Wednesday, 2017-02-08

« Tuesday, 2017-02-07 Index Thursday, 2017-02-09 »
*** salv-orlando has quit IRC00:11
*** hongbin has quit IRC00:24
*** saneax-_-|AFK is now known as saneax00:24
*** limao has joined #openstack-kuryr00:26
apuimedoivc_: ping00:38
*** tonanhngo has quit IRC00:57
ivc_apuimedo pong01:04
*** yedongcan has joined #openstack-kuryr01:16
*** tonanhngo has joined #openstack-kuryr01:17
*** tonanhngo has quit IRC01:22
*** yamamoto has quit IRC01:34
*** yamamoto has joined #openstack-kuryr01:38
*** yamamoto has quit IRC01:43
*** saneax is now known as saneax-_-|AFK02:10
*** yedongcan_ has joined #openstack-kuryr02:16
openstackgerritCao Xuan Hoang proposed openstack/kuryr master: Remove a py34 environment from tox  https://review.openstack.org/41614202:20
*** yedongcan_ has quit IRC02:31
openstackgerritCao Xuan Hoang proposed openstack/kuryr master: Remove support for py34  https://review.openstack.org/41614202:38
*** yamamoto has joined #openstack-kuryr02:40
*** yamamoto has quit IRC03:01
*** hongbin has joined #openstack-kuryr03:51
*** saneax-_-|AFK is now known as saneax03:58
*** limao has quit IRC04:18
*** limao has joined #openstack-kuryr04:18
*** hongbin has quit IRC04:27
*** yedongcan1 has joined #openstack-kuryr05:09
*** janki has joined #openstack-kuryr05:12
*** yedongcan has quit IRC05:12
*** tonanhngo has joined #openstack-kuryr05:25
*** salv-orlando has joined #openstack-kuryr05:45
*** salv-orlando has quit IRC05:50
openstackgerritMerged openstack/kuryr-kubernetes master: Resolve 'resourceVersion' conflicts  https://review.openstack.org/42294606:25
*** salv-orlando has joined #openstack-kuryr06:46
*** salv-orlando has quit IRC06:51
*** salv-orlando has joined #openstack-kuryr07:02
*** salv-orl_ has joined #openstack-kuryr07:05
*** salv-orlando has quit IRC07:09
*** tonanhngo has quit IRC07:25
*** pcaruana has joined #openstack-kuryr07:29
irenabapuimedo, hi07:29
*** david-lyle_ has joined #openstack-kuryr07:37
*** devvesa has joined #openstack-kuryr07:38
*** devvesa has quit IRC07:40
apuimedoivc_: For the life of me, I don't remember what I wanted last night07:47
apuimedoirenab: hi07:47
*** salv-orlando has joined #openstack-kuryr07:50
*** salv-orl_ has quit IRC07:52
*** yamamoto has joined #openstack-kuryr07:56
*** yamamoto has quit IRC07:56
*** yedongcan has joined #openstack-kuryr07:59
*** yedongcan1 has quit IRC07:59
*** tonanhngo has joined #openstack-kuryr08:03
*** tonanhngo has quit IRC08:04
*** salv-orlando has quit IRC08:10
openstackgerritMerged openstack/fuxi master: Enable Fuxi to use Manila share  https://review.openstack.org/37545208:10
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-libnetwork master: Moving from device_owner to tagging  https://review.openstack.org/42577208:10
apuimedoyedongcan: did you see my question on https://review.openstack.org/#/c/428002/08:14
apuimedo?08:14
yedongcanapuimedo: I had replied.08:16
* apuimedo going to look08:16
yedongcanapuimedo: we need "tag-ext" in show_extension, but in app we need "tag_ext".08:17
yedongcanapuimedo: thanks.08:17
openstackgerritMerged openstack/kuryr master: Remove support for py34  https://review.openstack.org/41614208:19
apuimedoGot it Thanks08:19
* apuimedo merging08:19
openstackgerritMerged openstack/fuxi master: Import _ from i18n.py  https://review.openstack.org/42966408:20
openstackgerritMerged openstack/fuxi master: Sync .gitignore from kuryr-libnetwork  https://review.openstack.org/41466008:20
*** yamamoto has joined #openstack-kuryr08:20
openstackgerritMerged openstack/kuryr-libnetwork master: Use Neutron tag-ext extension  https://review.openstack.org/42800208:24
*** yamamoto has quit IRC08:26
*** david-lyle_ has quit IRC08:28
*** ltomasbo is now known as ltomasbo|away08:44
*** ltomasbo|away is now known as ltomasbo08:44
*** salv-orlando has joined #openstack-kuryr08:57
openstackgerritdengshaolin proposed openstack/fuxi master: Add parameter 'region_name' for novaclient  https://review.openstack.org/43065809:00
*** garyloug has joined #openstack-kuryr09:01
yedongcanltomasbo: ping09:08
ltomasboyedongcan, hi!09:08
yedongcanltomasbo:  just review your patch: https://review.openstack.org/#/c/425772/, nice work.09:09
ltomasbothanks!09:09
yedongcanltomasbo: just one question, after port created, what's the attribute of device_owner?09:10
ltomasbonothing, is not set09:10
ltomasboif later the port is used as a subport associated to a trunk, it will have trunk:subport09:11
yedongcanokay, is there a bug for the inconsistencies attached?09:11
ltomasbodo you think we should include something?09:11
ltomasboI filled this one:https://bugs.launchpad.net/neutron/+bug/165744109:12
openstackLaunchpad bug 1657441 in neutron "Remove the set device_owner when attaching subports" [Undecided,New]09:12
yedongcanThanks, I'm ok with nothing09:12
ltomasboI hit that problem for the nested kuryr case09:12
ltomasbowhere, add_subport was actually changing the device_owner, and then kuryr was not removing the ports after container deletion09:13
ltomasboas the device_owner was not kuryr:container09:13
ltomasbothere were some discussions also here: https://review.openstack.org/#/c/419028/309:14
yedongcanok, will check. doesn't it support update port's device_owner?09:14
ltomasboyes, we were actually triggering update_port09:16
ltomasboafter add_subport09:16
ltomasbowith that aim09:16
ltomasbobut (as stated in the bug) it seems add_subport also calls update_port09:16
ltomasbodue to the patch I wanted to revert in neutron09:16
ltomasboand even if we trigger our update_port after the add_subport09:17
ltomasboit is (some times) executed in the wrong order09:17
ltomasboso, the update_port that applies is the one inside add_subport, leaving the device_owner to trunk:subport09:18
ltomasboinstead of kuryr:container09:18
apuimedoirenab: http://bofh.nikhef.nl/events/FOSDEM/2017/H.2215/weave_net_npc.mp4 weave talk on performing network policy for kubernetes (haven't seen it yet)09:18
ltomasbocould be fixed by waiting, but as we don't reallly needed, it is better to not delay the container booting time (IMO)09:18
yedongcanokay, got it. Seems get tag is the proper way.09:18
ltomasbo:D09:19
yedongcanltomasbo: thanks.09:19
apuimedoltomasbo: Delaying container boot time is against my religion09:22
apuimedo:-)09:22
ltomasbolol09:22
*** limao has quit IRC09:24
irenabapuimedo, thanks09:29
irenabapuimedo, regarding the meeting tomorrow. Any chance to make it 30 mins earlier?09:30
apuimedommm09:30
apuimedoivc_: vikasc: ltomasbo: does 30min earlier work for you?09:30
irenabI may have hard stop after 30 mins09:30
vikascapuimedo, yep09:31
ltomasbowould it work 30 minutes later?09:31
ltomasboor 1 hour before?09:31
apuimedoltomasbo: https://review.openstack.org/#/c/425772/ should also be done for kuryr-k8s, shouldn't it?09:32
apuimedoltomasbo: you mean at 13:30?09:32
apuimedocet09:32
ltomasbo15 CET would be perfect09:33
*** openstackgerrit has quit IRC09:34
ltomasbobut I can handle 14 CET too (if works better for all)09:34
irenab1 hour before yes, but not 30  mins later09:34
ltomasboso, let's go for 14 CET09:34
irenabgreat, thanks09:35
ltomasboapuimedo, regarding the change for kuryr-k8s, yes, agree09:35
ltomasboI'll work on that later too09:35
irenabapuimedo, https://docs.google.com/document/d/1TW3P4c8auWwYy-w_5afIPDcGNLK3LZf0m14943eVfVg/edit?ts=58877ea709:36
irenabproposal for multiple networks support by k8s09:36
apuimedothanks09:37
*** vikasc has quit IRC09:42
*** dims_ has joined #openstack-kuryr09:44
*** dims has quit IRC09:45
*** devvesa has joined #openstack-kuryr09:47
*** vikasc has joined #openstack-kuryr09:54
*** yedongcan has left #openstack-kuryr10:08
*** yamamoto has joined #openstack-kuryr10:17
*** yamamoto has quit IRC10:22
*** salv-orlando has quit IRC10:49
*** garyloug has quit IRC11:05
*** garyloug has joined #openstack-kuryr11:05
ltomasboirenab, ping11:12
*** yamamoto has joined #openstack-kuryr11:20
*** vikasc has quit IRC11:26
*** yamamoto has quit IRC11:26
*** limao has joined #openstack-kuryr11:33
*** salv-orlando has joined #openstack-kuryr11:33
*** vikasc has joined #openstack-kuryr11:43
*** limao has quit IRC11:50
irenabltomasbo: hi11:50
*** limao has joined #openstack-kuryr11:50
ltomasbohi11:53
ltomasboirenab, I saw your comments on the device_owner patch11:54
ltomasbothe comment about the tag_ext extension11:54
ltomasbodoes it means the API is changed?11:54
ltomasboI didn't think so11:54
*** limao has quit IRC11:55
ltomasboand the other question is about leaving device_owner, I don't have a strong opinion there, but there may be collisions with other components using this field (as the trunk port)11:56
ltomasboso, if we are not going to use it at all, why keep it?11:57
irenabltomasbo: trunk port is a bit unique case, and I am not sure why they set device owner, but for the common bare metal case, device owner should have nova equivalent12:05
ltomasboseems for baremetal (ironic) they are also using this12:06
ltomasboas well as for nova when booting a VM12:06
ltomasbothen again, I don't have a strong opinion here, and I don't mind to include it. Though it can break things if other people is relying on that (which they should not)12:07
irenabltomasbo: I also do not have strong opinion on this, just seems proper if the one who creates the port will set the owner12:09
irenabRegarding the extensions, check here https://review.openstack.org/#/c/428002/12:09
ltomasboirenab, I agree with you on that. But armax suggested to not use device_owner12:11
ltomasboand regarding the extension12:11
ltomasbothat is to check if the support for subnet/router tagging is available12:11
ltomasbobut then, the actual tagging happens with the normal neutron add tag API12:11
irenabltomasbo: you are correct about the extension, just somehow missed the check if the ext of tag is supported12:16
irenabregarding the device_owner, I think the advise was not to count on the value to do some logic based on it. Seems there is no proper definition on what is expected from that field12:18
irenabbut I do not have strong opinion if we should keep it, I tend to keep12:19
irenabltomasbo: check here about the device_owner field http://www.innervoice.in/blogs/2015/07/05/ports-in-openstack-neutron/12:25
irenablooks like to be aligned we should set something like compute:kuryr12:26
ltomasboirenab, ok, I'll keep it and modify it to compute:kuryr12:31
*** salv-orlando has quit IRC12:34
*** yamamoto has joined #openstack-kuryr12:35
*** dougbtv has joined #openstack-kuryr12:44
*** yamamoto has quit IRC12:48
*** devvesa has quit IRC13:00
apuimedoalraddarla: ping13:11
*** garyloug has quit IRC13:17
*** salv-orlando has joined #openstack-kuryr13:19
*** saneax is now known as saneax-_-|AFK13:32
*** roeyc has joined #openstack-kuryr13:33
*** janki has quit IRC13:38
*** devvesa has joined #openstack-kuryr13:40
*** janki has joined #openstack-kuryr13:46
*** mchiappe1o is now known as mchiappero13:48
*** yamamoto has joined #openstack-kuryr13:49
*** salv-orl_ has joined #openstack-kuryr13:50
*** salv-orlando has quit IRC13:52
*** yamamoto has quit IRC13:55
*** roeyc has quit IRC13:56
*** limao has joined #openstack-kuryr14:02
alraddarlaapuimedo, hi14:05
*** limao_ has joined #openstack-kuryr14:05
*** limao has quit IRC14:06
apuimedoalraddarla: nevermind. It was one thing about your patch, but I already found out :-)14:09
alraddarlaapuimedo, okay :)14:14
alraddarlaNow that https://review.openstack.org/#/c/426644/ is merged I can move the doc to libnetwork?14:14
apuimedoalraddarla: yes please!14:16
apuimedo:-)14:16
alraddarlajust that file?14:16
apuimedoalraddarla: any devref you see for kuryr-libnetwork14:18
alraddarlaOkay! There is also a k8 devref...do you want that moved to kuryr-kuberentes while I'm at it?14:19
alraddarlaapuimedo, https://github.com/openstack/kuryr/blob/master/doc/source/devref/k8s_api_watcher_design.rst   link for reference14:21
*** limao_ has quit IRC14:23
*** limao has joined #openstack-kuryr14:24
apuimedoalraddarla: interesting, we have to see if it conflicts 100% with the one in kuryr-kubernetes or if it adds something14:24
alraddarlaapuimedo, I'll look into it14:24
*** garyloug has joined #openstack-kuryr14:32
apuimedothanks alraddarla14:33
*** devvesa has quit IRC14:40
*** limao has quit IRC14:45
*** limao has joined #openstack-kuryr14:45
apuimedoivc_: irenab: ltomasbo: vikasc: so what about tomorrow 12 utc?14:50
ivc_apuimedo i'm ok with that14:51
ltomasbothat is 13 CET?14:51
apuimedoltomasbo: indeed14:51
ltomasboworks for me!14:51
apuimedook14:53
*** tonanhngo has joined #openstack-kuryr15:01
*** tonanhngo has quit IRC15:02
*** limao has quit IRC15:06
*** limao has joined #openstack-kuryr15:06
*** saneax-_-|AFK is now known as saneax15:25
*** hongbin has joined #openstack-kuryr15:26
*** limao_ has joined #openstack-kuryr15:27
*** limao has quit IRC15:30
*** saneax is now known as saneax-_-|AFK15:40
*** limao_ has quit IRC15:45
*** janki has quit IRC15:49
*** limao has joined #openstack-kuryr15:51
*** limao has quit IRC16:12
*** limao has joined #openstack-kuryr16:13
*** salv-orl_ has quit IRC16:19
ltomasboapuimedo: I was just checking the use of device_owner in kuryr-k8s16:21
ltomasboit is just set but not used anywhere16:21
apuimedoand?16:21
ltomasboso I think we don't need to modify anything there16:21
apuimedoltomasbo: not even for nested?16:21
ltomasbonop16:21
ltomasboI mean, we set it, but we only used it before in kuryr-libnetwork to remove the port16:22
ltomasbobut here, we already have the vif.id, so no need to search for the port to delete16:23
apuimedotrue16:25
apuimedoltomasbo: but the fact is that there's no kuryr mark on subports then16:25
ivc_ltomasbo i was actually considering device_owner for (now lacking) cleanup job16:25
apuimedoivc_: so then we should port over ltomasbo's libnetwork work16:25
*** saneax-_-|AFK is now known as saneax16:26
ltomasboI will not remove device_owner anyway16:26
ltomasboso, it is still there16:26
ltomasbothe only thing is that it is not used inside kuryr later one16:26
ltomasbos/one/on16:26
apuimedobut it will16:26
apuimedo:-)16:26
ivc_apuimedo ltomasbo what libnetwork is using?16:27
ivc_instead of device_owner16:27
apuimedoivc_: Ocata introduced tags for everything16:27
ltomasboI'm going to include tags16:27
ltomasbobut so far, just the fact that port_name = device_id (in kuryr-libnetwork16:27
ivc_apuimedo ltomasbo did you check how tag lookup is implemented behind the scenes on neutron side?16:28
apuimedoivc_: you mean filtering by tag?16:28
ivc_apuimedo yup16:28
apuimedoI did not16:28
ltomasboI think it is not yet there16:29
ivc_is the db query indexed or do they simply do "tag like '%TAG_TEXT%'"?16:29
ivc_if its the latter, than its pretty bad16:30
apuimedoI'd bet it's the latter16:30
*** limao has quit IRC16:30
apuimedobut something that could be improved and likely to be accepted16:30
*** limao has joined #openstack-kuryr16:30
ivc_apuimedo i was thinking that we use another approach16:31
ltomasbohttps://review.openstack.org/#/c/429621/16:31
ivc_s/use/could use/16:31
ivc_we could get trunk port by device_owner/device_id16:32
ivc_and that would include the collection of subports16:32
ivc_so the only question if we know device_id of a trunk16:32
ltomasboand what about the baremetal case?16:33
ltomasbohttps://github.com/openstack/neutron/blob/master/neutron/db/tag_db.py16:34
apuimedoltomasbo: is the trunk port ownership changed? Or only the subports16:34
apuimedo?16:34
ivc_the only problem with device_owner is with sub-ports16:34
ltomasbowell, the only we are aware of16:34
ltomasboand there may be others to come16:35
ltomasboI would prefer avoiding device_owner, as it seems there is no consensus on their usage16:35
ivc_ltomasbo ok, from tag_db.py it seems the lookup should be efficient (can only tell for sure after seeing actual db model/query, but even if there are problems, those could be fixed)16:37
ivc_but my point was also about the logic - i.e. we should only delete what we 'own'16:39
ltomasboyes yes, that is already working16:40
ltomasboand there will be a case for tagging and another one without tagging, and from ocata, it should be fine with tagging16:41
*** limao has quit IRC16:42
ivc_ltomasbo i mean i'm not too comfortable with us removing ports based on tags rather than device_owner16:42
ivc_what if a user decides to add a tag that would match ours?16:42
ltomasbofor the kubernetes case, it will not matter16:45
ltomasbofor the libnetwork16:45
ivc_apuimedo suggested that we use libnetwork approach in kuryr-k8s16:46
ltomasboit needs to match subnet_cidr, ip_address and tag name16:46
*** salv-orlando has joined #openstack-kuryr16:54
*** tonanhngo has joined #openstack-kuryr16:56
*** tonanhngo has quit IRC16:56
*** ivc_ has quit IRC17:36
*** ivc_ has joined #openstack-kuryr17:38
*** openstackgerrit has joined #openstack-kuryr17:39
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-libnetwork master: Moving from device_owner to tagging  https://review.openstack.org/42577217:39
openstackgerritOpenStack Proposal Bot proposed openstack/fuxi master: Updated from global requirements  https://review.openstack.org/43108517:47
*** ltomasbo is now known as ltomasbo|away17:52
*** roeyc has joined #openstack-kuryr17:54
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr master: Change DEVICE_OWNER to make it more Neutron compliant  https://review.openstack.org/43109117:55
*** portdirect has quit IRC18:35
*** janonymous has quit IRC18:35
*** sridhar_ram has quit IRC18:45
*** tonanhngo has joined #openstack-kuryr18:45
*** tonanhngo has quit IRC18:47
*** tonanhngo has joined #openstack-kuryr18:48
*** garyloug has quit IRC19:13
*** roeyc has quit IRC19:15
*** saneax is now known as saneax-_-|AFK19:15
*** roeyc has joined #openstack-kuryr19:16
*** roeyc has quit IRC19:29
*** sridhar_ram has joined #openstack-kuryr19:39
*** roeyc has joined #openstack-kuryr19:42
*** salv-orlando has quit IRC19:48
*** tonanhngo has quit IRC19:54
*** portdirect has joined #openstack-kuryr19:57
*** roeyc has quit IRC20:03
*** tonanhngo has joined #openstack-kuryr20:08
*** salv-orlando has joined #openstack-kuryr20:11
*** janonymous has joined #openstack-kuryr20:14
*** pcaruana has quit IRC20:43
*** salv-orlando has quit IRC20:56
*** salv-orlando has joined #openstack-kuryr21:40
*** yamamoto has joined #openstack-kuryr22:10
*** dougbtv has quit IRC22:40
*** tonanhngo has quit IRC23:12
*** ajo has quit IRC23:26
*** ajo has joined #openstack-kuryr23:27
« Tuesday, 2017-02-07 Index Thursday, 2017-02-09 »

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!