| *** wanghao has joined #openstack-mogan | 00:20 | |
| zhenguo | morning mogan! | 00:40 |
|---|---|---|
| *** openstackgerrit has joined #openstack-mogan | 00:55 | |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add placement with mogan devstack plugin https://review.openstack.org/488931 | 00:55 |
| *** liusheng has joined #openstack-mogan | 01:07 | |
| *** litao__ has joined #openstack-mogan | 01:11 | |
| wanghao | morning! | 01:18 |
| wanghao | liusheng: ping | 01:18 |
| liusheng | wanghao: pong | 01:18 |
| wanghao | liusheng: do you test guru in uwsgi. it do not work? | 01:19 |
| liusheng | zhenguo: yes, I don't know why :( | 01:19 |
| wanghao | liusheng: it just work under 2en way | 01:19 |
| liusheng | wanghao: yes, I just add the config option to mogan.conf, and then restart api server, but I cannot find any log | 01:20 |
| wanghao | liusheng: did you touch the file? | 01:20 |
| liusheng | wanghao: yes | 01:21 |
| wanghao | liusheng: mogan-api run with 2>/var/log/mogan/mogan-api-err.log? | 01:22 |
| liusheng | wanghao: yes, | 01:23 |
| wanghao | ok I will check it | 01:24 |
| liusheng | wanghao: let me try again | 01:24 |
| wanghao | liusheng: or you can withouth the 2> | 01:24 |
| wanghao | liusheng: and you will see the guru report in uwsgi console output | 01:24 |
| liusheng | wanghao: just restart mogan-api ? | 01:25 |
| wanghao | yes | 01:25 |
| wanghao | and then touch the file | 01:25 |
| wanghao | just uwsgi --ini /etc/mogan/mogan-uwsgi.ini | 01:26 |
| liusheng | wanghao: no, I use systemctl to restart api server | 01:27 |
| wanghao | liusheng: so you also can't see the uwsgi console output | 01:28 |
| liusheng | wanghao: but currently, the common way to restart services in devstack is using systemctl | 01:29 |
| liusheng | wanghao: it cannot support this way ? | 01:29 |
| wanghao | liusheng: what is your mogan-api process looking like. ps -ef | grep mogan-api | 01:34 |
| liusheng | wanghao: /usr/local/bin/uwsgi --ini /etc/mogan/mogan-uwsgi.ini | 01:35 |
| zhenguo | liusheng: do you know which config file to use with uwsgi | 01:36 |
| zhenguo | liusheng: like /etc/mogan/mogan.cnof | 01:37 |
| wanghao | liusheng: /usr/local/bin/uwsgi --ini /etc/mogan/mogan-uwsgi.ini 2>/var/log/mogan/mogan-api-err.log | 01:37 |
| wanghao | I use this, and it works in my env | 01:37 |
| wanghao | did you have path /var/log/mogan? | 01:37 |
| liusheng | wanghao: you mean add "2>/var/log/mogan/mogan-api-err.log" after the command ? | 01:37 |
| liusheng | wanghao: yes | 01:38 |
| wanghao | yes | 01:38 |
| wanghao | and then touch the file later | 01:38 |
| liusheng | wanghao: no I use "service devstack@mogan-api restart" to restart my api server | 01:38 |
| liusheng | wanghao: I touched the file | 01:38 |
| wanghao | so you need to add 2>/var/log/mogan/mogan-api-err.log | 01:38 |
| wanghao | and then see the report in this log file | 01:39 |
| liusheng | wanghao: so I cannot add that, unless I modify the systemd script :( | 01:39 |
| liusheng | wanghao: if so, the gmr cannot directly work with systemd ? | 01:39 |
| wanghao | liusheng: I think it can, with the systemd, where is your log? | 01:41 |
| liusheng | wanghao: with "journalctl", e.g. "journalctl -u devstack@mogan-api" | 01:42 |
| wanghao | liusheng: could you see other uwsgi log in journalctl? | 01:46 |
| liusheng | wanghao: yes | 01:46 |
| liusheng | wanghao: just specify the systemd startup script name after -u parameter | 01:47 |
| wanghao | liusheng: I find the devstack@mogan-api.service is not uwsgi way | 01:50 |
| wanghao | liusheng: it just wsgi: /usr/bin/python /usr/local/bin/mogan-api --config-file /etc/mogan/mogan.conf | 01:51 |
| liusheng | wanghao: why ?? | 01:51 |
| liusheng | wanghao: no, maybe you didn't re-stack your env recently, hah | 01:51 |
| wanghao | liusheng: what is your script in devstack@mogan-api.service | 01:52 |
| wanghao | liusheng: never mind, I will update it. | 01:52 |
| liusheng | wanghao: after the https://review.openstack.org/#/c/482057/ patch merged, if you install your devstack env, the api server will run under uwsgi | 01:53 |
| wanghao | I see | 01:53 |
| wanghao | liusheng: it work fine with systemctl | 01:58 |
| wanghao | liusheng: Jul 31 09:58:18 openstack uwsgi[17255]: ==== Green Threads ==== | 01:58 |
| wanghao | I can find the log in my env | 01:59 |
| wanghao | I also use the systemctl way | 01:59 |
| liusheng | wanghao: really, let me try | 01:59 |
| wanghao | if you didn't find the log, you can wait for a seconds, and then touch again | 01:59 |
| wanghao | and you will search the key word guru in your whole log | 02:00 |
| liusheng | wanghao: no I didn't fond that, I am installing a new env, will test in another devstack env | 02:03 |
| wanghao | okay | 02:08 |
| wanghao | https://docs.openstack.org/zaqar/latest/admin/gmr.html | 02:13 |
| wanghao | here is doc for uwsgi mode | 02:13 |
| wanghao | A GMR can be generated by “touch”ing the file which was specified in file_event_handler. The GMR will then output to standard error for that particular process. | 02:13 |
| wanghao | So technically, you can check it in standard error. | 02:13 |
| openstackgerrit | wanghao proposed openstack/mogan master: Support Guru in uwsgi way https://review.openstack.org/487738 | 02:16 |
| openstackgerrit | wanghao proposed openstack/mogan master: Support quota for keypairs in Mogan(part-three) https://review.openstack.org/485461 | 02:37 |
| *** deep-book-gk has joined #openstack-mogan | 02:46 | |
| *** deep-book-gk has left #openstack-mogan | 02:48 | |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add aggregates tests https://review.openstack.org/488296 | 03:54 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add aggregate nodes API https://review.openstack.org/487284 | 03:54 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add placement with mogan devstack plugin https://review.openstack.org/488931 | 04:48 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add aggregates tests https://review.openstack.org/488296 | 04:54 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan-specs master: Node aggregates https://review.openstack.org/470927 | 04:56 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Trivial-Fix: Correct test case name in tempest https://review.openstack.org/489089 | 05:00 |
| wanghao | zhenguo: add some comments on node aggregates spec. | 06:07 |
| zhenguo | wanghao: thanks | 06:07 |
| wanghao | zhenguo: : ) | 06:10 |
| zhenguo | wanghao: please find my replies | 06:17 |
| wanghao | zhenguo: sure | 06:17 |
| * zhenguo brb | 06:26 | |
| litao__ | zhenguo: I found the latest codes has added some flavor columns. How can i apply it to my devstack rapidly | 06:35 |
| litao__ | zhenguo: not restack my devstack. | 06:36 |
| zhenguo | litao__: seems you need to recreat your db | 06:36 |
| litao__ | zhenguo: How? | 06:37 |
| zhenguo | litao__: maybe just drop mogan database, and use dbsync but which will lose all data | 06:37 |
| litao__ | zhenguo: OK, i can restore my db data | 06:38 |
| zhenguo | litao__: ok | 06:38 |
| zhenguo | liusheng: does placement only need a config which has database and keystone config options? | 06:39 |
| wanghao | zhenguo: I see your reply, so I think maybe we need to make the alternaives more clear, since we implement the aggregates on mogan side basically. | 06:40 |
| zhenguo | wanghao: ok, the alternatives should be something like, we don't use placement any more, hah | 06:41 |
| liusheng | zhenguo: I think so, at least the username of placement need to be confiured in [keystone_authtoken] in nova.conf | 06:41 |
| wanghao | zhenguo: or just None, that's better I think | 06:42 |
| zhenguo | liusheng: ok | 06:42 |
| zhenguo | wanghao: hah, ok | 06:42 |
| wanghao | zhenguo: I check the placement API, since nova has host aggregates, so it put the aggregates id to resource provider | 06:43 |
| wanghao | zhenguo: so for Mogan, it's node aggregates like Nova did. | 06:43 |
| zhenguo | wanghao: yes, and we also need to put our node aggregate id there. | 06:43 |
| zhenguo | wanghao: so, with placement nova can't handle bare metal aggregates as well :D | 06:44 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan-specs master: Node aggregates https://review.openstack.org/470927 | 06:45 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add placement with mogan devstack plugin https://review.openstack.org/488931 | 07:00 |
| zhenguo | liusheng: I would like to add an aggregates attribute to flavor, wdyt? | 07:19 |
| liusheng | zhenguo: how to use that ? | 07:20 |
| zhenguo | liusheng: with key/value pairs | 07:20 |
| zhenguo | liusheng: as now we use query_filters of rp list instead of filters node key/values like before | 07:21 |
| zhenguo | liusheng: we can't use an extra specs to include everything | 07:21 |
| zhenguo | liusheng: we need to know which flavor attribute matches which filter | 07:21 |
| liusheng | zhenguo: we don't have extra now | 07:22 |
| zhenguo | liusheng: rp list only support resources, traits, aggregates query filter | 07:22 |
| zhenguo | liusheng: and now our flavor have resource and resource traits | 07:22 |
| liusheng | zhenguo:yes | 07:22 |
| zhenguo | liusheng: so we can't put aggregates info to those field right? | 07:23 |
| zhenguo | liusheng: before, I think we can put aggregate info to traits, but it's hard to use | 07:23 |
| zhenguo | liusheng: and I don't want to add extra specs back | 07:23 |
| liusheng | zhenguo: so we don't need schedule_hints ? | 07:23 |
| zhenguo | liusheng: not related to scheduler_hints I think :D | 07:24 |
| zhenguo | liusheng: only for aggregates | 07:24 |
| zhenguo | liusheng: I want to construct the list rp query filters | 07:24 |
| liusheng | zhenguo: sounds ok, but not sure if we need to add simlar things to flavor in future | 07:25 |
| openstackgerrit | Tao Li proposed openstack/mogan master: Manage existing BMs: Part-2 https://review.openstack.org/481544 | 07:25 |
| zhenguo | liusheng: when there's a new query filter, we may add a new filed in flavor | 07:25 |
| zhenguo | liusheng: but seems that's all, quanlitive, quantitive, and group | 07:26 |
| zhenguo | *qualitative | 07:27 |
| liusheng | zhenguo: for now, it looks like ok :) | 07:27 |
| zhenguo | liusheng: then every specified aggregate key/value will cost a rp list API call | 07:27 |
| zhenguo | liusheng: then will set set1 | set 2| setn to get the final rps | 07:28 |
| zhenguo | liusheng: mogan aggregate metadta key/value may result in a list of aggregates | 07:29 |
| liusheng | zhenguo: that I think is because the placement not matrue yet | 07:29 |
| zhenguo | liusheng: seems not | 07:29 |
| zhenguo | liusheng: one mogan metadata key, may have a list of aggregate | 07:29 |
| zhenguo | liusheng: but not sure whether rp will support all that conditions in future | 07:30 |
| zhenguo | liusheng: for now, we can just do like that | 07:30 |
| liusheng | zhenguo: ok, we can improve if placment provide more functionalities | 07:31 |
| zhenguo | liusheng: ok | 07:31 |
| liusheng | zhenguo: so we won't have something like metadata of flavor ? | 07:32 |
| liusheng | zhenguo: seems resources_traits isn't | 07:32 |
| zhenguo | liusheng: yes | 07:32 |
| zhenguo | liusheng: as we don't know how to handle the metadata | 07:32 |
| zhenguo | liusheng: we only extract the 3 query opetions from flavor | 07:32 |
| zhenguo | liusheng: *options | 07:32 |
| liusheng | zhenguo: seems yes | 07:33 |
| zhenguo | liusheng: it's not like the old way of filtering | 07:33 |
| liusheng | zhenguo: sure, we won't have filter like old ones | 07:34 |
| zhenguo | liusheng: ok, will draft a patch to see | 07:34 |
| liusheng | zhenguo: ok | 07:34 |
| * zhenguo brb | 07:38 | |
| openstackgerrit | Tao Li proposed openstack/mogan master: Manage existing BMs: Part-2 https://review.openstack.org/481544 | 07:43 |
| zhenguo | liusheng: when will we clean up /etc/systemd/system/devstack@placement-api.service? | 08:07 |
| liusheng | zhenguo: seems unstack.sh | 08:07 |
| zhenguo | liusheng: on my test it's not | 08:07 |
| liusheng | zhenguo: oh, maybe in stack.sh, not sure | 08:07 |
| zhenguo | liusheng: and if you changed some parameters, it's not override | 08:08 |
| liusheng | zhenguo: didn't try that | 08:08 |
| openstackgerrit | Merged openstack/mogan master: Remove legacy auth loading https://review.openstack.org/488937 | 08:17 |
| openstackgerrit | Merged openstack/mogan master: Trivial-Fix: Correct KeypairExists exception name https://review.openstack.org/488934 | 08:17 |
| openstackgerrit | Merged openstack/mogan master: Trivial-Fix: Correct test case name in tempest https://review.openstack.org/489089 | 08:17 |
| zhenguo | http://insight.qingcloud.com/#insight-live , QingCloud new bare metal service | 08:22 |
| zhenguo | liusheng, shaohe_feng, wanghao, litao__ ^^ | 08:22 |
| liusheng | zhenguo: thanks for the sharing :) | 08:24 |
| zhenguo | liusheng: hah | 08:24 |
| zhenguo | liusheng: they use the same API :( | 08:27 |
| * litao__ pls review: https://review.openstack.org/#/c/481544/ | 08:28 | |
| liusheng | zhenguo: what ? | 08:28 |
| zhenguo | liusheng: same API for vm and bm | 08:28 |
| liusheng | zhenguo: oh, like Nova does :( | 08:29 |
| zhenguo | liusheng: seems not | 08:29 |
| zhenguo | liusheng: they just follow the same API pattern | 08:29 |
| liusheng | zhenguo: oh, don't fimilar with | 08:31 |
| zhenguo | liusheng: hah | 08:32 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add aggregates tests https://review.openstack.org/488296 | 08:37 |
| zhenguo | liusheng: when to create this /etc/systemd/system/devstack@placement-api.service ? | 08:38 |
| zhenguo | liusheng: after manually clean up it, it's not created with stack.sh | 08:38 |
| zhenguo | liusheng: now I can succeed to start placement-api with uwsgi | 08:39 |
| zhenguo | liusheng: but can't under systemd | 08:39 |
| liusheng | zhenguo: really, it should be created by stack.sh | 08:39 |
| liusheng | zhenguo: let me change | 08:39 |
| liusheng | zhenguo: check | 08:39 |
| liusheng | zhenguo: did you try: systemctl daemon-reload | 08:45 |
| zhenguo | liusheng: I'm restacking now | 08:45 |
| zhenguo | liusheng: still, devstack@placement-api isn't created | 09:18 |
| zhenguo | liusheng: /etc/systemd/system/devstack@placement-api.service | 09:18 |
| liusheng | zhenguo: I guess may devstack will handle by "enabled services" | 09:21 |
| liusheng | zhenguo: you just start placement in mogan devstack plugin | 09:21 |
| zhenguo | liusheng: how does plugin work | 09:21 |
| liusheng | zhenguo: devstack may will create the syetemd script by the name of project in enabled services | 09:22 |
| liusheng | zhenguo: let me check | 09:23 |
| zhenguo | liusheng: seems yes, all mogan services there | 09:23 |
| liusheng | zhenguo: may you can call write_user_unit_file method in our plugin | 09:26 |
| *** wanghao has quit IRC | 09:26 | |
| liusheng | zhenguo: it is defined in devstack/common-functions | 09:27 |
| zhenguo | liusheng: let me check | 09:27 |
| liusheng | zhenguo: devstack/functions-common | 09:27 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add aggregates filed to flavor https://review.openstack.org/489151 | 09:29 |
| * zhenguo brb | 09:31 | |
| zhenguo | liusheng: yes, it will check whether service enabled in run_process | 09:45 |
| zhenguo | :q | 09:46 |
| zhenguo | liusheng: stop_process will also check enabled services | 09:52 |
| liusheng | zhenguo: but i didn | 09:53 |
| liusheng | zhenguo: didn't found any place to generate the file in devstack | 09:53 |
| liusheng | zhenguo: besides write_user_unit_file method function ,but the function only be called once for etcd3 | 09:54 |
| zhenguo | liusheng: no, it will called under run_process | 09:54 |
| liusheng | zhenguo: oh | 09:54 |
| zhenguo | liusheng: I will add a own run_proces sand stop_process method for placement | 09:56 |
| liusheng | zhenguo: may you can leverage the write_user_unit_file method :) | 09:57 |
| zhenguo | liusheng: not only that, | 09:57 |
| zhenguo | liusheng: run process will do other things | 09:57 |
| zhenguo | liusheng: you can check under the is_service_enabled | 09:57 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add placement with mogan devstack plugin https://review.openstack.org/488931 | 09:58 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan-specs master: Update new-flavor specs https://review.openstack.org/489167 | 10:23 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add resource aggregates filed to flavor https://review.openstack.org/489151 | 10:28 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add aggregates tests https://review.openstack.org/488296 | 10:38 |
| zhenguo | liusheng: hah, it finally passed the gate https://review.openstack.org/#/c/488931/ | 10:41 |
| liusheng | zhenguo: awesome :D | 10:42 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add placement with mogan devstack plugin https://review.openstack.org/488931 | 10:43 |
| zhenguo | liusheng: will remove the depends-on to test if it's work when user enable placement-api service with nova | 10:43 |
| liusheng | zhenguo: ok, let's see it | 10:44 |
| zhenguo | liusheng: ok | 10:44 |
| zhenguo | liusheng: please help to test this https://review.openstack.org/#/c/484690/ when you got time | 10:45 |
| liusheng | zhenguo: ok | 10:47 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan-specs master: Node aggregates https://review.openstack.org/470927 | 10:48 |
| * zhenguo brb | 11:03 | |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add placement with mogan devstack plugin https://review.openstack.org/488931 | 11:19 |
| *** litao__ has quit IRC | 11:50 | |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Add placement with mogan devstack plugin https://review.openstack.org/488931 | 12:08 |
| * zhenguo away | 12:09 | |
| *** openstackgerrit has quit IRC | 14:33 | |
| *** openstackgerrit has joined #openstack-mogan | 16:20 | |
| openstackgerrit | Spencer Yu proposed openstack/python-moganclient master: Switch from oslosphinx to openstackdocstheme https://review.openstack.org/489269 | 16:20 |
| *** liusheng has quit IRC | 18:09 | |
| *** liusheng has joined #openstack-mogan | 18:09 | |
| *** harlowja has joined #openstack-mogan | 18:11 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!