Tuesday, 2017-06-06

« Monday, 2017-06-05 Index Wednesday, 2017-06-07 »
*** vikasc has joined #openstack-zun00:04
*** jwcroppe has joined #openstack-zun00:32
*** jwcroppe has quit IRC01:08
*** catintheroof has quit IRC01:10
*** XueFeng has quit IRC01:14
*** lakerzhou2 has joined #openstack-zun01:32
*** caowei has joined #openstack-zun01:56
openstackgerritMerged openstack/zun-ui master: Updated from global requirements  https://review.openstack.org/47056902:01
openstackgerritMerged openstack/python-zunclient master: Updated from global requirements  https://review.openstack.org/47055802:04
*** Shunli has joined #openstack-zun02:08
*** hongbin has joined #openstack-zun02:26
*** chenying has quit IRC02:28
*** chenying has joined #openstack-zun02:28
*** jwcroppe has joined #openstack-zun02:29
hongbinhi all, team meeting will start in about 15 minutes02:44
hongbinat #openstack-meeting02:44
*** chenying has quit IRC02:54
*** chenying has joined #openstack-zun02:54
*** diga has joined #openstack-zun02:58
openstackgerritfeng.shengqin proposed openstack/zun master: [WIP]Persist infra container in DB  https://review.openstack.org/46753503:04
*** zsli_ has joined #openstack-zun03:05
*** Shunli has quit IRC03:07
*** Shunli has joined #openstack-zun03:13
*** zsli_ has quit IRC03:16
*** klkumar has joined #openstack-zun03:18
*** zhurong has joined #openstack-zun03:28
openstackgerritfeng.shengqin proposed openstack/zun master: Log the network configuration  https://review.openstack.org/47080903:36
openstackgerritMerged openstack/python-zunclient master: Skip run OSC tests on unit tests  https://review.openstack.org/46519003:55
*** janki has joined #openstack-zun04:05
hongbinlakerzhou2: hi zhoubin, a question04:07
hongbinlakerzhou2: does nova support create a vm in both tenant and provider network?04:08
hongbinperhaps, it is as simple as specifying two --nic ?04:10
*** lakerzhou2 has quit IRC04:12
*** hongbin has quit IRC04:16
openstackgerritRajat Sharma proposed openstack/zun master: Replace assertEqual([], items) with assertEmpty(items)  https://review.openstack.org/46835804:18
*** zhurong has quit IRC04:27
*** kevinz has joined #openstack-zun04:29
*** diga has quit IRC04:39
*** zhurong has joined #openstack-zun04:54
*** jwcroppe has quit IRC05:42
*** zsli_ has joined #openstack-zun05:48
*** zsli__ has joined #openstack-zun05:49
*** Shunli has quit IRC05:51
*** zsli_ has quit IRC05:52
*** zsli_ has joined #openstack-zun05:55
*** zsli__ has quit IRC05:57
*** zsli__ has joined #openstack-zun05:58
*** zsli_ has quit IRC06:00
*** XueFeng has joined #openstack-zun06:03
*** jwcroppe has joined #openstack-zun06:10
openstackgerritmiaohb proposed openstack/zun master: Add return the image ID in container commit  https://review.openstack.org/47086406:21
openstackgerritfeng.shengqin proposed openstack/python-zunclient master: OSC: return columns instead of using print_dict  https://review.openstack.org/47085006:24
*** zhurong has quit IRC06:32
openstackgerritfeng.shengqin proposed openstack/zun master: Persist infra container in DB  https://review.openstack.org/46753506:36
*** zhurong has joined #openstack-zun06:38
*** jwcroppe has quit IRC06:47
*** eliqiao has quit IRC07:09
*** eliqiao has joined #openstack-zun07:10
openstackgerritMerged openstack/zun master: Log the network configuration  https://review.openstack.org/47080907:16
openstackgerritMerged openstack/zun master: Updated from global requirements  https://review.openstack.org/47094107:16
*** klkumar has quit IRC07:21
*** klkumar has joined #openstack-zun07:23
*** yasemin has joined #openstack-zun07:24
*** jwcroppe has joined #openstack-zun07:29
*** nicolasbock has joined #openstack-zun07:31
openstackgerritfeng.shengqin proposed openstack/python-zunclient master: OSC: return columns instead of using print_dict  https://review.openstack.org/47085007:33
*** egonzalez has joined #openstack-zun07:37
openstackgerritfeng.shengqin proposed openstack/python-zunclient master: Delete the links info when showing a container  https://review.openstack.org/47125208:21
*** zhurong has quit IRC08:35
*** zhurong has joined #openstack-zun08:40
*** nicolasbock has quit IRC09:03
*** yasemin has quit IRC09:10
*** zsli__ has quit IRC09:30
*** jwcroppe has quit IRC09:33
*** zhurong has quit IRC09:33
*** yasemin has joined #openstack-zun09:42
*** kevinz has quit IRC10:14
*** caowei has quit IRC10:26
*** klkumar has quit IRC10:38
*** vikasc has quit IRC10:41
*** nicolasbock has joined #openstack-zun10:44
*** vikasc has joined #openstack-zun10:54
*** klkumar has joined #openstack-zun10:57
*** yasemin has quit IRC11:26
*** zhurong has joined #openstack-zun11:30
*** klkumar has quit IRC11:42
*** XueFeng has quit IRC11:51
*** XueFeng has joined #openstack-zun11:51
*** jwcroppe has joined #openstack-zun11:59
*** yasemin has joined #openstack-zun12:02
*** lakerzhou2 has joined #openstack-zun12:09
lakerzhou2hongbin, yes, VM can attach to multiple networks, tenant and/or providers12:10
*** klkumar has joined #openstack-zun12:11
*** zhurong has quit IRC12:12
*** janki has quit IRC12:21
*** XueFeng has quit IRC12:26
*** yasemin has quit IRC12:35
*** yasemin has joined #openstack-zun12:38
*** catintheroof has joined #openstack-zun12:42
*** vikasc has quit IRC12:48
*** lakerzhou2 has quit IRC12:59
*** vikasc has joined #openstack-zun13:05
*** raphaelehret has quit IRC13:08
*** lakerzhou2 has joined #openstack-zun13:23
*** lakerzhou1 has joined #openstack-zun13:24
*** lakerzhou2 has quit IRC13:28
*** klkumar has quit IRC13:58
*** janki has joined #openstack-zun14:02
*** lakerzhou2 has joined #openstack-zun14:02
*** lakerzhou1 has quit IRC14:06
*** klkumar has joined #openstack-zun14:10
*** raphaelehret has joined #openstack-zun14:12
raphaelehretHi! I am currently trying to install zun in devstack, but during the `./stack` command, I encounter the error "AttributeError: 'module' object has no attribute 'HostAddressOpt'" when the script executes the command `/usr/local/bin/zun-db-manage upgrade`.14:14
mkrairaphaelehret: Hi14:14
mkraiCan you paste the error log ?14:14
raphaelehretMy devstack version is Mitaka, as I wanted to try if zun was compatible with this version of Openstack. Is this error caused by compatibiliy?14:15
raphaelehrethttps://pastebin.com/raw/YCeSyn9c14:15
mkraiYes that is possibly a compatibility issue14:16
mkraiWhat's your oslo.cfg version?14:17
*** lakerzhou2 has quit IRC14:18
*** lakerzhou2 has joined #openstack-zun14:20
raphaelehretmhh, where is this file supposed to be located? Can't seem to find it in /etc | /usr/local/lib14:20
*** hongbin has joined #openstack-zun14:28
mkraisudo pip show oslo.config14:28
raphaelehretHo, my bad14:28
raphaelehret3.9.014:28
raphaelehretFull output : "Name: oslo.config Version: 3.9.0 Summary: Oslo Configuration API Home-page: https://wiki.openstack.org/wiki/Oslo#oslo.config Author: OpenStack Author-email: openstack-dev@lists.openstack.org License: UNKNOWN Location: /usr/local/lib/python2.7/dist-packages Requires: six, netaddr, stevedore, debtcollector"14:28
mkrairaphaelehret: Mitaka doesn't have HostAddressOpt option14:30
*** lakerzhou2 has quit IRC14:30
*** lakerzhou2 has joined #openstack-zun14:30
raphaelehretSo there is definitely compatibility involved, then14:30
raphaelehretmkrai: Thanks for this :) Do you know if zun is compatible with Zun?14:31
raphaelehretwith Newton*14:31
hongbinraphaelehret: officially no, zun doesn't have stable branch for either newton or ocata, but you might want to make several small tweak to make it work with newton if you want14:34
raphaelehrethongbin: I'll try this way, I guess14:35
raphaelehretMy production environment is currently Mitaka, I will migrate it into Newton soon enough, but Pike isn't stable ATM so I don't think it is suitable for prod14:36
hongbinraphaelehret: i see14:37
hongbinyes, we should start produce stable branch from now on14:37
hongbinunfortunately, it seems we cannot make a stable branch for newton anymore14:38
raphaelehretIs it out of the development cycle?14:38
hongbindont quite get your question, could you elaborate?14:39
raphaelehretIs it still possible to develop plugins/modules for newton and integrate it into Newton officially?14:41
raphaelehret(Excuse my for my poor English, I'm not a native english speaker :P)14:41
hongbinas far as i know, this is impossible officially14:42
hongbinsince zun is a new project, we were not participanting in the newton release in about one year ago14:42
raphaelehretOkay, thanks14:43
hongbin(np for the english speaking, i am not native english speaker as well)14:43
raphaelehretOne last question, is there any ressources available on the Web for the installation of Zun into Newton (official or unofficial)?14:44
hongbinthe closest one is: https://github.com/openstack/zun/blob/master/doc/source/dev/manual-devstack.rst , we are working on giving the official installation guide, but not there yet14:47
*** kevzha01 has joined #openstack-zun14:48
hongbinthe manual install guide is for master, but might work for newton after a few tweak14:49
*** kevzha01 has quit IRC14:52
*** kevzha01 has joined #openstack-zun14:52
raphaelehrethongbin: Okay, I'll definitely give it a try. Would it help for me to give a feedback about the needed modifications to make it work?14:53
raphaelehretFor you*14:53
raphaelehretI mean, would it be helpful for you if I give a feedback of my installation of Zun on Newton?14:54
*** kevzha01_ has joined #openstack-zun14:54
hongbinraphaelehret: that is for sure14:54
mkrairaphaelehret: That will be great :)14:54
raphaelehretOkay then! Thanks for your time :)14:56
hongbinthank you14:56
hongbinraphaelehret: btw, for the error you mentioned above, it might be resolved by reverting this commit: https://github.com/openstack/zun/commit/093c65aea30d69e8bf4623b218a4cba921886bf714:56
*** kevzha01 has quit IRC14:57
hongbinraphaelehret: feel free to ping me or mkrai if you encounter any error later and need help to resolve it14:58
raphaelehrethongbin: Let's try this14:59
*** pksingh has joined #openstack-zun15:03
*** kevzha01_ has quit IRC15:04
*** kevzha01_ has joined #openstack-zun15:04
*** kevzha01_ has quit IRC15:07
*** kevzha01_ has joined #openstack-zun15:07
pksinghhongbin: hello15:09
hongbinpksingh: hey pradeep15:09
pksinghhongbin: i am busy these days with someother stuffs, so just getting some time at night to work on zun15:09
pksinghhongbin: i think this will be over soon15:09
hongbinpksingh: that is totally fine :)15:10
openstackgerritMerged openstack/zun master: Change default scheduler to filter_scheduler  https://review.openstack.org/47006215:10
pksinghhongbin: btw, do you have any idea why we use oslo-versioned-object in openstack, from user's perspective15:10
pksinghhongbin: what are it's benifits?15:10
hongbinpksingh: it is for upgrade between compoenents15:11
pksinghhongbin: components means different openstack components, like nove and zun15:12
pksinghhongbin: or some service in a single project15:12
hongbinhongbin: some service in a single project15:12
pksinghhongbin: ok, so how does it help, can you little bit elaborate, if you have some time plz15:13
hongbinhongbin: for example, if you want to upgrade zun-api from version 1 to version 2, you will have zun-api (version 2) and zun-compute (version 1) after the upgrade15:13
hongbins/after the upgrade/during the upgrade/15:13
hongbinthat is what versioned object is for15:14
openstackgerritMerged openstack/zun master: Introduce cpu-policy for container deployment  https://review.openstack.org/42700715:14
hongbinit is for different version of compoenents to do rpc communication15:14
pksinghhongbin: ok, so it gives the result based on the consuming service version15:15
hongbinpksingh: yes, it suppose to do that15:15
pksinghhongbin: so it maintains a version for every service which is accessing the DB and if there is change in schema they wouldn't be affected?15:16
hongbinpksingh: in particular, maintains a version for every service which is doing RPC calls15:17
hongbinbut yes, eventually, it si for handling different db schema15:18
pksinghhongbin: so in our case zun-compute is not using rpc calls to access the DB?15:18
hongbinpksingh: no we don't15:18
raphaelehrethongbin: I have tried with the commit reverted, the script still fails with a quite similar error : https://pastebin.com/A3bJWWPB15:18
hongbinraphaelehret: perhaps revert this commit as well: https://github.com/openstack/zun/commit/72a79970803d8d5ba9c11903a7ba0e0fe8fe63ac15:19
hongbinpksingh: rpc layer and db layer are independent15:20
pksinghhongbin: and same with zun-api, which is accessing the database directly?15:20
hongbinpksingh: yes, zun-api should access the db directly15:21
hongbinpksingh: for security reason, we don't want to let zun-compute to access the db directly, but we are doing it right now15:21
pksinghhongbin: so you are saying, since zun api uses rpc calls, it will be benifitted from this versioning library/15:22
hongbinpksingh: yes15:23
pksinghhongbin: ok, but not zun-compute?15:23
hongbinpksingh: zun-compute is using versioned-object as well15:23
hongbinpksingh: i think there are several benefits the versioned-object provides15:24
hongbinpksingh: one of them is solving the upgrade problem15:24
pksinghhongbin: k15:25
hongbinpksingh: another one is for defining the rpc protocal15:25
pksinghhongbin: can you plz explain last point?15:25
hongbinpksingh: for rpc call, we need something to serialize object to data stream, and deserialize data stram to object15:25
pksinghhongbin: ok, got it15:26
hongbinpksingh: for example, serialize the container object into a json stream, before sending it to the message queue15:26
hongbinyes, this is what versioned-object is doing15:26
pksinghhongbin: ok thanks a lot for your time15:27
pksinghhongbin: you save a lot of my time15:27
hongbinpksingh: np, happy to help15:27
pksinghhongbin: thnx again15:27
pksinghhongbin: i will see if i can take any BP over this weekend15:27
pksinghhongbin: did you found how nova updates the Floating IP in DB?15:28
hongbinpksingh: ack, take your time15:28
pksinghhongbin: i did not get time to work on that15:28
hongbinpksingh: haven't looked into it yet15:28
pksinghhongbin: ok, no issue, i will look into that15:29
hongbinpksingh: ack, take your time :)15:29
*** kevzha01_ has quit IRC15:40
*** janki has quit IRC15:58
*** egonzalez has quit IRC16:05
*** pksingh has quit IRC16:21
*** klkumar has quit IRC16:51
*** pksingh has joined #openstack-zun17:05
*** pksingh has quit IRC17:06
*** pksingh has joined #openstack-zun17:33
*** nicolasbock has quit IRC17:41
*** nicolasbock has joined #openstack-zun17:42
pksinghhongbin: It seems nova get the floating IP update through events generated by neutron18:34
hongbinpksingh: interesting18:35
pksinghhongbin: when there is change in network, neutron generates an event and nova process that18:35
hongbinpksingh: could you point to the code?18:36
pksinghhongbin: https://github.com/openstack/nova/blob/5d95cb9dbca403790db4e9680919e6716fa5cb76/nova/compute/manager.py#L688118:36
pksinghhongbin: nova maintains such information in table at here https://github.com/openstack/nova/blob/5d95cb9dbca403790db4e9680919e6716fa5cb76/nova/db/sqlalchemy/models.py#L35518:37
pksinghhongbin: neutron generates it here https://github.com/openstack/neutron/blob/03c5283c69f1f5cba8a9f29e7bd7fd306ee0c123/neutron/notifiers/nova.py#L13318:38
hongbinpksingh: the event is delivered over rpc?18:40
pksinghhongbin: i have no idea about that, but i think should18:41
pksinghhongbin: otherwise who will process it, if it pub/sub?18:42
hongbinpksingh: i think another possibility is via rest api18:42
hongbinpksingh: but it seems it is via rpc18:42
pksinghhongbin: i was talking at nova channel, and they said to avoid neutron proxy18:42
pksinghhongbin: they also said plz dont copy there address field style, which they show during server detail18:43
pksinghhongbin: its not good18:43
hongbini see18:43
hongbinpksingh: what do you think about our cases18:45
pksinghhongbin: let me give you the chat snippet18:46
pksinghhongbin:  i am little bit weak in networking ;)18:46
pksinghhongbin: http://eavesdrop.openstack.org/irclogs/%23openstack-nova/%23openstack-nova.2017-06-06.log.html#t2017-06-06T15:37:5018:47
pksinghhongbin: at this time i am not sure, how to deal with it18:50
pksinghhongbin: but i am sure we should not use neutron proxy, because that will create dependency on neutron18:51
* hongbin is reading the nova logs18:51
hongbinpksingh: i see, then you think we should not show any floating ip in server details?18:52
pksinghhongbin: i think we need to show it, may be we can follow the same as nova follows18:55
pksinghhongbin: neutron, is using rest API, https://github.com/openstack/neutron/blob/03c5283c69f1f5cba8a9f29e7bd7fd306ee0c123/neutron/notifiers/nova.py#L24218:56
hongbinpksingh: that means nova has a rest api to receive external events18:57
hongbinpksingh: here you go: https://developer.openstack.org/api-ref/compute/#create-external-events-os-server-external-events18:58
pksinghhongbin: yes, so neutron generates the event, nova calls neutron api and updates the floating IP18:59
hongbinpksingh: ok, that means we cannot leverage the same approcah since it is specific for nova-neutron communication18:59
pksinghhongbin: yes, for that purpose there will be major change in neutron19:00
hongbinpksingh: personally, i think the nova implementation is very advanced, we possibly couldn't copy the same approach due to the amount of work19:00
pksinghhongbin: agree19:00
hongbinpksingh: do you have an idea for a simplier approach?19:01
pksinghhongbin: there remains two approaches, 1) go old nova way 2) add some periodic task(it will be large overhead for large number for containers)19:01
hongbinpksingh: i see19:02
pksinghhongbin: do you see any other way19:02
hongbinpksingh: i think we could let zunclient to retrieve the floating ip19:02
hongbinpksingh: that will reduce the complexity and performance overhead on the server19:03
pksinghhongbin: but if user uses rest then?19:03
hongbinpksingh: yes, that is a cons19:03
pksinghhongbin: and zun client will always calls neutron, whether fip is assigned or not19:04
hongbinpksingh: that is true19:04
hongbinpksingh: you decide :)19:05
hongbinpksingh: there are pros and cons of either approach19:06
pksinghhongbin: correct19:06
pksinghhongbin: i think in long term, we should also be notified if there is change in network19:08
pksinghhongbin: like if somebody deletes port then we should get notified19:08
hongbinpksingh: ok, i will look into the long term approach and figure out how to talk to the neutron people19:08
hongbinlet me create a bp to track that19:09
pksinghhongbin: regarding approaches, we can decide in team meeting, or i can drop am mail to team19:09
hongbinpksingh: sure19:09
hongbinsounds like a plan19:09
pksinghhongbin: ok, bye, going to sleep, will ttyl19:10
hongbinpksingh: good night19:10
pksinghhongbin: have a great day ahead19:10
pksinghbye19:10
*** pksingh has quit IRC19:11
*** Qiming has quit IRC19:13
*** Qiming has joined #openstack-zun19:17
*** harlowja has quit IRC19:29
*** harlowja has joined #openstack-zun19:50
*** nicolasbock has quit IRC20:35
*** lakerzhou2 has quit IRC21:28
*** raphaelehret has quit IRC21:29
*** catintheroof has quit IRC22:41
*** hongbin has quit IRC22:59
*** jwcroppe has quit IRC23:07
« Monday, 2017-06-05 Index Wednesday, 2017-06-07 »

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