| *** wanghao has joined #openstack-mogan | 00:32 | |
| *** openstackgerrit has joined #openstack-mogan | 00:51 | |
| openstackgerrit | wanghao proposed openstack/mogan master: Manage existing BMs: Part-1 https://review.openstack.org/479660 | 00:51 |
|---|---|---|
| openstackgerrit | wanghao proposed openstack/mogan master: Manage existing BMs: Part-1 https://review.openstack.org/479660 | 00:56 |
| zhenguo | morning mogan! | 00:59 |
| openstackgerrit | wanghao proposed openstack/mogan master: Refactor exception raise by using mogan exception https://review.openstack.org/475971 | 01:00 |
| *** liusheng has quit IRC | 01:04 | |
| *** liusheng has joined #openstack-mogan | 01:09 | |
| wanghao | morning! | 01:20 |
| wanghao | I'm back! | 01:20 |
| zhenguo | wanghao: o/ | 01:25 |
| zhenguo | liusheng: hi, maybe we should combine the 2ed and 3ed placement patches together, wdyt? | 01:26 |
| zhenguo | liusheng: as people may check the git log and got problems if only get 2ed patch | 01:27 |
| liusheng | zhenguo: ... maybe I should putted all the things in one patch | 01:36 |
| zhenguo | liusheng: hah | 01:36 |
| zhenguo | liusheng: the concern is that, the 2ed patch will break mogan if it merged without the following up patch, as it not consume resources in placement | 01:39 |
| liusheng | zhenguo: yes, I know that | 01:40 |
| liusheng | zhenguo: will merge the two | 01:40 |
| zhenguo | liusheng: when split big patches to small one, we should avoid that | 01:40 |
| zhenguo | liusheng: thanks | 01:40 |
| *** yushb has joined #openstack-mogan | 01:41 | |
| openstackgerrit | liusheng proposed openstack/mogan master: Clean the compute_node and compute_port objects and db interfaces https://review.openstack.org/478406 | 01:50 |
| openstackgerrit | liusheng proposed openstack/mogan master: Get rid of listing availability zone api and clean some legacy code https://review.openstack.org/478403 | 01:50 |
| openstackgerrit | liusheng proposed openstack/mogan master: Clean the methods about updating node resources to Mogan db https://review.openstack.org/478357 | 01:50 |
| openstackgerrit | liusheng proposed openstack/mogan master: Get rid of node listing api of Mogan https://review.openstack.org/478361 | 01:50 |
| openstackgerrit | liusheng proposed openstack/mogan master: Refactor the scheduler to use placement service https://review.openstack.org/477426 | 01:50 |
| openstackgerrit | Tao Li proposed openstack/mogan master: [WIP] Manage existing BMs: Part-2 https://review.openstack.org/481544 | 01:58 |
| *** wanghao_ has joined #openstack-mogan | 02:01 | |
| *** wanghao has quit IRC | 02:03 | |
| zhenguo | liusheng: seems the unstack will leave mogan services running | 02:13 |
| liusheng | zhenguo: really ? why ? | 02:13 |
| zhenguo | liusheng: not sure, but I just run unstack, and only mogan services left | 02:14 |
| liusheng | zhenguo: how you started mogan ? | 02:14 |
| zhenguo | liusheng: with stack | 02:14 |
| liusheng | zhenguo: the newest ? with systemd ? | 02:14 |
| zhenguo | liusheng: yes | 02:14 |
| liusheng | zhenguo: systemctl | 02:14 |
| zhenguo | liusheng: oh, just run unstack again, and it disappeared | 02:15 |
| zhenguo | liusheng: not sure why | 02:15 |
| liusheng | zhenguo: hah | 02:15 |
| zhenguo | liusheng:ｈａｈ | 02:15 |
| wxy | magic! | 02:15 |
| zhenguo | wxy: lol | 02:15 |
| zhenguo | liusheng: instead of adding aggregates information to flavor, maybe it's better to let users specify that? | 02:43 |
| liusheng | zhenguo: in schedule_hints ? | 02:44 |
| zhenguo | liusheng: no, add a new parameters | 02:44 |
| liusheng | zhenguo: but it is related with scheduling | 02:45 |
| zhenguo | liusheng: yes | 02:45 |
| liusheng | zhenguo: why don't put it in schedule_hints | 02:46 |
| zhenguo | liusheng: seems a good alternatives, just dont' like to put it into flavors | 02:46 |
| zhenguo | liusheng: but if so, we need to let users be able to show the aggregates infor | 02:46 |
| liusheng | zhenguo: yes, another proposal, we need to persistent request_spec, which should include schedule_hints | 02:47 |
| zhenguo | liusheng: why need to save scheduler_hints? | 02:47 |
| liusheng | zhenguo: oh, seems we don't have migration | 02:48 |
| liusheng | zhenguo: nova has improved that | 02:48 |
| zhenguo | liusheng: yes, we don't have migrations | 02:49 |
| zhenguo | liusheng: maybe we can do it in future, hah | 02:49 |
| liusheng | zhenguo: yes, not sure if we have the rquirement | 02:49 |
| zhenguo | liusheng: seems not yet | 02:50 |
| liusheng | zhenguo: you mean add a attribute to server object ? | 02:50 |
| zhenguo | liusheng: for what? | 02:51 |
| liusheng | zhenguo: for your new parameter | 02:51 |
| zhenguo | liusheng: I don't know, will update the node aggregates spec by today, maybe we can discuss the detail there | 02:52 |
| liusheng | zhenguo: ok, sure | 02:53 |
| zhenguo | liusheng: as I understand, we still have to maintain the node aggregates and metadata info on mogan side, placement just save the group relationship | 02:53 |
| zhenguo | liusheng: we need to filter the metadata ourselves and the got a aggregates list, then when call placement to get RPs, we can pass a member_of with the filtered aggreggates | 02:54 |
| liusheng | zhenguo: if so we still to keep consistency between mogan node aggregates metadata and placement aggregate ? | 03:03 |
| zhenguo | liusheng: yes, we create the aggreate to placement first then got the uuid and save it to mogan aggregate | 03:04 |
| liusheng | zhenguo: we need to expose apis for node aggregates ? | 03:06 |
| zhenguo | liusheng: yes | 03:06 |
| liusheng | zhenguo: e.e. node aggregate creating | 03:06 |
| zhenguo | liusheng: and aggregate metadata | 03:07 |
| liusheng | zhenguo: make sence | 03:07 |
| liusheng | zhenguo: make sense | 03:07 |
| zhenguo | liusheng: but not sure how to deal with the nodes | 03:07 |
| zhenguo | liusheng: seems we don't need to save nodes and aggregates relationship on mogan side | 03:08 |
| liusheng | zhenguo: don't need, that is in placement | 03:08 |
| zhenguo | liusheng: yes, but we need two proxy API :D | 03:08 |
| liusheng | zhenguo: yes | 03:09 |
| zhenguo | liusheng: but maybe it's ok | 03:09 |
| zhenguo | liusheng: wait for my specs | 03:09 |
| liusheng | zhenguo: thanks | 03:09 |
| zhenguo | liusheng: np | 03:09 |
| zhenguo | liusheng: you can think more about the server group and scheduler hints part | 03:09 |
| liusheng | zhenguo: ok | 03:10 |
| zhenguo | liusheng: seems they need a more smart server group, not like nova's affiniy and anti-affinity for only same host and not same host | 03:10 |
| liusheng | zhenguo: I am just think what our server group for | 03:11 |
| zhenguo | liusheng: I'm also not very clear, hah | 03:12 |
| zhenguo | liusheng: oh, for az, should we also save it with our aggregates metadata? | 03:12 |
| liusheng | zhenguo: we don't have affiniy and anti-affinity | 03:13 |
| zhenguo | liusheng: hah | 03:13 |
| zhenguo | liusheng: as we don't have a notion of rack | 03:13 |
| zhenguo | liusheng: and rack can be grouped by aggregates | 03:14 |
| liusheng | zhenguo: does chassis in Ironic means that ? | 03:14 |
| zhenguo | liusheng: no, ironic wants to get rid of chassis for a long time | 03:14 |
| liusheng | zhenguo: if so, seem server group is meaningless, at least for now | 03:17 |
| zhenguo | liusheng: does nova support rack affinity and anti-affinity | 03:17 |
| liusheng | zhenguo: you mean Nova + Ironic? | 03:18 |
| zhenguo | liusheng: no, for vms | 03:18 |
| liusheng | zhenguo: I didn't found that | 03:18 |
| zhenguo | liusheng: hah, | 03:18 |
| Kevin_Zheng | ... vms doesn't have rack context | 03:18 |
| Kevin_Zheng | you can add host agg for all hosts in one rack | 03:19 |
| zhenguo | so the server group on nova is just host affinity and anti-affinity? | 03:19 |
| zhenguo | what the metadata of server group used for? | 03:19 |
| liusheng | zhenguo: that also looks like metadata for node aggregates | 03:20 |
| zhenguo | liusheng: so the filter will also consider of server group metadata? | 03:20 |
| liusheng | zhenguo: may only for affinity and anti-affinity | 03:21 |
| zhenguo | seems yes | 03:21 |
| * zhenguo brb | 03:22 | |
| openstackgerrit | klyang proposed openstack/mogan master: switch from oslosphinx to openstackdocstheme https://review.openstack.org/481992 | 03:32 |
| zhenguo | liusheng: maybe server group affinity and anti-affinity should be associated with aggregates | 03:39 |
| zhenguo | liusheng: AZ is a special metadata of aggregates, maybe rack should also be | 03:40 |
| zhenguo | liusheng: AZ, DC, ROW, RACk... | 03:40 |
| liusheng | zhenguo: how to get rack info | 03:46 |
| zhenguo | liusheng: defined by operators when set aggregates metadata | 03:47 |
| zhenguo | liusheng: maybe , hah just a propsal, need to think more about this | 03:49 |
| *** yushb has quit IRC | 04:20 | |
| *** wanghao_ has quit IRC | 04:20 | |
| *** wanghao has joined #openstack-mogan | 04:20 | |
| *** yushb has joined #openstack-mogan | 04:37 | |
| *** yushb has quit IRC | 04:54 | |
| *** yushb has joined #openstack-mogan | 06:05 | |
| openstackgerrit | Xinran WANG proposed openstack/mogan master: Detach interface for server https://review.openstack.org/461767 | 06:06 |
| Xinran | liusheng, I have some quesiton of your comments of this patch: https://review.openstack.org/#/c/473323/ | 06:08 |
| Xinran | liusheng, pls check my replys | 06:09 |
| liusheng | Xinran: ok, thanks | 06:09 |
| Xinran | liusheng, thanks for your comments :) | 06:09 |
| liusheng | Xinran: replied you | 06:16 |
| Xinran | liusheng, thanks | 06:16 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan-specs master: Node aggregates https://review.openstack.org/470927 | 07:07 |
| zhenguo | liusheng: just draft the node aggregates spec which leverages placement service | 07:08 |
| liusheng | zhenguo: thanks, will review | 07:08 |
| zhenguo | liusheng: thansk | 07:08 |
| *** harlowja has quit IRC | 07:11 | |
| zhenguo | Xinran: seems you need to rebase this https://review.openstack.org/#/c/461767/ as the tempest failed with a bug https://bugs.launchpad.net/mogan/+bug/1702822 | 07:12 |
| openstack | Launchpad bug 1702822 in Mogan "tempest gate has been broken by missing lib greening" [Critical,Fix released] - Assigned to Liusheng (liusheng) | 07:12 |
| liusheng | how openstack can say this in IRC :-P | 07:13 |
| zhenguo | liusheng: not sure, hah | 07:13 |
| liusheng | zhenguo: lol | 07:13 |
| liusheng | oh, Xinran just pasted the link | 07:14 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan master: Detach interface for server https://review.openstack.org/461767 | 07:14 |
| *** wanghao_ has joined #openstack-mogan | 07:15 | |
| *** wanghao has quit IRC | 07:17 | |
| zhenguo | liusheng: do you know where can I find the placement log | 07:23 |
| liusheng | zhenguo: journalctl -u devstack@placement-api | 07:24 |
| zhenguo | liusheng: thanks | 07:24 |
| liusheng | zhenguo: np :D | 07:25 |
| *** litao__ has joined #openstack-mogan | 07:25 | |
| zhenguo | liusheng I don't like the journlctl way... | 07:34 |
| liusheng | zhenguo: me too :( | 07:35 |
| zhenguo | liusheng: hah | 07:35 |
| liusheng | zhenguo: you can still use screen, but it has be marked deprecated | 07:35 |
| zhenguo | liusheng: I would like to killl the mogan process and manually start it to get logs | 07:36 |
| liusheng | zhenguo: now devstack switch to use systemctl to start services | 07:36 |
| zhenguo | liusheng: I don't like it | 07:37 |
| liusheng | zhenguo: and seems it don't store logs, the journlctl can only see the recent logs | 07:37 |
| liusheng | zhenguo: hah | 07:37 |
| zhenguo | liusheng: it's hard to debug | 07:37 |
| openstackgerrit | Tao Li proposed openstack/mogan master: Fix the stuck in *ing in powering servers. https://review.openstack.org/476862 | 07:38 |
| liusheng | zhenguo: yes, so I use rpdb than pdb to debug now | 07:38 |
| zhenguo | liusheng: hah | 07:39 |
| zhenguo | liusheng: I never used it before | 07:39 |
| liusheng | zhenguo: hah, it is very like pdb | 07:39 |
| zhenguo | liusheng: will try if needed, thanks | 07:39 |
| openstackgerrit | Tao Li proposed openstack/mogan master: Fix reverting the network when plug vif failed https://review.openstack.org/477440 | 07:41 |
| liusheng | zhenguo: I am trying to run mogan-api under uwsgi | 07:43 |
| zhenguo | liusheng: cool | 07:43 |
| zhenguo | liusheng: Is there some problems? | 07:43 |
| liusheng | zhenguo: just start to try, hah | 07:43 |
| zhenguo | liusheng: hah | 07:43 |
| zhenguo | liusheng: on my test, there's some problems when do multi creations with your placement patch. | 08:00 |
| zhenguo | liusheng: the allocation will all use the same server uuid, | 08:00 |
| liusheng | zhenguo: you mean the allocations in placement table ? | 08:01 |
| zhenguo | liusheng: yes | 08:01 |
| zhenguo | liusheng: see my comments on the patch | 08:01 |
| liusheng | zhenguo: will check | 08:01 |
| * zhenguo brb | 08:03 | |
| openstackgerrit | Tao Li proposed openstack/mogan master: Fix the stuck in *ing in powering servers. https://review.openstack.org/476862 | 08:08 |
| litao__ | wanghao_: see my comments | 08:08 |
| wanghao_ | litao__: ok | 08:12 |
| wanghao_ | litao__: I see your point, I mean why not rollback status to original status before we do this action? | 08:17 |
| wanghao_ | not just 'fail' status | 08:17 |
| *** wanghao_ has quit IRC | 08:18 | |
| *** wanghao has joined #openstack-mogan | 08:18 | |
| *** wanghao has quit IRC | 08:19 | |
| *** wanghao has joined #openstack-mogan | 08:19 | |
| *** wanghao has quit IRC | 08:20 | |
| *** wanghao has joined #openstack-mogan | 08:20 | |
| *** wanghao has quit IRC | 08:21 | |
| *** wanghao has joined #openstack-mogan | 08:21 | |
| *** wanghao has quit IRC | 08:21 | |
| *** wanghao has joined #openstack-mogan | 08:22 | |
| *** wanghao has quit IRC | 08:22 | |
| *** wanghao has joined #openstack-mogan | 08:22 | |
| *** wanghao has quit IRC | 08:23 | |
| *** wanghao has joined #openstack-mogan | 08:23 | |
| *** wanghao has quit IRC | 08:24 | |
| *** wanghao has joined #openstack-mogan | 08:24 | |
| *** wanghao has quit IRC | 08:25 | |
| *** wanghao has joined #openstack-mogan | 08:25 | |
| openstackgerrit | liusheng proposed openstack/mogan master: WIP: suuport running api under uWSGI https://review.openstack.org/482057 | 08:47 |
| openstackgerrit | Merged openstack/python-moganclient master: Updated from global requirements https://review.openstack.org/478082 | 08:53 |
| openstackgerrit | Merged openstack/python-moganclient master: Make the floating ip can be show in netinfo command https://review.openstack.org/475625 | 08:53 |
| openstackgerrit | Xinran WANG proposed openstack/mogan master: Detach interface for server https://review.openstack.org/461767 | 09:09 |
| openstackgerrit | Zhenguo Niu proposed openstack/mogan-specs master: Node aggregates https://review.openstack.org/470927 | 09:09 |
| openstackgerrit | liusheng proposed openstack/mogan master: WIP: suuport running api under uWSGI https://review.openstack.org/482057 | 09:24 |
| zhenguo | liusheng: resource traits seems should be a list of string in flavor | 09:42 |
| zhenguo | liusheng: flavor resources can be a list of key/pairs, but traits seems don't need such pair | 09:42 |
| zhenguo | *key/value | 09:43 |
| liusheng | zhenguo: what's the effect ? | 09:43 |
| zhenguo | liusheng: just think traits should be a string to describe it instead of a=b | 09:44 |
| liusheng | zhenguo: looks like tags | 09:45 |
| zhenguo | liusheng: ye | 09:45 |
| liusheng | zhenguo: not sure why trait desgined that | 09:45 |
| zhenguo | liusheng: it's our flavor using such key/value pair, nova use a string, hah | 09:46 |
| liusheng | zhenguo: do you think how about fix the problem of multi-creation of the placement patches ? | 09:47 |
| liusheng | zhenguo: how about fix it in a following patch | 09:48 |
| zhenguo | liusheng: how | 09:48 |
| liusheng | zhenguo: since the servers uuids don't be passed to scheduler previously | 09:48 |
| liusheng | zhenguo: I mean fix it in a separated patch | 09:48 |
| zhenguo | liusheng: before the placement patches landed? | 09:49 |
| liusheng | zhenguo: after | 09:49 |
| zhenguo | liusheng: so the placement patch will break mogan | 09:49 |
| zhenguo | liusheng: I prefer to fix it with the patch, as it do effect the allocations | 09:50 |
| liusheng | zhenguo: a following patch, you can treat it as a part of this patches | 09:50 |
| liusheng | zhenguo: alright.. | 09:50 |
| zhenguo | liusheng: why not just fix with that patch? | 09:51 |
| zhenguo | liusheng: as we discussed, each patch should keep mogan work | 09:51 |
| liusheng | zhenguo: it didn't broken mogan.. | 09:51 |
| liusheng | zhenguo: ok | 09:51 |
| zhenguo | liusheng: ... | 09:51 |
| zhenguo | liusheng: hah | 09:52 |
| zhenguo | liusheng: on my test, when do multi creation, then remove on of them, all allocations got to be cleaned, then issue one create request, it will scheduled to a node which already deployed with a server | 09:52 |
| zhenguo | *one of them | 09:53 |
| liusheng | zhenguo: I am not sure if the problem also existing before these patches | 09:53 |
| zhenguo | liusheng: no | 09:53 |
| zhenguo | liusheng: only when using allocations | 09:54 |
| liusheng | zhenguo: previously, mogan don't set server_id to node object in scheduler ? | 09:54 |
| zhenguo | liusheng: IIRC, not in scheduler | 09:55 |
| liusheng | zhenguo: i see | 09:55 |
| openstackgerrit | liusheng proposed openstack/mogan master: Clean the compute_node and compute_port objects and db interfaces https://review.openstack.org/478406 | 10:01 |
| openstackgerrit | liusheng proposed openstack/mogan master: Get rid of listing availability zone api and clean some legacy code https://review.openstack.org/478403 | 10:01 |
| openstackgerrit | liusheng proposed openstack/mogan master: Clean the methods about updating node resources to Mogan db https://review.openstack.org/478357 | 10:01 |
| openstackgerrit | liusheng proposed openstack/mogan master: Get rid of node listing api of Mogan https://review.openstack.org/478361 | 10:01 |
| openstackgerrit | liusheng proposed openstack/mogan master: Refactor the scheduler to use placement service https://review.openstack.org/477426 | 10:01 |
| liusheng | zhenguo: can you help to test the patches in multi-creating case | 10:02 |
| liusheng | zhenguo: my env has a problem :( | 10:02 |
| openstackgerrit | Merged openstack/mogan master: switch from oslosphinx to openstackdocstheme https://review.openstack.org/481992 | 10:04 |
| *** yushb has quit IRC | 10:23 | |
| zhenguo | liusheng: sure | 10:46 |
| liusheng | zhenguo: thanks | 10:46 |
| zhenguo | liusheng: do you know why we still can't access the physical env | 10:47 |
| liusheng | zhenguo: I can acess the bmc dashboard, and tried to reboot the server, but cannot work | 10:49 |
| zhenguo | liusheng: seems after the vpn restart, we can't access it | 10:50 |
| liusheng | zhenguo: yes | 10:51 |
| zhenguo | liusheng: maybe need to wait zhaobo back | 10:51 |
| liusheng | zhenguo: let me try again | 10:52 |
| zhenguo | liusheng: hah | 10:52 |
| liusheng | zhenguo: zhaobo has off from work because he just has a baby :D | 10:53 |
| zhenguo | liusheng: hah, seems take a long time... I remember since last week | 10:53 |
| liusheng | zhenguo: yes | 10:54 |
| liusheng | zhenguo: let me try | 10:54 |
| zhenguo | liusheng: ok | 10:54 |
| Xinran | Could you pls help me to debug detach interface patch. It runs well in my local env, really have no idea why it doesn't work with jeckins. :( | 10:54 |
| zhenguo | Xinran: liusheng should be able to help with the tempest stuff | 10:55 |
| liusheng | ok, will try to review, but my env has been broken :( | 10:56 |
| Xinran | zhenguo, I'm cainiao | 10:56 |
| zhenguo | Xinran: hah | 10:57 |
| Xinran | liusheng, Thanks a lot! | 10:58 |
| liusheng | zhenguo: the server is rebooting again and again, because it cannot find a bootable device, it is verid | 11:02 |
| zhenguo | liusheng: the server failed... | 11:02 |
| openstackgerrit | liusheng proposed openstack/mogan master: WIP: suuport running api under uWSGI https://review.openstack.org/482057 | 11:29 |
| zhenguo | liusheng: on my test, multi-create still got the allocations with same consumer_id | 11:31 |
| liusheng | zhenguo: I just tested, it don't got two same consumer_id | 11:50 |
| liusheng | zhenguo: even my env broken and will create failed, but it can create two items in allocations table | 11:50 |
| liusheng | zhenguo: with different two consumer id | 11:51 |
| zhenguo | liusheng: let me restart the mogan services and test again | 11:57 |
| *** litao__ has quit IRC | 11:58 | |
| *** yushb has joined #openstack-mogan | 13:38 | |
| *** liusheng has quit IRC | 15:21 | |
| *** liusheng has joined #openstack-mogan | 15:22 | |
| *** yushb has quit IRC | 16:00 | |
| *** harlowja has joined #openstack-mogan | 16:37 | |
| *** Kevin_Zheng has quit IRC | 17:33 | |
| *** wanghao has quit IRC | 23:52 | |
| *** wanghao has joined #openstack-mogan | 23:53 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!