Monday, 2017-04-24

*** limao has joined #openstack-kuryr00:45
*** hongbin has quit IRC01:17
*** yamamoto has quit IRC01:33
*** yamamoto has joined #openstack-kuryr01:33
*** yedongcan has joined #openstack-kuryr01:37
*** yamamoto has quit IRC02:03
*** yamamoto has joined #openstack-kuryr02:28
*** limao has quit IRC02:48
*** limao has joined #openstack-kuryr02:48
*** yamamoto has quit IRC02:50
*** yamamoto has joined #openstack-kuryr02:53
*** yamamoto has quit IRC03:04
*** salv-orlando has joined #openstack-kuryr03:07
*** hongbin has joined #openstack-kuryr03:07
*** salv-orlando has quit IRC03:12
*** aojea has joined #openstack-kuryr03:14
*** aojea has quit IRC03:18
*** limao has quit IRC03:47
*** limao has joined #openstack-kuryr03:49
*** limao has quit IRC03:54
*** yamamoto has joined #openstack-kuryr04:07
*** salv-orlando has joined #openstack-kuryr04:08
*** salv-orlando has quit IRC04:13
*** limao has joined #openstack-kuryr04:18
*** limao has quit IRC04:33
*** jayahn has joined #openstack-kuryr04:36
*** hongbin has quit IRC04:39
*** salv-orlando has joined #openstack-kuryr04:57
*** limao has joined #openstack-kuryr05:00
*** aojea has joined #openstack-kuryr05:07
*** aojea has quit IRC05:45
*** aojea has joined #openstack-kuryr05:47
*** janki has joined #openstack-kuryr05:50
*** aojea has quit IRC05:52
*** kzaitsev_ws has joined #openstack-kuryr06:04
*** pksingh has joined #openstack-kuryr06:20
*** jayahn has quit IRC06:21
*** ltomasbo|away is now known as ltomasbo06:25
*** dimak_ has joined #openstack-kuryr06:26
*** pksingh has quit IRC06:30
*** pmannidi has quit IRC06:36
*** aojea has joined #openstack-kuryr06:48
*** pmannidi has joined #openstack-kuryr06:49
*** salv-orlando has quit IRC06:50
*** aojea has quit IRC06:54
*** pmannidi has quit IRC07:08
*** pmannidi has joined #openstack-kuryr07:21
*** pksingh has joined #openstack-kuryr07:23
*** pcaruana has joined #openstack-kuryr07:24
*** pksingh has quit IRC07:53
*** salv-orlando has joined #openstack-kuryr07:58
*** oanson has quit IRC08:23
*** oanson has joined #openstack-kuryr08:24
*** garyloug has joined #openstack-kuryr08:42
*** aojea has joined #openstack-kuryr08:50
*** aojea has quit IRC08:55
*** limao has quit IRC09:25
*** limao has joined #openstack-kuryr09:27
*** salv-orl_ has joined #openstack-kuryr09:49
*** salv-orlando has quit IRC09:52
*** jchhatbar has joined #openstack-kuryr09:59
*** janki has quit IRC10:00
openstackgerritLuis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Nested vif driver extension to enable ports reuse  https://review.openstack.org/43689310:03
*** limao has quit IRC10:05
*** egonzalez has joined #openstack-kuryr10:11
*** aojea has joined #openstack-kuryr10:14
*** jchhatbar has quit IRC10:20
*** oanson has quit IRC10:37
*** oanson has joined #openstack-kuryr10:37
*** janki has joined #openstack-kuryr10:53
*** yedongcan has left #openstack-kuryr11:12
*** salv-orl_ has quit IRC11:15
*** oanson has quit IRC11:28
*** oanson has joined #openstack-kuryr11:29
*** janki has quit IRC11:53
*** jchhatbar has joined #openstack-kuryr11:53
*** salv-orlando has joined #openstack-kuryr12:00
*** vikasc has quit IRC12:15
*** jchhatbar is now known as janki12:16
*** limao has joined #openstack-kuryr12:41
openstackgerritMerged openstack/kuryr-kubernetes master: Updated from global requirements  https://review.openstack.org/45593112:43
*** limao_ has joined #openstack-kuryr12:44
*** limao has quit IRC12:46
*** vikasc has joined #openstack-kuryr12:46
*** gsagie has joined #openstack-kuryr12:47
*** salv-orlando has quit IRC12:47
*** limao_ has quit IRC13:04
*** limao has joined #openstack-kuryr13:04
*** limao has quit IRC13:25
*** limao has joined #openstack-kuryr13:26
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: Delete docker.pid only if exists on CentOS/Fedora  https://review.openstack.org/45929213:27
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: Delete docker.pid only if exists on CentOS/Fedora  https://review.openstack.org/45929213:32
openstackgerritselvakumar proposed openstack/kuryr-kubernetes master: Handle LB Conflict exceptions during service deletion  https://review.openstack.org/45841313:39
openstackgerritDaniel Mellado proposed openstack/kuryr-kubernetes master: Delete docker.pid only if exists  https://review.openstack.org/45929213:41
*** hongbin has joined #openstack-kuryr13:41
*** limao has quit IRC13:46
*** limao has joined #openstack-kuryr13:47
apuimedovikasc: please review https://review.openstack.org/#/c/396048/613:48
openstackgerritAntoni Segura Puimedon proposed openstack/kuryr-kubernetes master: Spec for fuxi-kubernetes  https://review.openstack.org/45255413:48
openstackgerritSiyi Luo proposed openstack/kuryr-kubernetes master: Replace six.iteritems() with .items()  https://review.openstack.org/45931313:50
openstackgerritHongbin Lu proposed openstack/fuxi master: Make volume providers configurable in devstack  https://review.openstack.org/45769414:00
*** limao has quit IRC14:02
*** limao has joined #openstack-kuryr14:03
*** dimak_ has quit IRC14:17
openstackgerritMerged openstack/kuryr-libnetwork master: container connect both bridge and kuryr network in Rally  https://review.openstack.org/39604814:23
*** limao_ has joined #openstack-kuryr14:27
*** limao has quit IRC14:28
*** limao_ has quit IRC14:43
*** limao has joined #openstack-kuryr14:44
*** gsagie has quit IRC14:50
openstackgerritMerged openstack/kuryr-kubernetes master: Handle LB Conflict exceptions during service deletion  https://review.openstack.org/45841314:52
*** salv-orlando has joined #openstack-kuryr15:02
ltomasboapuimedo, irenab, ivc: the question I wanted to ask was about this: https://review.openstack.org/#/c/436875/5/kuryr_kubernetes/controller/handlers/vif.py15:05
*** limao has quit IRC15:05
ltomasbocomments at lines 46-4715:05
hongbinapuimedo: i am not sure if i can go to the ptg as well, so +1 for virtual15:05
irenabltomasbo: checking15:05
*** limao has joined #openstack-kuryr15:05
ltomasboshould I keep it as it is now (proposed by apuimedo and ok-ish from ivc side with a TODO/REVISIT)15:06
ivcltomasbo i don't want pool to be mentioned in handler in any way15:06
ivcltomasbo but we don't have necessary facilities for that yet15:06
ivcltomasbo so i'm ok with TODO/REVISIT15:06
apuimedothanks hongbin15:06
irenabltomasbo: I am ok with very visible TODO/REVISIT comment15:07
ltomasboivc, irenab: ok, I'll add that15:07
irenabbecause now its absolutly broken imho15:07
ivcltomasbo irenab apuimedo with Profile/Flavour we'll have set_... instead of self._drv_ = in __init__15:07
ivcltomasbo so the Profile will VIFHandler.set_vif_driver(VIFPool(NestedVIFDriver()))15:08
irenabivc: will try to want the recording asap15:08
ivcirenab ok, we discussed it with you already on irc earlier, so you prolly already know most of that stuff :)15:09
apuimedoirenab: I think you already "want" it15:09
apuimedo:-)15:10
irenab:-)15:10
ivc:P15:10
ltomasbo:D15:10
irenabivc: you always have some new fresh (crazy) ideas to share :-)15:10
apuimedoirenab: and that's why we love him15:11
apuimedo:-)15:11
ivcirenab i wonder if its good "crazy" or bad "crazy"? XD15:11
ltomasbo:D15:11
ltomasboivc, I guess it depends on the day :p15:11
irenabivc:  good of course15:11
ltomasbobtw ivc15:12
apuimedoI don't know what we would do without ivc's ideas and without irenab keeping us grounded, proposals and translating between our weird ideas15:12
apuimedoI'd probably bald15:12
ltomasboin your example VIFHandler.set_vif_driver(VIFPool(NestedVIFDriver()))15:12
*** janki has quit IRC15:13
ltomasbothen you take the vif driver as a composition of vif_pool which uses nestedVIF15:13
ltomasboivc ^^15:13
ivcltomasbo ideally it should be Pool(VIFDriver())15:13
ivcltomasbo yup composition all the way15:14
apuimedoivc: +1 for Pool instead of VIFPool15:14
apuimedo:-)15:14
ivcltomasbo oh we also have to change VIFDriver too so it can request/release/activate multiple VIFs in 1 batch15:14
irenabPool as for generic resources?15:14
ivc^15:14
ivcproxy-like object15:15
ltomasboivc, yep, actually right now the vif_pool_driver15:15
ltomasbois pretty much the same for both generic and nested15:15
apuimedoirenab: right15:15
ivcso Pool(VIFDriver()) will have same API as VIFDriver()15:15
ltomasboso, with a little bit of work, it could be even the same, just setting a differnt VIF driver (generic and nested)15:15
ivctho its pretty complex to implement and kinda complicated when you think about primary-key-like args / cache key base15:16
ivcPool is sort of a Cache15:16
ltomasboivc, I agree, it is sort of a proxy object with a cache15:16
irenabivc: ltomasbo : Are you already designing the follow-up patch?15:17
ltomasboirenab, what follow up patch?15:17
ivcltomasbo yup and in the end it will be VIFHandler.set_vif_driver( MultiCNIVIFDriver([Pool(GenericVIFDriver()), Pool(MACVLAN...()), Pool(TRUNK...())])))15:17
irenabltomasbo: the one after TODO/REVISIT15:18
ivcltomasbo MultiCNI is also more complex (we need some sort of discriminator to select the proper 'backend')15:18
ltomasboirenab, not yet, that we need the second call we were discussing during the meeting today15:18
ltomasbobut I'm happy to be involved in that! and work on the patch15:18
ltomasbobut I would like to keep it in mind to adapt the current one as much as possible15:19
ltomasboirenab, to simplify (if possible) the reshape15:19
irenabltomasbo: makes sense15:19
ltomasboin fact, I'll do a bit of thinking to try to leave both nested and generic pool drivers the same15:20
ltomasboso that it will be straightforward to adapt them to the next way of loading drivers15:20
ltomasbothough it may entail a bit of modification to the base drivers15:20
ivcltomasbo for it to work you need to change VIFDriver API15:20
ltomasbosuch as the request_vif plus request_vifs15:20
ivc^15:21
ltomasboivc, what do you mean?15:21
ltomasboadding the new request_vifs?15:21
* apuimedo got to go pick up child from school15:21
*** apuimedo is now known as apuimedo|away15:21
ivcltomasbo yup. that and release/activate_vifS too i think15:21
ltomasbonot sure about the activate15:22
ltomasboprobably not needed, but for sure for request/release15:22
ltomasbobut I think that is a reasonable change, right?15:22
ivcltomasbo thats for API consistency. tho release/activate will both be for-each loops15:22
ltomasboivc, irenab ^^15:22
ltomasboivc, not sure about that15:23
ltomasboivc, the ports are created in bulk creations15:23
ltomasbobut they do not need to be deleted in bulk creations15:23
ltomasboand for the activate, it is the same15:23
ltomasbothe fact that we create, let's say, 10 ports, does not mean we need to use them and wait for them to become active15:24
ltomasbobut when new ports are created, they will check (individually) if they are active or not15:24
ivcltomasbo yup but having only request_vifS feels inconsistent.15:24
ltomasbo(that may change though a bit for the macvlan case...)15:24
ivcAPI-wise15:24
ltomasboummm15:25
ltomasbothat's a good point15:25
ltomasbobut writing useless code is inconsistent too15:25
ivcremember that driver interface is API layer15:25
ivcthat is supposed to be reusable/extendable/external15:25
ltomasboI would prefer to add those when/if we find they are needed15:26
ivcwe need good API and i'm ok to duplicate code if necessary for a cleaner API15:26
ltomasboit is not duplicating code, my concern is creating code it is not going to be used15:26
irenabit looks like something bigger than just modifing the current patch. Maybe worht to have stand alone patch for the API change15:26
*** limao has quit IRC15:26
ltomasbobut that said15:26
ltomasbowe can create them at the API class15:27
ivcltomasbo yup15:27
*** limao has joined #openstack-kuryr15:27
ltomasboand live with the NotImplemented until needed15:27
ltomasbogreat!15:27
ltomasboagreed then!15:27
ivcltomasbo not exactly15:27
ltomasboumm, too fast :P15:27
ivcltomasbo the _vifS default implementations can be for-each _vif15:28
ivcltomasbo instead of NotImplemented15:28
ivcltomasbo i.e. VIFDriver.request_vifs(...): for vif in vifs: self.request_vif(vif)15:28
ltomasboyes, I know how if could be implemented15:29
ltomasbomy concern is that I don't see (right now) the point in the code where we will call release_vifS15:29
ivcltomasbo that was for your 'NotImplemented until needed'15:29
ltomasbofor sure, I request_vifS will be implemented (it is already in the patches)15:29
ivcltomasbo oh, well the pool can sure use some batch release15:30
ltomasboummm15:30
ltomasbothat could be another patch, as then you need to group them by trunk_id15:30
ivcltomasbo yup15:30
ltomasboand then group the ports15:30
ltomasboso it would be a few more line of code15:31
ivcltomasbo if we get to 'actors' some day, you can collect a batch of pool requests and execute in one go15:31
ltomasboso, if you are ok with that, I can leave it as NotImplemented for these 5 correlated patches, and hten have an extra one for the activate_vifs (if need it) and similarly for release_vifs15:31
ivcltomasbo right now afaik you are doing that on per-event basis15:31
ltomasboyes, the release if is per-event15:32
ltomasbothat is why there is no need for release_vifS15:32
ivcyup and that could be changed15:32
ltomasbobut, I was actually thinking of skiping some calls to recycle the pool15:32
ivcyup15:32
ltomasboso that I can do that on bulk requests...15:32
ivcnot skipping events, but postponing the action15:33
ltomasbobut I see this as an optimization, not like a ToDo in this patch serie15:33
ltomasboivc, yes, postponing/grouping15:33
ltomasboyep15:33
ivcyes but it aligns well with release_vifS API :)15:33
ltomasbotrying to make as fewer calls to neutron as possible15:33
ltomasboyes, it does!15:33
ivcgot it now? :)15:33
ltomasboivc, yes yes, I agree with you, and with including this at the API15:34
ivcXD15:34
ivci can be persuasive15:34
ltomasbojust wanting to ensure, that is not required for this patch serie (5 patchs enabling ports pool)15:34
ltomasbobut an extra one15:34
ltomasboivc, :D15:34
ivca TODO/REVISIT is always fine by me :)15:35
ltomasbo:D15:35
ltomasbosure! I'll include that15:35
ltomasbonice I'm not done with these patches and I already have a bunch of them in the ToDo list for later on...15:35
ivcits nice indeed. if you have TODO when you write your current impl means you think of future :)15:36
ltomasbo...or lazy to do it now...15:37
ltomasbojust joking...15:37
*** limao has quit IRC15:38
irenabivc: your last statement is very applicable to you15:41
ivcltomasbo its good to be lazy15:41
ivcirenab as in i dont write enough TODOs? :)15:42
ltomasbo:D15:42
ivcltomasbo not joking here. best code is the code not written15:42
ltomasboivc, yep, it has no bugs (yet)15:42
ivcand it is best if it runs on a powered-off server15:43
*** apuimedo|away is now known as apuimedo15:44
apuimedoivc: ltomasbo: irenab: how about a Thursday double feature to discuss mchiappero's and ltomasbo's patches?15:44
ivcapuimedo i'm busy on thursdays15:45
apuimedohow about tomorrow then?15:46
ltomasboapuimedo, we already discussed mine! :D15:46
ivcapuimedo and ltomasbo patches are in good shape15:46
apuimedogood then15:46
apuimedoless work15:46
ltomasbo:D15:46
apuimedothanks for sorting it out on the channel :-)15:46
ivcapuimedo for you, not for ltomasbo XD15:46
openstackgerritMerged openstack/fuxi master: Remove log translations  https://review.openstack.org/44782315:46
ltomasbobut it would be nice to have a follow-up call to the one we had the other day15:47
apuimedoivc: that's the work that matters15:47
apuimedoxD15:47
ltomasbo...to you...15:47
apuimedoltomasbo: that one for sure15:47
ltomasboxD15:47
apuimedoltomasbo: if you overwork I'll give you fuet to compensate15:47
ltomasboumm, nice15:47
ltomasbothough next time I'll go myself to grab it15:47
apuimedoltomasbo: but maybe you don't want dmellado doing the delivery15:47
ltomasbootherwise dmellado taxes are too high...15:48
apuimedohe may take a big chunk as delivery fee15:48
apuimedoright15:48
ivcapuimedo ltomasbo "will code for food" ? :)15:48
apuimedoltomasbo: I wouldn't mind visiting Madrid one day15:48
ltomasbosure! you are welcome!15:48
apuimedoI going a second time in my 30 years would not damage me15:48
ltomasbolol15:48
apuimedoivc: that's the best reason to code15:48
ltomasboto the office or to madrid?15:48
apuimedofor delicacies15:49
apuimedoltomasbo: both. I need an excuse15:49
ltomasbo:D15:49
apuimedootherwise the spanish secret service may get suspicious15:49
apuimedoxD15:49
ltomasboxDD15:49
apuimedowhat is that bearded guy with the strange accent going to madrid for15:49
*** salv-orl_ has joined #openstack-kuryr15:49
ltomasbolol15:51
*** salv-orlando has quit IRC15:52
apuimedoivc: ltomasbo: so how about tomorrow for macvlan discussion then?16:00
ltomasbosure, apuimedo when?16:00
ivcapuimedo requires mchiappero16:00
ltomasboI have a few other meetings...16:01
ivcapuimedo and i'd also prefer wed16:01
apuimedomchiappero: can't on wednesday16:01
apuimedohow about friday16:01
apuimedo?16:01
ivcfriday it is then16:02
apuimedoalright16:02
apuimedoI'll send an invite16:02
*** kzaitsev_ws has quit IRC16:10
*** ltomasbo is now known as ltomasbo|away16:19
*** salv-orl_ has quit IRC16:26
mchiapperofine with me!16:33
*** garyloug has quit IRC16:44
*** egonzalez has quit IRC16:51
openstackgerritOpenStack Proposal Bot proposed openstack/fuxi master: Updated from global requirements  https://review.openstack.org/45447416:52
*** salv-orlando has joined #openstack-kuryr17:38
*** tonanhngo has joined #openstack-kuryr17:45
*** salv-orlando has quit IRC18:45
*** salv-orlando has joined #openstack-kuryr19:35
openstackgerritMerged openstack/fuxi master: Make volume providers configurable in devstack  https://review.openstack.org/45769419:57
*** pcaruana has quit IRC20:01
*** tonanhngo_ has joined #openstack-kuryr20:12
*** tonanhngo has quit IRC20:14
*** tonanhngo has joined #openstack-kuryr20:20
*** tonanhngo_ has quit IRC20:22
openstackgerritOpenStack Proposal Bot proposed openstack/fuxi master: Updated from global requirements  https://review.openstack.org/45447420:25
apuimedoivc: I got it!20:27
apuimedoIt is possible to change the status of a service without implementing a darned cloudprovider20:28
apuimedoeverything can be done from the kuryr controller20:28
apuimedohttp://paste.openstack.org/show/607765/20:29
apuimedoI just had to create a role with patch permission not only to services, but also to services/status20:29
apuimedo:-)20:29
*** lihi has quit IRC20:52
*** lihi has joined #openstack-kuryr20:54
*** aojea has quit IRC21:26
*** salv-orl_ has joined #openstack-kuryr21:49
*** salv-orlando has quit IRC21:51
*** limao has joined #openstack-kuryr22:06
*** limao_ has joined #openstack-kuryr22:15
*** limao has quit IRC22:16
*** limao_ has quit IRC22:25
*** limao has joined #openstack-kuryr22:26
*** limao has quit IRC22:46
*** limao has joined #openstack-kuryr22:47
*** limao has quit IRC23:08
*** limao has joined #openstack-kuryr23:08
*** salv-orl_ has quit IRC23:10
*** limao has quit IRC23:28
*** tonanhngo has quit IRC23:43
*** tonanhngo has joined #openstack-kuryr23:55

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