04 |
| apuimedo | just describe | 16:04 |
| apuimedo | in json format | 16:04 |
| mchiappero | apuimedo: tnx | 16:08 |
| apuimedo | yw | 16:10 |
| mchiappero | describe do not support json | 16:12 |
| mchiappero | uhm is there any other way? | 16:15 |
| apuimedo | mmm | 16:15 |
| mchiappero | funnily I restarted kubelet and it started | 16:17 |
| mchiappero | (the container) | 16:17 |
| apuimedo | mchiappero: kubectl -o json get pods | 16:17 |
| mchiappero | it's totally random | 16:17 |
| apuimedo | mchiappero: it probably had ran out of retries before | 16:17 |
| apuimedo | and restarting the kubelet made cni be called again | 16:18 |
| *** tianquan has joined #openstack-kuryr | 16:19 | |
| apuimedo | mchiappero: when did it exactly get stuck? | 16:20 |
| mchiappero | not it's no longer working | 16:21 |
| mchiappero | it's always random | 16:21 |
| mchiappero | it doesn't seem to contain any annotation | 16:21 |
| mchiappero | I keep restarting the services all the time | 16:22 |
| mchiappero | sometimes I manage to run some tests | 16:22 |
| mchiappero | sometimes it doesn't even start a single container | 16:22 |
| apuimedo | mchiappero: you are using ivc_'s services patch (you mentioned services) | 16:22 |
| mchiappero | no | 16:23 |
| mchiappero | (have I?) | 16:23 |
| mchiappero | now it started a container but took ages | 16:23 |
| apuimedo | mchiappero: I assume it is devstack with your macvlan patch | 16:24 |
| apuimedo | nothing else | 16:24 |
| apuimedo | right? | 16:24 |
| mchiappero | right | 16:24 |
| mchiappero | basic devstack on the host | 16:24 |
| mchiappero | single VM, devstack in the VM too | 16:24 |
| mchiappero | as per VLAN instructions | 16:25 |
| *** yamamoto has joined #openstack-kuryr | 16:26 | |
| mchiappero | I thing I noticed is that it seems that randomly some macvlan interfaces do not work since the beginning | 16:26 |
| apuimedo | mchiappero: ok | 16:26 |
| apuimedo | mchiappero: what kind of not working? | 16:26 |
| mchiappero | no connectivity | 16:26 |
| mchiappero | sometimes I start a ping | 16:27 |
| mchiappero | and starts working after many seconds | 16:27 |
| apuimedo | mchiappero: isn't that a result of you forcing the active status? | 16:27 |
| apuimedo | which means it probably reutrns to kubelet before the allowed address pairs has been applied | 16:28 |
| mchiappero | but I have horizon open | 16:28 |
| mchiappero | and it shows it's been added | 16:28 |
| mchiappero | I thought the same | 16:28 |
| mchiappero | also I have the Noopfirewall | 16:28 |
| mchiappero | so I'm not sure it makes a difference anyway | 16:29 |
| apuimedo | mchiappero: maybe you can check the logs of the ovs agent | 16:30 |
| apuimedo | it should mention something about the address pair processing, I supposed | 16:30 |
| apuimedo | *suppose | 16:30 |
| *** yamamoto has quit IRC | 16:30 | |
| apuimedo | mchiappero: I thought that's only for sg, not for ip spoofing | 16:30 |
| mchiappero | oh ok | 16:31 |
| mchiappero | anyway, Neutron is broken | 16:32 |
| mchiappero | broken by design | 16:32 |
| mchiappero | you restart the switch | 16:33 |
| mchiappero | and loose you connectivity to the VM | 16:33 |
| mchiappero | I see no annotation | 16:36 |
| mchiappero | btw, any clue on: NetworkPlugin cni failed on the status hook for pod 'test-1-server-1' - Unexpected command output Device "eth0" does not exist | 16:38 |
| mchiappero | ? | 16:38 |
| apuimedo | mchiappero: jlibosva found a couple of bugs on neutron with ltomasbo's investigation | 16:38 |
| apuimedo | also, with the new push instead of pull, the brokenness reduces quite a bit | 16:39 |
| apuimedo | mchiappero: well, that message is that somehow the cni driver didn't create the macvlan device, didn't rename it | 16:39 |
| apuimedo | or something is up with that | 16:39 |
| apuimedo | gotta go pick up my son | 16:39 |
| apuimedo | I'll be online later | 16:39 |
| mchiappero | ok | 16:40 |
| ltomasbo | mchiappero, does the VM you created have a eth0? | 16:40 |
| mchiappero | I understand it's harmless | 16:40 |
| ltomasbo | or it is ens... something? | 16:40 |
| mchiappero | no | 16:40 |
| mchiappero | ens3 | 16:40 |
| ltomasbo | that is the problem (probably) | 16:40 |
| ltomasbo | you need to configure kuryr | 16:40 |
| ltomasbo | to indicate the VM nic is ens3 instead | 16:40 |
| mchiappero | it's already that way | 16:41 |
| ltomasbo | umm, seems it is not being read then, as it is trying eth0 | 16:41 |
| mchiappero | I don't think that's the problem | 16:41 |
| mchiappero | it works anyway | 16:41 |
| mchiappero | and picks the master right | 16:42 |
| ltomasbo | umm, so why it tries eth0 then? | 16:42 |
| mchiappero | I don't know :D | 16:43 |
| *** tianquan has quit IRC | 16:45 | |
| openstackgerrit | Hongbin Lu proposed openstack/kuryr-libnetwork master: Support creating from existing subnetpool https://review.openstack.org/426595 | 17:03 |
| openstackgerrit | Hongbin Lu proposed openstack/kuryr-libnetwork master: Support creating network with pool uuid https://review.openstack.org/424889 | 17:04 |
| dmellado | that kinda recalls me some weird vagrant bug with some OS's xD | 17:06 |
| dmellado | when it happened exactly the same thing | 17:06 |
| *** devvesa has quit IRC | 17:07 | |
| mchiappero | the bug is openstack | 17:11 |
| mchiappero | Gateway Timeout | 17:11 |
| mchiappero | The gateway did not receive a timely response from the upstream server or application. | 17:11 |
| mchiappero | you try to use horizon and then it timeouts | 17:11 |
| mchiappero | uhm I suspect the allowed_address_pairs had leftovers | 17:16 |
| mchiappero | ltomasbo: have you ever seen the "Unexpected command output Device "eth0" does not exist." message? | 17:18 |
| ltomasbo | no, I don't remember seen that one | 17:18 |
| mchiappero | uhm ok | 17:19 |
| mchiappero | later I'll test again the VLAN config | 17:19 |
| mchiappero | and check | 17:19 |
| ltomasbo | ok | 17:19 |
| mchiappero | uhm I often see neutron-server going to 100% | 17:23 |
| ltomasbo | ahh, that I see it too | 17:24 |
| ltomasbo | 100% of a core, right> | 17:24 |
| ltomasbo | ? | 17:24 |
| mchiappero | yes | 17:25 |
| mchiappero | ok, when do we start rewriting neutron? :) | 17:26 |
| mchiappero | I'm in | 17:26 |
| ltomasbo | :D | 17:26 |
| mchiappero | :D | 17:26 |
| *** yamamoto has joined #openstack-kuryr | 17:26 | |
| ltomasbo | we did some initial research on the time for creating 50 ports | 17:26 |
| ltomasbo | and some times it spends as much as 26 seconds just getting the device_settings from the server | 17:27 |
| mchiappero | cool! | 17:27 |
| mchiappero | but I don't understand | 17:27 |
| ltomasbo | but we saw a few point where we believe it may be improved | 17:27 |
| mchiappero | what is that? Is it python? unnecessary trendy decoupling? | 17:28 |
| mchiappero | bad code? bad algorithms? | 17:28 |
| ltomasbo | I'm going to test with ovs 2.6 instead of 2.5, as it seems there are some fixed that may help there | 17:28 |
| mchiappero | ok, let us know then :) | 17:28 |
| ltomasbo | getting some failures right now compiling the new ovs version, but I will compare results asap | 17:29 |
| *** yamamoto has quit IRC | 17:32 | |
| mchiappero | I'm afraid the allowed_address_pairs is the problem | 17:41 |
| mchiappero | not within kuryr, but probably the way it's implemented in neutron/ovs | 17:41 |
| mchiappero | it could well be the issue you saw | 17:44 |
| mchiappero | I'm afraid that with VMs is hard to see since they take a long time to start | 17:44 |
| mchiappero | with containers is different | 17:44 |
| ltomasbo | yep, could be | 17:47 |
| ltomasbo | I'm not using allowed_address_pairs anyway | 17:47 |
| ltomasbo | but agree with you with the different booting up speed of VMs and containers | 17:48 |
| *** mattmceuen has joined #openstack-kuryr | 17:48 | |
| ltomasbo | specially the amount of them happending at the same time | 17:48 |
| ltomasbo | it works good with a few containers at the same time | 17:48 |
| ltomasbo | but when I go over 10, things start to get randomly bad | 17:48 |
| ltomasbo | just to be clear, the case I was evaluating was not even the nested one, just the generic vif | 17:49 |
| mchiappero | to me it's about 6/8 containers along with 2/3 VMs | 17:50 |
| mchiappero | It's a bit of a concern though :/ | 17:51 |
| mchiappero | BTW, I'm sorry for being a bit over, but i've been spending days trying to get something working | 17:52 |
| mchiappero | it's super frustrating | 17:53 |
| mchiappero | and I have tons of other things to do | 17:53 |
| mchiappero | yeah, when I try to clean up after creating many containers I see the neutron ports getting remove slowly | 17:56 |
| *** garyloug has quit IRC | 18:01 | |
| ltomasbo | mchiappero, I know the feeling... | 18:10 |
| *** ltomasbo is now known as ltomasbo|away | 18:11 | |
| mchiappero | from the behaviour it looks like it's not linear | 18:16 |
| mchiappero | there is some O(N^2) logic | 18:16 |
| *** yamamoto has joined #openstack-kuryr | 18:28 | |
| *** yamamoto has quit IRC | 18:33 | |
| *** david-lyle has quit IRC | 19:14 | |
| *** salv-orl_ has joined #openstack-kuryr | 19:22 | |
| *** salv-orlando has quit IRC | 19:25 | |
| *** yamamoto has joined #openstack-kuryr | 19:30 | |
| *** yamamoto has quit IRC | 19:35 | |
| *** salv-orl_ has quit IRC | 19:37 | |
| *** salv-orlando has joined #openstack-kuryr | 19:38 | |
| *** salv-orlando has quit IRC | 20:08 | |
| *** yamamoto has joined #openstack-kuryr | 20:31 | |
| *** yamamoto has quit IRC | 20:35 | |
| *** salv-orlando has joined #openstack-kuryr | 21:09 | |
| apuimedo | vikasc: you up? | 21:29 |
| *** yamamoto has joined #openstack-kuryr | 21:32 | |
| *** yamamoto has quit IRC | 21:38 | |
| *** tianquan has joined #openstack-kuryr | 21:50 | |
| *** aojea has joined #openstack-kuryr | 21:54 | |
| *** tianquan has quit IRC | 22:41 | |
| *** aojea has quit IRC | 23:05 | |
| *** saneax-_-|AFK is now known as saneax | 23:25 | |
| *** yamamoto_ has joined #openstack-kuryr | 23:35 | |
| *** yamamoto_ has quit IRC | 23:40 | |
| *** pmannidi has joined #openstack-kuryr | 23:44 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!