Wednesday, 2015-06-03

*** hemna is now known as hemnafk00:01
*** ijw_ has joined #openstack-nova00:02
*** tjones1 has quit IRC00:03
*** tjones2 has joined #openstack-nova00:03
*** browne has joined #openstack-nova00:03
*** Alexandra_ has quit IRC00:06
*** ijw has quit IRC00:06
openstackgerritJens Rosenboom proposed openstack/nova: Fix some issues in devref for api_microversions  https://review.openstack.org/18713300:09
*** BadCub is now known as BadCub_Hiding00:10
*** shakamunyi has joined #openstack-nova00:12
*** barra204 has joined #openstack-nova00:12
*** gholler has quit IRC00:13
*** oro has quit IRC00:19
*** patrickeast has quit IRC00:20
*** ian_brown has joined #openstack-nova00:21
*** nelsnelson has quit IRC00:24
*** nelsnelson has joined #openstack-nova00:25
*** ijw_ has quit IRC00:27
*** ijw has joined #openstack-nova00:27
*** VW has joined #openstack-nova00:29
*** zhenguo has joined #openstack-nova00:33
*** Sukhdev has quit IRC00:35
*** xyang1 has quit IRC00:35
*** shashankhegde has quit IRC00:47
*** fawadkhaliq has quit IRC00:48
*** salv-orlando has joined #openstack-nova00:48
*** promulo__ has joined #openstack-nova00:52
*** yingjun has joined #openstack-nova00:53
*** gokrokve has quit IRC00:54
*** gokrokve has joined #openstack-nova00:55
*** salv-orlando has quit IRC00:55
*** mtanino has quit IRC00:56
*** promulo has quit IRC00:56
*** irenab_ has joined #openstack-nova00:57
*** bkopilov has quit IRC00:58
*** irenab has quit IRC00:58
*** irenab_ is now known as irenab00:58
*** tsekiyama has quit IRC00:59
*** gokrokve has quit IRC00:59
*** Raildo_ has joined #openstack-nova01:02
*** Raildo has quit IRC01:04
*** tjones2 has quit IRC01:05
*** tjones1 has joined #openstack-nova01:06
*** bkopilov has joined #openstack-nova01:08
*** nelsnelson has quit IRC01:12
*** superfly_ has joined #openstack-nova01:12
*** asettle has joined #openstack-nova01:13
*** superfly_ is now known as superflyy01:14
*** gokrokve has joined #openstack-nova01:14
openstackgerritZhenyu Zheng proposed openstack/nova: resize-revert can break anti-affinity policy  https://review.openstack.org/16011001:17
*** Raildo_ has quit IRC01:18
*** mmedvede has quit IRC01:20
*** vilobhmm1 has joined #openstack-nova01:20
*** vilobhmm has quit IRC01:21
*** angdraug has quit IRC01:21
*** barra204 has quit IRC01:24
*** shakamunyi has quit IRC01:24
*** superflyy has quit IRC01:24
*** mmedvede has joined #openstack-nova01:25
openstackgerritDavanum Srinivas (dims) proposed openstack/nova: Enable python34 tests for nova/tests/unit/objects/test*.py  https://review.openstack.org/18778501:29
*** alanf-mc has quit IRC01:31
*** browne has quit IRC01:31
*** ssurana has quit IRC01:31
*** ssurana has joined #openstack-nova01:32
*** tobe has joined #openstack-nova01:33
*** ijw has quit IRC01:34
*** ijw has joined #openstack-nova01:35
openstackgerritJens Rosenboom proposed openstack/nova-specs: Allow ip6 server search for non-admin  https://review.openstack.org/18783501:36
*** ssurana has quit IRC01:36
*** ian_brown has quit IRC01:39
*** baoli has quit IRC01:41
*** vilobhmm1 has quit IRC01:42
openstackgerritZhenyu Zheng proposed openstack/nova: add support for muti_statuses_query from migration  https://review.openstack.org/16963101:43
*** nelsnelson has joined #openstack-nova01:51
*** pixelbeat has quit IRC01:52
*** pixelbeat_ has joined #openstack-nova01:52
*** jyuso has joined #openstack-nova01:53
*** otter768 has joined #openstack-nova01:55
*** davideagnello has quit IRC01:58
*** JHelf has joined #openstack-nova01:58
*** otter768 has quit IRC01:59
*** VW has quit IRC02:04
*** VW has joined #openstack-nova02:05
*** nelsnelson has quit IRC02:05
*** shashankhegde has joined #openstack-nova02:06
*** vivek_nandavanam has joined #openstack-nova02:06
*** shashankhegde has quit IRC02:06
*** ianbrown has joined #openstack-nova02:07
*** salv-orlando has joined #openstack-nova02:08
*** rushiagr_away is now known as rushiagr02:09
*** asettle has quit IRC02:11
*** sdake has quit IRC02:11
*** asettle has joined #openstack-nova02:12
*** JHelf has quit IRC02:12
*** asettle has quit IRC02:13
*** salv-orlando has quit IRC02:14
*** bzhao has joined #openstack-nova02:15
*** tjones1 has quit IRC02:15
*** rfolco has quit IRC02:17
*** yamahata has quit IRC02:17
*** zxen has joined #openstack-nova02:20
*** mohitsharma has joined #openstack-nova02:20
openstackgerritDavanum Srinivas (dims) proposed openstack/nova: Enable python34 tests for nova/tests/unit/objects/test*.py  https://review.openstack.org/18778502:22
*** unicell has quit IRC02:22
*** baoli has joined #openstack-nova02:23
*** IanGovett has joined #openstack-nova02:26
*** haigang has quit IRC02:26
*** hshiina has quit IRC02:26
*** yingjun has quit IRC02:26
*** haigang has joined #openstack-nova02:27
*** annegentle has joined #openstack-nova02:28
*** yingjun has joined #openstack-nova02:29
*** haigang has quit IRC02:30
*** haigang has joined #openstack-nova02:32
*** achanda has joined #openstack-nova02:32
*** mpaolino has joined #openstack-nova02:33
*** haigang has quit IRC02:33
*** haigang has joined #openstack-nova02:34
openstackgerritAuktavian Garrett proposed openstack/nova: Use unique values for hypervisor_hostnames  https://review.openstack.org/18784402:35
*** Kennan2 has joined #openstack-nova02:37
*** Kennan has quit IRC02:38
*** haigang has quit IRC02:38
*** IanGovett has quit IRC02:39
*** rushiagr is now known as rushiagr_away02:40
*** mpaolino has quit IRC02:40
*** nelsnelson has joined #openstack-nova02:41
*** annegentle has quit IRC02:44
*** haigang has joined #openstack-nova02:46
*** Kennan2 is now known as Kennan02:49
*** baoli has quit IRC02:49
*** ijw has quit IRC02:50
openstackgerritDavanum Srinivas (dims) proposed openstack/nova: Enable python34 tests for nova/tests/unit/objects/test*.py  https://review.openstack.org/18778502:51
*** annegentle has joined #openstack-nova02:51
*** annegentle has quit IRC02:51
*** unicell has joined #openstack-nova02:52
*** tjones1 has joined #openstack-nova02:55
*** dims__ has quit IRC02:56
*** hshiina has joined #openstack-nova02:58
openstackgerritAlex Xu proposed openstack/nova: Add the rule of separate plugin for Nova REST API in devref  https://review.openstack.org/16291302:59
openstackgerritAlex Xu proposed openstack/nova: The devref for Nova stable API  https://review.openstack.org/16291202:59
*** gyee is now known as operator9903:00
*** ddaskal has joined #openstack-nova03:00
*** achanda has quit IRC03:02
*** achanda has joined #openstack-nova03:03
*** achanda has quit IRC03:08
*** unicell has quit IRC03:09
*** unicell has joined #openstack-nova03:10
*** mohitsharma has quit IRC03:13
*** asettle has joined #openstack-nova03:13
openstackgerritChangBo Guo(gcb) proposed openstack/nova: Filter out inactive compute firstly  https://review.openstack.org/18015503:14
*** asettle has quit IRC03:15
*** asettle has joined #openstack-nova03:15
*** tjones1 has left #openstack-nova03:16
*** mohitsharma has joined #openstack-nova03:17
*** gokrokve_ has joined #openstack-nova03:17
*** alanf-mc has joined #openstack-nova03:21
*** gokrokve has quit IRC03:21
*** gokrokve_ has quit IRC03:21
*** dboik has joined #openstack-nova03:23
*** alanf-mc has quit IRC03:25
*** dboik has quit IRC03:27
*** nelsnelson has quit IRC03:30
*** achanda has joined #openstack-nova03:34
*** hshiina has quit IRC03:34
*** yingjun has quit IRC03:34
*** kiran-r has joined #openstack-nova03:38
*** markvoelker has quit IRC03:44
*** iamjarvo has joined #openstack-nova03:49
*** Longgeek_ has quit IRC03:53
*** alexpilotti has joined #openstack-nova03:54
*** arnaud____ has joined #openstack-nova03:54
*** Longgeek has joined #openstack-nova03:54
*** otter768 has joined #openstack-nova03:55
*** browne has joined #openstack-nova03:55
*** Longgeek_ has joined #openstack-nova03:57
*** achanda has quit IRC03:58
*** achanda has joined #openstack-nova03:58
*** ddaskal has quit IRC03:59
*** changbl_ has joined #openstack-nova04:00
*** Longgeek has quit IRC04:00
*** otter768 has quit IRC04:01
*** hshiina has joined #openstack-nova04:01
*** Longgeek has joined #openstack-nova04:02
*** achanda has quit IRC04:03
*** Longgeek_ has quit IRC04:04
*** Longgeek_ has joined #openstack-nova04:08
openstackgerritNicolas Simonds proposed openstack/nova: Fix/implement revert-resize for RBD-backed images  https://review.openstack.org/18739504:08
openstackgerritNicolas Simonds proposed openstack/nova: imagebackend: Split out resize_image logic from create_image  https://review.openstack.org/18785704:08
openstackgerritNicolas Simonds proposed openstack/nova: imagebackend: Split out resize_image logic from create_image  https://review.openstack.org/18785704:09
openstackgerritNicolas Simonds proposed openstack/nova: Fix/implement revert-resize for RBD-backed images  https://review.openstack.org/18739504:10
*** salv-orlando has joined #openstack-nova04:10
*** Longgeek has quit IRC04:10
*** nelsnelson has joined #openstack-nova04:11
*** gabor has joined #openstack-nova04:12
*** mdrabe has joined #openstack-nova04:12
*** gabor has quit IRC04:12
*** Longgeek_ has quit IRC04:12
*** artom has quit IRC04:13
*** gabor-t has joined #openstack-nova04:14
*** pixelbeat_ has quit IRC04:14
*** Longgeek has joined #openstack-nova04:18
*** Longgeek has quit IRC04:18
*** nic has quit IRC04:18
*** Longgeek has joined #openstack-nova04:19
*** Longgeek has quit IRC04:20
*** Longgeek has joined #openstack-nova04:20
*** Longgeek_ has joined #openstack-nova04:23
*** Longgee__ has joined #openstack-nova04:25
*** Longgeek has quit IRC04:25
*** jecarey has joined #openstack-nova04:25
*** VW has quit IRC04:26
*** Longgeek_ has quit IRC04:26
*** Longgeek has joined #openstack-nova04:26
*** Longgeek_ has joined #openstack-nova04:28
*** mdrabe has quit IRC04:28
*** sorrison_laptop has joined #openstack-nova04:29
*** Longge___ has joined #openstack-nova04:29
*** Longgee__ has quit IRC04:30
*** coolsvap|afk is now known as coolsvap04:30
*** coolsvap is now known as coolsvap|afk04:31
*** Longgeek has quit IRC04:32
*** coolsvap|afk is now known as coolsvap04:32
*** Longgeek_ has quit IRC04:32
*** bzhao has quit IRC04:34
*** changbl_ has quit IRC04:40
*** krtaylor has quit IRC04:40
*** hshiina has quit IRC04:44
*** markvoelker has joined #openstack-nova04:45
*** markvoelker has quit IRC04:50
*** garyk has joined #openstack-nova04:53
*** ildikov has quit IRC04:54
*** kiran-r has quit IRC04:59
*** nkrinner has joined #openstack-nova05:03
*** mohitsharma has quit IRC05:05
*** ildikov has joined #openstack-nova05:13
openstackgerritAlex Xu proposed openstack/nova: Add the rule of separate plugin for Nova REST API in devref  https://review.openstack.org/16291305:15
openstackgerritAlex Xu proposed openstack/nova: The devref for Nova stable API  https://review.openstack.org/16291205:15
*** jecarey has quit IRC05:15
*** yamahata has joined #openstack-nova05:19
*** arnaud____ has quit IRC05:21
*** mpaolino has joined #openstack-nova05:23
*** mpaolino has quit IRC05:25
openstackgerritRitesh proposed openstack/nova: Users will not be allowed to delete instance during snapshot  https://review.openstack.org/18594205:25
openstackgerritRitesh proposed openstack/nova: Users will not be allowed to delete instance during snapshot  https://review.openstack.org/18594205:26
*** krtaylor has joined #openstack-nova05:27
*** exploreshaifali has joined #openstack-nova05:27
*** apuimedo has quit IRC05:27
*** hshiina has joined #openstack-nova05:36
*** nihilifer has joined #openstack-nova05:37
*** ildikov has quit IRC05:38
openstackgerritJens Rosenboom proposed openstack/nova-specs: Allow ip6 server search for non-admin  https://review.openstack.org/18783505:38
*** vilobhmm has joined #openstack-nova05:42
*** moshele has joined #openstack-nova05:42
*** exploreshaifali has quit IRC05:43
*** yingjun has joined #openstack-nova05:43
*** dulek has joined #openstack-nova05:46
*** jyuso has quit IRC05:49
fricklerMicrosoft Hyper-V CI check takes 4 hours to fail on a doc only check, good job: https://review.openstack.org/18713305:49
fricklerI fail to find some error in their log, too05:49
fricklers/check/patch/05:50
*** veena has joined #openstack-nova05:50
*** sudipto has joined #openstack-nova05:50
*** ankit_ag has joined #openstack-nova05:51
*** Maike has joined #openstack-nova05:52
*** vipul has quit IRC05:53
*** gabor-t has quit IRC05:54
*** vipul has joined #openstack-nova05:54
*** mpaolino has joined #openstack-nova05:55
*** mohitsharma has joined #openstack-nova05:56
*** otter768 has joined #openstack-nova05:56
*** tcammann_ has quit IRC05:58
*** iamjarvo has quit IRC05:58
*** slong_ has joined #openstack-nova05:58
*** lpetrut has joined #openstack-nova05:58
*** tcammann_ has joined #openstack-nova05:58
*** vilobhmm has quit IRC05:58
*** slong has quit IRC05:59
*** josecastroleon has joined #openstack-nova05:59
*** armax has quit IRC06:01
*** otter768 has quit IRC06:01
*** mpaolino has quit IRC06:01
*** ildikov has joined #openstack-nova06:04
*** mpaolino has joined #openstack-nova06:04
*** radek_ has joined #openstack-nova06:07
*** sahid has joined #openstack-nova06:07
*** stefanb has quit IRC06:08
*** mpaolino has quit IRC06:09
*** kiran-r has joined #openstack-nova06:10
*** asettle has quit IRC06:11
*** romainh has joined #openstack-nova06:11
*** shakamunyi has joined #openstack-nova06:16
*** barra204 has joined #openstack-nova06:17
*** mnestratov has quit IRC06:17
*** mnestratov has joined #openstack-nova06:17
*** stefanb has joined #openstack-nova06:22
*** belmoreira has joined #openstack-nova06:26
*** zxen has quit IRC06:27
*** sorrison_laptop has quit IRC06:27
*** slong__ has joined #openstack-nova06:27
openstackgerritRitesh proposed openstack/nova: Users will not be allowed to delete instance during snapshot  https://review.openstack.org/18594206:28
*** moshele is now known as moshele_06:28
*** moshele_ has quit IRC06:28
*** slong_ has quit IRC06:28
*** moshele has joined #openstack-nova06:29
*** mohitsharma has quit IRC06:29
*** mohitsharma has joined #openstack-nova06:29
*** nelsnelson has quit IRC06:30
*** markvoelker has joined #openstack-nova06:34
*** ianbrown has quit IRC06:35
*** Longgeek has joined #openstack-nova06:35
*** bzhao has joined #openstack-nova06:37
*** Longge___ has quit IRC06:37
*** markvoelker has quit IRC06:38
*** oro_ has joined #openstack-nova06:38
*** cfriesen has quit IRC06:39
*** slong__ has quit IRC06:40
openstackgerritChangBo Guo(gcb) proposed openstack/nova: libvirt: add missing iotune parse for LibvirtConfigGuestDisk  https://review.openstack.org/18788106:51
*** pradipta has joined #openstack-nova06:55
sudiptosahid, hello, regarding the review of https://review.openstack.org/185942 - i wanted to understand in the extensions.py - why should we not add a more generic base class of the exception which could be something like InstanceInvalidState ?06:58
*** sahid has quit IRC07:01
*** radek_ has quit IRC07:02
*** browne has quit IRC07:02
openstackgerritChangBo Guo(gcb) proposed openstack/nova: libvirt:add instance disk IO control for rbd backend  https://review.openstack.org/18788807:04
*** radek_ has joined #openstack-nova07:05
*** mpavone has joined #openstack-nova07:13
openstackgerritgaryk proposed openstack/nova: DB: add support for volume_type by BDMS  https://review.openstack.org/13999207:13
openstackgerritgaryk proposed openstack/nova: Object: add volume_type_id to block_device objects  https://review.openstack.org/13999407:14
*** Nakato has quit IRC07:16
*** exploreshaifali has joined #openstack-nova07:18
*** radek__ has joined #openstack-nova07:18
*** Nakato has joined #openstack-nova07:18
*** FL1SK has quit IRC07:18
*** gryf|afk is now known as _gryf07:19
*** haigang has quit IRC07:19
*** alex_klimov has joined #openstack-nova07:19
*** radek_ has quit IRC07:20
*** bradjones has quit IRC07:20
*** rotbeard has joined #openstack-nova07:22
*** bradjones has joined #openstack-nova07:22
openstackgerritRui Chen proposed openstack/nova: Refactor network API 'get_instance_nw_info'  https://review.openstack.org/14603607:24
*** mohitsharma has quit IRC07:27
*** matrohon has joined #openstack-nova07:27
*** mohitsharma has joined #openstack-nova07:27
*** jyuso1 has joined #openstack-nova07:29
*** radek_ has joined #openstack-nova07:35
*** radek__ has quit IRC07:37
openstackgerritRui Chen proposed openstack/nova: Fix resource leaking when consume_from_instance raise exception  https://review.openstack.org/18216507:38
*** pbandzi has joined #openstack-nova07:39
*** coolsvap is now known as coolsvap|afk07:41
*** belmoreira has quit IRC07:42
openstackgerritZhang Ni proposed openstack/nova-specs: Add volume type to create server API  https://review.openstack.org/18781207:44
*** jcoufal has joined #openstack-nova07:45
*** markus_z has joined #openstack-nova07:45
*** belmoreira has joined #openstack-nova07:46
*** coolsvap|afk is now known as coolsvap07:46
*** jistr has joined #openstack-nova07:51
*** tobe has quit IRC07:52
kaisers1Hi Nova! Small question: where exactly do i configure how much RAM can be used by nova-compute on a given nova-compute host? That is not set by the quota_RAM settings, isn't it?07:54
gilliardkaisers1: by nova-compute itself?07:55
*** otter768 has joined #openstack-nova07:57
*** dims_ has joined #openstack-nova07:58
kaisers1gilliard: yep. I have a default RDO installation (Juno) that by default seems to provide 8GB RAM top be shared among VMs on a given compute host. As the machine has more available i wanted to reconfigure this.07:59
kaisers1'...top ...' was '... to ...'07:59
*** ndipanov has joined #openstack-nova08:00
kaisers1gilliard: but i only find options for quotas (default something like 52GB) and reserved_host_memory_mb.08:00
kaisers1gilliard: The latter would make sense to me if nova would provide all available RAM to VMs...08:01
*** otter768 has quit IRC08:02
*** rushiagr_away is now known as rushiagr08:02
gilliardkaisers1: I only know of reserverd_host_memory_mb, but I think it does what you suggest...08:03
*** dims_ has quit IRC08:03
gilliardor, I *thought* it did ;)08:03
kaisers1gilliard: Weird. Maybe there's a different layer in tenant management, etc. that filters available memory per host? I've just a very big gap between physical memory available and memory allowed by nova compute when instantiating VMs :-D08:04
*** jlanoux has joined #openstack-nova08:05
openstackgerritAlex Xu proposed openstack/nova: Add the rule of separate plugin for Nova REST API in devref  https://review.openstack.org/16291308:08
openstackgerritAlex Xu proposed openstack/nova: The devref for Nova stable API  https://review.openstack.org/16291208:08
*** exploreshaifali has quit IRC08:08
gilliardyeah quotas will be per-tenant. Are you sure seeing boot requests fail for not-enough-memory reasons?08:08
*** Kennan2 has joined #openstack-nova08:11
*** Kennan has quit IRC08:11
*** Nakato has quit IRC08:13
*** lucasagomes has joined #openstack-nova08:14
*** tobe has joined #openstack-nova08:14
*** Nakato has joined #openstack-nova08:14
*** yingjun has quit IRC08:14
*** oro_ has quit IRC08:15
*** yingjun has joined #openstack-nova08:19
*** derekh has joined #openstack-nova08:20
*** apuimedo has joined #openstack-nova08:21
*** sahid has joined #openstack-nova08:22
*** markvoelker has joined #openstack-nova08:22
bauzaskaisers1: you can allocate more than what you want, since there are allocation ratios for RAM08:23
bauzaskaisers1: defaulting to 16 btw.08:23
*** vivek_nandavanam has quit IRC08:23
*** lpetrut has quit IRC08:23
moshelepczesno: hi08:25
bauzasgilliard: reserved_host_memory_mb is for making sure that you leave some space for else08:26
bauzaswhatelse08:26
*** danpb has joined #openstack-nova08:26
*** yassine_ has joined #openstack-nova08:27
bauzaskaisers1 : so the RAM thay you dedicate to Nova per compute would be "(total RAM - reserved_host_memory_mb) * ram_allocation_ratio"08:27
*** markvoelker has quit IRC08:27
bauzaskaisers1: it will give you how much RAM you can cap08:27
*** ZZelle has quit IRC08:28
*** ZZelle has joined #openstack-nova08:28
kaisers1bauzas: thanks for the info!08:30
kaisers1bauzas: But there has to be another (maybe project related?) setting. I'm looking at a 64GB RAM machine, reserved_host_memory_mb is 16384. Nova shows 8GB RAM available for VMs. I'd expect around 48GB to be available for VMs, no?08:34
*** heyongli has quit IRC08:34
kaisers1Maybe there are other (hypervisor? other?) related settings that influence this?08:34
bauzaskaisers1: your ram allocation ratio ?08:34
kaisers1No, the total RAM available for VMs08:35
bauzaskaisers1: I mean, could you please give it to me ?08:35
kaisers1bauzas: hmm, not set, what is the default value?08:36
bauzaskaisers1: 1608:37
kaisers1bauzas: setting this to 1.0 and restarting nova-compute....08:37
openstackgerritMarkus Zoeller (markus_z) proposed openstack/nova: WIP: docs: Add bug handling description to developer reference  https://review.openstack.org/18757108:37
bauzaskaisers1: technically, we're not following the $subject of this chan, but let's continue to see if there is bug08:37
gilliardBTW if you have the source checked out, run "tox -e genconfig" then see etc/nova/nova.conf.sample to find doc and defaults for all config options.08:38
kaisers1bauzas: oh, sry. Wich channel would be appropriate?08:38
bauzaskaisers1: #openstack08:38
bauzaskaisers1: but let's continue08:38
kaisers1bauzas: ok08:38
bauzasthe room is quite quiet now08:38
bauzaskaisers1: so, no instances ?08:38
bauzaskaisers1: check the n-cpu log, it will tell you the physical capacity and the ram in use08:39
kaisers1moment, checking on the changes i did08:39
bauzaskaisers1: here is how it calculates the usage https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L800-L82608:39
kaisers1hmm, now the available RAM is equal to the quota limit, yay!08:40
kaisers1~51GB, that's the default of the quota_RAM setting08:41
*** ianbrown has joined #openstack-nova08:41
bauzascool08:42
kaisers1Ok, from what i see now nova tells me 16GB are in use (the reserved_host_memory_mb? Is that show as 'in use'?). Up to a total of 52GB RAM is available for VMs. That's what i was aiming at08:42
bauzaskaisers1: yup for reserved_ram, see my link above08:43
kaisers1That's with no running guests.08:43
bauzaskaisers1: yeah08:43
kaisers1yep, will read now. Thanks a lot08:43
bauzaskaisers1: try to understand how the RT works08:43
openstackgerritDaniel Berrange proposed openstack/nova: docs: fixup libvirt NUMA testing docs to match reality  https://review.openstack.org/17877308:43
kaisers1gilliard: bauzas: To both of you!!08:43
bauzaskaisers1: https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L393 is the entry point, called every 60 secs08:44
bauzaskaisers1: follow the code, you'll see how Nova counts08:44
kaisers1bauzas: ack08:44
*** garyk has quit IRC08:44
*** garyk has joined #openstack-nova08:45
*** garyk has quit IRC08:46
*** garyk has joined #openstack-nova08:46
*** heyongli has joined #openstack-nova08:49
*** heyongli has quit IRC08:51
*** nkrinner has quit IRC08:52
openstackgerritJoe Gordon proposed openstack/nova: Prune development.environment doc  https://review.openstack.org/18288408:53
openstackgerritJoe Gordon proposed openstack/nova: Remove jenkins, launchpad and gerrit docs  https://review.openstack.org/18334208:53
openstackgerritJoe Gordon proposed openstack/nova: Make blueprints doc a reference for nova blueprints  https://review.openstack.org/18334308:53
openstackgerritJoe Gordon proposed openstack/nova: Remove unit_test doc  https://review.openstack.org/18369708:53
gilliardkaisers1: :)08:55
mdboothIs there any reason that we would have a requirement for oslo.db>=1.7.0 rather than the latest 1.8.0?08:56
*** ianbrown has quit IRC08:56
mdboothI submitted a patch which uses 1.8.0. Works on some jenkins tests, not on others.08:56
*** nkrinner has joined #openstack-nova08:57
mdboothOn the ones where it doesn't it seems it hasn't pulled in the latest oslo.db08:57
mdboothAnd the requirements is definitely bugged08:57
johnthetubaguymdbooth: we have been pinning some releases due to compatibility issues, but I don't remember anything about oslo.db08:58
mdboothjohnthetubaguy: It doesn't currently have an upper bound in requirements.txt08:58
johnthetubaguymdbooth: oh, you mean its just the old version thats installed in the test VM?08:59
mdboothYeah08:59
mdboothHowever, the old version meets the constraint in requirements.txt08:59
johnthetubaguymdbooth: hmm, is there anything in global requirements?08:59
mdboothSo that's obviously incorrect if I'm using new features08:59
mdboothjohnthetubaguy: No. Just wondering if there's any reason for that08:59
mdboothThere was an upper bound before L09:00
mdboothSo it was change to remove the upper bound, but the lower bound was not bumped09:00
johnthetubaguyyeah, we added a bound for the stable stuff09:00
johnthetubaguyunsure to be honest, dims and/or sdague would be almost certainly be able to tell you the answer09:00
mdboothjohnthetubaguy: Git says it was updated in a global removal of upper bounds, so likely no specific thought09:03
* mdbooth submits a patch to global-requirements.txt, waits for review, waits for committee approval, waits for filter-down merge in nova, rebases patch on updated requirements.txt, waits for patch review09:04
kashyap:-)09:05
mdboothAh, well. I submitted v1 of the patch over 6 months ago. I guess another 6 months won't hurt :)09:05
kashyapjohnthetubaguy: Do you have a comment on this thread (especially, shortening the release cycle to 2 months aspect) - http://lists.openstack.org/pipermail/openstack-dev/2015-February/057614.html09:05
*** lpetrut has joined #openstack-nova09:05
*** heyongli has joined #openstack-nova09:06
*** tellesnobrega has quit IRC09:07
*** heyongli has quit IRC09:07
mdboothStep 1 complete: https://review.openstack.org/18791809:08
*** dguitarbite has quit IRC09:09
*** dguitarbite has joined #openstack-nova09:09
*** dguitarbite has quit IRC09:10
openstackgerritPetrut Lucian proposed openstack/nova: Hyper-V: Fix SMBFS volume attach race condition  https://review.openstack.org/18767309:11
* danpb wonders if there's any cores around who can review a numa docs fix so we stop confusing our users https://review.openstack.org/#/c/178773/09:15
*** FL1SK has joined #openstack-nova09:15
*** claudiub has joined #openstack-nova09:17
*** paul-carlton has joined #openstack-nova09:20
*** oro_ has joined #openstack-nova09:21
*** afazekas has joined #openstack-nova09:22
*** lpetrut has quit IRC09:22
*** paul-carlton has quit IRC09:25
*** gcb has joined #openstack-nova09:27
fricklermdbooth: how about adding a reference to your nova patch into the commit message for the g-r update? and maybe also to the oslo.db update that introduced the function you need.09:33
*** gszasz has joined #openstack-nova09:34
mdboothfrickler: I could, and I considered that. However, I decided against it as that specific context seems irrelevant in the git history.09:35
mdboothIn fact, just the mention of update_match seems irrelevant. It's just an initial motivation.09:35
mdboothThere are presumably lots of potential users of new functionality.09:35
mdboothMore of a covering note thing.09:35
*** yingjun has quit IRC09:36
mdboothHowever, I could be convinced otherwise if it would speed the process.09:36
fricklermdbooth: well, obviously I can't promise it will really help in the end, but with the commit message in its current form, for me it raises the question "Where does he need that?" and it might help if that question had a direct answer09:39
openstackgerritMarkus Zoeller (markus_z) proposed openstack/nova: WIP: docs: Add bug handling description to developer reference  https://review.openstack.org/18757109:39
*** pixelbeat_ has joined #openstack-nova09:39
*** abhishekk has joined #openstack-nova09:44
*** paul-carlton has joined #openstack-nova09:44
*** ianbrown has joined #openstack-nova09:47
*** pradipta has quit IRC09:48
johnthetubaguykashyap: danpb: looking at the thread, we consider every commit to be a release, but we do have lots of freezes, which are awkward09:51
*** sudipto has quit IRC09:51
*** veena has quit IRC09:53
*** Kevin_Zheng has quit IRC09:55
*** Administrator_ has quit IRC09:55
markus_zmdbooth: About review https://review.openstack.org/#/c/148273/ What can I do there?09:56
*** ianbrown has quit IRC09:56
markus_zmdbooth: I think a change of the commit message would be sufficient to +1 it.09:57
*** lpetrut has joined #openstack-nova09:57
johnthetubaguydanpb: your email on release schedule changes, have you seen what ironic are looking to do? its an interesting twist on what you are proposing here09:57
*** otter768 has joined #openstack-nova09:58
*** dguitarbite has joined #openstack-nova09:59
markus_zjohnthetubaguy: Which email are you referring to?09:59
johnthetubaguydanpb: I have been doing some thinking about the release cycles, and honestly I think SemVer helps with a lot of these09:59
johnthetubaguymarkus_z: http://lists.openstack.org/pipermail/openstack-dev/2015-February/057614.html09:59
markus_zjohnthetubaguy: thanks09:59
*** paul-carlton has quit IRC09:59
sdaguemdbooth: you are correct, if you are submitting patches that use new oslo.db features, you should get g-r bumped first with that explanation. The minimum should be real.10:00
danpbjohnthetubaguy: yep, i saw their proposal, but I'm not a fan of the idea of doing feature based releases in general10:00
johnthetubaguydanpb: as in you prefer the time based releases?10:01
danpbjohnthetubaguy: i've seen too many projects go wrong with that, as they continually delay the release "just one more week" to let the feature get ready10:01
johnthetubaguyyeah, that so happens10:01
* kashyap notices SemVer = http://semver.org/10:01
danpbfrom what I've experianced strict time based releases are more effective - /if/ they are frequent enough10:01
*** tellesnobrega has joined #openstack-nova10:02
danpbwhich was why i suggested a 2 monthly release cycle, so we'd still match up with the 6 month cycle of other projects10:02
danpbbut be able to deliver a couple of extra intermediate releases10:02
openstackgerritBalazs Gibizer proposed openstack/nova: Add functional test for server group  https://review.openstack.org/14751410:02
johnthetubaguydanpb: so we do 6 weekly release right now, I am curious why they are not useful10:02
*** otter768 has quit IRC10:03
johnthetubaguyto be clear, they are not being consumed by people, so there must be something stopping that10:03
danpbwell if we tagged those 6 weekly releases as formal production deployable releases that'd work fine too10:03
mdboothmarkus_z: So you would like a link to the nova commit which uses it?10:03
johnthetubaguydanpb: so I am thinking we can do that now, as we move to semver10:03
danpbbut the messaging around them is that they are development snapshots which i think discourages people from using them10:03
johnthetubaguydanpb: gotcha10:03
johnthetubaguydanpb: I think the plan is liberty is 12.x and M is probably 12.x10:04
openstackgerritBalazs Gibizer proposed openstack/nova: scheduler:refactor group info in filter_properties  https://review.openstack.org/14827510:04
openstackgerritBalazs Gibizer proposed openstack/nova: api: add soft-affinity policies for server groups  https://review.openstack.org/14752010:04
openstackgerritBalazs Gibizer proposed openstack/nova: Add functional tests to soft affinity capabilities  https://review.openstack.org/14752110:04
openstackgerritBalazs Gibizer proposed openstack/nova: scheduler: add num of instance per host in a group  https://review.openstack.org/14827710:04
openstackgerritBalazs Gibizer proposed openstack/nova: object:count members per hosts in InstanceGroup  https://review.openstack.org/14751710:04
openstackgerritBalazs Gibizer proposed openstack/nova: scheduler: add soft-(anti-)affinity weighers  https://review.openstack.org/14751610:04
johnthetubaguydanpb: I am thinking we make liberty 1, something like 12.0.110:04
* gilliard notes happily that semver.org is semantically versioned10:04
johnthetubaguydanpb: and then liberty is 12.1.0 (the first stable release), then we start 13.0.0.dev110:05
*** dims_ has joined #openstack-nova10:06
danpbisn't that still giving  the impression that liberty 1 is a lesser thing compared to liberty-final10:06
johnthetubaguydanpb: I am thinking if we actually give every commit an automatic version, but only tag on the milestones, so each commit gets, like if we were on the 42nd commit since master opened for liberty it would be: 12.0.0.dev4210:06
johnthetubaguydanpb: so it kinda is, because the docs folks only really help with 12.110:07
johnthetubaguydanpb: I am thinking about upgrade here too10:07
johnthetubaguy12.0 -> 13.0 means we can drop live upgrade code at those moments10:07
johnthetubaguyand thats the point where we force people to upgrade10:07
johnthetubaguynow that doesn't have to be every 6 months, but we could start that way10:07
*** mohitsharma has quit IRC10:08
johnthetubaguydanpb: a way of looking at the milestones is that we are doing a mini tick/tock style release train10:09
*** markvoelker has joined #openstack-nova10:11
johnthetubaguydanpb: 12.0.1 is open for all sorts of features (risky stuff goes here), 12.0.2 is also open to all features, 12.0.3 is only priority features, 12.1 only adds bug fixes on 12.0.310:11
*** exploreshaifali has joined #openstack-nova10:12
danpbso that's more or less what we do now today10:12
dims_haypo: got a bunch of tests working python34 - https://review.openstack.org/#/c/187785/10:13
dims_oops he is not here10:13
mdboothdims_: Thanks!10:13
danpbwhich leads to a very bursty development cycle which creates instability due to crunch times when lots of big features are all shoved in at roughly the same time10:13
johnthetubaguydanpb: agreed, but thats probably easier to understand, and we can evolve that once we work out why we are doing what we are doing today (plus understanding what we are doing)10:13
dims_mdbooth: i went further to 0.10.0 :)10:13
danpba big part of my idea was to try to smooth out the cycle to eliminate the highs & lows10:13
mdboothdims_: tbh, I'm pretty sure I meant 0.10.010:13
danpbby making each time based release equivalent in the types of stuff accepted10:14
mdboothHowever, I couldn't work out how version numbers are calculated anyway10:14
dims_mdbooth: haha :)10:14
danpbso people don't feel them need to rush to shove in a big features when its not ready, just in order to avoid slipping another 6 months10:14
danpbiow, try and get rid of the idea that we have a specific time where we shove in big risky stuff10:15
*** markvoelker has quit IRC10:15
danpbas if we say we're allowed to shove in destablizing stuff at point X, it becomes a self-fullfilling prophecy10:16
johnthetubaguydanpb: agreed with the self-fullfilling prophecy, the "big risky stuff" is meant to be an exception, and generally not referring to a feature, but thats certainly not true any more10:17
johnthetubaguydanpb: so my take for this release, is that we should be better at communicating what we are doing and why, so we can have a better debate on what to do for M10:18
*** garyk has quit IRC10:18
johnthetubaguydanpb: its slower than I would want, but we get the feedback our users want stability over features, and thats what the process is optimising for, at some level10:18
*** garyk has joined #openstack-nova10:18
*** garyk has quit IRC10:19
danpbi guess if we can at least start improving the messaging around intermediate releases to encourage their use, that'd be a start in the right direction10:19
*** garyk has joined #openstack-nova10:19
danpbdon't have to change everything at once10:19
johnthetubaguydanpb: yeah, thats true10:19
johnthetubaguydanpb: I think when we started calling them a1 and a2 we broke the original messaging there10:20
johnthetubaguydanpb: I am cool with doing that, plus advertising the fact that every commit is "released" in some sense10:20
danpbbtw, another aspect of the proposal was that we stop marking specs against specific releases - allow specs to be submitted& reviewed & approved at any time and once approved allow their validity to remain in force for 12 months10:20
mdboothjohnthetubaguy: Ironically, though, stability can involve significant change without introducing new features.10:20
danpbso we avoid the tedium of having to re-submit a whole bunch of specs at the start of each cycle10:20
johnthetubaguydanp: then add that the 12.x -> 13.x is a forced upgrade point, to make sure we do live-upgrade with minimal technical debt10:21
sahiddanpb: +110:21
johnthetubaguydanpb: the backlog concept helps there somewhat10:21
johnthetubaguybut yeah, the current version to too heavy weight10:22
danpbmostly just to cut down on work we've creating for ourselves10:22
johnthetubaguymdbooth: you are totally right, the efforts to get more stability are quite de-stabilising, look at all the work to formalise the interface, rewrite cells so its more maintainable and a first class concept, plans to add task tracking and API, etc10:23
johnthetubaguydanpb: we have done a lot of that with abandoning milestone tracking, but yes, more of that is needed10:24
danpbyeah, that was an improvement10:24
johnthetubaguydanpb: just for the record, I think unlearning things is more important than adding process!10:24
johnthetubaguy(its also harder)10:24
*** sahid has quit IRC10:25
*** mwagner_afk has quit IRC10:26
*** Gomeler has quit IRC10:34
johnthetubaguydanpb: I am sending a proposal to the ML10:34
*** samueldmq has joined #openstack-nova10:37
samueldmqmorning10:37
openstackgerritLucas Alvares Gomes proposed openstack/nova: Ironic: Parse and validate Node's properties  https://review.openstack.org/17689010:38
*** qianli_ has quit IRC10:39
*** sudipto has joined #openstack-nova10:41
*** sudipto has quit IRC10:42
*** sudipto has joined #openstack-nova10:43
*** sudipto has quit IRC10:44
*** sudipto has joined #openstack-nova10:45
*** bzhao has quit IRC10:46
*** sudipto has quit IRC10:49
*** bzhao has joined #openstack-nova10:51
*** jyuso1 has quit IRC10:52
*** exploreshaifali has quit IRC10:54
*** garyk has quit IRC10:55
*** garyk has joined #openstack-nova10:55
* johnthetubaguy runs away for lunch10:56
*** garyk has quit IRC10:56
*** garyk has joined #openstack-nova10:56
*** vladikr has quit IRC10:56
openstackgerritgaryk proposed openstack/nova: VMware: add support for cores per socket  https://review.openstack.org/18794210:58
*** sdake has joined #openstack-nova11:02
*** flwang1 has joined #openstack-nova11:05
*** sudipto has joined #openstack-nova11:07
samueldmqsdague, hi, I'd like to talk about the dynamic policies + nova needs, you around ?11:07
sdaguesamueldmq: yep11:08
*** lyanchih has joined #openstack-nova11:08
samueldmqsdague, nice .. from what I understood, you main use case is to warn the user if he's changing the default and that may impact on the success of some operations11:09
sdaguesamueldmq: I also tried to put more details back on list posts11:09
samueldmqsdague, am I right?11:09
sdaguesamueldmq: that's one of the things that emerged as "we'd like to do that"11:09
samueldmqsdague, I saw a couple of messages in the ML11:09
samueldmqsdague, nice, what if we did that for all projects ?11:10
sdaguesamueldmq: I'm pro doing it for all projects11:10
samueldmqsdague, and the defaults (in code) would be managed by the policy management server11:10
sdaguethat's why I said "features we want in oslo.policy"11:10
*** exploreshaifali has joined #openstack-nova11:10
sdaguesamueldmq: ok, so where does the policy management server get it's seed of truth?11:10
samueldmqsdague, one of the ideas is to have a separate service for policy storage + management (changing the api's granularly, etc)11:11
*** lyanchih has quit IRC11:11
samueldmqsdague, I guess that would be in the code of the policy management service11:12
samueldmqsdague, that once started, will consider to load its database based on i) the defaults (probably in the code) + ii) the unified policy.json file11:12
sdaguehow does that work with Nova adding a feature, policy that goes with it, and it being CDed into production while all other components remain the same11:12
sdagueI feel like the unified policy.json file makes an implicit assumption of lock step deploys of all components all at once11:13
sdaguewhich ... isn't the real world11:13
samueldmqonce we have the unified policy + dynamic changes is accepted, services will need to update i) local policies for backwards compt + ii) unified policy11:14
samueldmqbut that's a different concern11:14
garykdanpb: thanks! will check if that works11:14
sdaguesamueldmq: I still don't understand how that anticipates version skew11:14
samueldmqsdague, so maybe I am not understanding your question11:15
samueldmqsdague, how do we keep the unified policy consistent with the inidividual ones from services ?11:15
samueldmqsdague, is that your question ?11:16
sdagueespecially when the components are upgrading on different schedules with new features being added all the time which add new policy bits11:16
samueldmqsdague, got it, I am not sure I have the complete answer for this question, but I will talk about this and define it better in the specs11:17
sdagueright, the staggered upgrade seems intractable if you assume that all policy starts from a central registry, instead of federated up11:18
*** lucasagomes has quit IRC11:18
sdagueit's kind of like saying "what if we remove all the db migrations from all the projects and have 1 db migration service instead"11:18
sdaguewhich works only if everything is going to lock step all the time11:18
samueldmqsdague, looks like we will need to have a new version of this service each time any service using it does have a new version11:18
sdaguesamueldmq: projects are designed to be Continuously Deployed11:19
sdagueso that's every git commit11:19
sdaguefor every project11:19
sdaguealso, there is no reason to assume it's linear. Rax has a 2 year old cinder, for instance.11:20
samueldmqsdague, hmm, good question then... I will bring this into discussion with other folks11:20
samueldmqsdague, yeah, and what if some api's have disappeared ?11:20
sdaguethis is why I wanted to get this into a wider discussion (and honestly, preferably on the mailing list, so we don't keep having point discussions and relays)11:21
samueldmqsdague, and you havve a 2 year old cinder, but running master on other components11:21
*** flwang1 has quit IRC11:21
sdagueyep11:21
samueldmqsdague, nice, got it11:21
sdaguewhich is why I think the only real answer is to have all projects that are going to be part of this implement a /policy end point to fetch their seed of truth11:21
sdaguehopefully I explained it reasonably on the list11:22
*** zul has quit IRC11:22
sdagueNova makes for a good test case for this because we've got a huge API, people do CD nova, and we're doing API changes with microversions (and potentially policy adds) many times a cycle11:22
sdagueI think that the original model of a unified file didn't account for much of that11:23
garykdanpb: with your suggestion of the topology do i need to change a scheduling filter?11:23
danpbnot with kvm at least11:24
openstackgerritAbhishek Kekane proposed openstack/nova: Extract cinder-client response to log request IDs  https://review.openstack.org/18794811:24
danpbany host can run vms with any cores/socket topology11:24
garykthe flavor has - "capabilities:cpu_info:topology:sockets": "2" - but the vanilla scheduling is failing11:24
danpbso there's no need to filter anything11:24
samueldmqsdague, yeah, at very least, if we go for that unified thing, changes in let's say nova which introduce new api's would be using 'depends-on' on the change in the policy management server11:24
sdaguesamueldmq: that's not good enough11:24
sdaguebecause you assume a time linear relationship11:25
sdaguenow in order to update nova, you have to deploy that policy server as well11:25
*** aix has quit IRC11:25
sdaguebut what if there were also depends on from Trove11:25
sdaguebut you aren't upgrading trove for 6 months11:25
samueldmqsdague, deployments do not use 'depends-on' :/11:25
*** rushiagr is now known as rushiagr_away11:25
sdaguesamueldmq: more importnatly, they shouldn't. A project needs to be able to roll forward independent of other projects11:26
sdaguethe policy server (which owns the policy source of truth) assumes a very specific deployment model that we've never required in OpenStack11:27
samueldmqsdague, if we go for implementing the /policy thing in each project11:28
samueldmqsdague, how do we handle old projects that don't know anything about the dynamic policies ?11:28
*** zul has joined #openstack-nova11:28
sdaguethey don't get dynamic policies until they support it11:28
garykdanpb: i get:  The capabilities '' couldn't be loaded from JSON: No JSON object could be decoded. I am trying to investigate. something is fishy11:28
samueldmqsdague, oh that's easy, they keep enforcing as they were doing11:28
sdaguebut they weren't going to get it anyway, because they would need code changes to fetch and enforce policy from keystone anyway11:29
samueldmqsdague, only get, enforcement is kept at the service side11:29
sdaguesamueldmq: sure11:29
danpbgaryk: i dunno what that flavor capabilities stuff is for - its not related to the cpu topology stuff i mentioned though11:29
samueldmqsdague, keystonemiddleware will be in charge of caching and updating the last policy for the endpoint it is service11:30
sdaguebut you still need code in a project to pull the dynamic computed policy back from keystone for checks11:30
samueldmqit is serving*11:30
sdagueok, sure11:30
samueldmqsdague, nice11:30
sdagueI also think this kind of approach is more big tent friendly11:30
*** ociuhandu has quit IRC11:30
garykdanpb: ok, thanks.11:30
samueldmqsdague, ++11:30
sdaguebecause any project that does this gets to have dynamic policy, and keystone doesn't need to know about it semantically. It just needs to comply to some interfaces11:31
samueldmqsdague, those points on the unified policy need to be discussed on the ML11:31
sdaguesamueldmq: yeh... so check out my last posts11:31
*** ferest has joined #openstack-nova11:31
samueldmqsdague, I will talk to ayoung, we should have a good answer for that, and then we reply the ML11:31
sdaguegreat11:31
ankit_agHi all, Jenkins is failing for os.utime() with Permission denied error on one of my patch https://review.openstack.org/#/c/185549/1, But it is not reproducible on local environment. So it looks like a jenkins issue for me. Is someone have any idea about this jenkins failure.11:32
samueldmqsdague, thanks11:32
sdaguethe trick of this being right is getting a very detailed specification for the /policy required interface, because the code implementing it will be different in different servers as they have different wsgi layers11:32
samueldmqsdague, ++11:32
ankit_agPlease refer Jenkins logs at http://logs.openstack.org/49/185549/1/check/check-tempest-dsvm-nova-v21-full/17c7e67/logs/screen-n-cpu.txt.gz#_2015-05-28_06_51_49_80911:32
samueldmqsdague, completely agree11:32
danpbgaryk: unless you are doing strict cpu pinning with no over commit, it is pointless trying to match guest cores/sockets to host cores/sockets topologies11:32
danpbas you are not going to get any performance benefit that isn't wiped out by the over commit usage11:33
garykdanpb: the problem is that the scheduler expects the host cpu info in a json format.11:33
samueldmqsdague, policies change in different timings for different projects, unifying them may be a trap when services changes its interfaces11:33
garykthe vmware driver does not return this as the VC does the scheduling. so we have a little conflict here11:34
garyki will see if we can report the sockets.11:34
danpbwhy are you trying to set schedular filters at all though11:34
samueldmqsdague, services versions would be very dependent on the new policy  management service version11:34
samueldmqsdague, I think that summarizes well the issue we have been discussing11:34
garykdanpb: i have not changed the scheduling filters. they are the defautl ones11:34
samueldmqsdague, :)11:34
danpb"capabilities:cpu_info:topology:sockets": "2"11:34
danpbthat's not a default setting is it11:35
garykdanpb: that is following your suggestion on review https://review.openstack.org/#/c/187942/11:35
garykwe need to pass the number of sockets to the driver.11:35
markus_zmdbooth: sorry, I was AFK. I meant that the commit message could provide more details what is solved by that change.11:36
*** ferest_ has joined #openstack-nova11:36
garykdanpb: this setting does not work with the scheduling filter ComputeCapabilitiesFilter11:36
markus_zmdbooth: Does it solve bug 1412436 or should the test be active when bug 1412436 is solved?11:36
openstackbug 1412436 in OpenStack Compute (nova) "Race in instance_create with security_group_destroy" [Low,Confirmed] https://launchpad.net/bugs/141243611:36
danpbgaryk: huh ? it shouldn't need anything more than the 2 lines of code i illustrated afaik11:37
markus_zuh, nice feature11:37
danpbthe capabilities:cpu_info:topology:sockets  thing is filtering host cpu topology - that's completely independent & unrelated to the question of setting guest CPU topology11:37
fricklerankit_ag: the same check passed before you did your recheck, so hopefully another recheck might solve it. there also seems to have been some issues with the nodepool earlier today11:38
sdaguesamueldmq: :)11:38
danpba host with 4 sockets and 2 cores per socket is fine hosting a guest with virtual topology 2 sockets and 4 cores11:38
sdaguesamueldmq: yeh, I think that's the crux of it. And that we actually want to free people from having to do lock step upgrades of everything all at once, not make them have to do more of that.11:39
*** ferest has quit IRC11:39
garykdanpb: https://github.com/openstack/nova/blob/master/nova/scheduler/filters/compute_capabilities_filter.py#L39 and https://github.com/openstack/nova/blob/master/nova/virt/vmwareapi/driver.py#L46111:39
*** ferest_ has quit IRC11:39
sdagueso any future direction needs to assume components will be less likely to all upgrade at once, not more11:39
jaypipesgood morning, all :)11:39
*** bzhao has quit IRC11:39
gilliardHi jaypipes11:39
garykdanpb: so the scheduling filter is expecting us to return the 'sockets' which we do not11:40
ankit_agfrickler: Thanks for your reply ! I have added recheck for 2 times and each time a different test fails, but the failure reason on n-cpu screen logs is same . i.e. Permission denied error11:40
danpbgaryk: so just don't set anything that triggers that filter11:40
sdaguejohnthetubaguy: when get back around for some interactive time, let me know. I need to talk through this API writeup to get past my current writers block on what's next.11:41
ankit_agfrickler: And that too with the code I have added os.utime(), So I think even jenkins passes on my patch, there may be other gate failures with change11:41
garykthe extra flavor spec '"capabilities:cpu_info:topology:sockets": "2" triggers that filter :(11:41
danpbso do not set that extra spec then11:42
*** markvoelker has joined #openstack-nova11:42
garykdanpb: that is why I passed vmware:cores_per_scoket - which cotradicts what you suggested inthe review11:42
danpbthat change i suggested does not need / use that extra spec11:42
garyklet me look again. maybe i have missed something11:43
garykdanpb: so it should be "hw:cpu_sockets" ?11:44
fricklerankit_ag: ah, I didn't look into your patch deeply and assumed the error was completely unrelated11:45
garykdanpb: that works. thanks!11:45
danpbgaryk: that sets the preferred number of sockets11:45
fricklerankit_ag: but it this case it is not unlikely that your code either has a new race condition or just triggers one that has existed already before11:46
danpbif you want to enforce it more strongly you can set  cpu_max_sockets instead or as well11:46
danpbthere's more docs here  http://specs.openstack.org/openstack/nova-specs/specs/juno/implemented/virt-driver-vcpu-topology.html11:46
garykdanpb: yes, that is what i am looking for. cool. thanks11:46
*** markvoelker has quit IRC11:46
*** baoli has joined #openstack-nova11:47
ankit_agfrickler: Yes IMO it might be related to jenkins environment because I have tried to reproduce it for all the possible race conditions on my local environment11:47
ankit_agfrickler: looks like this is relaated to the user's permissions on jenkins11:48
*** baoli has quit IRC11:49
*** sahid has joined #openstack-nova11:50
*** baoli has joined #openstack-nova11:50
*** dprince has joined #openstack-nova11:54
samueldmqsdague, yes, I agree, thanks11:56
*** tobe has quit IRC11:56
*** Raildo has joined #openstack-nova11:56
*** irenab has quit IRC11:56
*** irenab has joined #openstack-nova11:57
*** tobe has joined #openstack-nova11:57
*** mohitsharma has joined #openstack-nova11:57
sdaguesamueldmq: awesome. thanks for engaging. And lets get as much of the conversation driven on the mailing list so we can include as many folks in different timezones as possible. Especially as I think this has pretty far reaching implications for all openstack projects.11:58
openstackgerritDavanum Srinivas (dims) proposed openstack/nova: Enable python34 tests for nova/tests/unit/scheduler/test*.py  https://review.openstack.org/18796811:58
*** sudipto has quit IRC11:58
*** lucasagomes has joined #openstack-nova11:59
*** otter768 has joined #openstack-nova11:59
samueldmqsdague, ++ makes sense11:59
samueldmqsdague, I am glad you want to be involved and make it as good as we can11:59
*** aix has joined #openstack-nova12:00
markus_zkashyap: bauzas: We talked last week about the bug triaging process. I tried to clarify my newbie questions with the openstack wiki and packed my understanding in review 187571. Maybe you could have a look some day, no hurry.12:01
*** markvoelker has joined #openstack-nova12:01
kashyapYep, will read through.12:02
sdaguesamueldmq: yeh, happy to help. It's a really cool future vision of openstack to have these dynamic policies, I really want that to succeed. Just want to make sure it works with all the other things we also need.12:02
markus_zkashyap: Thank you!12:02
*** tobe has quit IRC12:02
samueldmqsdague, ++12:02
kashyapmarkus_z: Thanks for writing that up.12:03
bauzasmarkus_z: ack12:04
*** otter768 has quit IRC12:04
bauzasadd me as CC please12:04
*** lucasagomes is now known as lucas-hungry12:05
garykdanpb: all i can say is 'NICE'. Thanks!12:05
markus_zsdague: Some months ago I "discovered" the "OpenStack Bootstrapping Hour" on Youtube. I liked the format and it helped me a lot. Unfortunately there are no new sessions. May I ask what the current state of play is?12:07
markus_zbauzas: Thanks! done12:07
kashyapmarkus_z: I think it's waiting on volunteers :-)12:08
kashyapI.e. someone willing to commit the time to do it in their area of expertise.12:08
sdaguemarkus_z: the current state of play is .... I have a 8 month old daughter now, and had to prune some of my organizing activities12:08
*** VW has joined #openstack-nova12:08
*** lennyb has joined #openstack-nova12:09
bauzassdague: count me in12:09
sdaguegiven that I was organizing and hosting them all. I'd be happy to host again, but kind of need a producer to help line up content and help with logistics12:09
markus_zsdague: Oh, OK, that shifts some priorities :)12:09
*** rfolco has joined #openstack-nova12:09
bauzassdague: I can discuss that with mriedem12:10
*** derekh has quit IRC12:10
kashyapmarkus_z: You might want to rebase that change, there's a merge conflict.12:14
markus_zsdague: I was just curious and wanted to let you know that there is an audience for that type of format.12:14
sdaguemarkus_z: cool, thanks12:14
*** lennyb_ has quit IRC12:14
markus_zkashyap: Yepp, I'm rebasing. jogo is making changes faster than me :)12:15
openstackgerritgaryk proposed openstack/nova: VMware: add support for cores per socket  https://review.openstack.org/18794212:15
*** IanGovett has joined #openstack-nova12:17
*** eharney has joined #openstack-nova12:18
*** amotoki has joined #openstack-nova12:18
openstackgerritMarkus Zoeller (markus_z) proposed openstack/nova: WIP: docs: Add bug handling description to developer reference  https://review.openstack.org/18757112:18
markus_zjogo: I've seen you do a lot of work in the devref right now. Could you maybe have a look at review 187571 and tell me where this would make sense in your opinion?12:20
markus_zkashyap: From a previous patch set (same content): http://docs-draft.openstack.org/71/187571/4/check/gate-nova-docs/daa7215//doc/build/html/bugs.html12:21
*** rotbeard has quit IRC12:22
*** ivasev has joined #openstack-nova12:22
kashyapThanks. From a cursory look, the state flow looks quite helpful :-)12:26
*** artom has joined #openstack-nova12:26
markus_zYeah, the "blockdiag" package provides some cool features, I think. :)12:27
*** aysyanne has joined #openstack-nova12:28
*** artom_ has joined #openstack-nova12:29
*** artom has quit IRC12:31
*** kiranr has joined #openstack-nova12:34
*** kiran-r has quit IRC12:34
*** baoli has quit IRC12:34
openstackgerritDavanum Srinivas (dims) proposed openstack/nova: Enable python34 tests for nova/tests/unit/scheduler/test*.py  https://review.openstack.org/18796812:36
alex_xujaypipes: sdague I send the microversion spec to api-wg https://review.openstack.org/187112, but there is a little depend on what decision make on python-novaclient. Do we follow the ironic client implementation? If yes, we need add two more headers for api.12:36
jaypipesalex_xu: which two more headers?12:37
mdboothmarkus_z: Ah, I think we were talking about different bugs. The one you were referring to has dropped off my care-enough-to-push list :) I'm not going to devote further time to it, so rather than waste anybody else's time I might as well abandon it.12:38
alex_xujaypipes: if the requested version won't supported by server, ironic return the min and max vesion by http headers. that is a little different with nova12:38
*** kiranr has quit IRC12:39
jaypipesalex_xu: I see...12:40
jaypipesalex_xu: I would not be opposed to including that behaviour in Nova.12:40
markus_zmdbooth: That's a pity. I feel a bit responsible because of my -1.12:41
*** ociuhandu has joined #openstack-nova12:41
*** iamjarvo has joined #openstack-nova12:41
alex_xujaypipes: ok, got it. should I waiting for python-novaclient's nova-spec send out?12:41
mdboothmarkus_z: Nah, don't worry about it.12:41
markus_zmdbooth: Despite of the effort you want to spent on that, do you still think that this is an issue which should be solved?12:42
jaypipesalex_xu: sorry, not sure what you are asking... are you asking if you should wait until the API-WG patch is merged before proceeding to add functionality to nova to return those two headers?12:42
markus_zI have to admit, I didn't quite understand the impact.12:43
mdboothmarkus_z: It's just 1 of a zillion races.12:43
mdboothI saw it, so I attempted to document it in code.12:44
mdboothI couldn't think of a way to fix it.12:44
sdaguealex_xu: I'm confused what other headers are needed12:44
*** artom_ has quit IRC12:44
alex_xujaypipes: at summit we said we need nova-spec for python-novalcient microversion supported, so I think maybe we should waiting for what propose in that nova-spec. avoid if there are something different propose for version negotiate.12:44
mdboothmarkus_z: The impact is probably pretty low, tbh.12:44
alex_xusdague: the use-case 7A: https://github.com/openstack/ironic-specs/blob/master/specs/kilo/api-microversions.rst12:44
sdaguealex_xu: yeh, I signed up for doing that, but it's going to be next week at the earliest12:44
jaypipesalex_xu: but this is a nova server thing, not a novaclient thing, right?12:45
sdaguealex_xu: so... I think the return content should not be headers, it should be in the body12:45
sdaguethat seems really odd to return as headers in those cases12:45
alex_xujaypipes: ^ ^ see... sdague have another thought :)12:46
*** abhishekk has quit IRC12:46
markus_zmdbooth: I see, thanks for clarifying that for me!12:46
gilliardyes, sdague, to make it analagous with content-negotiation, having the min/max as an entity makes sense.12:47
gilliardrather than in headers12:47
sdagueyeh, agree with gilliard12:47
*** artom has joined #openstack-nova12:48
gilliarddoes ironic use headers in that case?12:48
sdagueyes, apparently12:48
alex_xusdague: emm...maybe we should refer to sub-error-code if the api-wg work out that.12:48
jaypipesgilliard: isn't content-negotiation done via HTTP request headers?12:48
sdaguejaypipes: not errors12:49
gilliardyes, but if content negotiation fails, the reason is explained in the body of the response.12:49
*** artom_ has joined #openstack-nova12:49
jaypipesah, gotcha, yes.12:49
sdagueright, you can specify multiple accepts with weights and the server returns best thing it can, but if it can't fullfill the request the error is in the body12:49
sdaguewe specifically didn't do the range / weight thing because... its' really complicated12:50
sdagueand requires way more conditional code on both sides12:50
sdagueit was in an early version of the microversions spec for nova12:50
johnthetubaguysdague: +1 baby steps forward, and see what clients actually need and why12:51
sdagueso, I think the only thing to do here is recommend that this gets done in the body instead of headers, and probably ask the ironic team if there was a compelling reason they chose headers here12:52
sdaguealex_xu: can you circle with them?12:52
openstackgerritRitesh proposed openstack/nova: Users will not be allowed to delete instance during snapshot  https://review.openstack.org/18594212:52
*** jcoufal has quit IRC12:52
*** artom has quit IRC12:52
alex_xusdague: ok, no problem12:52
sdaguealex_xu: it's probably worth defining in your recommendation a standard json payload with the mix / max that would be returned12:52
*** ankit_ag has quit IRC12:53
sdagueI think that could be done without needing the whole sub error system sorted, which feels like it's probably another cycle at least12:53
*** rwsu has joined #openstack-nova12:54
alex_xusdague: yes, agree we needn't whole sub error system now, just need a format good for add sub error system in the future12:54
*** jistr is now known as jistr|mtg12:57
*** derekh has joined #openstack-nova12:58
*** VW has quit IRC12:59
*** tsekiyama has joined #openstack-nova13:01
*** Raildo_ has joined #openstack-nova13:04
*** tsekiyama has quit IRC13:04
*** mdrabe has joined #openstack-nova13:05
*** kylek3h has joined #openstack-nova13:06
*** Raildo has quit IRC13:07
*** lucas-hungry is now known as lucasagomes13:08
*** unicell1 has joined #openstack-nova13:09
*** unicell has quit IRC13:10
*** Raildo__ has joined #openstack-nova13:12
*** nkrinner has quit IRC13:15
*** Raildo_ has quit IRC13:15
*** gholler has joined #openstack-nova13:17
*** dave-mcc_ has joined #openstack-nova13:19
garykdanpb: did the patch with the image metadata object get approved?13:20
*** Raildo__ has quit IRC13:21
*** dave-mccowan has quit IRC13:23
*** zzzeek has joined #openstack-nova13:23
dims_johnthetubaguy: i reworked the oslo-config-generator review a whole lot after the feedback from you and jogo - so please do take a quick look when you get a chance - https://review.openstack.org/#/c/180013/13:27
*** burt has joined #openstack-nova13:27
belmoreiraHi, we are running Juno and now I noticed that on "vcpu placement" the "cpuset" is explicitly defined13:28
johnthetubaguydims_: honestly, I still don't like maintaining those separate lists when really we just want to include all options13:28
johnthetubaguydims_: do you know what the reason was to do it that way?13:28
danpbgaryk: still waiting another core unfortunately13:28
belmoreirathis is causing us some problems with "cpuset" overcommitment13:28
dims_johnthetubaguy: yes, so we can generate multiple configuration files for different services with different set of options in each13:29
dims_johnthetubaguy: once this base lands, i can try to do that13:29
belmoreiracan't find any bug related. Do you have any ideia? Can the "cpuset" be disabled?13:30
dhellmannjohnthetubaguy, dims_ : also because the code scanning was fragile13:30
dims_ack dhellmann: ++13:30
johnthetubaguydims_: so thats a good point, but in theory can't we just want the code from the cmd point13:30
danpbperhaps dansmith will be able to do another review of it soon https://review.openstack.org/#/c/76234/13:31
johnthetubaguydhellmann: dims_: I just have the feeling that list will be just as fragile, to be honest, while it was nasty, I quite liked when we change the sample in tree when adding a config to prove it gets updated13:31
johnthetubaguybut I haven't tried to maintain the scanning code, so I am not really in a position to comment13:32
*** jecarey has joined #openstack-nova13:32
dims_so johnthetubaguy please take our word for it? :)13:32
dhellmannare we talking about how the discovery works, or are we talking about gating on a generated file in git?13:32
*** mriedem_away is now known as mriedem13:32
johnthetubaguydhellmann: the former really13:33
dansmithdanpb: I was waiting for garyk to drop his -113:33
*** jcoufal has joined #openstack-nova13:33
*** moshele has quit IRC13:33
dims_johnthetubaguy: we are not going to change back to the old discovery mechanism, so this is it or nova keeps doing what its doing13:34
dhellmannjohnthetubaguy: ok. and the problem is that now we have to be explicit about what options are defined?13:34
garykdanpb: i am currently addressing your comments. i'll drop the -1 if it does not break the code with this landing.13:34
garykdanpb: dansmith: i will post a patch above this one addressing my concerns13:34
johnthetubaguydhellmann: yeah, I worry about missing stuff when its added, tests might help there13:34
dansmithgaryk: well, if it really needs addressing then the patch needs to come first, right?13:35
dansmithgaryk: we don't want to break vmware knowingly even with a patch in the queue13:35
dhellmannjohnthetubaguy: yeah, some of the other projects have unit tests to verify that plugins define the options they should, but I don't think that's necessarily going to resolve your concern here13:35
garykdansmith: i will post the patch in a few minutes13:35
dansmithokay thanks13:35
*** baoli has joined #openstack-nova13:35
danpbgaryk: this patch can't break anything, since it merely adds the object which nothing uses13:35
*** moshele has joined #openstack-nova13:36
johnthetubaguydhellmann: yeah, a good test would help a lot I think13:36
danpbonly once we convert drivers over to use the object is there a risk of stuff breaking13:36
garykdanpb: understood13:36
dhellmannjohnthetubaguy: what would the test do?13:36
garykdanpb: my concern is that we do not want to break existing api usage… that will be challenging.13:36
danpbnothing about this image meta stuff is easy :-)13:37
johnthetubaguydhellmann: sorry, in a meeting, let me get back to you, just something that proves it changes, and the variable is present might do the trick, not 100% sure13:37
dansmithdanpb: we also don't want to merge something that we just have to change, which I guess is what I thought he was saying might be necessary13:37
danpbi've found and fixed soooooo many bugs in libvirt from doing this work13:37
bauzasgaryk: honestly it has to be done in a separate change since it's not needed yet...13:37
dhellmannjohnthetubaguy: ok -- doing this explicitly gives us a few features for other projects that we don't want to lose by going back to scanning code, so we need *some* sort explicit list13:38
*** iamjarvo has quit IRC13:38
dhellmanndims_: did you have a version of that patch using more than one entry point?13:38
garykbauzas: i was not sure that we needed to be honest about it all.13:39
danpbgaryk: so we just need to extend the constants in the network model module to cover the missing ones13:39
openstackgerritAndrey Pavlov proposed openstack/nova: Extract helper method to get image metadata from volume  https://review.openstack.org/18746113:39
garyki have changed my score as this is blocking a number of thinsg and i am in favor of the change. i have some concrens which i think that we agree can be done in another patch which i am doing13:39
garykdanpb: agreed. that is what i am doing now13:39
dims_dhellmann: the patch has 14 entrypoints - https://review.openstack.org/#/c/180013/12/setup.cfg,cm13:39
danpbok, ping me when you have it ready for review13:39
garykok, will do. thanls13:40
*** gokrokve has joined #openstack-nova13:41
dansmithgaryk: are they changes that will affect the object schema? meaning, would they be things that would technically require a version bump/13:41
dhellmanndims_: oh, I should have looked, I thought this draft just had one13:41
mdboothDoes anybody here understand pbr?13:41
dansmithgaryk: if not, then se should go ahead and merge this.. if so, then we should wait, IMHO13:41
danpbdansmith: they'd be adding some further keys to an enum field13:41
*** sudipto has joined #openstack-nova13:41
dansmithokay13:41
danpbso they'd invalidate the version hash, but don't think we need to bump the version number for that13:41
*** VW has joined #openstack-nova13:42
garykdansmith: i will post in 5 min.13:42
*** whenry has joined #openstack-nova13:42
dansmithdanpb: normally we would, but not in this case until it's used. However, unless it's laborious, it's still cleaner to do that than to have an unversioned change in the history we have to explain13:42
* mdbooth is about to start some pbr rtfs13:42
danpbok13:43
dhellmannmdbooth: you should come chat in #openstack-oslo13:43
dansmithgaryk: okay, thanks13:43
*** dan_ has joined #openstack-nova13:43
*** dan_ is now known as Guest8367913:43
kashyapmdbooth: Maybe Alan Pevec from stable maint might know.13:43
*** changbl_ has joined #openstack-nova13:43
dansmithdanpb: aside from that, I'd be +W, so lets just peek at his changes and either merge or squash and assume this will be merged today one way or the other :)13:44
mriedemmdbooth: if you have pbr questions i'd ask lifeless or mordred13:45
*** IanGovett has quit IRC13:46
mdboothmriedem: dhellmann has directed me to #openstack-oslo13:46
*** apuimedo has quit IRC13:47
dims_johnthetubaguy: do i wait for someone else's input on that oslo-config-generator patch? about ready to give up on it13:48
*** vladikr has joined #openstack-nova13:50
openstackgerritgaryk proposed openstack/nova: Network: add in additional networks types  https://review.openstack.org/18800813:50
garykdansmith: danpb: ^13:50
mordredwhat did I do?13:50
dansmithgaryk: the point of this work is to get rid of things like VirtualE1000 in the external API, right?13:51
*** cfriesen has joined #openstack-nova13:51
dansmithgaryk: so that and PCNet32 at least need to not be added13:52
*** xyang1 has joined #openstack-nova13:52
*** mwagner_afk has joined #openstack-nova13:52
dansmither, wait13:52
garykdansmith: we are unable to do that. there are people who have this stuff defined as image metadat data today in production.13:52
danpbdansmith: yes indeed, the names need to be canonicalized a bit13:52
dansmithmaybe this is southbound?13:52
danpbgaryk: the enumfield has the ability to canonicalize legacy names for back compat13:52
dansmithright13:52
dansmithgaryk: one of the intents of this work is to stop that delta from widening and getting people to use standard names for standard things going forward13:53
danpbgaryk: eg see SCSIModel at line 425 in this  https://review.openstack.org/#/c/76234/36/nova/objects/fields.py13:53
*** nelsnelson has joined #openstack-nova13:53
*** oro_ has quit IRC13:54
garykdanpb: ok, understood.13:54
*** BadCub_Hiding is now known as BadCub13:54
dansmithdanpb: I'm not sure what to do about the vmxnet, vmxnet3, and sriov ones though13:55
garykwe need to create new values for those13:55
*** claudiub has quit IRC13:55
danpbdansmith: garyk i've commented on the review showing how we can deal with it13:57
*** gokrokve has quit IRC13:57
garykdanpb: :) i was already doing what you suggested13:58
dansmithugh13:58
*** gokrokve has joined #openstack-nova13:58
dansmithgaryk: is vmxnet3 incompatible with vmxnet? or is it just newer such that an older driver will still work with it?13:58
garykdansmith: they are different drivers. the vmxnet3 is newer13:59
garykwe have people using both13:59
*** jistr|mtg is now known as jistr13:59
*** dulek has quit IRC14:00
*** coolsvap is now known as coolsvap|afk14:00
*** otter768 has joined #openstack-nova14:00
*** ildikov has quit IRC14:00
*** tonytan4ever has joined #openstack-nova14:00
danpbdansmith: vmxnet is to vmxnet3 as  e1000 is to e1000e, so it makes sense for them to be separate14:00
dansmithdanpb: the e1000 driver will drive an e1000e though, won't it?14:01
dansmithhttp://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100180514:02
dansmiththis says they're not compatible though, so yeah, I guess14:02
danpbwell that's a guest OS question14:02
*** dboik has joined #openstack-nova14:02
danpbthis is about host side hardware setup14:02
*** Kennan2 has quit IRC14:02
dansmithdanpb: no, I meant I thought that e1000e just added some new transfer modes or something such that any e1000 driver should be able to drive the e1000e without noticing14:02
*** moshele has quit IRC14:03
*** dprince has quit IRC14:03
danpbaccording to that doc, we ought to have vmxnet2 as well14:03
*** dprince has joined #openstack-nova14:03
*** apmelton_away is now known as apmelton14:03
*** Kennan has joined #openstack-nova14:03
*** ildikov has joined #openstack-nova14:03
*** moshele has joined #openstack-nova14:04
* danpb has a new found level of hatred for mox and the incredible difficulty in figuring out its failures14:04
danpbsomeone needs to take a chainsaw to unit tests and rip out all trace of mox14:05
*** otter768 has quit IRC14:05
mdboothdanpb: I'll buy you a subscription to chainsaw monthly.14:05
*** mpavone has quit IRC14:06
garykdanpb: i am not aware of anyone using that14:06
dansmithdanpb: so you're just going to squash this into your patch yes?14:06
danpbdansmith: i guess so14:07
*** armax has joined #openstack-nova14:08
dansmithI'm kinda conflicted on adding vmxnet2, etc, because I really wish we could be a little more opinionated and not support ancient things that nobody has any reason to actually use14:08
*** thangp has joined #openstack-nova14:08
kashyapWhat is recommended to invoke unit tests locally? Between: `tox -e py27 nova.tests.unit.virt` and  ./run_tests.sh nova.tests.unit.virt14:08
dansmithat least enumerating the supported values is a major step, so I should be less annoyed, but..14:08
sdaguekashyap: tox -e py27 nova.tests.unit.virt14:09
sdagueor use my fun emacs integration :)14:09
kashyapsdague: Thanks. Why `tox` as opposed to run_tests.sh (which uses virtualenv I guess)?14:09
* kashyap goes to rtfm of `tox`14:10
sdaguevirtual env isolation mostly14:10
sdagueit's a handy common dispatch point given the amount of other setup that's done14:10
sdagueand it's repeatability is high, as you don't need to setup your local system in advance, which means less support requests for people setting up unit tests14:11
*** hshiina has quit IRC14:11
*** nickchase has joined #openstack-nova14:12
*** bkopilov has quit IRC14:12
*** sudipto has quit IRC14:12
danpbsdague: is it possible to run a single test with tox ?14:12
kashyapOkay, /me recalls seeing a thread about "fastest way to run tests" or something. Goes to look for that.14:12
sdagueyes, if you name it correctly14:12
* danpb currently does "$ .venv/bin/python -m testtools.run nova.tests.unit.virt.libvirt.test_driver.LibvirtDriverTestCase.test_finish_revert_migration_preserves_disk_bus"14:12
sdagueit's just passing it down to testr14:13
dansmithdanpb: usually just 'tox -epy27 name-of-test' is enough14:13
dansmithdanpb: if the test is reasonably uniquely named14:13
kashyapI remember learning something like this to run individual tests:  $ python -m testtools.run nova.tests.unit.virt.libvirt.test_driver14:13
* danpb tries it14:13
sdaguedansmith: well, it's a regex, so it can often match more names14:13
*** Gomeler has joined #openstack-nova14:13
kashyapdansmith: Thanks for the new tip14:13
dansmithsdague: right, hence the uniqueness constraint14:13
*** triggerz is now known as tobasco14:14
danpbthat's why people name test cases like     test_this_is_the_feature_I_want_test_no_really_this_one_not_that_other_one14:14
sdagueso, in all seriousness, if you are running emacs, I did build a tool that does inline testing - https://github.com/sdague/stacktest14:14
tobascothe fuck is up with this vif_type=binding_failed, upgraded from juno to kilo and shit breaks like hell14:14
sdaguewhich is basically just doing what danpb is doing with calling testtools appropriately14:15
*** FL1SK has quit IRC14:15
sdagueusing tox on single tests is kind slow because it still does testr discovery14:15
danpbsdague:  ah nice emacs magic !14:16
sdaguedanpb: yep, based on the nosetests module that did similar things14:16
kashyapdanpb: lol (About test case naming)14:16
*** sahid has quit IRC14:17
bauzasfor a direct run of a single test, I'm using ./runtests.sh -d14:17
bauzaswhich prevents the testr discovery14:17
garyktobasco: not sure that the explitives will help14:19
beaglestobasco: I'm assuming you are running neutron (afaik, that's the only place that comes from). It is a somewhat unfortunate mechanism to convey a potentially transient port binding failure in neutron. You might want to look over at neutron for reasons it might be happening. There is also a related patch up for review that attempts to fail early14:19
beaglestobasco, https://review.openstack.org/#/c/156112/14:19
beaglestobasco, so that might give you some more context14:19
*** mtanino has joined #openstack-nova14:20
openstackgerritgaryk proposed openstack/nova: Network: add in additional networks types  https://review.openstack.org/18800814:20
garykdanpb: dansmith: i think i need to write tests for the conversions.14:21
*** mdbooth has quit IRC14:21
tobascobeagles: ty im just scratching my head over here trying to get it working, upgrades is a pain in the ass14:21
* beagles nods14:22
dansmithsdague: can haz +W? https://review.openstack.org/#/c/187690/14:24
danpbgaryk: yep, would look good with a test case in  test_fields.py14:24
garykdanpb: sure, i am on it14:24
*** zz_jgrimm is now known as jgrimm14:25
*** david-lyle has quit IRC14:25
*** tsekiyama has joined #openstack-nova14:27
*** derekh has quit IRC14:28
openstackgerritArtom Lifshitz proposed openstack/nova-specs: Support --all-tenants in floatingips-list  https://review.openstack.org/17138914:29
*** VW has quit IRC14:30
*** rwsu has quit IRC14:31
kashyaptobasco: From about a year ago, a test-only, workaround (note: 'vif_plugging_is_fatal = False' is *not* useful if you want to setup credentials b/n Nova <-> Neutron) - https://www.redhat.com/archives/rdo-list/2014-May/msg00079.html14:31
*** iamjarvo has joined #openstack-nova14:32
*** iamjarvo has quit IRC14:32
*** iamjarvo has joined #openstack-nova14:33
openstackgerritjaveme proposed openstack/nova: VMware: Prefer ShutdownGuest to PowerOffVM_Task  https://review.openstack.org/18163414:33
* bauzas writes too many emails today14:35
*** exploreshaifali has quit IRC14:36
openstackgerritgaryk proposed openstack/nova: Network: add in additional networks types  https://review.openstack.org/18800814:36
garykdanpb: dansmith: with tests et al14:36
*** barra204 has quit IRC14:36
*** shakamunyi has quit IRC14:36
* dansmith thinks this should be sqashed14:37
dansmithor squashed even14:37
openstackgerritDaniel Stelter-Gliese proposed openstack/nova: Extract device path after connecting volume when booting LXC instances from volumes  https://review.openstack.org/18802514:38
*** raildo has joined #openstack-nova14:39
garykdanpb: do you want to squash or should i?14:39
*** VW has joined #openstack-nova14:39
*** artom_ is now known as artom14:40
*** lan_ has joined #openstack-nova14:40
*** Maike has quit IRC14:41
*** Jeffrey4l has joined #openstack-nova14:41
danpbgaryk: i'll take care of it14:42
*** afazekas has quit IRC14:43
garykdanpb: tx14:43
openstackgerritQin Zhao proposed openstack/nova: Ensure to store context in thread local after spawn/spawn_n  https://review.openstack.org/18314414:44
*** VW has quit IRC14:44
*** lan__ has joined #openstack-nova14:45
*** lan_ has quit IRC14:48
*** nihilifer has quit IRC14:50
*** nikhil_k is now known as nikhil_k-brb14:52
*** VW has joined #openstack-nova14:53
openstackgerritAlvaro Lopez Garcia proposed openstack/nova: scheduler: weight nodes based on their cached images  https://review.openstack.org/17330414:54
*** agireud has joined #openstack-nova14:58
*** dguitarbite has quit IRC14:59
openstackgerritDaniel Berrange proposed openstack/nova: objects: define the ImageMeta & ImageMetaProps objects  https://review.openstack.org/7623415:01
*** gholler has quit IRC15:05
*** lan_ has joined #openstack-nova15:05
*** kiran-r has joined #openstack-nova15:05
*** Gomeler has quit IRC15:05
*** Gomeler has joined #openstack-nova15:07
*** lan__ has quit IRC15:08
*** mdrabe has quit IRC15:09
*** changbl_ has quit IRC15:10
*** jgrimm is now known as zz_jgrimm15:10
*** iamjarvo has quit IRC15:10
*** changbl_ has joined #openstack-nova15:11
*** kiran-r has quit IRC15:13
*** iamjarvo has joined #openstack-nova15:14
*** gholler has joined #openstack-nova15:16
*** marun has joined #openstack-nova15:16
openstackgerritDavanum Srinivas (dims) proposed openstack/nova: Enable python34 tests for nova/tests/unit/scheduler/test*.py  https://review.openstack.org/18796815:17
*** mestery_ has joined #openstack-nova15:17
*** ddaskal has joined #openstack-nova15:18
*** rwsu has joined #openstack-nova15:19
*** mestery has quit IRC15:20
sdaguedanpb: looking15:20
*** mdbooth has joined #openstack-nova15:20
sdagueoh, markmcclain got you already15:21
dansmithsdague: yep, thanks15:21
*** claudiub has joined #openstack-nova15:22
*** artom has quit IRC15:23
*** alex_klimov has quit IRC15:23
*** alexpilotti has quit IRC15:24
*** jecarey has quit IRC15:24
openstackgerritWalter A. Boring IV (hemna) proposed openstack/nova-specs: Use os-brick library  https://review.openstack.org/18436015:25
dansmithjaypipes: danpb just added a couple more vmware-specific values, if you want to send this: https://review.openstack.org/#/c/76234/3715:25
bauzasdansmith: jaypipes: yeah, thanks danpb15:26
*** jecarey has joined #openstack-nova15:26
kashyapFolks, is there a `dnf builddep` kind of a thing that I can run before I invoke `tox`, so that running `tox -e py27 [. . .]` won't fail because it can't find deps 'libpq-fe.h' (postgres-devel)15:26
* kashyap just the deps installs manually. But, sometimes, I just invoke `tox` and go do something else, only to find 30 min later, it failed due to some system dep.15:27
*** jcoufal has quit IRC15:27
kashyaps/just/just installs/15:28
bauzaskashyap: there is a wiki page listing all system deps to install, sec checking15:28
bauzas(and this is not in devref AFAIK)15:28
*** RawInfoSec has joined #openstack-nova15:28
bauzasoh no15:29
jaypipesdansmith: done.15:29
bauzaskashyap: http://docs.openstack.org/developer/nova/devref/development.environment.html#using-devstack15:29
dansmithjaypipes: thanks15:29
bauzasjaypipes: thanks you help my rebasing hell15:29
jaypipesheh15:29
kashyapbauzas: It just contains regular DevStack info, and I already am looking at the unit tests page in devref15:29
openstackgerritJay Pipes proposed openstack/nova: Create RequestSpec object  https://review.openstack.org/14552815:29
kashyapbauzas: It doesn't enumerate any system deps to be installed. Or am I just blind?15:30
bauzas kashyap: it does :)15:30
*** lan_ has quit IRC15:30
*** mestery_ is now known as mestery15:30
*** artom has joined #openstack-nova15:30
bauzasjaypipes: thanks for that, but since I'm now playing with the object in the FilterScheduler, I feel the PITA so probably I should -W it15:31
jaypipeskk15:31
kashyapbauzas: You're right, I was blind. Thanks.15:31
*** RawInfoSec has left #openstack-nova15:33
openstackgerritAndrew Laski proposed openstack/nova-specs: Proposal for cells v2 scheduling interaction  https://review.openstack.org/14148615:35
*** kiran-r has joined #openstack-nova15:35
*** mohitsharma has quit IRC15:35
mriedemmelwitt: now that https://review.openstack.org/#/c/185141/ is merged and that pbr fix, i think we're ready for a python-novaclient release15:36
mriedemanything we really want in here before we release? https://review.openstack.org/#/q/status:open+project:openstack/python-novaclient+branch:master,n,z15:38
*** david-lyle has joined #openstack-nova15:39
*** Sukhdev has joined #openstack-nova15:39
mriedemmelwitt: any second thoughts on this one that has a +2? https://review.openstack.org/#/c/175603/15:40
*** zz_jgrimm is now known as jgrimm15:40
*** romainh has left #openstack-nova15:40
mriedemdebate was over whether or not unit tests would be useful15:40
*** hemnafk is now known as hemna15:41
*** claudiub|2 has joined #openstack-nova15:42
*** claudiub has quit IRC15:43
*** artom has quit IRC15:44
*** gszasz has quit IRC15:46
*** david-lyle has quit IRC15:47
*** gholler has quit IRC15:48
*** alexpilotti has joined #openstack-nova15:48
*** patrickeast has joined #openstack-nova15:48
*** mdbooth has quit IRC15:49
*** nikhil_k-brb is now known as nikhil_k15:50
*** mdbooth has joined #openstack-nova15:50
danpbcan someone remind me the equivalent of   mox.IsA(objects.Instance)  for using with "mock" and its assert_called_with method15:52
*** mohitsharma has joined #openstack-nova15:52
*** david-lyle has joined #openstack-nova15:53
*** busterswt has joined #openstack-nova15:54
ndipanovdanpb, there is none really - you have to look into the call tuple and do assertEqual etc.15:55
danpbeww15:56
ndipanovthere is a catch all value you can use but if you want to check just the type15:56
ndipanovyou can't I think15:56
* danpb goes to try to write an equiv15:58
bauzasdanpb: you mean self.assertIsInstance(obj, type) ?15:58
bauzasdanpb: ie. here self.assertIsInstance(my_obj, objects.Instance)15:59
dansmithno15:59
*** dulek has joined #openstack-nova15:59
dansmithdanpb: there might be some thing around "provide a callable for the argument" so that it calls it to check the value15:59
dansmiththere are a few semi-hidden things for doing stuff like that that I never can remember16:00
bauzasaaaha16:01
*** otter768 has joined #openstack-nova16:01
*** matrohon has quit IRC16:01
bauzasyou want to check the type of the arg ?16:01
bauzasgotcha16:01
bauzasdanpb: there are the mock_calls that you can play with16:02
*** gholler has joined #openstack-nova16:02
*** oro_ has joined #openstack-nova16:02
bauzasdanpb: http://www.voidspace.org.uk/python/mock/helpers.html#calls-as-tuples16:02
bauzasdanpb: but that's hacky16:02
bauzasdanpb: or use mock.ANY16:03
danpbthis does the trick  http://paste.fedoraproject.org/228511/3473851416:03
*** rushiagr_away is now known as rushiagr16:03
danpbbasically using the same approach as  mock.ANY but considering only a specific type16:03
claudiub|2hello folks. If you have some time available, could you take a look at the hyper-v nova rescue commit? It's been through quite a lot of review and it seems good to go. https://review.openstack.org/#/c/127159/  Thanks!16:03
bauzasdanpb: yeah I sometimes have this PITA16:04
*** FL1SK has joined #openstack-nova16:04
andreykurilinsdague: Hi! Since we need a spec for implementation of microversion support on client side, should I write it?16:04
mdboothCan anybody remind me what the timescale is for global-requirements to nova requirements?16:04
bauzasdanpb: since 2 objects are different16:04
ndipanovdanpb, nice - a bit dirty but nice16:04
mdboothDo I recall it happens approximately weekly?16:04
sdagueandreykurilin: sure, if you want to take a first pass. I have a bunch of what's needed in my head, but I'm not getting to it until next week16:05
danpbndipanov: that's basically what  mox.IsA is doing too16:05
bauzasdanpb: wouldn't it be possible to create the mock object and stub the method to return it ?16:05
*** artom has joined #openstack-nova16:05
openstackgerritMarkus Zoeller (markus_z) proposed openstack/nova: libvirt: use log file and serial console at the same time  https://review.openstack.org/18805816:05
*** arnaud____ has joined #openstack-nova16:05
andreykurilinsdague: nice, I can start working on it from tomorrow16:05
mriedemdanpb: wonder if there is anything in mock that can do this with the spec kwarg stuff on mocks16:05
bauzasdanpb: eg. mock.patch.objects(objects.MyObject, 'get_by_something', return_value=my_fake_obj)16:05
mriedemi know spec prevents calling things on the mock that aren't defined in the type16:06
sdagueandreykurilin: sounds great, let me know when it's up, I'll try to throw in bits that emerged from the summit16:06
bauzasmriedem: http://www.voidspace.org.uk/python/mock/helpers.html#calls-as-tuples16:06
*** otter768 has quit IRC16:06
danpbbauzas: that's alot more cumbersome imho16:06
bauzasmriedem: and http://www.voidspace.org.uk/python/mock/mock.html#mock.Mock.mock_calls16:06
openstackgerritMarkus Zoeller (markus_z) proposed openstack/nova: libvirt: use log file and serial console at the same time  https://review.openstack.org/18805816:06
mriedembauzas: yeah i know you can get the call args and check isinstance on those16:06
*** jgrimm is now known as zz_jgrimm16:07
mriedembut that's a bit heavy like danpb said16:07
*** jecarey has quit IRC16:07
bauzastotally agreed16:07
danpbthis approach makes the test clearer to read16:07
bauzashence my (18:05:59) bauzas: danpb: eg. mock.patch.objects(objects.MyObject, 'get_by_something', return_value=my_fake_obj)16:07
mriedemdidn't mox have an IsA or something?16:07
danpbbauzas: that's rather obscure too imho16:07
mriedemwonder why mock doesn't have that16:07
*** belmoreira has quit IRC16:07
danpbmriedem: yes, that's what i'm attempting to replicate here16:07
*** jlanoux has quit IRC16:07
bauzasdanpb: why? you stub the method you know you'll get the object, it makes the test more isolatedd16:08
andreykurilinsdague: It looks like spec from ironic can be used. It describes a lot of stuff and, btw, current status of patches mostly implements it.16:08
bauzasdanpb: unless you're explicitely creating the object within the method you wanna test16:08
danpbbauzas: that's exactly what i'm doing16:09
*** zz_jgrimm is now known as jgrimm16:09
bauzasdanpb: mmm16:09
bauzasdanpb: nevermind, I'll review your change if I can, because apparently I'm missing some piece16:10
*** jecarey has joined #openstack-nova16:10
*** moshele has quit IRC16:10
*** gholler_ has joined #openstack-nova16:11
dansmithI think danpb's solution looks good and we should have it in one of our common test utils things16:11
*** jistr has quit IRC16:11
* danpb wonders if mock has a receptive upstream16:12
*** gokrokve_ has joined #openstack-nova16:12
*** artom has quit IRC16:12
*** afazekas has joined #openstack-nova16:12
danpbhmm, last upload to pypi  2012  ... guess that answers that question16:12
*** dulek has quit IRC16:13
dansmithwell, I was going to say, I don't think it changes much16:13
dansmithnot sure if that's by design or not16:13
bauzasdansmith: the AnyInstance() stuff would help a lot the contributors indeed16:14
*** kiranr has joined #openstack-nova16:14
dansmitheff that, it'd help *me* a lot :)16:14
bauzasdansmith: since lots of us are wondering why obj1 != obj2 if the fields are equal16:14
*** kiran-r has quit IRC16:14
*** dulek has joined #openstack-nova16:14
*** gholler has quit IRC16:14
*** gokrokve has quit IRC16:14
dansmithwell, that's not really what he's trying to solve16:15
ndipanovdansmith, danpb it's part of stdlib16:15
bauzasis the problem that is_called_with complains that the callable is different ?16:15
ndipanovso your upstream is python16:15
dansmithoh?16:15
danpbndipanov: oh, i see it merged into base16:15
danpbso what's on pypi is stale16:16
bauzasndipanov: I've been heard that's called mock3, nope ?16:16
dansmithis it in stdlib in 2.7?16:16
ndipanovnope16:16
ndipanovonly in python 316:16
dansmithright, okay16:16
*** unicell1 has quit IRC16:16
*** vilobhmm has joined #openstack-nova16:16
ndipanovbauzas, not sure I think that you're thinking of mox316:16
* bauzas *facepalm*16:17
ndipanovthat iirc was the fork of the orriginal project that also got abandoned16:17
ndipanovwell it took patches but did no releases16:17
ndipanovthere was a very detailed thread about this on openstack-dev some time ago16:17
openstackgerritBaodong (Robert) Li proposed openstack/nova-specs: Support live migration with macvtap SR-IOV  https://review.openstack.org/13607716:18
*** dims_ has quit IRC16:18
*** dims_ has joined #openstack-nova16:18
*** artom has joined #openstack-nova16:20
*** dims__ has joined #openstack-nova16:20
*** jamielennox is now known as jamielennox|away16:21
*** barra204 has joined #openstack-nova16:21
*** david-lyle has quit IRC16:22
*** mlavalle has joined #openstack-nova16:22
mlavallejohnthetubaguy: ping16:22
johnthetubaguymlavalle: hello16:22
*** sriatom has joined #openstack-nova16:23
mlavallejohnthetubaguy: hi. during this session in Vancouver https://etherpad.openstack.org/p/YVR-nova-network I mentioned to you this spec https://review.openstack.org/#/c/90150/16:23
*** alanf-mc has joined #openstack-nova16:24
*** dims_ has quit IRC16:24
johnthetubaguymlavalle: ah, yes16:24
mlavallejohnthetubaguy: this spec addresses point 4 in the 'Priority' section of the etherpad16:24
mlavallejohnthetubaguy: when you have a chance, please help us to get it approved. I responded in gerrit to your concerns from previous review16:25
johnthetubaguymlavalle: thats more a reference to the Nova priorities agreed here: http://specs.openstack.org/openstack/nova-specs/priorities/kilo-priorities.html16:25
*** Mike_D_laptop has joined #openstack-nova16:26
mlavallejohnthetubaguy: yeah, so the spec I am talking about fits under the nova-network / neutron migration title16:26
johnthetubaguymlavalle: https://review.openstack.org/#/c/187272/ is the review for the liberty priorities16:27
johnthetubaguymlavalle: yeah, thats not actually in the list of priorities for liberty at this point16:27
johnthetubaguy...possibly incorrectly16:27
bauzasjohnthetubaguy: PaulMurray: I was discussing with some HP Grenoble folks and we were okay to propose HP Grenoble as another offer for the M midcycle16:28
johnthetubaguymlavalle: so I am about 4 weeks behind spec reviews at this point while I am writing up things from the summit, hopefully that will get fixed soon16:28
*** barra204 has quit IRC16:28
johnthetubaguybauzas: for M, totally we can propose that, I am hoping someone steps up to get this agreed enough so we can talk about it at the midcycle this time16:29
bauzasjohnthetubaguy: yeah, my point is not to say I don't want Bristol, just to say there is an alternative :)16:29
mlavallejohnthetubaguy: great, please keep this spec in your radar screen. Let me know if there is something I can do to help. I am willing to help implement also on the nova side. I will need guidance, but will be glad to help16:29
johnthetubaguymlavalle: cool, thank you, its roughly on my radar, just swamped right now, sorry16:30
mlavallejohnthetubaguy: :-)16:30
*** yassine_ has quit IRC16:31
*** arnaud____ has quit IRC16:31
*** dulek has quit IRC16:33
*** tjones1 has joined #openstack-nova16:33
*** marun has quit IRC16:33
*** marun has joined #openstack-nova16:35
*** dulek has joined #openstack-nova16:36
*** kiranr has quit IRC16:36
*** _kiran_ has joined #openstack-nova16:36
*** gabor-t has joined #openstack-nova16:36
*** nic has joined #openstack-nova16:37
*** marzif_ has joined #openstack-nova16:39
*** _kiran_ has quit IRC16:41
*** gokrokve has joined #openstack-nova16:46
*** hanzhang has quit IRC16:47
*** hanzhang has joined #openstack-nova16:48
*** thangp_ has joined #openstack-nova16:49
*** thangp has quit IRC16:49
*** gokrokve_ has quit IRC16:49
bauzasalaski: while I'm writing ReqSpec and using it, I just rediscovered that HostManager expects task_state16:50
bauzasalaski: I remember that we discussed on that one saying it would be another way to get it, but do you remember how ?16:50
*** gokrokve has quit IRC16:51
alaskibauzas: I was saying that I don't think we need to get it.  It's only ever set one way when passing requests to the scheduler16:52
alaskiso you can just assume it's a certain thing16:52
*** mohitsharma has quit IRC16:52
alaskibut it's likely that what HostManager is doing with it could be accomplished another way16:52
*** sriatom is now known as _sri16:53
bauzasalaski: mmm16:53
*** ssurana has joined #openstack-nova16:53
bauzasalaski: http://docs.openstack.org/developer/nova/devref/vmstates.html#create-instance-states I should always expect 'scheduling' but that's not something I like16:54
bauzasalaski: the problem is, I want to iterate over all the modules by providing an object and backport it to the dicts for the upper methods16:54
*** garyk has quit IRC16:54
vilobhmmjaypipes : hi16:54
bauzasalaski: I mean southbound16:55
*** afazekas has quit IRC16:55
bauzasalaski: so, I'm reconstructing the request_spec dict and the instance_props16:55
jaypipesvilobhmm: hi (though I don't really have time to chat right now... I can do async chat ;)16:55
*** Longgeek has quit IRC16:55
bauzasalaski: because if not, it will be an ugly fat change16:55
vilobhmmjaypipes : sure :) just a small request…can you please review https://review.openstack.org/#/c/138607/16:55
*** claudiub|2 has quit IRC16:55
vilobhmmbauzas : you too : ^^16:56
bauzasalaski: so I could just query the instance based on the id, and put it back16:56
jaypipesvilobhmm: yep, it's on my long list :)16:56
bauzasalaski: but due to the layer isolation, I would prefer to stick with what the object provides16:56
bauzasvilobhmm: oh yeah...16:56
*** dulek has quit IRC16:58
*** belmoreira has joined #openstack-nova16:58
sdagueandreykurilin: links?16:58
alaskibauzas: I think I understand you, but HostState is using it for one very specific thing right?  we should look at how that can be satisfied without needing task_state16:59
bauzasalaski: gotcha16:59
bauzasanyway, meeting time16:59
bauzasneed a recreation with cells16:59
*** markus_z has quit IRC17:01
*** unicell has joined #openstack-nova17:01
*** thangp_ has quit IRC17:03
*** patrickeast has quit IRC17:03
*** lucasagomes is now known as lucas-dinner17:04
*** thangp has joined #openstack-nova17:06
*** ildikov has quit IRC17:07
openstackgerritDavid Bingham proposed openstack/nova: Allow libvirt cleanup completion when serial ports already released  https://review.openstack.org/17997817:07
*** amotoki has quit IRC17:09
*** afazekas has joined #openstack-nova17:09
*** david-lyle has joined #openstack-nova17:13
*** david-lyle has quit IRC17:13
*** josecastroleon has quit IRC17:14
*** marun has quit IRC17:14
*** david-lyle has joined #openstack-nova17:15
*** rushiagr is now known as rushiagr_away17:15
*** marun has joined #openstack-nova17:15
*** nic1 has joined #openstack-nova17:16
*** kaisers has quit IRC17:16
*** pbandzi has quit IRC17:17
*** nic has quit IRC17:17
*** kaisers1 has quit IRC17:17
openstackgerritOpenStack Proposal Bot proposed openstack/nova: Updated from global requirements  https://review.openstack.org/18809917:20
*** yamahata has quit IRC17:21
*** kaisers has joined #openstack-nova17:22
*** kaisers1 has joined #openstack-nova17:22
*** iamjarvo has quit IRC17:24
openstackgerritPetrut Lucian proposed openstack/nova: Hyper-V: restart serial console workers after instance power change  https://review.openstack.org/16308017:24
*** patrickeast has joined #openstack-nova17:24
*** dboik has quit IRC17:26
*** adrian_otto has joined #openstack-nova17:26
*** iamjarvo has joined #openstack-nova17:26
*** lpetrut has quit IRC17:27
*** dboik has joined #openstack-nova17:27
*** adrian_otto has quit IRC17:27
*** david-lyle has quit IRC17:28
openstackgerritOpenStack Proposal Bot proposed openstack/nova: Updated from global requirements  https://review.openstack.org/18809917:29
mriedemjohnthetubaguy: dansmith: you guys ok with a novaclient release? assuming yes | don't are17:31
mriedem*care17:31
dansmithgo for gate blocker17:32
mriedemha17:32
mriedemwe are go for gate blocker17:32
mriedemcount it down17:32
* johnthetubaguy nodes with approval17:34
johnthetubaguylol17:34
johnthetubaguynods17:34
*** Sukhdev has quit IRC17:34
openstackgerritDan Smith proposed openstack/nova: Convert test objects to new field formats  https://review.openstack.org/18771917:36
openstackgerritDan Smith proposed openstack/nova: Begin the transition to an explicit object registry  https://review.openstack.org/18771817:36
openstackgerritDan Smith proposed openstack/nova: Use fields from oslo.versionedobjects  https://review.openstack.org/18772017:36
openstackgerritDan Smith proposed openstack/nova: Remove assertRemotes() from objects tests  https://review.openstack.org/18810917:36
openstackgerritDan Smith proposed openstack/nova: Fix an objects layering violation in compute/api  https://review.openstack.org/18811017:36
openstackgerritDan Smith proposed openstack/nova: Replace metaclass registry with explicit opt-in registry from oslo  https://review.openstack.org/18811117:36
openstackgerritDan Smith proposed openstack/nova: Make readonly field tests use exception from o.vo  https://review.openstack.org/18811217:36
*** danpb has quit IRC17:36
openstackgerritOpenStack Proposal Bot proposed openstack/nova: Updated from global requirements  https://review.openstack.org/18809917:38
*** xgsa has quit IRC17:38
*** iamjarvo has quit IRC17:39
*** burt has quit IRC17:40
*** yamahata has joined #openstack-nova17:40
*** garyk has joined #openstack-nova17:41
openstackgerritDaniel Berrange proposed openstack/nova: compute: only use non_inheritable_image_properties if snapshotting  https://review.openstack.org/18725017:41
openstackgerritDaniel Berrange proposed openstack/nova: compute: remove get_image_metadata method  https://review.openstack.org/18725117:41
openstackgerritDaniel Berrange proposed openstack/nova: libvirt: make instance compulsory in blockinfo APIs  https://review.openstack.org/18811817:41
openstackgerritDaniel Berrange proposed openstack/nova: WIP: libvirt: convert to use nova.objects.ImageMeta  https://review.openstack.org/18811917:41
*** kiran-r has joined #openstack-nova17:41
johnthetubaguyndipanov: sorry about your email on testing, I totally miss read what you put, makes more sense to me now!17:44
ndipanovjohnthetubaguy, yeah when I read it back I totally could see how someone could read that like I argue against unit tests17:44
ndipanov:)17:44
ndipanovjohnthetubaguy, maybe we should move your email to a separate thread in [Nova]17:45
ndipanovI am particularly interested in documenting "tribal knowledge"17:46
johnthetubaguyndipanov: oh, totally, I went off on one17:46
johnthetubaguyndipanov: I mean its a hard nut to crack, but I think we need to make some progress with the "big things"17:47
*** adrian_otto has joined #openstack-nova17:47
openstackgerritgaryk proposed openstack/nova: Compute: tidy up legacy treatment for vif types  https://review.openstack.org/18812217:48
mriedemwait we aren't doing unit tests anymore?17:48
mriedemsweet17:48
* johnthetubaguy makes gun shooting noises like the crazy rich texan in the Simpsons17:50
mriedemyeeehaaawww!17:51
*** iamjarvo has joined #openstack-nova17:52
*** agireud has quit IRC17:53
*** busterswt has quit IRC17:57
mriedemit is done https://pypi.python.org/pypi/python-novaclient/2.26.017:59
openstackgerritmelanie witt proposed openstack/nova: Handle cells race condition deleting unscheduled instance  https://review.openstack.org/18812618:00
*** kiran-r has quit IRC18:02
*** otter768 has joined #openstack-nova18:02
*** belmoreira has quit IRC18:02
*** VW_ has joined #openstack-nova18:03
*** VW has quit IRC18:05
*** tjones2 has joined #openstack-nova18:06
*** tjones2 has left #openstack-nova18:06
melwittmriedem: awesome, thanks. is there another thing we need to run to make the milestone things go to Fix Released?18:06
mriedemmelwitt: i have to do it manually since the script doesn't work apparently18:06
*** otter768 has quit IRC18:07
*** dboik has quit IRC18:07
*** dboik has joined #openstack-nova18:07
*** afazekas has quit IRC18:07
*** tjones1 has quit IRC18:07
mriedemmelwitt: johnthetubaguy: dansmith: sdague: now that novaclient 2.26.0 is out with the deprecation warnings for using the volume APIs/CLIs, i'm thinking we should bump the minimum required version global-requirements so anyone using it pulls that in and gets aware of the deprecation as soon as possible18:09
mriedemanyone against that?18:09
sdaguemriedem: fine by me18:09
johnthetubaguymriedem: cunning, worth a try18:09
*** apuimedo has joined #openstack-nova18:09
mriedemcool, i didn't want to have to release Mujina on anyone18:10
* mriedem looked that up in the last 20 minutes18:10
*** adrian_otto has quit IRC18:10
*** ildikov has joined #openstack-nova18:11
kragniz>The Nova team is beside themselves with glee18:11
kragnizthat script never ceases to entertain18:11
mriedemthe nova team is mildly apathetic about releasing novaclient 2.27.018:11
mriedemthat can be the next one18:11
lifelessmriedem: so thats actually against the g-r policy18:11
* mriedem goes to read the readme in the requirements repo18:12
lifelessmriedem: specifically http://git.openstack.org/cgit/openstack/requirements/tree/README.rst#n18118:13
mriedemyes i just read that18:13
lifelessmriedem: I'm not saying this is right18:13
lifelessjust that I read it yesterday and so had it paged in :)18:14
mriedemdamn your brain18:14
lifelessthe gate will see your warnings immediately as we don't upper cap most things18:14
mriedemfwiw, we're updating packages weekly if not daily18:14
mriedemand have just gotten used to it18:14
mriedemsince oslo releases weekly18:14
mriedemanyway, i'll leave it out and check the gate logs to see if there is any obvious usage that needs to be cleaned up18:15
openstackgerritNicolas Simonds proposed openstack/nova: Fix/implement revert-resize for RBD-backed images  https://review.openstack.org/18739518:15
*** agireud has joined #openstack-nova18:16
*** marzif_ has quit IRC18:19
*** VW_ has quit IRC18:19
*** marzif_ has joined #openstack-nova18:19
*** VW has joined #openstack-nova18:20
*** annegentle has joined #openstack-nova18:23
*** yamahata has quit IRC18:25
*** yamahata has joined #openstack-nova18:25
*** lpetrut has joined #openstack-nova18:27
*** xgsa has joined #openstack-nova18:27
*** adrian_otto has joined #openstack-nova18:27
*** artom_ has joined #openstack-nova18:30
openstackgerritQiu Yu proposed openstack/nova: Force iSCSI disconnect after timeout  https://review.openstack.org/16781518:31
*** artom has quit IRC18:31
*** david-lyle has joined #openstack-nova18:33
*** artom_ has quit IRC18:33
*** ijw has joined #openstack-nova18:35
*** adrian_otto has quit IRC18:38
*** gokrokve has joined #openstack-nova18:39
*** aix has quit IRC18:39
*** flaper87 has quit IRC18:44
*** mlavalle has quit IRC18:46
*** jecarey has quit IRC18:49
*** baoli has quit IRC18:52
openstackgerritNicolas Simonds proposed openstack/nova: Fix/implement revert-resize for RBD-backed images  https://review.openstack.org/18739518:53
openstackgerritNicolas Simonds proposed openstack/nova: imagebackend: Split out resize_image logic from create_image  https://review.openstack.org/18785718:53
*** pradipta has joined #openstack-nova18:53
*** mordred has quit IRC18:55
*** mordred has joined #openstack-nova18:55
*** adrian_otto has joined #openstack-nova18:56
*** marzif_ has quit IRC18:57
*** barra204 has joined #openstack-nova18:57
*** ijw has quit IRC19:00
dims__mriedem: dansmith: how are we able to compare a variable of type nova.db.sqlalchemy.models.Quota to a integer? (https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L3431). where's the magic that converts user_quotas[res] into an integer so the if condition works...does not work in python 3, hence asking :)19:02
*** barra204 has quit IRC19:02
*** alanf-mc has quit IRC19:03
*** ssurana has quit IRC19:06
mriedemdims__: i thought user_quotas was updated before _calculate_overquota was called19:07
*** alanf-mc has joined #openstack-nova19:07
mriedemdims__: i think i've wondered about that myself though fwiw, it's not easy to trace how those dicts are populated throughout that path19:07
*** agireud_ has joined #openstack-nova19:08
*** sjcherry has joined #openstack-nova19:08
*** flaper87 has joined #openstack-nova19:08
*** ssurana has joined #openstack-nova19:09
*** ssurana has quit IRC19:09
*** flaper87 has quit IRC19:09
*** flaper87 has joined #openstack-nova19:09
*** agireud has quit IRC19:09
*** agireud_ is now known as agireud19:09
mriedemdims__: https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L306319:09
dims__mriedem: i printed out user_quotas[res] : <nova.db.sqlalchemy.models.Quota object at 0x10d219d90>19:10
mriedemuser_quotas[res] should be mapping to the hard_limit int here https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/models.py#L44319:10
dims__mriedem: both in py3 and py219:10
mriedemit's mutated somewhere then19:11
dansmithdims__: I'm touched that you think that's a thing I know about :)19:11
dims__mriedem: right don't know ehre19:11
dims__:(19:11
mriedemdims__: basically you have to find out what in http://git.openstack.org/cgit/openstack/nova/tree/nova/quota.py is eventually passing in user_quotas to db.quota_reserve19:11
*** adrian_otto has quit IRC19:11
dims__dansmith: :)19:11
mriedemsince http://git.openstack.org/cgit/openstack/nova/tree/nova/quota.py gets a bit goofy with changing things19:11
*** ssurana has joined #openstack-nova19:11
mriedemdims__: it's probably this http://git.openstack.org/cgit/openstack/nova/tree/nova/quota.py#n24819:12
mriedemit fills in user_quotas with Quota objects from project_quotas19:12
mriedembut those should also be ints https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L307819:13
dims__mriedem: right, that's not the problem the python statement i linked to above someone works in py2 but not py319:13
mriedemidk, does something in the sqlalchemy model coerce the value?19:13
dims__mriedem: if i print "user_quotas[res] >= 0" it fails with TypeError: unorderable types: Quota() >= int()19:13
dims__right mriedem19:14
mriedemdims__: can you get a traceback when it fails?19:14
dims__yep19:14
dims__right on that line :)19:14
mriedembut the full trace going back to nova.quota19:14
dims__http://paste.openstack.org/show/260221/19:15
dims__zzzeek: around?19:16
zzzeekdims__: hey19:16
mriedemdims__: are you sure that test case isn't just borked?19:16
dims__zzzeek: in https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L3431 - user_quotas[res] is of type nova.db.sqlalchemy.models.Quota19:17
dims__zzzeek: and somehow in python2 the >= against integer works fine19:17
dims__zzzeek: and in python3 it doesn't19:17
dims__zzzeek: fails with http://paste.openstack.org/show/260221/19:17
zzzeekdims__: in py3k it’s not valid to compare disparate types19:17
mriedemdims__: the tests are wrong19:17
mriedemhttps://github.com/openstack/nova/blob/master/nova/tests/unit/db/test_db_api.py#L11219:17
mriedemuser_quotas[resource] = db.quota_create(context, project_id,19:18
mriedemthe test is setting a Quota model dict like thing in the user_quotas dict that's passed to db.api.quota_reserve19:18
zzzeekdims__: in that code you’d want to have an integer on the other side, if that’s an attribute of the Quota object perhaps, or the user_quotas dict is mis-popualted as it seems mriedem is saying19:18
mriedemwhich is wrong since user_quotas should be mapping of resources to limit integers19:18
zzzeekdims__: see?  this is py3k finding a bug :)19:18
mriedemyay!19:19
mriedemso...when can we just move to java?19:19
mriedemwith typing?19:19
mriedeman serializable objects19:19
dims__haha19:19
*** apuimedo has quit IRC19:19
dims__mriedem: i have a whole bunch of these19:19
mriedemso i guess python 3 is just becoming more java like huh19:19
mriedemdims__: is this a recent regression? i thought the nova py34 job was passing last week?19:20
dims__hmmm. now i have to fix the test!19:20
mriedemshouldn't be hard19:20
*** pradipta has quit IRC19:20
dims__mriedem: py34 had just 2 tests :) i have 2 reviews now that brings it to like 1500+ and now working on another one19:21
mriedemuser_quotas[resource] = db.quota_create(context, project_id,...).hard_limit19:21
mriedemoh i didn't realize it was filtering stuff out19:21
dims__mriedem: look at the tox.ini19:21
dims__the idea was we got all the dependencies lined up so we could enable the test job last weeek19:22
mriedemha19:22
mriedemhttps://github.com/openstack/nova/blob/master/tox.ini#L4419:22
mriedemwell as long as nova.tests.unit.test_versions works i think we can ship it19:22
dims__mriedem: what the hell was the tests testing? :)19:23
mriedemthe db api one? idk19:23
mriedemobviously not enough19:23
mriedemwhen i refactored quota_reserve in kilo i added a slew of unit tests for that path but they were using mock19:23
dims__zzzeek: thanks!19:24
zzzeekdims__: youre welcome19:24
dims__mriedem: dansmith: so i take a stab at adding the .hard_limit in a separate patch?19:24
mriedemdims__: is there a patch you're looking at?19:25
mriedemotherwise yes this is a bug in the test code19:25
mriedemyou could even open a bug for tracking it19:25
dims__mriedem: only on my hard disk19:25
dims__mriedem: let me figure out a bit and throw in a bug and review19:26
*** jgrimm is now known as zz_jgrimm19:26
dims__who do we know from bluebox?19:36
*** VW has quit IRC19:38
*** iamjarvo has quit IRC19:39
*** adrian_otto has joined #openstack-nova19:39
dansmithdims__: jlk19:39
dims__mriedem: here's the bug and a way to repro it in python2 - https://bugs.launchpad.net/nova/+bug/146166519:40
openstackLaunchpad bug 1461665 in OpenStack Compute (nova) "nova.tests.unit.db.test_db_api sets up quotas incorrectly" [Undecided,New]19:40
*** igordcard_ has quit IRC19:40
*** claudiub has joined #openstack-nova19:40
mriedemdims__: want me to push the change?19:41
dims__mriedem: sure thanks19:41
dims__dansmith: ah. thanks19:42
mriedemi hail my new blue box overlords19:43
*** artom has joined #openstack-nova19:45
*** artom has quit IRC19:47
*** artom has joined #openstack-nova19:47
*** ociuhandu has quit IRC19:48
*** Mike_D_laptop has quit IRC19:52
*** _sri has quit IRC19:53
*** armax has quit IRC19:53
*** alanf-mc has quit IRC19:55
*** burt has joined #openstack-nova19:55
bauzasalaski: around ?19:56
alaskibauzas: yes19:57
*** gabor-t has quit IRC19:57
*** adrian_otto has quit IRC19:57
bauzasalaski: I'm looking at https://github.com/openstack/nova/blob/master/nova/scheduler/host_manager.py#L296-L30119:57
*** flaper87 has quit IRC19:57
*** VW has joined #openstack-nova19:58
bauzasalaski: honestly, I don't figure out how to check for that without the task_state :( do you have another idea ?19:58
*** barra204 has joined #openstack-nova19:58
bauzasalaski: I can suppose that vm_state is always BUILDING but I need to figure out if all calls to the scheduler are done this way19:58
*** flaper87 has joined #openstack-nova19:59
*** VW_ has joined #openstack-nova19:59
*** thangp has quit IRC20:00
*** iamjarvo has joined #openstack-nova20:00
alaskibauzas: right.  my thinking was that since you have to synthesize an instance from the request spec when this is called in the scheduler you can just set it to building.  But really I would probably change the api to have an inc_ip_ops=False parameter that can be set, and let the caller decide rather than having HostManager care about vm/task_states20:01
*** xyang1 has quit IRC20:01
*** patrickeast_ has joined #openstack-nova20:01
*** dave-mcc_ has quit IRC20:02
bauzasalaski: I see20:02
bauzasalaski: there are a few number of callers for select_dests()20:02
*** patrickeast has quit IRC20:02
*** patrickeast_ is now known as patrickeast20:02
bauzasalaski: but we would then have to check the instance.task_state each time before calling select_dest()20:03
*** otter768 has joined #openstack-nova20:03
bauzasalaski: which is a bit hard IMHO20:03
*** VW has quit IRC20:03
bauzasalaski: just wonder if it would be acceptable to port this in the scheduler client20:03
*** alanf-mc has joined #openstack-nova20:03
*** barra204 has quit IRC20:03
*** flaper87 has quit IRC20:03
alaskibauzas: we can assume that if the scheduler is calling this that the instance is in motion, and therefore would be an iop, right?20:03
*** marun_ has joined #openstack-nova20:03
*** marun has quit IRC20:03
bauzasalaski: ie. the sched client would check vm_state and task_state and provide the parameter right20:03
bauzasalaski: probably, but I can't be sure about that20:04
bauzasalaski: I would have just said that the conditional would be True each time, but I just wonder if that's actually right20:04
alaskiI don't know, but I would expect that when it's called from the scheduler it is20:05
bauzasalaski: I mean, there are so many corner cases when scheduling and you know that we don't have a clear state20:05
*** dave-mccowan has joined #openstack-nova20:05
*** mtanino has quit IRC20:06
bauzasalaski: mmm20:06
*** jecarey has joined #openstack-nova20:06
alaskianother option would be to add vm_state/task_state to requestspec, but not persist them.  though that's not my preference20:07
bauzasalaski: hardly checking http://docs.openstack.org/developer/nova/devref/vmstates.html20:07
bauzasalaski: yeah, I will have to do this for the context too20:07
*** otter768 has quit IRC20:07
bauzasalaski: and an ugly thing20:07
bauzasalaski: that => https://github.com/openstack/nova/blob/master/nova/scheduler/filter_scheduler.py#L12620:08
*** salv-orlando has quit IRC20:09
*** salv-orlando has joined #openstack-nova20:09
*** baoli has joined #openstack-nova20:10
alaskiugh20:10
alaskithat looks like it just needs to move out into a new parameter to host_manager methods20:11
alaskior just set host_manager.config_options20:11
alaskireally shouldn't be in filter_properties20:12
*** baoli has quit IRC20:13
*** alanf-mc has quit IRC20:13
bauzasalaski: I'll just provide it as a non-field20:13
bauzasalaski: discussed that with dansmith20:13
*** baoli has joined #openstack-nova20:13
bauzasalaski: like a getter/setter20:14
*** alanf-mc has joined #openstack-nova20:14
alaskiokay.  since that doesn't appear to ever go over the wire, or need to at least20:14
bauzasalaski: yeah20:14
bauzasalaski: here io_ops can't be done that way, since all the non-fields are not provided over the wire20:15
*** adrian_otto has joined #openstack-nova20:15
*** lucas-dinner has quit IRC20:15
bauzasalaski: I mean the task_state20:15
*** alanf-mc has quit IRC20:16
bauzasalaski: so we honestly have 2 possibilities :20:16
bauzas1/ add a parameter to select_dest() considering the need for io_ops20:16
*** flaper87 has joined #openstack-nova20:16
openstackgerritmelanie witt proposed openstack/nova: Handle cells race condition deleting unscheduled instance  https://review.openstack.org/18812620:16
bauzas2/ add some bad thing in the spec object20:17
bauzasmmm20:17
bauzas3/ figure out which cases are not io_ops20:17
*** adrian_otto has quit IRC20:17
alaskiI think we can safely assume that if the scheduler calls consume_from_instance that iops should be incremented20:17
alaskibut that may be proved wrong later20:17
*** alexpilotti has quit IRC20:17
bauzasalaski: exactly :(20:18
*** gokrokve has quit IRC20:18
bauzasalaski: I don't want to bet on it20:18
dims__dansmith: https://review.openstack.org/#/c/187720/ reminds me the hash calculation may end up being python version specific, i have not been able to get the same hashes with python320:18
alaskibauzas: I think it can be audited easily enough, there are few calls to the scheduler20:18
alaskibauzas: should only be boot and resize/migrate, and both of those would increment io_ops20:19
openstackgerritmelanie witt proposed openstack/nova: Handle cells race condition deleting unscheduled instance  https://review.openstack.org/18812620:19
bauzasalaski: mmm, I should probably add a DNM change to add something so Tempest could just check that20:19
bauzasalaski: a log saying 'eh, here is the state of the instance before calling out the scheduler'20:20
alaskibauzas: sure.  I just can't think of a case where the scheduler would be called and not result in an io_ops.  it is a precursor to moving an instance20:20
bauzasalaski: agreed20:21
bauzasalaski: if we consume from an instance, that should be an IO op20:21
bauzasalaski: the opposite I'm thinking is a check_dest() API method, which would return True or False by checking all filters20:21
bauzasin that case, the hostmanager wouldn't be consuming usage20:22
*** tjones1 has joined #openstack-nova20:22
alaskiright20:22
cfriesenwho knows stuff about the nova interface to glance?  I think we've found a problem that can lead to glance leaving a file open.20:23
bauzasalaski: okay, will write a DNM and see but I think it should be true that way - or leave a big comment saying to throw me tomatoes if I'm wrong20:23
*** ddaskal has quit IRC20:25
*** busterswt has joined #openstack-nova20:25
openstackgerritNicolas Simonds proposed openstack/nova: libvirt: Make the guest console configurable  https://review.openstack.org/12252320:27
*** ociuhandu has joined #openstack-nova20:29
*** ivasev has quit IRC20:29
*** Sukhdev has joined #openstack-nova20:29
*** matrohon has joined #openstack-nova20:29
mriedemcfriesen: probably jaypipes20:29
cfriesenmriedem: thx20:30
dansmithdims__: okay, that calculation is done in o.vo now and it passes both versions of python, FWIW20:30
dansmithdims__: I mean, it's done in o.vo and nova will soon use that implementation20:30
mriedemdansmith: have you ever thought about logging how long we waited for an external event to come back from neutron?20:32
*** adrian_otto has joined #openstack-nova20:33
*** alanf-mc has joined #openstack-nova20:33
*** Nastooh_ has joined #openstack-nova20:33
dansmithmriedem: I think we did in a test version before landing it whilst making sure it was working20:33
mriedemwe have some some logging around preparing to wait for events and getting them back, but you have to diff the timestamps in the logs20:33
dansmithmriedem: but it seems like a reasonable thing to log all the time, yeah20:33
mriedemi'll take a look at adding something20:33
dims__dansmith: i see, will check my work again then20:33
dansmithdepends on whether you want to just time the wait (easy) or the prepare...wait end20:33
Nastooh_No dice in openstack-containers, maybe here: Can't ssh, externally or via netns, to a nova-docker instance. I can ping it, netns and externally; however. Any suggestions?20:35
mriedemdansmith: not sure yet, i'm looking into the code to see why i have a 10 second black box here after getting the vif plug event back from neutron and then timeout http://paste.openstack.org/show/260394/20:35
dansmithmriedem: I don't see any timeout in that paste20:36
mriedemdansmith: it's right after the error20:37
mriedemit's an rpc_response_timeout20:37
mriedemi didn't put it in the paste but that's what it was20:37
mriedemcopy_qcow2_image takes 28 seconds, vif plug event takes about 6 seconds, processing that takes at least 10 seconds and we timeout20:38
mriedemwith random stuff in between taking time in a 60 second rpc timeout20:38
dansmithah, okay20:38
*** vladikr has quit IRC20:41
*** adrian_otto has quit IRC20:43
*** radek_ has quit IRC20:44
*** matrohon has quit IRC20:46
*** baoli has quit IRC20:46
claudiubhello folks. If you have a bit of time, could you take a look at the hyper-v nova rescue patch? https://review.openstack.org/#/c/127159/ It has been reviewed quite a lot for some time, Jenkins is happy, CI is happy, we are happy, but we would be happier to get some extra reviews on it. :D20:49
openstackgerritSylvain Bauza proposed openstack/nova: DO NOT MERGE: Ajouter un log moche pour le scheduler  https://review.openstack.org/18817820:54
*** apuimedo has joined #openstack-nova20:54
openstackgerritDaniel Stelter-Gliese proposed openstack/nova: Extract device path after connecting volume when booting LXC instances from volumes  https://review.openstack.org/18802520:56
mriedemdansmith: i guess network vif plug wasn't my issue, it was this turd http://paste.openstack.org/show/260548/20:57
mriedemwhich we also see upstream20:57
*** baoli has joined #openstack-nova20:57
*** adrian_otto has joined #openstack-nova20:58
*** marun_ has quit IRC20:58
dansmithmriedem: oh lovely21:00
openstackgerritOpenStack Proposal Bot proposed openstack/nova: Updated from global requirements  https://review.openstack.org/18809921:00
*** adrian_otto has quit IRC21:01
*** angdraug has joined #openstack-nova21:01
openstackgerritsean mooney proposed openstack/nova: fix AggregateTypeAffinityFilter multi values support  https://review.openstack.org/14188321:02
*** burt has quit IRC21:02
flwangmriedem: dansmith:  could you pls revisit https://review.openstack.org/#/c/164494/ ? the nova backup fix, thanks21:03
*** sabari has quit IRC21:03
*** swamireddy has quit IRC21:03
*** nickchase has quit IRC21:04
*** tsekiyama has quit IRC21:05
*** annegentle has quit IRC21:06
*** samueldmq has quit IRC21:06
*** adrian_otto has joined #openstack-nova21:08
*** artom has quit IRC21:10
openstackgerritChris Friesen proposed openstack/nova: don't leave glance with open file descriptor on error  https://review.openstack.org/18817921:13
cfriesenjaypipes: (and anyone else interested)   ^21:13
*** Gomeler has quit IRC21:13
cfriesenif anyone has  a more elegant way to handle this, please let me know21:13
*** iamjarvo has quit IRC21:14
*** dboik_ has joined #openstack-nova21:17
*** sabari has joined #openstack-nova21:18
*** sjcherry has quit IRC21:20
*** dboik has quit IRC21:20
*** dboik_ has quit IRC21:21
*** alex_klimov has joined #openstack-nova21:21
mriedemcfriesen: could you write a simple unit test such that data is a mock and on like the 1st call to write() you get the IOError but then assert that it keeps going (len(image_chunks) == 2 or something in the test)21:22
mriedemand finally assert that the IOError is raised21:23
*** swamireddy has joined #openstack-nova21:24
*** iamjarvo has joined #openstack-nova21:25
*** baoli has quit IRC21:28
*** baoli has joined #openstack-nova21:29
*** lpetrut has quit IRC21:34
*** dontalton has joined #openstack-nova21:39
*** busterswt has quit IRC21:40
openstackgerritBaodong (Robert) Li proposed openstack/nova: Handle port delete event  https://review.openstack.org/17939021:42
*** alexpilotti has joined #openstack-nova21:45
*** iamjarvo has quit IRC21:49
*** iamjarvo has joined #openstack-nova21:49
*** mwagner_afk has quit IRC21:49
*** iamjarvo has quit IRC21:50
*** ianbrown has joined #openstack-nova21:52
*** dprince has quit IRC21:53
*** mwagner_afk has joined #openstack-nova21:53
*** lpetrut has joined #openstack-nova21:55
*** lpetrut has quit IRC21:56
*** mriedem has quit IRC21:57
*** ianbrown has quit IRC22:00
*** agireud has quit IRC22:01
*** armax has joined #openstack-nova22:03
*** otter768 has joined #openstack-nova22:04
*** tsekiyama has joined #openstack-nova22:04
*** ianbrown has joined #openstack-nova22:05
*** otter768 has quit IRC22:09
*** tsekiyama has quit IRC22:09
*** gholler__ has joined #openstack-nova22:14
*** mwagner_afk has quit IRC22:16
*** dave-mccowan has quit IRC22:18
*** gholler_ has quit IRC22:18
*** dave-mccowan has joined #openstack-nova22:20
*** jamielennox|away is now known as jamielennox22:20
*** dave-mccowan has quit IRC22:24
*** oro_ has quit IRC22:26
*** VW_ has quit IRC22:27
*** Sukhdev has quit IRC22:29
*** adrian_otto has quit IRC22:34
*** mriedem has joined #openstack-nova22:34
*** gholler__ has quit IRC22:35
*** mriedem has quit IRC22:37
*** mriedem has joined #openstack-nova22:38
*** kylek3h has quit IRC22:39
openstackgerriteeway hsu proposed openstack/nova: Fixes referenced path in nova/doc/README.rst  https://review.openstack.org/18819722:39
*** adrian_otto has joined #openstack-nova22:40
*** vilobhmm has quit IRC22:41
*** vilobhmm has joined #openstack-nova22:41
*** VW has joined #openstack-nova22:41
*** tonytan4ever has quit IRC22:41
*** tonytan4ever has joined #openstack-nova22:42
*** VW_ has joined #openstack-nova22:43
*** VW_ has quit IRC22:43
*** adrian_otto has quit IRC22:43
*** VW_ has joined #openstack-nova22:43
*** vilobhmm1 has joined #openstack-nova22:44
*** VW has quit IRC22:45
*** vilobhmm has quit IRC22:46
*** tonytan4ever has quit IRC22:46
*** barra204 has joined #openstack-nova22:48
*** VW_ has quit IRC22:48
*** tpatil has joined #openstack-nova22:50
*** whenry has quit IRC22:50
*** barra204 has quit IRC22:53
*** apuimedo has quit IRC22:53
*** nic1 has quit IRC22:58
*** Sukhdev has joined #openstack-nova22:59
*** baoli has quit IRC23:01
*** Raildo_ has joined #openstack-nova23:03
*** destiny has quit IRC23:04
*** tjones1 has quit IRC23:06
*** jecarey has quit IRC23:06
*** tjones1 has joined #openstack-nova23:07
*** aysyanne has quit IRC23:08
*** pixelbeat_ has quit IRC23:11
*** vladikr has joined #openstack-nova23:11
flwangjaypipes: ping23:14
flwangre glance v2 in nova23:15
*** angdraug has quit IRC23:18
*** tsekiyama has joined #openstack-nova23:25
mikaldims__: you around?23:25
*** tjones1 has quit IRC23:33
*** tjones1 has joined #openstack-nova23:34
*** sdake has quit IRC23:35
*** VW has joined #openstack-nova23:35
*** dontalton has quit IRC23:37
*** Raildo_ has quit IRC23:39
*** tellesnobrega has quit IRC23:40
*** rwsu has quit IRC23:41
*** VW has quit IRC23:44
dims__mikal: pong23:48
*** dims_ has joined #openstack-nova23:51
*** bzhao has joined #openstack-nova23:52
openstackgerritTomoki Sekiyama proposed openstack/nova: libvirt: Make behavior of os_require_quiesce consistent  https://review.openstack.org/16781923:53
mikaldims__: I am trying to remember what we agreed to with that oslo.conconcurrenty patch we talked about at the summit. I think we agreed I should send the minimum viable patch against master in public and then we'd back port it? Is that your recollection?23:55
*** dims__ has quit IRC23:55
dims_mikal: yes. that's right.23:56
mikaldims_: ok, cool. I've been off sick so I wasn't 100% sure I'd made that up.23:56
mikaldims_: I'm work on progressing that today then.23:56
*** claudiub has quit IRC23:57

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!