Tuesday, 2017-10-10

« Monday, 2017-10-09 Index Wednesday, 2017-10-11 »
*** gouthamr has joined #openstack-kuryr00:21
*** caowei has joined #openstack-kuryr01:04
*** kiennt26 has joined #openstack-kuryr01:43
*** c00281451 has joined #openstack-kuryr01:56
*** hongbin has quit IRC02:11
*** wangbo_ has joined #openstack-kuryr02:42
*** kiennt26 has quit IRC02:49
*** kiennt26 has joined #openstack-kuryr02:59
*** premsankar has quit IRC03:28
*** kiennt26 has quit IRC03:31
*** hongbin has joined #openstack-kuryr03:32
*** kiennt26 has joined #openstack-kuryr03:32
*** caowei has left #openstack-kuryr04:08
*** gouthamr has quit IRC04:14
*** wangbo_ has quit IRC04:28
*** pcaruana has joined #openstack-kuryr04:32
*** oikiki has quit IRC04:35
*** kiennt26 has quit IRC04:36
*** hongbin has quit IRC04:40
*** pcaruana has quit IRC04:58
*** wangbo_ has joined #openstack-kuryr05:01
*** pmannidi has quit IRC05:08
*** janki has joined #openstack-kuryr05:10
*** pmannidi has joined #openstack-kuryr05:25
irenabapuimedo: related to the question from yesterday: https://stackoverflow.com/questions/28505305/changing-pods-labels-on-the-fly05:37
*** caowei has joined #openstack-kuryr05:40
apuimedoirenab: https://review.openstack.org/#/c/498698/2205:43
apuimedoI answered your question05:43
apuimedoirenab: pffff, that stackoverflow answer means a lot of trouble05:44
irenabapuimedo: how is the socket file permissions set?05:44
apuimedoirenab: it should be deployment05:44
irenabI somehow miss this part05:44
apuimedobut usually it is owned by the user that created it, so05:44
apuimedoif the controller runs as root05:44
irenabmaybe worth to document it05:45
apuimedoonly root can write to it05:45
apuimedosimilar ot the docker.sock05:45
irenabso it has nothing to do with keystone at all, right?05:45
irenabwho is allowed to call the manager?05:46
irenabshould it be similar to get proper env settings as with calling openstack cli?05:46
apuimedoirenab: nope05:51
apuimedono relation to keystone05:51
apuimedono, as I said, it is like with the docker daemon05:51
apuimedoyou handle it with linux file permissions05:51
apuimedowhoever is 'root' in the cluster controllers can interact with it05:52
*** pmannidi has quit IRC05:52
irenabok, got it05:54
openstackgerritMerged openstack/kuryr-kubernetes master: Add devstack ports pool configuration options  https://review.openstack.org/51053205:56
openstackgerritMerged openstack/kuryr-kubernetes master: Migrate to Zuulv3  https://review.openstack.org/50938306:06
*** pmannidi has joined #openstack-kuryr06:09
*** yamamoto has quit IRC06:11
apuimedoltomasbo: irenab: This conversation raises a good question though06:14
apuimedowhen the controller runs containerized...06:14
apuimedoThe way to interact with it via socket file will not be that consumable06:15
ltomasbosorry I didn't realise about the conversation here06:17
ltomasboand yes, if you change the file permissions the manager cannot access the socket06:17
ltomasboso it is handled with that, in the same way as it is done in the cni split06:17
ltomasboapuimedo, I didn't test it with the containerized version, right06:18
*** wangbo_ has quit IRC06:20
*** edisonxiang has quit IRC06:20
*** wangbo_ has joined #openstack-kuryr06:21
openstackgerritMerged openstack/kuryr-libnetwork master: Migrate to Zuulv3  https://review.openstack.org/50935606:22
apuimedoltomasbo: well... It's not just about testing, it's that the approach won't quite work06:27
apuimedoxD06:27
apuimedobecause it may be scheduled to run on any worker node06:27
apuimedoand how do you get then to the socket file06:27
apuimedoyou'll have to do06:27
apuimedokubectl get po06:27
apuimedofind the pod06:27
apuimedoand then do kubectl exec06:27
apuimedoto execute the command inside06:27
apuimedoit's a bit ugly06:27
apuimedobut for the patch, let's keep it as it is06:28
apuimedowe just need to find a better way going forward06:28
*** livelace-link has joined #openstack-kuryr06:32
*** livelace-link has quit IRC06:33
ltomasboapuimedo, can this be part of the controller?06:35
ltomasboapuimedo, I mean, that is a process that is spawn at running time by the controller depend on the driver and config option06:35
*** livelace has joined #openstack-kuryr06:36
ltomasboso, I assume it will be just enabling access to /run/kuryr when creating the controller06:36
*** livelace-link has joined #openstack-kuryr06:38
*** livelace-link has quit IRC06:38
*** kiennt26 has joined #openstack-kuryr06:42
*** livelace-link has joined #openstack-kuryr06:47
dmelladokiennt26: around?07:01
*** pcaruana has joined #openstack-kuryr07:04
*** livelace-link has quit IRC07:09
apuimedoltomasbo: 'access to /run/kuryr' on which machine?07:10
apuimedoxD07:10
ltomasbowherever the controller pod is running07:10
*** karimb has joined #openstack-kuryr07:11
ltomasboit needs modifications, but I guess it should be included into the pod daemon set, but not sure how...07:11
ltomasboperhaps dulek knows better about how we can do this07:12
apuimedoI would not do it in the worker nodes07:12
apuimedoltomasbo: I'm actually considering that maybe the best is to have a CRD in Kubernetes for the pools07:13
apuimedothen you would just do a kubectl create -f mypool07:13
apuimedoand the pool object, if only specifying ports07:13
apuimedosorry07:13
apuimedosize07:13
apuimedoit would mean that the controller has to fulfill it07:14
apuimedoirenab: ltomasbo: dulek: dmellado: ivc_: what do you think ^^ ?07:14
ltomasbolet me process that....:)07:15
dmelladoapuimedo: same here07:15
dmelladoI'm still low on caffeine07:15
apuimedodmellado: I'm up since 6am XD07:15
ltomasboapuimedo, so, this pool object will be all the pools? a specific pool?07:15
ltomasboapuimedo, why?? :D07:16
dmelladoapuimedo: you're no human07:16
dmelladoxD07:16
dmelladoand I know that you'll go take your nap in a while xD07:16
apuimedoltomasbo: baby woke up all peed, so I had to change it and feed it and make it sleep again, and since it took 45 minutes , I was not sleepy anymore07:16
apuimedoltomasbo: can you please address https://review.openstack.org/#/c/509989/ ?07:17
irenabapuimedo: you will miss this time when you are asleep to take cae of a baby and not wit for them to be back home at 6:00 am …07:18
apuimedoirenab: I don't think so... Life is so much better with the old one that already goes to school and sleeps like a marmot07:19
*** c00281451 is now known as zengchen07:19
apuimedothe baby is cute though07:19
irenabwait till he is 1607:19
apuimedoso it compensates it a bit07:19
ltomasboapuimedo, sure, I was reading emails, and other stuff (odl fighting), I'll address your comments asap07:19
apuimedoirenab: maybe he'll be like me and he won't party07:19
irenabapuimedo: :-)07:19
apuimedoltomasbo: did you sort out the connectivity between the two servers?07:20
ltomasbonop07:20
apuimedo:/07:20
apuimedoalso, https://review.openstack.org/#/c/510157/ would be nice to merge07:20
apuimedo(needs ut)07:20
ltomasboactually I'm testing other controllers deployment (as suggested by other guys) and it is not even working07:20
ltomasboand as it seems we need to move to nitrogen, I'm testing with that too07:20
ltomasboapuimedo, I'll do too!07:21
ltomasboas stack.sh takes some time, I can do some ut in between...07:21
ltomasboalso, irenab, can you take a look at this one when you have time: https://review.openstack.org/#/c/509364/07:23
irenabapuimedo: regarding pools CRD, would it be driven by controller or user?07:24
irenabapuimedo: ltomasbo : I wonder if we should change the defualt for devstack on port setting name. It was very usefull and now I always forget to change the default …07:25
ltomasboirenab, sure, actually, even if the default is false07:26
ltomasboit makes sense for devstack to set it to true07:26
ltomasboso, I agree with you07:26
irenabapuimedo: ^^ ?07:27
ltomasboand it is just 1 line change :)07:27
irenabyes, otherwise matching ports to pods requires investigation07:28
irenabregarding the pools manager, I guess we can add the current patch as dev feature. but for production grade, we should have more solid approach, maybe CRD07:30
*** wangbo_ has quit IRC07:31
*** wangbo has joined #openstack-kuryr07:32
dulekltomasbo, apuimedo: I've just catched up with scrollback.07:35
dulekltomasbo, apuimedo: Regarding socket in containerized Kuryr - we can always mount a directory on the host as a volume to place the socket in.07:36
dulekltomasbo, apuimedo: But to be honest - I'm wondering if we don't just need a regular API service?07:37
*** egonzalez has joined #openstack-kuryr07:39
ltomasbodulek, yep, that could be another option07:39
*** kiennt26 has quit IRC07:43
apuimedoirenab: driven by both07:43
*** kiennt26 has joined #openstack-kuryr07:43
openstackgerritMerged openstack/kuryr-kubernetes master: Add exception handling to _get_in_use_ports  https://review.openstack.org/50936407:49
*** pmannidi has quit IRC07:54
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Speed up neutron request_vifs function  https://review.openstack.org/50998907:56
*** garyloug has joined #openstack-kuryr07:58
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add Pool Manager to handle subports  https://review.openstack.org/49869808:10
*** wangbo has quit IRC08:43
*** wangbo has joined #openstack-kuryr08:53
apuimedoirenab: please review https://review.openstack.org/#/c/509989/09:17
*** kiennt26 has quit IRC09:19
irenabok09:21
openstackgerritMerged openstack/kuryr-kubernetes master: Speed up neutron request_vifs function  https://review.openstack.org/50998909:40
openstackgerritMerged openstack/kuryr-kubernetes master: Add Pool Manager to handle subports  https://review.openstack.org/49869809:48
*** kiennt26 has joined #openstack-kuryr09:52
*** kiennt26 has quit IRC09:53
*** phuoc_ has quit IRC10:06
dmelladoapuimedo: irenab ltomasbo meeting?10:31
irenabjoining10:31
dmelladootherwise I'll go get lunch xD10:31
janonymous:D10:31
dmelladoapuimedo: ' I woke up at 6am man ' !10:32
dmelladodon't tell me that now you got asleep xD10:32
ltomasbojoining10:32
irenabdmellado: you maybe right10:32
dmelladoI'll start writing quotes10:33
dmelladoapuimedo: 'I woke up at 6am'10:33
dmelladoltomasbo: 'I'll write a test'10:33
dmelladoxD10:33
dmelladoand so xD10:33
ltomasbojejejeje10:33
dmelladodmellado: 'ZuulV3 will improve this'10:33
ltomasbodmellado, you can set them at the channel topic! :D10:34
dmelladoxD10:34
irenabI think apuimedo made himself confused again with utc/cet10:34
dmelladohold on, I'll call him10:34
janonymous:D10:34
apuimedono, no10:35
apuimedoI'll join10:35
apuimedojust a sec10:35
apuimedoI was finishing one patch10:35
dmelladoapuimedo: yeah, a 'patch'10:35
dmelladoI like 'finishing patches' after lunch too10:35
dmelladoxD10:35
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add list and show pool commands to Pool Manager  https://review.openstack.org/50441010:36
dmelladohttps://sepoles.files.wordpress.com/2015/10/siesta.png10:36
dmelladobtw ltomasbo you did a good thing not coming to the office today10:38
dmelladoalexa's becoming worse than ever10:38
dmellado'alexa'10:38
ltomasboyep10:41
ltomasboI had enough last week...10:41
dmelladoyou and me both xD10:41
ltomasboapuimedo, irenab, I had to press the rebase in this one: https://review.openstack.org/#/c/504410 (so +W was lost)10:44
irenabdone10:44
ltomasbothanks!10:52
*** wangbo has quit IRC10:55
*** wangbo has joined #openstack-kuryr11:08
*** yamamoto has joined #openstack-kuryr11:38
*** yamamoto has quit IRC11:44
*** yamamoto has joined #openstack-kuryr11:44
*** yamamoto has quit IRC11:47
*** yamamoto has joined #openstack-kuryr11:48
*** yamamoto has quit IRC11:49
openstackgerritMerged openstack/kuryr-kubernetes master: Add list and show pool commands to Pool Manager  https://review.openstack.org/50441011:50
*** yamamoto has joined #openstack-kuryr11:55
openstackgerritMichał Dulko proposed openstack/kuryr-kubernetes master: Get rid of code patching in containers  https://review.openstack.org/51085211:57
zengchenhi, guys. will the vPTG be held on schedule? I still can not join in the meeting.12:03
zengchenping apuimedo12:18
*** yamamoto has quit IRC12:27
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr master: Updated from global requirements  https://review.openstack.org/49604812:27
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr-libnetwork master: Updated from global requirements  https://review.openstack.org/50266312:27
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr-kubernetes master: Updated from global requirements  https://review.openstack.org/50978112:27
*** yamamoto has joined #openstack-kuryr12:29
*** yamamoto has quit IRC12:33
*** wangbo has quit IRC12:48
*** kzaitsev1pi has quit IRC13:05
*** kzaitsev_pi has joined #openstack-kuryr13:06
*** lihi has quit IRC13:12
*** leyal has quit IRC13:13
*** dimak has quit IRC13:14
*** oanson has quit IRC13:14
apuimedozengchen: no. no. The fuxi session will most likely be next week13:14
*** irenab has quit IRC13:14
*** irenab has joined #openstack-kuryr13:15
*** dimak has joined #openstack-kuryr13:15
*** leyal has joined #openstack-kuryr13:15
*** oanson has joined #openstack-kuryr13:15
*** lihi has joined #openstack-kuryr13:16
zengchenapuimedo:got it. thanks.13:25
*** caowei has quit IRC13:28
openstackgerritMichał Dulko proposed openstack/kuryr-kubernetes master: Get rid of code patching in containers  https://review.openstack.org/51085213:33
*** karimb has quit IRC13:35
apuimedodulek: have you tried https://review.openstack.org/#/c/510852/ ?13:36
apuimedowe do not install any rootwrap13:36
dulekapuimedo: I've did. Privsep is used by os_vif.13:37
dulekapuimedo: os_vif needs to have root to connect stuff, right?13:37
apuimedodulek: I know it is used by os_vif, I wrote the patch when I encountered the bug of it relying on rootwrap and the fork mode not being configurable13:37
apuimedohttps://review.openstack.org/47648913:38
dulekapuimedo: Hm. So what were your problems with running with the default?13:39
dulekapuimedo: I haven't hit any, but maybe I've omitted a case when testing?13:39
apuimedoIIRC it would just fail to do the vif.plug13:40
dulekapuimedo: So if I was able to ping pods from the host - was it enough to hit the problem?13:41
apuimedomaybe I'm forgetting some catch13:43
apuimedowhat was the environment? baremetal?13:43
*** karimb has joined #openstack-kuryr13:43
dulekapuimedo: My local CentOS VM.13:44
dulekapuimedo: I've just run DevStack in there.13:44
apuimedoI wonder if maybe it would show on pod-in-vm scenario13:44
apuimedonow I don't remember13:44
openstackgerritMerged openstack/kuryr-libnetwork master: Updated from global requirements  https://review.openstack.org/50266313:45
*** gouthamr has joined #openstack-kuryr13:46
dulekapuimedo: Hm… This makes me upset, as I really hate this arbitrary patching. How should I test this before we're sure it's not breaking anything?13:46
apuimedodulek: I understand13:47
apuimedoI would say, if you can't reproduce it in openshift, get rid of the patch13:47
dulek"os_vif.plug() at cni fails with privsep issue when using ovs-firewall" - okay, I don't think I have ovs-firewall configured?13:47
apuimedoaha!13:49
apuimedoxD13:49
dulekapuimedo: But sorry, I don't get how "However, CNI is already run as root. So this makes little sense." is an issue?13:49
dulekapuimedo: I mean doing "sudo sudo command" is equivalent of "sudo command", right?13:50
apuimedodulek: well, it's retarded to use sudo when we're already root13:50
apuimedothe privsep fork mode is much cleaner13:50
dulekapuimedo: I agree, but it shouldn't break things.13:50
dulekapuimedo: And doing sudo as root is IMO cleaner than arbitrary patching when building a container. :P13:51
apuimedodulek: sudo doesn't need to be on the system though13:51
apuimedoyou have it due to devstack13:51
dulekapuimedo: Oh. CNI is run on the host, not inside the container…13:52
dulekapuimedo: That's a place we have no control of… Makes sense.13:52
*** kiennt26 has joined #openstack-kuryr13:53
*** livelace has quit IRC13:53
apuimedoyup13:53
*** hongbin has joined #openstack-kuryr13:54
apuimedodulek: what we should do is redo my oslo.privsep patch13:54
apuimedoand make one to os.vif13:54
apuimedoit's such a drag though13:54
apuimedosome days I'd just drop os.vif xD13:54
dulekapuimedo: I think I should be able to convince dhelmann to merge it.13:54
apuimedodulek: if you manage that I'll buy you a beer13:55
apuimedoI failed miserably at that13:55
dulekapuimedo: I guess discussion happened on IRC and that's why there's not much on review?13:55
apuimedodulek: just a bit13:56
dulekapuimedo: I also wonder why isn't a patch to os_vif enough? Like os_vif should make it possible to set how it will call start().13:56
apuimedonot so much13:56
apuimedowell, I wanted this patch to be merged, then you didn't need to change os_vif13:56
apuimedo:-)13:56
dulekapuimedo: :D13:57
apuimedoif you want to go doug's way, then you need to patch both13:57
dulekOkay, I'll look at it.13:57
openstackgerritDaniel Mellado proposed openstack/kuryr-tempest-plugin master: [WIP] Add scenario test manager  https://review.openstack.org/51089614:04
*** phuoc_ has joined #openstack-kuryr14:04
*** lihi has quit IRC14:05
*** leyal has quit IRC14:06
*** dimak has quit IRC14:06
*** oanson has quit IRC14:06
*** irenab has quit IRC14:06
*** dimak has joined #openstack-kuryr14:07
*** irenab has joined #openstack-kuryr14:07
*** leyal has joined #openstack-kuryr14:07
*** oanson has joined #openstack-kuryr14:07
*** lihi has joined #openstack-kuryr14:08
*** atoth has joined #openstack-kuryr14:11
*** kiennt26 has quit IRC14:22
*** janki has quit IRC14:27
hongbinirenab, apuimedo : ping14:34
apuimedohongbin: pong14:47
*** jgriffith has quit IRC14:48
hongbinapuimedo: hi toni14:48
hongbinapuimedo: want to request your review on the SR-IOV patches: https://review.openstack.org/#/c/500420/ and https://review.openstack.org/#/c/508778/14:48
hongbinapuimedo: i need those patches merged and make a new release of kuryr to continue the kuryr-libnetwork work :)14:49
apuimedohongbin: on it!14:49
hongbinapuimedo: thx14:49
apuimedoI see that zuul has -1 the first14:49
apuimedoboth actually14:50
apuimedodo you know why?14:50
hongbinzuul v3 is unstable now14:50
hongbinbut jenkin has +1, which should be fine14:50
apuimedook. I'll review today14:54
hongbinthx14:54
*** jgriffith_ has joined #openstack-kuryr14:58
*** zengchen has quit IRC15:15
*** gianpietro has joined #openstack-kuryr15:17
*** gianpietro has quit IRC15:17
*** egonzalez has quit IRC15:57
*** pcaruana has quit IRC16:07
*** gouthamr has quit IRC16:10
*** aojea has joined #openstack-kuryr16:35
*** garyloug has quit IRC16:54
*** gouthamr has joined #openstack-kuryr16:56
*** gouthamr has quit IRC17:00
*** gouthamr has joined #openstack-kuryr17:01
dulekapuimedo: Oh well, does http://paste.openstack.org/show/623265/ look familiar to you? I'm getting that in kuryr-daemon container when running without privsep patch.17:03
dulekI guess I'll just port the patch to kuryr-daemon version…17:04
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr master: Updated from global requirements  https://review.openstack.org/49604818:54
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr-kubernetes master: Updated from global requirements  https://review.openstack.org/50978118:54
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr master: Updated from global requirements  https://review.openstack.org/49604819:10
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr-kubernetes master: Updated from global requirements  https://review.openstack.org/50978119:10
*** aojea has quit IRC19:16
*** aojea has joined #openstack-kuryr19:24
*** aojea has quit IRC19:41
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr master: Updated from global requirements  https://review.openstack.org/49604820:22
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr-kubernetes master: Updated from global requirements  https://review.openstack.org/50978120:22
*** gianpietro has joined #openstack-kuryr20:30
*** openstackgerrit has quit IRC20:33
*** openstackgerrit has joined #openstack-kuryr20:50
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr master: Updated from global requirements  https://review.openstack.org/49604820:50
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr-kubernetes master: Updated from global requirements  https://review.openstack.org/50978120:50
*** gouthamr has quit IRC21:06
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr master: Updated from global requirements  https://review.openstack.org/49604821:06
openstackgerritOpenStack Proposal Bot proposed openstack/kuryr-kubernetes master: Updated from global requirements  https://review.openstack.org/50978121:06
*** zarka has joined #openstack-kuryr21:22
*** heeba has joined #openstack-kuryr21:31
*** heeba has quit IRC21:34
*** zarka_ has joined #openstack-kuryr21:45
*** zarka_ has quit IRC21:46
*** zarka has quit IRC21:49
*** zarka has joined #openstack-kuryr21:49
*** zarka has quit IRC21:56
*** zarka has joined #openstack-kuryr22:00
*** oikiki has joined #openstack-kuryr22:06
*** zarka_ has joined #openstack-kuryr22:07
*** zarka has quit IRC22:07
*** zarka has joined #openstack-kuryr22:08
*** zarka has joined #openstack-kuryr22:08
*** zarka has quit IRC22:11
*** oikiki has quit IRC22:11
*** heeba has joined #openstack-kuryr22:12
*** zarka has joined #openstack-kuryr22:12
*** zarka has quit IRC22:14
*** zarka has joined #openstack-kuryr22:15
*** heeba has quit IRC22:15
*** yamamoto has joined #openstack-kuryr22:19
*** zarka has quit IRC22:29
*** zarka has joined #openstack-kuryr22:33
*** zarka_ has joined #openstack-kuryr22:35
*** asd has joined #openstack-kuryr22:38
*** zarka has quit IRC22:38
*** asd has quit IRC22:39
*** zarka_ has quit IRC22:39
*** zarka has joined #openstack-kuryr22:40
*** jgriffith_ has quit IRC22:46
*** zarka has quit IRC22:50
*** zarka has joined #openstack-kuryr22:50
*** oikiki has joined #openstack-kuryr22:51
*** zarka has quit IRC22:54
*** karimb has quit IRC23:03
*** hongbin has quit IRC23:22
*** gouthamr has joined #openstack-kuryr23:40
« Monday, 2017-10-09 Index Wednesday, 2017-10-11 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!