*** MVenesio has joined #openstack-nova | 00:02 | |
vilobhmm1 | jaypipies : ping | 00:03 |
---|---|---|
vilobhmm1 | jaypipes : ^^ | 00:03 |
*** annegentle has quit IRC | 00:04 | |
*** Murali has quit IRC | 00:06 | |
vilobhmm1 | the whole reason for introducing the start() and leave() interface was that in the tooz driver that I have introduced I use start() for start/initialization and join() for joining the group for clear speration of functionality you can see here https://review.openstack.org/#/c/172502/4/nova/servicegroup/drivers/tz.py . In your feedback you mentioned that you dont want to introduce few more interfaces to nova/servicegroup/ | 00:06 |
vilobhmm1 | jaypipes : ^^ | 00:06 |
*** melwitt has joined #openstack-nova | 00:10 | |
*** otter768 has joined #openstack-nova | 00:11 | |
*** patrickeast has quit IRC | 00:14 | |
*** achanda has joined #openstack-nova | 00:16 | |
*** achanda_ has quit IRC | 00:19 | |
*** achanda has quit IRC | 00:21 | |
*** tsekiyam_ has joined #openstack-nova | 00:22 | |
*** openstackgerrit has quit IRC | 00:22 | |
*** openstackgerrit has joined #openstack-nova | 00:23 | |
*** tsekiya__ has joined #openstack-nova | 00:24 | |
*** tsekiyam_ has quit IRC | 00:24 | |
openstackgerrit | melanie witt proposed openstack/python-novaclient: Don't lookup service url when bypass_url is given https://review.openstack.org/175622 | 00:24 |
*** tsekiya__ has quit IRC | 00:24 | |
*** tsekiyama has quit IRC | 00:26 | |
*** sdake_ has joined #openstack-nova | 00:31 | |
*** claudiub has quit IRC | 00:33 | |
*** arif-ali has joined #openstack-nova | 00:33 | |
*** openstack has joined #openstack-nova | 00:36 | |
*** Administrator_ has joined #openstack-nova | 00:37 | |
*** Marga_ has quit IRC | 00:38 | |
*** eliqiao has joined #openstack-nova | 00:39 | |
*** shihanzhang has quit IRC | 00:44 | |
openstackgerrit | zhangtralon proposed openstack/nova: remove the CONF.allow_migrate_to_same_host https://review.openstack.org/118604 | 00:48 |
*** penick has joined #openstack-nova | 00:48 | |
*** shaohe_feng1 has joined #openstack-nova | 00:49 | |
*** promulo__ has joined #openstack-nova | 00:52 | |
*** emagana has quit IRC | 00:54 | |
*** promulo_ has quit IRC | 00:55 | |
alex_xu | cfriesen: sorry, just wake up, I saw the patch already merged, it's fine, let's improve it later :) | 00:59 |
*** browne has quit IRC | 00:59 | |
*** Marga_ has joined #openstack-nova | 01:00 | |
*** zzzeek has quit IRC | 01:01 | |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require from compute_node_get_all_by_* from db layer https://review.openstack.org/175334 | 01:02 |
*** shaohe_feng1 has quit IRC | 01:03 | |
*** dboik has joined #openstack-nova | 01:05 | |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require from compute_node_get_all_by_* from db layer https://review.openstack.org/175334 | 01:06 |
*** dboik_ has joined #openstack-nova | 01:06 | |
openstackgerrit | Eli Qiao proposed openstack/nova: Remove db layer hard-code permission checks for v2.1 cells https://review.openstack.org/150350 | 01:06 |
*** shaohe_robot has quit IRC | 01:07 | |
*** dsanders has quit IRC | 01:07 | |
*** igordcard has quit IRC | 01:08 | |
*** joefides has joined #openstack-nova | 01:09 | |
*** dboik has quit IRC | 01:10 | |
*** park_heijlong has joined #openstack-nova | 01:10 | |
*** joefides_ has joined #openstack-nova | 01:10 | |
*** Nic1 has quit IRC | 01:12 | |
*** joefides has quit IRC | 01:13 | |
*** MVenesio has quit IRC | 01:19 | |
*** zhenguo has joined #openstack-nova | 01:21 | |
*** fawadkhaliq has joined #openstack-nova | 01:25 | |
*** popw has joined #openstack-nova | 01:25 | |
*** fawadkhaliq has quit IRC | 01:29 | |
openstackgerrit | Eli Qiao proposed openstack/nova: Neutron: Check port binding status https://review.openstack.org/156112 | 01:30 |
*** mwagner_lap has joined #openstack-nova | 01:37 | |
*** erkules_ has joined #openstack-nova | 01:39 | |
*** erkules has quit IRC | 01:41 | |
*** browne has joined #openstack-nova | 01:42 | |
*** mmedvede has quit IRC | 01:44 | |
*** jecarey has joined #openstack-nova | 01:45 | |
*** vilobhmm1 has quit IRC | 01:45 | |
*** mmedvede has joined #openstack-nova | 01:48 | |
*** melwitt has quit IRC | 01:51 | |
*** baoli has quit IRC | 01:52 | |
*** baoli has joined #openstack-nova | 01:52 | |
*** thedodd has quit IRC | 01:53 | |
*** bkopilov has quit IRC | 01:53 | |
*** unicell has quit IRC | 01:59 | |
*** hightall has joined #openstack-nova | 02:02 | |
*** ijw has quit IRC | 02:02 | |
*** VW_ has joined #openstack-nova | 02:05 | |
eliqiao1 | alex_xu: hi there? | 02:10 |
alex_xu | eliqiao1: yea there | 02:10 |
eliqiao1 | alex_xu: a question about remove admin policy check from db layer. | 02:11 |
alex_xu | eliqiao1: yea, sure | 02:11 |
eliqiao1 | alex_xu: for those db api method, there is no any rest api calling, how to deal with them , directly remove them? | 02:12 |
eliqiao1 | alex_xu: they are all called by nova code | 02:12 |
eliqiao1 | alex_xu: I remember that we have seen those kinds of cases, can you recall them? | 02:12 |
alex_xu | eliqiao1: yea, we can remove them directly | 02:12 |
eliqiao1 | alex_xu: any test cases required? | 02:13 |
alex_xu | eliqiao1: this one https://review.openstack.org/#/c/150349/ | 02:13 |
alex_xu | eliqiao1: not test needn't | 02:13 |
eliqiao1 | alex_xu: cool, thanks | 02:13 |
alex_xu | eliqiao1: np, :) | 02:14 |
*** liusheng has quit IRC | 02:16 | |
*** rfolco has quit IRC | 02:31 | |
*** achanda has joined #openstack-nova | 02:35 | |
*** harlowja is now known as harlowja_away | 02:35 | |
*** VW_ has quit IRC | 02:36 | |
*** zhangtralon has joined #openstack-nova | 02:36 | |
*** harlowja_away is now known as harlowja | 02:37 | |
*** fawadkhaliq has joined #openstack-nova | 02:39 | |
*** VW_ has joined #openstack-nova | 02:39 | |
*** Marga_ has quit IRC | 02:40 | |
*** Marga_ has joined #openstack-nova | 02:40 | |
*** dsanders has joined #openstack-nova | 02:40 | |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require from compute_node_get_all_by_* from db layer https://review.openstack.org/175334 | 02:41 |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require for compute_node_create/update/delete from db layer https://review.openstack.org/175693 | 02:41 |
*** Marga_ has quit IRC | 02:42 | |
*** exploreshaifali has quit IRC | 02:46 | |
*** exploreshaifali has joined #openstack-nova | 02:47 | |
*** mmedvede has quit IRC | 02:56 | |
openstackgerrit | Li Yingjun proposed openstack/nova: Add missing delete policies in the sample file https://review.openstack.org/175251 | 02:57 |
openstackgerrit | shihanzhang proposed openstack/nova: Add config option to override url for versions https://review.openstack.org/160266 | 02:57 |
*** mmedvede has joined #openstack-nova | 02:59 | |
openstackgerrit | zhangtralon proposed openstack/nova: filter_class_names is never passed https://review.openstack.org/143285 | 02:59 |
*** xuhanp has joined #openstack-nova | 03:02 | |
*** xuhanp_ has joined #openstack-nova | 03:02 | |
*** tojuvone has joined #openstack-nova | 03:08 | |
*** eliqiao1 has quit IRC | 03:09 | |
*** park_heijlong has quit IRC | 03:14 | |
*** park_heijlong has joined #openstack-nova | 03:15 | |
*** artom has quit IRC | 03:19 | |
*** bkopilov has joined #openstack-nova | 03:19 | |
*** Marga_ has joined #openstack-nova | 03:22 | |
*** jogo has quit IRC | 03:22 | |
*** fawadkhaliq has quit IRC | 03:22 | |
*** jyuso1 has quit IRC | 03:23 | |
*** park_heijlong has quit IRC | 03:26 | |
*** exploreshaifali has quit IRC | 03:30 | |
openstackgerrit | Tomi Juvonen proposed openstack/nova-specs: New nova API call to mark nova-compute down https://review.openstack.org/169836 | 03:31 |
openstackgerrit | heijlong proposed openstack/nova: Merge create-backup functional tests of v2 and v2.1 https://review.openstack.org/174226 | 03:34 |
openstackgerrit | heijlong proposed openstack/nova: Merge extension-info functional tests of v2 and v2.1 https://review.openstack.org/175701 | 03:34 |
*** Marga_ has quit IRC | 03:35 | |
openstackgerrit | Vincent Hou proposed openstack/nova: Correct the order of the parameters to call swap_volume https://review.openstack.org/174307 | 03:36 |
*** Marga_ has joined #openstack-nova | 03:36 | |
*** zhangtralon has quit IRC | 03:38 | |
*** achanda has quit IRC | 03:41 | |
*** RuiChen has joined #openstack-nova | 03:41 | |
*** annashen has joined #openstack-nova | 03:42 | |
*** otter768 has quit IRC | 03:42 | |
*** pixelbeat has quit IRC | 03:44 | |
*** pixelbeat has joined #openstack-nova | 03:44 | |
*** jecarey has quit IRC | 03:45 | |
*** dsanders has quit IRC | 03:47 | |
*** lpetrut has joined #openstack-nova | 03:49 | |
*** deepthi has joined #openstack-nova | 03:50 | |
*** Zxen has quit IRC | 03:51 | |
*** mmedvede has quit IRC | 03:54 | |
*** ajayaa has joined #openstack-nova | 03:54 | |
*** harlowja is now known as harlowja_away | 03:56 | |
*** mmedvede has joined #openstack-nova | 04:02 | |
*** annashen has quit IRC | 04:05 | |
*** annashen has joined #openstack-nova | 04:06 | |
*** ajayaa has quit IRC | 04:08 | |
*** annashen has quit IRC | 04:11 | |
*** iamjarvo has joined #openstack-nova | 04:14 | |
*** jogo has joined #openstack-nova | 04:15 | |
*** jogo has quit IRC | 04:17 | |
*** achanda has joined #openstack-nova | 04:17 | |
*** baoli has quit IRC | 04:18 | |
*** Marga_ has quit IRC | 04:18 | |
*** Murali has joined #openstack-nova | 04:19 | |
*** jogo has joined #openstack-nova | 04:20 | |
*** pixelbeat has quit IRC | 04:21 | |
*** VW_ has quit IRC | 04:21 | |
openstackgerrit | Michael Still proposed openstack/nova-specs: console: convert consoles code to use objects framework https://review.openstack.org/165838 | 04:25 |
*** ishant has joined #openstack-nova | 04:26 | |
*** dsanders has joined #openstack-nova | 04:26 | |
*** otter768 has joined #openstack-nova | 04:28 | |
*** fawadkhaliq has joined #openstack-nova | 04:30 | |
*** dsanders has quit IRC | 04:30 | |
*** unicell has joined #openstack-nova | 04:32 | |
*** otter768 has quit IRC | 04:38 | |
*** Murali_ has joined #openstack-nova | 04:38 | |
*** Murali has quit IRC | 04:42 | |
*** Murali_ is now known as Murali | 04:42 | |
*** dave-mccowan has quit IRC | 04:44 | |
*** Sukhdev has joined #openstack-nova | 04:44 | |
*** annashen has joined #openstack-nova | 04:45 | |
*** Murali has quit IRC | 04:49 | |
*** vilobhmm1 has joined #openstack-nova | 04:50 | |
*** Longgeek has joined #openstack-nova | 04:52 | |
*** zhenguo has quit IRC | 04:57 | |
*** mmedvede has quit IRC | 04:59 | |
*** unicell has quit IRC | 05:01 | |
*** flaper87 has quit IRC | 05:01 | |
*** emagana has joined #openstack-nova | 05:02 | |
*** flaper87 has joined #openstack-nova | 05:04 | |
*** mmedvede has joined #openstack-nova | 05:13 | |
*** emagana has quit IRC | 05:16 | |
*** emagana has joined #openstack-nova | 05:17 | |
*** neelashah has quit IRC | 05:20 | |
*** emagana has quit IRC | 05:21 | |
*** nkrinner has joined #openstack-nova | 05:21 | |
*** dsanders has joined #openstack-nova | 05:22 | |
*** ajayaa has joined #openstack-nova | 05:25 | |
openstackgerrit | Alex Xu proposed openstack/nova-specs: Virt driver support custom cpu model based on flavor/image https://review.openstack.org/170397 | 05:28 |
openstackgerrit | Alex Xu proposed openstack/nova-specs: Virt driver to support cpu features https://review.openstack.org/170396 | 05:28 |
*** Administrator_ has quit IRC | 05:28 | |
*** jcoufal has joined #openstack-nova | 05:31 | |
*** iamjarvo has quit IRC | 05:31 | |
*** iamjarvo has joined #openstack-nova | 05:37 | |
*** arnaud___ has joined #openstack-nova | 05:50 | |
*** irenab has joined #openstack-nova | 05:50 | |
*** sudipto has joined #openstack-nova | 05:52 | |
*** achanda has quit IRC | 05:52 | |
*** iamjarvo has quit IRC | 05:54 | |
openstackgerrit | Alex Xu proposed openstack/nova: Add rpc client side version control for conductor task API https://review.openstack.org/174152 | 05:54 |
openstackgerrit | Alex Xu proposed openstack/nova: Update rpc version aliases for kilo https://review.openstack.org/174148 | 05:54 |
*** ildikov has quit IRC | 05:57 | |
*** achanda has joined #openstack-nova | 05:57 | |
*** garyk has joined #openstack-nova | 06:02 | |
*** achanda has quit IRC | 06:03 | |
*** dsanders has quit IRC | 06:04 | |
*** ishant has quit IRC | 06:05 | |
*** Murali has joined #openstack-nova | 06:06 | |
*** armax_ has quit IRC | 06:06 | |
*** armax has joined #openstack-nova | 06:07 | |
*** armax has quit IRC | 06:07 | |
*** armax has joined #openstack-nova | 06:08 | |
*** armax has left #openstack-nova | 06:08 | |
*** armax has joined #openstack-nova | 06:08 | |
*** Administrator_ has joined #openstack-nova | 06:10 | |
*** armax has quit IRC | 06:10 | |
*** zhangtralon has joined #openstack-nova | 06:10 | |
*** armax has joined #openstack-nova | 06:10 | |
*** armax has quit IRC | 06:10 | |
openstackgerrit | shihanzhang proposed openstack/nova: Add config option to override url for versions https://review.openstack.org/160266 | 06:14 |
*** lpetrut has quit IRC | 06:18 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Imported Translations from Transifex https://review.openstack.org/171943 | 06:18 |
openstackgerrit | Alex Xu proposed openstack/nova: Soft delete system_metadata when destroy instance https://review.openstack.org/110875 | 06:21 |
openstackgerrit | Alex Xu proposed openstack/nova: Let soft-deleted instance_system_metadata readable https://review.openstack.org/109201 | 06:21 |
*** afazekas_ has joined #openstack-nova | 06:21 | |
*** annashen has quit IRC | 06:23 | |
openstackgerrit | Divya K Konoor proposed openstack/nova: Control create/delete flavor api permissions using policy.json https://review.openstack.org/175739 | 06:28 |
*** salv-orlando has quit IRC | 06:29 | |
openstackgerrit | Timofey Durakov proposed openstack/nova: Removed 'PYTHONHASHSEED=0' from tox.ini https://review.openstack.org/163877 | 06:33 |
openstackgerrit | Timofey Durakov proposed openstack/nova: Fixed test in test_tracker to work with random PYTHONHASHSEED https://review.openstack.org/174356 | 06:33 |
*** jamielennox is now known as jamielennox|away | 06:34 | |
*** doude has joined #openstack-nova | 06:34 | |
*** doron_afk has joined #openstack-nova | 06:34 | |
*** otter768 has joined #openstack-nova | 06:39 | |
*** dsanders has joined #openstack-nova | 06:39 | |
*** doron_afk is now known as doron | 06:41 | |
*** joefides_ has quit IRC | 06:43 | |
*** Sukhdev has quit IRC | 06:44 | |
*** dikonoor has joined #openstack-nova | 06:44 | |
*** otter768 has quit IRC | 06:44 | |
*** sahid has joined #openstack-nova | 06:45 | |
*** vilobhmm11 has joined #openstack-nova | 06:50 | |
*** vilobhmm1 has quit IRC | 06:51 | |
*** joefides has joined #openstack-nova | 06:52 | |
*** doron is now known as doron_afk | 06:53 | |
*** kaisers has joined #openstack-nova | 06:54 | |
*** cfriesen has quit IRC | 06:54 | |
*** vilobhmm11 has quit IRC | 06:54 | |
dikonoor | alex_xu: On https://review.openstack.org/#/c/175739/ , the build has failed. I am a little new to OpenStack contribution. I don't see any logs under screen-logs under the pci-test failure | 06:54 |
alex_xu | dikonoor: let me check | 06:55 |
dikonoor | alex_xu: ok..Thanks | 06:55 |
alex_xu | dikonoor: it's fine just a third-party ci failed | 06:55 |
alex_xu | dikonoor: let's waiting for more time | 06:56 |
dikonoor | alex_xu: ok..sure..Thanks | 06:56 |
alex_xu | dikonoor: np, thanks for the patch | 06:56 |
dikonoor | alex_xu: I noticed similar problem with view host details. But I am not able to pinpoint the recent change that has caused it. | 06:58 |
dikonoor | 9b0d73e660af434481a0a9b6d6a3bab7 - - -] User does not have admin privileges | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors Traceback (most recent call last): | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors File "/usr/lib/python2.7/site-packages/nova/api/openstack/compute/contrib/hypervisors.py", line 147, in show | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors service = self.host_api.service_get_by_compute_host(context, hyp.host) | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors File "/usr/lib/python2.7/site-packages/nova/compute/api.py", line 3451, in service_get_by_compute_host | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors return objects.Service.get_by_compute_host(context, host_name) | 06:58 |
*** zhangtralon has quit IRC | 06:58 | |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors File "/usr/lib/python2.7/site-packages/nova/objects/base.py", line 163, in wrapper | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors result = fn(cls, context, *args, **kwargs) | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors File "/usr/lib/python2.7/site-packages/nova/objects/service.py", line 151, in get_by_compute_host | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors db_service = db.service_get_by_compute_host(context, host) | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors File "/usr/lib/python2.7/site-packages/nova/db/api.py", line 139, in service_get_by_compute_host | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors use_slave=use_slave) | 06:58 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hypervisors File "/usr/lib/python2.7/site-packages/nova/db/sqlalchemy/api.py", line 214, in wrapper | 06:59 |
dikonoor | 2015-04-19 21:34:22.194 23179 TRACE nova.api.openstack.compute.contrib.hyp | 06:59 |
alex_xu | dikonoor: that's raised from db layer, right? | 07:00 |
dikonoor | alex_xu : The exception comes from the require_admin_context decorator used in the nova db sqlalchemy layer but I think that decorator was always there. | 07:00 |
*** arnaud___ has quit IRC | 07:00 | |
*** arnaud___ has joined #openstack-nova | 07:01 | |
alex_xu | dikonoor: yea, we plan to clean up those decorator in the db layer. that's part of work in this bp https://blueprints.launchpad.net/nova/+spec/nova-api-policy-final-part | 07:01 |
*** arnaud___ has quit IRC | 07:01 | |
alex_xu | dikonoor: for the compute node related db call is work by eliqiao, there is one of patch https://review.openstack.org/175334 | 07:02 |
*** Nikolay_St has joined #openstack-nova | 07:02 | |
*** mestery has quit IRC | 07:03 | |
*** zhangtralon has joined #openstack-nova | 07:03 | |
dikonoor | alex_xu: ok..Let me check if any of the existing gerrit reviews cover this methos | 07:04 |
*** eliqiao has quit IRC | 07:04 | |
*** eliqiao has joined #openstack-nova | 07:04 | |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge server groups tests between v2 and v2.1 https://review.openstack.org/174181 | 07:04 |
*** romainh has joined #openstack-nova | 07:04 | |
StevenK | zigo: I figured out the assertRaisesRegex thing. You need to update python-testtools to the latest upstream. This may require updating unittest2, and packaging up traceback2 and linecache2. | 07:04 |
alex_xu | dikonoor: not yet as know, I guess eliqiao is working on it | 07:05 |
*** browne has quit IRC | 07:06 | |
*** dsanders has quit IRC | 07:09 | |
dikonoor | alex_xu: ok, the exception I pointed out comes from method service_get_by_compute_host and the patch in https://review.openstack.org/#/c/175334/ from eliqiao is not covering that. | 07:10 |
*** zhenguo has joined #openstack-nova | 07:10 | |
alex_xu | dikonoor: yea, I guess he will work out that on separated patch | 07:10 |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge availability_zone tests between v2 and v2.1 https://review.openstack.org/171017 | 07:11 |
*** kaisers has quit IRC | 07:11 | |
alex_xu | dikonoor: we don't want to cleanup all the db call in the single patch, that's hard for review | 07:11 |
*** eliqiao has quit IRC | 07:11 | |
*** pcaruana has quit IRC | 07:11 | |
*** eliqiao has joined #openstack-nova | 07:12 | |
gilliard | Hello, Nova! | 07:12 |
dikonoor | alex_xu , eliqiao : Is there a LP bug that should be opened and assigned to eligiao for this ? | 07:13 |
sahid | hello gilliard | 07:13 |
sudipto | bauzas, Hello ! | 07:14 |
alex_xu | dikonoor: There already have bp https://blueprints.launchpad.net/openstack/?searchtext=nova-api-policy-final-part for it, so needn't a LP bug | 07:14 |
openstackgerrit | shihanzhang proposed openstack/nova: Optimize validate_networks https://review.openstack.org/175742 | 07:14 |
dikonoor | alex_xu ,eliqiao : oh, right.you pointed me to that. would this patch make it for Kilo? | 07:14 |
dikonoor | alex_xu : I mean not the bp; specifically those that eliqiao is working on | 07:15 |
alex_xu | dikonoor: no, it's won't. | 07:15 |
dikonoor | alex_xu : ok | 07:16 |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require for compute_node(get_all/search_by_hyperviso) from db https://review.openstack.org/175744 | 07:16 |
openstackgerrit | Eli Qiao proposed openstack/nova: API: Add policy enforcement test cases for pci API https://review.openstack.org/175745 | 07:16 |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge preserve_ephemeral_rebuild tests between v2 & v2.1 https://review.openstack.org/173284 | 07:17 |
*** jyuso1 has joined #openstack-nova | 07:18 | |
*** jichen has joined #openstack-nova | 07:20 | |
*** arif-ali has quit IRC | 07:20 | |
*** arif-ali has joined #openstack-nova | 07:21 | |
*** kaisers has joined #openstack-nova | 07:21 | |
*** dsanders has joined #openstack-nova | 07:23 | |
zigo | StevenK: Thanks for letting me know! I'll see what I can do for that, though probably it's going to be nothing until Jessie is out (ie: in 4 days from now). | 07:23 |
*** flaper87 has quit IRC | 07:23 | |
*** flaper87 has joined #openstack-nova | 07:23 | |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge simple tenant usage tests between v2 and v2.1 https://review.openstack.org/175243 | 07:23 |
*** xuhanp_ has quit IRC | 07:28 | |
*** xuhanp has quit IRC | 07:28 | |
*** afazekas_ has quit IRC | 07:32 | |
alex_xu | dikonoor: just small pep8 problem need fix https://review.openstack.org/#/c/175739/ | 07:32 |
dikonoor | alex_xu: thanks for pointing that out :) ..I completely forgot to run pep8..my basd..Let me fix it right away | 07:32 |
alex_xu | dikonoor: np | 07:33 |
*** mpaolino has joined #openstack-nova | 07:33 | |
*** afazekas_ has joined #openstack-nova | 07:36 | |
alex_xu | dikonoor: how about we change https://github.com/openstack/nova/blob/master/nova/tests/unit/api/openstack/compute/contrib/test_flavor_manage.py#L445 | 07:37 |
alex_xu | and https://github.com/openstack/nova/blob/master/nova/tests/unit/api/openstack/compute/contrib/test_flavor_manage.py#L467 | 07:37 |
alex_xu | dikonoor: to return fakes.HTTPRequest.blank(url, use_admin_context=False) | 07:38 |
alex_xu | dikonoor: that can make sure there isn't any hard-code permission checks in the code | 07:38 |
alex_xu | dikonoor: It means use non-admin context in the unittest, if all the unittest passed, that means there isn't any hard-code permission checks | 07:39 |
dikonoor | alex_xu : Thanks for pointing those out..I will make those changes as part of the next patch | 07:39 |
alex_xu | dikonoor: thanks | 07:39 |
*** penick has quit IRC | 07:39 | |
*** tralon_ has joined #openstack-nova | 07:42 | |
*** jamielennox|away is now known as jamielennox | 07:43 | |
*** zhangtralon has quit IRC | 07:46 | |
*** mestery has joined #openstack-nova | 07:46 | |
*** jistr has joined #openstack-nova | 07:47 | |
openstackgerrit | Hans Lindgren proposed openstack/nova: Make ComputeNode.service_id nullable to match db schema https://review.openstack.org/169284 | 07:50 |
openstackgerrit | Hans Lindgren proposed openstack/nova: Don't query compute_node through service object in nova-manage https://review.openstack.org/168907 | 07:50 |
*** moshele has joined #openstack-nova | 07:51 | |
openstackgerrit | Hans Lindgren proposed openstack/nova: Avoid querying for Service in resource tracker https://review.openstack.org/169288 | 07:52 |
*** dsanders has quit IRC | 07:53 | |
*** erkules_ is now known as erkules | 07:53 | |
*** erkules has joined #openstack-nova | 07:53 | |
*** apuimedo_ has joined #openstack-nova | 07:54 | |
*** Marga_ has joined #openstack-nova | 07:56 | |
*** foexle has joined #openstack-nova | 07:58 | |
*** arnaud___ has joined #openstack-nova | 08:02 | |
*** doron_afk is now known as doron | 08:06 | |
*** arnaud___ has quit IRC | 08:06 | |
*** zz_johnthetubagu is now known as johnthetubaguy | 08:08 | |
*** Nikolay_St has quit IRC | 08:10 | |
*** ildikov has joined #openstack-nova | 08:11 | |
*** johnthetubaguy is now known as zz_johnthetubagu | 08:15 | |
*** Administrator_ has quit IRC | 08:16 | |
*** lennyb has joined #openstack-nova | 08:17 | |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: scheduler: weight nodes based on their cached images https://review.openstack.org/173304 | 08:17 |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: Add new image cache monitor in compute nodes https://review.openstack.org/173303 | 08:17 |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: compute monitors: set update delta from base class https://review.openstack.org/173987 | 08:17 |
*** derekh has joined #openstack-nova | 08:18 | |
*** eliqiao1 has joined #openstack-nova | 08:19 | |
*** eliqiao1 has quit IRC | 08:19 | |
*** eliqiao1 has joined #openstack-nova | 08:19 | |
*** eliqiao has quit IRC | 08:20 | |
*** zz_johnthetubagu is now known as johnthetubaguy | 08:20 | |
*** lucasagomes has joined #openstack-nova | 08:23 | |
*** mmedvede has quit IRC | 08:23 | |
*** eliqiao1 has quit IRC | 08:24 | |
openstackgerrit | Adelina Tuvenie proposed openstack/nova: Adds Hyper-V vNic hot-plug implementation https://review.openstack.org/138255 | 08:31 |
openstackgerrit | Bartosz Fic proposed openstack/nova: Live migration failure in API doesn't set vm_state to Error https://review.openstack.org/168916 | 08:33 |
*** matrohon has joined #openstack-nova | 08:35 | |
*** mmedvede has joined #openstack-nova | 08:35 | |
openstackgerrit | Bartosz Fic proposed openstack/nova: Live migration failure in API doesn't set vm_state to Error https://review.openstack.org/168916 | 08:36 |
sahid | ndipanov: any chance to get a review here https://review.openstack.org/#/c/172079/ ? | 08:36 |
*** eglynn has joined #openstack-nova | 08:37 | |
*** ankit_ag has joined #openstack-nova | 08:38 | |
*** eliqiao has joined #openstack-nova | 08:39 | |
*** otter768 has joined #openstack-nova | 08:40 | |
*** Nikolay_St has joined #openstack-nova | 08:41 | |
openstackgerrit | garyk proposed openstack/nova: VMware: enable configuring of console delay https://review.openstack.org/140669 | 08:41 |
*** Longgeek has quit IRC | 08:42 | |
*** pixelbeat has joined #openstack-nova | 08:42 | |
*** mmedvede has quit IRC | 08:42 | |
*** eliqiao has quit IRC | 08:44 | |
*** eliqiao1 has joined #openstack-nova | 08:44 | |
*** eliqiao1 has left #openstack-nova | 08:44 | |
*** eliqiao has joined #openstack-nova | 08:44 | |
*** otter768 has quit IRC | 08:44 | |
*** coolsvap|afk is now known as coolsvap | 08:45 | |
openstackgerrit | Divya K Konoor proposed openstack/nova: Control create/delete flavor api permissions using policy.json https://review.openstack.org/175739 | 08:46 |
ndipanov | johnthetubaguy, ah I see that ttx has started setting rc2 targets | 08:46 |
johnthetubaguy | ndipanov: we are doing that right now | 08:46 |
*** coolsvap is now known as coolsvap|afk | 08:46 | |
*** sdake_ has quit IRC | 08:47 | |
*** redbeard has joined #openstack-nova | 08:47 | |
ndipanov | johnthetubaguy, was there a meeting or some such about what gets a pass? | 08:47 |
johnthetubaguy | ndipanov: its just the weekly release sync up meeting | 08:49 |
ndipanov | johnthetubaguy, ok well - let me know if you need any input from me on the PCI bugs | 08:52 |
ndipanov | that I targeted | 08:52 |
*** f13o has joined #openstack-nova | 08:54 | |
*** tralon_ has quit IRC | 08:54 | |
redbeard | any hint to trigger the instance healing process manually? when I associate a floating IP in icehouse while the instance is in build or spawning state, the cache needs 10 to 20 minutes to "get" the floating IP information. the ip itself is associated + works in time | 08:55 |
*** rajesht has joined #openstack-nova | 08:58 | |
openstackgerrit | Ankit Agrawal proposed openstack/nova: Detach volume after deleting the instance https://review.openstack.org/145738 | 09:01 |
*** arnaud___ has joined #openstack-nova | 09:03 | |
johnthetubaguy | ndipanov: just looking at the merged ones, this is the current list for RC2 https://launchpad.net/nova/+milestone/kilo-rc2 which was basically everything | 09:03 |
johnthetubaguy | all but one deprecation commit that had a string in it | 09:03 |
*** mmedvede has joined #openstack-nova | 09:03 | |
*** doron is now known as doron_afk | 09:04 | |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: libvirt: make live_migration_uri flag dependant on virt_type https://review.openstack.org/175780 | 09:05 |
ndipanov | johnthetubaguy, the one we acked by garyk yesterday seems to be missing | 09:05 |
ndipanov | and it seemed important | 09:05 |
ndipanov | let me try to find that | 09:05 |
johnthetubaguy | ndipanov: hmm, maybe its not in master yet? which one is that? | 09:06 |
ndipanov | https://review.openstack.org/#/c/173794/ | 09:06 |
ndipanov | ah it is targetted | 09:06 |
ndipanov | ok | 09:06 |
johnthetubaguy | ndipanov: thats in the list I think | 09:06 |
ndipanov | sorry about that :( | 09:06 |
johnthetubaguy | np | 09:06 |
ndipanov | yeah looks like quite a list | 09:07 |
*** arnaud___ has quit IRC | 09:07 | |
garyk | johnthetubaguy: ndipanov: i backported that one to stable/kilo. just need some stables cores to ack it | 09:08 |
sahid | aloga: i have a question about https://review.openstack.org/#/c/173987/ | 09:09 |
ndipanov | tbh I am a bit confused about stable/kilo and proposed/kilo right now | 09:09 |
*** fawadkhaliq has quit IRC | 09:11 | |
openstackgerrit | garyk proposed openstack/nova: VMware: add in folder support on VC https://review.openstack.org/165060 | 09:11 |
*** ildikov has quit IRC | 09:11 | |
*** Longgeek has joined #openstack-nova | 09:14 | |
openstackgerrit | garyk proposed openstack/nova: VMware: improve instance names on VC https://review.openstack.org/166608 | 09:17 |
openstackgerrit | garyk proposed openstack/nova: VMware: remove unused method https://review.openstack.org/172896 | 09:21 |
aloga | sahid: hi | 09:24 |
*** ildikov has joined #openstack-nova | 09:24 | |
*** fawadkhaliq has joined #openstack-nova | 09:25 | |
aloga | sahid: if it's according your suggestion in https://review.openstack.org/#/c/173987/ | 09:26 |
aloga | I do agree | 09:27 |
*** fawadkhaliq has quit IRC | 09:27 | |
aloga | :) | 09:27 |
sahid | aloga: cool | 09:28 |
sahid | thanks :) | 09:29 |
johnthetubaguy | ndipanov: any chance you could do some numa rc2 back ports for me please? https://launchpad.net/nova/+milestone/kilo-rc2 the ones in the new state | 09:30 |
ndipanov | johnthetubaguy, yes | 09:31 |
ndipanov | need to step out for an hour or 2 - is it ok if it's PM europe time? | 09:32 |
aloga | sahid: you're welcome! | 09:32 |
johnthetubaguy | ndipanov: thats fine, it will see if there are some easier ones I can do, some of them seemed to conflict a bit, but I probably just got the order wrong | 09:32 |
ndipanov | johnthetubaguy, also - which branch should they be posted to? | 09:33 |
ndipanov | yeah order is important | 09:33 |
ndipanov | let me try | 09:33 |
johnthetubaguy | ndipanov: stable/kilo basically | 09:33 |
johnthetubaguy | ndipanov: as I understand it | 09:33 |
*** hightall has quit IRC | 09:33 | |
*** lpetrut has joined #openstack-nova | 09:36 | |
*** alexpilotti has quit IRC | 09:37 | |
*** fawadkhaliq has joined #openstack-nova | 09:38 | |
*** fawadkhaliq has quit IRC | 09:38 | |
*** fawadkhaliq has joined #openstack-nova | 09:38 | |
*** salv-orlando has joined #openstack-nova | 09:42 | |
*** Marga_ has quit IRC | 09:42 | |
openstackgerrit | Divya K Konoor proposed openstack/nova: Control create/delete flavor api permissions using policy.json https://review.openstack.org/175739 | 09:43 |
*** apuimedo_ has quit IRC | 09:45 | |
ndipanov | johnthetubaguy, done but gerrit hates me for not putting change ids in the footer | 09:49 |
ndipanov | let me try again | 09:49 |
ndipanov | ok done - it starts here https://review.openstack.org/#/c/175787/2 | 09:52 |
ndipanov | johnthetubaguy, ^^ | 09:52 |
ndipanov | ah I see you posted some already... | 09:52 |
ndipanov | ah well nvm | 09:52 |
*** lpetrut has quit IRC | 09:54 | |
openstackgerrit | dave-mcnally proposed openstack/nova: rebuild_instance doesn't detach cinder volumes correctly https://review.openstack.org/172951 | 09:56 |
johnthetubaguy | ndipanov: no worries, I can drop mine if you got the full chain | 09:58 |
*** park_heijlong has joined #openstack-nova | 10:00 | |
*** tojuvone has quit IRC | 10:00 | |
ndipanov | johnthetubaguy, no need | 10:00 |
ndipanov | the change ids are the same | 10:01 |
*** apuimedo_ has joined #openstack-nova | 10:02 | |
johnthetubaguy | ndipanov: ah, OK | 10:03 |
johnthetubaguy | cools | 10:03 |
openstackgerrit | Radoslav Gerganov proposed openstack/nova: Consolidate the APIs for getting consoles https://review.openstack.org/148509 | 10:03 |
johnthetubaguy | ndipanov: thanks for sorting those, big help | 10:03 |
*** boris-42 has quit IRC | 10:05 | |
*** kaisers has quit IRC | 10:06 | |
ndipanov | johnthetubaguy, np - thanks for considering them for the release | 10:06 |
*** tojuvone has joined #openstack-nova | 10:07 | |
*** boris-42 has joined #openstack-nova | 10:08 | |
*** oro has joined #openstack-nova | 10:10 | |
*** BobBall_AWOL is now known as BobBall | 10:10 | |
*** yamahata has quit IRC | 10:12 | |
*** tojuvone has quit IRC | 10:17 | |
*** tojuvone has joined #openstack-nova | 10:20 | |
*** IanGovett has joined #openstack-nova | 10:22 | |
*** mpaolino has quit IRC | 10:29 | |
sudipto | PaulMurray, Hi, can you point me to any example of a RT plugin in the code? | 10:31 |
*** fawadkhaliq has quit IRC | 10:32 | |
*** fawadkhaliq has joined #openstack-nova | 10:32 | |
sudipto | PaulMurray, nova.compute.resources - i guess that has 2 | 10:34 |
sudipto | rather 1 | 10:34 |
*** xyang1 has quit IRC | 10:35 | |
PaulMurray | sudipto, hi - yes, only the vcpu | 10:35 |
*** lpetrut has joined #openstack-nova | 10:36 | |
PaulMurray | sudipto, the extensible RT plugins are only implemented at the RT - so the values do arrive at the scheduler, but the consumer plugins are not there | 10:36 |
*** fawadk has joined #openstack-nova | 10:36 | |
PaulMurray | sudipto, the work was stalled because the whole area is likely to be upturned soon with resource objects | 10:37 |
PaulMurray | sudipto, what do you want to use them for? | 10:37 |
*** fawadkhaliq has quit IRC | 10:38 | |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: scheduler: weight nodes based on their cached images https://review.openstack.org/173304 | 10:39 |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: Add new image cache monitor in compute nodes https://review.openstack.org/173303 | 10:39 |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: compute monitors: set update delta from base class https://review.openstack.org/173987 | 10:39 |
*** otter768 has joined #openstack-nova | 10:41 | |
*** fawadkhaliq has joined #openstack-nova | 10:41 | |
*** fawadk has quit IRC | 10:43 | |
*** fawadk has joined #openstack-nova | 10:44 | |
*** otter768 has quit IRC | 10:45 | |
*** sahid has quit IRC | 10:46 | |
*** fawadkhaliq has quit IRC | 10:48 | |
*** fawadkhaliq has joined #openstack-nova | 10:49 | |
*** fawadkhaliq has quit IRC | 10:50 | |
*** fawadkhaliq has joined #openstack-nova | 10:50 | |
*** fawadk has quit IRC | 10:51 | |
*** arnaud___ has joined #openstack-nova | 10:52 | |
*** Hosam has joined #openstack-nova | 10:52 | |
*** fawadk has joined #openstack-nova | 10:53 | |
*** sahid has joined #openstack-nova | 10:54 | |
*** sahid has quit IRC | 10:54 | |
*** fawadkhaliq has quit IRC | 10:55 | |
*** fawadkhaliq has joined #openstack-nova | 10:55 | |
*** sahid has joined #openstack-nova | 10:55 | |
*** arnaud___ has quit IRC | 10:56 | |
sudipto | PaulMurray, sorry had stepped out for a bit. I wanted to implement something like a memory b/w utilization as a RT plugin. | 10:56 |
sudipto | PaulMurray, I have implemented it using the monitors currently - but that would mean that the data is populated in the metrics column and is available to the host_states on the controller. | 10:56 |
sudipto | PaulMurray, hence essentially it doesn't use the RT. | 10:57 |
sudipto | PaulMurray, would you have a suggestion on which framework fits better for this - the monitors or the RT? | 10:58 |
*** fawadk has quit IRC | 10:58 | |
*** deepthi has quit IRC | 10:59 | |
PaulMurray | sudipto, RT deals with values that are accounted - generally they are artificial values, like disk space or memory that are added up form the number of instances and their nominal usage. These values may also translate into limits on actual usage in the hypervisor but not necessarily. | 11:01 |
sudipto | PaulMurray, I wouldn't want to tie this with the extra specs of instances. | 11:01 |
PaulMurray | sudipto, metrics are used to monitor actual use of resources | 11:01 |
sudipto | PaulMurray, So metrics sounds like the right place for this.... | 11:01 |
*** jistr has quit IRC | 11:02 | |
*** nihilifer has quit IRC | 11:03 | |
PaulMurray | sudipto, so if you want to monitor the current use of memory b/w on the host, as opposed to saying "count 5 for this flavor" then metrics sounds like it | 11:03 |
*** nihilifer has joined #openstack-nova | 11:03 | |
sudipto | PaulMurray, hmm yeah - and then nothing should theoretically stop us from writing a filter that would just use this right? Something on the lines of this spec: https://review.openstack.org/#/c/154423/5/specs/liberty/approved/cache-aware-weigher.rst | 11:04 |
sudipto | PaulMurray, would this just require an implementation then or a spec per your advice? | 11:04 |
*** jistr has joined #openstack-nova | 11:04 | |
*** tojuvone has quit IRC | 11:07 | |
PaulMurray | sudipto, I think you can probably just do it. These are simple plugins, so specs seems heavy weight - the filter would be quicker and easier to write than the spec. | 11:08 |
PaulMurray | sudipto, Do it and see if anyone complains | 11:08 |
sudipto | PaulMurray, Ok sounds good. Thanks Paul. | 11:09 |
*** fawadkhaliq has quit IRC | 11:09 | |
sudipto | PaulMurray, Just one final thing, does it then go as a bug or still a BP? (sorry, i am new to openstack) | 11:09 |
*** f13o has quit IRC | 11:10 | |
*** claudiub has joined #openstack-nova | 11:12 | |
*** fawadkhaliq has joined #openstack-nova | 11:13 | |
*** Hosam has quit IRC | 11:13 | |
dikonoor | alex_xu:On the recent comments you gave for https://review.openstack.org/#/c/175739/ , I guess you are referring to the spaces there..? | 11:13 |
alex_xu | dikonoor: yes, those are unrelated change, we shouldn't make | 11:14 |
*** fawadkhaliq has quit IRC | 11:15 | |
dikonoor | alex_xu: I don't see those spaces when I open in a notepad or gedit.Let me try | 11:15 |
*** fawadkhaliq has joined #openstack-nova | 11:16 | |
alex_xu | dikonoor: thanks | 11:16 |
PaulMurray | sudipto, do a BP - it can be approved without a spec if its trivial. You need to set the target and then you can put your patches up referring to that bp | 11:16 |
sudipto | PaulMurray, Ok. Thanks! | 11:16 |
PaulMurray | Some of this is described here: http://docs.openstack.org/infra/manual/developers.html#working-on-specifications-and-blueprints | 11:18 |
PaulMurray | but not sure where nova specific bits are | 11:18 |
*** mleroy has quit IRC | 11:19 | |
*** Hosam has joined #openstack-nova | 11:20 | |
*** Hosam has quit IRC | 11:23 | |
*** mleroy has joined #openstack-nova | 11:24 | |
sudipto | PaulMurray, Thanks. Sorry i should have been better at reading those. | 11:27 |
*** otter768 has joined #openstack-nova | 11:27 | |
openstackgerrit | Radoslav Gerganov proposed openstack/nova: Consolidate the APIs for getting consoles https://review.openstack.org/148509 | 11:27 |
*** lucasagomes is now known as lucas-hungry | 11:29 | |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require from compute_node_get_all_by_* from db layer https://review.openstack.org/175334 | 11:30 |
openstackgerrit | Eli Qiao proposed openstack/nova: API: Add policy enforcement test cases for pci API https://review.openstack.org/175745 | 11:30 |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require for compute_node(get_all/search_by_hyperviso) from db https://review.openstack.org/175744 | 11:30 |
openstackgerrit | Eli Qiao proposed openstack/nova: API: remove admin require for compute_node_create/update/delete from db layer https://review.openstack.org/175693 | 11:30 |
*** fawadkhaliq has quit IRC | 11:30 | |
*** fawadkhaliq has joined #openstack-nova | 11:32 | |
*** Kevin_Zheng has joined #openstack-nova | 11:33 | |
*** MVenesio has joined #openstack-nova | 11:34 | |
*** fawadkhaliq has quit IRC | 11:35 | |
openstackgerrit | Divya K Konoor proposed openstack/nova: Control create/delete flavor api permissions using policy.json https://review.openstack.org/175739 | 11:35 |
*** mwagner_lap has quit IRC | 11:38 | |
*** alexpilotti has joined #openstack-nova | 11:40 | |
openstackgerrit | Eli Qiao proposed openstack/nova: Object: Improve Instance _save_flavor https://review.openstack.org/160075 | 11:43 |
openstackgerrit | Eli Qiao proposed openstack/nova: Object: Merge db accesses for extra field of instance https://review.openstack.org/160076 | 11:43 |
jaypipes | morning all. | 11:44 |
claudiub | mornooning. :D | 11:45 |
*** dave-mccowan has joined #openstack-nova | 11:46 | |
*** jcoufal_ has joined #openstack-nova | 11:47 | |
*** otter768 has quit IRC | 11:50 | |
*** jcoufal has quit IRC | 11:50 | |
sahid | claudiub: can i have you review here please https://review.openstack.org/#/c/155730/ | 11:53 |
*** yingjun has joined #openstack-nova | 11:54 | |
claudiub | sahid: sure. :) | 11:54 |
*** MVenesio has quit IRC | 11:56 | |
openstackgerrit | Jay Pipes proposed openstack/nova: Fix race in server group policy validation https://review.openstack.org/164762 | 11:57 |
openstackgerrit | Jay Pipes proposed openstack/nova: Fix race in affinity server group scheduling https://review.openstack.org/169489 | 11:57 |
*** atuvenie_ has joined #openstack-nova | 12:01 | |
*** atuvenie has joined #openstack-nova | 12:01 | |
*** markvoelker has joined #openstack-nova | 12:02 | |
*** markvoelker_ has joined #openstack-nova | 12:03 | |
*** markvoelker_ has quit IRC | 12:04 | |
*** markvoelker_ has joined #openstack-nova | 12:05 | |
*** raildo has joined #openstack-nova | 12:06 | |
*** kaisers has joined #openstack-nova | 12:06 | |
*** markvoelker has quit IRC | 12:07 | |
*** thangp has joined #openstack-nova | 12:09 | |
*** tojuvone has joined #openstack-nova | 12:11 | |
*** bkopilov has quit IRC | 12:12 | |
*** atuvenie_ has quit IRC | 12:12 | |
*** jyuso1 has quit IRC | 12:16 | |
*** sahid has quit IRC | 12:16 | |
*** dprince has joined #openstack-nova | 12:17 | |
openstackgerrit | Brent Eagles proposed openstack/nova: WIP: Refactor Nova VIF model https://review.openstack.org/167287 | 12:18 |
*** salv-orlando has quit IRC | 12:18 | |
openstackgerrit | Eli Qiao proposed openstack/nova: Add policy check for extension_info https://review.openstack.org/173210 | 12:18 |
*** lpetrut has quit IRC | 12:19 | |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: scheduler: weight nodes based on their cached images https://review.openstack.org/173304 | 12:19 |
openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: Add new image cache monitor in compute nodes https://review.openstack.org/173303 | 12:19 |
*** mwagner_lap has joined #openstack-nova | 12:19 | |
*** salv-orlando has joined #openstack-nova | 12:21 | |
*** aix has joined #openstack-nova | 12:21 | |
*** sudipto has quit IRC | 12:22 | |
*** yingjun has quit IRC | 12:22 | |
*** baoli has joined #openstack-nova | 12:23 | |
openstackgerrit | Sergey Nikitin proposed openstack/nova: Changed backporting of embedded Network object when backporting a FixedIP https://review.openstack.org/173723 | 12:25 |
*** yingjun has joined #openstack-nova | 12:27 | |
*** yingjun has quit IRC | 12:28 | |
*** joefides has quit IRC | 12:30 | |
openstackgerrit | John Garbutt proposed openstack/nova: Fix inefficient transaction usage in floating_ip_bulk_destroy https://review.openstack.org/140769 | 12:32 |
*** MVenesio has joined #openstack-nova | 12:35 | |
*** lucas-hungry is now known as lucasagomes | 12:35 | |
*** neelashah has joined #openstack-nova | 12:37 | |
openstackgerrit | garyk proposed openstack/nova: VMware: enable a resize of instance with no root disk https://review.openstack.org/173802 | 12:37 |
*** thangp has quit IRC | 12:38 | |
openstackgerrit | John Garbutt proposed openstack/nova: Cleanup unnecessary session creation in floating_ip_deallocate https://review.openstack.org/140770 | 12:38 |
*** arnaud___ has joined #openstack-nova | 12:40 | |
*** zhangtralon has joined #openstack-nova | 12:41 | |
*** lpetrut has joined #openstack-nova | 12:42 | |
*** mpaolino has joined #openstack-nova | 12:42 | |
*** jistr has quit IRC | 12:42 | |
*** jistr has joined #openstack-nova | 12:43 | |
*** arnaud___ has quit IRC | 12:44 | |
*** lpetrut has quit IRC | 12:46 | |
*** VW_ has joined #openstack-nova | 12:50 | |
*** ajayaa has quit IRC | 12:52 | |
openstackgerrit | John Garbutt proposed openstack/nova: Sorting and pagination params used as filters https://review.openstack.org/147298 | 12:52 |
openstackgerrit | shihanzhang proposed openstack/nova: Optimize validate_networks https://review.openstack.org/175742 | 12:55 |
*** irenab has quit IRC | 12:58 | |
*** tdurakov has quit IRC | 12:58 | |
*** baoli has quit IRC | 13:00 | |
*** Marga_ has joined #openstack-nova | 13:01 | |
*** jistr is now known as jistr|mtg | 13:01 | |
*** rajesht has quit IRC | 13:01 | |
*** baoli has joined #openstack-nova | 13:02 | |
*** xyang1 has joined #openstack-nova | 13:04 | |
openstackgerrit | Radoslav Gerganov proposed openstack/nova: Consolidate the APIs for getting consoles https://review.openstack.org/148509 | 13:05 |
*** Longgeek has quit IRC | 13:05 | |
*** lpetrut has joined #openstack-nova | 13:06 | |
*** salv-orlando has quit IRC | 13:06 | |
*** sandywalsh has quit IRC | 13:11 | |
*** kiran-r has joined #openstack-nova | 13:11 | |
*** VW_ has quit IRC | 13:11 | |
johnthetubaguy | sdague: I was looking at the DB downgrade change, https://review.openstack.org/#/c/175010/3 do we need to follow up an drop the shadow tables at some point? | 13:11 |
*** joefides has joined #openstack-nova | 13:12 | |
sdague | well, dropping the shadow tables is a whole other can of worms, right? | 13:12 |
jaypipes | johnthetubaguy, sdague, other cores: https://review.openstack.org/#/c/173913/ could use an extra review pls. thx in advance! | 13:12 |
jaypipes | johnthetubaguy: yeah, dropping the shadow stuff is a big thing... one sec, grfabbing a link to relevant spec.. | 13:13 |
johnthetubaguy | sdague: yeah, its more data lossey, but yeah there is more to it | 13:13 |
jaypipes | johnthetubaguy: https://review.openstack.org/#/c/137669/ | 13:13 |
johnthetubaguy | jaypipes: yeah, but I got the impression the shadow tables were only really used for downgrades, but I probably got the wrong end of the stick, I guess folks could use them for archive | 13:14 |
sdague | johnthetubaguy: yeh, no, the shadow tables are about archive | 13:14 |
*** changbl has quit IRC | 13:14 | |
johnthetubaguy | sdague: my bad | 13:14 |
sdague | they don't have any impact on downgrade as far as I know | 13:14 |
jaypipes | no worries, johnthetubaguy :) | 13:15 |
jaypipes | johnthetubaguy: it's a mess.. | 13:15 |
johnthetubaguy | jaypipes: agreed, our internal tools that rely on them, more so | 13:15 |
sdague | jaypipes: agreed... grab your shovel | 13:15 |
*** sandywalsh has joined #openstack-nova | 13:15 | |
johnthetubaguy | jaypipes: I mean the soft_deleted stuff, rather than shadow | 13:15 |
jaypipes | sdague: Bring a Broom! | 13:15 |
johnthetubaguy | sdague: yeah, I thought we did something like leave old stuff in shadow tables, so you can restore from shadow instead of downgrade, but I don't see how that could ever had worked though… oops | 13:16 |
johnthetubaguy | anyways, ignore my rambling | 13:17 |
*** zhenguo has quit IRC | 13:17 | |
*** sdake has joined #openstack-nova | 13:19 | |
*** atuvenie has quit IRC | 13:21 | |
*** sdake_ has joined #openstack-nova | 13:22 | |
openstackgerrit | Sean Dague proposed openstack/python-novaclient: wip: add ips to novaclient server manager https://review.openstack.org/175603 | 13:23 |
openstackgerrit | Sean Dague proposed openstack/python-novaclient: refactor functional test base class to no inherit from tempest_lib https://review.openstack.org/175854 | 13:23 |
*** busterswt has joined #openstack-nova | 13:23 | |
sdague | jaypipes: you are going to need way more than a broom :) | 13:23 |
jaypipes | sdague: :) | 13:23 |
alex_xu | johnthetubaguy: This is ready https://review.openstack.org/#/c/175739/ | 13:24 |
*** Longgeek has joined #openstack-nova | 13:24 | |
*** sdake has quit IRC | 13:24 | |
*** Longgeek has quit IRC | 13:24 | |
*** vladikr has joined #openstack-nova | 13:24 | |
*** Longgeek has joined #openstack-nova | 13:25 | |
*** sdake has joined #openstack-nova | 13:25 | |
*** atuvenie has joined #openstack-nova | 13:25 | |
johnthetubaguy | alex_xu: hmm, but now its not backwards compatible with the old check right? | 13:26 |
*** irenab has joined #openstack-nova | 13:26 | |
*** sdake_ has quit IRC | 13:28 | |
alex_xu | johnthetubaguy: yes, the old code always pass the admin context down to the code, so that means it always can pass the db layer admin checks. | 13:29 |
*** zhangtralon has quit IRC | 13:30 | |
alex_xu | johnthetubaguy: so flavor-manage's API can be configured by policy.json before. https://review.openstack.org/#/c/175739/ remove the wrong admin checks. | 13:30 |
johnthetubaguy | alex_xu: but it didn't seem to read that way in the commit that removed it, do you remember where you saw that? | 13:31 |
alex_xu | johnthetubaguy: REST API always call this https://github.com/openstack/nova/blob/master/nova/compute/flavors.py#L175 , it always pass the admin context | 13:32 |
alex_xu | johnthetubaguy: and the REST API never pass the user's context down to the code https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/contrib/flavormanage.py#L93 | 13:33 |
*** mriedem_away is now known as mriedem | 13:36 | |
*** kiran-r has quit IRC | 13:37 | |
*** doron_afk has quit IRC | 13:38 | |
*** sudipto has joined #openstack-nova | 13:38 | |
sudipto | dansmith, Hi Dan, I have got the +1s now from sahid and bauzas https://review.openstack.org/#/c/173187/ could you please take a look when you can...thanks!! | 13:39 |
*** zzzeek has joined #openstack-nova | 13:39 | |
dansmith | sudipto: no | 13:40 |
dansmith | sudipto: I have a struct "no +2s without coffee" policy | 13:40 |
dansmith | er, strict :) | 13:40 |
mriedem | johnthetubaguy: what in god's green earth is a doodle pool? | 13:40 |
alex_xu | looks like /me's English doesn't works again | 13:40 |
dansmith | damn, messed that all up | 13:40 |
*** Nikolay_St has quit IRC | 13:40 | |
sudipto | dansmith, :D ok, when you can that's why :-) | 13:40 |
*** EmilienM has quit IRC | 13:40 | |
*** EmilienM has joined #openstack-nova | 13:41 | |
johnthetubaguy | alex_xu: ah, I see, I missed that bit | 13:41 |
johnthetubaguy | alex_xu: thanks | 13:41 |
*** doron has joined #openstack-nova | 13:41 | |
alex_xu | johnthetubaguy: np :) cool my English works | 13:41 |
johnthetubaguy | mriedem: http://doodle.com/ | 13:41 |
mriedem | oic | 13:42 |
*** dboik_ has quit IRC | 13:43 | |
*** baoli has quit IRC | 13:44 | |
*** fawadkhaliq has joined #openstack-nova | 13:44 | |
*** ankit_ag has quit IRC | 13:44 | |
*** Murali has quit IRC | 13:44 | |
*** baoli_ has joined #openstack-nova | 13:45 | |
*** cbits has joined #openstack-nova | 13:45 | |
vladikr | johnthetubaguy, Hi, could you please remove your -2 from the libosinfo patches? https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/libvirt-hardware-policy-from-libosinfo,n,z | 13:45 |
*** burt has joined #openstack-nova | 13:45 | |
sudipto | bauzas, PaulMurray : Just out of curiosity, can the resource monitors be made optional? | 13:46 |
*** emagana has joined #openstack-nova | 13:48 | |
*** zz_jgrimm is now known as jgrimm | 13:49 | |
*** otter768 has joined #openstack-nova | 13:50 | |
johnthetubaguy | vladikr: the spec needs to be resubmitted for liberty, using the previously-approved: kilo git commit tag should make that quicker | 13:51 |
openstackgerrit | Alex Xu proposed openstack/nova: Add policy check for extension_info https://review.openstack.org/173210 | 13:51 |
*** armax_ has joined #openstack-nova | 13:51 | |
johnthetubaguy | vladikr: ping me once the spec is up for review, and I can speed that all up for you | 13:51 |
*** emagana_ has joined #openstack-nova | 13:51 | |
*** yamahata has joined #openstack-nova | 13:52 | |
vladikr | johnthetubaguy, got it , thanks | 13:52 |
*** salv-orlando has joined #openstack-nova | 13:52 | |
*** emagana has quit IRC | 13:52 | |
*** armax_ has quit IRC | 13:52 | |
*** armax_ has joined #openstack-nova | 13:52 | |
johnthetubaguy | vladikr: no problem | 13:53 |
*** amotoki has joined #openstack-nova | 13:54 | |
*** ildikov has quit IRC | 13:54 | |
*** salv-orlando has quit IRC | 13:54 | |
*** VW_ has joined #openstack-nova | 13:54 | |
*** otter768 has quit IRC | 13:55 | |
tojuvone | Jay, John,... changed the scope for this BP to have API to mark nova-compute down as wanted in review: https://review.openstack.org/#/c/169836/ | 13:56 |
tojuvone | Should I in meantime also start to write BP for CLI and perhaps for notification... | 13:56 |
*** Sukhdev has joined #openstack-nova | 13:57 | |
tojuvone | and perhaps single API to mark all VMs on failing host to "error" state.. and BP for CLI to that and PB perhaps for notification | 13:57 |
*** cfriesen has joined #openstack-nova | 13:58 | |
jaypipes | tojuvone: sure, that would be great, thank you! | 13:59 |
*** salv-orlando has joined #openstack-nova | 14:00 | |
*** tonytan4ever has joined #openstack-nova | 14:00 | |
*** jecarey has joined #openstack-nova | 14:00 | |
*** dboik has joined #openstack-nova | 14:01 | |
*** tojuvone has quit IRC | 14:02 | |
*** ijw has joined #openstack-nova | 14:02 | |
*** doron is now known as doron_afk | 14:02 | |
*** dikonoor has quit IRC | 14:02 | |
*** ijw has quit IRC | 14:02 | |
*** ijw has joined #openstack-nova | 14:03 | |
*** tojuvone has joined #openstack-nova | 14:03 | |
*** sudipto has quit IRC | 14:05 | |
*** sudipto has joined #openstack-nova | 14:05 | |
*** eharney has joined #openstack-nova | 14:06 | |
*** mtanino has joined #openstack-nova | 14:06 | |
*** apmelton_away is now known as apmelton | 14:07 | |
tojuvone | lost connection for a while.. but at got the one comment from jaypipes before that. Great, so can start to work for total 6 BPs then if seen worth while to have them over the topic. | 14:07 |
*** emagana_ has quit IRC | 14:08 | |
*** afazekas_ has quit IRC | 14:08 | |
*** emagana has joined #openstack-nova | 14:08 | |
*** jcoufal_ has quit IRC | 14:08 | |
*** armax_ has quit IRC | 14:08 | |
*** ildikov has joined #openstack-nova | 14:09 | |
*** armax has joined #openstack-nova | 14:09 | |
openstackgerrit | Dan Smith proposed openstack/nova-specs: Add robustify-evacuate spec https://review.openstack.org/161444 | 14:10 |
tojuvone | 3 BPs: API to mark nova-compute down, notification to that and CLI. 3 BPs: mark VMs related to failed nova-compute (host) down with single API call, notification to that and CLI. | 14:10 |
*** annegentle has joined #openstack-nova | 14:10 | |
*** cbits has left #openstack-nova | 14:12 | |
*** bkopilov has joined #openstack-nova | 14:12 | |
*** park_heijlong has quit IRC | 14:12 | |
*** lpetrut has quit IRC | 14:13 | |
*** popw has quit IRC | 14:13 | |
tojuvone | So jsut stated here if wanted some discussion. Otherwise surely discussed in those BPs :) | 14:14 |
*** atuvenie has quit IRC | 14:15 | |
johnthetubaguy | tojuvone: it would be great if the wider picture was described in an overarching dev ref document, I was talking to VW_ about something very related to this | 14:15 |
*** garyk has quit IRC | 14:17 | |
VW_ | ha - indeed - we were just discussing. Let me take a look at the bluepring | 14:17 |
VW_ | bluprint even | 14:17 |
openstackgerrit | Joshua Hesketh proposed openstack/nova: Don't require max_number for migrate_flavor_data https://review.openstack.org/175890 | 14:17 |
*** iamjarvo has joined #openstack-nova | 14:18 | |
*** iamjarvo has quit IRC | 14:18 | |
*** iamjarvo has joined #openstack-nova | 14:19 | |
*** iamjarvo has quit IRC | 14:19 | |
*** iamjarvo has joined #openstack-nova | 14:20 | |
tojuvone | VW_ so last version of BP is to mark service (nova-compute) down. The idea of whole is that external system can detect host failed fast and report to Nova to change all related states fast. | 14:21 |
dansmith | alex_xu: are you still around? | 14:21 |
tojuvone | Also fencinf would be guaranteed. No scheduling for that host and evacuate coudl happen immediately | 14:22 |
VW_ | ah, yes tojuvone - I remember this one now. I'm fine if something changes the compute state, but I'd agree with johnthetubaguy that doesn't always mean the instances are down as well | 14:23 |
VW_ | so I'd be leery to automatically change all states | 14:23 |
tojuvone | As single API for host couldn't be solution to mark nova-compute and VMs to down/error, there is currently only this one BP for nova-compute | 14:23 |
tojuvone | now it would be to make more BPs. | 14:23 |
tojuvone | Yes, that is why everybody interested of HA would even currently have own monitoring... | 14:24 |
tojuvone | now it would be to have API if the monitoring can know or if it can fence (power down) | 14:25 |
VW_ | well, it's a little more that that. I think nova computes should be completely expendable. We actually run them as a VM on the host - which means the compute can die and the host and instances are just fine. | 14:25 |
*** jistr|mtg is now known as jistr | 14:26 | |
tojuvone | So not solution for all cases | 14:26 |
VW_ | but anyway, I like the idea of the system quickly knowing a compute (or underlying host) is down and disabling the compute | 14:26 |
*** atuvenie has joined #openstack-nova | 14:27 | |
tojuvone | Yes and like I am now working also with this opensource project Doctor in OPNFV, that should have Telco (NFV) capability to detect host faults and fence... | 14:27 |
tojuvone | So when that shoudl be ready there should be infra tool for everybody to use | 14:28 |
*** annegentle has quit IRC | 14:28 | |
*** annegentle has joined #openstack-nova | 14:28 | |
*** arnaud___ has joined #openstack-nova | 14:29 | |
*** annashen has joined #openstack-nova | 14:29 | |
openstackgerrit | Sean Dague proposed openstack/nova: remove downgrade support from our database migrations https://review.openstack.org/175010 | 14:29 |
openstackgerrit | Vladik Romanovsky proposed openstack/nova-specs: Libvirt hardware policy from libosinfo https://review.openstack.org/175895 | 14:29 |
tojuvone | Great. How about VMs. Would the collective API to mark host related VMs to erro be good (or what would be the state)? | 14:30 |
tojuvone | anyhow that has API to go VM by VM | 14:30 |
*** browne has joined #openstack-nova | 14:30 | |
*** dboik has quit IRC | 14:30 | |
*** lpetrut has joined #openstack-nova | 14:30 | |
*** dboik has joined #openstack-nova | 14:31 | |
*** rushil has joined #openstack-nova | 14:31 | |
vladikr | johnthetubaguy, I have resubmitted the bp for Liberty: https://review.openstack.org/#/c/175895 | 14:32 |
*** arnaud___ has quit IRC | 14:33 | |
*** annashen has quit IRC | 14:34 | |
*** Kevin_Zheng has quit IRC | 14:34 | |
johnthetubaguy | tojuvone: I just think we need to step back a bit a look at the wider picture before approving these changes, its getting very confusing right now, I will try write something up about that all soon | 14:35 |
*** achanda has joined #openstack-nova | 14:35 | |
*** iamjarvo has quit IRC | 14:36 | |
bauzas | johnthetubaguy: tojuvone: sorry I was afk for the last 2 days because I had to run a meetup but I would like to review your changes again | 14:37 |
*** iamjarvo has joined #openstack-nova | 14:38 | |
*** claudiub has quit IRC | 14:38 | |
openstackgerrit | Sean Dague proposed openstack/python-novaclient: refactor functional test base class to no inherit from tempest_lib https://review.openstack.org/175854 | 14:38 |
openstackgerrit | Sean Dague proposed openstack/python-novaclient: wip: add ips to novaclient server manager https://review.openstack.org/175603 | 14:38 |
*** Marga_ has quit IRC | 14:40 | |
*** pixelbeat has quit IRC | 14:41 | |
tojuvone | Ok. The OPNFV Doctor link is in the spec to have some Telco perspective, but really any HA interested would want all those VM and nova-compute states fast and reliably (meaning VMs are surely down if nova-compute marked down). And all this infra side monitoring HW specifically.. would be for external tool. Even configured by deployment to be perfect. | 14:41 |
tojuvone | I could look some material if you would like. | 14:43 |
*** achanda has quit IRC | 14:43 | |
openstackgerrit | Dan Smith proposed openstack/nova: Wedge DB migrations if flavor migrations are not complete https://review.openstack.org/174480 | 14:43 |
dansmith | sdague: just checking in on the grenade branching progress... | 14:43 |
*** artom has joined #openstack-nova | 14:44 | |
openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: Adds support for x509 certificates as keypairs https://review.openstack.org/136458 | 14:44 |
openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: Implements 'microversions' api type - Part 1 https://review.openstack.org/152569 | 14:44 |
openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: Deprecate v1.1 and remove v3 https://review.openstack.org/169378 | 14:44 |
openstackgerrit | Andrey Kurilin proposed openstack/python-novaclient: Implements 'microversions' api type - Part 2 https://review.openstack.org/167408 | 14:44 |
*** jecarey has quit IRC | 14:44 | |
sdague | dansmith: currently blocked on a potential oslo bug | 14:46 |
*** thomasem has quit IRC | 14:46 | |
dansmith | sdague: okay | 14:46 |
sdague | https://bugs.launchpad.net/oslo-incubator/+bug/1446583 | 14:46 |
openstack | Launchpad bug 1446583 in Keystone "services no longer reliably stop in stable/kilo" [Critical,In progress] - Assigned to Julien Danjou (jdanjou) | 14:46 |
alex_xu | dansmith: I'm here | 14:47 |
openstackgerrit | Dan Smith proposed openstack/nova-specs: Add admin-query-any-keypair.rst https://review.openstack.org/175579 | 14:47 |
dansmith | alex_xu: cool, could use your eyes on that ^ | 14:47 |
sdague | that actually has prevented grenade from passing on the d-g change that would enable stable/kilo upgrades | 14:47 |
sdague | because the services aren't stopping on the kilo side all the time | 14:47 |
alex_xu | dansmith: sure, will check now | 14:47 |
dansmith | sdague: le suck | 14:47 |
tojuvone | Thanks guys for the comments and interest :) Turning to more idle mode, evening here and home alone with 3 kids. | 14:47 |
*** iamjarvo has quit IRC | 14:48 | |
johnthetubaguy | vladikr: thanks, paperwork should be sorted for you now | 14:48 |
dansmith | alex_xu: thanks, no rush | 14:48 |
sdague | dansmith: all the branches are there now, so that's the last patch that has to land to unblock you | 14:48 |
dansmith | sdague: okay, I'm not in a rush, I just want to make sure I don't keep us blocked longer than necessary | 14:49 |
sdague | yep | 14:49 |
alex_xu | dansmith: np, will review asap :) | 14:49 |
sdague | I'll let you know as soon as it's ready to go | 14:49 |
sdague | https://review.openstack.org/175010 removes the class vars that confused everyone, even though they've been there for 2 years :) | 14:50 |
*** marun_ has joined #openstack-nova | 14:53 | |
*** marun_ is now known as marun | 14:54 | |
*** iamjarvo has joined #openstack-nova | 14:55 | |
*** Marga_ has joined #openstack-nova | 14:55 | |
*** sileht has quit IRC | 14:57 | |
*** penick has joined #openstack-nova | 14:57 | |
*** zhpan188 has joined #openstack-nova | 14:59 | |
bauzas | all, nova-scheduler meeting now in #openstack-meeting | 15:00 |
*** panbalag has joined #openstack-nova | 15:00 | |
*** zhpan188 has left #openstack-nova | 15:01 | |
dansmith | I don't understand this RST complaint: http://logs.openstack.org/79/175579/2/check/gate-nova-specs-docs/bae69f8/console.html | 15:01 |
dansmith | it's on a line with ``foo bar`` | 15:02 |
dansmith | which seems right to me | 15:02 |
*** wznoinsk has quit IRC | 15:03 | |
jogo | lennyb: o/ | 15:04 |
*** hemnafk is now known as hemna | 15:04 | |
*** edleafe is now known as not_edleafe | 15:05 | |
panbalag | Hi, I'm looking for Travis Ankrom. Could you please ping me if you are on this list? | 15:05 |
*** jecarey has joined #openstack-nova | 15:05 | |
mtreinish | dansmith: I think it's the space at the end of the line | 15:05 |
*** not_edleafe is now known as edleafe | 15:05 | |
*** cbader has joined #openstack-nova | 15:05 | |
*** dprince has quit IRC | 15:05 | |
dansmith | mtreinish: you mean before the closing ``? | 15:05 |
mtreinish | dansmith: when I pulled it locally having the space before the trailing `` screwed up the syntax highlighting on my editor | 15:06 |
mtreinish | dansmith: yeah | 15:06 |
openstackgerrit | Sean Dague proposed openstack/python-novaclient: refactor functional test base class to no inherit from tempest_lib https://review.openstack.org/175854 | 15:06 |
openstackgerrit | Sean Dague proposed openstack/python-novaclient: wip: add ips to novaclient server manager https://review.openstack.org/175603 | 15:06 |
dansmith | mtreinish: hmm, okay I copied another example of that | 15:06 |
dansmith | mtreinish: ah, yeah, that changes my highlighting too | 15:06 |
gilliard | does seem to fix it, on http://rst.ninjs.org/ at least. | 15:06 |
gilliard | Or you can insert a space after the first `` | 15:07 |
vladikr | johnthetubaguy, thanks a lot! | 15:07 |
*** apuimedo_ has quit IRC | 15:07 | |
*** fawadkhaliq has quit IRC | 15:07 | |
openstackgerrit | Dan Smith proposed openstack/nova-specs: Add admin-query-any-keypair.rst https://review.openstack.org/175579 | 15:08 |
*** dave-mccowan has quit IRC | 15:08 | |
*** xgsa has quit IRC | 15:11 | |
*** jaypipes has quit IRC | 15:11 | |
*** beagles is now known as beagles_brb | 15:11 | |
*** mestery has quit IRC | 15:11 | |
panbalag | Hi, has anyone configured SSL options in Nova? | 15:12 |
*** mestery has joined #openstack-nova | 15:12 | |
openstackgerrit | Joshua Hesketh proposed openstack/nova: Fix max_number for migrate_flavor data https://review.openstack.org/175890 | 15:12 |
*** xgsa has joined #openstack-nova | 15:13 | |
*** atuvenie_ has joined #openstack-nova | 15:13 | |
*** lpetrut1 has joined #openstack-nova | 15:14 | |
*** masayukig has quit IRC | 15:15 | |
*** atuvenie has quit IRC | 15:15 | |
*** baoli_ has quit IRC | 15:15 | |
*** yamahata has quit IRC | 15:16 | |
*** lpetrut has quit IRC | 15:16 | |
*** baoli has joined #openstack-nova | 15:17 | |
*** penick has quit IRC | 15:18 | |
artom | Is the openstack etherpad broken for anyone else? Ex: https://etherpad.openstack.org/p/nova-libvirt-meeting-agenda I get a 'Uncaught Error: Attempt to load undefined module' | 15:20 |
bauzas | artom: can see it perfectly | 15:20 |
*** masayukig has joined #openstack-nova | 15:21 | |
artom | bauzas, o_O | 15:21 |
bauzas | artom: try refreshing your cache | 15:21 |
artom | What the crap's going on with my browser then | 15:21 |
artom | bauzas, yep, clearing the cache fixed it, thanks! | 15:22 |
bauzas | np | 15:22 |
*** dave-mccowan has joined #openstack-nova | 15:22 | |
*** MVenesio has quit IRC | 15:23 | |
*** joefides has quit IRC | 15:23 | |
*** tojuvone has quit IRC | 15:24 | |
*** joefides has joined #openstack-nova | 15:24 | |
dansmith | upstream etherpad was upgraded recently, | 15:24 |
*** tojuvone has joined #openstack-nova | 15:24 | |
*** MVenesio has joined #openstack-nova | 15:24 | |
bauzas | johnthetubaguy: do you know if there is any BP for helping fixing the quota issues we have ? | 15:24 |
dansmith | and etherpad is notorious for requiring aggressive cache flushing on the browser side | 15:24 |
*** eglynn has quit IRC | 15:24 | |
*** IanGovett has quit IRC | 15:25 | |
bauzas | johnthetubaguy: can only find https://blueprints.launchpad.net/nova/+spec/lock-free-quota-management | 15:25 |
*** wznoinsk has joined #openstack-nova | 15:26 | |
*** dguitarbite has joined #openstack-nova | 15:27 | |
*** Marga_ has quit IRC | 15:27 | |
*** arnaud___ has joined #openstack-nova | 15:27 | |
bauzas | dansmith: I would just love a read-only mode for etherpad just giving plain HTML without using JS | 15:27 |
bauzas | like ?raw | 15:28 |
dansmith | there is that I think | 15:28 |
*** fawadkhaliq has joined #openstack-nova | 15:28 | |
mriedem | so i guess we're doing rc2 huh | 15:28 |
bauzas | IIRC, the read-only mode still requires javascript to update the content | 15:28 |
*** doron_afk is now known as doron | 15:29 | |
dansmith | https://etherpad.openstack.org/p/nova-libvirt-meeting-agenda/export/txt | 15:29 |
bauzas | dansmith: I owe you a beer | 15:29 |
dansmith | bauzas: noted :) | 15:29 |
* bauzas giving free vouchers | 15:29 | |
johnthetubaguy | bauzas: not sure I have seen one | 15:29 |
bauzas | johnthetubaguy: ack | 15:30 |
johnthetubaguy | mriedem: dansmith: ttx has opened RC2 for us this morning, basically took all bugs with a tag and already committed on master at this point | 15:30 |
johnthetubaguy | bauzas: I think we need one though... | 15:30 |
dansmith | johnthetubaguy: okay cool | 15:30 |
bauzas | johnthetubaguy: https://etherpad.openstack.org/p/liberty-nova-summit-ideas L49 is providing some context, but not yet BPs | 15:30 |
*** Sukhdev has quit IRC | 15:31 | |
*** annashen has joined #openstack-nova | 15:31 | |
*** quaa has joined #openstack-nova | 15:31 | |
*** pask81 has joined #openstack-nova | 15:31 | |
dansmith | johnthetubaguy: oh you mean "already took" as "already +2/merged" ? | 15:32 |
johnthetubaguy | dansmith: stuff that is FixCommitted on master already, allowing those to be cherry picked on stable/kilo | 15:33 |
*** fawadkhaliq has quit IRC | 15:33 | |
dansmith | johnthetubaguy: okay, but we can still decide what to actually apply, right? | 15:34 |
dansmith | there are several already pending against kilo, but they were proposed/kilo I think | 15:34 |
*** fawadkhaliq has joined #openstack-nova | 15:34 | |
johnthetubaguy | dansmith: they got moved I think to stable/kilo | 15:34 |
johnthetubaguy | dansmith: if we don't like whats merged there, we can sure revert them, not sure who was doing the approve on stable, not checked that | 15:35 |
dansmith | ah, mine were all abandoned and I guess recreated | 15:35 |
johnthetubaguy | dansmith: mriedem: you can see it here, if that helps: https://launchpad.net/nova/+milestone/kilo-rc2 | 15:35 |
*** browne has quit IRC | 15:35 | |
*** changbl has joined #openstack-nova | 15:36 | |
*** annashen has quit IRC | 15:36 | |
*** david-lyle has quit IRC | 15:36 | |
dansmith | so this is the list of pending ones: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:stable/kilo,n,z | 15:36 |
dansmith | all the others seem to have already merged? | 15:36 |
mriedem | yeah | 15:36 |
*** thangp has joined #openstack-nova | 15:36 | |
mriedem | https://review.openstack.org/#/q/status:merged+project:openstack/nova+branch:stable/kilo,n,z | 15:36 |
*** IanGovett has joined #openstack-nova | 15:36 | |
johnthetubaguy | ah, I guess ttx merged those: https://review.openstack.org/#/c/174078/ them seemed OK for backporting when I checked them, but happy to revert that if required | 15:37 |
mordred | so that I can learn something ... | 15:39 |
dansmith | sigh, well, | 15:39 |
mordred | https://bugs.launchpad.net/nova/+bug/1223345 | 15:39 |
openstack | Launchpad bug 1223345 in OpenStack Compute (nova) "Soft-deleted instance files deleted by periodic cleanup task" [High,Fix released] - Assigned to Hans Lindgren (hanlind) | 15:39 |
mordred | why is it that someone wouldn' want soft-deleted instanaces cleaned up? | 15:39 |
mordred | and/or what if someone _did_ want soft-deletes to just diaf | 15:39 |
dansmith | mordred: because soft-deleted instances are supposed to be undeletable | 15:39 |
mordred | aroo? | 15:39 |
dansmith | johnthetubaguy: I guess the rest of the ones up there are important, and not worth reverting anything already merged | 15:40 |
mordred | dansmith: can I translate that to mean "it is the intent of nova to keep soft-deleted rows in the database until the end of time" ? | 15:40 |
dansmith | mordred: there is a "restore instance" call that undeletes a thing | 15:40 |
mordred | ah | 15:40 |
*** amotoki has quit IRC | 15:41 | |
dansmith | mordred: well, I think in the case of soft-deleted instances there is an interval for which they're still available, IIRC, but I don't really know | 15:41 |
*** andreykurilin__ has joined #openstack-nova | 15:41 | |
dansmith | mordred: but as you know, nova has a confused relationship with deleted things | 15:41 |
mordred | it just seems that with a cleanup thread, having a longer undelete-available interval would be friendlier to operators | 15:41 |
mordred | dansmith: well, I'm learning this | 15:41 |
*** kaisers has quit IRC | 15:41 | |
mordred | dansmith: I don't really think I fully grokked the extent of the confusing relationship | 15:41 |
ttx | johnthetubaguy: we shouldn't have merged that one ? | 15:41 |
ttx | johnthetubaguy: is it the bug is invalid or the fix in invalid ? | 15:42 |
dansmith | mordred: we have soft delete of instances, plus soft-delete of rows, plus shadow tables, plus zero ability to define when something can be moved from soft-deleted to shadow without breaking things | 15:42 |
dansmith | mordred: (AFAIK) | 15:42 |
mordred | dansmith: that's so exciting | 15:42 |
clarkb | mordred: dansmith is there a way as a user to indicate when deleting an instance that i neverwant to be able to restore the instance so just really delete it? | 15:43 |
johnthetubaguy | ttx: not sure at this point, just checking in general with other folks and if they are happy with all the ones we backported | 15:43 |
dansmith | mordred: it's like russian roulette.. with your production DB :) | 15:43 |
*** amotoki has joined #openstack-nova | 15:43 | |
*** amotoki has quit IRC | 15:43 | |
dansmith | clarkb: soft delete is different than delete from a user perspective I think | 15:43 |
johnthetubaguy | dansmith: thanks for checking that lift for me, they all seemed worth backporting | 15:43 |
dansmith | clarkb: i.e. delete means "I don't want to maybe undelete" | 15:43 |
mriedem | there is a command to move deleted rows to shadow tables http://git.openstack.org/cgit/openstack/nova/tree/nova/cmd/manage.py#n920 | 15:44 |
openstackgerrit | Doug Hellmann proposed openstack/python-novaclient: Update README to work with release tools https://review.openstack.org/175947 | 15:44 |
ttx | johnthetubaguy: looks like it was originally proposed for release by a "Dan Smith". Probably an alias. | 15:44 |
mriedem | but that doesn't 'delete' things | 15:44 |
clarkb | dansmith: I don't think that is what we are experiencing | 15:44 |
dansmith | ttx: some of the proposed things were good, others I think needed some discussion | 15:44 |
clarkb | but mordred has actual details | 15:44 |
dansmith | ttx: not worth it at this point | 15:44 |
mordred | I don't suppose there is an option which says "just perform the SQL delete and don't do soft-deletes or shadowtables kthxbia" is there? | 15:44 |
dansmith | mordred: nope | 15:45 |
dansmith | mordred: we never actually delete anything ever | 15:45 |
dansmith | mordred: the closest is moving something from real to shadow tables | 15:45 |
dansmith | mordred: except for a few cases of smaller things, but nothing you have visibility to I tink | 15:45 |
ttx | dansmith, johnthetubaguy: ok well, keep me posted | 15:45 |
mordred | dansmith: well - I mean - infra creates and deletes 20k instances a day on our clouds | 15:45 |
dansmith | mordred: clarkb: just in defense of everyone where, this has been the case since the dawn of time, in case it's not clear :) | 15:46 |
mordred | dansmith: so we wind up producing HUGE amounts of data in those tables over time | 15:46 |
johnthetubaguy | mordred: we actually go and delete those records after a month now | 15:46 |
dansmith | mordred: right, actual cleaning of the DB is left as "an exercise for the operator" :/ | 15:46 |
mordred | dansmith: that's so friendly | 15:46 |
johnthetubaguy | mordred: and yes, the tables got very large | 15:46 |
dansmith | mordred: see ^ :) | 15:46 |
mordred | johnthetubaguy: but by we - you mean rackspace, not nova? | 15:46 |
dansmith | yes, that's what he means | 15:46 |
dansmith | my "we" is nova, his "we" is actual people | 15:47 |
johnthetubaguy | mordred: yeah, sorry, I mean public cloud at rackspace in that case, different hat on there for a moment | 15:47 |
mordred | johnthetubaguy: if both of the public cloud providers are now deleting the rows on a cronjob - why not add an option to nova.conf to have the cleanup thread delete rows when it runs? | 15:47 |
*** baoli has quit IRC | 15:47 | |
mriedem | johnthetubaguy talks in the royal 'we' now that he's PTL :) | 15:47 |
johnthetubaguy | dansmith: I just use we for both and let folks guess, my bad | 15:47 |
dansmith | johnthetubaguy: heh | 15:47 |
mordred | I mean, it seems like nova is already doing that sort of thing - cleaning up after itself would be a friendly thing for the operators | 15:47 |
dansmith | we[0] == rax, w[1] == nova | 15:47 |
mordred | dansmith: :) | 15:48 |
johnthetubaguy | lol | 15:48 |
dansmith | mordred: well, part of the problem is, it's dangerous and hard to actually know when it's okay in some cases | 15:48 |
dansmith | mordred: so we shift the risk to the operators | 15:48 |
mordred | dansmith: that seems like a bug | 15:48 |
dansmith | mordred: I don't really disagree :) | 15:48 |
mordred | :) | 15:48 |
dansmith | mordred: now that we have flavors separated from instances, we could get a lot closer | 15:49 |
johnthetubaguy | mordred: for context, turns out we updated our script so it doesn't delete so fast if it sees the read slaves lagging behind the master too log, sigh, turns out its deployment specific, but I should get that uploaded somewhere public | 15:49 |
dansmith | mordred: but in the past, it has been because of linkages among data structures | 15:49 |
*** baoli has joined #openstack-nova | 15:49 | |
mordred | dansmith: hang on ... | 15:49 |
* dansmith waits for an incoming missile from mordred | 15:49 | |
mordred | dansmith: you're telling me taht even though we do the thing that makes me unhappy and actually cause sqlalchemy to produce foreign keys in the database which is the most insane stupid thing ever for scalability | 15:49 |
mordred | we ALSO don't actually gain the data protection from doing so? | 15:50 |
mordred | so all of teh production scalability cost with none of the tenuous benefits? | 15:50 |
mriedem | we don't use cascading deletes on FKeys either | 15:50 |
dansmith | mordred: we do in some cases, but not all I think probably, yeah | 15:50 |
dansmith | yeahg | 15:50 |
* mordred gets out the stabbing stick | 15:50 | |
dansmith | mordred: again, this predates many of the folks here | 15:50 |
dansmith | so don't stab us :) | 15:50 |
mordred | dansmith: oh - I know who to blame ... | 15:51 |
mriedem | i wouldn't mind a good stabbing | 15:51 |
*** atuvenie_ has quit IRC | 15:51 | |
* mordred stabs mriedem on general principle | 15:51 | |
*** fawadkhaliq has quit IRC | 15:51 | |
dansmith | so violent! | 15:51 |
* mriedem enjoys the warm blood oozing | 15:51 | |
*** matrohon has quit IRC | 15:51 | |
mordred | so I now have 3 things to rage-code on in nova ... | 15:51 |
bauzas | mordred: there was an abandoned spec about something on trigger https://review.openstack.org/#/c/132656/ | 15:51 |
*** fawadkhaliq has joined #openstack-nova | 15:51 | |
*** otter768 has joined #openstack-nova | 15:51 | |
* johnthetubaguy feels like I should get the use of some kind of helpline number for use by community members | 15:52 | |
dansmith | at least for mordred ... | 15:52 |
*** baoli has quit IRC | 15:52 | |
mriedem | what were the other 2 things? | 15:52 |
bauzas | eh :) | 15:52 |
dansmith | you had to ask... | 15:52 |
mordred | well - thing one is I'm going to go absolutely postal on the assinine table locks related to quotas | 15:52 |
*** sdake is now known as sdake-afk | 15:52 | |
mordred | beacuse BONGHITS | 15:53 |
mriedem | you missed 420 by a day | 15:53 |
mriedem | sorry | 15:53 |
johnthetubaguy | mordred: best bit, it doesn't work | 15:53 |
mordred | thing two is I'm going to finish that spec that was just linked because nova should clean after itself | 15:53 |
dansmith | the quota stuff is problematic, I think that's well known | 15:53 |
mordred | thing three is that I'm going to delete all foreign keys with prejudice because they're stupid and not giving any value | 15:53 |
*** baoli_ has joined #openstack-nova | 15:53 | |
bauzas | mordred: be happy, we just removed one | 15:54 |
mriedem | given we don't cascade anything on the FKeys i'd agree with that | 15:54 |
mordred | bauzas: yay! | 15:54 |
mordred | mriedem: I know of literally zero large-scale mysql sites that use them | 15:54 |
bauzas | mordred: but wrt live upgrades, we are still paying debt by keeping them till the next release | 15:54 |
mordred | and I know of literally no large-scale postgres sites :) | 15:54 |
sdague | mordred: they all emulate it in the application? | 15:54 |
mordred | sdague: yup | 15:54 |
mordred | sdague: it's too costly to do them in teh db | 15:55 |
mordred | the db is the hardest scaling point to have things centralized | 15:55 |
*** atuvenie_ has joined #openstack-nova | 15:55 | |
dansmith | well, | 15:55 |
sdague | so... isn't that the whole point of the R in RDMS ? | 15:55 |
*** ijw has quit IRC | 15:55 | |
mordred | same reason using stored procedures is a nono in large deployments | 15:55 |
mordred | sdague: yse | 15:55 |
dansmith | with our objects facade we have a good place to do that emulation | 15:55 |
dansmith | as we do for flavors now | 15:55 |
mordred | sdague: but you don't need it enforced by the engine | 15:55 |
mordred | you just want the relational algebra | 15:55 |
*** ijw has joined #openstack-nova | 15:55 | |
sdague | ok | 15:55 |
sdague | dansmith: yeh, what's left in the stale references? | 15:56 |
*** otter768 has quit IRC | 15:56 | |
dansmith | sdague: parse error | 15:56 |
*** dboik has quit IRC | 15:56 | |
mordred | however - quotas and deletes are top on my list | 15:56 |
sdague | so, flavor is now delete safe, because we keep a copy of the flavor we were started with | 15:57 |
mordred | well, after I finish coding my "please give me the ssh-host-key of the instance" feature | 15:57 |
*** dboik has joined #openstack-nova | 15:57 | |
*** sileht has joined #openstack-nova | 15:57 | |
dansmith | sdague: yeah | 15:57 |
sdague | what are the other problematic relationships that still exist and can explode if we just deleted stuff | 15:57 |
*** dboik has quit IRC | 15:58 | |
*** dboik has joined #openstack-nova | 15:58 | |
artom | mordred, the bit that mriedem linked though, http://git.openstack.org/cgit/openstack/nova/tree/nova/cmd/manage.py#n920 | 15:58 |
dansmith | sdague: probably lots. IPs, networks, | 15:58 |
dansmith | sdague: I could pull up the model if you want, but there are things | 15:58 |
artom | That + something that empties out the shadow tables == a really ugly delete, no? | 15:58 |
*** armax has left #openstack-nova | 15:58 | |
*** Sukhdev has joined #openstack-nova | 15:58 | |
sdague | yeh, ips and networks seem a little trickier | 15:59 |
dansmith | yeah | 15:59 |
*** dboik has quit IRC | 15:59 | |
dansmith | metadata entries | 15:59 |
*** dboik has joined #openstack-nova | 15:59 | |
dansmith | actually, most everything | 16:00 |
dansmith | consoles, dns stuff | 16:00 |
*** armax has joined #openstack-nova | 16:00 | |
*** ijw has quit IRC | 16:00 | |
dansmith | basically you just have to run through almost all models, delete where instance_uuid=$uuid | 16:00 |
dansmith | then you can delete the real instance | 16:00 |
dansmith | most things are instance-centric | 16:00 |
dansmith | so if you delete them before the instance, you're okay | 16:00 |
dansmith | but there are things the instance depends on | 16:01 |
sdague | dansmith: shouldn't it be the other way around? | 16:01 |
*** tsekiyama has joined #openstack-nova | 16:01 | |
sdague | delete the instance first | 16:01 |
bauzas | sdague: you can't, FK | 16:01 |
*** armax has left #openstack-nova | 16:01 | |
mriedem | right, everything has a fkey on instances.uuid | 16:01 |
ndipanov | what's the problem with FKs again | 16:01 |
dansmith | sdague: well, things that are discoverable from the API would hit InstanceNotFound errors if they get listed while you've got the instance deleted | 16:01 |
ndipanov | I missed that class | 16:01 |
mriedem | so you have to do what i did in my cleanup script for null instances.uuid | 16:01 |
*** armax has joined #openstack-nova | 16:02 | |
*** vilobhmm1 has joined #openstack-nova | 16:02 | |
*** david-lyle has joined #openstack-nova | 16:02 | |
bauzas | ndipanov: the problem is at scale how to purge the DB | 16:02 |
*** Marga_ has joined #openstack-nova | 16:02 | |
dansmith | sdague: then there are things like the instance->info_cache dependency | 16:02 |
*** baoli_ has quit IRC | 16:02 | |
bauzas | ndipanov: and FKs make the cleanup a little harder | 16:02 |
dansmith | not sure which way that needs to go | 16:02 |
sdague | right, gotcha | 16:02 |
ndipanov | bauzas, well for the right reasons | 16:03 |
ndipanov | so that you don't have to think about that sort of consistency in your app | 16:03 |
bauzas | ndipanov: here, the question is : should that be abstracted for operators ? | 16:03 |
*** redbeard has quit IRC | 16:04 | |
*** atuvenie_ has quit IRC | 16:04 | |
bauzas | ndipanov: atm, this is a DIY stuff | 16:04 |
*** baoli has joined #openstack-nova | 16:04 | |
ndipanov | bauzas, abstracted as - something like shadow-tables so taht you can clean things up but that doesn't suck? | 16:04 |
bauzas | ndipanov: abstracted like - I want to ask Nova to really delete my deleted instances forever | 16:05 |
*** sahid has joined #openstack-nova | 16:05 | |
bauzas | I'm just a little concerned about all the locks we could get on the tables, wrt all the FKs we have :) | 16:06 |
dansmith | can someone +W this please? | 16:06 |
dansmith | https://review.openstack.org/#/c/175890/2 | 16:06 |
bauzas | it's like "push the button and we freeze all the tables to your convenience" | 16:06 |
*** pixelb has joined #openstack-nova | 16:06 | |
artom | Pfft, the real solution is to get Western Digital to sponsor OpenStack and ship a free disk or two to all operators. | 16:06 |
dansmith | johnthetubaguy: that's a bug, but probably not worth putting in RC2 and just backporting, but I'll let you decide | 16:06 |
*** ildikov has quit IRC | 16:06 | |
mordred | ndipanov: anything that is for the reason "so you don't have to think about the consistency in your app" is a mistake | 16:07 |
mordred | because the app's job is to think about the consistency | 16:07 |
mordred | and there is a model class | 16:07 |
mordred | that encapsulates that | 16:07 |
ndipanov | mordred, well why use relational dbs then | 16:07 |
mordred | that the logic about how to perform basic tasks is split across two places and almost nobody understands how to safely perform a task like "delete teh instance records" is a clear indication that the theory is not working | 16:07 |
mordred | ndipanov: because there are HUGE benefits to the data model | 16:08 |
dansmith | mordred: probably worth noting that you're talking about using them at crazy scale, because the point of the constraints is defining what you want the DB to enforce for you in a normal case | 16:08 |
mordred | ndipanov: you don't need db level enforcement of the foreign keys to reap them | 16:08 |
*** ijw has joined #openstack-nova | 16:08 | |
mordred | dansmith: yes | 16:08 |
mordred | this is why defining them in the sqlalchemy model is a GREAT idea | 16:08 |
mordred | actually applying them at the db level is the thing that does not scale and provides no benefit | 16:08 |
ndipanov | mordred, k - in that case I'd say we should look at whether our data model is sane for how we use it vs. blanket kill FKs | 16:08 |
dansmith | ndipanov: thought you knew, mordred has two settings: "off" and "kill with fire" | 16:09 |
mordred | ndipanov: well, there are zero cases in which a production database should actually have FKs applied to it - but I agree, working on the data model is actually the important task - not fighting with me about megascale database things :) | 16:09 |
bauzas | yay, let's denormalize stuff ! | 16:09 |
*** atuvenie_ has joined #openstack-nova | 16:10 | |
mordred | well | 16:10 |
mordred | we having the good table and relationships == good | 16:10 |
mordred | having some summary tables (materialized views) == performance optimization, also good | 16:10 |
ndipanov | agreed | 16:10 |
*** ijw_ has joined #openstack-nova | 16:11 | |
bauzas | +1 | 16:11 |
*** moshele has quit IRC | 16:11 | |
*** dsanders has joined #openstack-nova | 16:11 | |
mordred | quotas, for instance, should be being run on a rollup/summary table rather than directly on the underlying information tables | 16:11 |
mordred | because you can get actual atomic update operations that way | 16:11 |
mordred | and if somethign goes wrong, you can always rebuild the summary table from the base data | 16:12 |
ndipanov | I don't know - I get the argument that with fk's your encoding logic into your db schema and in your app | 16:12 |
artom | mordred, isn't it the RDBMS's job to ensure atomicity? | 16:12 |
mordred | so no need for locking _at_ _all_ and blinding fast speed and at least as semantically correct as current | 16:12 |
mordred | artom: only if you want to write academic papers | 16:12 |
*** ijw has quit IRC | 16:12 | |
ndipanov | I just can't help but think that killing them you trade one class of bugs for another | 16:12 |
mordred | ndipanov: put them in teh app model - awesome | 16:13 |
artom | mordred, why is it bad in real life? | 16:13 |
mordred | artom: because they are a scaling bottleneck | 16:13 |
artom | Ah, that's a different thing | 16:13 |
mordred | and if your app doesn't understand the object model you're screwed anyway | 16:13 |
mordred | also - once you shard data into more than one db | 16:13 |
mordred | you can't use them anyway | 16:13 |
artom | mordred, if you want scale don't chose RDBMs | 16:13 |
mordred | artom: wrong | 16:13 |
artom | That's the real problem | 16:13 |
mordred | artom: categorically untrue | 16:14 |
bauzas | artom: not really, it's about how you shard your model | 16:14 |
*** doron is now known as doron_afk | 16:14 | |
mriedem | mordred: are any other openstack projects using views? | 16:14 |
mordred | artom: and I will happily point you at facebook adn google as my examples | 16:14 |
mordred | mriedem: you don't actually want VIEWs | 16:14 |
openstackgerrit | Dan Smith proposed openstack/nova: Adding user_id handling to keypair index api call https://review.openstack.org/70485 | 16:14 |
mordred | mriedem: you want app-generated summary tables | 16:14 |
mordred | views in mysql are not materialized and are a terrible idea in production | 16:14 |
artom | mordred, I know Facebook uses MySQL, don't know how they scale it though. | 16:14 |
mordred | artom: I do :) | 16:14 |
artom | mordred, didn't know Google used MySQL at any sort of scale. | 16:15 |
artom | mordred, well, how? :) | 16:15 |
mordred | artom: the main money maker for google, adsense, runs 100% on mysql | 16:15 |
ndipanov | so once you get rid of the FKs, you also say I won't use a number of feature given to me by MCC your db implements and you have to be careful about thos in your code too | 16:15 |
mordred | artom: sharding | 16:15 |
mordred | ndipanov: yes | 16:15 |
ndipanov | sometimes that's OK | 16:15 |
mordred | but you have to be careful in your code _Anyway) | 16:15 |
ndipanov | but sometimes you open yourself to ugly race conditions | 16:15 |
mordred | sure | 16:15 |
ndipanov | right that;s true | 16:15 |
mordred | main thing is - the db - rdbms or not, is not a magic wand | 16:16 |
mordred | it's not going to design an app for yoy | 16:16 |
artom | mordred, and in what way sharding prevents FKs/atomicity? | 16:16 |
ndipanov | yes true | 16:16 |
mordred | artom: FKs don't work across servers | 16:16 |
mordred | artom: when part of your model is on one server and part is on another server, you MUST handle the logic in your code | 16:16 |
artom | mordred, wait, model? | 16:16 |
bauzas | artom: usually, you want to shard per col | 16:17 |
bauzas | artom: and not per row | 16:17 |
mordred | artom: yes. say you have your instances split across 10 servers - and there isa whole other server for flavors | 16:17 |
artom | bauzas, ah! | 16:17 |
mordred | bauzas: you might want to do both | 16:17 |
*** beagles_brb is now known as beagles | 16:17 | |
mordred | sharding by row is known as horizontal sharding, sharding by column as vertical | 16:17 |
bauzas | mordred: yay, was just giving the reason why it doesn't work :) | 16:17 |
artom | But then haven't you basically recreated a key-value store in SQL? | 16:17 |
mordred | and big places tend to use a combination of both | 16:17 |
mordred | bauzas: :) | 16:17 |
mordred | artom: nope | 16:18 |
artom | Since all the logic is in the code anyways? | 16:18 |
*** romainh has left #openstack-nova | 16:18 | |
mordred | artom: because on a per-server basis you still get the benefits of indexes and joins | 16:18 |
bauzas | artom: that's just trade-off | 16:18 |
mordred | artom: so what you have is sharded indexing | 16:18 |
bauzas | +1 | 16:18 |
*** fawadkhaliq has quit IRC | 16:18 | |
mordred | and properly design indexes and joins are super useful | 16:18 |
artom | Key-value stores tend to have indeces (or equivalents), no? | 16:19 |
mordred | they're getting there | 16:19 |
bauzas | artom: in general, you shard wrt your model to make joins locally | 16:19 |
*** doron_afk has quit IRC | 16:19 | |
mordred | and key/value stores are useful in some cases | 16:19 |
bauzas | artom: that would speed up local joins while your app can still scale | 16:19 |
mordred | yup | 16:19 |
mordred | artom: often times people will use a key-value layer on top of the rdbms layer as a smart-cache | 16:20 |
*** fawadkhaliq has joined #openstack-nova | 16:20 | |
mordred | so instead of making a summary table | 16:20 |
mordred | you make a key-value entry with the summary in the blob | 16:20 |
*** iamjarvo has quit IRC | 16:20 | |
mordred | but you also but the main data into the rdbms | 16:20 |
mordred | so that you don't have to care about the data longevity in the key-value | 16:20 |
mordred | since most of them do not have the production usage to be at the point of knowing their failure domains well enough to trust them as the ultimate store of all of your things | 16:21 |
mordred | there isn't a single right answer for any of this - other tahn FKs being bong :) | 16:21 |
*** jistr has quit IRC | 16:21 | |
*** Sukhdev has quit IRC | 16:22 | |
artom | One day too late for bong ;) | 16:22 |
*** doude has quit IRC | 16:22 | |
*** xgsa has quit IRC | 16:23 | |
artom | And yet after all this, we're no closer to having a clean delete in Nova :) | 16:23 |
*** fawadkhaliq has quit IRC | 16:23 | |
*** greghaynes has joined #openstack-nova | 16:23 | |
artom | Since I don't think a patch called "DELETE ALL FKS FOREVAR" will get merged anytime soon ;) | 16:24 |
mriedem | oh idk | 16:24 |
*** vilobhmm1 has quit IRC | 16:24 | |
artom | (Thanks for the answers though, I've learned) | 16:24 |
mriedem | push a patch with the db migration that does it, run it through check to see if there are bugs | 16:24 |
bauzas | mriedem: agreed, I think it's doable | 16:25 |
bauzas | artom: I made the move for the service/compute FK and that was really not an issue | 16:25 |
mriedem | yeah i don't really see a problem with it since the FKeys don't do anything since they don't cascade anything | 16:25 |
artom | Maybe nothing currently existing will break, but I feel like that's opening up a whole slew of new bugs as ndipanov said. | 16:26 |
bauzas | artom: the main problem I had was that I was providing a new column for doing the denormalization and *that* was the problem | 16:26 |
ndipanov | so a quick search on google seems to imply taht innodb should not have a massive overhead on locking for FKs | 16:26 |
mriedem | artom: but we can still orphan deleted stuff with fkeys today since we don't have ondelete=cascade | 16:26 |
ndipanov | iow - if there is massive locking caused by FKs - your model or queries are likely wrong | 16:27 |
bauzas | ndipanov: I can ask some peers I know about that, but I think it's not really a black-or-white answer | 16:27 |
*** dsanders has quit IRC | 16:27 | |
*** dsanders has joined #openstack-nova | 16:27 | |
ndipanov | what do you mean | 16:27 |
bauzas | ndipanov: that depends on your reads and writes mostly | 16:27 |
ndipanov | I literally just said the exact same thing | 16:27 |
* mriedem is surprised no one has summoned zzzeek yet | 16:28 | |
* zzzeek is making his lunch so will be here in 10 | 16:28 | |
bauzas | ndipanov: then I missed it :) | 16:28 |
* mriedem goes to lunch | 16:28 | |
ndipanov | well I said it in a slightly inverted way | 16:28 |
*** pask81 has quit IRC | 16:29 | |
*** dprince has joined #openstack-nova | 16:29 | |
bauzas | ndipanov: well you implied that having locks means that your model is wrong IIUC | 16:29 |
*** BobBall has quit IRC | 16:29 | |
bauzas | ndipanov: I'm not that explicit | 16:30 |
*** jaypipes has joined #openstack-nova | 16:30 | |
*** jlvillal has quit IRC | 16:30 | |
*** exploreshaifali has joined #openstack-nova | 16:30 | |
*** jlvillal has joined #openstack-nova | 16:30 | |
bauzas | ndipanov: like table A and table B have relationships, but both can be written for good reasons | 16:30 |
ndipanov | bauzas, so as an example - if most of your modifications include something that will lock the related table | 16:30 |
ndipanov | then that field should not be a FK for sure | 16:30 |
bauzas | I see | 16:31 |
artom | mriedem, so there's no cascade... then FKs are currently used just to... enforce relational integrity? | 16:31 |
*** tojuvone has quit IRC | 16:31 | |
ndipanov | otherwise - it seems to me that you get stuff for mostly free | 16:31 |
artom | And not even that, since, well, delete | 16:31 |
ndipanov | iow - you need to consider the behavior of your app | 16:32 |
*** annashen has joined #openstack-nova | 16:32 | |
ndipanov | I have no idea what mordred was seeing wrt quotas but I can well believe that it;s fubar | 16:32 |
zzzeek | looks like its something about FKs. Please use adequate normalization and please use FKs. they work great. that is all | 16:32 |
ndipanov | thanks! :) | 16:32 |
ndipanov | what I am not sure about is "nuke FKs" as a remedy :) | 16:33 |
mordred | zzzeek: he. | 16:33 |
mordred | zzzeek: I do and always have disagreed with you | 16:33 |
mordred | but that's ok | 16:33 |
mordred | zzzeek: FKs in sqlalchemy model == yes | 16:33 |
mordred | zzzeek: FKs applied to db == silly | 16:34 |
artom | mordred, wait, you can separate the two? I thought it was pretty much a 1:1 mapping... | 16:34 |
zzzeek | mordred: it’s nice to know that a DB will never have orphaned records that crash the app | 16:35 |
zzzeek | artom: theres the notion that the schema has FK-like relationships, then there is the constraint itself | 16:35 |
*** Mike_D_laptop has joined #openstack-nova | 16:35 | |
mordred | zzzeek: sure. but for all the reasons listed above, if you depend on the rdbms catching taht for you, you are limiting how well you can scale your app if you need to grow past a single server | 16:35 |
artom | zzzeek, oh, so pulling an example out of my ass, something like "network_id" in the instances table without actually setting the FK constraint? | 16:36 |
mordred | zzzeek: which is why you should for sure design as if you had them | 16:36 |
mordred | zzzeek: and in fact, in dev/test you shoudl apply them to the db | 16:36 |
zzzeek | mordred: OK there’s a lot in there. “grow past a single server”, that’s a whole ball of wax | 16:36 |
mordred | so that all of your tests will catch it | 16:36 |
mordred | zzzeek: yah | 16:36 |
zzzeek | can mean many things | 16:36 |
*** Marga_ has quit IRC | 16:36 | |
zzzeek | mordred: if tables are vertically sharded across DBs, then yes, having an FK is clearly not worth it | 16:36 |
*** annashen has quit IRC | 16:37 | |
zzzeek | mordred: if tables are horizontally sharded, then you are probably crazy anyway :) | 16:37 |
*** Marga_ has joined #openstack-nova | 16:37 | |
mordred | zzzeek: :) | 16:37 |
mordred | zzzeek: what if they're both? | 16:37 |
zzzeek | mordred: we are 95% in agreement. if youre just saying, “turn them off in production”, but otherwise they’re fully present both conceptially as well as physically in test environments, fine | 16:37 |
mordred | zzzeek: yes. that's all I'm saying | 16:37 |
*** iamjarvo has joined #openstack-nova | 16:37 | |
zzzeek | mordred: sure | 16:38 |
*** mpaolino has quit IRC | 16:38 | |
*** Marga_ has quit IRC | 16:38 | |
ndipanov | well you can do that even now no? | 16:38 |
mordred | zzzeek: in fact, thanks - I think I should be clearer when I describe my POV on this | 16:38 |
*** Marga_ has joined #openstack-nova | 16:38 | |
*** arnaud___ has quit IRC | 16:39 | |
zzzeek | in Oracle we jsut emit some system setting to do it. on MySQL, I dont know that there’s just a schema- or table-wise setting for it, you maybe have to emit different DDL. our migration setups dont quite ahve that feature right now though it is doable | 16:39 |
*** derekh has quit IRC | 16:39 | |
ndipanov | I see | 16:39 |
*** ijw_ is now known as ijw | 16:40 | |
*** exploreshaifali has quit IRC | 16:40 | |
openstackgerrit | Sean Dague proposed openstack/python-novaclient: add ips to novaclient server manager https://review.openstack.org/175603 | 16:41 |
sdague | mordred: ^^^ gives you the extra call you wanted in novaclient | 16:43 |
*** browne has joined #openstack-nova | 16:43 | |
*** sdake-afk has quit IRC | 16:44 | |
*** jichen has quit IRC | 16:45 | |
mordred | sdague: yay! | 16:45 |
*** annegentle has quit IRC | 16:47 | |
*** READ10 has joined #openstack-nova | 16:47 | |
ndipanov | interesting discussion - we should do more of those... | 16:47 |
mordred | ndipanov: I agree! | 16:48 |
*** dsanders has quit IRC | 16:49 | |
*** kaisers has joined #openstack-nova | 16:50 | |
bauzas | dansmith: re: https://review.openstack.org/#/c/164762/9 I wonder why we couldn't put the logic outside of the object | 16:50 |
*** mpaolino has joined #openstack-nova | 16:50 | |
bauzas | dansmith: I saw you were not against the idea of adding a new remotable method for a specific usecase hence my wonders | 16:51 |
bauzas | dansmith: I don't really like having myspecificmethodformysingleuse() being remotable | 16:52 |
bauzas | well, I'm not saying having it remotable is bad, just that it makes things worst | 16:52 |
dansmith | bauzas: well, the less specific it is, the more data we send over rpc just to do the filtering on the other side | 16:53 |
bauzas | dansmith: agreed, but here the payload is really slow | 16:53 |
dansmith | bauzas: I agree that adding them for no reason is just debt, but if there's a good reason, it makes sense | 16:54 |
bauzas | s/slow/small | 16:54 |
bauzas | dansmith: and like I said, it's just a matter of how many instances we have per group | 16:54 |
bauzas | see the other https://review.openstack.org/#/c/169489/6/nova/objects/instance_group.py,cm | 16:54 |
bauzas | dansmith: in 2 patches, we're just adding 2 new remotable methods for just a single usecase that can be done in the same loop | 16:55 |
dansmith | in this case, I'm not sure doing this outside the remotable is still the same thing | 16:56 |
bauzas | dansmith: iterating over the list of computes is the same | 16:56 |
dansmith | you'd have to request all the instances on all hosts that are represented in the group, which could be large for even a small group, right? | 16:56 |
bauzas | dansmith: so we can do both checks in the same loop | 16:56 |
bauzas | dansmith: yeah that's O(n) | 16:57 |
*** harlowja_away is now known as harlowja | 16:57 | |
bauzas | dansmith: but that sounds a small price to pay IMHO, in particular if 2 methods are doing that | 16:57 |
bauzas | dansmith: so we can just do the loop once | 16:58 |
dansmith | but instances are really large | 16:58 |
dansmith | and also, | 16:58 |
*** annashen has joined #openstack-nova | 16:59 | |
bauzas | well, we're getting the list of hosts whose at least one instance is in the server group | 16:59 |
dansmith | I feel like there is a certain amount of atomicity we may want to have for something like this that we can't guarantee across multiple calls | 16:59 |
dansmith | granted, the point is that ids are monotonically increasing, but that could be different with a different backend | 17:00 |
bauzas | dansmith: that's also another concern I didn't raise | 17:00 |
bauzas | dansmith: because I don't like checking against IDs | 17:00 |
bauzas | that's just assuming that it won't change | 17:00 |
bauzas | dansmith: so before commenting saying I was disagreeing, I have to think why it's just all of that needed | 17:01 |
dansmith | bauzas: right, but the semantics of the call do not imply comparison by id, which means we could change it later without the client needing to know | 17:01 |
bauzas | like saying that we add an instance in a group before even making sure that instance can be booted seems like a bit optimistic to me | 17:01 |
dansmith | I don't follow | 17:02 |
bauzas | sorry | 17:02 |
bauzas | the problem is that the instances being booted are already part of the group, right ? | 17:02 |
dansmith | we have to put it into the group/host before we start booting it and then if it fails, we roll back | 17:02 |
dansmith | no | 17:02 |
dansmith | it's that the instance is assigned to the host | 17:02 |
*** vilobhmm1 has joined #openstack-nova | 17:04 | |
bauzas | dansmith: so the validation call is getting the host, I see | 17:04 |
*** vilobhmm11 has joined #openstack-nova | 17:04 | |
*** davideagnello has joined #openstack-nova | 17:04 | |
bauzas | dansmith: in that case, why not filtering out get_hosts using the instance_uuid ? | 17:04 |
bauzas | I mean using the exclude arg ? | 17:05 |
dansmith | sorry, I'm really confused | 17:05 |
*** Sukhdev has joined #openstack-nova | 17:05 | |
dansmith | we filter by the uuids in the group | 17:05 |
bauzas | I think I'm confused too | 17:06 |
bauzas | okay, before discussing further I need to better understand what the problem is | 17:06 |
bauzas | because I probably misunderstood | 17:06 |
davideagnello | hello, I have been trying out Tempest and Rally for integration and scenario testing for stackforge project Cue. Nova has tests in both frameworks, I would like to get some feedback tempest vs. rally, using one or both. | 17:06 |
mriedem | davideagnello: you probably want to be in the #openstack-qa channel | 17:07 |
davideagnello | it looks like we are going to use tempest for integration tests and rally for e2e scenario testing | 17:07 |
davideagnello | mriedem: ok, thanks. Just wanted to get some feedback from Nova group on using these frameworks. Will post there as well. | 17:08 |
mriedem | nova doesn't gate on rally, only tempest right now | 17:08 |
*** david-lyle has quit IRC | 17:08 | |
*** vilobhmm1 has quit IRC | 17:08 | |
davideagnello | mriedem: ok, any reason to not include rally as well? | 17:09 |
mriedem | i'm not really sure honestly, it's come up before | 17:10 |
dansmith | davideagnello: answers to those questions are found in -qa :) | 17:10 |
*** atuvenie_ has quit IRC | 17:10 | |
davideagnello | dansmith: thanks, posted there as well | 17:10 |
* dansmith goes there to watch the fireworks | 17:11 | |
davideagnello | mriedem: ok I see | 17:11 |
openstackgerrit | Vilobh Meshram proposed openstack/nova: WIP: Replace service groups with the tooz groups https://review.openstack.org/172502 | 17:11 |
*** ijw has quit IRC | 17:12 | |
*** ijw has joined #openstack-nova | 17:12 | |
*** Sukhdev has quit IRC | 17:12 | |
*** ijw has quit IRC | 17:17 | |
*** fawadkhaliq has joined #openstack-nova | 17:21 | |
*** patrickeast has joined #openstack-nova | 17:22 | |
*** asselin has quit IRC | 17:24 | |
*** fawadkhaliq has quit IRC | 17:26 | |
*** tonytan_brb has joined #openstack-nova | 17:26 | |
*** achanda has joined #openstack-nova | 17:26 | |
*** tonytan4ever has quit IRC | 17:27 | |
*** sdake has joined #openstack-nova | 17:28 | |
*** lucasagomes is now known as lucas-dinner | 17:29 | |
*** ajayaa has joined #openstack-nova | 17:29 | |
*** asselin has joined #openstack-nova | 17:29 | |
*** lpetrut1 has quit IRC | 17:30 | |
*** ijw has joined #openstack-nova | 17:31 | |
*** kmartin has joined #openstack-nova | 17:31 | |
*** VW_ has quit IRC | 17:32 | |
*** annegentle has joined #openstack-nova | 17:32 | |
openstackgerrit | Brent Eagles proposed openstack/nova: WIP: Add support for VIF plugin script to Nova https://review.openstack.org/162470 | 17:33 |
*** VW_ has joined #openstack-nova | 17:34 | |
*** nkrinner has quit IRC | 17:34 | |
*** VW_ has quit IRC | 17:34 | |
*** VW_ has joined #openstack-nova | 17:35 | |
*** rdo has quit IRC | 17:36 | |
*** asselin has quit IRC | 17:36 | |
*** penick has joined #openstack-nova | 17:36 | |
*** aix has quit IRC | 17:37 | |
*** asselin has joined #openstack-nova | 17:37 | |
*** dguitarbite has quit IRC | 17:38 | |
*** annegentle has quit IRC | 17:38 | |
*** rdo has joined #openstack-nova | 17:38 | |
*** annegentle has joined #openstack-nova | 17:39 | |
*** dsanders has joined #openstack-nova | 17:40 | |
*** iamjarvo has quit IRC | 17:44 | |
*** fawadkhaliq has joined #openstack-nova | 17:48 | |
*** mpaolino has quit IRC | 17:48 | |
openstackgerrit | Brent Eagles proposed openstack/nova: WIP: Add support for VIF plugin script to Nova https://review.openstack.org/162470 | 17:49 |
*** otter768 has joined #openstack-nova | 17:52 | |
*** samueldmq has quit IRC | 17:53 | |
*** annegentle has quit IRC | 17:53 | |
*** otter768 has quit IRC | 17:57 | |
*** vilobhmm11 has quit IRC | 17:59 | |
*** joefides has quit IRC | 17:59 | |
*** tonytan_brb is now known as tonytan4ever | 17:59 | |
*** vilobhmm1 has joined #openstack-nova | 17:59 | |
*** vilobhmm1 has quit IRC | 17:59 | |
*** tjones1 has joined #openstack-nova | 17:59 | |
*** joefides has joined #openstack-nova | 17:59 | |
*** vilobhmm1 has joined #openstack-nova | 18:00 | |
*** melwitt has joined #openstack-nova | 18:00 | |
*** vilobhmm1 has quit IRC | 18:00 | |
*** vilobhmm1 has joined #openstack-nova | 18:00 | |
*** vilobhmm11 has joined #openstack-nova | 18:02 | |
*** vilobhmm12 has joined #openstack-nova | 18:02 | |
*** dsanders has quit IRC | 18:03 | |
sahid | an interesting fix ready to be merged here https://review.openstack.org/#/c/173187/ | 18:03 |
sahid | mriedem: ^ perhaps you can check it | 18:03 |
cfriesen | bauzas: dansmith: sorry, didn't see you were discussing my proposed changes. In both cases we could move the logic to nova-compute. Ultimately it would mean calling objects.InstanceList.get_by_filters() and getting all the instances in the group (serializing them and passing them over the wire back to compute manager). The advantage of the new remotable methods is that the RPC response is just one or more hostnames instead | 18:04 |
cfriesen | of the full object for all instances in the group. | 18:04 |
dansmith | cfriesen: I'm not sure we were :) | 18:04 |
mriedem | sahid: has ndipanov seen that change? | 18:04 |
*** vilobhmm1 has quit IRC | 18:04 | |
cfriesen | dansmith: https://review.openstack.org/#/c/164762/9 and https://review.openstack.org/#/c/169489/6/nova/objects/instance_group.py,cm | 18:04 |
dansmith | cfriesen: right, I said that | 18:05 |
sahid | ah i don't think, perhaps it was to ask him | 18:05 |
sahid | i will do tomorrow morning | 18:05 |
sahid | it was better | 18:05 |
*** joefides has quit IRC | 18:05 | |
*** annegentle has joined #openstack-nova | 18:05 | |
vilobhmm12 | sahid : ping | 18:05 |
mriedem | sahid: i can take a look, i just know ndipanov is more familiar with the cpu pinning/numa stuff than me | 18:05 |
sahid | mriedem: yes sure it's right | 18:06 |
sahid | vilobhmm12: yes ? | 18:06 |
*** dsanders has joined #openstack-nova | 18:06 | |
*** vilobhmm11 has quit IRC | 18:07 | |
vilobhmm12 | are you still working on this https://review.openstack.org/#/c/168876/ ? | 18:07 |
vilobhmm12 | since its related to something I have been working as part of moving servicegroup to tooz so thought will check with you | 18:07 |
vilobhmm12 | sahid : ^6 | 18:08 |
vilobhmm12 | sahid : ^^ | 18:08 |
sahid | vilobhmm12: well i got complans for Hans so i expected to get other reviews | 18:08 |
sahid | complains | 18:08 |
*** dsanders has quit IRC | 18:08 | |
*** dsanders has joined #openstack-nova | 18:09 | |
dansmith | sahid: "complaints" .. with a T :) | 18:09 |
vilobhmm12 | ok… | 18:09 |
*** sdake_ has joined #openstack-nova | 18:09 | |
sahid | vilobhmm12: how it is related to your work? basically it just remove the method get_all | 18:09 |
*** eglynn has joined #openstack-nova | 18:09 | |
sahid | dansmith: yes :/ needs to do more efforts with english | 18:10 |
sahid | i need ! | 18:10 |
dansmith | sahid: me too :) | 18:10 |
cfriesen | dansmith: bauzas: we will only ever do one of the checks...for anti-affinity we need the hosts of all earlier instances, and for affinity we need the host of the earliest instance. | 18:10 |
*** annegentle has quit IRC | 18:10 | |
cfriesen | dansmith: bauzas: I suppose one option would be a single remotable function to return the hosts of earlier instances, along with the instance ID of the lowest-id group member running on that host. Then the affinity case could just pick the host with the lowest id number. Seems a little hacky though. | 18:12 |
*** sdake has quit IRC | 18:13 | |
sahid | vilobhmm12: i need to go now, don't hesitate to let comments on the review if you feel something is wrong or whatever, i will chck tomorrow morning. | 18:14 |
*** sahid has quit IRC | 18:14 | |
*** Nic has joined #openstack-nova | 18:14 | |
vilobhmm12 | sahid : sure you can also have a look at https://review.openstack.org/#/c/138607 and https://review.openstack.org/#/c/172502/ let me know your feedback | 18:15 |
vilobhmm12 | good night :) | 18:15 |
*** achanda has quit IRC | 18:15 | |
*** VW_ has quit IRC | 18:15 | |
*** johnthetubaguy is now known as zz_johnthetubagu | 18:16 | |
lennyb | exit | 18:17 |
*** ildikov has joined #openstack-nova | 18:17 | |
*** lennyb has quit IRC | 18:17 | |
*** garyk has joined #openstack-nova | 18:17 | |
*** fawadkhaliq has quit IRC | 18:18 | |
*** isd1 has joined #openstack-nova | 18:18 | |
*** fawadkhaliq has joined #openstack-nova | 18:18 | |
*** achanda has joined #openstack-nova | 18:20 | |
*** VW_ has joined #openstack-nova | 18:21 | |
*** doude has joined #openstack-nova | 18:22 | |
*** mriedem has quit IRC | 18:29 | |
*** vilobhmm1 has joined #openstack-nova | 18:33 | |
*** vilobhmm12 has quit IRC | 18:33 | |
*** unicell has joined #openstack-nova | 18:34 | |
*** mriedem has joined #openstack-nova | 18:35 | |
alaski | bauzas: http://logs.openstack.org/10/175010/5/check/check-tempest-dsvm-cells/78f39a5/console.html#_2015-04-21_15_20_11_613 | 18:36 |
*** openstackgerrit has quit IRC | 18:37 | |
alaski | mriedem: is there a way to track job success/failures over time? | 18:37 |
mriedem | alaski: yeah.... | 18:37 |
mriedem | http://graphite.openstack.org/ | 18:38 |
*** openstackgerrit has joined #openstack-nova | 18:38 | |
alaski | cool. though I hate the graphite interface | 18:38 |
mriedem | me too | 18:38 |
mriedem | http://graphite.openstack.org/render/?width=586&height=308&_salt=1429641586.192&target=stats.zuul.pipeline.check.job.check-devstack-dsvm-cells.SUCCESS&target=stats.zuul.pipeline.check.job.check-devstack-dsvm-cells.None&target=stats.zuul.pipeline.check.job.check-devstack-dsvm-cells.FAILURE | 18:39 |
*** david-lyle has joined #openstack-nova | 18:39 | |
mriedem | ? | 18:39 |
alaski | mriedem: that works, I think. Thanks | 18:41 |
mriedem | http://graphite.openstack.org/render/?width=586&height=308&_salt=1429641672.456&target=stats.zuul.pipeline.check.job.check-devstack-dsvm-cells.SUCCESS&target=stats.zuul.pipeline.check.job.check-devstack-dsvm-cells.None&target=stats.zuul.pipeline.check.job.check-devstack-dsvm-cells.FAILURE&from=-7days | 18:41 |
mriedem | that's the last 7 days | 18:41 |
mriedem | looks like it's definitely trending in the right way | 18:41 |
*** davideagnello has quit IRC | 18:41 | |
*** sudipto has quit IRC | 18:41 | |
mriedem | i see there is a havana cells job in infra somewhere, we can probably delete that now :) | 18:42 |
alaski | hah, yeah | 18:42 |
mtreinish | mriedem: bah, how else will you know havana cells works :) | 18:42 |
alaski | I'm suprised there are so many successes before today though | 18:42 |
mriedem | we've been treinished | 18:42 |
mriedem | yeah i was going to see what logstash shows | 18:43 |
mtreinish | mriedem: my name is a verb? | 18:43 |
mriedem | it is now | 18:43 |
mriedem | alaski: http://goo.gl/1ltSFw | 18:44 |
mriedem | that shows the job only passing in the last 24 hours | 18:44 |
mriedem | less even | 18:44 |
*** davideagnello has joined #openstack-nova | 18:44 | |
alaski | should be around 9:31 EDT this morning | 18:44 |
mriedem | shit, let's make it voting! | 18:45 |
alaski | mriedem: oh, your graph is devstack not tempest | 18:45 |
mriedem | ah, exercises | 18:45 |
alaski | mriedem: heck yeah! | 18:45 |
alaski | not really, but I'm going to send an email asking people to pay attention | 18:45 |
mriedem | http://graphite.openstack.org/render/?width=586&height=308&_salt=1429641955.059&from=-7days&target=stats.zuul.pipeline.check.job.check-tempest-dsvm-cells.SUCCESS&target=stats.zuul.pipeline.check.job.check-tempest-dsvm-cells.FAILURE | 18:46 |
mriedem | very little blue in that | 18:46 |
mriedem | so yeah, that's right | 18:46 |
*** david-ly_ has joined #openstack-nova | 18:46 | |
*** david-lyle has quit IRC | 18:46 | |
mriedem | now i love the graphite ui | 18:46 |
alaski | stupid question, how do you get the link? | 18:46 |
dansmith | right click on the image | 18:46 |
alaski | I have a graph, but no idea how to link to it | 18:46 |
dansmith | copy image url | 18:46 |
mriedem | just right click, copy url | 18:46 |
alaski | oh | 18:47 |
dansmith | it's plotting server-side not client-side | 18:47 |
clarkb | I approved a couple of the test skips for the cells job yesterday | 18:47 |
* dansmith runs graphite at home per clarkb | 18:47 | |
alaski | dansmith: good to know, I wouldn't have expected that | 18:47 |
*** ZZelle has quit IRC | 18:48 | |
*** ZZelle has joined #openstack-nova | 18:48 | |
mriedem | alaski: so why wouldn't we want the cells job to be voting now since it's a baseline? | 18:58 |
*** wuhg has quit IRC | 18:59 | |
alaski | I thought we typically let jobs soak a bit. But mainly I can't guarantee there's not another test that fails periodically that we haven't seen yet | 18:59 |
mriedem | maybe we circle back next week | 19:00 |
mriedem | anything major would shake out in a week | 19:00 |
alaski | yeah, I would feel good after that | 19:00 |
alaski | even later this week really, just want a few good days of data | 19:02 |
mriedem | works for me | 19:03 |
*** melwitt has quit IRC | 19:03 | |
*** melwitt has joined #openstack-nova | 19:04 | |
*** eglynn has quit IRC | 19:07 | |
*** joefides has joined #openstack-nova | 19:07 | |
*** sdake has joined #openstack-nova | 19:07 | |
*** ajayaa has quit IRC | 19:07 | |
*** iamjarvo has joined #openstack-nova | 19:10 | |
*** iamjarvo has quit IRC | 19:10 | |
*** iamjarvo has joined #openstack-nova | 19:11 | |
*** sdake_ has quit IRC | 19:11 | |
mgagne | guys, I would like someone to review the regression introduced by bug #1423885 | 19:12 |
openstack | bug 1423885 in python-novaclient "nova flavor-show inconsistent with mixed case in names" [Low,Fix released] https://launchpad.net/bugs/1423885 - Assigned to Abhishek Talwar (abhishek-talwar) | 19:12 |
*** jichen has joined #openstack-nova | 19:13 | |
melwitt | mgagne: looking.. | 19:15 |
*** changbl has quit IRC | 19:16 | |
openstackgerrit | garyk proposed openstack/nova: VMware: expand support for Opaque networks https://review.openstack.org/165750 | 19:18 |
*** READ10 has quit IRC | 19:18 | |
*** david-ly_ has quit IRC | 19:18 | |
*** changbl has joined #openstack-nova | 19:19 | |
*** quaa has quit IRC | 19:19 | |
bauzas | alaski: \o/ | 19:30 |
bauzas | alaski: saw your email, thanks for noticing me :) | 19:30 |
bauzas | dansmith: cfriesen: sorry, was afk (2100CEST now), but we can discuss yup | 19:31 |
bauzas | cfriesen: I'm trying to take time thinking about your problem and how to solve it | 19:31 |
*** VW_ has quit IRC | 19:31 | |
dansmith | bauzas: can you look at mriedem's comments on this: test -d /home/ro.pi | 19:31 |
dansmith | oops | 19:31 |
dansmith | https://review.openstack.org/#/c/173187/17 | 19:31 |
bauzas | eh, +1 with mriedem | 19:31 |
bauzas | lol | 19:31 |
dansmith | that ^ | 19:32 |
*** iamjarvo has quit IRC | 19:32 | |
alaski | bauzas: heh | 19:32 |
*** VW_ has joined #openstack-nova | 19:33 | |
*** jcoufal has joined #openstack-nova | 19:33 | |
*** Hosam has joined #openstack-nova | 19:34 | |
cfriesen | bauzas: no worries. the main problems are that instances can be scheduled/booted in parallel, there are multiple things that can cause an instance to be rescheduled, and the chosen host isn't reflected in the DB right away | 19:35 |
*** alexpilotti has quit IRC | 19:36 | |
artom | Going back to the database delete thing from earlier, what would be the implications of adding a feature that gives operators an option to *actually* delete stuff when it's deleted, not just soft-delete? | 19:39 |
artom | In that case, the DIY would be "have proper backups" instead of "clean the DB" | 19:40 |
*** achanda has quit IRC | 19:41 | |
*** Marga_ has quit IRC | 19:42 | |
melwitt | artom: well, the way it's supposed to work is operators can run a nova-manage db archive which will delete things with deleted != 0 and move them to the shadow tables, but last I remember the archive tool is broken | 19:44 |
artom | melwitt, I just tried, it is. | 19:44 |
artom | And there are other criticisms levied against it, like complexity and slowness. | 19:45 |
mriedem | i thought zzzeek and/or sdague had a fix for the archive deleted rows thing? | 19:45 |
melwitt | mriedem: I think the data loss bug got fixed, but archive still can't work completely because sysmeta is never deleted (soft-deleted) or something like that | 19:46 |
mriedem | oh that super old bug | 19:46 |
melwitt | yeah :( | 19:46 |
artom | In my case it was simply FKs errors (hah) | 19:46 |
melwitt | artom: right. that's because of the sysmeta, that has to also be deleted for the instance row delete to go through | 19:47 |
*** lpetrut has joined #openstack-nova | 19:47 | |
alaski | https://review.openstack.org/#/c/109201/ for the system_metadata soft-delete | 19:47 |
melwitt | there are some patches up about it but they're not ironed out all the way yet | 19:47 |
artom | Ok, so fixing archive_deleted_rows is definitely an option :) | 19:48 |
artom | Still, hypothetically speaking (as a backup plan or whatever), could giving operators a config option to do real deletes instead of soft deletes be considered? | 19:50 |
*** fawadkhaliq has quit IRC | 19:50 | |
mriedem | artom: you can always post a spec | 19:51 |
artom | mriedem, ok, so no pitchforks hostility ;) | 19:51 |
artom | (Though that may come later :D ) | 19:52 |
mriedem | i currently don't have any pitchforks | 19:52 |
artom | He said with his hands suspiciously hidden. | 19:52 |
mriedem | plus, i've already bled out from this morning's shanking | 19:52 |
melwitt | artom: that, I don't know. I'd imagine it's been debated on the dev mailing list before but I haven't seen any recent discussion. and as mentioned you can propose a spec about it. you can post on the ML to get preliminary feedback before you go all in to the spec | 19:53 |
artom | melwitt, awesome. I'll search and post, thanks! | 19:53 |
*** otter768 has joined #openstack-nova | 19:53 | |
artom | mriedem, I kinda want to capitalize that and make it into a regular event now. The Shanking. | 19:54 |
mriedem | there is probably already a b rate movie somewhere with that title, i don't want to get into trademark disputes | 19:56 |
*** otter768 has quit IRC | 19:58 | |
artom | Worse than a B movie, a YouTube B-movie: https://www.youtube.com/watch?v=WIdDKNoUBxQ | 20:00 |
artom | (And it's actually hilarious) | 20:01 |
*** tjones1 has quit IRC | 20:04 | |
*** MVenesio has quit IRC | 20:04 | |
*** MVenesio has joined #openstack-nova | 20:04 | |
*** baoli has quit IRC | 20:05 | |
*** Mike_D_laptop has quit IRC | 20:06 | |
mriedem | dansmith: https://review.openstack.org/#/c/173187/ isn't on boot request, it's on compute init_host | 20:06 |
mriedem | so when that blows up, your compute service is dead | 20:07 |
mriedem | which is why i wanted more debug info in the logs | 20:07 |
dansmith | ohh | 20:07 |
mriedem | the admin is setting the config options so it shouldn't be bogus | 20:07 |
*** annashen has quit IRC | 20:08 | |
dansmith | yeah, that's totally different | 20:08 |
dansmith | I thought we were actually verifying a cpu pinning request against what was available | 20:09 |
*** andreykurilin__ has quit IRC | 20:09 | |
mriedem | naw, this would be pretty low level for that :) | 20:11 |
mriedem | i don't know shit about your cloud, but i know i want pcpu #5! | 20:12 |
mriedem | however, i wouldn't be surprised if there is a blueprint in someone's head for supporting something that low level | 20:12 |
*** andreykurilin__ has joined #openstack-nova | 20:12 | |
dansmith | that's the goal of our cpu pinning work, no? | 20:14 |
*** achanda has joined #openstack-nova | 20:14 | |
mriedem | is it? | 20:14 |
mriedem | idk | 20:14 |
mriedem | sounds crazy | 20:14 |
dansmith | agreed, but... | 20:15 |
mriedem | soo....can we allow booting to a specific vmware datastore? | 20:15 |
mriedem | is that a thing? | 20:15 |
mriedem | i'm sure there are all kinds of vmware specific features people want for things like that | 20:15 |
*** jcoufal has quit IRC | 20:15 | |
*** jwang_ has quit IRC | 20:16 | |
*** jwang_ has joined #openstack-nova | 20:16 | |
dansmith | http://specs.openstack.org/openstack/nova-specs/specs/juno/approved/virt-driver-cpu-pinning.html | 20:16 |
dansmith | I guess you only get to set policy | 20:16 |
*** ildikov has quit IRC | 20:17 | |
dansmith | when I hear "pinning" I think of the hypervisor's notion of it, which is very low-level | 20:17 |
dansmith | this is like "cpu dedication" or something :) | 20:17 |
openstackgerrit | Tim Pownall proposed openstack/nova: Cleaning up debug messages from previous change in vmops.py https://review.openstack.org/175248 | 20:20 |
*** matrohon has joined #openstack-nova | 20:20 | |
*** dboik has quit IRC | 20:20 | |
*** david-lyle has joined #openstack-nova | 20:21 | |
*** dboik has joined #openstack-nova | 20:21 | |
*** VW_ has quit IRC | 20:24 | |
*** igordcard has joined #openstack-nova | 20:29 | |
*** ildikov has joined #openstack-nova | 20:30 | |
*** vladikr has quit IRC | 20:31 | |
*** Sukhdev has joined #openstack-nova | 20:31 | |
*** iamjarvo has joined #openstack-nova | 20:33 | |
*** matrohon has quit IRC | 20:34 | |
*** tjones1 has joined #openstack-nova | 20:39 | |
*** baoli has joined #openstack-nova | 20:42 | |
*** raildo has quit IRC | 20:43 | |
*** neelashah has quit IRC | 20:44 | |
*** emagana has quit IRC | 20:46 | |
bauzas | cfriesen: sorry was diverted | 20:47 |
*** Marga_ has joined #openstack-nova | 20:47 | |
*** alexpilotti has joined #openstack-nova | 20:51 | |
*** openstackgerrit has quit IRC | 20:52 | |
*** foexle has quit IRC | 20:52 | |
*** openstackgerrit has joined #openstack-nova | 20:52 | |
*** changbl has quit IRC | 20:57 | |
*** VW_ has joined #openstack-nova | 21:00 | |
*** annashen has joined #openstack-nova | 21:01 | |
*** VW_ has quit IRC | 21:01 | |
cfriesen | dansmith: mriedem: while we're talking about low level hypervisor stuff...(okay it's a stretch)...any thoughts on https://blueprints.launchpad.net/nova/+spec/cpu-model-per-flavor-or-image or https://blueprints.launchpad.net/nova/+spec/numa-node-pinning ? | 21:01 |
*** lucas-dinner has quit IRC | 21:01 | |
*** VW_ has joined #openstack-nova | 21:02 | |
mriedem | cfriesen: no opinion? | 21:03 |
*** iamjarvo has quit IRC | 21:03 | |
*** zz_johnthetubagu is now known as johnthetubaguy | 21:04 | |
cfriesen | mriedem: fair enough. so far nobody seems to have any opinions. | 21:05 |
*** annashen has quit IRC | 21:06 | |
mriedem | cfriesen: i guess you'd need a spec to get those | 21:06 |
cfriesen | mriedem: I proposed specs for each | 21:06 |
mriedem | oh | 21:06 |
openstackgerrit | Andrew Laski proposed openstack/nova-specs: WIP Store RequestSpec before scheduling https://review.openstack.org/169901 | 21:06 |
openstackgerrit | Andrew Laski proposed openstack/nova-specs: Persist the RequestSpec object https://review.openstack.org/176078 | 21:06 |
vilobhmm1 | jaypipes : can you please have a look at https://review.openstack.org/#/c/172502/6 | 21:06 |
mriedem | cfriesen: well, i'd talk to danpb, ndipanov, bauzas and/or sahid on those | 21:06 |
*** jecarey has quit IRC | 21:07 | |
cfriesen | mriedem: okay, thanks. | 21:07 |
bauzas | wazzup ? | 21:07 |
*** haomaiwang has quit IRC | 21:07 | |
*** exploreshaifali has joined #openstack-nova | 21:07 | |
*** haomaiwang has joined #openstack-nova | 21:07 | |
mriedem | https://www.youtube.com/watch?v=mO1NyXlnL84 | 21:07 |
bauzas | zactly | 21:07 |
cfriesen | bauzas: I'm looking for specs reviews, mriedem suggested I should talk to a list of people including you | 21:07 |
bauzas | cfriesen: oh the numa stuff | 21:08 |
cfriesen | numa pinning and cpu model specification | 21:08 |
*** jichen has quit IRC | 21:08 | |
*** vilobhmm1 has quit IRC | 21:08 | |
* bauzas reading back | 21:09 | |
*** vilobhmm11 has joined #openstack-nova | 21:09 | |
cfriesen | bauzas: the cpu model one seems like something that would make sense...finer grained control instead of per compute node. the numa one I could see people wanting it to be more general (tying it more indirectly to features/devices rather than to a specific node) but I thought I'd put it out there to see what people thought. | 21:10 |
cfriesen | make that per compute service | 21:10 |
*** vilobhmm12 has joined #openstack-nova | 21:11 | |
*** vilobhmm12 has quit IRC | 21:11 | |
*** vilobhmm13 has joined #openstack-nova | 21:11 | |
bauzas | CCing both | 21:11 |
*** vilobhmm11 has quit IRC | 21:12 | |
clarkb | hello nova. Now that kilo is a thing it should be safe to merge https://review.openstack.org/#/c/168701/2 to master (with the string changes) then figure out how to backport that to kilo without string chagnes to fix the bug right? | 21:13 |
clarkb | getting that change in should allow us to run neutron with ipv4 and ipv6 enabled in the gate so that we can actually test that ipv6 works | 21:13 |
*** marun has quit IRC | 21:14 | |
openstackgerrit | Andrew Laski proposed openstack/nova-specs: WIP Don't store instance in RequestSpec object https://review.openstack.org/176083 | 21:15 |
sc68cal | clarkb: s/actually test that ipv6 works/test IPv6 more fully/ | 21:16 |
clarkb | sc68cal: we do zero testing of it today iirc | 21:16 |
clarkb | sc68cal: at least for things like nova floating ip attachment which that fix is for | 21:16 |
sc68cal | clarkb: no, tempest has tests for intra-tenant connectivity checks | 21:16 |
sc68cal | it just has to use v4 to ssh into the node currently ;) | 21:17 |
clarkb | sc68cal: but none of that is ipv6 in the gate | 21:17 |
clarkb | because cirros doesn't support ipv6 in dropbear iirc | 21:17 |
clarkb | nor does it have ping6 | 21:17 |
clarkb | so next step after this is to get a new cirros image that lets us do ^ | 21:17 |
sc68cal | clarkb: true, and I have a TODO to take a look at how cirros is built, to tweak the makefiles to make that happen | 21:18 |
clarkb | that and have tempest properly assert some stuff about ipv6 | 21:18 |
*** dboik_ has joined #openstack-nova | 21:18 | |
sc68cal | for busybox | 21:18 |
clarkb | ya busybox can do it, I just don't think cirros is building it with the correct flag | 21:18 |
sc68cal | downside is cirros uses BZR. YUCK! | 21:18 |
clarkb | hopefully a simple change | 21:18 |
sc68cal | clarkb++ | 21:18 |
*** lpetrut has quit IRC | 21:19 | |
sc68cal | jaypipes: if you get the chance - check scrollback and possibly take a look at https://review.openstack.org/#/c/168701/ ? | 21:19 |
*** iamjarvo has joined #openstack-nova | 21:20 | |
*** eharney has quit IRC | 21:20 | |
*** jgrimm is now known as zz_jgrimm | 21:20 | |
clarkb | the nice thing about getting the bit flipped here is we can make any subsequent changes self testing | 21:21 |
*** lpetrut has joined #openstack-nova | 21:21 | |
sc68cal | clarkb: oh? | 21:21 |
*** dboik has quit IRC | 21:21 | |
*** thangp has quit IRC | 21:21 | |
clarkb | sc68cal: ya if we get the devstack change in to enable ipv6 by default then the addition of new cirros image, new tempest tests, etc is all self testing | 21:21 |
sc68cal | clarkb: ah good point | 21:22 |
*** lpetrut has quit IRC | 21:22 | |
*** dprince has quit IRC | 21:23 | |
jaypipes | vilobhmm13: is that ready to remove the WIP: in the commit message then? :) | 21:25 |
jaypipes | sc68cal: you got it. | 21:25 |
sc68cal | jaypipes: thanks :) | 21:26 |
*** tonytan4ever has quit IRC | 21:27 | |
vilobhmm13 | jaypipes : jenkins gave a +1; only if you bless https://review.openstack.org/#/c/172502/6 I can remove WIP :) | 21:30 |
*** sdake_ has joined #openstack-nova | 21:31 | |
*** emagana has joined #openstack-nova | 21:32 | |
*** doude has quit IRC | 21:33 | |
*** sdake has quit IRC | 21:34 | |
*** haomaiwang has quit IRC | 21:35 | |
*** haomaiwang has joined #openstack-nova | 21:36 | |
mriedem | vilobhmm13: -2 so you can stop asking for jaypipes to give a +1 :) | 21:37 |
mriedem | vilobhmm13: wip changes for a blueprint are -2 by process until the spec is approved | 21:38 |
*** iamjarvo has quit IRC | 21:38 | |
vilobhmm13 | mriedem : not asking ; that was in a funny way.…i get that the blueprint for this is not yet accepted and hence so i have WIP in the heading i guess | 21:40 |
mriedem | vilobhmm13: sure | 21:40 |
mriedem | that's fine, i have one like that myself | 21:40 |
mriedem | https://review.openstack.org/#/c/168966/ | 21:40 |
*** zz_jgrimm is now known as jgrimm | 21:40 | |
*** haomaiwang has quit IRC | 21:40 | |
*** ijw has quit IRC | 21:41 | |
vilobhmm13 | hmm | 21:41 |
*** thomasem has joined #openstack-nova | 21:42 | |
*** haomaiwang has joined #openstack-nova | 21:42 | |
openstackgerrit | Matt Riedemann proposed openstack/nova: Allow refreshing network info_cache in get_nw_info_for_instance https://review.openstack.org/171390 | 21:43 |
openstackgerrit | Matt Riedemann proposed openstack/nova: Force a refresh of the network info_cache if not set in os-floating-ips https://review.openstack.org/171391 | 21:43 |
*** jecarey has joined #openstack-nova | 21:44 | |
*** mriedem is now known as mriedem_away | 21:44 | |
*** VW_ has quit IRC | 21:45 | |
*** dboik_ has quit IRC | 21:46 | |
*** artom has quit IRC | 21:46 | |
*** Hosam has quit IRC | 21:46 | |
*** dboik has joined #openstack-nova | 21:47 | |
*** mwagner_lap has quit IRC | 21:47 | |
*** patrickeast has quit IRC | 21:51 | |
mriedem_away | bauzas: dansmith: i have to head out but left some more comments in https://review.openstack.org/#/c/173187/ | 21:52 |
mriedem_away | i'm not sure if i'm seeing something completely different or what, | 21:52 |
mriedem_away | but we're obviously not on the same page :) | 21:52 |
*** mtanino has quit IRC | 21:52 | |
*** joefides has quit IRC | 21:53 | |
mriedem_away | anyway, gotta go | 21:53 |
*** busterswt has quit IRC | 21:53 | |
*** emagana has quit IRC | 21:53 | |
*** sdake has joined #openstack-nova | 21:53 | |
*** emagana has joined #openstack-nova | 21:54 | |
*** otter768 has joined #openstack-nova | 21:54 | |
*** mtanino has joined #openstack-nova | 21:55 | |
*** burt has quit IRC | 21:56 | |
openstackgerrit | Matt Riedemann proposed openstack/nova: Force a refresh of the network info_cache if not set in os-floating-ips https://review.openstack.org/171391 | 21:57 |
*** sdake_ has quit IRC | 21:57 | |
*** johnthetubaguy is now known as zz_johnthetubagu | 21:58 | |
*** yamahata has joined #openstack-nova | 21:58 | |
*** emagana has quit IRC | 21:59 | |
*** otter768 has quit IRC | 21:59 | |
*** harlowja is now known as harlowja_away | 22:00 | |
*** gyee has joined #openstack-nova | 22:01 | |
*** vilobhmm13 has quit IRC | 22:03 | |
*** vilobhmm1 has joined #openstack-nova | 22:03 | |
*** baoli has quit IRC | 22:03 | |
*** dave-mccowan has quit IRC | 22:03 | |
*** vilobhmm1 has quit IRC | 22:03 | |
*** VW_ has joined #openstack-nova | 22:03 | |
*** isd1 has quit IRC | 22:04 | |
*** vilobhmm1 has joined #openstack-nova | 22:04 | |
*** harlowja_away is now known as harlowja | 22:05 | |
*** emagana has joined #openstack-nova | 22:10 | |
*** vilobhmm1 has quit IRC | 22:12 | |
*** vilobhmm1 has joined #openstack-nova | 22:12 | |
*** yamahata has quit IRC | 22:13 | |
*** andreykurilin__ has quit IRC | 22:14 | |
*** vilobhmm1 has quit IRC | 22:14 | |
*** vilobhmm1 has joined #openstack-nova | 22:15 | |
*** patrickeast has joined #openstack-nova | 22:18 | |
*** dave-mccowan has joined #openstack-nova | 22:19 | |
*** patrickeast has quit IRC | 22:21 | |
*** iamjarvo has joined #openstack-nova | 22:22 | |
bauzas | mriedem_away: no worries, I need to pause and think about your comments | 22:23 |
*** patrickeast has joined #openstack-nova | 22:24 | |
openstackgerrit | melanie witt proposed openstack/python-novaclient: Revert "nova flavor-show command is inconsistent" https://review.openstack.org/176106 | 22:24 |
*** VW_ has quit IRC | 22:24 | |
*** marun has joined #openstack-nova | 22:26 | |
*** Sukhdev has quit IRC | 22:31 | |
*** mwagner_lap has joined #openstack-nova | 22:32 | |
*** annegentle has joined #openstack-nova | 22:32 | |
openstackgerrit | Tracy Jones proposed openstack/nova: Cleanup of compute api reboot method https://review.openstack.org/127427 | 22:34 |
*** annegentle has quit IRC | 22:38 | |
*** BadCub is now known as BadCub_Away | 22:38 | |
*** joefides has joined #openstack-nova | 22:39 | |
*** jecarey has quit IRC | 22:39 | |
*** joefides_ has joined #openstack-nova | 22:41 | |
*** irenab has quit IRC | 22:41 | |
*** irenab has joined #openstack-nova | 22:42 | |
*** joefides has quit IRC | 22:44 | |
*** haomaiwang has quit IRC | 22:46 | |
*** annegentle has joined #openstack-nova | 22:46 | |
*** haomaiwang has joined #openstack-nova | 22:47 | |
*** haomaiwang has quit IRC | 22:49 | |
*** dboik_ has joined #openstack-nova | 22:50 | |
*** haomaiwang has joined #openstack-nova | 22:50 | |
*** dboik has quit IRC | 22:53 | |
*** Longgeek has quit IRC | 22:54 | |
*** penick has quit IRC | 22:58 | |
*** dboik_ has quit IRC | 22:59 | |
*** tjones1 has quit IRC | 22:59 | |
*** annegentle has quit IRC | 23:00 | |
*** penick has joined #openstack-nova | 23:02 | |
*** tjones1 has joined #openstack-nova | 23:02 | |
*** penick has quit IRC | 23:03 | |
*** annashen has joined #openstack-nova | 23:03 | |
*** annashen has quit IRC | 23:08 | |
*** apmelton is now known as apmelton_away | 23:10 | |
*** hemna is now known as hemnafk | 23:12 | |
vilobhmm1 | dansmith, jaypipes, melwitt : nova memcache servicegroup driver internally making calls to database is that expected https://github.com/openstack/nova/blob/master/nova/servicegroup/drivers/mc.py#L81 ? I mean shoudln't the memcache driver fetch store info from memcache ? | 23:14 |
vilobhmm1 | while discussing with harlowja we saw this | 23:15 |
vilobhmm1 | dims : ^^ | 23:15 |
harlowja | vilobhmm1 thats also sorta an issue for the nova disable command right? | 23:16 |
harlowja | *mainly for that (i think?) | 23:16 |
*** markvoelker_ has quit IRC | 23:16 | |
melwitt | vilobhmm1: what do you mean by store info? the code you link is pulling all services related to the group, then for each checking last heartbeat in memcache to determine up/down | 23:18 |
vilobhmm1 | melwitt : fetching info is the write word please disregard *store* as its just a fetch call | 23:19 |
openstackgerrit | Eric Brown proposed openstack/nova: VMware: Don't raise exception on resize of 0 disk https://review.openstack.org/173170 | 23:19 |
vilobhmm1 | i mean shouldn't the fetch also be from memcache ? like for zk.py it fetches info from the znode tree stored in zookeeper | 23:19 |
melwitt | vilobhmm1: okay. so there it's getting the list of services from the db, but then for each of those fetching from memcache in the is_up() function | 23:20 |
melwitt | that is, I think only the heartbeats are stored in memcache, not the services themselves | 23:21 |
vilobhmm1 | melwitt : seems like that whereas in case of zookeeper driver even the services info is stored in zookeeper; thats not the case in case of memcache driver (ideally it should be so) | 23:29 |
*** Guest53909 has quit IRC | 23:31 | |
*** Guest53909 has joined #openstack-nova | 23:31 | |
vilobhmm1 | melwitt, dansmith, jaypipes, dims : example if you see https://gist.github.com/vilobhmm/1044d2598e3fe2b4d92d here the nova.services table can be manipulated through an admin interface in the form of nova-manage disable/enable etc and programatically even the servicegroup driver(esp mc.py and db.py) uses the same table (nova.services) underlying. Say some customer tries to use zk.py in which case the zookeper itself stores | 23:35 |
*** sdake_ has joined #openstack-nova | 23:35 | |
*** penick has joined #openstack-nova | 23:36 | |
*** sdake has quit IRC | 23:38 | |
*** iamjarvo has quit IRC | 23:40 | |
melwitt | vilobhmm1: yeah, I see your point. if anything I would think zk driver should do similar as the others. I don't have history about why it's different, or whether it does anything to sync state with the nova db, if someone were to update disable/enable | 23:40 |
vilobhmm1 | melwitt : yep thats the point…the info doesn't seem in sync and which is not good :( | 23:42 |
dansmith | vilobhmm1: really, no need to ping everyone on the planet with every comment | 23:42 |
dansmith | vilobhmm1: also, most of nova-manage directly dabbles in the database.. that's kinda the point | 23:42 |
vilobhmm1 | ok | 23:42 |
dansmith | vilobhmm1: if it's doing something in the DB that causes it to not work for a different technology, it probably needs to be refactored to use an abstract interface | 23:43 |
vilobhmm1 | dansmith : so how is the services info sync maintanied in case of zookeeper sg driver | 23:43 |
vilobhmm1 | sure | 23:44 |
vilobhmm1 | at present | 23:44 |
*** exploreshaifali has quit IRC | 23:44 | |
dansmith | vilobhmm1: I don't know, I don't look at that code much. It gets so little attention, that I'm skeptical that much other than the db driver gets really used | 23:44 |
vilobhmm1 | oh okay | 23:45 |
*** sdake_ has quit IRC | 23:46 | |
*** zhenguo has joined #openstack-nova | 23:47 | |
melwitt | vilobhmm1: at a glance, it looks like nothing is done to sync it. if true, I think that driver needs some work on it to behave like the others, pull services from the db and query zk for the up/down status | 23:47 |
vilobhmm1 | melwitt: True…I agree …may be https://review.openstack.org/#/c/138607 should help lets see | 23:51 |
openstackgerrit | melanie witt proposed openstack/python-novaclient: Don't lookup service url when bypass_url is given https://review.openstack.org/175622 | 23:55 |
*** otter768 has joined #openstack-nova | 23:55 | |
*** penick has quit IRC | 23:56 | |
*** rwsu has quit IRC | 23:56 | |
*** melwitt has quit IRC | 23:56 | |
*** dsanders has quit IRC | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!