Monday, 2015-11-30

« Saturday, 2015-11-28 Index Tuesday, 2015-12-01 »
*** openstack has joined #openstack-nova15:45
BobBallnotmorgan: But when you make changes to the plugin, you also need to increment the XenAPI plugin version number (see https://git.openstack.org/cgit/openstack/nova/tree/plugins/xenserver/xenapi/etc/xapi.d/plugins/nova_plugin_version and https://git.openstack.org/cgit/openstack/nova/tree/nova/virt/xenapi/client/session.py#n84) - as long as you give a default for glance_path in download_vhd then it's backwards compatable and you can just d15:45
*** FL1SK has joined #openstack-nova15:45
*** akshai has joined #openstack-nova15:46
*** Sukhdev_ has joined #openstack-nova15:46
*** flaper87 has joined #openstack-nova15:46
*** lpetrut has joined #openstack-nova15:47
*** rfolco has joined #openstack-nova15:47
notmorganBobBall: ok15:48
*** e0ne has joined #openstack-nova15:48
notmorganBobBall: yeah thats is the plan, the default will be ""15:48
notmorganBobBall: so no change in behavior15:48
BobBallawesome15:48
notmorganBobBall: now, i *do* want to look into backporting this change for the genral case, but thats is a convo to have once this lands15:49
*** salv-orlando has joined #openstack-nova15:49
*** cfarquhar has joined #openstack-nova15:49
*** cfarquhar has quit IRC15:49
*** cfarquhar has joined #openstack-nova15:49
*** lpetrut has quit IRC15:49
notmorganBobBall: not sure how will that will work for the plugin w/ versioning15:49
BobBallbackporting for the general case?  What do you mean there?15:49
*** scheuran has quit IRC15:49
*** dims has joined #openstack-nova15:49
notmorganBobBall: general case = not with a plugin15:50
openstackgerritAndrew Bogott proposed openstack/nova: Prevent scheduling of instances on full hosts.  https://review.openstack.org/24225115:50
*** shaohe_feng has joined #openstack-nova15:50
notmorganBobBall: the other codepath that i already have an easy fix for.15:50
notmorganBobBall: but let me land this / finish the code then we will discuss backports if possible15:50
notmorganand how that is handled :)15:50
BobBallok15:50
notmorganbecause can't backport till this lands anyway15:50
*** burt has joined #openstack-nova15:51
*** rlrossit has quit IRC15:51
*** corvus is now known as jeblair15:52
*** garyk has quit IRC15:52
*** amotoki has quit IRC15:53
*** rlrossit has joined #openstack-nova15:53
*** agireud has joined #openstack-nova15:53
*** garyk has joined #openstack-nova15:54
johnthetubaguypkholkin: rpodolyaka: I would like to see what others think about that, still seems a bit odd to me. Added a comment around some more tests and a link to some utils we have to help with the decorator15:54
*** rook has quit IRC15:55
*** Kevin_Zheng has quit IRC15:56
openstackgerritShaoHe Feng proposed openstack/nova-specs: Attach/detach SR-IOV interface  https://review.openstack.org/13991015:56
*** Kevin_Zheng has joined #openstack-nova15:56
*** jistr has quit IRC15:57
*** lucas-afk is now known as lucasagomes15:58
*** jasondotstar has joined #openstack-nova15:58
*** sdague has joined #openstack-nova15:59
*** mriedem has joined #openstack-nova16:00
openstackgerritDan Smith proposed openstack/nova: Actually pass the migration data object down to the virt drivers  https://review.openstack.org/24821116:00
*** artom has joined #openstack-nova16:00
*** bkopilov has quit IRC16:00
*** e0ne has quit IRC16:02
openstackgerritRyan McNair proposed openstack/nova: Add retry logic for detaching device using LibVirt  https://review.openstack.org/22785116:02
*** rook has joined #openstack-nova16:03
*** zhangjn has quit IRC16:05
*** kevinbenton_ has quit IRC16:07
*** ccarmack has joined #openstack-nova16:07
*** kevinbenton has joined #openstack-nova16:07
*** Marga_ has joined #openstack-nova16:11
*** irenab has joined #openstack-nova16:12
*** Marga_ has quit IRC16:13
*** EinstCrazy has quit IRC16:13
*** Marga_ has joined #openstack-nova16:14
*** rook has quit IRC16:14
openstackgerritChuck Carmack proposed openstack/python-novaclient: Remove python 2.6 support from python-novaclient  https://review.openstack.org/24887416:14
*** tjones has joined #openstack-nova16:14
*** rook has joined #openstack-nova16:15
openstackgerritEd Leafe proposed openstack/nova: Add better help text to scheduler options  https://review.openstack.org/24718116:15
*** arif-ali_ has joined #openstack-nova16:15
*** arif-ali has quit IRC16:15
*** arif-ali_ is now known as arif-ali16:15
*** openstackstatus has joined #openstack-nova16:16
*** ChanServ sets mode: +v openstackstatus16:16
*** pm90__ has joined #openstack-nova16:16
*** pm90_ has quit IRC16:17
*** pm90_ has joined #openstack-nova16:17
openstackgerritBalazs Gibizer proposed openstack/nova: api: add soft-affinity policies for server groups  https://review.openstack.org/14752016:18
openstackgerritBalazs Gibizer proposed openstack/nova: scheduler: fill RequestSpec.instance_group.members  https://review.openstack.org/14827716:18
openstackgerritBalazs Gibizer proposed openstack/nova: scheduler: add soft-(anti-)affinity weighers  https://review.openstack.org/14751616:18
bauzasmriedem: johnthetubaguy: all the reno files have been written for M116:18
bauzashttps://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:reno_nova-m1,n,z16:18
*** atuvenie has joined #openstack-nova16:19
*** zenoway has quit IRC16:19
*** zenoway has joined #openstack-nova16:20
*** pm90__ has quit IRC16:20
*** hemnafk is now known as hemna16:20
mriedemok16:20
*** alexschm_ has quit IRC16:20
*** jlanoux has joined #openstack-nova16:20
openstackgerritMorgan Fainberg proposed openstack/nova: Don't drop uri path when selecting glance api  https://review.openstack.org/25113816:21
notmorganBobBall: ^ there is my stab at it, still need to chase tests down to be 100% sure we don't regress. comments welcome.16:21
BobBallLooks good to me - but let's wait to see what the XenServer CI says ;)16:24
notmorganBobBall: sure! and then i might poke at you again re testing if i can't chase some of this stuff down to ensure we don't start dropping the URI again.16:24
notmorganBobBall: :)16:24
*** zenoway has quit IRC16:24
notmorganBobBall: and seriously thnx for the help.16:24
*** mdrabe has joined #openstack-nova16:25
BobBallAwesome.  H2H and yeah, let's not drop it in future :) Can we configure devstack to set up a path?  If so, it'd be easy to either add it just to the XenServer CI or the generic CI to prove everything works?16:26
bauzasmriedem: I'd also love to get reviews for the novaclient reno change https://review.openstack.org/#/c/248895/16:27
bauzasmriedem: because we need to have reno in place the next time we cut a release16:28
bauzas(which is 3.0 IIUC)16:28
*** shaohe_feng has quit IRC16:29
mriedembauzas: ok16:29
mriedemwrapping up some havana psirt backport reviews on this fine morning :)16:29
*** baoli has quit IRC16:29
notmorganBobBall: i'm working on a strategy to have all services sub-url mounted - see https://gist.github.com/morganfainberg/c9b73fcd2b76d3a95b1416:30
*** vladnikityuk has quit IRC16:30
notmorganBobBall: so i expect we can do something to this effect for a "real" test long term. but shorter term just having a unit test that ensures the right value is there would be a big step forward :)16:30
*** baoli has joined #openstack-nova16:31
BobBallunit testing the xapi plugins is an interesting question indeed... I've got a meeting scheduled to talk about that one in February I think - the 30th if I remember rightly.16:32
*** mjura has quit IRC16:33
*** haypo_ has joined #openstack-nova16:34
*** mwagner_lap has quit IRC16:35
*** mwagner_lap has joined #openstack-nova16:35
*** vladnikityuk has joined #openstack-nova16:35
*** rook has quit IRC16:37
*** tjones has left #openstack-nova16:37
*** mjura_ has joined #openstack-nova16:38
*** mjura has joined #openstack-nova16:38
*** achanda_ has quit IRC16:41
*** paul-carlton1 has quit IRC16:43
*** mjura has quit IRC16:44
*** mgoddard_ has joined #openstack-nova16:44
*** xyang1 has joined #openstack-nova16:45
*** pm90__ has joined #openstack-nova16:45
*** pm90_ has quit IRC16:45
*** LukeHinds has quit IRC16:46
*** pm90_ has joined #openstack-nova16:46
*** haypo_ has quit IRC16:48
*** baoli has quit IRC16:48
*** mgoddard__ has quit IRC16:48
*** pm90__ has quit IRC16:49
*** baoli has joined #openstack-nova16:50
notmorgansdague: responded to your -1/comment on the glance uri patch.16:50
notmorgansdague: i think a quick land it and then solve the bigger shift to catalog is the better approach - mostly because this feels like something we can potentially backport, where moving to the catlog is potentially a lot more effort16:51
*** sacharya has joined #openstack-nova16:52
*** kmartin has joined #openstack-nova16:53
mriedembauzas: some wording issues in https://review.openstack.org/#/c/250849/16:53
*** armax has joined #openstack-nova16:54
bauzasmriedem: ack, respinning16:54
openstackgerritMarkus Zoeller (markus_z) proposed openstack/nova: config options: centralize libvirt options  https://review.openstack.org/25090716:54
*** atuvenie has quit IRC16:54
sdaguenotmorgan: that's how we end up with cruft that never leaves the code base16:54
sdagueand the url parsing ends up with all kinds of weird edge case and bugs, because you've now encoded all these assumptions in16:56
*** sacharya has quit IRC16:56
openstackgerritSylvain Bauza proposed openstack/nova: Add note for HyperV 2008 drop of support  https://review.openstack.org/25140816:57
openstackgerritSylvain Bauza proposed openstack/nova: Add note for force_config_drive opt change  https://review.openstack.org/25085216:57
notmorgansdague: well, like i said, I'm def. working on this cleanup and will continue on it. *shrug* i still view small fixes as a win especially in this case.16:57
openstackgerritSylvain Bauza proposed openstack/nova: Add note for deprecating local conductor  https://review.openstack.org/25084916:57
openstackgerritSylvain Bauza proposed openstack/nova: Add note for removing conductor RPC API v2  https://review.openstack.org/25086516:57
openstackgerritSylvain Bauza proposed openstack/nova: Add note for automatic determination of compute_rpc version by service  https://review.openstack.org/25143216:57
openstackgerritSylvain Bauza proposed openstack/nova: Add note for dropping InstanceV1 objects  https://review.openstack.org/25085916:57
openstackgerritSylvain Bauza proposed openstack/nova: Add note for Virtuozzo supporting snapshots  https://review.openstack.org/25142016:57
*** sacharya has joined #openstack-nova16:58
notmorgansdague: but i have a specific goal that will be incorporating these cleanup items, just the work to do catalog work is a bit further down the road16:58
*** achanda has joined #openstack-nova16:59
sdaguenotmorgan: I don't really see this as a small fix, because it also means that the user application needs to basically embed the same logic16:59
sdagueso it's creating an anti pattern16:59
sdaguewhere as referencing glance like we do cinder in the code would actually start converging us in the right direction, and support a pattern we want people to use17:00
*** coolsvap has quit IRC17:01
*** ildikov has quit IRC17:01
mriedemdansmith: are you happy with this release note for deprecation of local conductor? https://review.openstack.org/#/c/250849/17:01
mriedemit's pretty straight forward17:01
openstackgerritdane-fichter proposed openstack/nova: Functional test for ephemeral storage encryption  https://review.openstack.org/24007517:01
*** mgoddard_ has quit IRC17:02
*** otter768 has joined #openstack-nova17:02
*** mgoddard has joined #openstack-nova17:02
*** rook has joined #openstack-nova17:03
*** jaosorior has quit IRC17:03
*** rdo_ has quit IRC17:03
*** gcb has quit IRC17:05
*** coolsvap has joined #openstack-nova17:05
*** derekh has joined #openstack-nova17:05
*** Sukhdev_ has quit IRC17:06
*** krtaylor has joined #openstack-nova17:06
*** otter768 has quit IRC17:07
mriedembauzas: some issues with https://review.openstack.org/#/c/250852/ and https://review.openstack.org/#/c/251408/ - the rest i'm +W on17:07
notmorgansdague: well then i am going to have to defer a bit on the fix here. which i'm fine with, but just not to the point where i'm digging into how services communitcate to other services and the fixes i will be proposing along those lines.17:08
*** salv-orl_ has joined #openstack-nova17:08
*** sahid has quit IRC17:09
openstackgerritTang Chen proposed openstack/nova: Trivial: Fix a typo in power_state.py  https://review.openstack.org/25147417:10
*** paul-carlton1 has joined #openstack-nova17:11
*** salv-orlando has quit IRC17:11
bauzasmriedem: ack, will respin later17:11
*** Sukhdev_ has joined #openstack-nova17:12
openstackgerritTracy Jones proposed openstack/nova: Cleanup of compute api reboot method  https://review.openstack.org/12742717:13
sdaguenotmorgan: yeh, the problem is that api_servers is a minority configuration approach, so in the majority approach you won't support this at all because people are still configuring host & port as separate vars in config17:13
*** ccarmack has quit IRC17:13
notmorgansdague: that is a valid point. and i'm happier with that answer.17:13
notmorgansdague: :)17:13
sdaguenotmorgan: sorry, I've been in this code a ton17:14
*** eharney has joined #openstack-nova17:14
notmorgansdague: no worries :)17:14
mriedembauzas: do we need reno on stable/liberty novaclient? https://review.openstack.org/#/c/248895/17:14
*** EinstCrazy has joined #openstack-nova17:14
sdagueand the mess of it means I want no more "just one more fix"17:14
mriedembauzas: the email said that was optional17:14
notmorgansdague: i really only ran across this cause i'm working on a suburl mounted PoC17:14
sdaguenotmorgan: yep, I 100% support us getting there17:14
notmorgansdague: there is a lot of changes i'm catalogging to make this work, and this was the only showstopper of a bug btw i've really hiy17:14
notmorganhit*17:14
notmorgansdague: everything else pretyt much just works17:15
notmorgannow, i have to L7 route traffic, but that is fine. it shows we can do it17:15
sdaguebut introducing https://github.com/openstack/nova/blob/1734ce7101982dd95f8fab1ab4815bd258a33744/nova/volume/cinder.py#L42 to glance isn't a huge amount of work17:15
notmorganfixing services to run in wsgi containers is orthogonal to just "does this all work" :)17:15
sdagueand should solve this as well17:15
sdagueplus moves the whole ball forward17:15
*** boris-42_ has quit IRC17:15
mriedembauzas: nvm, i guess we need to since it's in the master toc tree17:16
notmorgansdague: nod. i'll poke at that in a bit, i just need to get to the point where i'm digging into that step before i take it on17:16
sdaguenotmorgan: sure17:16
notmorgansdague: because it's not the only change i need to roll in that kind of space.17:16
markus_zbauzas: If you have time, there is advice necessary in: https://review.openstack.org/#/c/246476/17:16
sdaguenotmorgan: ok, sure, I'm just saying the reason you hit this as the only big show stopper, is because glance was left off the list when jamielennox converted the services to use the catalog17:16
sdaguefor unknown reasons17:16
notmorgansdague: because glanceclient is special.17:17
notmorgansdague: or was waaaay more special until recently17:17
sdaguehmmm, ok17:17
paul-carlton1sdague, do you have time to look at https://review.openstack.org/#/c/228828/ please?17:21
*** EinstCrazy has quit IRC17:21
*** signed8bit is now known as signed8bit_ZZZzz17:21
*** ccarmack has joined #openstack-nova17:22
*** Piet has joined #openstack-nova17:22
*** vladnikityuk has quit IRC17:24
*** signed8bit_ZZZzz is now known as signed8bit17:27
*** rdo has joined #openstack-nova17:29
*** ccarmack has quit IRC17:30
*** vladnikityuk has joined #openstack-nova17:30
*** nic has joined #openstack-nova17:31
*** vladnikityuk has quit IRC17:31
*** vladnikityuk has joined #openstack-nova17:31
*** sacharya has quit IRC17:31
*** ccarmack has joined #openstack-nova17:33
*** diogogmt has joined #openstack-nova17:33
*** jlanoux has quit IRC17:36
sdaguepaul-carlton1: yeh, that interace looks like what I think we should do17:36
paul-carlton1alex_xu, and others seemed to want the original approach?17:38
*** nkrinner has quit IRC17:38
paul-carlton1I'm happy either way, but we need a consensus17:38
*** rook has quit IRC17:38
*** rook has joined #openstack-nova17:39
*** unicell has quit IRC17:39
*** mjura_ has quit IRC17:40
*** rdopiera has quit IRC17:41
*** ildikov has joined #openstack-nova17:42
*** smurke has joined #openstack-nova17:43
markus_zAny core here who has interest in this libvirt spec? https://review.openstack.org/#/c/234291/17:44
*** edmondsw has joined #openstack-nova17:44
openstackgerritAndrey Kurilin proposed openstack/python-novaclient: [microversions] Increase max version to 2.7  https://review.openstack.org/25148317:45
openstackgerritAndrey Kurilin proposed openstack/python-novaclient: [microversions] Add support for 2.8  https://review.openstack.org/25148417:45
openstackgerritAndrey Kurilin proposed openstack/python-novaclient: [microversions] Increase max version to 2.9  https://review.openstack.org/25148517:45
openstackgerritAndrey Kurilin proposed openstack/python-novaclient: [microversions] WIP: Add support for 2.10  https://review.openstack.org/25148617:45
*** Marga_ has quit IRC17:45
*** klkumar has quit IRC17:50
*** markus_z has quit IRC17:50
*** penick has joined #openstack-nova17:51
*** hblixt has joined #openstack-nova17:53
openstackgerritdane-fichter proposed openstack/nova: Functional test for ephemeral storage encryption  https://review.openstack.org/24007517:53
*** achanda has quit IRC17:54
openstackgerritRoman Podoliaka proposed openstack/nova: novnc: move config options into [novnc] group  https://review.openstack.org/22211317:56
*** salv-orl_ has quit IRC17:56
*** salv-orlando has joined #openstack-nova17:57
openstackgerritPushkar Umaranikar proposed openstack/nova: Modify VM's updated_at field on volume actions  https://review.openstack.org/24717617:57
*** bkopilov has joined #openstack-nova17:57
*** paul-carlton1 has quit IRC18:00
*** salv-orlando has quit IRC18:01
*** derekh has quit IRC18:02
*** penick_ has joined #openstack-nova18:02
*** ctrath1 has joined #openstack-nova18:03
*** penick has quit IRC18:04
*** penick_ is now known as penick18:04
*** jwang_ has quit IRC18:04
*** ctrath has quit IRC18:04
Lyncos1Hi, I'm I at the right place to get help with kilo and hugepages ?18:04
*** Kevin_Zheng has quit IRC18:06
ndipanovLyncos1, yeah18:06
Lyncos1I'm struggling setting 1GB hugepages... I always get :    Returning exception Page size 1048576 is not supported by the host. to caller18:07
Lyncos1when my numa for that host seems ok:18:07
Lyncos1http://pastebin.com/QxVBHThB18:08
Lyncos1from the mysql db ... I see  50x 1GB hugepages right ?18:08
*** jwang has joined #openstack-nova18:08
Lyncos1this is the full log: http://pastebin.com/byTJhSXg18:09
openstackgerritPushkar Umaranikar proposed openstack/nova: Modify VM's updated_at field on volume actions  https://review.openstack.org/24717618:09
Lyncos1it dosen't seems to go into the NUMA filter... not sure ?  in that aggregate I(with numa pinning and hugepages) I only have 1 host18:09
ndipanovLyncos1, how's your flavor set?18:10
*** harlowja has joined #openstack-nova18:10
*** mikeit has joined #openstack-nova18:10
ndipanovLyncos1, it's best not to specify the exact page size unless necessary and just go with large18:10
Lyncos1hw:cpu_policy = dedicated ,  hw:mem_page_size = 104857618:10
*** baoli_ has joined #openstack-nova18:10
Lyncos1Ok so In flavor I just set Large and it will use the largest possible ?18:11
*** edmondsw has quit IRC18:11
*** changbl has quit IRC18:12
ndipanovyeah18:12
ndipanovlarge18:12
ndipanovnot Large18:12
Lyncos1damn18:12
Lyncos1seems better18:12
*** paul-carlton1 has joined #openstack-nova18:12
Lyncos1now I get:   unable to create backing store for hugepages: Permission denied18:12
Lyncos1:-018:12
ndipanovLyncos1, cool18:12
Lyncos1I guess it is permission on mount point18:13
*** baoli has quit IRC18:14
*** changbl has joined #openstack-nova18:14
*** adam_g has quit IRC18:14
*** adam_g has joined #openstack-nova18:14
*** adam_g has quit IRC18:15
*** adam_g has joined #openstack-nova18:15
*** unicell has joined #openstack-nova18:16
Lyncos1 ndipanov: I changed perms to  775  and group kvm .. still not working .. any idea ?18:17
*** ndipanov has quit IRC18:17
*** anshprat has quit IRC18:17
*** baoli_ has quit IRC18:18
*** ctrath1 has quit IRC18:20
*** artom has quit IRC18:20
*** nic has quit IRC18:20
*** EinstCrazy has joined #openstack-nova18:21
*** davidlenwell has quit IRC18:21
*** ctrath has joined #openstack-nova18:21
*** davideagnello has joined #openstack-nova18:22
*** Guest44042 has quit IRC18:22
*** nikhil|afk has joined #openstack-nova18:22
*** nic has joined #openstack-nova18:22
*** swamireddy has quit IRC18:22
*** davidlenwell has joined #openstack-nova18:23
*** Guest83268 has quit IRC18:23
*** rods has quit IRC18:23
*** krtaylor has quit IRC18:23
*** renatoarmani has joined #openstack-nova18:23
*** renatoarmani has quit IRC18:24
*** brad[] has quit IRC18:25
*** EinstCrazy has quit IRC18:25
*** ctrath has quit IRC18:26
*** crinkle has quit IRC18:27
*** rods has joined #openstack-nova18:27
*** mgagne has joined #openstack-nova18:28
*** mgagne is now known as Guest6345318:28
*** doug-fish has quit IRC18:28
*** ctrath has joined #openstack-nova18:28
*** ssurana has joined #openstack-nova18:28
PietHi folks, we wanted to let you know about a results presentation for the Nova Networks to Neutron Migration Study this Tuesday, December 1st at 10AM MST / 12PM EST / 5:00PM UTC.  A link to the virtual room for the presentation can be found at https://wiki.openstack.org/wiki/HorizonUsability_Testing#December_2015_Nova_Networks_to_Neutron_Migration_Study18:29
*** artom has joined #openstack-nova18:29
*** crinkle has joined #openstack-nova18:29
*** penick has quit IRC18:29
*** aginwala has joined #openstack-nova18:31
*** penick has joined #openstack-nova18:32
*** pm90__ has joined #openstack-nova18:32
*** aweeks has quit IRC18:33
*** ndipanov has joined #openstack-nova18:33
*** paul-carlton1 has quit IRC18:33
*** claudiub has quit IRC18:34
*** pm90_ has quit IRC18:35
*** pm90___ has joined #openstack-nova18:35
*** cfriesen has joined #openstack-nova18:36
*** brad[] has joined #openstack-nova18:36
*** hogepodge has joined #openstack-nova18:37
*** jistr has joined #openstack-nova18:37
*** pm90__ has quit IRC18:38
*** Guest63453 has quit IRC18:38
*** Guest63453 has joined #openstack-nova18:38
*** Guest63453 is now known as mgagne18:39
openstackgerritSurojit Pathak proposed openstack/nova: Fix wrong CPU metric value in metrics_filter  https://review.openstack.org/24382518:40
*** krtaylor has joined #openstack-nova18:41
*** paul-carlton_ has quit IRC18:42
*** gszasz has quit IRC18:43
*** rook has quit IRC18:43
*** swamireddy has joined #openstack-nova18:46
*** hparekh has quit IRC18:47
*** jichen has joined #openstack-nova18:47
*** Marga_ has joined #openstack-nova18:48
*** mikeit has quit IRC18:51
*** sripriya has joined #openstack-nova18:51
*** vilobhmm11 has joined #openstack-nova18:52
*** achanda has joined #openstack-nova18:52
*** suro-patz has joined #openstack-nova18:54
*** hparekh has joined #openstack-nova18:55
*** PaulMurray is now known as ptm_away18:55
*** vilobhmm111 has joined #openstack-nova18:55
*** penick has quit IRC18:56
*** salv-orlando has joined #openstack-nova18:57
*** markmcclain has joined #openstack-nova18:57
*** vilobhmm11 has quit IRC18:58
openstackgerritaginwala proposed openstack/nova: fix free_gb value when scheduler receives None from compute  https://review.openstack.org/25076718:59
*** anshprat has joined #openstack-nova19:01
*** rotbeard has quit IRC19:02
*** aginwala has quit IRC19:02
*** otter768 has joined #openstack-nova19:03
*** sileht has quit IRC19:05
*** sulo has quit IRC19:06
*** aginwala has joined #openstack-nova19:07
*** sulo has joined #openstack-nova19:07
*** FL1SK has quit IRC19:07
*** penick has joined #openstack-nova19:08
*** otter768 has quit IRC19:08
*** jaosorior has joined #openstack-nova19:08
*** changbl has quit IRC19:09
*** rook has joined #openstack-nova19:09
*** unicell1 has joined #openstack-nova19:12
mriedemhas anyone seen this before? http://logs.openstack.org/94/242594/2/check/gate-tempest-dsvm-nova-v20-api/085f243/logs/screen-n-cpu.txt.gz?level=TRACE#_2015-11-30_18_03_48_00319:14
mriedemlooks like a race while deleting an instance that's still building19:14
*** unicell has quit IRC19:14
dansmithmriedem: yep, acked that relnote19:14
*** tjones has joined #openstack-nova19:14
*** jaosorior has quit IRC19:15
*** salv-orlando has quit IRC19:19
*** markmc has quit IRC19:19
*** salv-orlando has joined #openstack-nova19:20
*** markmc has joined #openstack-nova19:20
*** thangp has joined #openstack-nova19:20
*** Lyncos1 has left #openstack-nova19:21
*** xek has quit IRC19:21
*** zenoway has joined #openstack-nova19:23
*** aginwala is now known as aginwala8719:24
*** aginwala87 is now known as aginwala19:24
*** mkoderer has quit IRC19:26
*** aginwala has quit IRC19:27
*** mkoderer has joined #openstack-nova19:28
*** zhenq has joined #openstack-nova19:34
cfriesenhypothetically speaking...supposing I wanted to add some additional debug logs to scheduler filters (so that we would see exactly why a scheduling operation failed), would that require a blueprint or spec?  And would it make sense to make those messages available to end-users or would that be exposing too much data about the cloud internals?19:35
cfriesenI'm assuming that if we did expose it to end-users that'd require a spec.19:36
*** sacharya has joined #openstack-nova19:36
*** markmcclain has quit IRC19:41
*** markmcclain has joined #openstack-nova19:41
*** ssurana has quit IRC19:42
*** lucasagomes is now known as lucas-dinner19:42
*** ssurana has joined #openstack-nova19:42
*** aginwala has joined #openstack-nova19:43
*** penick has quit IRC19:48
mriedemexposing why you hit NoValidHost is going to require a spec because of security impacts19:50
*** penick has joined #openstack-nova19:50
mriedemadding debug logs does not require a blueprint or spec19:50
*** whenry_ has joined #openstack-nova19:51
mriedemcfriesen: ^19:51
cfriesenmriedem: cool.  What do you think about the idea of exposing why someone hit NoValidHost?  We're doing it internally, but our use-case is a bit different from the large-scale public cloud scenario.19:52
mriedemcfriesen: i'd suggest talking to edleafe since he looked at that in kilo19:54
edleafecfriesen: you mean give them the filters that rejected all the hosts?19:55
cfriesenedleafe: basically, yes.  The intent is to avoid the end-user having to go to the cloud admin to figure out why they couldn't schedule an instance19:56
cfriesensome of the more tricky ones are numa topology stuff, or pci passthrough19:56
edleafecfriesen: Ah, so like a summary of sorts?19:56
edleafecfriesen: IOW, "you asked for too much RAM"19:56
edleafecfriesen: without revealing details about the provider's physical hosts?19:57
ccarmackmriedem: I saw that build - delete race today also http://logs.openstack.org/43/226043/11/check/gate-tempest-dsvm-postgres-full/a9e02da/logs/screen-n-cpu.txt.gz?level=ERROR19:57
cfriesenedleafe: that'd be most likely to get in, I suspect.19:58
edleafecfriesen: but that's the point of keeping it accessible to admins only19:58
*** edtubill has joined #openstack-nova19:59
cfriesenedleafe: right, but if the admin is going to tell them "you asked for too much RAM", anyways, it seems like we might be able to just tell them that directly19:59
edleafecfriesen: well, it's complicated :)20:00
cfriesenedleafe: or maybe we just want to keep a person in the loop?  Either way, I think some of the more complicated filters could use better error logs20:00
cfriesennumatopologyfilter in particular has many different things that can fail20:00
edleafecfriesen: like they might have hosts with enough RAM, but those were rejected for some other reason20:00
*** penick has quit IRC20:03
cfriesenedleafe: fair enough.  It's probably not worth trying to get people to agree to a spec. :)   But I'm planning on trying to improve the error messages to make the admin's job easier at least, and it sounds like I can do that without a blueprint.  (Which simplifies things.)20:03
*** busterswt has joined #openstack-nova20:03
edleafecfriesen: agreed. But remember that better almost always means 'more info', which means bigger logs, which some admins will fight, which means...20:05
*** sacharya has quit IRC20:05
*** penick has joined #openstack-nova20:08
*** mpryc has joined #openstack-nova20:08
cfriesenedleafe: at a minimum, I think we could make it only log on a scheduler failure20:08
cfriesentry to cut down on the logs as much as possible.  maybe even have a config option of people really don't like it20:09
edleafecfriesen: right now the filtering info is only logged when no host is found20:11
edleafecfriesen: I don't think that yeat another config option is going to go over well :)20:11
*** doug-fish has joined #openstack-nova20:14
*** Marga_ has quit IRC20:19
cfriesenedleafe: yeah, was going to start without it and see if someone asks for it. :)20:19
*** tongli has joined #openstack-nova20:21
*** eharney has quit IRC20:22
*** rfolco has quit IRC20:24
sdaguecfriesen: honestly, if you first solve the problem of a plain english answer as a LOG.warn() for ops to easily see, that would be the first step20:24
sdaguethen a separate conversation about optionally exposing that all the way back to users20:25
sdagueand all the ops would love you if you got that LOG.warn in there20:25
mriedemthis bdm not found thing is everywhere http://logs.openstack.org/13/194413/7/check/gate-tempest-dsvm-neutron-full/9f99f8e/logs/screen-n-cpu.txt.gz?level=TRACE#_2015-11-30_15_51_27_61820:26
mriedemwe must have merged a race20:27
*** jistr has quit IRC20:27
mriedemw/o logstash we don't know when it was introduced really, but maybe not hard to find20:27
*** omnipresent has joined #openstack-nova20:29
*** ssurana has quit IRC20:29
mriedemsdague: see when this was merged? https://review.openstack.org/#/c/241179/20:29
mriedemhttps://review.openstack.org/#/c/241179/9/nova/context.py20:29
mriedemdifferent timestamp formats screwing up logstash and blowing things to hell?20:30
mriedemjeblair: ^20:30
*** ssurana has joined #openstack-nova20:30
*** achanda has quit IRC20:31
*** wznoinsk has joined #openstack-nova20:31
*** pm90___ has quit IRC20:31
*** pm90_ has joined #openstack-nova20:31
*** zenoway has quit IRC20:32
*** pm90__ has joined #openstack-nova20:32
*** apoorvad has quit IRC20:32
*** pm90___ has joined #openstack-nova20:33
*** pm90__ has quit IRC20:33
*** RichardRaseley has joined #openstack-nova20:34
*** pm90__ has joined #openstack-nova20:35
*** salv-orlando has quit IRC20:35
*** salv-orlando has joined #openstack-nova20:35
*** pm90_ has quit IRC20:37
*** eharney has joined #openstack-nova20:37
*** pm90___ has quit IRC20:38
*** aginwala has quit IRC20:39
wznoinskhi all20:40
wznoinskjohnthetubaguy: are you around?20:40
*** aginwala has joined #openstack-nova20:42
*** EinstCrazy has joined #openstack-nova20:42
*** matrohon has joined #openstack-nova20:44
bauzas:r20:44
bauzasmriedem: lemme try to pair your bluetooth brain for understanding why you -1'd https://review.openstack.org/#/c/251408/3 :D20:45
mriedembauzas: heh, it'd help if i posted the comments in ps220:45
mriedemyou had rev'ed ps3 while i was reviewing20:46
mriedemso i lost the comment20:46
*** EinstCrazy has quit IRC20:47
bauzasah ah20:47
*** rook has quit IRC20:48
*** Marga_ has joined #openstack-nova20:50
*** pm90__ has quit IRC20:51
*** doug-fish has quit IRC20:51
openstackgerritSylvain Bauza proposed openstack/nova: Add note for HyperV 2008 drop of support  https://review.openstack.org/25140820:53
*** belmoreira has joined #openstack-nova20:53
openstackgerritSylvain Bauza proposed openstack/nova: Add note for force_config_drive opt change  https://review.openstack.org/25085220:53
openstackgerritSylvain Bauza proposed openstack/nova: Add note for removing conductor RPC API v2  https://review.openstack.org/25086520:53
openstackgerritSylvain Bauza proposed openstack/nova: Add note for automatic determination of compute_rpc version by service  https://review.openstack.org/25143220:53
openstackgerritSylvain Bauza proposed openstack/nova: Add note for dropping InstanceV1 objects  https://review.openstack.org/25085920:53
openstackgerritSylvain Bauza proposed openstack/nova: Add note for Virtuozzo supporting snapshots  https://review.openstack.org/25142020:53
*** raildo is now known as raildo-afk20:53
*** dims_ has joined #openstack-nova20:53
*** pm90_ has joined #openstack-nova20:53
*** doug-fish has joined #openstack-nova20:54
*** pm90__ has joined #openstack-nova20:54
*** doug-fis_ has joined #openstack-nova20:55
*** dims has quit IRC20:56
*** mfisch` has quit IRC20:56
*** mfisch has joined #openstack-nova20:57
*** busterswt has quit IRC20:57
*** mfisch is now known as Guest715020:57
*** doug-fi__ has joined #openstack-nova20:57
*** pm90_ has quit IRC20:58
*** doug-fish has quit IRC20:58
*** sfinucan has joined #openstack-nova20:59
*** doug-fis_ has quit IRC21:00
*** tongli has quit IRC21:00
bauzassdague: I know this one has been +W'd but I'd love to get your insights https://review.openstack.org/#/c/247552/21:00
bauzassdague: in case something looks bad, we can just try to modify that in a later change21:01
*** Guest7150 has quit IRC21:01
*** doug-fi__ has quit IRC21:02
*** aginwala has quit IRC21:02
*** dims_ has quit IRC21:03
openstackgerritmelanie witt proposed openstack/nova: Add methods for RequestContext to switch db connection  https://review.openstack.org/16190621:03
*** otter768 has joined #openstack-nova21:04
*** Piet has quit IRC21:04
*** changbl has joined #openstack-nova21:04
*** doug-fish has joined #openstack-nova21:04
*** Marga_ has quit IRC21:05
bauzasmelwitt: dansmith: alaski: you could be interested in https://review.openstack.org/#/c/211753/21:05
*** Marga_ has joined #openstack-nova21:06
sdaguemriedem: where are you seeing that impacting log timestamps?21:06
sdaguebauzas: ok, I'd not yet seen that patch21:07
mriedemsdague: it changed the timestamp in the request context, but looking closer it looks like it correctly copied the isotime and strtime methods from oslo.utils' timeutils module21:07
mriedem*changed the method called to set the timestamp in the request context21:07
mriedemand merged on 11/24 which is when logstash went haywire21:08
sdagueyeh, so honestly, I think the timestamp parsing thing is a bit of a red herring as well21:08
sdaguewe had unparsable stuff before21:08
sdagueand it was fine21:08
*** otter768 has quit IRC21:08
*** doug-fish has quit IRC21:09
sdaguebauzas: so, the thing I don't like about that patch is that it logs once, and goes silent forever21:09
bauzassdague: if for each check, it goes an exception, right?21:10
*** Marga_ has quit IRC21:10
sdagueyeh21:10
sdaguehonestly, I kind of feel like it would be better to store the timestamp of the disconnected time21:10
mriedemsdague: heh, i think this is the cause for the bdm not found race https://review.openstack.org/#/c/194063/21:10
sdagueand if we're past some threshold, warn again21:10
mriedemagain, which was a rehash of a previous revert21:10
bauzastbh, I was rather thinking of passing the exception up, so the service would be killed21:11
bauzassdague: ^21:11
sdaguebauzas: wait, killing the service under what circumstance?21:11
bauzasbecause if something goes unexpected, the operator has to understand why21:11
sdaguebauzas: how would you do a rolling upgrade?21:11
bauzassdague: not thinking of the MessagingTimeout exception, just the catch-all exception21:12
mriedemyup http://logstash.openstack.org/#dashboard/file/logstash.json?query=message:%5C%22BDMNotFound:%20No%20Block%20Device%20Mapping%20with%20id%5C%22%20AND%20message:%5C%22_default_block_device_names%5C%22%20AND%20tags:%5C%22screen-n-cpu.txt%5C%2221:12
bauzaswe're only expecting something getting wrong with the communication to the controller21:12
mriedemstarted spiking on 11/2721:12
bauzassdague: the other exceptions are related to a situation where the compute is running, ie. when not upgradingi t21:13
*** aginwala has joined #openstack-nova21:13
sdaguecrashing the compute seems bad21:13
sdagueyeh, I'm ok with the catchall there honestly21:13
*** rook has joined #openstack-nova21:14
sdagueI'd just like to not go entirely silent on messaging timeouts21:14
bauzasokay, then that's me who overthought that21:14
*** dims has joined #openstack-nova21:14
sdagueyeh, we should always try to keep the compute up, because if it crashes then we lose the ability to know it's state in the rest of openstack21:15
sdaguethe only time we should have the compute exit is if it's going to corrupt data horribly21:16
*** READ10 has joined #openstack-nova21:16
bauzasgood point21:17
openstackgerritMatt Riedemann proposed openstack/nova: Revert "Detach volume after deleting instance with no host"  https://review.openstack.org/25154321:18
*** rdo has quit IRC21:18
dansmithand crashing the compute because of something you can send it over the network is, like, about the worst thing ever21:19
*** raildo-afk is now known as raildo21:19
*** FL1SK has joined #openstack-nova21:19
*** aginwala has quit IRC21:19
*** Marga_ has joined #openstack-nova21:19
bauzasokay, it seems I provided a brilliant idea... :)21:20
* bauzas jk21:20
bauzasI just usually don't like hiding stuff below the carpet21:21
bauzasthat said, I totally get the point where a compute has to be resilient21:21
mriedemstevemar: i'm not sure what i'm looking at in https://review.openstack.org/#/c/251538/1/openstackclient/compute/v2/server.py21:21
mriedemthere is no shared flag for the live migration api21:21
stevemarbah, no dtroyer21:22
stevemarmriedem: so how does someone perform a shared-migration?21:22
mriedemstevemar: they have shared storage setup with the compute21:22
mriedemit's automagic21:22
stevemarmriedem: is that something new?21:23
mriedemno21:23
stevemarmriedem: i think there is a misunderstanding about how this command should look like :(21:24
mriedemcan you just add an --easy-button option?21:24
stevemarmriedem: it sounds like there should only be one option, --migrate21:24
stevemarmriedem: file a bug and i'll mark it as wishlist21:24
*** vladikr has quit IRC21:25
mriedemso we have a big bag of dicts here in the libvirt driver https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L622421:25
mriedemdansmith has a series up to objectify that mess21:25
*** rdo has joined #openstack-nova21:26
dansmithand what a mess it is21:26
dansmithholy crap.21:26
mriedemstevemar: from the api docs:21:26
mriedem"block_migrationplainxsd:booleanSet to True to migrate local disks by using block migration. If the source or destination host uses shared storage and you set this value to True, the live migration fails."21:26
mriedemscrew you user if you don't know if the backing storage is shared or not21:26
*** mpryc has quit IRC21:26
mriedemgreat UX ^21:27
mriedemstevemar: you're probably best served at looking at the tempest tests for live migration21:27
mriedemhttps://github.com/openstack/tempest/blob/master/tempest/api/compute/admin/test_live_migration.py21:28
notmorgansdague: so... with the conversion to use the catalog... do we need to support the old options at all (old_opts set > new way?) or can we just cut it over21:32
notmorgansdague: i ask becasue it's much easier to do the "ignore the old options" method21:32
sdaguenotmorgan: we need to deprecate the old options21:32
notmorganhm. ok trying to figure out how to make this fall through.21:32
notmorganthen.21:32
*** doug-fish has joined #openstack-nova21:32
sdagueif the old options are set, they win, but we throw a deprecation warning21:33
notmorganyah21:33
notmorganhm21:33
sdagueif they aren't we use this new thing21:33
notmorganlike i said, trying to make this fall through... this is really awfully tacked together from waaaaaay outside the code21:33
notmorganit's going to be a pretty sizable change to convert over because it's passing the host and what not down like 5+ layers21:34
*** zenoway has joined #openstack-nova21:34
*** doug-fis_ has joined #openstack-nova21:35
*** doug-fis_ has quit IRC21:35
*** doug-fis_ has joined #openstack-nova21:35
sdagueyeh, though it looks like honestly a bunch of this code collapses in the process21:35
sdaguebecause eventually we assemble all to this - https://github.com/openstack/nova/blob/1734ce7101982dd95f8fab1ab4815bd258a33744/nova/image/glance.py#L15421:36
*** doug-fish has quit IRC21:37
notmorgansdague: it sortof does.21:37
notmorgansdague: part of the issue is that it also allows for direct parsing of the image location21:37
wznoinskjohnthetubaguy: whenever you have a free moment, could you have a look at https://blueprints.launchpad.net/nova/+spec/nova-hot-add-vcpus-support, feedback much appreciated (I'm trying to make before m1)21:37
*** aginwala has joined #openstack-nova21:37
notmorganso you can say "image is <<<here>>>" or so it looks21:38
mriedemdansmith: is there even a good reason we have the block migration option with live migrate? seems that the compute and virt drivers already figure that all out anyway and validate based on it https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L520521:38
sdaguenotmorgan: where do you see that?21:38
notmorganwhich isn't really a "use glance from catalog"21:38
*** omnipresent has quit IRC21:38
notmorgansec, looking for it again21:38
notmorganwas a few sections back from where i am staring now21:38
sdagueso, there is a lot of gorp in here for sure21:38
dansmithmriedem: well, I think it's one of those things that an operator may want to avoid.. like "do this migration if you can without block migration"21:39
mriedemi guess i forgot that it's an admin-only api by default21:40
mriedemand the admin should know if it's using shared storage or not21:40
*** READ10 has quit IRC21:40
notmorgansdague: https://github.com/openstack/nova/blob/1734ce7101982dd95f8fab1ab4815bd258a33744/nova/image/glance.py#L65921:40
notmorgansdague: or am i mis-reading this?21:40
*** Piet has joined #openstack-nova21:41
notmorganthis is def. hard to follow (and i'm ignoreing the XEN plugin atm)21:41
mriedemBobBall: is the xenserver ci in the toilet?21:41
sdaguegive me a few, this is a function I don't think I've been in before21:41
mriedemseems to be failing all day21:41
notmorgansdague: np. thnx for looking21:41
*** dims_ has joined #openstack-nova21:42
*** baoli has joined #openstack-nova21:43
*** smurke has quit IRC21:43
*** baoli has quit IRC21:43
*** dims has quit IRC21:43
cfriesenmriedem: there's a spec open for making nova live migration automatically figure out whether it's shared or not21:43
*** smurke has joined #openstack-nova21:44
dansmithcfriesen: this is not shared or not21:44
sdaguenotmorgan: the only place this seems to be used is in a xen conversion function21:44
dansmithAIUI, block migration can generate a crapton of network traffic21:44
dansmithwhich is the case where storage is *not* shared, and thus may want to be avoided21:44
notmorgansdague: ok21:45
sdaguenotmorgan: yeh, and the whole href path I think is never used21:45
sdaguehttps://github.com/openstack/nova/blob/1734ce7101982dd95f8fab1ab4815bd258a33744/nova/virt/xenapi/image/vdi_through_dev.py#L6421:45
sdagueso I think you are always returning from here - https://github.com/openstack/nova/blob/1734ce7101982dd95f8fab1ab4815bd258a33744/nova/image/glance.py#L67521:46
notmorgansdague: i see a ref in hyperv snapshotops21:46
notmorganthat does pass in the image_id... so might be safe. looking21:46
cfriesendansmith: right, it's separate from the shared instance storage.  but nova can figure out both cases.  this is what I was thinking of: https://review.openstack.org/#/c/24554321:46
notmorganyeah i think it is *just* the ID in that case too21:47
cfriesenmriedem: ^  feel free to review. :)  Personally I think we should just remove the block_migration flag entirely rather than making it optional, but Alex wants to be able to use it as a hint to the scheduler eventually.21:47
sdaguenotmorgan: so, this does show up a few other places, enough so that I'm not fully confident that no one uses the href version without more looking21:48
sdaguehowever, it seems unlikely21:48
notmorgansdague: yeah.21:48
notmorgansdague: if i could collapse all of that magic down just down into the glanceclientwrapper i'd be happy.21:48
*** mfisch has joined #openstack-nova21:48
notmorganbecause then it's easy to detect "old opts, new opts"21:48
*** mfisch has quit IRC21:48
*** mfisch has joined #openstack-nova21:48
dansmithcfriesen: right, but I'm not sure ops always want it to figure it out21:48
sdaguenotmorgan: yeh, I think that's definitely doable21:48
dansmithcfriesen: because if you have two nodes that might be on the same NFS, you want to live migrate if you can, else not go and do block migration or something21:49
*** aginwala has quit IRC21:49
dansmithcfriesen: the shared/not-shared flag seemed superfluous, but the block migration one is not the same, IIRC21:49
notmorgani can *probably* just pass down image_id and then do the same splitout if needed.21:49
notmorgansdague: have i mentioned i don't like rabbitholes like this :P21:50
notmorganat least this one isn't insane21:50
cfriesendansmith: then it should probably be renamed.  as it stands, it's misleading since it appears to be asking for information that nova already knows (ie do we need to do a block migration or not)21:50
sdagueright, well this is one of the reasons I -1ed fast, because it would just paper over some of this, and I think it's actually fixable not with a huge amount of effort21:50
cfriesendansmith: and if you want to limit the live migration to nodes that share storage, isn't that something that host aggregates could be used for?21:51
dansmithcfriesen: I interpret is "I am willing (or not) to do a block migration" and thus I think it's fine21:51
notmorgansdague: maybe.21:51
notmorgansdague: for what it's worth this has taken ~3x the time to dig through as the previous stuff. :P21:51
notmorganat least.21:51
dansmithcfriesen: regardless, I'm not the expert here, so you should talk to some of the live  migration people about it21:51
cburgesscfriesen dansmith I just left feedback. I like the overall idea but I think the existing block migrate flag should be kept as is and a new auto-magic flag added.21:51
*** rook has quit IRC21:51
notmorganbut i admit i'm not a nova expert21:51
sdaguenotmorgan: sure, but the end result is going to pay it forward in debt reduction21:51
cfriesendansmith: so would you support Alex's suggestion of making it optional (which would mean "do whatever is needed to migrate this instance")?21:52
*** rook has joined #openstack-nova21:52
notmorgansdague: maybe depends on how frustrated i get :P [so far not too bad]21:52
*** baoli has joined #openstack-nova21:52
dansmithcfriesen: I haven't read the spec and I'd need to see how that affects the API21:52
cfriesencburgess: you don't think the microversion bump is sufficient?21:53
dansmithcfriesen: microversion bump to change a default-if-not-specified thing is kinda sucky21:54
cburgessThe short answer is no. The reason I say that is because as you stated you can't figure out if something is a block migrate before hand via the nova API. I suspect some sites use the combination of the command and flag to accomplish this.21:54
cburgessAs in... "I want to live migrate this but *only* if its *not* block.21:54
*** jichen has quit IRC21:55
cburgessThats what some people want to do today.21:55
*** matrohon has quit IRC21:55
dansmithcburgess: right, exactly21:55
cburgessI like having a "figure it out and make it work" flag. I absolutely want that.21:55
cburgessBut I think you need to have the ability to do the former.21:55
cfriesencburgess: okay, I can see it.21:55
cburgessIf you want to change the flags around and add more and make it possible to still do that great. But your current spec doesn't seem to support that type of use case.21:55
*** penick has quit IRC21:56
*** andrewbogott has quit IRC21:56
*** andrewbogott has joined #openstack-nova21:56
cburgessSo to be clear.. its not that I think the flag needs to not change. As long as you do a microversion change away, but you need to support the other use case. I'll update my comment on the spec to more properly reflect that.21:56
*** mwagner_lap has quit IRC21:57
notmorgansdague: oooh i think i can do this in one spot already. it'll mean later patches to do cleanup21:57
*** rook has quit IRC21:57
notmorgansdague: i think i found the magic place to do the heavy lifting to start.21:57
sdaguenotmorgan: ok, cool21:58
notmorgansdague: if i just do the work in get_api_servers() that preserves the same functionality everywhere else.21:58
sdagueI need to start the dinner routine, but I'm happy to help review to get this into landable state laster this week21:58
sdaguenotmorgan: yep21:58
sdaguethat sounds reasonable21:58
notmorgansdague: then we can do surgical cleanup on the rest after21:58
sdaguesure21:58
notmorgansdague: cool will ping you (maybe tomorrow) on this change then21:58
sdaguesounds good21:58
*** penick has joined #openstack-nova21:59
*** aginwala has joined #openstack-nova21:59
*** pm90__ has quit IRC21:59
*** mriedem is now known as mriedem_away22:00
*** Guest87643 has joined #openstack-nova22:03
cfriesendoes anyone know why we convert from COW to a pure image when doing a cold migration with qcow2 backend?  This is unintuitive, since if the backing store file is already present on the destination we should be able to copy only the delta file.22:05
*** Marga__ has joined #openstack-nova22:05
*** thorst_ has quit IRC22:05
*** Marga_ has quit IRC22:06
*** thorst_ has joined #openstack-nova22:06
*** markvoelker has joined #openstack-nova22:07
*** Marga__ has quit IRC22:09
*** thorst_ has quit IRC22:10
*** pm90_ has joined #openstack-nova22:14
*** rlrossit has left #openstack-nova22:17
*** vilobhmm111 has quit IRC22:18
*** mpryc has joined #openstack-nova22:18
*** vilobhmm11 has joined #openstack-nova22:19
*** eharney has quit IRC22:20
*** thangp has quit IRC22:22
*** ZZelle has joined #openstack-nova22:23
*** omnipresent has joined #openstack-nova22:28
*** zenoway has quit IRC22:32
*** lykinsbd has quit IRC22:35
*** ccarmack has left #openstack-nova22:42
*** doug-fis_ has quit IRC22:44
*** mpryc has left #openstack-nova22:44
*** doug-fish has joined #openstack-nova22:44
belmoreiraHi, I need your help to recover from an issue related with flavor migration in Kilo22:47
belmoreiraWe use cells and the flavor migration is not updating instance_extra in the api cell DB when the stop api is called from the compute node22:47
belmoreiraThis happens when a user shutdown an instance for example...22:47
belmoreiraIn cell DB everything seems updated correctly22:47
belmoreiraThen the user can't list VMs anymore22:48
*** loquacities has left #openstack-nova22:48
dansmithbelmoreira: running all kilo code at the top and bottom?22:48
*** doug-fish has quit IRC22:49
belmoreira@dansmith I still have some nodes in Juno. But all are "juno compat"22:49
*** tjones has quit IRC22:49
*** doug-fish has joined #openstack-nova22:49
dansmithbelmoreira: which kind of nodes?22:50
dansmithbelmoreira: either way, running a kilo api cell and a juno child cell is probably not going to work22:50
melwittbelmoreira: are the failures to update with only juno computes or any compute?22:50
belmoreira@dansmith: only compute nodes22:50
*** doug-fis_ has joined #openstack-nova22:50
belmoreiradansmith: If I run "nova stop <uuid>" everything is updated correctly. The problem that I see is when an instance is shutdown and is the compute node that detects it and calls stop api22:52
dansmithbelmoreira: well, given the lack of any care we give to live upgrade scenarios on cells, I'm not at all surprised that being in that arrangement results in pain22:52
dansmithbelmoreira: yeah, which is wonky in cells, I think22:52
*** penick has quit IRC22:52
*** kylek3h has quit IRC22:53
belmoreirain the case that I described the cell DB is updated correctly but not the api cell DB. Misses the instance_extra22:53
dansmithright22:54
*** doug-fish has quit IRC22:54
*** pm90_ has quit IRC22:54
belmoreiraI'm starting to have a lot of projects in this state... not sure how can recover22:54
*** doug-fis_ has quit IRC22:55
belmoreiramelwitt: all compute nodes (juno or kilo)22:55
* dansmith got a phone call22:55
*** Marga_ has joined #openstack-nova22:56
*** Marga_ has quit IRC22:57
*** Marga_ has joined #openstack-nova22:57
*** mwagner_lap has joined #openstack-nova22:57
*** yamahata has joined #openstack-nova22:58
melwittbelmoreira: it's a known problem that a mix of kilo and juno cells won't work. if kilo -> kilo instance update isn't propagating to the api, that would be a new problem22:58
belmoreiramelwitt: is kilo->kilo but still using juno compat22:59
*** tjones has joined #openstack-nova23:00
melwittbelmoreira: what do you mean by "using juno compat"?23:00
dansmithmelwitt: rpc api23:01
dansmithbut still,23:01
dansmithnone of that stuff is done properly in cells23:01
dansmithwhich is why the mix of versions doesn't work23:01
*** Marga_ has quit IRC23:01
*** omnipresent has quit IRC23:02
belmoreiramelwitt: upgrade_levels: compute=juno23:02
*** dims has joined #openstack-nova23:02
dansmithbelmoreira: that's only in-cell communication23:02
*** doug-fish has joined #openstack-nova23:03
*** dims_ has quit IRC23:03
melwittokay. yeah. I expect the kilo computes to succeed in updating api, but the juno computes won't23:03
dansmithyeah23:03
dansmithand I'm sure there are some compat gotchas in the flavor stuff when getting an older update over the cells link23:03
vilobhmm11melwitt, dansmith : if you can have a look at https://review.openstack.org/#/c/222423/ it would be helpful.23:03
vilobhmm11servicegroup spec23:03
*** changbl has quit IRC23:04
*** Marga_ has joined #openstack-nova23:04
*** otter768 has joined #openstack-nova23:05
belmoreiraI'm not running upgrade_levels in the api cell... should I? I'm only running upgrade levels in child cell controllers and compute nodes23:05
*** mdrabe has quit IRC23:05
*** changbl has joined #openstack-nova23:05
dansmithbelmoreira: it won't matter :)23:05
belmoreirayeah, is what I thought23:06
*** aginwala_ has joined #openstack-nova23:06
melwittbelmoreira: this explains more if you haven't seen it yet https://bugs.launchpad.net/nova/+bug/147918123:06
openstackLaunchpad bug 1479181 in OpenStack Compute (nova) "Cells: Build instance doesn't work with kilo api, juno compute " [Low,Invalid]23:06
*** doug-fish has quit IRC23:07
*** ctrath has quit IRC23:07
*** edtubill has quit IRC23:09
*** otter768 has quit IRC23:09
*** aginwala has quit IRC23:10
belmoreiramelwitt: thanks23:11
belmoreiradansmith, melwitt: do you know how people recovered from https://review.openstack.org/#/c/186505/23:12
belmoreirabecause the consequence is more or less the same23:13
*** smurke_ has joined #openstack-nova23:13
*** smurke has quit IRC23:13
dansmithbelmoreira: ran flavor migrations again I expect23:13
belmoreiraI was waiting to have everything in kilo and remove the upgrade_levels before doing it...23:14
*** breitz has quit IRC23:15
*** breitz has joined #openstack-nova23:15
dansmithwell, it's already happening in the background any time one of your kilo nodes saves an instance23:16
*** zenoway has joined #openstack-nova23:16
*** burt has quit IRC23:17
belmoreiradansmith: fair enough23:18
*** smurke_ has quit IRC23:19
*** zenoway has quit IRC23:20
*** ijuwang has quit IRC23:21
*** suro-patz1 has joined #openstack-nova23:23
*** suro-patz has quit IRC23:24
openstackgerritEd Leafe proposed openstack/nova: Docs: update the concept guide for Host topics  https://review.openstack.org/24994323:25
belmoreiradansmith, melwitt: thank you23:27
*** ZZelle has quit IRC23:33
*** pm90_ has joined #openstack-nova23:34
*** suro-patz1 has quit IRC23:36
*** barra204 has quit IRC23:38
*** shakamunyi has joined #openstack-nova23:38
*** jerrygb has quit IRC23:39
*** jerrygb has joined #openstack-nova23:39
*** xyang1 has quit IRC23:41
*** suro-patz has joined #openstack-nova23:43
*** signed8bit is now known as signed8bit_ZZZzz23:48
*** hemna is now known as hemnafk23:48
*** penick has joined #openstack-nova23:54
*** zenoway has joined #openstack-nova23:54
*** RichardRaseley has quit IRC23:54
*** RichardRaseley has joined #openstack-nova23:56
*** zenoway has quit IRC23:58
*** suro-patz has quit IRC23:59
*** suro-patz has joined #openstack-nova23:59
« Saturday, 2015-11-28 Index Tuesday, 2015-12-01 »

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