| *** wanghao has joined #openstack-mogan | 00:38 | |
| zhenguo | morning mogan! | 00:46 |
|---|---|---|
| zhenguo | liusheng: hi, should we keep this https://review.openstack.org/#/c/478403/ ? | 00:52 |
| liusheng | morning | 00:52 |
| liusheng | zhenguo: not sure, it is related with your spec | 00:53 |
| liusheng | zhenguo: how do you want to design the az info with node aggregate metadata ? | 00:54 |
| zhenguo | liusheng: yes, with the aggregate metadata | 00:54 |
| zhenguo | liusheng: should we keep the az list interface and just return the default az | 00:55 |
| liusheng | zhenguo: only defaut az ? if so I don't think so | 00:56 |
| liusheng | zhenguo: an api to read one config option | 00:56 |
| zhenguo | liusheng: we will add the logic soon | 00:56 |
| zhenguo | liusheng: as it get rid of the az interface and the api doc, which we will add back again | 00:57 |
| liusheng | so you think should we still need the az api ? | 00:59 |
| liusheng | zhenguo: if we record az info into the aggregates metadata | 01:00 |
| zhenguo | liusheng: nova als record az info in the aggregate metadata | 01:01 |
| liusheng | zhenguo: yes, we will add node aggregates api, the return node aggregates will include all the medatadata, right, how about add an request option to the aggregate list api to only return the az info ? | 01:03 |
| liusheng | zhenguo: a personal opinion | 01:03 |
| zhenguo | liusheng: only admin can get aggregate info | 01:03 |
| liusheng | zhenguo: non-admin can get az list ? | 01:04 |
| zhenguo | liusheng: sure | 01:04 |
| zhenguo | liusheng: az list is just for non-admins | 01:04 |
| zhenguo | liusheng: when claiming a server common users can specify which az whey want the server in | 01:05 |
| liusheng | zhenguo: I don't think so, in Nova, az list is for admin or owners, az details is only for admins | 01:06 |
| zhenguo | liusheng: but we don't have az details | 01:06 |
| liusheng | zhenguo: yes, they need to specify, but they don't need to know all the az list | 01:07 |
| liusheng | zhenguo: yes | 01:07 |
| zhenguo | liusheng: ? | 01:07 |
| zhenguo | liusheng: maybe not all the list, but a list of az they can specify | 01:08 |
| zhenguo | liusheng: so, still need a api | 01:08 |
| liusheng | zhenguo ok, will abandon that | 01:09 |
| zhenguo | liusheng: not abandon it | 01:09 |
| zhenguo | liusheng: the logic related to compute_nodes need to be removed | 01:10 |
| liusheng | zhenguo: sure, I know | 01:10 |
| zhenguo | liusheng: ok, thanks | 01:10 |
| openstackgerrit | wanghao proposed openstack/mogan master: Manage existing BMs: Part-1 https://review.openstack.org/479660 | 01:22 |
| *** liujiong has joined #openstack-mogan | 01:24 | |
| openstackgerrit | liusheng proposed openstack/mogan master: Clean the compute_node and compute_port objects and db interfaces https://review.openstack.org/478406 | 01:27 |
| openstackgerrit | wanghao proposed openstack/mogan master: Manage existing BMs: Part-1 https://review.openstack.org/479660 | 01:29 |
| *** wanghao has quit IRC | 01:31 | |
| *** wanghao has joined #openstack-mogan | 01:32 | |
| *** liujiong_lj has joined #openstack-mogan | 01:36 | |
| *** wanghao_ has joined #openstack-mogan | 01:37 | |
| *** liujiong has quit IRC | 01:37 | |
| *** wanghao_ has quit IRC | 01:37 | |
| *** wanghao_ has joined #openstack-mogan | 01:38 | |
| *** wanghao has quit IRC | 01:39 | |
| *** liujiong_lj is now known as liujiong | 01:39 | |
| *** wanghao has joined #openstack-mogan | 01:42 | |
| *** wanghao_ has quit IRC | 01:45 | |
| openstackgerrit | liusheng proposed openstack/mogan master: Clean the compute_node and compute_port objects and db interfaces https://review.openstack.org/478406 | 01:59 |
| openstackgerrit | liusheng proposed openstack/mogan master: Tempest: get rid of the default flavor created by devstack https://review.openstack.org/483672 | 02:03 |
| zhenguo | seems we still have time to modify the presentations. | 02:05 |
| zhenguo | liusheng: the presentation content was updated according to some suggestions from shaohe_feng's colleague, please take a look | 02:07 |
| liusheng | zhenguo: thanks | 02:09 |
| liusheng | zhenguo: Bare Metal or Baremetal ? compute or computing ? | 02:11 |
| zhenguo | liusheng: Bare Metal | 02:11 |
| liusheng | zhenguo: seems yes | 02:12 |
| zhenguo | liusheng: not sure compute or computing | 02:12 |
| zhenguo | liusheng: but seems nova is compute service | 02:12 |
| liusheng | zhenguo: yes | 02:13 |
| liusheng | zhenguo: is the "for" in tittle a bit redundant ? | 02:13 |
| zhenguo | liusheng: not sure | 02:13 |
| zhenguo | liusheng: but it's suggested by a native speaker. lol | 02:14 |
| liusheng | zhenguo: hah | 02:14 |
| liusheng | zhenguo: not sure, seems Ironic like to say they are Bare Metal provisioning | 02:15 |
| zhenguo | liusheng: yes, I also find that ironic use provisioning service | 02:15 |
| zhenguo | liusheng: anyway, I think both are ok | 02:16 |
| *** litao__ has joined #openstack-mogan | 02:16 | |
| liusheng | zhenguo: yes, it is may hard to avoid to analysis the relationship between Mogan and Ironic during the presentation | 02:17 |
| liusheng | zhenguo: the abstract looks good! | 02:18 |
| zhenguo | liusheng: they may also look into the abstract to see what's the title meaning, hah | 02:18 |
| liusheng | zhenguo: to my understanding, the "without twists and truns" is comparing with Nova+Ironic, right? | 02:20 |
| zhenguo | liusheng: yes | 02:20 |
| liusheng | zhenguo: but seems we didn't mention Nova + Ironic in abstract | 02:20 |
| zhenguo | liusheng; the abstract should be less than 1000 chars | 02:21 |
| liusheng | zhenguo: :( | 02:21 |
| zhenguo | liusheng: but we still have links to mogan wiki | 02:22 |
| zhenguo | liusheng: but the wiki is also a bit out of date | 02:22 |
| liusheng | zhenguo: how about adding a "comparing with Nova+Ironic deployment" before "no twists and turns" in the first line of abstract | 02:23 |
| zhenguo | liusheng: but it's not deployment | 02:24 |
| liusheng | zhenguo: why ? | 02:24 |
| zhenguo | liusheng: for twists and turns we don't mean deployment, but the driver architecture, the api, the model | 02:25 |
| liusheng | zhenguo: yes, the deployment I don't mean only deploying | 02:25 |
| zhenguo | liusheng: but for others I think they think it's the deploying | 02:26 |
| liusheng | zhenguo: in some openstack docs it can also means a model | 02:26 |
| zhenguo | liusheng: I think maybe people can understand the twist and truns here | 02:27 |
| liusheng | zhenguo: .. ok | 02:27 |
| zhenguo | liusheng: hah, but we need to update the wiki soon | 02:27 |
| liusheng | zhenguo: yes | 02:27 |
| zhenguo | liusheng: like placement should be added there | 02:28 |
| liusheng | zhenguo: will check | 02:28 |
| zhenguo | liusheng: I will prepare the ppt for openstack china days, then update the wiki | 02:28 |
| zhenguo | liusheng: will send you a draft version soon | 02:28 |
| liusheng | zhenguo: thanks, the ppt can be resue in summit. if we are selected | 02:29 |
| zhenguo | liusheng: btw, do you have experience to write '软文' | 02:29 |
| zhenguo | liusheng:ｙｅｓ | 02:29 |
| liusheng | zhenguo: lol | 02:29 |
| zhenguo | liusheng: hah | 02:29 |
| liusheng | zhenguo: no :( | 02:29 |
| zhenguo | liusheng: we may need some blogs | 02:29 |
| zhenguo | liusheng: it's ok to write chinese blog first | 02:30 |
| liusheng | zhenguo: the author should have reputation. hah | 02:30 |
| zhenguo | liusheng: hah | 02:30 |
| litao__ | zhenguo: Pls see my comment in patch https://review.openstack.org/#/c/476862/8 | 06:36 |
| zhenguo | litao__: ok | 06:37 |
| openstackgerrit | liusheng proposed openstack/mogan master: Clean the compute_node and compute_port objects and db interfaces https://review.openstack.org/478406 | 06:39 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add indexes to tables https://review.openstack.org/483203 | 06:55 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Adds aggregates DB model and API https://review.openstack.org/482786 | 07:01 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Adds aggregates DB model and API https://review.openstack.org/482786 | 07:02 |
| zhenguo | liusheng: https://review.openstack.org/#/c/478406/ , the gate failed again? | 07:17 |
| liusheng | zhenguo: Failed to retrieve filtered list of resource providers from placement API for filters {'resources': u'CUSTOM_GOLD:1'} | 07:21 |
| liusheng | zhenguo: not sure why the resources is CUSTOM_GOLD | 07:21 |
| zhenguo | liusheng: yes, why we still got a CUSTOM_GOLD | 07:25 |
| zhenguo | liusheng: maybe need to rebase the patch | 07:25 |
| zhenguo | liusheng: it's still the old flavor, need a rebase to get it update | 07:29 |
| liusheng | zhenguo: I have rebased | 07:30 |
| liusheng | zhenguo: I got the reason | 07:34 |
| liusheng | zhenguo: we need this patch: https://review.openstack.org/#/c/483672/ to fix | 07:35 |
| zhenguo | liusheng: why? | 07:37 |
| zhenguo | liusheng: I don't understand why we still got CUSTOM_GOLD | 07:37 |
| liusheng | zhenguo: the flavor related tempest tests will create flavors with CUSTOM_GOLD, if tempest run parallelly, we use the first one of listing flavors to create server, the first one may be the flavor created by flavor tests | 07:38 |
| liusheng | zhenguo: but I didn't know why the above patch cannot pass the gate, and the log show : No such resource class CUSTOM_BAREMETAL_1CPU_1280MBRAM_10GBDISK | 07:40 |
| zhenguo | liusheng: you mean in resource_setup? | 07:40 |
| liusheng | zhenguo: the flavor tests and server tests may running in different threads, so it is possible the listing flavors can get the flavor created by flavor tests | 07:41 |
| zhenguo | liusheng: ok, got it | 07:41 |
| liusheng | zhenguo: and I have check the gate log, it actually is | 07:41 |
| liusheng | zhenguo: our default resource class is CUSTOM_BAREMETAL_1CPU_1280MBRAM_10GBDISK right ? | 07:42 |
| zhenguo | liusheng: seems yes | 07:42 |
| liusheng | zhenguo: hmm, it is strange, the above patch's log show no such resource class | 07:43 |
| zhenguo | liusheng: do you have a env running with the latest code | 07:44 |
| zhenguo | liusheng: not sure if placement will do some twist for the resource class name | 07:44 |
| liusheng | zhenguo: yes, I mean the https://review.openstack.org/#/c/483672/ cannot pass the gate | 07:44 |
| liusheng | zhenguo: but I have checked in my local env, it is that resource class | 07:45 |
| zhenguo | liusheng: on placement db? | 07:45 |
| liusheng | zhenguo: yes | 07:45 |
| zhenguo | liusheng: so you can create server successfully in your local env? | 07:46 |
| liusheng | zhenguo: yes | 07:46 |
| zhenguo | liusheng: weird | 07:47 |
| litao__ | liusheng: If I update the latest code from mogan, should i need to update other codes like placement? | 08:18 |
| liusheng | litao__: no, Mogan don't depends on the latest placement | 08:19 |
| zhenguo | liusheng: I got the reason | 08:27 |
| zhenguo | liusheng: in tempest, it's CUSTOM_BAREMETAL_1CPU_1024MBRAM_10GBDISK | 08:28 |
| zhenguo | liusheng: not 1280MBRAM | 08:28 |
| liusheng | zhenguo: ... haha | 08:28 |
| zhenguo | liusheng: you can t ry | 08:28 |
| liusheng | zhenguo: let me modify | 08:28 |
| zhenguo | liusheng: ok | 08:28 |
| openstackgerrit | liusheng proposed openstack/mogan master: Tempest: get rid of the default flavor created by devstack https://review.openstack.org/483672 | 08:31 |
| litao__ | liusheng: thanks | 08:39 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Update devstack flavor description https://review.openstack.org/484291 | 08:49 |
| litao__ | liusheng: In _update_available_resources， the all_rps is None from placement. | 09:03 |
| liusheng | litao__: did you find error when getting rps ? | 09:05 |
| zhenguo | shaohe_feng: need you +2 on this https://review.openstack.org/#/c/483672/ | 09:07 |
| zhenguo | liusheng: it finally passed the gate | 09:08 |
| liusheng | zhenguo: thanks | 09:08 |
| zhenguo | liusheng: np | 09:08 |
| * litao__ liusheng: no more useful message, just TypeError: 'NoneType' object is not iterable | 09:13 | |
| litao__ | liusheng: ☹️ | 09:14 |
| liusheng | litao__: yes, this is an little issue, but the root cause is it retrive resources providers failed. there should be error log. see: get_filtered_resource_providers method in mogan/scheduler/client/report.py | 09:15 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Update devstack flavor description https://review.openstack.org/484291 | 09:16 |
| litao__ | liusheng: The log should in mogan-engine, but I can't find it | 09:21 |
| liusheng | litao__: mogan/scheduler/client/report.py:189 | 09:24 |
| litao__ | liusheng: Yes, I see in code, but nothing in logs | 09:24 |
| liusheng | litao__: only when retrive rps failed, it can return a None | 09:24 |
| *** wanghao has quit IRC | 09:34 | |
| *** liujiong has quit IRC | 10:06 | |
| openstackgerrit | Merged openstack/mogan master: Update devstack flavor description https://review.openstack.org/484291 | 11:04 |
| openstackgerrit | Merged openstack/mogan master: Remove the private database config option https://review.openstack.org/483723 | 11:05 |
| *** litao__ has quit IRC | 11:52 | |
| zhenguo | liusheng: seems we already got aggregates and rps cache with the SchedulerReportClient, | 11:57 |
| zhenguo | liusheng: but is that singleton? | 11:57 |
| liusheng | zhenguo: seems not | 12:00 |
| zhenguo | liusheng: I find two places import scheduler client | 12:00 |
| zhenguo | liusheng: should we avoid that | 12:01 |
| zhenguo | liusheng: make only one client import in the engine manager | 12:01 |
| liusheng | zhenguo: oh, for scheduler_client it is not singleton, but for one schedule_client's report clients, it is singleton | 12:05 |
| liusheng | zhenguo: will check the effects | 12:05 |
| openstackgerrit | Merged openstack/mogan master: Tempest: get rid of the default flavor created by devstack https://review.openstack.org/483672 | 12:06 |
| zhenguo | liusheng: you mean report client is singleton? | 12:06 |
| zhenguo | liusheng: that's what I expected | 12:06 |
| liusheng | zhenguo: you can see the client/__init__.py | 12:06 |
| liusheng | zhenguo: there is a class loader wrapper | 12:06 |
| zhenguo | liusheng: aha, yes | 12:08 |
| * zhenguo away | 12:23 | |
| *** harlowja has joined #openstack-mogan | 17:17 | |
| *** harlowja has quit IRC | 19:32 | |
| *** harlowja has joined #openstack-mogan | 20:56 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!