Friday, 2017-07-28

*** wanghao has joined #openstack-mogan00:39
*** liusheng has quit IRC00:52
*** litao__ has joined #openstack-mogan00:57
*** liusheng has joined #openstack-mogan00:57
*** wanghao has quit IRC00:59
*** wanghao has joined #openstack-mogan01:00
zhenguomorning mogan!01:05
zhenguoliusheng: seems we misused rp aggregate, set will replace it and set [] just means remove all aggregates01:08
*** harlowja has quit IRC01:10
liushengzhenguo: so that is a good news ?01:10
liushengzhenguo: hah01:10
zhenguoliusheng: yes, I have updated the patch01:10
zhenguoliusheng: it works now01:10
liushengzhenguo: cool01:10
litao__zhenguo,liusheng ,shaohe_feng ,wanghao pls review patch:
wanghaolitao__: cool01:48
zhenguolitao__: ok01:49
*** wanghao_ has joined #openstack-mogan01:58
*** wanghao has quit IRC02:01
liushenglitao__: just left some comments in your patch02:08
* zhenguo brb02:11
litao__liusheng: ok,  thanks for your -1  :)02:26
liushenglitao__: lol02:26
zhenguoliusheng: maybe you can re+A the aggregates patches if no objections, I will add a following up tests patch02:35
liushengzhenguo: ok02:36
zhenguoliusheng: thanks02:36
*** wanghao_ has quit IRC02:39
openstackgerritMerged openstack/mogan master: Updated from global requirements
*** wanghao has joined #openstack-mogan02:44
*** wanghao has quit IRC02:51
*** wanghao has joined #openstack-mogan02:51
*** wanghao has quit IRC03:00
litao__liusheng:  could we set  multiple resource classes in flavor?03:03
liushenglitao__: that is possible03:09
litao__liusheng:  what do you mean 'fully match the resource_class and flavor['resources']'?03:12
liushenglitao__: when we create a server, we can specify a flavor with resources field. scheduler will select a node that must match all the condictons in flavor['resources']. so I think we should apply that in server atoptation03:13
liushenglitao__: I found you use a "in" in your patch03:14
openstackgerritMerged openstack/mogan master: Adds aggregates DB model and API
openstackgerritMerged openstack/mogan master: Add aggregate object
litao__liusheng: But adoption is a little different from creating servers. The baremetal node is user-specific, so we just need to ensure the baremetal node's resource class is in flavor resources.03:23
liushenglitao__: why ? if so you need to chose a flavor that can match the node03:23
litao__liusheng: Because the flavor is created by user and specified in adopting nodes.03:26
openstackgerritZhenguo Niu proposed openstack/mogan master: Clean up some unused codes
liushengzhenguo: so when create the flavor, you need to keep the resources field can match the nodes'03:27
liushenglitao__: ^^ sorry , sent to wrong person03:27
liushengzhenguo: may zhenguo can also give a suggestion, hah03:27
zhenguoliusheng: I need to read the patch first :D03:28
litao__liusheng: Currently yes03:28
litao__zhenguo: Give us advice03:28
liushenglitao__: we'd better to avoid different behaviors about matching flavor and nodes between creating and adoptation03:29
litao__liusheng: yes, need more discussion03:30
litao__liusheng: have a lunch first.03:31
zhenguoliusheng, litao__: currently, how do we handle the match if we specify flavor resources like GOLD:1, SILVER:1?03:31
liushengzhenguo: that looks a bad flavor, hah03:31
zhenguoliusheng: hah03:31
zhenguoliusheng: but that could be possible as we don't have limitations03:32
liushengzhenguo: as we are Mogan, for Nova, the resources can be like: RAM_MB:1024, DISK_GB:10003:32
liushengzhenguo: in Nova, it need to fully match the flavor and compute node03:32
liushengzhenguo: but when request to placement, it will try to match all the conditions03:33
zhenguoliusheng: yes, can that mean both GOLD and SILVER are ok?03:33
liushengzhenguo: unless we add the logical in our code, the placement cannot perform that03:34
zhenguoliusheng: if more than one resources specified, we only get the first one now?03:34
*** wanghao has joined #openstack-mogan03:35
liushengzhenguo: no, placement will try to select a node with "and" all the conditions in flavor['resources']03:35
zhenguoliusheng: so we just pass down the flavor resources to placement now, right?03:36
liushengzhenguo: I am not sure if Mogan will have the siatuation, but maybe in future, the resources can be putted into other resources than node ?03:36
liushengzhenguo: yes03:36
zhenguoliusheng: that's ok03:36
zhenguoliusheng: as we may use nested resource in future03:36
liushengzhenguo: ok03:37
* zhenguo brb03:38
liushengzhenguo: is there possible any shared hardware can be freely allocated for node provision ?03:38
*** wanghao has quit IRC03:39
*** wanghao has joined #openstack-mogan03:41
zhenguoliusheng: you mean someting like accelerators provided by Cyborg?03:50
liushengzhenguo: not sure, but like some hardware that isn't belong to a node, but can be allocated to any node04:00
*** harlowja has joined #openstack-mogan04:32
openstackgerritZhenguo Niu proposed openstack/mogan master: Retrieve availability zone from aggregate
openstackgerritZhenguo Niu proposed openstack/mogan master: Add aggregate nodes API
openstackgerritZhenguo Niu proposed openstack/mogan master: Add node list support
openstackgerritZhenguo Niu proposed openstack/mogan master: Add aggregate API
openstackgerritZhenguo Niu proposed openstack/mogan master: Add aggregates tests
openstackgerritliusheng proposed openstack/mogan master: WIP: add server groups db models and api
openstackgerritZhenguo Niu proposed openstack/mogan master: Add aggregate nodes API
openstackgerritZhenguo Niu proposed openstack/mogan master: Add aggregates tests
*** harlowja has quit IRC05:28
openstackgerritZhenguo Niu proposed openstack/mogan-specs master: Node aggregates
*** harlowja has joined #openstack-mogan05:41
openstackgerritwanghao proposed openstack/mogan master: Support quota for keypairs in Mogan(part-three)
litao__liusheng:  so what your suggestion for matching the flavor resource classes.06:14
liushenglitao__: just ensure fully match the flavor06:16
litao__liusheng: Also comparing the values?06:17
liushenglitao__: sure06:17
* litao__ liusheng: OK, I will try06:18
zhenguoliusheng: can you help to check whether this work ? thanks06:36
liushengzhenguo: ok06:36
* zhenguo brb06:38
openstackgerritliusheng proposed openstack/mogan master: Add server groups db models and api
*** harlowja has quit IRC06:48
*** wanghao_ has joined #openstack-mogan06:48
openstackgerritliusheng proposed openstack/mogan master: Fix uwsgi doc to avoid misunderstanding
*** wanghao has quit IRC06:52
openstackgerritZhenguo Niu proposed openstack/mogan master: Add aggregate nodes API
litao__liusheng: For ironic, the  flavor resource class value must be 1, right?07:48
liushenglitao__: as discussed above, for now, it is, but not sure in future07:49
litao__liusheng: OK07:50
*** wanghao_ has quit IRC07:57
zhenguolitao__, liusheng: for node resource_class is must be 1, as we just want 1 node08:00
zhenguolitao__, liusheng: but we may introduce nics resource_class later, then it can be more than 108:00
liushengzhenguo: yes, so I said for now. hah08:01
zhenguoliusheng: hah, yes, and we can't claim more than 1 node for a server08:01
liushengzhenguo: sure08:01
liushengzhenguo: btw, about the resource updating process08:01
liushengzhenguo: we disscussed yestoday, it seems impossible a rp in placement, but the node doesn't exist in ironic08:02
zhenguoliusheng: not exist in ironic means with bad status?08:03
liushengzhenguo: oh sorry, I mean my previous concern about when rescheduling, we need to keep allocations but delete rp08:03
liushengzhenguo: so maybe re-reconsider the ps1 of patch ? lol08:04
zhenguoliusheng: seems placement doesn't support to delete rp but left allocations, right?08:05
liushengzhenguo: yes, so must delete allocations before deleting rp08:06
litao__thank you two :)08:07
zhenguoliusheng: then when it come back, what will happen?08:08
zhenguoliusheng: seems there may be a dead server still associate with the node, right?08:10
liushengzhenguo: hmm.. yes, seems need to consider again08:10
*** wanghao has joined #openstack-mogan08:12
liushengzhenguo: how aobut just skip this rp, if there is a server associated with it ?08:12
zhenguoliusheng: how?08:16
liushengzhenguo: before delete rp in placement, query server by the rp, if there is a server associated with the rp, just skip08:17
liushengzhenguo: wdyt ?08:17
zhenguoliusheng: skip means not delete it?08:17
liushengzhenguo: yes,08:17
*** openstackgerrit has quit IRC08:18
zhenguoliusheng: and keep the allocations08:18
liushengzhenguo: yes08:18
zhenguoliusheng: make sense to me08:18
liushengzhenguo: ok, will update the patch08:19
zhenguoliusheng: okay08:20
zhenguoliusheng: how's the tempest without nova going?08:22
liushengzhenguo: I need to re-stack my env to test08:25
zhenguoliusheng: ok, and please help to test the aggregates when you got time :D08:25
liushengzhenguo: ok08:25
liushengzhenguo: the first server group patch just uploaded :)08:26
zhenguoliusheng: hah, then the folowing up patches maybe related to aggregates08:26
liushengzhenguo: seems not yet, hah08:27
zhenguoliusheng: lol08:27
zhenguoliusheng: before you can specify server groups08:27
liushengzhenguo: yes08:27
*** openstackgerrit has joined #openstack-mogan09:15
openstackgerritZhenguo Niu proposed openstack/mogan master: Add aggregate nodes API
*** wanghao has quit IRC09:29
openstackgerritliusheng proposed openstack/mogan master: Improve resources updating process
*** wanghao has joined #openstack-mogan11:57
*** litao__ has quit IRC11:59
*** wanghao has quit IRC11:59
*** wanghao has joined #openstack-mogan12:08
*** wanghao has quit IRC12:09
openstackgerritMerged openstack/mogan master: Clean up some unused codes
*** harlowja has joined #openstack-mogan14:51
openstackgerritMerged openstack/mogan master: Fix uwsgi doc to avoid misunderstanding
*** harlowja has quit IRC15:51
*** openstackstatus has quit IRC16:41
*** openstack has joined #openstack-mogan16:45
*** harlowja has joined #openstack-mogan16:54
*** harlowja_ has joined #openstack-mogan16:55
*** harlowja has quit IRC16:59
-openstackstatus- NOTICE: The Gerrit service on is being taken offline for roughly 5 minutes to perform a database backup and reconfiguration17:13
*** openstackgerrit has quit IRC17:33
*** openstackstatus has quit IRC18:11
*** openstack has joined #openstack-mogan18:15
*** openstackstatus has quit IRC18:42
*** openstack has joined #openstack-mogan18:46
*** liusheng has quit IRC19:45
*** liusheng has joined #openstack-mogan19:46

Generated by 2.15.3 by Marius Gedminas - find it at!