*** dboik has joined #openstack-nova | 00:00 | |
*** promulo__ has joined #openstack-nova | 00:01 | |
*** tjones1 has quit IRC | 00:01 | |
*** mriedem has quit IRC | 00:01 | |
*** Marga_ has quit IRC | 00:02 | |
*** dboik has quit IRC | 00:02 | |
*** ijw has quit IRC | 00:03 | |
*** claudiub has quit IRC | 00:04 | |
*** promulo_ has quit IRC | 00:05 | |
*** otter768 has joined #openstack-nova | 00:12 | |
openstackgerrit | Chris Friesen proposed openstack/nova: Fix binding:host_id for neutron on evacuate https://review.openstack.org/173615 | 00:14 |
---|---|---|
*** IanGovett has quit IRC | 00:14 | |
*** dave-mccowan has quit IRC | 00:16 | |
*** otter768 has quit IRC | 00:17 | |
mriedem1 | handle_lifecycle_event bites me again | 00:17 |
*** baoli has joined #openstack-nova | 00:18 | |
mriedem1 | libvirt graceful shutdown will stop instances, which triggers the event to nova to stop the instance (call stop api) which changes the vm_state to stopped. on libvirt restart, the instances are started by libvirt again but sync_instance_power_states sees the instance is supposed to be stopped (according to the nova db), so it stops the running instance | 00:18 |
mriedem1 | gah | 00:18 |
mriedem1 | sdague: ^ | 00:18 |
mriedem1 | dansmith: have you heard of a problem like that? | 00:23 |
dansmith | hah | 00:23 |
*** r-daneel has quit IRC | 00:24 | |
dansmith | you can configure a vm in libvirt to either start or not on startup | 00:24 |
dansmith | I would think that we should be telling libvirt to do nothing with it and let us either start it (or not) on init_host() | 00:24 |
*** sbfox has quit IRC | 00:25 | |
*** salv-orlando has joined #openstack-nova | 00:25 | |
mriedem1 | this is in /etc/sysconfig/libvirt-guests right? | 00:25 |
mriedem1 | ON_SHUTDOWN=shutdown | 00:26 |
mriedem1 | with a timeout | 00:26 |
dansmith | no | 00:26 |
mriedem1 | ok, was reading this https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/6/html/Virtualization_Administration_Guide/sub-sect-Shutting_down_rebooting_and_force_shutdown_of_a_guest_virtual_machine-Manipulating_the_libvirt_guests_configuration_settings.html | 00:28 |
mriedem1 | ON_SHUTDOWN doc basically says be careful with this (because you could eff up nova's power state syncing for sure bro!) | 00:28 |
dansmith | there is an onboot= oncrash=, etc settings in libvirt | 00:28 |
*** jwcroppe has joined #openstack-nova | 00:28 | |
mriedem1 | oh | 00:28 |
dansmith | I thought we had a way to set what we wanted that to be.. looking | 00:29 |
*** browne has quit IRC | 00:29 | |
dansmith | mriedem1: https://libvirt.org/formatdomain.html#elementsEvents | 00:29 |
dansmith | I also though there was controls in there to change what we do on libvirtd start | 00:29 |
dansmith | there also used to be something like /etc/libvirt/autostart I think | 00:30 |
*** heyongli_ has joined #openstack-nova | 00:30 | |
dansmith | which may be what I'm thinking of | 00:30 |
dansmith | virsh autostart | 00:31 |
dansmith | yeah, "virsh autostart foo" leaves me with /etc/libvirt/qemu/autostart/foo.xml | 00:32 |
*** heyongli_ has quit IRC | 00:32 | |
dansmith | which is a link to /etc/libvirt/qemu/foo.xml | 00:32 |
*** salv-orlando has quit IRC | 00:32 | |
*** neelashah has joined #openstack-nova | 00:32 | |
dansmith | so unless nova is running the equivalent of that command, I'd not think the instances would be restarting | 00:33 |
dansmith | mriedem1: ^ | 00:33 |
mriedem1 | dansmith: unless the operator is configuring something outside nova for this? | 00:33 |
dansmith | or that, of course | 00:33 |
mriedem1 | i assume that's what's happening here but trying to get specific recreate steps | 00:33 |
*** nelsnelson has quit IRC | 00:33 | |
dansmith | gotta run to dinner | 00:33 |
*** dims_ has joined #openstack-nova | 00:35 | |
openstackgerrit | Chris Friesen proposed openstack/nova: Fix binding:host_id for neutron on evacuate https://review.openstack.org/173615 | 00:35 |
*** dims has quit IRC | 00:36 | |
*** nelsnelson has joined #openstack-nova | 00:36 | |
*** nelsnelson has quit IRC | 00:39 | |
*** emagana has quit IRC | 00:40 | |
*** emagana has joined #openstack-nova | 00:41 | |
*** IanGovett has joined #openstack-nova | 00:42 | |
*** jogo has quit IRC | 00:42 | |
*** neelashah has quit IRC | 00:45 | |
*** Marga_ has joined #openstack-nova | 00:47 | |
*** emagana has quit IRC | 00:48 | |
*** IanGovett has quit IRC | 00:49 | |
*** annashen has joined #openstack-nova | 00:49 | |
mriedem1 | guh, apparently handle_lifecycle_event was added in grizzly and resume_guests_state_on_host_boot was added back in like essex? | 00:50 |
*** kmartin has quit IRC | 00:51 | |
mriedem1 | hell diablo even https://github.com/openstack/nova/commit/ba4946d0d3c73e5d9f67f42203d103bf98563458 | 00:51 |
*** Marga_ has quit IRC | 00:52 | |
*** pixelb has joined #openstack-nova | 00:53 | |
*** Marga_ has joined #openstack-nova | 00:53 | |
*** annashen has quit IRC | 00:55 | |
*** sbfox has joined #openstack-nova | 00:55 | |
*** promulo_ has joined #openstack-nova | 00:59 | |
*** slong has joined #openstack-nova | 01:00 | |
*** yuntongjin has joined #openstack-nova | 01:01 | |
*** shaohe_feng has joined #openstack-nova | 01:02 | |
*** heyongli_ has joined #openstack-nova | 01:03 | |
*** shaohe_feng1 has joined #openstack-nova | 01:03 | |
*** promulo__ has quit IRC | 01:03 | |
*** Marga_ has quit IRC | 01:03 | |
*** slong_ has joined #openstack-nova | 01:03 | |
heyongli_ | baoli: ping | 01:05 |
*** slong has quit IRC | 01:05 | |
*** tjones1 has joined #openstack-nova | 01:06 | |
*** shaohe_feng has quit IRC | 01:07 | |
*** shaohe_feng1 has quit IRC | 01:07 | |
*** shaohe_feng has joined #openstack-nova | 01:08 | |
*** eliqiao has joined #openstack-nova | 01:09 | |
*** yuntongjin has quit IRC | 01:09 | |
*** achanda has quit IRC | 01:10 | |
*** jogo has joined #openstack-nova | 01:10 | |
*** yuntongjin has joined #openstack-nova | 01:10 | |
*** sbfox has quit IRC | 01:12 | |
*** marun has quit IRC | 01:13 | |
*** pixelb has quit IRC | 01:13 | |
*** sdake_ has joined #openstack-nova | 01:17 | |
*** sdake has quit IRC | 01:21 | |
*** dboik_ has joined #openstack-nova | 01:21 | |
baoli | heyongli_: Hi | 01:26 |
heyongli_ | hi, baoli | 01:27 |
openstackgerrit | zhangjl proposed openstack/nova: Check if destination_type in block_device_mapping_v2 https://review.openstack.org/173357 | 01:28 |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge V2 and V2.1 cloudpipe functional tests https://review.openstack.org/167481 | 01:28 |
alex_xu | cfriesen: mriedem1 just FYI, for the evaucate case, there already patch from sdague https://review.openstack.org/#/c/169827/ | 01:29 |
*** slong_ has quit IRC | 01:31 | |
*** MVenesio has quit IRC | 01:31 | |
mriedem1 | alex_xu: i like cfriesen's patch a lot more, it's much cleaner if it works | 01:31 |
*** jwcroppe has quit IRC | 01:32 | |
alex_xu | mriedem1: This one https://review.openstack.org/#/c/173615/ ? will take look at it | 01:33 |
*** dsanders has quit IRC | 01:34 | |
*** Sukhdev has quit IRC | 01:34 | |
*** setuid has quit IRC | 01:36 | |
*** popw has joined #openstack-nova | 01:36 | |
alex_xu | mriedem1: just quick look at it, I'm not sure it is right, For nova-network case, there also need some cleanup on the source host(for unshelve, reschedule case). And those cleanup and host only for one instance | 01:37 |
*** vilobhmm1 has quit IRC | 01:37 | |
alex_xu | mriedem1: mriedem1 setup_networks_on_host is for normal generic network setup, not just for one instance | 01:38 |
alex_xu | mriedem1: anyway let me dig into more later | 01:38 |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge remote consoles tests between v2 and v2.1 https://review.openstack.org/173639 | 01:43 |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] Store context in local store after spawn_n https://review.openstack.org/173112 | 01:47 |
*** erkules has quit IRC | 01:48 | |
*** erkules_ has joined #openstack-nova | 01:48 | |
mriedem1 | alex_xu: ok, i didn't know it was broken for nova-network too | 01:50 |
mriedem1 | i duplicated cfriesen's bug against the original one though | 01:51 |
*** sdake has joined #openstack-nova | 01:52 | |
*** yamahata has quit IRC | 01:53 | |
*** baoli has quit IRC | 01:53 | |
alex_xu | mriedem1: ok | 01:54 |
*** baoli has joined #openstack-nova | 01:54 | |
*** baoli has quit IRC | 01:54 | |
*** VW_ has joined #openstack-nova | 01:54 | |
*** park_heijlong has joined #openstack-nova | 01:55 | |
*** sdake_ has quit IRC | 01:56 | |
*** kaisers has joined #openstack-nova | 01:58 | |
*** kaisers1 has quit IRC | 01:59 | |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge server rescue tests between v2 and v2.1 https://review.openstack.org/173651 | 02:09 |
*** achanda has joined #openstack-nova | 02:10 | |
openstackgerrit | Lan Qi Song proposed openstack/nova: Vmware:Fix how VolumeOps is getting host information https://review.openstack.org/135182 | 02:13 |
*** otter768 has joined #openstack-nova | 02:13 | |
*** changbl has joined #openstack-nova | 02:14 | |
*** setuid has joined #openstack-nova | 02:15 | |
*** unicell has quit IRC | 02:16 | |
*** achanda has quit IRC | 02:17 | |
*** harlowja is now known as harlowja_away | 02:17 | |
*** otter768 has quit IRC | 02:18 | |
*** VW_ has quit IRC | 02:18 | |
*** dims_ has quit IRC | 02:21 | |
*** tjones1 has quit IRC | 02:22 | |
*** tjones1 has joined #openstack-nova | 02:22 | |
*** VW_ has joined #openstack-nova | 02:23 | |
*** vilobhmm1 has joined #openstack-nova | 02:28 | |
*** yamahata has joined #openstack-nova | 02:28 | |
*** jecarey has joined #openstack-nova | 02:28 | |
*** salv-orlando has joined #openstack-nova | 02:28 | |
openstackgerrit | ShaoHe Feng proposed openstack/nova: Move unlock_override policy enforcement into V2.1 REST API layer https://review.openstack.org/162168 | 02:29 |
*** browne has joined #openstack-nova | 02:32 | |
*** patrickeast has quit IRC | 02:40 | |
*** Nic1 has quit IRC | 02:41 | |
*** salv-orlando has quit IRC | 02:41 | |
*** flwang has quit IRC | 02:48 | |
openstackgerrit | Matt Riedemann proposed openstack/nova: Remove power_state.BUILDING https://review.openstack.org/173658 | 02:48 |
*** xylan has joined #openstack-nova | 02:51 | |
*** Nic has joined #openstack-nova | 02:55 | |
*** Nic has quit IRC | 02:55 | |
*** tjones1 has quit IRC | 02:59 | |
*** flwang has joined #openstack-nova | 03:01 | |
*** mriedem has joined #openstack-nova | 03:02 | |
*** tjones1 has joined #openstack-nova | 03:03 | |
*** mriedem1 has quit IRC | 03:04 | |
*** VW_ has quit IRC | 03:04 | |
*** otter768 has joined #openstack-nova | 03:05 | |
*** dsanders has joined #openstack-nova | 03:08 | |
*** mriedem has quit IRC | 03:08 | |
*** tjones1 has quit IRC | 03:12 | |
*** achanda has joined #openstack-nova | 03:13 | |
*** jwcroppe has joined #openstack-nova | 03:14 | |
*** tjones1 has joined #openstack-nova | 03:15 | |
*** tjones1 has left #openstack-nova | 03:15 | |
*** Sukhdev has joined #openstack-nova | 03:17 | |
*** achanda has quit IRC | 03:17 | |
openstackgerrit | Alex Xu proposed openstack/nova: Remove instance group member when delete instance https://review.openstack.org/173664 | 03:17 |
*** Longgeek has joined #openstack-nova | 03:18 | |
*** sbfox has joined #openstack-nova | 03:24 | |
*** aswadr has joined #openstack-nova | 03:25 | |
openstackgerrit | Rui Chen proposed openstack/nova: Fix empty instance nw_info cache https://review.openstack.org/146036 | 03:25 |
*** tojuvone has joined #openstack-nova | 03:33 | |
*** dsanders has quit IRC | 03:34 | |
*** tjones1 has joined #openstack-nova | 03:34 | |
*** tiagogomes has quit IRC | 03:37 | |
*** tiagogomes has joined #openstack-nova | 03:37 | |
*** eliqiao has quit IRC | 03:41 | |
*** eliqiao1 has joined #openstack-nova | 03:41 | |
*** Longgeek has quit IRC | 03:41 | |
*** eliqiao has joined #openstack-nova | 03:43 | |
*** eliqiao1 has quit IRC | 03:45 | |
*** VW_ has joined #openstack-nova | 03:47 | |
*** achanda has joined #openstack-nova | 03:49 | |
*** tjones1 has quit IRC | 03:52 | |
*** tjones1 has joined #openstack-nova | 03:53 | |
*** aswadr has quit IRC | 03:54 | |
*** amotoki_ has joined #openstack-nova | 03:54 | |
openstackgerrit | ShaoHe Feng proposed openstack/nova: remove the redundant policy check for SecurityGroupsOutputController https://review.openstack.org/159369 | 03:55 |
openstackgerrit | Ken'ichi Ohmichi proposed openstack/python-novaclient: WIP: Fix version-list failure https://review.openstack.org/173670 | 03:55 |
*** heyongli_ has quit IRC | 03:55 | |
*** oro has joined #openstack-nova | 03:57 | |
*** otter768 has quit IRC | 04:01 | |
*** oro has quit IRC | 04:04 | |
*** sgordon has quit IRC | 04:04 | |
*** dsanders has joined #openstack-nova | 04:07 | |
*** krtaylor has quit IRC | 04:07 | |
*** tjones1 has quit IRC | 04:07 | |
*** tjones1 has joined #openstack-nova | 04:09 | |
*** sgordon has joined #openstack-nova | 04:10 | |
*** krtaylor has joined #openstack-nova | 04:10 | |
*** sudipto has joined #openstack-nova | 04:10 | |
*** sdake has quit IRC | 04:11 | |
*** tjones1 has quit IRC | 04:13 | |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge security groups tests between v2 and v2.1 https://review.openstack.org/173671 | 04:18 |
*** annashen has joined #openstack-nova | 04:21 | |
*** dsanders has quit IRC | 04:22 | |
*** amotoki_ has quit IRC | 04:24 | |
openstackgerrit | Lianhao Lu proposed openstack/nova: Refactor the bandwidth_usage object's remotable method https://review.openstack.org/173213 | 04:25 |
*** annashen has quit IRC | 04:26 | |
*** salv-orlando has joined #openstack-nova | 04:29 | |
*** ankit_ag has joined #openstack-nova | 04:29 | |
*** VW_ has quit IRC | 04:34 | |
*** sbfox has quit IRC | 04:34 | |
*** rajesht has joined #openstack-nova | 04:35 | |
*** shyama has joined #openstack-nova | 04:36 | |
*** rushiagr_away is now known as rushiagr | 04:36 | |
*** salv-orlando has quit IRC | 04:36 | |
*** Longgeek has joined #openstack-nova | 04:37 | |
*** rushiagr is now known as rushiagr_away | 04:38 | |
*** rushiagr_away is now known as rushiagr | 04:38 | |
*** sbfox has joined #openstack-nova | 04:38 | |
*** dave-mccowan has joined #openstack-nova | 04:39 | |
*** Longgeek has quit IRC | 04:44 | |
openstackgerrit | Chris Friesen proposed openstack/nova: Fix binding:host_id for neutron on evacuate https://review.openstack.org/173615 | 04:44 |
*** sudipto has quit IRC | 04:47 | |
*** dsanders has joined #openstack-nova | 04:49 | |
*** achanda has quit IRC | 04:50 | |
*** achanda has joined #openstack-nova | 04:50 | |
*** xyang1 has quit IRC | 04:52 | |
*** dsanders has quit IRC | 04:57 | |
*** jwcroppe has quit IRC | 05:01 | |
*** unicell has joined #openstack-nova | 05:01 | |
*** mrodden has quit IRC | 05:12 | |
*** apuimedo has joined #openstack-nova | 05:12 | |
*** ajayaa has joined #openstack-nova | 05:14 | |
*** heyongli_ has joined #openstack-nova | 05:16 | |
*** Sukhdev has quit IRC | 05:17 | |
*** heyongli_1 has joined #openstack-nova | 05:19 | |
*** ritesh has quit IRC | 05:20 | |
*** heyongli_ has quit IRC | 05:21 | |
*** cfriesen has quit IRC | 05:21 | |
*** nkrinner has joined #openstack-nova | 05:23 | |
*** ociuhandu has quit IRC | 05:29 | |
*** yuntongjin has quit IRC | 05:34 | |
*** lan has quit IRC | 05:35 | |
*** lan has joined #openstack-nova | 05:35 | |
openstackgerrit | Rui Chen proposed openstack/nova: Fix empty instance nw_info cache https://review.openstack.org/146036 | 05:37 |
openstackgerrit | Rui Chen proposed openstack/nova: Clean up network_api.get_instance_nw_info https://review.openstack.org/173680 | 05:37 |
*** ociuhandu has joined #openstack-nova | 05:38 | |
*** ankit8188 has joined #openstack-nova | 05:38 | |
*** ankit8188 has quit IRC | 05:38 | |
*** Nikolay_St has quit IRC | 05:44 | |
*** dsanders has joined #openstack-nova | 05:48 | |
*** Longgeek has joined #openstack-nova | 05:48 | |
*** dulek has joined #openstack-nova | 05:49 | |
openstackgerrit | Zhenzan Zhou proposed openstack/nova: Bypass ironic server not available issue https://review.openstack.org/173681 | 05:50 |
*** salv-orlando has joined #openstack-nova | 05:50 | |
*** bkopilov has quit IRC | 05:53 | |
*** mrodden has joined #openstack-nova | 05:55 | |
*** ociuhandu has quit IRC | 05:56 | |
*** bkopilov has joined #openstack-nova | 05:59 | |
*** salv-orlando has quit IRC | 06:00 | |
*** otter768 has joined #openstack-nova | 06:02 | |
*** unicell has quit IRC | 06:02 | |
*** isd has joined #openstack-nova | 06:03 | |
*** TobiasE has joined #openstack-nova | 06:03 | |
*** TobiasE has quit IRC | 06:05 | |
*** garyk has joined #openstack-nova | 06:06 | |
*** tjones1 has joined #openstack-nova | 06:06 | |
*** otter768 has quit IRC | 06:07 | |
*** ajayaa has quit IRC | 06:07 | |
*** browne has quit IRC | 06:09 | |
*** hdd has joined #openstack-nova | 06:09 | |
*** sudipto has joined #openstack-nova | 06:11 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Imported Translations from Transifex https://review.openstack.org/171943 | 06:18 |
*** alex_xu has quit IRC | 06:18 | |
*** ajayaa has joined #openstack-nova | 06:24 | |
*** bkopilov has quit IRC | 06:24 | |
*** tjones1 has left #openstack-nova | 06:26 | |
*** armax has quit IRC | 06:28 | |
*** jcoufal has joined #openstack-nova | 06:29 | |
*** alex_xu has joined #openstack-nova | 06:30 | |
*** sahid has joined #openstack-nova | 06:32 | |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge V2 and V2.1 hypervisor functional tests https://review.openstack.org/167082 | 06:35 |
*** RuiChen has joined #openstack-nova | 06:42 | |
*** irenab has joined #openstack-nova | 06:44 | |
*** unicell has joined #openstack-nova | 06:45 | |
*** RuiChen has quit IRC | 06:46 | |
*** achanda has quit IRC | 06:48 | |
*** vilobhmm1 has quit IRC | 06:49 | |
*** bkopilov has joined #openstack-nova | 06:50 | |
*** Mike_D_laptop has joined #openstack-nova | 06:51 | |
*** rfolco has quit IRC | 06:55 | |
*** deepthi has joined #openstack-nova | 06:55 | |
*** bkopilov has quit IRC | 06:56 | |
*** isd has quit IRC | 06:58 | |
*** RuiChen has joined #openstack-nova | 06:59 | |
openstackgerrit | ShaoHe Feng proposed openstack/nova: remove the redundant policy check for SecurityGroupsOutputController https://review.openstack.org/159369 | 07:03 |
openstackgerrit | Ken'ichi Ohmichi proposed openstack/python-novaclient: Don't use SessionClient for version-list https://review.openstack.org/173670 | 07:03 |
*** jeblair has quit IRC | 07:05 | |
*** sbfox has quit IRC | 07:07 | |
*** jeblair has joined #openstack-nova | 07:07 | |
*** salv-orlando has joined #openstack-nova | 07:08 | |
*** rwsu has quit IRC | 07:09 | |
*** bkopilov has joined #openstack-nova | 07:11 | |
*** doron_afk has joined #openstack-nova | 07:11 | |
*** markus_z has joined #openstack-nova | 07:12 | |
*** jichen has joined #openstack-nova | 07:12 | |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge console & console_output tests between v2 and v2.1 https://review.openstack.org/171440 | 07:12 |
*** moshele has joined #openstack-nova | 07:19 | |
openstackgerrit | Sudipta Biswas proposed openstack/nova: Allow discrete blocks of online pCPUs for pinning https://review.openstack.org/173187 | 07:22 |
sudipto | bauzas, Hi, made some changes to the review from last night. Didn't have to write a new function, since there's already a _host.get_online_cpus() function present in the latest Kilo code... | 07:23 |
openstackgerrit | ShaoHe Feng proposed openstack/nova: Move unlock_override policy enforcement into V2.1 REST API layer https://review.openstack.org/162168 | 07:23 |
*** jistr has joined #openstack-nova | 07:24 | |
*** annashen has joined #openstack-nova | 07:24 | |
openstackgerrit | Ghanshyam Mann proposed openstack/nova: Share multiple create server tests between v2 and v2.1 https://review.openstack.org/172801 | 07:25 |
gilliard | Good $time_of_day! | 07:27 |
openstackgerrit | Sergey Nikitin proposed openstack/nova: Fixed incorrect dhcp_server value during nova-network creation https://review.openstack.org/173362 | 07:28 |
openstackgerrit | Sudipta Biswas proposed openstack/nova: Allow discrete blocks of online pCPUs for pinning https://review.openstack.org/173187 | 07:29 |
*** annashen has quit IRC | 07:29 | |
openstackgerrit | Ken'ichi Ohmichi proposed openstack/python-novaclient: Add min/max microversions to version-list cmd https://review.openstack.org/173711 | 07:30 |
*** ankit_ag has quit IRC | 07:30 | |
*** RuiChen has quit IRC | 07:31 | |
*** eglynn has quit IRC | 07:33 | |
*** ankit_ag has joined #openstack-nova | 07:33 | |
*** eglynn has joined #openstack-nova | 07:33 | |
*** rfolco has joined #openstack-nova | 07:39 | |
*** dsanders has quit IRC | 07:41 | |
*** romainh has joined #openstack-nova | 07:41 | |
*** c0m0 has joined #openstack-nova | 07:42 | |
*** jlanoux has joined #openstack-nova | 07:42 | |
*** bkopilov has quit IRC | 07:45 | |
*** Nikolay_St has joined #openstack-nova | 07:45 | |
*** belmoreira has joined #openstack-nova | 07:47 | |
*** eglynn has quit IRC | 07:47 | |
*** bkopilov has joined #openstack-nova | 07:49 | |
*** rushiagr is now known as rushiagr_away | 07:49 | |
*** naggappan has joined #openstack-nova | 07:49 | |
*** bkopilov has quit IRC | 07:54 | |
*** bkopilov has joined #openstack-nova | 07:54 | |
ankit_ag | sahid: hi | 07:59 |
*** dsanders has joined #openstack-nova | 08:01 | |
*** rushiagr_away is now known as rushiagr | 08:01 | |
*** tjones1 has joined #openstack-nova | 08:02 | |
*** otter768 has joined #openstack-nova | 08:02 | |
*** tjones1 has quit IRC | 08:04 | |
*** tjones1 has joined #openstack-nova | 08:05 | |
*** eglynn has joined #openstack-nova | 08:05 | |
sahid | hello ankit_ag | 08:06 |
*** otter768 has quit IRC | 08:07 | |
ankit_ag | sahid: I want to discuss about your comment on https://review.openstack.org/#/c/173296 | 08:09 |
*** dave-mccowan has quit IRC | 08:09 | |
*** bkopilov has quit IRC | 08:10 | |
ankit_ag | sahid: if delete is defined, in that case we do not need to explicitly check for the condition "changes-since" | 08:10 |
ankit_ag | and thats the existing behaviour | 08:11 |
ankit_ag | sahid: "changes-since" condition is to set 'deleted'=False if 'deleted' is not passed in to filters, so if delete is defined then we do not need to explicitly set that flag again. | 08:13 |
*** mpavone has joined #openstack-nova | 08:14 | |
*** zzz_pczesno is now known as pczesno | 08:14 | |
*** shyama has quit IRC | 08:15 | |
*** eglynn has quit IRC | 08:16 | |
sahid | ankit_ag: the comment says "No 'changes-since', so we only want non-deleted servers" that makes me think in all cases if changes-since is not defined we want deleted=False | 08:17 |
*** bkopilov has joined #openstack-nova | 08:17 | |
*** lucasagomes has joined #openstack-nova | 08:17 | |
sahid | we are breakin whether deleted is defined and equal to True and changes-since not defined | 08:18 |
*** ujjain has joined #openstack-nova | 08:20 | |
*** dsanders has quit IRC | 08:20 | |
sahid | ankit_ag: hum let me double check... | 08:21 |
ankit_ag | sahid: yes sure, because by looking at the master code, it is clear that change-since needs to check only if deleted is not defined in filters | 08:23 |
*** yamahata has quit IRC | 08:25 | |
*** a5t3r1x has joined #openstack-nova | 08:25 | |
a5t3r1x | hello:-) | 08:26 |
a5t3r1x | quick question. | 08:26 |
a5t3r1x | it is possible to use eth1:0 as the flat_interface? | 08:26 |
*** ndipanov has joined #openstack-nova | 08:27 | |
*** eglynn has joined #openstack-nova | 08:27 | |
sahid | ankit_ag: replied | 08:28 |
*** inhumanitas has quit IRC | 08:29 | |
ankit_ag | sahid: yes, that makes sense. Thank you ! | 08:29 |
*** ociuhandu has joined #openstack-nova | 08:31 | |
*** bkopilov has quit IRC | 08:39 | |
*** heyongli_1 has quit IRC | 08:43 | |
*** pixelb has joined #openstack-nova | 08:44 | |
*** shaohe_feng has quit IRC | 08:45 | |
*** bkopilov has joined #openstack-nova | 08:45 | |
openstackgerrit | Sergey Nikitin proposed openstack/nova: Changed backporting of embedded Network object when backporting a FixedIP https://review.openstack.org/173723 | 08:47 |
*** doron_afk is now known as doron_mtg | 08:49 | |
*** shaohe_feng has joined #openstack-nova | 08:49 | |
*** park_heijlong has quit IRC | 08:50 | |
*** bkopilov has quit IRC | 08:51 | |
*** bkopilov has joined #openstack-nova | 08:52 | |
*** shaohe_feng has quit IRC | 08:55 | |
*** krtaylor has quit IRC | 08:57 | |
*** bkopilov has quit IRC | 08:57 | |
*** pask81 has joined #openstack-nova | 08:59 | |
*** pask81 has quit IRC | 08:59 | |
openstackgerrit | Eli Qiao proposed openstack/nova: Libvirt: pre check cpuset when do LM https://review.openstack.org/173729 | 09:02 |
*** warewang has joined #openstack-nova | 09:02 | |
*** Kevin_Zheng has joined #openstack-nova | 09:03 | |
*** shyama has joined #openstack-nova | 09:03 | |
*** pask81 has joined #openstack-nova | 09:03 | |
*** krtaylor has joined #openstack-nova | 09:04 | |
*** sudipto has quit IRC | 09:05 | |
*** Marga_ has joined #openstack-nova | 09:06 | |
openstackgerrit | Eli Qiao proposed openstack/nova: Libvirt: pre check cpuset when do LM https://review.openstack.org/173729 | 09:08 |
*** ipj has joined #openstack-nova | 09:08 | |
ipj | hey guys | 09:08 |
ipj | I'm having serious problems with instance resizing | 09:08 |
ipj | the command "nova resize" etc is working | 09:08 |
ipj | and I get no errors | 09:08 |
ipj | but the status of the VM does not change... simply, the resize takes no effect | 09:08 |
ipj | anyone can help me? | 09:09 |
*** sdake has joined #openstack-nova | 09:10 | |
*** jamie_h has joined #openstack-nova | 09:12 | |
*** eliqiao has quit IRC | 09:13 | |
kashyap | ipj: You might get more responses on #openstack for such questions, this channel is mostly for upstream Nova development | 09:15 |
ipj | ok kashyap, thanks | 09:15 |
*** pask81 has quit IRC | 09:15 | |
openstackgerrit | Sergey Nikitin proposed openstack/nova: Changed backporting of embedded Network object when backporting a FixedIP https://review.openstack.org/173723 | 09:16 |
*** sudipto has joined #openstack-nova | 09:21 | |
*** pask81 has joined #openstack-nova | 09:22 | |
*** derekh has joined #openstack-nova | 09:25 | |
*** Marga_ has quit IRC | 09:28 | |
*** pask81 has quit IRC | 09:28 | |
*** Marga_ has joined #openstack-nova | 09:29 | |
*** abhishekk has joined #openstack-nova | 09:29 | |
*** TobiasE has joined #openstack-nova | 09:31 | |
openstackgerrit | Tracy Jones proposed openstack/nova: Cleanup of compute api reboot method https://review.openstack.org/127427 | 09:31 |
*** pask81 has joined #openstack-nova | 09:31 | |
*** gszasz has joined #openstack-nova | 09:36 | |
*** gszasz_ has joined #openstack-nova | 09:37 | |
*** garyk has quit IRC | 09:39 | |
*** tjones1 has quit IRC | 09:39 | |
*** garyk has joined #openstack-nova | 09:40 | |
*** tjones1 has joined #openstack-nova | 09:41 | |
*** gszasz has quit IRC | 09:41 | |
*** ZZelle has quit IRC | 09:44 | |
ipj | kashyap, I dunno if you are curious: the VM resizing is working from a small flavor to a bigger one, but not in the other direction | 09:45 |
kashyap | ipj: Yeah, I remember testing that a long time ago, | 09:46 |
*** tjones1 has quit IRC | 09:46 | |
kashyap | and IIRC, it wasn't considered a priority | 09:47 |
ipj | kashyap, may this be a problem related to LVM? | 09:47 |
kashyap | At one point (later 2013) Nova used to return bogus success for a resize operation, but it was fixed in later releases -- https://bugzilla.redhat.com/show_bug.cgi?id=1045515 | 09:48 |
openstack | bugzilla.redhat.com bug 1045515 in openstack-nova "Re-size of an ACTIVE Nova instance returns success (albeit, bogus)" [Unspecified,Closed: currentrelease] - Assigned to ndipanov | 09:48 |
moshele | hi can nova cores review https://review.openstack.org/#/c/168317/ and https://review.openstack.org/#/c/168680/ ? | 09:48 |
kashyap | ipj: Hmm, I don't test a whole lot with LVM, so I can't say it's that for sure. | 09:49 |
*** liusheng has quit IRC | 09:51 | |
*** jamielennox is now known as jamielennox|away | 09:52 | |
*** gszasz has joined #openstack-nova | 09:56 | |
*** gszasz has quit IRC | 09:57 | |
*** jhesketh has quit IRC | 09:57 | |
*** gszasz_ has quit IRC | 09:58 | |
rajesht | ipj: IMO resize down is not supported | 09:59 |
rajesht | ipj: https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L6225 | 09:59 |
*** jhesketh has joined #openstack-nova | 10:00 | |
*** amotoki has joined #openstack-nova | 10:00 | |
*** gszasz has joined #openstack-nova | 10:00 | |
*** Longgeek has quit IRC | 10:02 | |
*** Longgeek_ has joined #openstack-nova | 10:02 | |
kashyap | Instead of throwing up hands with reason "Unable to resize disk down", it should throw a little more reasonable error: "Invalid operation: resize disk down" or something like that. "unable" implies there's some error on the user's part. | 10:03 |
*** otter768 has joined #openstack-nova | 10:03 | |
*** otter768 has quit IRC | 10:08 | |
ipj | rajesht, thanks | 10:09 |
ipj | imho, this should be notified to the user in some fashion | 10:09 |
ipj | I could never understand that it isn't supported | 10:09 |
Mike_D_laptop | rajesht: Hi! Why you think so? | 10:09 |
Mike_D_laptop | rajesht: Ohh :)I forgot... never mind. It's possible for instance which booted from volume... | 10:11 |
*** irenab has quit IRC | 10:12 | |
rajesht | Mike_D_laptop: np :) | 10:13 |
*** sudipto has quit IRC | 10:16 | |
*** doron_mtg has quit IRC | 10:21 | |
*** dheeraj-gupta-4 has joined #openstack-nova | 10:22 | |
*** moshele has quit IRC | 10:22 | |
*** achanda has joined #openstack-nova | 10:22 | |
*** dguitarbite has joined #openstack-nova | 10:24 | |
*** moshele has joined #openstack-nova | 10:25 | |
*** annashen has joined #openstack-nova | 10:26 | |
*** jcoufal has quit IRC | 10:27 | |
*** lpetrut has joined #openstack-nova | 10:27 | |
*** achanda has quit IRC | 10:27 | |
*** jcoufal has joined #openstack-nova | 10:27 | |
*** IanGovett has joined #openstack-nova | 10:28 | |
*** sudipto has joined #openstack-nova | 10:28 | |
*** annashen has quit IRC | 10:31 | |
*** romainh has quit IRC | 10:33 | |
*** ociuhandu has quit IRC | 10:34 | |
*** claudiub has joined #openstack-nova | 10:34 | |
*** amotoki_ has joined #openstack-nova | 10:40 | |
*** amotoki has quit IRC | 10:43 | |
*** sdake has quit IRC | 10:46 | |
*** romainh has joined #openstack-nova | 10:48 | |
*** pask81 has quit IRC | 10:48 | |
*** zhenguo has quit IRC | 10:48 | |
*** asselin_ has quit IRC | 10:49 | |
*** popw has quit IRC | 10:52 | |
*** ujjain has quit IRC | 10:59 | |
*** Kevin_Zheng has quit IRC | 11:02 | |
*** warewang has quit IRC | 11:02 | |
*** deepthi has quit IRC | 11:02 | |
*** Marga_ has quit IRC | 11:02 | |
*** amotoki has joined #openstack-nova | 11:02 | |
ankit_ag | alex_xu: hi | 11:04 |
*** amotoki_ has quit IRC | 11:05 | |
*** irenab has joined #openstack-nova | 11:06 | |
*** bkopilov has joined #openstack-nova | 11:07 | |
*** zz_johnthetubagu is now known as johnthetubaguy | 11:07 | |
*** dims has joined #openstack-nova | 11:08 | |
*** irenab has quit IRC | 11:09 | |
*** romainh has left #openstack-nova | 11:09 | |
*** ferest has joined #openstack-nova | 11:10 | |
*** ferest has quit IRC | 11:11 | |
*** ferest has joined #openstack-nova | 11:11 | |
*** romainh has joined #openstack-nova | 11:12 | |
*** ferest has quit IRC | 11:13 | |
*** gszasz has quit IRC | 11:16 | |
*** popw has joined #openstack-nova | 11:20 | |
openstackgerrit | Mh Raies proposed openstack/nova: Enable tenant/user specific instance filtering https://review.openstack.org/139025 | 11:21 |
*** gszasz has joined #openstack-nova | 11:21 | |
*** mwagner_lap has quit IRC | 11:21 | |
*** doron_mtg has joined #openstack-nova | 11:21 | |
*** Longgeek has joined #openstack-nova | 11:29 | |
*** otter768 has joined #openstack-nova | 11:31 | |
*** Longgeek_ has quit IRC | 11:31 | |
bauzas | johnthetubaguy: around ? | 11:33 |
johnthetubaguy | bauzas: I am | 11:33 |
johnthetubaguy | bauzas: although I am running to lunch in a sec, can I help? | 11:34 |
bauzas | johnthetubaguy: I saw your comments on https://review.openstack.org/#/c/141486/ | 11:34 |
johnthetubaguy | ah, yes | 11:34 |
bauzas | johnthetubaguy: just wanted to discuss with you for it, because I'm working on the scaling out scheduler | 11:34 |
bauzas | johnthetubaguy: see https://etherpad.openstack.org/p/liberty-nova-summit-ideas L67 | 11:34 |
johnthetubaguy | ah, OK, cools | 11:34 |
bauzas | johnthetubaguy: I'm thinking about having a shared-state HostState | 11:35 |
bauzas | johnthetubaguy: like Google Omega | 11:35 |
bauzas | johnthetubaguy: but go to lunch, and we'll discussing later on | 11:35 |
johnthetubaguy | bauzas: so I am mostly agreed with omega, from when I remember reading that last | 11:36 |
*** ipj has quit IRC | 11:36 | |
johnthetubaguy | bauzas: its partly why I don't like the two level scheduler, unless its a clean partition | 11:36 |
bauzas | johnthetubaguy: I also thought about Kafka, so maybe an hangout could be cool - unfortunately, we're close to RC2 so you would probably be just overworking now | 11:36 |
*** ociuhandu has joined #openstack-nova | 11:37 | |
johnthetubaguy | bauzas: more pressed for time with internal release stuff at the moment, but we should catch up on that, let me go get something to eat, and get back to you | 11:37 |
bauzas | johnthetubaguy: by remembering the most papers I saw, I think a divide-and-conquer scheduler (ie. a sharding one) is less better than a shared-state one | 11:37 |
johnthetubaguy | bauzas: its not in the way I sharded it though | 11:38 |
johnthetubaguy | bauzas: I was presenting a partitioned shared state system, not a two level system | 11:38 |
bauzas | johnthetubaguy: we have the chance to have a read-only HostState for the scheduler | 11:38 |
bauzas | johnthetubaguy: oh like Omega then ? | 11:39 |
johnthetubaguy | bauzas: thats unrelated, while useful, will talk more later | 11:39 |
johnthetubaguy | bauzas: yeah, its quite similar actually | 11:39 |
bauzas | johnthetubaguy: okay, I'll write out a spec | 11:39 |
bauzas | johnthetubaguy: just for opening discussion | 11:39 |
johnthetubaguy | bauzas: yes, the basic idea I have is two things | 11:39 |
bauzas | johnthetubaguy: because at the moment, it's just braindump | 11:39 |
bauzas | johnthetubaguy: I saw : Kafka for providing states, and partitioned shared-state for many scheds ? | 11:40 |
johnthetubaguy | bauzas: if you can split your infrastructure into groups based on the incomming request (we do that with groups of flavors, onmetal vs ssd vs spinny disk, etc) | 11:40 |
johnthetubaguy | thats not two level, its sharded | 11:40 |
johnthetubaguy | the other bit yeah, is efficient updating of the shared state | 11:41 |
johnthetubaguy | kafaka is a good messaging system that does what we need for that, I think | 11:41 |
bauzas | agreed, we have the same direction | 11:41 |
johnthetubaguy | OK cools | 11:41 |
johnthetubaguy | talk more later | 11:41 |
bauzas | okay, let me draft out some manifesto spec | 11:41 |
johnthetubaguy | bauzas: thanks, sounds good | 11:42 |
*** doron_mtg is now known as doron_afk | 11:48 | |
*** aix has quit IRC | 11:50 | |
*** beagles has joined #openstack-nova | 11:52 | |
*** tojuvone has quit IRC | 11:52 | |
*** otter768 has quit IRC | 11:56 | |
*** jecarey has quit IRC | 11:58 | |
*** doron has joined #openstack-nova | 12:01 | |
*** doron_afk has quit IRC | 12:02 | |
sudipto | bauzas, johnthetubaguy Can openstack currently perform resize of VMs without rebooting it for KVM? | 12:04 |
*** raildo has joined #openstack-nova | 12:06 | |
*** dprince has joined #openstack-nova | 12:08 | |
*** irenab has joined #openstack-nova | 12:08 | |
*** mwagner_lap has joined #openstack-nova | 12:09 | |
*** MVenesio has joined #openstack-nova | 12:13 | |
*** irenab has quit IRC | 12:13 | |
johnthetubaguy | sudipto: no driver can do that today, AFAIK, there are a few ideas that keep being raised though | 12:14 |
johnthetubaguy | sudipto: scale up isn't considered very cloudy, so its not something thats been priorities up to this point | 12:15 |
*** irenab has joined #openstack-nova | 12:16 | |
*** irenab has quit IRC | 12:17 | |
openstackgerrit | garyk proposed openstack/nova: Resource tracker: unable to restart nova compute https://review.openstack.org/173794 | 12:18 |
*** aysyd has joined #openstack-nova | 12:18 | |
*** irenab has joined #openstack-nova | 12:18 | |
bauzas | johnthetubaguy: sudipto: yeah agreed and that sounds very driver specific | 12:19 |
bauzas | even guest specific | 12:19 |
*** irenab has quit IRC | 12:20 | |
*** irenab has joined #openstack-nova | 12:20 | |
sudipto | johnthetubaguy, Hmm, ok. That answers my question. Libvirt doesn't have the support AFAIK...but you say that it's not a cloud use case coz? | 12:21 |
sudipto | trying to understand :) | 12:21 |
johnthetubaguy | sudipto: its just not a high priority cloud use case, because with cattle VMs you just start a new bigger VM, or ideally, more smaller VMs, its "pet" VMs that would need to be resized with zero downtime, and we are not really optimising for those | 12:22 |
sudipto | johnthetubaguy, Okk... | 12:23 |
johnthetubaguy | sudipto: having said that, I keep seeing blueprints for this crop us, so it may get somewhere soon | 12:25 |
*** amotoki has quit IRC | 12:25 | |
johnthetubaguy | as bauzas mentioned, there are lots issues making it work in general, but a lot of the "hotplug" work should be possible, even if some number of users need to reboot to see the extra capacity thats added, at least they could then choose when to reboot | 12:26 |
johnthetubaguy | (but I have my xenapi head on talking about that, not sure about libvirt) | 12:26 |
sudipto | johnthetubaguy, Yeah makes sense... | 12:26 |
*** amotoki has joined #openstack-nova | 12:27 | |
bauzas | johnthetubaguy: the main problem I see with this approach is that an image can be cloud-ready or not, and thus can't support resizing | 12:27 |
bauzas | johnthetubaguy: I mean live resizing | 12:28 |
*** aix has joined #openstack-nova | 12:28 | |
johnthetubaguy | bauzas: think more about hotplugging RAM and CPU, rather than resizing a disk | 12:28 |
sudipto | bauzas, yeah | 12:28 |
sudipto | bauzas, I was talking about RAM and CPU | 12:28 |
bauzas | johnthetubaguy: I remember my lovely operator days where I had to play with Windows guest images on OpenStack | 12:28 |
johnthetubaguy | right, thats very doable, just we don't do it | 12:29 |
*** zhenguo has joined #openstack-nova | 12:29 | |
sudipto | johnthetubaguy, assuming the driver should also support it which i think it currently doesn't. Atleast for Libvirt. | 12:29 |
johnthetubaguy | right, the code is not there, but its possible to add it | 12:30 |
sudipto | johnthetubaguy, Thanks! That helps me understand it! | 12:30 |
sudipto | bauzas, Thanks to you too! | 12:30 |
johnthetubaguy | no problem | 12:30 |
*** amotoki has quit IRC | 12:31 | |
*** dheeraj-gupta-4 has left #openstack-nova | 12:32 | |
*** eharney has joined #openstack-nova | 12:34 | |
*** dkliban_afk is now known as dkliban | 12:35 | |
*** irenab has quit IRC | 12:35 | |
rajesht | dansmith: you around ? | 12:35 |
*** irenab has joined #openstack-nova | 12:36 | |
*** vladikr has joined #openstack-nova | 12:39 | |
*** ZZelle has joined #openstack-nova | 12:39 | |
*** baoli has joined #openstack-nova | 12:42 | |
*** jaypipes has quit IRC | 12:42 | |
*** otter768 has joined #openstack-nova | 12:43 | |
rajesht | dims: you around ? | 12:43 |
openstackgerrit | garyk proposed openstack/nova: VMware: enable a resize of instance with no root disk https://review.openstack.org/173802 | 12:44 |
*** irenab has quit IRC | 12:45 | |
*** otter768 has quit IRC | 12:46 | |
*** smccully has joined #openstack-nova | 12:46 | |
*** irenab has joined #openstack-nova | 12:46 | |
*** rasoto_ has joined #openstack-nova | 12:47 | |
*** irenab has quit IRC | 12:47 | |
*** annegentle has joined #openstack-nova | 12:50 | |
openstackgerrit | sahid proposed openstack/nova: compute: make test rpcapi more flexible https://review.openstack.org/164177 | 12:51 |
*** naggappan has quit IRC | 12:51 | |
bauzas | alaski: quick update on the idempotent tests, it seems it's probably not due to the race condition we mentioned | 12:52 |
rajesht | johnthetubaguy: you around ? | 12:52 |
johnthetubaguy | rajesht: kind of yes | 12:52 |
*** baoli_ has joined #openstack-nova | 12:54 | |
*** garyk has quit IRC | 12:54 | |
bauzas | alaski: because the Tempest assertion is failing before we enter the instance.save() cell condition for _from_api | 12:54 |
johnthetubaguy | rajesht: can I help? | 12:55 |
rajesht | johnthetubaguy: there is an issue after adding V4 RPC api, if instance deleted during resize, we are not able to restart the n-cpu | 12:55 |
rajesht | johnthetubaguy: pl check the bug https://bugs.launchpad.net/nova/+bug/1444300 | 12:55 |
openstack | Launchpad bug 1444300 in OpenStack Compute (nova) "nova-compute service doesn't restart if resize operation fails" [Undecided,New] - Assigned to Rajesh Tailor (rajesh-tailor) | 12:55 |
rajesht | the decorators around the ComputeManager methods don't get parameters as kwargs, that why they failed with KeyError | 12:56 |
rajesht | johnthetubaguy: correct me If I'm wrong | 12:56 |
*** jaypipes has joined #openstack-nova | 12:56 | |
*** baoli has quit IRC | 12:56 | |
*** rasoto_ has left #openstack-nova | 12:56 | |
johnthetubaguy | rajesht: I don't know, sounds like a bug I just saw: https://review.openstack.org/#/c/173794/ | 12:57 |
*** claudiub has quit IRC | 12:57 | |
rajesht | johnthetubaguy: let me check, If this patch fixes this issue | 12:58 |
*** busterswt has joined #openstack-nova | 13:01 | |
*** sald has joined #openstack-nova | 13:05 | |
*** jecarey has joined #openstack-nova | 13:06 | |
*** neelashah has joined #openstack-nova | 13:06 | |
*** sudipto has quit IRC | 13:07 | |
*** dboik_ has quit IRC | 13:08 | |
*** adalbas has quit IRC | 13:12 | |
openstackgerrit | sahid proposed openstack/nova: servicegroup: remove get_all method never used as public https://review.openstack.org/168876 | 13:12 |
*** dims has quit IRC | 13:13 | |
*** burt has joined #openstack-nova | 13:14 | |
*** irenab has joined #openstack-nova | 13:14 | |
*** dims has joined #openstack-nova | 13:15 | |
rajesht | johnthetubaguy: with https://review.openstack.org/#/c/173794/ patch, I'm able to start the n-cpu service | 13:15 |
rajesht | johnthetubaguy: but the problem still remains which I have mentioned in https://bugs.launchpad.net/nova/+bug/1444300 LP bug | 13:16 |
openstack | Launchpad bug 1444300 in OpenStack Compute (nova) "nova-compute service doesn't restart if resize operation fails" [Undecided,New] - Assigned to Rajesh Tailor (rajesh-tailor) | 13:16 |
rajesht | johnthetubaguy: The issue which I have reported is because of newly added RPC V4.0 API, where methods gets parameter in kwargs from rpcapi | 13:17 |
rajesht | johnthetubaguy: but don't pass these kwargs as args to RPC 3.x API, | 13:18 |
*** thomasem has joined #openstack-nova | 13:18 | |
rajesht | johnthetubaguy: methods of ComputeManager class (RPC 3.x) expects objects as kwargs, but gets as args from RPC 4.0 _ComputeV4Proxy class. | 13:19 |
johnthetubaguy | rajesht: hmm, interesting, we had something related for live-migrate, dansmith might be able to help take a look at that | 13:19 |
*** thomasem has quit IRC | 13:19 | |
rajesht | IMO, the same will fail for each method defined in 4.0 because the decorators around 3.x expects kwargs | 13:19 |
alaski | bauzas: where did you see that? | 13:20 |
bauzas | alaski: local debugging | 13:20 |
johnthetubaguy | rajesht: a lot of the decorators use a util method to fix that, but not all of them | 13:20 |
alaski | bauzas: did it fail with a different error? | 13:20 |
johnthetubaguy | rajesht: I can't dig into that right now, I am in the middle of another bug I am afraid, sorry to not be more help right this second | 13:21 |
bauzas | alaski: sec, reproducing the behaviour | 13:21 |
rajesht | johnthetubaguy: np, thanks for your time | 13:21 |
bauzas | alaski: just to be clear, instance.save() is called when terminating here https://github.com/openstack/tempest/blob/master/tempest/thirdparty/boto/test_ec2_instance_run.py#L126 | 13:22 |
dims | rajesht: pong | 13:22 |
rajesht | dims: hi | 13:22 |
bauzas | alaski: I left a pdb trace within _handle_cell_update_from_api and it got executed only when it was terminating | 13:23 |
*** MVenesio has quit IRC | 13:23 | |
rajesht | dims: could you please take a look at this LP bug https://bugs.launchpad.net/nova/+bug/1444300 | 13:23 |
openstack | Launchpad bug 1444300 in OpenStack Compute (nova) "nova-compute service doesn't restart if resize operation fails" [Undecided,New] - Assigned to Rajesh Tailor (rajesh-tailor) | 13:23 |
rajesht | dims: IMO it is because of newly added RPC V4.0 | 13:23 |
*** david-lyle has quit IRC | 13:24 | |
alaski | bauzas: okay. I may need to trace this locally then to see what's going on | 13:24 |
rajesht | dims: I was discussing with johnthetubaguy related to this only. | 13:24 |
bauzas | alaski: I just checked | 13:25 |
*** dboik has joined #openstack-nova | 13:25 | |
bauzas | alaski: so https://github.com/openstack/tempest/blob/master/tempest/thirdparty/boto/test_ec2_instance_run.py#L101 | 13:25 |
bauzas | alaski: later calls https://github.com/openstack/nova/blob/master/nova/api/ec2/cloud.py#L1405-L1418 | 13:26 |
*** thomasem has joined #openstack-nova | 13:26 | |
bauzas | alaski: but compute_api.create() is actually calling the main API | 13:26 |
bauzas | alaski: because the cells API returns the super method | 13:26 |
bauzas | alaski: https://github.com/openstack/nova/blob/master/nova/compute/cells_api.py#L199 | 13:27 |
*** VW_ has joined #openstack-nova | 13:27 | |
*** mriedem has joined #openstack-nova | 13:27 | |
*** cbader has joined #openstack-nova | 13:27 | |
*** irenab has quit IRC | 13:27 | |
*** irenab has joined #openstack-nova | 13:29 | |
bauzas | alaski: and https://github.com/openstack/nova/blob/master/nova/compute/api.py#L919 calls .update() instrad | 13:29 |
bauzas | instead even | 13:29 |
*** annashen has joined #openstack-nova | 13:29 | |
alex_xu | ankit_ag: hi | 13:29 |
alaski | bauzas: it eventually calls save in create_db_entry... | 13:30 |
bauzas | alaski: but as it's done on the main compute.api, it's not having a cell_name right ? | 13:30 |
*** r-daneel has joined #openstack-nova | 13:31 | |
alaski | oh, it calls create() | 13:31 |
alaski | bauzas: right, it does not | 13:31 |
ankit_ag | alex_xu: hi, I was looking for your help in https://review.openstack.org/#/c/171523/ | 13:31 |
bauzas | alaski: so https://github.com/openstack/nova/blob/master/nova/objects/instance.py#L770 returns False even if it's called somewhere | 13:32 |
alex_xu | ankit_ag: ok, reading your reply | 13:32 |
alaski | bauzas: yeah. but it's calling create() and doesn't need to be synced at this point | 13:33 |
bauzas | alaski: agreed | 13:33 |
*** garyk has joined #openstack-nova | 13:33 | |
bauzas | alaski: and then .update() | 13:33 |
*** annashen has quit IRC | 13:33 | |
ankit_ag | should we make these changes or it is fine to not allow an user to delete an instance by name when it is in soft_deleted state | 13:33 |
dansmith | rajesht: johnthetubaguy: although that seems similar, I don't think it is. That looks like the delete deletes the instance files under the resize, which we shouldn't let it do | 13:34 |
alaski | bauzas: and then the ec2 api looks up the instance and adds system_metadata for the client token | 13:34 |
*** ipj has joined #openstack-nova | 13:34 | |
*** ipj has quit IRC | 13:35 | |
johnthetubaguy | dansmith: ah, gotcha, didn't look too closely, but I guess we always want to allow delete | 13:35 |
rajesht | dansmith: IMO we need to use safe_utils.getcallargs in errors_out_migration decorator ? | 13:35 |
dansmith | johnthetubaguy: I guess, but seems like resize/migrate would be a good time to *not* allow it | 13:35 |
ankit_ag | because for me it is an issue, whether we document it in the api documentation, because for all the remaining cases we are allowing user to delete instance by name and id, so why not in case of soft_deleted instances. | 13:35 |
bauzas | alaski: yeah https://github.com/openstack/nova/blob/master/nova/api/ec2/cloud.py#L1423 | 13:35 |
dansmith | rajesht: that may be collateral damage, yeah | 13:36 |
johnthetubaguy | dansmith: thats certainly easier | 13:36 |
bauzas | alaski: within there, we have an instance.save() | 13:36 |
johnthetubaguy | dansmith: I mean we see folks delete stuff half way through building, and that can create some mess, but still you want it, in some way | 13:36 |
dansmith | johnthetubaguy: it just seems like if we're in the middle of the copy, the easiest thing to do is wait for it to finish and then do the delete.. | 13:37 |
dansmith | johnthetubaguy: rajesht: is this a regression or just a thing you found/ | 13:37 |
ankit_ag | alex_xu: I am looking for ypur opinion about the delete api functionality | 13:37 |
johnthetubaguy | dansmith: yeah | 13:37 |
*** jwcroppe has joined #openstack-nova | 13:38 | |
rajesht | dansmith: yes | 13:38 |
*** jwcroppe has quit IRC | 13:38 | |
dansmith | johnthetubaguy: obviously I'd like to be able to signal the thread, stop the copy immediately, etc, but that likely requires another rpc call to the other side to tell it to clean up, etc | 13:38 |
bauzas | alaski: but unless I'm wrong, the created instance doesn't still has a cell_name field | 13:38 |
alex_xu | ankit_ag: I'm ok fix the first problem in the separated patch, but I'm still thinking about the second one | 13:38 |
bauzas | alaski: so it doesn't still sync | 13:39 |
johnthetubaguy | dansmith: agreed | 13:39 |
*** jwcroppe has joined #openstack-nova | 13:39 | |
bauzas | alaski: I can trace it, sec | 13:39 |
alaski | bauzas: that may be true, unless it has synced up from the cell at that point | 13:39 |
*** ozialien has joined #openstack-nova | 13:39 | |
alaski | bauzas: but either way, it should save the system_metadata properly | 13:39 |
alaski | bauzas: in the api db | 13:39 |
*** neelashah has quit IRC | 13:40 | |
ankit_ag | alex_xu: By looking at the souce code, second issue looks like it is by design there | 13:40 |
*** neelashah has joined #openstack-nova | 13:40 | |
ankit_ag | so I have filed a doumentation issue for the same | 13:40 |
*** garyk has quit IRC | 13:41 | |
*** ferest has joined #openstack-nova | 13:41 | |
alex_xu | ankit_ag: what the behavior if status=deleted without deleted=true? | 13:42 |
bauzas | alaski: sec, running again the tests | 13:42 |
ankit_ag | alex_xu: For the first issue I have submitted separate patch to handle 'deletedd' filter . But for 'soft_deleted' I need to handle in the same patch | 13:42 |
alex_xu | ankit_ag: yes | 13:42 |
alex_xu | ankit_ag: and how about we make https://review.openstack.org/#/c/171523/ depend on this fix https://review.openstack.org/#/c/173296/? | 13:43 |
bauzas | alaski: confirmed, instance.cell_name is None | 13:43 |
*** ferest has quit IRC | 13:43 | |
alaski | bauzas: okay. but does it save the system_metada properly? | 13:43 |
alaski | bauzas: and does that get overwritten at some point later? | 13:44 |
*** neelashah has quit IRC | 13:44 | |
bauzas | alaski: just to be clear, I'm following the path for instance.save() | 13:45 |
ankit_ag | alex_xu: https://review.openstack.org/#/c/171523/ is not dependent https://review.openstack.org/#/c/173296/, because that only handles deleted filter and not 'soft_deleted' | 13:45 |
bauzas | alaski: so, as cell_name is None in https://github.com/openstack/nova/blob/master/nova/api/ec2/cloud.py#L1435 | 13:45 |
alex_xu | ankit_ag: ok, I see now | 13:46 |
*** hdd has quit IRC | 13:46 | |
*** irenab has quit IRC | 13:46 | |
alex_xu | ankit_ag: I checked the code, the status=deleted works without deleted=true. If this is right, I think we can use stauts=soft-deleted without soft_delted=true also | 13:47 |
*** triggerz has joined #openstack-nova | 13:47 | |
*** mtanino has joined #openstack-nova | 13:48 | |
triggerz | what is done when a vm is in the "spawning" state? | 13:48 |
*** markvoelker has joined #openstack-nova | 13:48 | |
ankit_ag | alex_xu: yes, thats correct. But on master status=soft_deleted does not work without deleted=true | 13:49 |
*** READ10 has joined #openstack-nova | 13:49 | |
alex_xu | ankit_ag: I prefer to fix that | 13:49 |
bauzas | alaski: so, yeah, confirmed, at any moment when saving, we're syncing | 13:49 |
*** hdd has joined #openstack-nova | 13:50 | |
alex_xu | ankit_ag: but looks like that can be fix in separated patch also | 13:51 |
alex_xu | ankit_ag: a fix https://review.openstack.org/#/c/171523/ can depend on | 13:51 |
ankit_ag | alex_xu: in sqlalchemy.api 'deleted' filter is applied on top of the 'soft_deleted' filter whlie fetching instance list | 13:52 |
*** eliqiao has joined #openstack-nova | 13:52 | |
ndipanov | pczesno, hey | 13:52 |
ndipanov | u there? | 13:52 |
alex_xu | ankit_ag: status=delete work with this way https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/servers.py#L184 | 13:53 |
*** dkliban has left #openstack-nova | 13:53 | |
alex_xu | ankit_ag: does status=soft-delete can work with that way also? | 13:53 |
mtanino | Hi nova core team, Can I get a target of this bug? https://bugs.launchpad.net/nova/+bug/1377161 | 13:54 |
openstack | Launchpad bug 1377161 in OpenStack Compute (nova) "If volume-attach API is failed, Block Device Mapping record will remain" [Low,In progress] - Assigned to Mitsuhiro Tanino (mitsuhiro-tanino) | 13:54 |
ankit_ag | alex_xu: I'll check this and will come back to you. Thank you ! | 13:54 |
*** neelashah has joined #openstack-nova | 13:54 | |
*** Nikolay_St has quit IRC | 13:54 | |
alex_xu | ankit_ag: you are welcome, free to ping me if you still find some trouble with this way | 13:55 |
ankit_ag | alex_xu: status='deleted' is only allowed for admin user | 13:56 |
ankit_ag | but we need soft_deleted instances in case of delete api for normal user as well | 13:56 |
eliqiao | ankit_ag: alex_xu: I am not sure that will works, I remember that I have a patch to fix that, https://review.openstack.org/#/c/118641/1 | 13:58 |
ankit_ag | alex_xu: and if we do not add that admin context check, then normal user will be able to list all the soft deleted instances with this change | 13:58 |
*** artom has joined #openstack-nova | 13:59 | |
eliqiao | that admin check is explict, so I think soft-deleted instance can only be seen by admin , it is by designed. | 13:59 |
*** matrohon has joined #openstack-nova | 14:00 | |
*** rushil has joined #openstack-nova | 14:01 | |
ankit_ag | eliqiao: yup, so in my patch, we are not allowing normal user to see soft_deleted instances, but he should be able to delete that instance | 14:01 |
eliqiao | ankit_ag: there is a configure option in nova config file, it is a time slot, after sometime, the soft-deleted instances will be hard deleted by nova automaticly. | 14:02 |
bauzas | alaski: I think I got the idea | 14:02 |
bauzas | alaski: for the third instance being booted with the same token than the first one, it should enter that https://github.com/openstack/nova/blob/master/nova/api/ec2/cloud.py#L1367-L1374 | 14:03 |
ankit_ag | eliqiao: you are right, but what if you want to delete that instance immidiatly | 14:04 |
*** doron is now known as doron_afk | 14:04 | |
ankit_ag | currently it is allowed to call delete by instance-id, but by name it gives not found error | 14:04 |
*** Nikolay_St has joined #openstack-nova | 14:04 | |
*** nkrinner has quit IRC | 14:06 | |
*** eliqiao1 has joined #openstack-nova | 14:07 | |
eliqiao1 | ankit_ag: do you mean, it(soft-deleted) can be delete by uuid, but not by name? | 14:07 |
ankit_ag | eliqiao: yes, thats correct | 14:07 |
*** ujjain has joined #openstack-nova | 14:08 | |
eliqiao1 | ankit_ag: if that so , I think we need fix something. | 14:08 |
bauzas | alaski: got it | 14:08 |
ankit_ag | eliqiao: so I have submitted a patch in novaclient, for which we need to make these change in nova as well | 14:08 |
*** eliqiao has quit IRC | 14:08 | |
eliqiao1 | ankit_ag: cool, do you have a bug reported against it? | 14:09 |
ankit_ag | eliqiao: yes, let me get that for you | 14:09 |
eliqiao1 | ankit_ag: thanks :) | 14:10 |
*** zz_jgrimm is now known as jgrimm | 14:11 | |
ankit_ag | eliqiao: https://bugs.launchpad.net/nova/+bug/1436849 | 14:11 |
openstack | Launchpad bug 1436849 in OpenStack Compute (nova) "Inconsistent delete API in case of soft-delete" [Low,In progress] - Assigned to Ankit Agrawal (ankitagrawal) | 14:11 |
alaski | bauzas: what did it end up being? | 14:11 |
ankit_ag | eliqiao, alex_xu: I request your opinion on this issue | 14:11 |
bauzas | alaski: self.compute_api.get_all_system_metadata returns zero results | 14:12 |
ankit_ag | Thank you for your time and help :) | 14:12 |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Fix proxying finish_resize and resize_instance in _ComputeV4Proxy https://review.openstack.org/173897 | 14:12 |
dims | rajesht: see if this will help? ^^^ | 14:12 |
*** markvoelker_ has joined #openstack-nova | 14:13 | |
ndipanov | dansmith, https://review.openstack.org/#/c/173394/ | 14:13 |
ndipanov | it would be really good if we could slam this in so that I can propose it for RC2 | 14:13 |
bauzas | alaski: it should return one instance | 14:13 |
ndipanov | dansmith, and I have a feeling you will get what it is about very quickly ;) | 14:14 |
alaski | bauzas: right. did the system_metadata not get saved for some reason? | 14:14 |
*** abhishekk has quit IRC | 14:14 | |
bauzas | alaski: that's what I'm diggin | 14:14 |
bauzas | alaski: doesn't look to be overrided by the cells_api so that's just checking the API DB | 14:15 |
eliqiao1 | ankit_ag: yw, I will post some comments on the bug if I found something useful. | 14:15 |
dansmith | ndipanov: we just merged a thing to remove that None from the else clause, right bauzas ? | 14:15 |
alaski | bauzas: yeah, it's a bit confusing to me at this point | 14:15 |
dansmith | or was that a different thing? | 14:15 |
ndipanov | dansmith, yes | 14:15 |
ndipanov | I did that and broke it | 14:15 |
ndipanov | git blame it on me :) | 14:15 |
bauzas | dansmith: mmm ? | 14:15 |
bauzas | dansmith: missing context | 14:15 |
dansmith | bauzas: ndipanov's patch above | 14:15 |
bauzas | dansmith: yeah, that reminds me something, but I have to check | 14:16 |
*** markvoelker has quit IRC | 14:16 | |
*** ozialien has quit IRC | 14:17 | |
ndipanov | dansmith, bauzas https://review.openstack.org/#/c/169245/ | 14:17 |
bauzas | yeah that one | 14:17 |
*** tonytan4ever has joined #openstack-nova | 14:17 | |
ndipanov | hmmm I didn't actually remove the none there... | 14:18 |
ndipanov | ah right | 14:18 |
ndipanov | because we didn't even use it | 14:18 |
dansmith | no, that's not the one I was thinking of | 14:18 |
mtanino | dims: Hi, Can I get a target of this bug? I found your name on the bug. https://bugs.launchpad.net/nova/+bug/1377161 | 14:19 |
openstack | Launchpad bug 1377161 in OpenStack Compute (nova) "If volume-attach API is failed, Block Device Mapping record will remain" [Low,In progress] - Assigned to Mitsuhiro Tanino (mitsuhiro-tanino) | 14:19 |
ndipanov | because we stashed it on the filter | 14:19 |
*** ozialien has joined #openstack-nova | 14:19 | |
ndipanov | dansmith, anyway - there is an explanation why that None is needed in my comment | 14:19 |
*** achanda has joined #openstack-nova | 14:20 | |
dims | mtanino: there was a change in nova that got abandoned. | 14:20 |
bauzas | dansmith: oh you meant the sched client thing ? | 14:20 |
dansmith | I dunno, I'm doing too many things at once right now | 14:21 |
dims | mtanino: ah there's another one https://review.openstack.org/#/c/166695/ | 14:21 |
bauzas | yeah, we merged a lot of PCI changes recently... :/ | 14:21 |
BobBall | mriedem: Citrix CI is back working and commenting (clearly not voting); see https://review.openstack.org/#/c/168876/ (still currently the wrong format, sorry) | 14:21 |
*** irenab has joined #openstack-nova | 14:22 | |
dims | mtanino: what would you like the "Importance" field to be? | 14:22 |
dims | mtanino: can you please add some notes in the bug report so we can flip it? | 14:22 |
mriedem | BobBall: ok, are you asking that voting be turned back on? | 14:22 |
mtanino | dims: sure. The final patch is https://review.openstack.org/#/c/166695/ | 14:22 |
*** cfriesen has joined #openstack-nova | 14:23 | |
*** achanda has quit IRC | 14:24 | |
ankit_ag | eliqiao: That will be really helpful. Thank you | 14:24 |
mtanino | dims: I think the importance is medium because this bug causes unintentional volume detach. | 14:24 |
BobBall | mriedem: If you're happy to enable it, sure - but we won't use it yet. I want to leave it running for a day or so to check the pass rate and make sure there isn't another race lurking somewhere that'd cause more frustration. | 14:24 |
mriedem | BobBall: ok, that was my thinking | 14:25 |
dims | mtanino: done | 14:25 |
mriedem | BobBall: when you're comfortable with it please ping me or johnthetubaguy or mikal | 14:25 |
BobBall | Will do. | 14:25 |
openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Nova serial console access support https://review.openstack.org/145004 | 14:25 |
mtanino | dims: Thank you :) Is it possible to get merge this during Kilo or need to wait liberty1? | 14:25 |
*** ozialien has quit IRC | 14:26 | |
dims | mtanino: one of the cores will figure that out (i am not a core). it's probably late for Kilo if you ask me | 14:26 |
mtanino | dims: I see. Thanks. | 14:27 |
sdague | alaski / johnthetubaguy - so the gate back up is in part caused by 180 nodes stuck in deleting state in RAX. Any chance we could get someone to look at it | 14:27 |
mtanino | dims: you are in the nova bug team? | 14:27 |
dims | mtanino: see guidance from the team - http://markmail.org/message/ct45huk4cty6p2mx | 14:27 |
bauzas | ndipanov: dansmith: looking at your comment on https://review.openstack.org/#/c/173394/1/nova/scheduler/host_manager.py,cm | 14:28 |
johnthetubaguy | sdague: we should be able to get someone to look, what region are you seeing that? | 14:28 |
bauzas | ndipanov: dansmith: so you say that it never nullify the pci_requests field ? | 14:28 |
sdague | johnthetubaguy: see #openstack-infra, mordred is pinging you there with details | 14:28 |
mtanino | dims: Thank you for the link. | 14:28 |
*** Marga_ has joined #openstack-nova | 14:28 | |
bauzas | ndipanov: because of https://github.com/openstack/nova/blob/feef80c1f88cc28f8625b19ebf43b190fae0ca08/nova/pci/request.py#L193 ? | 14:28 |
openstackgerrit | Andrew Laski proposed openstack/nova: Store context in local store after spawn_n https://review.openstack.org/171299 | 14:29 |
*** browne has joined #openstack-nova | 14:29 | |
*** popw has quit IRC | 14:30 | |
alaski | sdague: looks like johnthetubaguy has it, but I can help push if necessary | 14:31 |
*** armax has joined #openstack-nova | 14:31 | |
*** ankit_ag has quit IRC | 14:32 | |
*** dboik_ has joined #openstack-nova | 14:32 | |
*** rushiagr is now known as rushiagr_away | 14:33 | |
*** achanda has joined #openstack-nova | 14:34 | |
*** emagana has joined #openstack-nova | 14:36 | |
*** dboik has quit IRC | 14:36 | |
*** markvoelker has joined #openstack-nova | 14:36 | |
*** markvoel_ has joined #openstack-nova | 14:38 | |
*** russellb has quit IRC | 14:38 | |
ndipanov | yes | 14:38 |
ndipanov | bauzas, | 14:38 |
ndipanov | that will always set an object on the instance | 14:38 |
ndipanov | now the ObjectList | 14:38 |
ndipanov | has __len__ defined | 14:39 |
*** hdd has quit IRC | 14:39 | |
ndipanov | meaning insance.pci_requests would be False | 14:39 |
ndipanov | in a boolean context | 14:39 |
ndipanov | but since we use the completely broken | 14:40 |
ndipanov | no no no let me rephrase that | 14:40 |
ndipanov | COMPLETELY EFFIN BROKEN | 14:40 |
ndipanov | obj_to_primitive | 14:40 |
*** markvoelker_ has quit IRC | 14:40 | |
bauzas | ndipanov: I think I understood you, it transforms it into a list of dicts ? | 14:40 |
ndipanov | it get's converted to {"instance_uuid": UUD} dict | 14:40 |
bauzas | ndipanov: orly? | 14:40 |
ndipanov | oyeah | 14:41 |
*** markvoel_ has quit IRC | 14:41 | |
ndipanov | which is Truthy | 14:41 |
ndipanov | it's just a mess | 14:41 |
*** russellb has joined #openstack-nova | 14:42 | |
*** markvoelker has quit IRC | 14:42 | |
bauzas | ndipanov: obj_to_primitive calls the objects primitives too | 14:42 |
bauzas | ndipanov: so that means that something is wrong in there | 14:42 |
openstackgerrit | Timofey Durakov proposed openstack/nova: Fixed nova-network dhcp-hostsfile update during live-migration https://review.openstack.org/173913 | 14:43 |
bauzas | ndipanov: it just flattens the nested objects into dicts with key and flatten value | 14:43 |
bauzas | ndipanov: the main problem I see with that is that obj_to_prim() != db_primitive | 14:44 |
*** achanda has quit IRC | 14:45 | |
*** otter768 has joined #openstack-nova | 14:46 | |
ndipanov | bauzas, well that could be one way to define it | 14:47 |
ndipanov | bauzas, for InstancePciRequests it would be {'instance_uuid': my_uuid, 'requests': []} | 14:48 |
ndipanov | which is very wrong in the context of an instance list | 14:48 |
ndipanov | er | 14:48 |
ndipanov | object list | 14:48 |
*** sald has quit IRC | 14:48 | |
*** dboik_ has quit IRC | 14:48 | |
bauzas | ndipanov: well, that's because the model is broken | 14:48 |
ndipanov | anyway | 14:49 |
bauzas | ndipanov: InstancePCIRequests is considered as one object, not a list | 14:49 |
ndipanov | acutally yes | 14:49 |
*** dboik has joined #openstack-nova | 14:49 | |
ndipanov | will correct myself | 14:49 |
bauzas | ndipanov: if it was a list of requests, it would be primitived into a list | 14:49 |
*** igordcard has quit IRC | 14:49 | |
ndipanov | lol that's another bug | 14:49 |
bauzas | I mean an ObjectList becomes a list of dicts | 14:50 |
ndipanov | if it happens to be a ObjectList | 14:50 |
ndipanov | yes | 14:50 |
ndipanov | but if you just have a list of objects (which is a field) | 14:50 |
ndipanov | then it would just stay objects | 14:50 |
*** igordcard has joined #openstack-nova | 14:50 | |
bauzas | you mean a nested objectlist as a field ? | 14:50 |
*** otter768 has quit IRC | 14:51 | |
ndipanov | bauzas, ListOfObjectsField but on a class that is not a subclass of ObjectList | 14:52 |
bauzas | ndipanov: IMHO https://github.com/openstack/nova/blob/master/nova/objects/instance_pci_requests.py | 14:52 |
bauzas | ndipanov: is wrong | 14:52 |
ndipanov | why is it wrong? | 14:52 |
bauzas | ndipanov: oh no no | 14:52 |
bauzas | my bad | 14:52 |
ndipanov | then NUMA is wrong since we do the same thing for cells | 14:52 |
*** hdd has joined #openstack-nova | 14:52 | |
bauzas | ndipanov: no no, the model is good | 14:53 |
bauzas | ndipanov: I missed the nested object as a fieldf | 14:53 |
ndipanov | so if you put it on an object that is not a list | 14:53 |
ndipanov | it will happily pass objects from obj_to_pirmitive | 14:53 |
*** annegentle has quit IRC | 14:54 | |
ndipanov | so you will have [<InstancePCiRequest at xxx>] for example | 14:54 |
ndipanov | now pass that to the serializer over RPC | 14:54 |
bauzas | ndipanov: mmm | 14:54 |
ndipanov | it will work because it will find objects and seralize them with their versions and maybe drop stuff | 14:54 |
ndipanov | when it tries to do compat | 14:55 |
bauzas | ndipanov: base.obj_to_primitive() on a InstancePCIRequests object will generate | 14:55 |
ndipanov | a dict with a uuid and a list of OBJECT instances | 14:55 |
bauzas | {'uuid': 'bla', 'requests': [{'count': 1, 'spec': 'foo'...}']} | 14:56 |
ndipanov | nope | 14:56 |
bauzas | https://github.com/openstack/nova/blob/master/nova/objects/base.py#L901-L906 | 14:57 |
ndipanov | yeah bauzas I know where it is - it's not how it works | 14:57 |
ndipanov | pull it up and try it | 14:57 |
bauzas | ndipanov: checking locally | 14:57 |
ndipanov | it will recursively call itself with a list | 14:58 |
ndipanov | which hits the last branch AKA get's returned as is | 14:58 |
*** kmartin has joined #openstack-nova | 14:58 | |
ndipanov | and you can't change that as you break all the codez | 14:58 |
ndipanov | that scheduler code actually expects it to work like that | 14:58 |
*** rwsu has joined #openstack-nova | 15:00 | |
ndipanov | anyway dansmith https://review.openstack.org/#/c/173394/ I'd much appreciate | 15:01 |
dansmith | ndipanov: when I can | 15:01 |
ndipanov | dansmith, but like 2015 right? :) | 15:01 |
*** jkraj has joined #openstack-nova | 15:02 | |
ndipanov | anyway - back to the other PCI disaster | 15:02 |
bauzas | ndipanov: damn | 15:03 |
ndipanov | bauzas, nice eh? | 15:04 |
ndipanov | so when I said - rip it out and burn it FIRST before anything else in the scheduler | 15:05 |
ndipanov | I was not kidding | 15:05 |
bauzas | ndipanov: yeah, that's not an instance of ObjectListBase, that's just a list... | 15:05 |
openstackgerrit | Chris Friesen proposed openstack/nova: libvirt: check qemu version for NUMA & hugepage support https://review.openstack.org/170780 | 15:05 |
*** xyang1 has joined #openstack-nova | 15:05 | |
*** TobiasE has quit IRC | 15:07 | |
mriedem | huzzah | 15:07 |
mriedem | compute manager cleanup_host shuts down the instance events handling | 15:07 |
*** thedodd has joined #openstack-nova | 15:07 | |
mriedem | we need to do the same for virt driver lifecycle events | 15:07 |
mriedem | easy peasy | 15:07 |
*** belmoreira has quit IRC | 15:08 | |
*** shyama has quit IRC | 15:10 | |
*** ijw has joined #openstack-nova | 15:10 | |
*** shyama has joined #openstack-nova | 15:11 | |
*** rushiagr_away is now known as rushiagr | 15:11 | |
*** markvoelker has joined #openstack-nova | 15:12 | |
*** c0m0 has quit IRC | 15:13 | |
openstackgerrit | Chris Friesen proposed openstack/nova: libvirt: check qemu version for NUMA & hugepage support https://review.openstack.org/170780 | 15:13 |
*** doron_afk is now known as doron | 15:14 | |
openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Nova serial console access support https://review.openstack.org/145004 | 15:16 |
*** markvoelker has quit IRC | 15:16 | |
*** dsanders has joined #openstack-nova | 15:17 | |
*** sbfox has joined #openstack-nova | 15:18 | |
*** c0m0 has joined #openstack-nova | 15:19 | |
openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Nova serial console access support https://review.openstack.org/145004 | 15:20 |
*** Nikolay_St has quit IRC | 15:20 | |
*** hdd has quit IRC | 15:20 | |
*** doron is now known as doron_afk | 15:20 | |
*** yingjun has joined #openstack-nova | 15:20 | |
*** yingjun has quit IRC | 15:22 | |
*** sbfox has quit IRC | 15:24 | |
*** sbfox has joined #openstack-nova | 15:25 | |
mriedem | cfriesen: are you still waiting on an answer on the pkvm stuff in https://review.openstack.org/#/c/170780/ ? | 15:25 |
mriedem | for _has_numa_support? | 15:25 |
*** ozialien has joined #openstack-nova | 15:26 | |
*** VW_ has quit IRC | 15:26 | |
mriedem | cfriesen: if so, i'd maybe leave the arch checks for ppc64 out and risk assume it doesn't work | 15:26 |
*** jcoufal_ has joined #openstack-nova | 15:26 | |
mriedem | cfriesen: you know it doesn't work with zkvm and you need the version constraint check on qemu | 15:26 |
mriedem | a separate pkvm patch could come later that adds ppc64 to supported_archs if pkvm does support those | 15:26 |
mriedem | jgrimm: krtaylor: ^ | 15:26 |
jgrimm | mriedem, agreed | 15:27 |
*** jamie_h has quit IRC | 15:28 | |
*** tonytan4ever has quit IRC | 15:29 | |
*** jcoufal has quit IRC | 15:29 | |
*** Marga_ has quit IRC | 15:30 | |
*** Marga_ has joined #openstack-nova | 15:30 | |
*** doron_afk has quit IRC | 15:31 | |
*** tonytan4ever has joined #openstack-nova | 15:32 | |
*** russellb has quit IRC | 15:32 | |
*** ajayaa has quit IRC | 15:34 | |
*** browne has quit IRC | 15:34 | |
*** sdake has joined #openstack-nova | 15:35 | |
*** VW_ has joined #openstack-nova | 15:35 | |
*** thomasem has quit IRC | 15:35 | |
*** russellb has joined #openstack-nova | 15:36 | |
*** dboik_ has joined #openstack-nova | 15:37 | |
openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Implement nova rescue https://review.openstack.org/127159 | 15:38 |
*** thomasem has joined #openstack-nova | 15:38 | |
cfriesen | mriedem: jgrimm: okay, will respin. | 15:38 |
*** Sukhdev has joined #openstack-nova | 15:38 | |
*** thomasem has quit IRC | 15:39 | |
mriedem | ndipanov: just read this https://bugzilla.redhat.com/show_bug.cgi?id=912284 - maybe we should update the config option help for that? | 15:39 |
openstack | bugzilla.redhat.com bug 912284 in openstack-nova "with resume_guests_state_on_host_boot=True rebooting host leaves VM's in Error state" [High,Closed: errata] - Assigned to beagles | 15:39 |
*** sdake_ has joined #openstack-nova | 15:39 | |
*** thomasem has joined #openstack-nova | 15:39 | |
mriedem | ndipanov: i.e. something along the lines of it's racy and might not work, so don't assume things will be happy if, for example, you're taking your compute host down for maintenance | 15:39 |
ndipanov | mriedem, hmm | 15:40 |
*** dboik has quit IRC | 15:40 | |
ndipanov | mriedem, I am trying to think of a case when that option is useful actually | 15:41 |
openstackgerrit | Chris Friesen proposed openstack/nova: libvirt: check qemu version for NUMA & hugepage support https://review.openstack.org/170780 | 15:41 |
*** matrohon has quit IRC | 15:42 | |
mriedem | ndipanov: looking at the libvirt driver and how it's handled, it appears it will only reboot the guest if it's down and the nova db thinks it's power state is running | 15:42 |
*** sdake has quit IRC | 15:42 | |
mriedem | ndipanov: but the option was added in diablo and races with the lifecycyle event handling added in grizzly | 15:42 |
ndipanov | mriedem, though this was a long time ago so things might have changed | 15:42 |
mriedem | ndipanov: yeah - i'm pulling some info together for a bug report. i see some things on ask.o.o about this option being set to True and it not really working | 15:43 |
mriedem | so the config option help doesn't really make it clear that it's not a cure all | 15:43 |
ndipanov | also we added some code to try to detect transition states and try to remedy that | 15:43 |
ndipanov | and I bet this was not tested with it as well | 15:43 |
cfriesen | mriedem: does it actually race? I thought the resume guest code ran really early at compute startup | 15:44 |
mriedem | cfriesen: the race is on shutdown | 15:44 |
mriedem | libvirt stops the guests which sends the stopped event to the compute manager which tries to call the stop api on the instance | 15:44 |
mriedem | while the compute service is doing down | 15:44 |
mriedem | so the instance vm_state/task_state can be wonky wrt the power_state on the hypervisor | 15:44 |
cfriesen | mriedem: ah, okay. maybe we should stop nova-compute prior to stopping the guests | 15:45 |
*** lpetrut has quit IRC | 15:45 | |
*** amotoki has joined #openstack-nova | 15:45 | |
mriedem | and you can hit cases where we actually do update the db power state before shutdown so nova thinks the instance is stopped, but when libvirt comes back up and restarts the guests, it sends the started event which nova gets and b/c the nova db thinks the instance should be stopped, it shuts it down | 15:45 |
mriedem | cfriesen: stopping the guests is handled asynchronously by libvirt | 15:46 |
mriedem | i'm going to push a change to stop handling lifecycle events from the driver when we're compute service is doing down, like we do with the neutron events listener | 15:46 |
cfriesen | mriedem: okay, so shutdown nova-compute before telling libvirt to shutdown? | 15:46 |
mriedem | cfriesen: well, if you do just an os reboot | 15:46 |
mriedem | that's not up to you really i guess, unless you are controlling the lifecycle processing of your services on the os | 15:47 |
cfriesen | mriedem: don't we still have a race if libvirt gets the shutdown signal before nova-compute does? | 15:48 |
mriedem | anyway, someone was freaking out on me because CONF.resume_guests_state_on_host_boot=True wasn't making their caviar dreams come true | 15:48 |
mriedem | cfriesen: yeah, possibly | 15:48 |
cfriesen | mriedem: I work on dedicated stuff...we always control the lifecycle. :) | 15:48 |
*** dboik_ has quit IRC | 15:48 | |
mriedem | it probably ultimately comes down to, if you're doing maintenance, plan for it and migrate the instances, | 15:49 |
*** dboik has joined #openstack-nova | 15:49 | |
mriedem | if shit crashes, you have to recover by resettting state and then start the instance, something like that | 15:49 |
cfriesen | mriedem: we had fun trying to get DOR working reasonably | 15:49 |
cfriesen | who's in charge of the Intel PCI CI? It seems to be failing. | 15:51 |
mriedem | baoli_: ^? | 15:51 |
*** garyk has joined #openstack-nova | 15:51 | |
baoli_ | mriedem: Hi | 15:52 |
ndipanov | PaulMurray, there? | 15:52 |
ndipanov | I found a very interesting issue backporting your patch to stable | 15:53 |
openstackgerrit | Chris Friesen proposed openstack/nova: libvirt: check qemu version for NUMA & hugepage support https://review.openstack.org/170780 | 15:53 |
cfriesen | mriedem: ^ commit message updated with arch info | 15:53 |
*** annegentle has joined #openstack-nova | 15:54 | |
cfriesen | baoli: Intel PCI CI seems to be failing with hardly any logs | 15:54 |
cfriesen | baoli_: http://198.175.100.33/170780/6/ | 15:55 |
PaulMurray | ndipanov, hi | 15:55 |
PaulMurray | ndipanov, whats the issue? | 15:56 |
ndipanov | hey so... I'm hitting this: http://logs.openstack.org/26/173226/3/check/gate-tempest-dsvm-large-ops/f136941/logs/screen-n-cond.txt.gz?level=INFO | 15:56 |
ndipanov | and am wondering how xome wi didn't hit that with kilo | 15:56 |
*** jwcroppe has quit IRC | 15:56 | |
PaulMurray | its this: 100000000000 | 15:56 |
PaulMurray | that number is too big for a db int filed | 15:57 |
*** jwcroppe has joined #openstack-nova | 15:57 | |
PaulMurray | your wondering how I spotted that :) | 15:57 |
ndipanov | yeah I know what it is | 15:57 |
PaulMurray | oh, ok :( | 15:57 |
ndipanov | but what I don't get is how come original patch wasnt hitting this | 15:57 |
PaulMurray | I fixed the big number -is it in the same patch? | 15:58 |
baoli_ | cfriesen: you can contact yonglihe_ for that, I think | 15:58 |
ndipanov | PaulMurray, you did?! | 15:59 |
ndipanov | show me :) | 15:59 |
PaulMurray | tyring to find the patch | 15:59 |
*** sdake has joined #openstack-nova | 15:59 | |
ndipanov | PaulMurray, awesome!! | 15:59 |
PaulMurray | the number never gets saved before because it is over-ridden by the calculations | 15:59 |
PaulMurray | so I changed it to be a sensible number - its disk size and is supposed to be in GB but the test has it in bytes | 16:00 |
*** vilobhmm1 has joined #openstack-nova | 16:00 | |
PaulMurray | looking - hold on | 16:00 |
*** unicell1 has joined #openstack-nova | 16:00 | |
ndipanov | PaulMurray, right - I got that | 16:00 |
ndipanov | and am wondering if that's what we actually want to be doing here as well | 16:01 |
ndipanov | i.e. create the record with no values and then re-do them | 16:01 |
*** Marga_ has quit IRC | 16:01 | |
*** unicell has quit IRC | 16:01 | |
PaulMurray | they need a value because the db fields are NOT NULL | 16:01 |
*** jwcroppe has quit IRC | 16:01 | |
ndipanov | ugh | 16:01 |
ndipanov | yes | 16:02 |
ndipanov | looking at it | 16:02 |
*** vilobhmm11 has joined #openstack-nova | 16:02 | |
PaulMurray | we looked at doing the calculation before saving first time, but it was no different from not doing the patch | 16:02 |
ndipanov | PaulMurray, exactly | 16:02 |
*** sdake_ has quit IRC | 16:03 | |
*** derekh has quit IRC | 16:03 | |
*** emagana has quit IRC | 16:03 | |
bauzas | alaski: quick q, still investigating the EC2 problem, do you know when the instance.cell_name is provided ? | 16:03 |
*** jistr has quit IRC | 16:03 | |
*** jistr has joined #openstack-nova | 16:04 | |
*** vilobhmm1 has quit IRC | 16:04 | |
*** FL1SK has quit IRC | 16:04 | |
ndipanov | PaulMurray, that's actually kind of bad | 16:05 |
*** jlanoux has quit IRC | 16:05 | |
ndipanov | PaulMurray, but if you can dig up the patch that'd be great | 16:06 |
mriedem | cfriesen: on the qemu version check change, i'm more or less +2, left some comments on how the tests gross me out though - we really need some decoupling in there and move that low-level host.has_min_version testing to test_host.py, but that could be a separate cleanup | 16:06 |
dansmith | bauzas: so are you good with ndipanov's patch? I'm just starting to look at it... | 16:07 |
PaulMurray | ndipanov, we had quite a bit of debate about it - don't like it myself, but thats life | 16:08 |
bauzas | dansmith: you mean the juno backport of PaulMurray ? | 16:09 |
bauzas | dansmith: yeah | 16:09 |
bauzas | dansmith: except that the large-ops job is saying no | 16:09 |
dansmith | bauzas: no the consume_from_instance one | 16:09 |
bauzas | dansmith: oh | 16:09 |
bauzas | dansmith: sorry, many stuff on my mind | 16:09 |
*** jistr has quit IRC | 16:09 | |
dansmith | bauzas: I missed the discussion between you guys, and thus I don't really understand the problem | 16:09 |
* bauzas still has to write a spec | 16:09 | |
bauzas | dansmith: oh | 16:09 |
*** emagana has joined #openstack-nova | 16:10 | |
bauzas | dansmith: well, the problem is that base.obj_to_prim() is buggy because a list is not an ObjectListBase :) | 16:10 |
*** _gryf is now known as _gryf|afk | 16:10 | |
ndipanov | PaulMurray, well I guess nova compute failing like that is actually pretty bad since there is no easy way for users to remedy it | 16:10 |
openstackgerrit | jichenjc proposed openstack/nova: Catch SnapshotNotFound exception at os-volumes https://review.openstack.org/173965 | 16:10 |
bauzas | dansmith: so it doesn't primitive it | 16:10 |
bauzas | dansmith: so yeah, basically, we need that hack IMHO | 16:10 |
ndipanov | but not sure what the solution would be other than set all the used ones to 0... | 16:11 |
*** jwcroppe has joined #openstack-nova | 16:11 | |
ndipanov | *_used that is | 16:11 |
ndipanov | dansmith, see my comment inline on the patch | 16:11 |
dansmith | ndipanov: yeah, I've read it | 16:11 |
cfriesen | mriedem: thanks...I agree that the unit tests are kind of gross. I basically did the minimum to get things working. | 16:11 |
dansmith | ndipanov: obj_to_primitive is not buggy, it's just that we use it when we shouldn't, yes? | 16:12 |
dansmith | I mean, it does what it's supposed to do | 16:12 |
dansmith | we just keep using it to hack around code that hasn't been converted yet because it's ugly and hard | 16:12 |
ndipanov | dansmith, well it should have never ever been used on anything that goes over rpc | 16:12 |
bauzas | dansmith: the real problem is that we primitive the first instance in order to get an instance_props field for the req_spec | 16:12 |
*** markvoelker has joined #openstack-nova | 16:13 | |
*** rushil has quit IRC | 16:13 | |
ndipanov | and req_spec is one of these | 16:13 |
mriedem | cfriesen: yeah, i understand | 16:13 |
bauzas | dansmith: instead, we should just trust RPC serialization and go ahead with passing an obj in the spec | 16:13 |
*** eliqiao1 has quit IRC | 16:13 | |
*** tiagogomes has left #openstack-nova | 16:13 | |
bauzas | dansmith: ndipanov: btw. https://review.openstack.org/#/c/173316/ | 16:13 |
*** dosaboy has quit IRC | 16:13 | |
ndipanov | plus it is pretty buggy - see what happens when an object has a ListOfObjects field and is not an instance of ListBase | 16:13 |
bauzas | dansmith: ndipanov: ^ RequestSpec object spec again... | 16:14 |
*** unicell1 has quit IRC | 16:14 | |
*** markus_z has quit IRC | 16:14 | |
*** dosaboy has joined #openstack-nova | 16:14 | |
ndipanov | bauzas, added myself | 16:14 |
dansmith | ndipanov: so, on line 265, if we just changed it to this: | 16:14 |
dansmith | ndipanov: pci_requests = instance.get(pci_requests) or None | 16:14 |
bauzas | ndipanov: cool | 16:14 |
dansmith | wouldn't that do what we want? | 16:15 |
*** jistr has joined #openstack-nova | 16:15 | |
ndipanov | dansmith, no | 16:15 |
bauzas | ndipanov: no | 16:15 |
bauzas | oop | 16:15 |
bauzas | dansmith: no | 16:15 |
bauzas | dansmith: the problem is that this is an object | 16:15 |
bauzas | not an empty list or dict | 16:15 |
cfriesen | Procedural question...I posted a couple of specs at https://review.openstack.org/#/c/168982/ and https://review.openstack.org/#/c/169026/ but have gotten very little feedback. Is that a sign or is it just early days yet? | 16:16 |
dansmith | bauzas: you mean we actually have an instance dict with a real object inside it? | 16:16 |
bauzas | cfriesen: eh, wait a couple more :) | 16:16 |
*** irenab has quit IRC | 16:16 | |
bauzas | dansmith: exactly | 16:16 |
dansmith | ah, okay | 16:16 |
ndipanov | dansmith, no | 16:16 |
ndipanov | we have an instance dict | 16:16 |
*** annegentle has quit IRC | 16:16 | |
ndipanov | that has a pci_requests field | 16:16 |
ndipanov | that is an object | 16:16 |
ndipanov | er | 16:16 |
ndipanov | a dict | 16:16 |
ndipanov | such as | 16:17 |
bauzas | ndipanov: yeah, that's what I said | 16:17 |
dansmith | huh? | 16:17 |
openstackgerrit | jichenjc proposed openstack/nova: Catch SnapshotNotFound exception at os-volumes https://review.openstack.org/173965 | 16:17 |
bauzas | >>> objects.base.obj_to_primitive(pci_mult) | 16:17 |
bauzas | {'instance_uuid': '48dc327e-4597-4767-8eaf-61e640455799', 'requests': [InstancePCIRequest(alias_name=<?>,count=1,is_new=<?>,request_id=<?>,spec=<?>)]} | 16:17 |
dansmith | you two just said different htings | 16:17 |
ndipanov | that ^^ | 16:17 |
ndipanov | that is obj_to_primitive | 16:17 |
bauzas | so, I was unclear | 16:17 |
*** irenab has joined #openstack-nova | 16:17 | |
*** jistr has quit IRC | 16:17 | |
bauzas | primitiving InstancePCIRequests makes this | 16:18 |
ndipanov | and to make matter worse | 16:18 |
ndipanov | there is a lot of code that actually relies on this | 16:18 |
dansmith | okay, what bauzas showed is what I was asking, but ndipanov you said dict of dict | 16:18 |
*** arnaud___ has joined #openstack-nova | 16:18 | |
ndipanov | well this is inside the instance_ref dict | 16:18 |
mriedem | russellb: vishy: garyk: pixelb: nova stable cores, can you take a look at this squashed backport? there are several bugs lined up around this racy code going back to at least icehouse: https://review.openstack.org/#/c/163378/ | 16:18 |
*** dboik_ has joined #openstack-nova | 16:18 | |
dansmith | ndipanov: right, that's what I said | 16:18 |
mriedem | and i have a change on master for kvm that depends on these to backport | 16:18 |
ndipanov | so it's {... 'pci_requests': that} | 16:18 |
dansmith | ndipanov: it's an instance dict with a non-primitive (real object) element | 16:19 |
dansmith | we think it's dicts all the way down, but it's not | 16:19 |
bauzas | dansmith: that's not the instance dict | 16:19 |
ndipanov | right | 16:19 |
mriedem | russellb: vishy: garyk: pixelb: this is the kvm change i plan on backporting https://review.openstack.org/#/c/172775/ | 16:19 |
bauzas | dansmith: instance['pci_requests'] is that | 16:19 |
ndipanov | same happens for any Object that has a ListOfObjectss field | 16:19 |
*** markvoelker has quit IRC | 16:19 | |
dansmith | bauzas: yeah | 16:19 |
dansmith | ndipanov: and we're not fixing obj_to_primitive() because why? | 16:19 |
*** lpetrut has joined #openstack-nova | 16:19 | |
alaski | bauzas: instance.cell_name is set in instance_update_at_top. | 16:20 |
bauzas | alaski: then I'm lost :) | 16:20 |
ndipanov | dansmith, because it changes what goes over RPC | 16:20 |
*** dave-mccowan has joined #openstack-nova | 16:20 | |
ndipanov | if we fix it | 16:20 |
*** browne has joined #openstack-nova | 16:20 | |
bauzas | alaski: that's around 3 hours I'm following the path, and my eyes are blinking :D | 16:20 |
ndipanov | fix it without changing scheduler code | 16:21 |
dansmith | ndipanov: so, if I may take a detour: | 16:21 |
dansmith | ndipanov: all this stuff has been broken like a hundred times in the last month | 16:21 |
ndipanov | yes | 16:21 |
ndipanov | approx | 16:21 |
dansmith | ndipanov: why should we continue to fix any of it until we have some functional tests to show that it works? | 16:21 |
bauzas | dansmith: I think we can safely assume this as a slow price, and fix the obj_to_primitive() stuff on a later patch, because then it could make some side-effect | 16:21 |
dansmith | I mean really, we're just playing whack-a-mole and it's ridiculous | 16:21 |
*** dboik has quit IRC | 16:22 | |
bauzas | dansmith: eh, come playing with cells job | 16:22 |
ndipanov | dansmith, well because there's companies that pay people outrages salaries to work on this shit | 16:22 |
dansmith | bauzas: the cells job at least has tests to show when it's working and not | 16:22 |
bauzas | :) | 16:22 |
ndipanov | and then we tell them to go screw themselves | 16:22 |
ndipanov | so I say we take the middle road | 16:22 |
bauzas | dansmith: just an expression of my frustration to not quickly fix those bugs :) | 16:22 |
ndipanov | and try to make it not completely broken | 16:22 |
ndipanov | while we agree that it's bad | 16:22 |
bauzas | agreed | 16:22 |
*** emagana has quit IRC | 16:22 | |
ndipanov | and also add tests | 16:22 |
bauzas | we need to move carefully with PCI | 16:23 |
PaulMurray | ndipanov, found it ! | 16:23 |
ndipanov | PaulMurray, \o/ | 16:23 |
alaski | bauzas: okay, I'll work on reproducing locally and see what's going on | 16:23 |
dansmith | but this stuff is >1cycle old and it still has so little testing that it's breaking this often | 16:23 |
PaulMurray | ndipanov, my fix for that big number went away in a rebase because sean landed https://review.openstack.org/#/c/165376/ | 16:23 |
dansmith | so when do we do that? | 16:23 |
*** mpavone has quit IRC | 16:23 | |
ndipanov | dansmith, well as I pointed out on one of the reviews that fixes the breakage | 16:24 |
bauzas | alaski: still continuing to work on, basically looking at the DB API why it doesn't find the same metadata instance | 16:24 |
ndipanov | of PCI | 16:24 |
ndipanov | it was actually broken by | 16:24 |
ndipanov | object fuckery (as I put it) | 16:24 |
PaulMurray | ndipanov, the number comes from nova/virt/fake.py #L113 before his change | 16:24 |
ndipanov | so... the answer is - like my relationship status on FB | 16:24 |
bauzas | ndipanov: I also helped to break the PCI stuff | 16:24 |
ndipanov | "it's complicated" | 16:24 |
alaski | bauzas: okay. that's my big question. instance.system_metadata = something; instance.save() should work regardles of cell_name being set | 16:24 |
bauzas | alaski: you got it | 16:25 |
PaulMurray | ndipanov, my fix just changed the definition to be in GB | 16:25 |
*** dave-mccowan has quit IRC | 16:25 | |
dansmith | ndipanov: bauzas: breaking it is understandable, but continuing to patch it without tests is not, IMHO | 16:25 |
bauzas | dansmith: hence my lookaheads to the PCI CI | 16:25 |
*** dave-mccowan has joined #openstack-nova | 16:25 | |
* bauzas wonders if it's grammatically correct | 16:25 | |
*** welldannit has quit IRC | 16:25 | |
ndipanov | pci + numa is really from February | 16:26 |
dansmith | bauzas: the problem is, PCI CI is complicated, but I don't think we need full environments to test some of this level of stupid brokenness | 16:26 |
ndipanov | so it's not really been there for >1 cycle | 16:26 |
bauzas | dansmith: agreed, it could be identified with mockups | 16:26 |
dansmith | there should be reasonable functional tests in our tree to protect against *this* level of broken | 16:26 |
bauzas | dansmith: high debt IMHO | 16:27 |
bauzas | dansmith: but sounds like it could be identified as a prio ? | 16:27 |
dansmith | I surely hope so | 16:27 |
bauzas | or part of an existing prio, like CI | 16:28 |
dansmith | on this specific patch, there are no tests that actually verify the change, right? just a change to make it continue to work | 16:28 |
dansmith | right ndipanov ? | 16:28 |
*** ociuhandu has quit IRC | 16:28 | |
mriedem | russellb: pixelb: garyk: vishy: adam_g: stable/juno is also race failing w/o this patch: https://review.openstack.org/#/c/170295/ | 16:28 |
ndipanov | dansmith, let me check | 16:29 |
*** jecarey has quit IRC | 16:29 | |
*** rajesht has quit IRC | 16:29 | |
mriedem | can someone just throw me on nova-stable-maint please? :P | 16:29 |
mriedem | dansmith: ^ | 16:29 |
ndipanov | dansmith, that is correct yes | 16:29 |
ndipanov | unit tests could have caught this actually | 16:30 |
bauzas | mriedem: oh man, lots of pain with the cells job related to the Tempest test you're mentioning | 16:30 |
*** sudipto has joined #openstack-nova | 16:30 | |
garyk | mriedem: done | 16:31 |
mriedem | bauzas: cells shouldn't be running on stable/juno | 16:31 |
mriedem | garyk: thanks | 16:31 |
bauzas | mriedem: yeah but cells were impacted by that Tempest change too | 16:31 |
bauzas | anyway | 16:32 |
*** emagana has joined #openstack-nova | 16:32 | |
*** annashen has joined #openstack-nova | 16:32 | |
*** Mike_D_laptop has quit IRC | 16:32 | |
*** dave-mccowan has quit IRC | 16:32 | |
*** dave-mccowan has joined #openstack-nova | 16:33 | |
openstackgerrit | Sudipta Biswas proposed openstack/nova: Allow discrete blocks of online pCPUs for pinning https://review.openstack.org/173187 | 16:33 |
*** shaohe_feng has joined #openstack-nova | 16:33 | |
ndipanov | PaulMurray, not sure hwo this patch fixes it | 16:34 |
*** Mike_D_laptop has joined #openstack-nova | 16:36 | |
*** annashen has quit IRC | 16:37 | |
*** annashen has joined #openstack-nova | 16:37 | |
sudipto | mriedem, can ask for a review: https://review.openstack.org/173187 ? | 16:38 |
*** EmilienM is now known as EmilienM|afk | 16:38 | |
PaulMurray | ndipanov, sorry, lost track of converstaion - do you mean you are not sure how sean's patch fixes it? | 16:39 |
*** Mike_D_laptop has quit IRC | 16:40 | |
openstackgerrit | jichenjc proposed openstack/nova: Catch OverQuota in volume create function https://review.openstack.org/173985 | 16:40 |
*** c0m0 has quit IRC | 16:40 | |
ndipanov | yeah | 16:41 |
ndipanov | PaulMurray, ^ | 16:41 |
*** Marga_ has joined #openstack-nova | 16:42 | |
*** arnaud___ has quit IRC | 16:42 | |
*** arnaud___ has joined #openstack-nova | 16:43 | |
*** unicell has joined #openstack-nova | 16:43 | |
*** shyama has quit IRC | 16:43 | |
*** markvoelker has joined #openstack-nova | 16:44 | |
PaulMurray | ndipanov, nova/virt/fake.py has the driver used in the tests. The resource tracker gets resource values from it in update_available_resources() | 16:45 |
PaulMurray | ndipanov, sean's patch changes the fake so that initially the values are different - so the big number doesn't occur any more | 16:45 |
ndipanov | PaulMurray, ah so the big number wasn't hardcoded | 16:46 |
*** unicell has quit IRC | 16:46 | |
*** unicell1 has joined #openstack-nova | 16:46 | |
ndipanov | this patch makes the tracking actually happen from instances right | 16:46 |
*** romainh has left #openstack-nova | 16:46 | |
ndipanov | PaulMurray, never mind I'm blind | 16:47 |
ndipanov | I need to stop working | 16:47 |
ndipanov | thanks a lot for this! | 16:47 |
PaulMurray | ndipanov, before his patch it was hard coded - sean's change makes the fake give values based on instances - yes, your right | 16:47 |
*** arnaud___ has quit IRC | 16:47 | |
*** otter768 has joined #openstack-nova | 16:47 | |
PaulMurray | ndipanov, np - glad to help | 16:47 |
cfriesen | sudipto: is it ever valid to have online_pcpus be empty? | 16:47 |
sudipto | cfriesen, Good question. I don't know if the system would be functional in that case? (Just wondering) | 16:48 |
ndipanov | PaulMurray, I'd still prefer to set those to 0 TBH but that's another topic | 16:48 |
*** zzzeek has joined #openstack-nova | 16:48 | |
PaulMurray | ndipanov, yes, I agree its not ideal | 16:49 |
cfriesen | sudipto: also, if available_ids is more than online_pcpus that would seem like a functional bug. The operator has told nova to use certain cpus and then offlined one or more of them. | 16:49 |
bauzas | alaski: so, now I haven't changed anything and now the problem is not reproduced... | 16:49 |
bauzas | alaski: I can't no longer get a failure | 16:49 |
bauzas | alaski: so, that's racy | 16:50 |
*** markvoelker has quit IRC | 16:50 | |
sudipto | cfriesen, You mean the validity of the bug itself? | 16:50 |
alaski | bauzas: yeah. it went away in Jenkins for a while too | 16:51 |
sudipto | cfriesen, like available ids is [2, 3, 4] and online_pcpus is [2, 3] coz 4 got offline? | 16:52 |
*** otter768 has quit IRC | 16:52 | |
*** Mike_D_laptop has joined #openstack-nova | 16:52 | |
sudipto | cfriesen, and at that time if nova-compute is restarted, total_pcpus count would be 2. and the older code would kick in (the code which was there before this review) - and throw an invalid error. | 16:53 |
cfriesen | sudipto: whoops...misread it. what does "available_ids <= online_pcpus" actually do for two lists? | 16:53 |
*** irenab has quit IRC | 16:54 | |
sudipto | cfriesen, these are sets - and hence we use the <= to determine if available_ids is a subset equal to the online_pcpus set | 16:54 |
cfriesen | sudipto: for the existing code should we be checking against online_pcpus instead of total_pcpus? | 16:57 |
*** raildo has quit IRC | 16:58 | |
sudipto | cfriesen, yeah that's what motivated me to file this bug, but then every host is not meant to support NUMA topology - and hence i had to retain the old behavior as well. | 16:58 |
cfriesen | sudipto: or is self._conn.getInfo()[2] only reporting online cpus? Also, isn't online/offline separate from NUMA? | 17:00 |
*** ozialien has quit IRC | 17:00 | |
sudipto | cfriesen, self._conn.getInfo()[2] is only returning a number - for the total number of pcpus. | 17:00 |
cfriesen | sudipto: assuming it maps to https://libvirt.org/html/libvirt-libvirt-host.html#virNodeInfo then it's actually reporting the number of "active" CPUs. | 17:01 |
sudipto | cfriesen, right, it's reporting the "number" of active CPUs - but let's say your pcpu numbers which are online are discrete in nature. For example : [5, 55, 89] | 17:02 |
cfriesen | sudipto: so my concern about total vs online is already dealt with. | 17:02 |
*** unicell1 has quit IRC | 17:02 | |
cfriesen | sudipto: yes, still need to deal with discontiguous CPUs | 17:02 |
*** unicell has joined #openstack-nova | 17:02 | |
sudipto | cfriesen, and let's say you specify vcpus as: 5, 55, 89 in nova.conf, but the pcpus older number would return 3 and it nova-compute would refuse to start... | 17:03 |
sudipto | cfriesen, yeah | 17:03 |
cfriesen | sudipto: maybe change the commit message to " | 17:04 |
cfriesen | If the IDs form a subset | 17:04 |
cfriesen | of the total available pcpus on the host, pinning is allowed. | 17:04 |
cfriesen | sudipto: bah, stupid paste....I meant "If the IDs form a subste of the online pCPUs on the host..." | 17:04 |
sudipto | cfriesen, change this line to the one you suggested: " if the | 17:05 |
sudipto | cpuset ID exceeds the total number of online pcpus. | 17:05 |
sudipto | "? | 17:05 |
*** raildo has joined #openstack-nova | 17:06 | |
cfriesen | sudipto: the current line has "...form a subset of the total available pCPUs...", I'm suggesting that you be explicit that you are checking against "online" pCPUs. (instead of "available") | 17:07 |
*** tonytan4ever has quit IRC | 17:07 | |
sudipto | cfriesen, ah yeah - makes sense. Let me change that right away. | 17:08 |
*** dboik_ has quit IRC | 17:08 | |
*** dboik has joined #openstack-nova | 17:08 | |
*** amotoki has quit IRC | 17:10 | |
sudipto | cfriesen, check now? | 17:10 |
sudipto | bauzas, I wrote the UTs you asked for. | 17:11 |
sudipto | https://review.openstack.org/#/c/173187/ | 17:11 |
*** eglynn has quit IRC | 17:13 | |
*** adalbas has joined #openstack-nova | 17:13 | |
*** ndipanov has quit IRC | 17:15 | |
*** sudipto has quit IRC | 17:16 | |
*** apuimedo has quit IRC | 17:17 | |
*** deepakcs has joined #openstack-nova | 17:19 | |
*** lucasagomes has quit IRC | 17:20 | |
deepakcs | Hi all, can someone take a look at https://review.openstack.org/168805 and provide reviews pls | 17:21 |
deepakcs | dansmith, ^ | 17:21 |
*** jichen has quit IRC | 17:22 | |
*** aix has quit IRC | 17:22 | |
*** asselin has joined #openstack-nova | 17:23 | |
dansmith | deepakcs: you haven't replied to my comments | 17:24 |
*** ijw has quit IRC | 17:25 | |
*** bkopilov has quit IRC | 17:26 | |
*** ijw has joined #openstack-nova | 17:26 | |
*** ijw has quit IRC | 17:27 | |
*** ijw has joined #openstack-nova | 17:28 | |
*** sudipto has joined #openstack-nova | 17:29 | |
*** bkopilov has joined #openstack-nova | 17:30 | |
sudipto | bauzas, Not sure if i got disconnected in between, i just wanted to mention to you that I have added the UTs per your comment for the review https://review.openstack.org/#/c/173187/ | 17:32 |
*** hdd has joined #openstack-nova | 17:32 | |
*** sahid has quit IRC | 17:32 | |
*** ajayaa has joined #openstack-nova | 17:32 | |
deepakcs | dansmith, sorry, i didn't get notification for that :) | 17:34 |
dansmith | deepakcs: the string freeze concerns are now moot of course | 17:34 |
deepakcs | dansmith, https://review.openstack.org/#/c/168805/5/nova/virt/libvirt/driver.py (what excatly u refer to by remove whitespace ?) | 17:34 |
dansmith | deepakcs: but it's always nice to see if you have comments before begging for reviews in here :) | 17:34 |
deepakcs | dansmith, yes, i was sure i didn't since somehow the notification didn't come / maybe i missed it. my bad | 17:35 |
dansmith | deepakcs: the leading whitespace between the # and the first character.. that's just crazy | 17:35 |
*** achanda has joined #openstack-nova | 17:35 | |
*** tonytan4ever has joined #openstack-nova | 17:36 | |
deepakcs | dansmith, ok, i thought that space was needed, if not, will remove it in next patch | 17:36 |
deepakcs | dansmith, thanks for ur review, will send next patch soon | 17:36 |
*** harlowja_away is now known as harlowja | 17:36 | |
*** sbfox has quit IRC | 17:36 | |
*** deepakcs has quit IRC | 17:37 | |
*** Nic has joined #openstack-nova | 17:37 | |
*** Sukhdev has quit IRC | 17:38 | |
*** sudipto has quit IRC | 17:39 | |
*** bkopilov has quit IRC | 17:39 | |
*** bkopilov has joined #openstack-nova | 17:39 | |
*** cbits has joined #openstack-nova | 17:43 | |
*** bkopilov has quit IRC | 17:44 | |
*** moshele has quit IRC | 17:45 | |
*** sbfox has joined #openstack-nova | 17:46 | |
*** signed8bit has joined #openstack-nova | 17:47 | |
*** markvoelker has joined #openstack-nova | 17:47 | |
*** johnthetubaguy is now known as zz_johnthetubagu | 17:48 | |
*** gszasz has quit IRC | 17:48 | |
*** marun has joined #openstack-nova | 17:49 | |
*** Marga_ has quit IRC | 17:49 | |
*** Marga_ has joined #openstack-nova | 17:49 | |
*** erlon has joined #openstack-nova | 17:50 | |
*** jecarey has joined #openstack-nova | 17:50 | |
*** Longgeek has quit IRC | 17:51 | |
*** adalbas has quit IRC | 17:51 | |
*** markvoelker has quit IRC | 17:52 | |
*** Mike_D_laptop has quit IRC | 17:52 | |
*** mpaolino has joined #openstack-nova | 17:52 | |
*** shaohe_feng1 has joined #openstack-nova | 17:54 | |
*** jecarey has quit IRC | 17:55 | |
cfriesen | deepakcs: I think the suggestion is to remove the multiple spaces, leaving just a single space. | 17:57 |
*** annashen has quit IRC | 17:57 | |
*** shaohe_feng has quit IRC | 17:57 | |
cfriesen | whoops, he's gone | 17:57 |
*** sbfox has quit IRC | 17:57 | |
*** rfolco has quit IRC | 17:57 | |
*** andreykurilin has joined #openstack-nova | 17:58 | |
*** openstackstatus has quit IRC | 17:58 | |
*** openstackstatus has joined #openstack-nova | 18:00 | |
*** ChanServ sets mode: +v openstackstatus | 18:00 | |
*** apuimedo has joined #openstack-nova | 18:00 | |
*** FL1SK has joined #openstack-nova | 18:02 | |
*** ozialien has joined #openstack-nova | 18:02 | |
*** annashen has joined #openstack-nova | 18:02 | |
*** sbfox has joined #openstack-nova | 18:03 | |
*** ijw has quit IRC | 18:03 | |
*** cbits has quit IRC | 18:03 | |
*** patrickeast has joined #openstack-nova | 18:03 | |
-openstackstatus- NOTICE: Gerrit has stopped emitting events so Zuul is not alerted to changes. We will restart Gerrit shortly to correct the problem. | 18:03 | |
*** ChanServ changes topic to "Gerrit has stopped emitting events so Zuul is not alerted to changes. We will restart Gerrit shortly to correct the problem." | 18:03 | |
*** ijw has joined #openstack-nova | 18:03 | |
*** signed8bit is now known as signed8bit_ZZZzz | 18:04 | |
*** annegentle has joined #openstack-nova | 18:04 | |
*** ijw has quit IRC | 18:04 | |
*** ijw has joined #openstack-nova | 18:05 | |
*** signed8bit_ZZZzz is now known as signed8bit | 18:05 | |
*** mpaolino has quit IRC | 18:09 | |
*** MVenesio has joined #openstack-nova | 18:13 | |
mriedem | i can't live without gerrit | 18:13 |
*** dosaboy has quit IRC | 18:13 | |
*** rfolco has joined #openstack-nova | 18:14 | |
*** dosaboy has joined #openstack-nova | 18:14 | |
*** Marga_ has quit IRC | 18:15 | |
*** apuimedo has quit IRC | 18:16 | |
*** VW_ has quit IRC | 18:18 | |
*** dosaboy has quit IRC | 18:20 | |
*** dosaboy has joined #openstack-nova | 18:21 | |
*** EmilienM|afk is now known as EmilienM | 18:22 | |
*** dosaboy has quit IRC | 18:22 | |
*** dosaboy has joined #openstack-nova | 18:24 | |
*** dosaboy has quit IRC | 18:24 | |
*** dosaboy has joined #openstack-nova | 18:24 | |
*** rajesht has joined #openstack-nova | 18:26 | |
*** tonytan4ever has quit IRC | 18:26 | |
*** ChanServ changes topic to "This channel is for Nova development. For support of Nova deployments, please use #openstack. -=- master is open for Lemming development" | 18:26 | |
-openstackstatus- NOTICE: Gerrit has been restarted. New patches, approvals, and rechecks between 17:30 and 18:20 UTC may have been missed by Zuul and will need rechecks or new approvals added. | 18:26 | |
*** signed8bit is now known as signed8bit_ZZZzz | 18:28 | |
*** Nikolay_St has joined #openstack-nova | 18:30 | |
openstackgerrit | Michael Bayer proposed openstack/nova: Add an index to virtual_interfaces.uuid https://review.openstack.org/154243 | 18:32 |
*** annegentle has quit IRC | 18:35 | |
*** annegentle has joined #openstack-nova | 18:35 | |
openstackgerrit | Michael Bayer proposed openstack/nova: Add an index to virtual_interfaces.uuid https://review.openstack.org/174018 | 18:35 |
*** eharney has quit IRC | 18:36 | |
*** andreykurilin has quit IRC | 18:36 | |
openstackgerrit | Michael Bayer proposed openstack/nova: Add an index to virtual_interfaces.uuid https://review.openstack.org/174018 | 18:37 |
*** andreykurilin has joined #openstack-nova | 18:40 | |
*** rajesht has quit IRC | 18:40 | |
*** matrohon has joined #openstack-nova | 18:42 | |
*** tonytan4ever has joined #openstack-nova | 18:43 | |
*** jcoufal_ has quit IRC | 18:45 | |
*** rushil has joined #openstack-nova | 18:46 | |
*** Marga_ has joined #openstack-nova | 18:46 | |
*** otter768 has joined #openstack-nova | 18:48 | |
*** sbfox has quit IRC | 18:49 | |
*** Nic1 has joined #openstack-nova | 18:51 | |
*** Marga_ has quit IRC | 18:51 | |
*** Nic has quit IRC | 18:51 | |
*** andreykurilin has quit IRC | 18:52 | |
*** otter768 has quit IRC | 18:53 | |
*** hdd has quit IRC | 18:53 | |
*** matrohon has quit IRC | 18:54 | |
*** baoli_ has quit IRC | 18:56 | |
*** ijw has quit IRC | 18:57 | |
*** VW_ has joined #openstack-nova | 19:00 | |
*** pixelb has quit IRC | 19:01 | |
*** samueldmq has joined #openstack-nova | 19:02 | |
openstackgerrit | Michael Bayer proposed openstack/nova: Add an index to virtual_interfaces.uuid https://review.openstack.org/154243 | 19:03 |
samueldmq | hi, how long approximately nova unit tests take to run ? | 19:04 |
*** annashen has quit IRC | 19:04 | |
*** eharney has joined #openstack-nova | 19:04 | |
*** jichen has joined #openstack-nova | 19:06 | |
*** matrohon has joined #openstack-nova | 19:07 | |
*** ijw has joined #openstack-nova | 19:08 | |
dansmith | samueldmq: depends entirely on your machine | 19:09 |
dansmith | i.e. could be 5 minutes and could be 30 | 19:09 |
*** ijw has quit IRC | 19:13 | |
*** stevemar has joined #openstack-nova | 19:17 | |
*** jaypipes has quit IRC | 19:18 | |
*** gszasz has joined #openstack-nova | 19:20 | |
*** jaypipes has joined #openstack-nova | 19:21 | |
*** jwcroppe has quit IRC | 19:21 | |
*** jwcroppe has joined #openstack-nova | 19:22 | |
*** dboik_ has joined #openstack-nova | 19:25 | |
*** jwcroppe has quit IRC | 19:26 | |
*** Nikolay_St has quit IRC | 19:27 | |
*** ijw has joined #openstack-nova | 19:28 | |
*** dboik has quit IRC | 19:29 | |
*** Nikolay_St has joined #openstack-nova | 19:29 | |
*** oro has joined #openstack-nova | 19:30 | |
*** david-lyle has joined #openstack-nova | 19:30 | |
*** ijw_ has joined #openstack-nova | 19:30 | |
*** VW_ has quit IRC | 19:30 | |
*** armax has quit IRC | 19:31 | |
samueldmq | dansmith, yeah I know, but for example, for you, how long it takes ? | 19:32 |
dansmith | samueldmq: it takes like 300s for me right now, IIRC | 19:32 |
*** ijw has quit IRC | 19:32 | |
samueldmq | dansmith, k, I am asking because I get curious after reading a post from sdague at https://dague.net/ | 19:33 |
samueldmq | dansmith, 'testr discovery takes 20 - 30s on the Nova source tree' | 19:33 |
sdague | samueldmq: I'm at about 2 minutes on my box | 19:33 |
*** ijw has joined #openstack-nova | 19:33 | |
sdague | maybe 3 | 19:33 |
* dansmith does a run | 19:33 | |
samueldmq | dansmith, k thx | 19:33 |
samueldmq | sdague, hi - I was wondering how useful/appreciated a tool like https://folding.stanford.edu/ would be for openstack | 19:34 |
*** VW_ has joined #openstack-nova | 19:34 | |
sdague | samueldmq: in what context? | 19:34 |
samueldmq | sdague, basically people register and their unit tests run distributed on all the registered machines | 19:34 |
sdague | samueldmq: we already have a giant CI pipeline | 19:35 |
samueldmq | sdague, unit tests which are already run in parallel | 19:35 |
sdague | http://status.openstack.org/zuul/ | 19:35 |
samueldmq | sdague, yeah but people run tests locally some times before submitting to jenkins, right? | 19:35 |
*** Hosam has joined #openstack-nova | 19:35 | |
sdague | samueldmq: sure | 19:35 |
sdague | do you expect they would get faster than 5 minute turn around from that? | 19:36 |
*** oro has quit IRC | 19:36 | |
samueldmq | sdague, and that should be fun: "be part of the infra, yes, yourself" | 19:36 |
dansmith | yeah, I'm at 3.5 minutes on my quadcore box | 19:36 |
samueldmq | sdague, yes I do think | 19:36 |
samueldmq | sdague, registered ppl would have the .tox env up to date | 19:36 |
*** ijw_ has quit IRC | 19:36 | |
samueldmq | sdague, each computer would get a list of tests to run, and return the results | 19:37 |
sdague | samueldmq: it seems like a lot of effort, for not a ton of gain. But feel free to try :) | 19:37 |
samueldmq | sdague, yeah, actually I am not convinced it's worth it | 19:37 |
samueldmq | sdague, but I am sure it would bring fun :p | 19:37 |
samueldmq | sdague, maybe a poc, I will mull it a bit more | 19:37 |
*** signed8bit_ZZZzz is now known as signed8bit | 19:38 | |
samueldmq | sdague, thanks! | 19:38 |
*** signed8bit is now known as signed8bit_ZZZzz | 19:39 | |
*** ijw has quit IRC | 19:39 | |
dansmith | that 3.5 minutes is with one test taking 120s and two taking 60s | 19:39 |
dansmith | yikes | 19:39 |
*** oro has joined #openstack-nova | 19:40 | |
sdague | oh, you must have the dbs configured correctly | 19:40 |
samueldmq | dansmith, one single test taking 120s, did I misread? | 19:40 |
dansmith | nova.tests.unit.compute.test_compute_mgr.ComputeManagerUnitTestCase.test_swap_volume_volume_api_usage | 19:41 |
dansmith | takes 120s ^ | 19:41 |
dansmith | sdague: eh? | 19:41 |
samueldmq | oO | 19:41 |
dansmith | sdague: *correctly* ? | 19:41 |
sdague | oh, I assumed it was the mysql tests | 19:41 |
dansmith | that's a unit unit test, shouldn't be using db | 19:41 |
dansmith | no | 19:41 |
sdague | they take real time | 19:41 |
dansmith | the first db test is 14s | 19:41 |
dansmith | s/first/longest/ | 19:41 |
sdague | right, but there are the oportunistic migration tests with real databases | 19:41 |
sdague | those take longer iirc | 19:42 |
dansmith | yeah, not those | 19:42 |
dansmith | right | 19:42 |
mriedem | test_compute_mgr.ComputeManagerUnitTestCase is NoDBTestCase is the point | 19:42 |
mriedem | shouldn't take 2 minutes | 19:42 |
sdague | but yeh, nova.tests.unit.compute.test_compute_mgr.ComputeManagerUnitTestCase.test_swap_volume_volume_api_usage seems wrong | 19:42 |
mriedem | did someone throw a sleep in :) | 19:42 |
sdague | but, I'm stepping out for a few | 19:42 |
samueldmq | mriedem, ++ haha | 19:42 |
dansmith | the error paths are taking 60s each | 19:47 |
*** alex_xu has quit IRC | 19:48 | |
*** alex_xu has joined #openstack-nova | 19:49 | |
*** markvoelker has joined #openstack-nova | 19:50 | |
dansmith | ah, I bet I got it | 19:50 |
*** ozialien has quit IRC | 19:50 | |
*** ijw has joined #openstack-nova | 19:50 | |
*** kmartin has quit IRC | 19:51 | |
*** jwcroppe has joined #openstack-nova | 19:51 | |
dansmith | hmm, maybe not | 19:51 |
*** rushil has quit IRC | 19:52 | |
*** jwcroppe has quit IRC | 19:53 | |
*** bkopilov has joined #openstack-nova | 19:54 | |
*** markvoelker has quit IRC | 19:55 | |
* mriedem runs it locally too | 19:57 | |
mriedem | the test is scary to look at | 19:57 |
dansmith | mriedem: I've almost got it licked | 19:57 |
*** ajayaa has quit IRC | 19:58 | |
mriedem | hmm, rpc timeout? | 19:58 |
mriedem | ignored rpc timeout? | 19:58 |
mriedem | two of them | 19:58 |
*** jichen has quit IRC | 19:58 | |
dansmith | yeah | 19:59 |
mriedem | nice | 19:59 |
dansmith | patch in a sec :) | 19:59 |
mriedem | back in the day we talked about mocking out conductor from within test_compute_mgr to blow up if you got to conductor from those tests | 19:59 |
*** jichen has joined #openstack-nova | 19:59 | |
dims | dansmith: that V4Proxy, one more bug showed up - https://review.openstack.org/#/c/173897/ possible fix | 19:59 |
openstackgerrit | Matt Riedemann proposed openstack/nova: compute: stop handling virt lifecycle events in cleanup_host() https://review.openstack.org/174069 | 20:00 |
dansmith | man, I suck | 20:00 |
dansmith | on the upside, this test used to take 120s and now takes 0.15s :) | 20:00 |
*** smccully has quit IRC | 20:01 | |
dims | haha | 20:01 |
*** oro has quit IRC | 20:01 | |
mriedem | dims: ah, the decorator could probably also be fixed | 20:01 |
mriedem | in how it normalizes the args/kwargs into a dict | 20:02 |
mriedem | i had to do that in @reverts_task_state | 20:02 |
*** Marga_ has joined #openstack-nova | 20:02 | |
openstackgerrit | Dan Smith proposed openstack/nova: Fix test_swap_volume_api_usage() https://review.openstack.org/174073 | 20:03 |
dansmith | merge that sumbitch ^ ! | 20:03 |
*** nkrinner has joined #openstack-nova | 20:03 | |
mriedem | dims: left a comment | 20:04 |
*** annashen has joined #openstack-nova | 20:04 | |
*** rushiagr is now known as rushiagr_away | 20:06 | |
*** jwcroppe has joined #openstack-nova | 20:07 | |
*** annashen has quit IRC | 20:08 | |
*** jwcroppe has quit IRC | 20:08 | |
*** annashen has joined #openstack-nova | 20:08 | |
*** jecarey has joined #openstack-nova | 20:08 | |
*** sbfox has joined #openstack-nova | 20:08 | |
dansmith | well, that saved like 15 seconds off my runtime | 20:09 |
*** ayoung has joined #openstack-nova | 20:12 | |
*** dave-mccowan has quit IRC | 20:12 | |
ayoung | Did we really leave the network specification off the common cli params? | 20:12 |
ayoung | $ openstack server create --flavor m1.medium --image "fedora-21-atomic" --key-name ayoung-pubkey ayoung-test --security-group default | 20:12 |
ayoung | ERROR: openstack Multiple possible networks found, use a Network ID to be more specific. (HTTP 400) (Request-ID: req-f16e86d0-877b-4d73-947e-712422040873) | 20:12 |
ayoung | and yet...no reference to network id in the --help string? | 20:12 |
ayoung | dims, you ever done this? | 20:13 |
*** dave-mccowan has joined #openstack-nova | 20:13 | |
dims | ayoung: never used the openstack command to boot if that's what you are asking :) | 20:14 |
ayoung | is that what --nic are for? | 20:14 |
samueldmq | dansmith, nice, 0.15 s :) | 20:14 |
ayoung | dims, yes that is what I am asking, but only cuz I see you and know you and don't want to address an empty room | 20:14 |
openstackgerrit | Dan Smith proposed openstack/nova: Fix test_attach_interface_failure() https://review.openstack.org/174080 | 20:15 |
openstackgerrit | Dan Smith proposed openstack/nova: Fix test_set_admin_password_bad_state() https://review.openstack.org/174081 | 20:15 |
dansmith | samueldmq: two more ^^ | 20:15 |
dims | ayoung: ack, looking one sec | 20:15 |
dansmith | for a total of 240s of time back | 20:15 |
ayoung | dims, think I got it | 20:15 |
ayoung | openstack server create --flavor m1.medium --image "fedora-21-atomic" --key-name ayoung-pubkey ayoung-test --security-group default --nic net-id=63258623-1fd5-497c-b62d-e0651e03bdca | 20:15 |
ayoung | that created something anyway | 20:16 |
samueldmq | dansmith, nice, I'll take a look, thanks :) | 20:16 |
dims | ayoung: haha :) | 20:16 |
*** rushil has joined #openstack-nova | 20:16 | |
*** jecarey has quit IRC | 20:17 | |
*** jecarey has joined #openstack-nova | 20:18 | |
dims | ayoung: looking through, you can specify net-id or port-id | 20:20 |
ayoung | dims, so I was doing this by trial and erro. I started by running openstack server create, and filling in the params, just like the N00Bi3 I am | 20:21 |
ayoung | and the error message didn't say nothing about ports | 20:21 |
ayoung | looking at the help string, I can see | 20:22 |
dims | ayoung: am still trying to figure out how to get "-h" to spit out the help for --nic | 20:22 |
ayoung | --nic <nic-config-string> | 20:22 |
ayoung | Specify NIC configuration (optional extension) | 20:22 |
ayoung | openstack help server create | 20:22 |
ayoung | I was able to guess the format for that param from the old nova cli | 20:23 |
ayoung | nova boot --flavor m1.tiny --image cirros-0.3.3-x86_64 --nic net-id=DEMO_NET_ID \ | 20:24 |
ayoung | --security-group default --key-name demo-key demo-instance1 | 20:24 |
*** rushiagr_away is now known as rushiagr | 20:24 | |
dims | right | 20:25 |
*** kmartin has joined #openstack-nova | 20:25 | |
*** dboik_ has quit IRC | 20:26 | |
*** dboik has joined #openstack-nova | 20:26 | |
mriedem | dansmith: +2 on those but someone will probably just add another test later that reintroduces the problem | 20:27 |
mriedem | i'd still like to just mock out self.compute.conductor_api globally in test_compute_mgr to blow up if not mocked specifically for a test case | 20:27 |
dansmith | yeah, I'm not sure hans' use_local=True is the right thing to do there etiher | 20:27 |
dansmith | right, I think I'd prefer that | 20:28 |
dansmith | have you ever tried to see what might fail? | 20:28 |
*** hdd has joined #openstack-nova | 20:28 | |
dansmith | replacing it with a big magicmock would let you easily override one call that you want to do a thing | 20:28 |
dansmith | i.e. maybe it doesn't need to blow up | 20:28 |
mriedem | self.compute.conductor_api = mock.Mock(side_effect=test.TestingException('f u'))? | 20:28 |
*** dboik has quit IRC | 20:28 | |
mriedem | i'd have to test it quick | 20:29 |
*** VW_ has quit IRC | 20:29 | |
dansmith | I guess that works since it's private | 20:29 |
*** dboik has joined #openstack-nova | 20:29 | |
*** baoli has joined #openstack-nova | 20:30 | |
mriedem | dansmith: http://paste.openstack.org/show/204037/ | 20:32 |
mriedem | that works on that swap volume test | 20:32 |
mriedem | it takes .153 seocnds | 20:32 |
*** Nic1 has quit IRC | 20:32 | |
mriedem | running the full class | 20:33 |
dansmith | it will there because it throws away any exception that it hits there | 20:33 |
dansmith | yeah | 20:33 |
dansmith | I still think making it blow up is probably not necessary | 20:33 |
dims | ayoung: https://review.openstack.org/174091 | 20:33 |
*** VW_ has joined #openstack-nova | 20:33 | |
*** Mike_D_laptop has joined #openstack-nova | 20:33 | |
mriedem | dansmith: all other tests pass in the test class | 20:33 |
dansmith | if you expect something back, then getting back the resulting MagicMock from any call should make you explode | 20:33 |
ayoung | dims, has anyone told you today that you are awesome? | 20:33 |
ayoung | No? | 20:33 |
ayoung | Then let me be the first. | 20:33 |
dansmith | mriedem: can you try without making it explode? | 20:34 |
dansmith | mriedem: actually, that's probably wrong | 20:34 |
ayoung | dims, http://paste.fedoraproject.org/211444/91300841/ | 20:34 |
dansmith | mriedem: that will only explode if you call the conductor_api as a callable | 20:35 |
ayoung | :) | 20:35 |
dansmith | mriedem: so I think you're doing what I was asking :) | 20:35 |
dansmith | mriedem: if you want to make it explode, you should do something like this: | 20:35 |
mriedem | dansmith: runs the same as just a mock w/o the side_effect | 20:35 |
dansmith | right | 20:35 |
dansmith | it needs to be something like: | 20:35 |
dansmith | m = mock.MagicMock() | 20:36 |
mriedem | shit, you could just make it None | 20:36 |
mriedem | if you wanted it to explode :) | 20:36 |
mriedem | m.callable = False | 20:36 |
dansmith | self.conductor_api = m.__getattr__.return_value.side_effect = Foo | 20:36 |
mriedem | actually | 20:36 |
mriedem | self.compute.conductor_api = mock.NonCallableMock() | 20:36 |
mriedem | bam | 20:36 |
dansmith | the reason I'm asking is because presumably we could want to actually check that one of those methods is called | 20:37 |
mriedem | yeah, idk, i think i'm moving on :) | 20:37 |
dansmith | mriedem: NonCallableMock doesn't make it explode | 20:38 |
mriedem | yeah, i know | 20:38 |
mriedem | now | 20:38 |
dansmith | hah | 20:38 |
*** tonytan4ever has quit IRC | 20:38 | |
mriedem | i have crazy donut sugar energy and running ADHD | 20:38 |
dansmith | lol | 20:38 |
samueldmq | dansmith, well, that saved ~100 s for me | 20:39 |
dansmith | samueldmq: then you have a slow computer :) | 20:39 |
dansmith | samueldmq: but glad to hear it :) | 20:39 |
samueldmq | dansmith, actually I am on a quad-core 20gb ram workstation :p | 20:41 |
dansmith | then you're doing something very wrong | 20:41 |
samueldmq | dansmith, don't know why it's taking so long | 20:41 |
dansmith | my 195s runs are coming from a 16GB quad i7 | 20:41 |
*** stevemar has quit IRC | 20:41 | |
dansmith | samueldmq: try nuking .testrepository | 20:42 |
*** vladikr has quit IRC | 20:42 | |
mriedem | was going to say that | 20:42 |
samueldmq | dansmith, hmm, so .. I cloned nova code, then tox -e py27 | 20:42 |
dansmith | it shits its own bed now and then and will cause single-threaded runs | 20:42 |
mriedem | i was having slow unit test runs before and a big .testrepository was something someone pointed out | 20:42 |
mriedem | but i don't think it helped me | 20:42 |
samueldmq | dansmith, ran it again and get 362 s ... got your patches and then 288 s | 20:42 |
dansmith | samueldmq: it takes a couple runs to get the optimal ordering, yeah | 20:42 |
dansmith | samueldmq: is yours quad with HT? | 20:43 |
dansmith | samueldmq: intel or AMD? | 20:43 |
samueldmq | dansmith, let me check | 20:43 |
dansmith | mine is quad i7 with HT, so eight threads | 20:43 |
dansmith | the closer to 240s that those patches save, the more linear your workload is being run | 20:44 |
dansmith | so 100s would make sense for only four threads I guess | 20:44 |
*** gyee has joined #openstack-nova | 20:44 | |
*** gyee has quit IRC | 20:44 | |
dansmith | across eight, it's not much gain for me | 20:44 |
samueldmq | dansmith, yeah, intel .. quad-core Intel(R) Xeon(R) CPU E3-1225 v3 @ 3.20GHz | 20:45 |
*** jichen has quit IRC | 20:45 | |
samueldmq | dansmith, maybe I need more runs to get the optimal ordering ? | 20:45 |
mriedem | dansmith: btw, what do you think about this? https://review.openstack.org/#/c/174069/ | 20:45 |
mriedem | pretty simple | 20:45 |
dansmith | samueldmq: should only take a few | 20:45 |
dansmith | samueldmq: that is only four threads | 20:46 |
dansmith | samueldmq: and based on core2 I think | 20:46 |
samueldmq | dansmith, k ... I will re-run to see what happens :) | 20:46 |
samueldmq | dansmith, note that I am not a nova guy .. I am just curious about this | 20:46 |
dansmith | so definitely going to be slower | 20:46 |
samueldmq | dansmith, I am mostly on keystone side | 20:47 |
dansmith | samueldmq: I have a machine with two of those procs in it (so 8 cores) and it runs only slightly faster than my i7 quad w/HT | 20:47 |
*** belmoreira has joined #openstack-nova | 20:47 | |
samueldmq | dansmith, hmm, so why ? | 20:47 |
samueldmq | dansmith, shared memory is the bottleneck? | 20:47 |
*** jecarey has quit IRC | 20:47 | |
dansmith | samueldmq: no, it's an older proc | 20:48 |
*** baoli has quit IRC | 20:48 | |
*** oro has joined #openstack-nova | 20:49 | |
*** baoli has joined #openstack-nova | 20:49 | |
*** otter768 has joined #openstack-nova | 20:49 | |
*** thedodd has quit IRC | 20:49 | |
samueldmq | dansmith, btw, what is missing for nova side to support keystone v3? | 20:50 |
samueldmq | dansmith, someone to work on it ? | 20:50 |
dansmith | samueldmq: I dunno, I'm not to the right person to ask | 20:50 |
dansmith | I think jogo might know more | 20:50 |
*** dboik_ has joined #openstack-nova | 20:51 | |
samueldmq | jogo, you around ? ^ | 20:51 |
mriedem | did we release kilo rc1 as kilo 2015.1? no rc2? why is there a stable/kilo now? | 20:52 |
mriedem | or just not doing proposed branches anymore? | 20:52 |
mriedem | dhellmann: ^? | 20:52 |
*** otter768 has quit IRC | 20:54 | |
*** dboik has quit IRC | 20:55 | |
dhellmann | mriedem: we've had to rename proposed/kilo to stable/kilo to make some of the tools work properly. We have not yet released RC2, though. ttx will be describing this in more detail on the ML tomorrow | 20:56 |
*** gszasz has quit IRC | 20:56 | |
*** rushil has quit IRC | 20:57 | |
mriedem | like requirements sync | 20:58 |
*** Mike_D_laptop has quit IRC | 21:00 | |
*** raildo has quit IRC | 21:00 | |
*** gyee has joined #openstack-nova | 21:00 | |
*** aysyd has quit IRC | 21:05 | |
*** Marga_ has quit IRC | 21:05 | |
samueldmq | dansmith, jogo, gotta go now, will be back later to ask about keystone v3 support :) | 21:07 |
*** ZZelle has quit IRC | 21:07 | |
*** mwagner_lap has quit IRC | 21:08 | |
*** jecarey has joined #openstack-nova | 21:08 | |
*** matrohon has quit IRC | 21:09 | |
*** jwcroppe has joined #openstack-nova | 21:10 | |
*** Marga_ has joined #openstack-nova | 21:10 | |
*** jwcroppe has quit IRC | 21:10 | |
*** dprince has quit IRC | 21:12 | |
*** Sukhdev has joined #openstack-nova | 21:13 | |
*** apuimedo has joined #openstack-nova | 21:15 | |
*** rfolco has quit IRC | 21:15 | |
*** jkraj has quit IRC | 21:16 | |
jogo | samueldmq: AFAIk nothing is missing for nova keystone v3 support | 21:16 |
morganfainberg | jogo, we need a way to run devstack w/o v2 enabled | 21:16 |
morganfainberg | jogo, and a way to turn off v2 checks in tempest... then we will see *what* breaks. I expect every service to have minor errors | 21:17 |
morganfainberg | jogo, on the short list to hit as soon as we're past release | 21:17 |
morganfainberg | jogo, but nova shouldn't care v2 or v3 at this point afaik | 21:17 |
morganfainberg | samueldmq, ^ cc | 21:18 |
*** Marga_ has quit IRC | 21:19 | |
*** Marga_ has joined #openstack-nova | 21:19 | |
jogo | morganfainberg: good idea, will we make devstack gate default to keystone v3 before the summit? | 21:22 |
morganfainberg | jogo, default? maaaaybe | 21:23 |
morganfainberg | available: likely | 21:23 |
jogo | that is a good start | 21:23 |
*** Vek has quit IRC | 21:24 | |
*** rushiagr is now known as rushiagr_away | 21:24 | |
samueldmq | morganfainberg, jogo nice, makes sense | 21:24 |
jogo | samueldmq: so if you find any issues please file bugs! | 21:25 |
*** andreykurilin has joined #openstack-nova | 21:25 | |
jogo | samueldmq: AFAIK the big thing is nova doesn't use domain scoped tokens | 21:25 |
jogo | so if your token is domain scoped things should just error | 21:25 |
*** dboik_ has quit IRC | 21:25 | |
*** dboik has joined #openstack-nova | 21:26 | |
samueldmq | jogo, ++ | 21:26 |
morganfainberg | jogo, you get a weird error in nova about not being in the project | 21:26 |
morganfainberg | but yes it does | 21:26 |
*** dave-mcc_ has joined #openstack-nova | 21:28 | |
samueldmq | jogo, I will sync up with morganfainberg the plan to get the devstack running and then see the errors | 21:28 |
samueldmq | jogo, nova is a good starting point | 21:28 |
*** nkrinner has quit IRC | 21:28 | |
samueldmq | I think most of the things that will break are due to code that is hard-coded for v2 | 21:29 |
samueldmq | morganfainberg, ^right? | 21:29 |
*** lpetrut has quit IRC | 21:29 | |
morganfainberg | samueldmq, yes | 21:30 |
*** dave-mccowan has quit IRC | 21:30 | |
samueldmq | nice, need to go now, talk to you later | 21:31 |
samueldmq | morganfainberg, jogo thanks | 21:31 |
*** busterswt has quit IRC | 21:32 | |
*** achanda has quit IRC | 21:34 | |
*** tjones1 has joined #openstack-nova | 21:34 | |
*** jwcroppe has joined #openstack-nova | 21:36 | |
*** achanda has joined #openstack-nova | 21:37 | |
*** achanda has quit IRC | 21:37 | |
cfriesen | bauzas: you on? | 21:38 |
*** achanda has joined #openstack-nova | 21:38 | |
bauzas | cfriesen: still up yeah | 21:38 |
cfriesen | bauzas: running into an issue with _filter_pools_for_numa_cells...noticed that you touched it somewhat recently. you know anything about that or who I should pester | 21:39 |
*** armax has joined #openstack-nova | 21:40 | |
cfriesen | bauzas: or at least you touched the pci/stats.py file. :) | 21:41 |
bauzas | cfriesen: oh yeah, because of an issue I had | 21:41 |
cfriesen | bauzas: okay, looks like author of code is Przemyslaw Czesnowicz | 21:42 |
bauzas | cfriesen: what's the problem, if I can help ? | 21:42 |
cfriesen | bauzas: well, it's calling utils.pci_device_prop_match(), but it appears that the pci_dev specifies the numa node as a string, but the specs specifies numa node as an int | 21:43 |
cfriesen | bauzas: so they don't compare properly | 21:43 |
*** tjones1 has quit IRC | 21:44 | |
bauzas | cfriesen: I pass | 21:46 |
*** dims_ has joined #openstack-nova | 21:46 | |
cfriesen | bauzas: :) | 21:46 |
*** jamielennox|away is now known as jamielennox | 21:46 | |
bauzas | cfriesen: ndipanov could help but he's offline | 21:46 |
*** andreykurilin has quit IRC | 21:46 | |
bauzas | cfriesen: I don't know the IRC nick of Przemyslaw | 21:47 |
*** annegentle has quit IRC | 21:47 | |
*** tjones1 has joined #openstack-nova | 21:48 | |
*** tjones1 has left #openstack-nova | 21:48 | |
*** dims has quit IRC | 21:48 | |
*** mwagner_lap has joined #openstack-nova | 21:48 | |
cfriesen | bauzas: will keep digging. wondering if it has anything to do with the change you made...it used to do a jsonutils.loads(stats) but you changed it to call to_dict() | 21:50 |
bauzas | for the stats ? | 21:50 |
cfriesen | bauzas: __init__() for PciDeviceStats writes self.pools | 21:51 |
*** jecarey has quit IRC | 21:51 | |
bauzas | cfriesen: yeah, because I objectified some bits in the API | 21:51 |
cfriesen | bauzas: so we're seeing pci_dev: {'count': 2, u'numa_node': u'0', 'vendor_id': u'8086', u'physical_network': u'group0-data0', 'product_id': u'10ed'} | 21:51 |
*** annegentle has joined #openstack-nova | 21:52 | |
*** dboik_ has joined #openstack-nova | 21:52 | |
cfriesen | bauzas: wondering if all those values should actually be ints and not unicode | 21:52 |
bauzas | cfriesen: what says the object ? | 21:53 |
*** vilobhmm11 has quit IRC | 21:53 | |
*** tjones2 has joined #openstack-nova | 21:53 | |
*** markvoelker has joined #openstack-nova | 21:53 | |
*** tjones2 has left #openstack-nova | 21:54 | |
*** sbfox has quit IRC | 21:54 | |
*** VW_ has quit IRC | 21:55 | |
*** irenab has joined #openstack-nova | 21:55 | |
cfriesen | bauzas: ah, okay. PciDevicePool shows product/vendor as strings. where's the numa node coming from? | 21:55 |
*** pixelb has joined #openstack-nova | 21:56 | |
*** dboik has quit IRC | 21:56 | |
bauzas | cfriesen: just to explain why I modified this, that's because the compute api proxy was changed to get the compute_nodes using the object method | 21:57 |
*** dboik_ has quit IRC | 21:57 | |
bauzas | so the nested pci_device_pool key was no longer a raw pci_stats but rather something objectified | 21:57 |
*** markvoelker has quit IRC | 21:58 | |
bauzas | in order to not change more than what was needed, I had to primitive that/these pci objects into something which was the opposite of from_pci_stats | 21:58 |
flwang | garyk: ping | 21:58 |
bauzas | cfriesen: hence the addition of to_dict() | 21:59 |
flwang | garyk: adam_g: could you pls review https://review.openstack.org/#/c/168759/ ? | 21:59 |
*** sbfox has joined #openstack-nova | 22:01 | |
cfriesen | bauzas: okay. will keep digging | 22:02 |
*** annegentle has quit IRC | 22:03 | |
*** patrickeast has quit IRC | 22:04 | |
*** artom has quit IRC | 22:05 | |
*** tjones1 has joined #openstack-nova | 22:07 | |
*** neelashah has quit IRC | 22:08 | |
*** MVenesio has quit IRC | 22:09 | |
*** hdd has quit IRC | 22:10 | |
*** tjones1 has quit IRC | 22:12 | |
openstackgerrit | Belmiro Moreira proposed openstack/nova: No disable reason defined for new services https://review.openstack.org/136645 | 22:14 |
*** yamahata has joined #openstack-nova | 22:14 | |
*** ijw has quit IRC | 22:18 | |
*** belmoreira has quit IRC | 22:18 | |
*** ijw has joined #openstack-nova | 22:20 | |
*** VW_ has joined #openstack-nova | 22:20 | |
*** annegentle has joined #openstack-nova | 22:20 | |
*** patrickeast has joined #openstack-nova | 22:21 | |
*** VW__ has joined #openstack-nova | 22:22 | |
*** VW_ has quit IRC | 22:22 | |
*** burt has quit IRC | 22:24 | |
*** annegentle has quit IRC | 22:24 | |
*** dboik has joined #openstack-nova | 22:26 | |
*** dboik_ has joined #openstack-nova | 22:27 | |
*** tjones1 has joined #openstack-nova | 22:28 | |
*** ijw has quit IRC | 22:30 | |
*** annashen has quit IRC | 22:30 | |
*** dboik has quit IRC | 22:31 | |
*** annashen has joined #openstack-nova | 22:31 | |
*** tjones1 has quit IRC | 22:33 | |
*** annegentle has joined #openstack-nova | 22:36 | |
*** jecarey has joined #openstack-nova | 22:36 | |
*** mriedem has quit IRC | 22:38 | |
*** tjones1 has joined #openstack-nova | 22:39 | |
*** annegentle has quit IRC | 22:41 | |
*** VW__ has quit IRC | 22:41 | |
*** jecarey has quit IRC | 22:41 | |
*** tjones1 has quit IRC | 22:46 | |
*** tjones1 has joined #openstack-nova | 22:46 | |
*** jwcroppe has quit IRC | 22:48 | |
*** achanda has quit IRC | 22:48 | |
*** patrickeast has quit IRC | 22:48 | |
*** otter768 has joined #openstack-nova | 22:50 | |
openstackgerrit | Tracy Jones proposed openstack/nova: Cleanup of compute api reboot method https://review.openstack.org/127427 | 22:50 |
*** Marga_ has quit IRC | 22:51 | |
*** markvoelker_ has joined #openstack-nova | 22:53 | |
*** otter768 has quit IRC | 22:54 | |
openstackgerrit | Tracy Jones proposed openstack/nova: Cleanup of compute api reboot method https://review.openstack.org/127427 | 22:56 |
*** baoli has quit IRC | 22:57 | |
*** achanda has joined #openstack-nova | 22:58 | |
*** baoli has joined #openstack-nova | 22:58 | |
*** apuimedo has quit IRC | 22:59 | |
*** oro has quit IRC | 22:59 | |
openstackgerrit | Tracy Jones proposed openstack/nova: Cleanup of compute api reboot method https://review.openstack.org/127427 | 23:00 |
*** erlon has quit IRC | 23:01 | |
*** zzzeek has quit IRC | 23:01 | |
*** baoli has quit IRC | 23:03 | |
*** Marga_ has joined #openstack-nova | 23:03 | |
*** erkules_ is now known as erkules | 23:03 | |
*** hemna is now known as hemnafk | 23:05 | |
openstackgerrit | Alex Xu proposed openstack/nova: Update rpc version aliases for kilo https://review.openstack.org/174148 | 23:06 |
*** jaypipes has quit IRC | 23:08 | |
*** Sukhdev has quit IRC | 23:12 | |
*** Marga_ has quit IRC | 23:13 | |
*** tjones1 has left #openstack-nova | 23:15 | |
*** otter768 has joined #openstack-nova | 23:16 | |
*** patrickeast has joined #openstack-nova | 23:16 | |
*** irenab has quit IRC | 23:17 | |
*** asettle has quit IRC | 23:19 | |
*** Hosam has quit IRC | 23:19 | |
*** Vek has joined #openstack-nova | 23:20 | |
*** annashen has quit IRC | 23:21 | |
*** jaypipes has joined #openstack-nova | 23:21 | |
*** Marga_ has joined #openstack-nova | 23:23 | |
openstackgerrit | Alex Xu proposed openstack/nova: Add rpc client side version control for conductor task API https://review.openstack.org/174152 | 23:24 |
*** tjones1 has joined #openstack-nova | 23:28 | |
*** jwcroppe has joined #openstack-nova | 23:30 | |
*** Madkiss has quit IRC | 23:31 | |
*** Madkiss_ has joined #openstack-nova | 23:31 | |
*** Madkiss_ is now known as Madkiss | 23:31 | |
*** armax_ has joined #openstack-nova | 23:32 | |
*** tjones1 has quit IRC | 23:33 | |
*** promulo__ has joined #openstack-nova | 23:33 | |
*** tjones1 has joined #openstack-nova | 23:33 | |
*** mestery_ has joined #openstack-nova | 23:33 | |
*** tteggel_ has joined #openstack-nova | 23:34 | |
*** kmartin_ has joined #openstack-nova | 23:34 | |
*** aloga_ has joined #openstack-nova | 23:35 | |
*** francois has joined #openstack-nova | 23:35 | |
*** frickler_ has joined #openstack-nova | 23:35 | |
*** HenryG_ has joined #openstack-nova | 23:35 | |
*** triggerz_ has joined #openstack-nova | 23:35 | |
*** salv-orl_ has joined #openstack-nova | 23:36 | |
*** luv has quit IRC | 23:38 | |
*** mdbooth has quit IRC | 23:38 | |
*** aloga has quit IRC | 23:38 | |
*** kmartin has quit IRC | 23:38 | |
*** bkopilov has quit IRC | 23:38 | |
*** HenryG has quit IRC | 23:38 | |
*** triggerz has quit IRC | 23:38 | |
*** luv_ has joined #openstack-nova | 23:38 | |
*** mdbooth_ has joined #openstack-nova | 23:38 | |
*** luv_ is now known as luv | 23:38 | |
*** bkopilov has joined #openstack-nova | 23:38 | |
*** lan_ has joined #openstack-nova | 23:38 | |
*** cburgess_ has joined #openstack-nova | 23:38 | |
*** mdbooth_ is now known as mdbooth | 23:38 | |
*** markvoelker has joined #openstack-nova | 23:39 | |
*** rpodolyaka has quit IRC | 23:39 | |
*** jesusaurus has quit IRC | 23:39 | |
*** francois1 has quit IRC | 23:39 | |
*** armax has quit IRC | 23:39 | |
*** salv-orlando has quit IRC | 23:39 | |
*** lan has quit IRC | 23:39 | |
*** promulo_ has quit IRC | 23:39 | |
*** heyongli has quit IRC | 23:39 | |
*** frickler has quit IRC | 23:39 | |
*** mestery has quit IRC | 23:39 | |
*** tteggel has quit IRC | 23:39 | |
*** cburgess has quit IRC | 23:39 | |
*** heyongli has joined #openstack-nova | 23:39 | |
*** rpodolyaka has joined #openstack-nova | 23:39 | |
*** heyongli has quit IRC | 23:39 | |
*** heyongli has joined #openstack-nova | 23:39 | |
*** armax_ is now known as armax | 23:39 | |
*** markvoelker_ has quit IRC | 23:40 | |
*** jesusaurus has joined #openstack-nova | 23:40 | |
*** markvoelker has quit IRC | 23:40 | |
*** markvoelker has joined #openstack-nova | 23:40 | |
*** markvoelker_ has joined #openstack-nova | 23:41 | |
*** emagana has quit IRC | 23:43 | |
*** tjones1 has left #openstack-nova | 23:43 | |
*** markvoelker has quit IRC | 23:45 | |
*** tjones1 has joined #openstack-nova | 23:50 | |
*** oro has joined #openstack-nova | 23:53 | |
*** Marga_ has quit IRC | 23:54 | |
*** tjones1 has quit IRC | 23:57 | |
*** dboik_ has quit IRC | 23:59 | |
*** wolsen_ is now known as wolsen | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!