Tuesday, 2016-09-27

*** salv-orlando has joined #openstack-cinder00:01
*** chris_morrell has quit IRC00:04
*** 7IZAAPZ3F has quit IRC00:04
*** ducttape_ has joined #openstack-cinder00:05
*** salv-orlando has quit IRC00:05
*** ducttape_ has quit IRC00:06
*** yangyapeng has quit IRC00:07
*** ducttape_ has joined #openstack-cinder00:10
*** kaisers has joined #openstack-cinder00:16
*** kaisers2 has joined #openstack-cinder00:17
*** kaisers4 has quit IRC00:17
*** kaisers1 has quit IRC00:17
*** lixiaoy1 has joined #openstack-cinder00:18
*** ducttape_ has quit IRC00:23
*** leeantho_ has quit IRC00:24
*** lixiaoy1 has quit IRC00:28
*** markvoelker has quit IRC00:30
*** baojg has quit IRC00:32
*** Apoorva has quit IRC00:34
*** david-lyle_ has joined #openstack-cinder00:38
*** laughterwym has joined #openstack-cinder00:38
*** laughterwym has quit IRC00:39
*** ducttape_ has joined #openstack-cinder00:39
*** laughterwym has joined #openstack-cinder00:39
*** zhangguoqing has joined #openstack-cinder00:40
*** david-lyle has quit IRC00:41
*** ducttape_ has quit IRC00:49
*** bardia has quit IRC00:50
*** bardia has joined #openstack-cinder00:51
*** bardia has joined #openstack-cinder00:51
*** gershon_ has joined #openstack-cinder00:52
*** gershon has joined #openstack-cinder00:52
*** GB21 has joined #openstack-cinder00:52
*** mriedem is now known as mriedem_debate00:54
*** aohuanxuan has joined #openstack-cinder01:00
*** salv-orlando has joined #openstack-cinder01:02
*** Apoorva has joined #openstack-cinder01:05
*** salv-orlando has quit IRC01:06
*** kaisers_ has quit IRC01:14
*** EinstCrazy has joined #openstack-cinder01:15
*** yangyapeng has joined #openstack-cinder01:19
*** mtanin___ has joined #openstack-cinder01:23
*** mtanino has quit IRC01:24
*** diablo_rojo has joined #openstack-cinder01:24
*** sdake has quit IRC01:28
*** markvoelker has joined #openstack-cinder01:31
*** aspiers has quit IRC01:34
*** diablo_rojo has quit IRC01:34
*** harlowja has quit IRC01:35
*** sage_ has quit IRC01:36
*** markvoelker has quit IRC01:36
*** sage_ has joined #openstack-cinder01:37
*** gershon_ has quit IRC01:37
*** gershon has quit IRC01:37
*** aspiers has joined #openstack-cinder01:39
*** ducttape_ has joined #openstack-cinder01:50
*** ducttape_ has quit IRC01:54
*** bkopilov has quit IRC01:55
*** lprice has joined #openstack-cinder01:58
*** raunak has quit IRC01:59
*** haplo37_ has quit IRC02:00
*** haplo37_ has joined #openstack-cinder02:02
*** bardia has quit IRC02:02
*** salv-orlando has joined #openstack-cinder02:03
*** ebalduf has joined #openstack-cinder02:07
*** salv-orlando has quit IRC02:07
*** aohuanxuan has quit IRC02:09
*** ebalduf has quit IRC02:09
*** rhefner has joined #openstack-cinder02:21
*** aohuanxuan has joined #openstack-cinder02:22
*** nicolasbock has quit IRC02:29
*** GB21 has quit IRC02:37
*** Lee1092 has joined #openstack-cinder02:41
*** david-lyle_ has quit IRC02:48
*** mtanin___ has quit IRC02:49
*** david-lyle has joined #openstack-cinder02:49
*** zengyingzhe__ has quit IRC02:50
*** zengyingzhe__ has joined #openstack-cinder02:50
*** ducttape_ has joined #openstack-cinder02:51
*** GB21 has joined #openstack-cinder02:54
*** sdake has joined #openstack-cinder02:55
jgriffithsmcginnis: scottda so what's the rule/goal for new api modules? V3, V2, use contrib? All of the above?02:56
*** ducttape_ has quit IRC02:56
*** itzdilip has joined #openstack-cinder02:58
openstackgerritTommyLike proposed openstack/cinder-specs: Add volume type filter to Get-Pools  https://review.openstack.org/36274702:58
*** mriedem_debate has quit IRC02:59
*** adisky has joined #openstack-cinder03:00
*** sudipto has joined #openstack-cinder03:03
*** sudipto_ has joined #openstack-cinder03:04
*** david-lyle has quit IRC03:04
zhengyingood morning03:04
*** Apoorva has quit IRC03:06
*** bkopilov has joined #openstack-cinder03:09
*** sdake has quit IRC03:15
*** sdake has joined #openstack-cinder03:23
*** lprice1 has joined #openstack-cinder03:31
*** lprice has quit IRC03:31
*** wanghao_ has joined #openstack-cinder03:32
sudiptozhengyin, good morning, may i ask you a question w.r.t LVM?03:32
*** markvoelker has joined #openstack-cinder03:32
*** raunak has joined #openstack-cinder03:36
*** markvoelker has quit IRC03:38
*** sdake has quit IRC03:40
*** sdake has joined #openstack-cinder03:41
*** aswadr_ has joined #openstack-cinder03:45
*** lkuchlan has joined #openstack-cinder03:47
*** sdake has quit IRC03:47
openstackgerritTuan Luong-Anh proposed openstack/cinder: TrivialFix: Remove default=None when set value in Config  https://review.openstack.org/37719403:50
*** karthikp- has quit IRC03:52
*** GB21 has quit IRC03:53
*** hemna_ has joined #openstack-cinder03:59
*** hemna_ has quit IRC04:00
*** rmcall_ has joined #openstack-cinder04:13
*** rmcall has quit IRC04:14
*** rmcall_ is now known as rmcall04:14
*** sdake has joined #openstack-cinder04:15
*** links has joined #openstack-cinder04:19
*** sdake_ has joined #openstack-cinder04:22
*** sdake has quit IRC04:24
*** shausy has joined #openstack-cinder04:27
*** coolsvap has quit IRC04:32
*** GB21 has joined #openstack-cinder04:36
*** sudipto_ has quit IRC04:39
*** sudipto has quit IRC04:39
*** sdake_ has quit IRC04:40
*** sdake has joined #openstack-cinder04:43
*** gouthamr has quit IRC04:44
*** gouthamr has joined #openstack-cinder04:45
*** gouthamr has quit IRC04:45
*** psachin has joined #openstack-cinder04:45
*** sdake_ has joined #openstack-cinder04:46
*** sdake has quit IRC04:48
*** bks has joined #openstack-cinder04:52
*** ducttape_ has joined #openstack-cinder04:53
*** salv-orlando has joined #openstack-cinder04:54
*** ducttape_ has quit IRC04:57
*** lpetrut has joined #openstack-cinder04:58
*** sdake_ has quit IRC04:58
*** kaisers_ has joined #openstack-cinder05:03
*** pgadiya has joined #openstack-cinder05:05
*** laughter_ has joined #openstack-cinder05:07
*** pdeore has joined #openstack-cinder05:07
*** pdeore has quit IRC05:08
*** pdeore has joined #openstack-cinder05:08
*** pdeore has quit IRC05:09
*** pdeore has joined #openstack-cinder05:09
*** laughterwym has quit IRC05:11
*** pdeore has quit IRC05:12
*** jgregor has joined #openstack-cinder05:12
*** pdeore has joined #openstack-cinder05:13
*** jgregor has quit IRC05:13
*** pbandark has joined #openstack-cinder05:16
*** sudipto has joined #openstack-cinder05:20
*** sudipto_ has joined #openstack-cinder05:20
*** snarwade has quit IRC05:21
*** snarwade has joined #openstack-cinder05:21
*** ducttape_ has joined #openstack-cinder05:25
*** ducttape_ has quit IRC05:27
*** ducttape_ has joined #openstack-cinder05:27
*** ducttape_ has quit IRC05:27
*** lpetrut has quit IRC05:29
*** cknight has joined #openstack-cinder05:36
openstackgerritTony Xu proposed openstack/python-brick-cinderclient-ext: Add oslo.concurrency to requirements  https://review.openstack.org/37724705:38
*** Poornima has joined #openstack-cinder05:53
*** rajinir has quit IRC05:55
*** arecknag has joined #openstack-cinder06:02
*** jay_ has joined #openstack-cinder06:03
*** mjura has joined #openstack-cinder06:04
*** nherciu has joined #openstack-cinder06:05
*** mjura has quit IRC06:11
*** mjura has joined #openstack-cinder06:12
*** liverpooler has joined #openstack-cinder06:14
*** liverpooler has quit IRC06:14
*** liverpooler has joined #openstack-cinder06:14
openstackgerritAnh Tran proposed openstack/python-cinderclient: Import module instead of object  https://review.openstack.org/37726306:19
*** laughter_ has quit IRC06:22
*** bluex has quit IRC06:22
*** laughterwym has joined #openstack-cinder06:22
*** cknight has quit IRC06:23
openstackgerritCao Xuan Hoang proposed openstack/os-brick: Docstrings should not start with a space  https://review.openstack.org/37726506:27
*** bluex has joined #openstack-cinder06:30
*** raunak has quit IRC06:36
*** pcaruana has joined #openstack-cinder06:45
*** GB21 has quit IRC06:47
*** rcernin has joined #openstack-cinder06:59
*** salv-orl_ has joined #openstack-cinder07:01
*** salv-orlando has quit IRC07:04
*** salv-orl_ has quit IRC07:05
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Imported Translations from Zanata  https://review.openstack.org/37728307:07
*** GB21 has joined #openstack-cinder07:09
*** ducttape_ has joined #openstack-cinder07:10
*** GB21 has quit IRC07:13
*** GB21 has joined #openstack-cinder07:13
*** ducttape_ has quit IRC07:15
*** GB21 has quit IRC07:15
*** GB21 has joined #openstack-cinder07:15
*** nkrinner_afk is now known as nkrinner07:20
*** alonma has joined #openstack-cinder07:24
*** xek has joined #openstack-cinder07:25
*** mjura has quit IRC07:34
*** mjura has joined #openstack-cinder07:35
*** hoonetorg has quit IRC07:42
*** anteaya has quit IRC07:48
*** anteaya has joined #openstack-cinder07:49
*** jordanP has joined #openstack-cinder07:50
*** hoonetorg has joined #openstack-cinder07:59
*** e0ne has joined #openstack-cinder07:59
*** zzzeek has quit IRC08:00
*** zzzeek has joined #openstack-cinder08:00
*** e0ne has quit IRC08:01
openstackgerritxianming.mao proposed openstack/cinder: add launchpad url in setup.cfg  https://review.openstack.org/36668608:11
*** ducttape_ has joined #openstack-cinder08:12
*** ducttape_ has quit IRC08:17
*** jay_ has quit IRC08:18
*** laughterwym has quit IRC08:27
*** mdovgal has joined #openstack-cinder08:29
*** e0ne has joined #openstack-cinder08:30
*** Digvijay has joined #openstack-cinder08:43
*** bhagyashri has joined #openstack-cinder08:44
*** lpetrut has joined #openstack-cinder08:46
*** mjura has quit IRC08:49
*** hoonetorg has quit IRC08:51
*** GB21 has quit IRC08:57
*** hoonetorg has joined #openstack-cinder09:03
*** GB21 has joined #openstack-cinder09:10
*** alonma has quit IRC09:12
*** ducttape_ has joined #openstack-cinder09:12
*** alonma has joined #openstack-cinder09:13
*** namnh has joined #openstack-cinder09:13
namnhHi everyone. Do you know how to setup Devstack in oder that Cinder can use NFS backend. :) Thanks09:15
*** obutenko_ has joined #openstack-cinder09:16
openstackgerritAnh Tran proposed openstack/python-cinderclient: TrivialFix: Removed redundant 'the'  https://review.openstack.org/37738909:16
*** alonma has quit IRC09:17
*** ducttape_ has quit IRC09:18
openstackgerritAnh Tran proposed openstack/python-cinderclient: TrivialFix: Removed redundant 'the'  https://review.openstack.org/37738909:20
*** alonma has joined #openstack-cinder09:20
e0nenamnh: hi. you have to setup NFS server manually and configure devstack to use NFS driver09:20
namnhe0ne, thanks for your repley. Yes, correct :) Would you mind giving me some comments.09:22
namnhe0ne, I tried to do some ways but it was not correct. :(09:23
*** alonma has quit IRC09:25
*** e0ne has quit IRC09:27
*** e0ne has joined #openstack-cinder09:27
*** mvk has quit IRC09:27
*** GB21 has quit IRC09:28
*** namnh has quit IRC09:28
*** namnh has joined #openstack-cinder09:29
e0nenamnh: do you have any errors?09:29
namnhe0ne: I cannot assign cinder-volumn using NFS, it means that when I finished installing Cinder by Devstack, the cinder.conf still setup using LVM09:33
*** laughterwym has joined #openstack-cinder09:35
*** laughterwym has quit IRC09:39
e0nenamnh: did you add CINDER_ENABLED_BACKENDS=nfs:nfs to your devstack config?09:41
*** GB21 has joined #openstack-cinder09:41
*** namnh has quit IRC09:42
*** itzdilip has quit IRC09:43
*** akapil has joined #openstack-cinder09:45
*** akapil has quit IRC09:50
*** jay_ has joined #openstack-cinder09:51
*** winston-d has quit IRC09:53
*** yangyapeng has quit IRC09:53
*** yangyapeng has joined #openstack-cinder09:53
*** rhefner has quit IRC09:54
*** haplo37_ has quit IRC09:56
*** Dinesh_Bhor has quit IRC09:57
*** jmccrory has quit IRC09:57
*** cebruns__ has joined #openstack-cinder09:57
*** cebruns_ has quit IRC09:57
*** aohuanxuan has quit IRC09:58
*** mvk has joined #openstack-cinder09:58
*** haplo37_ has joined #openstack-cinder09:59
*** jmccrory has joined #openstack-cinder10:00
*** melwitt has quit IRC10:01
*** melwitt has joined #openstack-cinder10:01
*** melwitt is now known as Guest320310:02
*** EinstCrazy has quit IRC10:03
*** EinstCrazy has joined #openstack-cinder10:04
*** yangyapeng has quit IRC10:05
*** sdague has joined #openstack-cinder10:08
*** EinstCrazy has quit IRC10:08
*** alonma has joined #openstack-cinder10:11
*** akapil has joined #openstack-cinder10:11
*** aohuanxuan has joined #openstack-cinder10:14
*** Lee1092 has quit IRC10:14
*** ducttape_ has joined #openstack-cinder10:15
*** ducttape_ has quit IRC10:19
*** sdake has joined #openstack-cinder10:27
*** portdirect has quit IRC10:39
*** salv-orlando has joined #openstack-cinder10:39
*** scottda has quit IRC10:40
*** alonma has quit IRC10:40
*** gcb has joined #openstack-cinder10:46
*** akapil has quit IRC10:47
*** akapil has joined #openstack-cinder10:48
*** alonma has joined #openstack-cinder10:48
*** nicolasbock has joined #openstack-cinder10:50
*** alonma has quit IRC10:50
*** alonma has joined #openstack-cinder10:51
*** nkrinner has quit IRC10:54
*** pdeore has quit IRC11:01
*** sudipto has quit IRC11:06
*** sudipto_ has quit IRC11:06
*** nkrinner has joined #openstack-cinder11:07
*** sdake has quit IRC11:12
*** ducttape_ has joined #openstack-cinder11:15
*** nherciu_ has joined #openstack-cinder11:17
*** yangyapeng has joined #openstack-cinder11:17
*** nherciu has quit IRC11:19
*** ducttape_ has quit IRC11:20
*** salv-orlando has quit IRC11:27
johnthetubaguyjgriffith: hemna: ildikov: https://review.openstack.org/373203 is updated after the meeting yesterday11:28
johnthetubaguyjgriffith: hemna: ildikov: feels like we need a call back to say os-brick has finished attaching things, so other volumes can re-use that host connection, or something like that, but I am not sure11:29
openstackgerritMerged openstack/cinder: Imported Translations from Zanata  https://review.openstack.org/37728311:32
*** sudipto has joined #openstack-cinder11:38
*** sudipto_ has joined #openstack-cinder11:39
*** salv-orlando has joined #openstack-cinder11:40
*** Poornima has quit IRC11:42
*** pgadiya has quit IRC11:43
*** akapil has quit IRC11:45
*** narayrak has joined #openstack-cinder11:49
*** bkopilov has quit IRC11:50
*** wanghao_ has quit IRC11:50
*** breitz has quit IRC11:56
*** jgregor has joined #openstack-cinder11:56
*** pgadiya has joined #openstack-cinder11:57
*** sandanar has joined #openstack-cinder12:03
openstackgerritMerged openstack/python-cinderclient: TrivialFix: Removed redundant 'the'  https://review.openstack.org/37738912:04
*** sandanar_ has joined #openstack-cinder12:04
*** sandanar has quit IRC12:05
*** Digvijay has quit IRC12:06
*** kumarb has joined #openstack-cinder12:06
*** bks has quit IRC12:06
*** yangyapeng has quit IRC12:10
*** jseiler__ has quit IRC12:10
*** rhe00 has joined #openstack-cinder12:14
*** rhedlind has quit IRC12:15
*** yangyapeng has joined #openstack-cinder12:16
*** gouthamr has joined #openstack-cinder12:16
*** ducttape_ has joined #openstack-cinder12:16
*** sudipto_ has quit IRC12:16
*** sudipto has quit IRC12:16
*** sudipto has joined #openstack-cinder12:17
*** tpsilva has joined #openstack-cinder12:18
openstackgerritJohnnyChou proposed openstack/cinder: Add support for Infortrend GS Series products  https://review.openstack.org/37527412:20
*** ducttape_ has quit IRC12:20
*** jungleboyj has joined #openstack-cinder12:24
openstackgerritGorka Eguileor proposed openstack/cinder: Add cleanable base object and cleanup request VO  https://review.openstack.org/30302012:27
openstackgerritGorka Eguileor proposed openstack/cinder: Make c-vol use workers table for cleanup  https://review.openstack.org/30302112:27
openstackgerritGorka Eguileor proposed openstack/cinder: [WIP] Add remaining operations to cluster  https://review.openstack.org/35596812:27
openstackgerritGorka Eguileor proposed openstack/cinder: [WIP] Support A/A on Scheduler operations  https://review.openstack.org/34422612:27
openstackgerritGorka Eguileor proposed openstack/cinder: [PoC][Don't review] Testing delay  https://review.openstack.org/35306912:27
openstackgerritGorka Eguileor proposed openstack/cinder: Support A/A in attach/detach operations  https://review.openstack.org/31857312:27
openstackgerritGorka Eguileor proposed openstack/cinder: [WIP] Allow triggering cleanup from API  https://review.openstack.org/36301012:27
openstackgerritGorka Eguileor proposed openstack/cinder: Support A/A in delete actions and get_capabilities  https://review.openstack.org/34422512:27
openstackgerritGorka Eguileor proposed openstack/cinder: [WIP] Cosmetic changes to scheduler  https://review.openstack.org/34604112:27
*** sdake has joined #openstack-cinder12:28
*** aohuanxuan has quit IRC12:29
*** markvoelker has joined #openstack-cinder12:29
*** edmondsw has joined #openstack-cinder12:30
*** portdirect has joined #openstack-cinder12:33
*** jgregor has quit IRC12:34
openstackgerritJohnnyChou proposed openstack/cinder: Add support for Infortrend GS Series products  https://review.openstack.org/37766512:34
openstackgerritIvan Kolodyazhny proposed openstack/python-cinderclient: set_metadata throws AttributeError: id  https://review.openstack.org/30847512:35
*** salv-orlando has quit IRC12:45
*** jungleboyj has quit IRC12:50
*** kumarb has quit IRC12:50
*** porrua has joined #openstack-cinder12:56
*** david-lyle has joined #openstack-cinder12:56
*** GB21 has quit IRC12:57
*** ducttape_ has joined #openstack-cinder12:58
*** rlrossit has joined #openstack-cinder13:03
openstackgerritDavid Sariel proposed openstack/cinder: Updated from global requirements  https://review.openstack.org/37768213:06
*** vnogin has joined #openstack-cinder13:08
*** jay_ has quit IRC13:09
*** laughterwym has joined #openstack-cinder13:09
*** rmcall has quit IRC13:10
*** pgadiya has quit IRC13:11
*** salv-orlando has joined #openstack-cinder13:12
*** alonma has quit IRC13:14
*** dustins has joined #openstack-cinder13:15
*** mriedem has joined #openstack-cinder13:19
*** alonma has joined #openstack-cinder13:20
*** tongli has joined #openstack-cinder13:23
*** merooney has joined #openstack-cinder13:29
openstackgerritEric Harney proposed openstack/cinder: Remove GlusterFS volume driver  https://review.openstack.org/37702813:33
*** akapil has joined #openstack-cinder13:34
*** rooneym has joined #openstack-cinder13:40
*** pdeore has joined #openstack-cinder13:40
*** scottda has joined #openstack-cinder13:40
*** shyama has joined #openstack-cinder13:40
*** salv-orl_ has joined #openstack-cinder13:42
*** bkopilov has joined #openstack-cinder13:43
*** rlrossit has quit IRC13:45
*** mugsie__ is now known as mugsie13:45
*** salv-orlando has quit IRC13:46
*** laughterwym has quit IRC13:47
*** eharney has joined #openstack-cinder13:48
*** shausy has quit IRC13:50
*** rlrossit_ has joined #openstack-cinder13:51
*** lkuchlan has left #openstack-cinder13:51
*** laughterwym has joined #openstack-cinder13:52
*** karthikp_ has joined #openstack-cinder13:53
*** jordanP has quit IRC13:54
*** arecknag has quit IRC13:56
*** laughterwym has quit IRC13:56
*** rajinir has joined #openstack-cinder13:56
*** jgregor has joined #openstack-cinder13:59
*** eharney has quit IRC14:00
openstackgerritJiong Liu proposed openstack/cinder: Remove default=None when set value in Config  https://review.openstack.org/37771514:02
*** mtanino has joined #openstack-cinder14:02
*** laughterwym has joined #openstack-cinder14:03
*** Julien-zte has joined #openstack-cinder14:11
scottdajgriffith: We wanted to move old and new api modules to core, if they belong there. So new core features should be in v3. New features that aren't implemented by everyone (so, traditional extensions) should go in contrib.14:13
*** alonma has quit IRC14:14
ildikovjohnthetubaguy: thank for the update!14:14
johnthetubaguyildikov: no worries, hopefully that helps14:15
johnthetubaguyildikov: hit a question about needing an extra API call, but added the TODO in context in the spec14:15
johnthetubaguyjgriffith: curious on your take on the updates spec14:15
ildikovjohnthetubaguy: yeap, I've just checked your earlier ping14:16
ildikovjohnthetubaguy: I'll check that and if we are unsure about that part we can bring this up on the next meeting14:17
*** eharney has joined #openstack-cinder14:18
*** jordanP has joined #openstack-cinder14:18
*** liverpooler has quit IRC14:18
*** sdake has quit IRC14:24
*** sudipto_ has joined #openstack-cinder14:24
*** GB21 has joined #openstack-cinder14:26
smcginnisscottda: That was my understanding of the past conversations as well.14:30
*** diablo_rojo has joined #openstack-cinder14:32
*** jungleboyj has joined #openstack-cinder14:33
*** zhangguoqing has quit IRC14:34
*** sdake has joined #openstack-cinder14:37
*** ducttape_ has quit IRC14:40
*** xinli has joined #openstack-cinder14:40
erlonscottda: https://gist.github.com/anonymous/6430e858967c9d97d460515815d4d56c14:42
erlonscottda: migration test script result14:42
scottda@erlon Are those NFS failures for ceph and gluster to be expected14:43
erlonscottda: no, don't know why are failing, haven't investigated14:44
erlonscottda: but the tests make easy to find out problems like this14:44
*** shyama has quit IRC14:44
scottdaerlon: Cool.14:44
erlonscottda: I only added 3 backends, but there's 614:45
*** gershon_ has joined #openstack-cinder14:45
*** gershon has joined #openstack-cinder14:45
erlonscottda: I think we will have to choose some of then, or find a way to select the backends14:45
scottdaerlon: Yes, it would be nice if the script read a conf file or something, so the backends could be easily changed.14:46
erlonbecause 6x6 will take, 80 minutes to run14:46
*** GB21 has quit IRC14:47
erlonscottda: I mean select the backends based on the code14:47
erlonscottda: for example, changes in the ceph driver will run all ceph tests14:48
scottdaerlon: OK. We should also get some feedback from infra on this, based on how long it takes to run, etc.14:53
erlonscottda: tough there are changes that can affect migration that can't  be related with any specific driver14:53
erlonscottda: ok14:53
*** cknight has joined #openstack-cinder14:53
*** adisky has quit IRC14:53
jgriffithscottda: smcginnis I think we need to talk through what we're doing here again14:54
jgriffithscottda: smcginnis I filed this bug last night https://bugs.launchpad.net/cinder/+bug/162792114:55
openstackLaunchpad bug 1627921 in Cinder "clean up API versions" [Undecided,New]14:55
*** cknight has quit IRC14:56
jgriffithscottda: the duplicated frameworks and managers inherriting from each other in multiple places is kind of wasteful and confusing14:56
jgriffithscottda: and I'm not sure I agree that anything new has to go under V3, because you create a problem with consumers and how they pick things up14:56
jgriffithscottda: in particular we've set things up as an all or nothing (you're V3, or you're V2)14:57
smcginnisjgriffith: Can't remember entirely now, but my hope was v3 could just inherit from v2 since they are initially the same.14:57
smcginnisjgriffith: Then override it as uniqueness is introduced.14:57
scottdajgriffith: Sorry, in a meeting ATM..can discuss later..14:57
jgriffithscottda: and the redirect back to V2 trick is neat, but I'm not sure that is a great end user experience14:57
jgriffithsmcginnis: yeah, that's in fact what's implemented and makes sense.... until one tries to actually do something with the code or add a new module :(14:58
smcginnisjgriffith: Any change should just go in v3 for the most part.14:59
jgriffithsmcginnis: I guess my point was I just *really* think that Nova's direction on this is vastly superior (and I don't say that about many things)14:59
smcginnisjgriffith: Guess I need to take a closer look at how they've done things.15:00
jgriffithsmcginnis: sure, but then I'm maintaining a V1, V2 and V3 set of API modules.15:00
jgriffithsmcginnis: and we still have all the contrib stuff hanging out there :(15:00
smcginnisYeah, hoping most of that can go away and just be absorbed into the normal API.15:00
jgriffithsmcginnis: maybe I'm crazy; but it seems like a more straight forward approach15:00
jgriffithsmcginnis: less code duplication, and rather easy to find things and understand what's going on15:01
smcginnisLooks good on paper.15:01
jgriffithsmcginnis: I am not a fan of being *clever* for the sake of being clever :)15:01
jgriffithsmcginnis: hehe... or an LCD screen :)15:01
jgriffithsmcginnis: I don't want to cause trouble on this, but I think we might be creating a mess for ourselves15:02
e0nejgriffith, smcginnis: just FYI, we've got the same problem with cinderclietn15:02
smcginnisjgriffith: Well, part of the intent was to reduce code duplication and hopefully make it simpler. But if that's not turning out to be the case then I'm all for trying another approach.15:03
jgriffithe0ne: yeah, that's where I started on this last night :(15:03
jgriffithe0ne: My dog had to cover his ears because I was using very offensive language :)15:03
e0nejgriffith, smcginnis: I'm going to refactor cinderclient side once, I'll finish cinderclient functional tests15:03
jgriffithsmcginnis: I think it's increased both of those things15:04
smcginnise0ne: Awesome!15:04
jgriffithsmcginnis: we now have a xxxManager in each directory V1, V2 and V315:04
e0nejgriffith, smcginnis: I could help with cinder side too if needed15:04
e0nejgriffith, smcginnis: for note: almost the same managers:(15:04
jgriffithsmcginnis: back when Nova was moving away from all of those I didn't understand the direction, but now looking at the end result it makes a lot of sense to me15:05
*** mriedem is now known as mriedem_meeting15:05
smcginnisAny work on simplifying the code would be very welcome. :)15:05
jgriffithsmcginnis: I started messing with it last night briefly and I don't think it's that difficult to do15:05
jgriffithsmcginnis: there is a LOT of detail/grunt work though which kinda sucks15:06
e0neI not a fun of OOP stuff, but in this case it's really helpful15:06
jgriffithe0ne: OOP?15:06
*** pcaruana has quit IRC15:06
e0neobject oriented programming15:06
jgriffithe0ne: oh.. haah15:06
jgriffithe0ne: well I think that's what's wrong with our approach, too much OOP hackery :)15:06
*** rcernin has quit IRC15:07
e0nejgriffith: :)15:07
*** yangyape_ has joined #openstack-cinder15:07
jgriffithanyway... I need to get my patches updated so that everything works and post them15:07
*** yangyapeng has quit IRC15:07
jgriffithjust wanted to raise this quickly15:07
jgriffithand I guess I'll have to figure out how to add these things to V3 and get Nova to talk to V3 :(15:08
*** pdeore has quit IRC15:11
*** alonma has joined #openstack-cinder15:15
sudipto_e0ne, hi, can i ask you questions w.r.t os-brick?15:16
e0nesudipto_: hi, sure15:16
*** jgregor has quit IRC15:16
*** jgregor has joined #openstack-cinder15:17
*** haplo37__ has joined #openstack-cinder15:17
sudipto_e0ne, per this code, https://github.com/openstack/os-brick/blob/5dfa56d8f40799b95c9e6d18c8fc42c263b3f6c4/os_brick/initiator/linuxscsi.py#L241 - it states that the multi path device file names can have various patterns, but seems to only address one such pattern. I thought of pushing a patch to include the other patterns, but also wanted to understand - why we have not done that already?15:17
*** openstackgerrit has quit IRC15:18
sudipto_On my RHEL host, i see that if the user friendly names are turned off - the device names are scsi-**** - however that doesn't get detected because we don't check for it in the above code.15:18
*** openstackgerrit has joined #openstack-cinder15:19
*** jgregor1 has joined #openstack-cinder15:19
*** alonma has quit IRC15:19
*** crose has joined #openstack-cinder15:20
*** jgregor has quit IRC15:21
sudipto_I wrote a bug for this here: https://bugs.launchpad.net/os-brick/+bug/1628163 and plan to propose a patch to fix it. But wanted to understand from you on whether it's a valid one to do.15:23
openstackLaunchpad bug 1628163 in os-brick "OS brick does not have a check for all types of multipath device paths" [Undecided,New] - Assigned to Sudipta Biswas (sbiswas7)15:23
*** arch-nemesis has joined #openstack-cinder15:24
scottdajgriffith: Sorry, I'm back now. Looks like you all have solved the worlds problems already...15:25
*** rlrossit_ has quit IRC15:26
openstackgerritJohnnyChou proposed openstack/cinder: Add support for Infortrend GS Series products  https://review.openstack.org/37766515:29
mtaninoe0ne: Hi15:31
mtaninoe0ne: I've tried your local-attach feature recently, and it works fine! it's nice15:32
*** diablo_rojo has quit IRC15:33
*** jgregor1 has quit IRC15:34
*** jgregor has joined #openstack-cinder15:36
e0nesudipto_: IMO, it's because multipath feature is not widely used and tested :(15:38
*** hemna_ has joined #openstack-cinder15:38
e0nehemna: ^^15:39
*** hemna_ has quit IRC15:39
e0nemtanino: hi! great news, thanks!15:39
mtaninoe0ne: I have a question about that. Do you have a plan to support FC storage for local-attach?15:39
*** psachin has quit IRC15:39
sudipto_e0ne, ah ok - just curious a most frequently used feature with LVM would be? Or is it that LVM itself is not so widely used? (Consider me a storage noob)15:39
e0nemtanino: unfortunately, I didn't have FC storage to test it15:40
hemnae0ne, what's up15:40
mtaninoe0ne: Ah I see, that's why the feature doesn't support FC storage..15:40
e0nehema: sudipto_ is asking about multipath stuff15:40
*** ducttape_ has joined #openstack-cinder15:40
e0nemtanino: it should be easy to enable it for FC if you have such env15:41
e0nehemna: https://bugs.launchpad.net/os-brick/+bug/162816315:41
openstackLaunchpad bug 1628163 in os-brick "OS brick does not have a check for all types of multipath device paths" [Undecided,New] - Assigned to Sudipta Biswas (sbiswas7)15:41
hemnasudipto_, the pattern was chosen because regardless of how multipathd is configured (friendly names on or off) that pattern is always there if it's running.15:41
hemnathe others don't show up all the time.15:42
hemnaand so searching for them isn't helpful15:42
sudipto_hemna, ok - on my host they appear as scsi-***15:42
hemnabecause /dev/disk/by-id/dm-uuid-mpath-%(wwn)s  is always there.15:42
mtaninoe0ne: I see. I don't have FC storage too. But you mean FC storage can be supported by technically.15:42
hemnaso it's not a bug that we don't look for the other paths15:43
hemnait's intentional15:43
sudipto_hemna, ok - but on my machine i couldn't see the dm-uuid pattern. That's one of the reasons it went on to another fix: https://review.openstack.org/#/c/324248/15:43
jgriffithscottda: LOL.. hardly!15:43
e0nemtanino: yes15:43
mtaninoe0ne: gocha. Thank you.15:43
e0nemtanino: you're welcome15:43
sudipto_and i was running mitaka - so that failed as well, because the fix above was not present.15:43
hemnasudipto_, your problem is not a bug in brick15:44
sudipto_Is there something i can do to configure the multi path ids to be of the pattern you mentioned?15:44
e0nemtanino: I'll try to get CI working with this feature during O release, maybe, we can test it with FC too15:44
jgriffithscottda: TLDR,  we should probably take another look at how we've done API versions, and consider aligning more with how Nova is handling things15:44
hemnayour problem is a misconfigured multipath daemon and a non functioning udev rules15:44
scottdajgriffith: OK. how is Nova doing it?15:44
mtaninoe0ne: It's nice.15:44
hemnayour bug even states that15:44
jgriffithscottda: I find our code a bit confusing and lots of duplicate15:44
hemnamultipath -l fails15:44
hemnathat's because udev failed15:44
jgriffithscottda: single modules for each API manager15:44
sudipto_hemna, well that does not fail once i apply the fix that i stated in the review...15:44
jgriffithscottda: handle the versioning inside of said module15:45
sudipto_https://review.openstack.org/#/c/324248/ - this one.15:45
scottdajgriffith: OK, I'll have a look at Nova.15:45
jgriffithscottda: get rid of this /v1, /v2, /v315:45
johnthetubaguyso the version number is global in nova15:45
johnthetubaguyjust hidden in the header15:45
sudipto_I am OK to mark the bug as invalid, if you could just help me understand what i might be doing wrong in the multi path config?15:45
jgriffithscottda: and create a MUCH simpler routing mechanism15:45
jgriffithjohnthetubaguy: +115:45
scottdajgriffith: Well, not to shirk responsibility, but the v1 to v2 switched pre-dates my involvement15:45
jgriffithjohnthetubaguy: scottda which IMO is a much more efficient way to go about it15:45
johnthetubaguyI think you will have to leave v1 and v2 there now though, just add it into v215:45
jgriffithscottda: not saying your responsible :)15:46
jgriffithscottda: BUT15:46
hemnawhy are we doing anything with v2 ?15:46
*** ducttape_ has quit IRC15:46
scottdaI'll claim responsibility for the v3 stuff :)15:46
e0nehemna: we fix bugs even for v115:46
jgriffithscottda: what I am saying is "cp -R v1/*, V2/*, V3/*, contrib/* "api/block/"15:46
hemnav1 has been deprecated for some time15:46
hemnawe shouldn't be touching it IMO15:46
sudipto_hemna, and when i put in the fix to look for pattern scsi-* the disk gets attached properly15:46
jgriffithscottda: then have a single router into those managers, and handle version differences in the module15:46
*** sandanar_ has quit IRC15:46
jgriffithscottda: V2 was not your problem, but it was our first mistake15:47
jgriffithscottda: at the time though I don't think we really knew any better, and we certainly didn't know we'd end up on the micro-versioning strategy etc15:47
scottdaYeah, my first pass on v3 was to do as we did in v1 -> v2, and copy everything. That sucked...15:47
e0nehemna: I don't agree with you. e.g. we've got 500 error from v1/v2/v3 APIs if some param is in a wrong format. we should fix such bugs for all versions15:47
scottdaSo then I went with inheriting everything in v3 from v215:48
jgriffithscottda: yeah... and what we have now is *slightly* better as it inherits from V2, but I think we should dump that whole business altogether15:48
scottdaAnd that is why we end up with duplicate code in v3 for any function that is changed.15:48
jgriffithscottda: right, and what I'm saying is we can avoid that15:48
scottdajgriffith: I'm all for doing things better. I'll get up to speed on the Nova way. I'm happy to help or work with you or whatever.15:48
jgriffithscottda: Nova has done a lot of work on this, and improved it via trial by fire15:48
jgriffithscottda: we should leverage the work done by those that went  before us :)15:49
hemnasudipto_, I want to see your logs and find out why line 491 returned None15:49
johnthetubaguydid you see the nice blog post on some of this stuff?15:49
johnthetubaguyits really good context15:49
jgriffithscottda: cool15:49
*** diablo_rojo has joined #openstack-cinder15:49
jgriffithscottda: I started playing with it last night, but I MUST finish the attach stuff first, then the local disk stuff15:49
johnthetubaguyscottda: jgriffith: I was thinking this one: https://dague.net/2015/06/05/the-nova-api-in-kilo-and-beyond-2/15:49
jgriffithjohnthetubaguy: you mean sdague 's or someobody elses?15:49
hemnasudipto_, besides, the patch you mentioned landed 9 weeks ago, so I'm not sure what the problem is15:50
johnthetubaguyjgriffith: yeah, sean'15:50
jgriffithjohnthetubaguy: yeah :)  That's a pretty nice write up15:50
scottdajohnthetubaguy: yeah, that's good stuff. Especially for describing microversions15:50
sudipto_hemna, no there isn't a problem with that patch - it's just that, i wondered why i ended up having to use that patch at all15:50
jgriffithscottda: anyway, I'd love to work on this with you if you're interested15:50
sudipto_i will get you the logs in a bit.15:50
jgriffithbut it's going to be a little bit before I have the bandwidth15:50
scottdajgriffith: yes, I am.15:50
jgriffithwe'll chat in Barcelona15:51
hemnasudipto_, ?  that patch is part of brick now.15:51
scottdajgriffith: I hear you on bandwidth. We can talk in Barcelona indeed.15:51
hemnasudipto_, I don't get what you are saying.  sorry.15:51
johnthetubaguyjgriffith: would love your take on this question when you get chance: https://review.openstack.org/#/c/373203/5/specs/ocata/approved/cinder-new-attach-apis.rst@13115:52
sudipto_hemna, ok - basically i all i am saying is - if we had looked for the right pattern (scsi-*) in the os-brick code here: https://github.com/openstack/os-brick/blob/5dfa56d8f40799b95c9e6d18c8fc42c263b3f6c4/os_brick/initiator/linuxscsi.py#L250 - we probably would always find the WWN?15:52
jgriffithjohnthetubaguy: looking now15:52
johnthetubaguyjgriffith: awesome, thanks15:52
jgriffithjohnthetubaguy: ahhh15:52
jgriffithjohnthetubaguy: well, so I took a different approach on this15:52
jgriffithjohnthetubaguy: part of what I wanted to solve was the raciness of the req/ack phase of this15:53
jgriffithjohnthetubaguy: so what I ended up doing was when you do "create_attachment" I mark the volume as "in-use"15:53
*** diablo_rojo has quit IRC15:53
jgriffithjohnthetubaguy: from Cinder's perspective I consider it attached/connected/in-use15:53
jgriffithat that point and time15:53
hemnasudipto_, the real problem is why didn't your box have the /dev/disk/by-id/dm-uuid-mpath-%(wwn)s or /dev/mapper/%(wwn)s paths15:54
jgriffithjohnthetubaguy: and I don't re-use attachments15:54
jgriffithjohnthetubaguy: so if you do multi-attach I create a new attachment record15:54
johnthetubaguyjgriffith: right, thats what I have assumed too15:54
sudipto_hemna, agreed...can you tell me how i could generate such device names? (considering i just installed mutlipath and didn't do much - just started the daemon)15:54
jgriffithjohnthetubaguy: that way we don't care or try and "guess" things when you remove_attachment(attachment-id)15:54
johnthetubaguyjgriffith: totally agreed, but there is a bigger issue here I think15:55
jgriffithjohnthetubaguy: ruh roh :)15:55
johnthetubaguyjgriffith: so you add the first volume15:55
hemnasudipto_, those paths are created by udev and the daemon itself based upon udev rules and how multpathd is configured15:55
*** pdeore has joined #openstack-cinder15:55
johnthetubaguyjgriffith: the first person to create_attachment gets to tell os-brick to do the volume host connect bit, but it fails for some reasson15:56
hemnasudipto_, https://github.com/openstack/os-brick/blob/5dfa56d8f40799b95c9e6d18c8fc42c263b3f6c4/os_brick/initiator/linuxscsi.py#L221-L23415:56
*** pbandark has quit IRC15:56
hemnathat is the standard pattern that I noticed when researching this ages ago15:56
johnthetubaguyjgriffith: in the mean time, a second volume is attached on another VM, expecting to re-use that connection underneath15:56
johnthetubaguyjgriffith: and well that fails15:56
*** laughterwym has quit IRC15:56
hemnasudipto_, what distro and version are you using ?15:56
sudipto_hemna, i am using RHEL 7.215:56
hemnasudipto_, and what version of udev and multipathd are you using?15:56
jgriffithjohnthetubaguy: hmm... I think I see what you're saying15:56
*** laughterwym has joined #openstack-cinder15:57
johnthetubaguyjgriffith: not totally sure what I want to do, but I am thinking we call back to say its good, and only then folks can "reuse" that connection15:57
sudipto_hemna, evice-mapper-multipath-0.4.9-8515:57
jgriffithjohnthetubaguy: I was going the opposite direction; I was working on telling Cinder the connection on the Nova side failed and go back to cinder and remove the attachment15:57
johnthetubaguyjgriffith: I guess they need to go in a holding loop in the mean time15:57
johnthetubaguyjgriffith: right, thats fine, but its too late, i.e. racey15:57
jgriffithjohnthetubaguy: I'm trying to parse out the complications with the two connections to same host15:58
sudipto_hemna, udevadm --version15:58
scottdajgriffith: johnthetubaguy But we should still be in attaching state, right?15:58
*** GB21 has joined #openstack-cinder15:58
sudipto_i am not sure if it's the right way to get you the udev version though15:58
hemnaok that's close to ubuntu 14.04's versions15:58
jgriffithscottda: nope15:58
scottdaso the second attempt to attach should not proceed?15:58
johnthetubaguyscottda: I am propsing we should be15:58
jgriffithscottda: when you call "create_attachment" if that succeeds it's marked as attached15:58
hemnamight be helpful to list out the paths that you did get15:58
hemnasomething isn't right15:58
jgriffithscottda: johnthetubaguy well, if we do that then there's no point in the changes I made :(15:58
jgriffithscottda: johnthetubaguy I mean the concept of create_attachment actually creating an attachment record is sort of not in the right place15:59
sudipto_hemna, something like this: lrwxrwxrwx 1 root root  9 Sep 27 15:20 scsi-3600507680280865df8000000000000cd -> ../../sdj15:59
johnthetubaguythere huge purpose in most of this still, better handing of multiple attachments, better handing of the info sent to os-brick, etc, etc15:59
scottdajgriffith: yes, we were talking yesterday about creating it during reserve()15:59
jgriffithjohnthetubaguy: scottda then we just go back to a seperate initialize and attach... which is *ok* I guess but that has a host of races15:59
jgriffithscottda: you can't :(15:59
sudipto_i am trying this for the first time today, if you let me know what exact info you need, i can get you that...you may have to let me know of how it get it though... :-|16:00
jgriffithscottda: if you look at the code the attachment record has things like the connector etc16:00
johnthetubaguyjgriffith: I don't think we do have those races, Cinder can block the bad situations, if you handle state on the attachment16:00
hemnasudipto_, and you don't have a /dev/disk/by-id/dm-uuid-mpath-<WWN> ?16:00
hemnathat should always be there16:00
johnthetubaguyjgriffith: it should mostly be covered in my spec, ish16:00
jgriffithjohnthetubaguy: that may be true16:00
* scottda needs pictures16:01
jgriffithjohnthetubaguy: hmmm...16:01
sudipto_hemna, i have something like dm-uuid-LVM - for some16:01
*** laughterwym has quit IRC16:01
sudipto_but not the pattern you state for sure16:01
johnthetubaguyscottda: does this help: https://review.openstack.org/#/c/373203/5/specs/ocata/approved/cinder-new-attach-apis.rst@16316:01
*** pdeore has quit IRC16:01
sudipto_even though this doesn't seem to resemble the WWN16:02
jgriffithjohnthetubaguy: so are you just proposing a notification back to Cinder when Nova is done?16:02
scottdajohnthetubaguy: That's good, but still contains the problem you described re: failure to connect the volume with os-brick, right?16:03
jgriffithjohnthetubaguy: so rather than create_attachment setting the state to "in-use" it leaves it as 'attaching', then makes the final transition when Nova comes back and says yay | nay ?16:03
johnthetubaguyjgriffith: yeah, I am typing that up16:03
jgriffithjohnthetubaguy: that's easy enough16:03
scottdaBut as jgriffith said, that's what we already have, isn't it?16:03
jgriffithscottda: it's different16:03
jgriffithscottda: because all the logic and work is still taking place in create_attachment16:04
*** karthikp_ has quit IRC16:04
scottdaBut the callback to change state sounds like the equivalent of the current attach() call16:04
johnthetubaguyits subtle, but different16:04
jgriffithscottda: johnthetubaguy IIUC the only thing we're doing is adding an ack back to Cinder saying things are complete on the Nova side16:04
johnthetubaguyscottda: oh, maybe it is, yeah16:04
jgriffithscottda: johnthetubaguy success or fail16:04
johnthetubaguyjgriffith: I think DELETE can be the fail now16:05
*** haplo37_ has quit IRC16:05
scottdaHow is that different from calling attach to set cinder DB from attaching -> in-use ?16:05
johnthetubaguyscottda: its not really16:05
jgriffithscottda: but attach currently does a LOT more than just set the DB state16:05
jgriffithjohnthetubaguy: I'm mixed on how I fee about this16:05
jgriffithjohnthetubaguy: I *kinda* see the case you're trying to cover, but I'm not sure I see it as a problem16:06
hemnasudipto_, do you have a custom /etc/multipath.conf ?16:06
jgriffithjohnthetubaguy: I also wanted to keep Cinder as just a provider and nothing more16:06
*** pdeore has joined #openstack-cinder16:06
jgriffithjohnthetubaguy: the more we try to be clever and introduce logic the more problems we have16:06
sudipto_hemna, nope16:06
johnthetubaguyjgriffith: so the problems exist, I guess the question is how to fix them in a robust way16:07
jgriffithjohnthetubaguy: in other words... Cinder just provides block resources; what the consumer does with them is up to them16:07
*** haplo37_ has joined #openstack-cinder16:07
scottdaWell, this diagram may be wrong, but it looks like Attach does does the DB change: https://drive.google.com/file/d/0B1Kp6K43HLHyRkFMZHVHWDJ3amM/view?pref=2&pli=116:07
jgriffithjohnthetubaguy: sorry I'm slow; can you walk me through the problem you're solving again?16:07
jgriffithscottda: it does, but it does MORE than that too16:07
johnthetubaguythe problem is, some of your drivers violate those rules16:07
jgriffithjohnthetubaguy: +116:07
jgriffithjohnthetubaguy: we're f'ing borked on the Cinder side because of all the snow-flake shit pushed up in to the manager16:08
johnthetubaguyjgriffith: so two volumes attaching on the same host, if done serially they will share one connection to the volume backend16:08
scottdajgriffith: I love the idea of Cinder being just a provider. Too bad we cannot just export a volume, give the connector to Nova/Ironic/whoever, and let them do as they please.16:08
jgriffithjohnthetubaguy: oh... well, I had actually thought we wouldn't do that :)16:08
johnthetubaguyproblem is some providers require that16:09
johnthetubaguyvolume providers, that is16:09
jgriffithjohnthetubaguy: yeah, I forgot about that little detail :)16:09
*** diablo_rojo has joined #openstack-cinder16:09
jgriffithjohnthetubaguy: anyway... continue please ;)16:09
johnthetubaguystepping back, I think in reality there are two attachments, one to the host, and one to the instance16:10
johnthetubaguybut I don't think we want that in the API, its way too confusing16:10
johnthetubaguyso the bug...16:10
jgriffithjohnthetubaguy: so let's keep going on your example,  I think that problem is solved but I could be wrong16:10
jgriffithjohnthetubaguy: so currently what I put in place and planned for:16:10
jgriffith1.  attachment-1 gets created16:11
johnthetubaguyyou have volume A attaching, and create attach is called, but os-brick isn't setup16:11
johnthetubaguynow volume B is attaching, on the same host16:11
johnthetubaguyit wants to reuse the conneciton to the back end, its already in place, so it can, and so it proceeds16:11
*** salv-orl_ has quit IRC16:11
*** pdeore has quit IRC16:11
johnthetubaguynow we get back to volume A, os-brick fails, and so things go boom16:11
johnthetubaguyvolume A just DELETEs its attachment16:12
hemnasudipto_, maybe dump out your /dev/disk/by-id dir ?16:12
johnthetubaguybut now volume B will never work, and spots that much later16:12
*** salv-orlando has joined #openstack-cinder16:12
sudipto_hemna, alright in a sec16:12
johnthetubaguyor volume B tries to use the thing before its actually created bt the volume A thread16:12
jgriffithjohnthetubaguy:  got ya16:12
hemnasudipto_, pastebin please :)16:13
sudipto_hemna, ok - was thinking paste.openstack - but will do pastern16:13
jgriffithjohnthetubaguy: I'm thinking.....16:13
*** raunak has joined #openstack-cinder16:14
scottdajohnthetubaguy: your idea about 2 attaches is right, so maybe we need to states. something like "connected" (to compute host) and "attached" (to instance)16:14
jgriffithjohnthetubaguy: so let's call the nova side of this and the things it does with brick the "connection" to avoid ambiguity16:14
scottdaThen, the "connected" state could be used to help with the drivers that multi-plex the connection.16:14
sudipto_hemna, http://pastebin.com/HuZZxXEe16:14
jgriffithjohnthetubaguy: so the concern is that the connection process itself can be a race, and if the first one in fails we're doomed16:15
johnthetubaguyjgriffith: yeah16:15
sudipto_hemna, i do see the wwn-% pattern - however they don't seem to match t16:16
jgriffithjohnthetubaguy: To be clear though, is this a problem for Volume-A and Volume-B OR is it actually a multi-attach problem?16:16
johnthetubaguyscottda: I was thinking we just allow one attachment to be "attaching" or "detaching" at any one time, per host, if required, but thats kinda confusing16:16
jgriffithjohnthetubaguy: I'm not familiar with drivers that share a single target for multiple volumes16:16
hemnasudipto_, the entire directory listing if possible16:16
johnthetubaguyjgriffith: I think its a problem without multi-attach16:16
scottdajgriffith: I think it's a problem for drivers that share connection to the same host, regardless of multi-attach16:16
sudipto_hemna, that's the entire thing ...16:16
johnthetubaguyjgriffith: I thought the fiber channel stuff maybe?16:16
jgriffithjohnthetubaguy: just revert that stuff and we'll all be beter off :)16:17
*** mvk has quit IRC16:17
*** cknight has joined #openstack-cinder16:17
diablo_rojodulek if you want to get this in I took my - Workflow off https://review.openstack.org/#/c/290885/1716:17
johnthetubaguyjgriffith: yeah... its like the live-migrate of the drivers16:17
jgriffithjohnthetubaguy: so the only problem I have with this is that I think Nova or Brick should be owning the details16:18
hemnasudipto_, how about /dev/mapper as well as multipath -ll16:18
johnthetubaguyjgriffith: but Nova doesn't know the driver, and brick has no state16:18
hemnathe /dev/disk/by-id listing doesn't look like multipath devices to me16:18
jgriffithjohnthetubaguy: yeah :(16:18
johnthetubaguyjgriffith: I think basically this is moving the state os-brick should have into the server16:19
*** akerr has joined #openstack-cinder16:19
*** sudipto_ has quit IRC16:19
*** sudipto has quit IRC16:19
jgriffithjohnthetubaguy: so the problem I'm having is in the case of a shared connection like you describe we don't currently have any mechanism to deal with it today either16:20
*** sudipto has joined #openstack-cinder16:20
*** sudipto_ has joined #openstack-cinder16:20
jgriffithjohnthetubaguy: Unless there's driver magic taking place during initialize-connection or attach-volume16:20
sudiptohemna, the multi path -l does not give any o/p16:21
jgriffithjohnthetubaguy: which very well could be16:21
*** nkrinner is now known as nkrinner_afk16:21
jgriffithjohnthetubaguy: this raises a bigger problem with trying to make this change :(16:21
hemnasudipto, multipath -ll doesn't return anything ?16:21
johnthetubaguyjgriffith: so deep down I keep thinking everytime we don't tell Cinder the truth of whats happening, we hit a snag16:21
sudiptohemna, yeah16:21
jgriffithjohnthetubaguy: one of the reasons I ended up where I did was because I can't change 80+ cinder drivers16:21
hemnathen you don't have multipath devices16:22
johnthetubaguyjgriffith: it feels good saying when we are done16:22
scottdajohnthetubaguy: That's the gist of many problems, we hide the truth with trickery16:22
jgriffithjohnthetubaguy: so I wrapped up existing calls16:22
johnthetubaguyscottda: +116:22
*** akapil has quit IRC16:22
hemnasudipto, something else is very wrong.  multipathd and udev aren't playing nice.16:22
*** cknight1 has joined #openstack-cinder16:22
johnthetubaguyjgriffith: so thats still OK here, I think... because I think this is extra state that drivers could eventually use if needed16:22
sudiptohemna, oh k :(16:23
sudiptohemna, basically if multi path -ll doesn't yield any o/p - the fix would be this right: https://review.openstack.org/#/c/324248/ ?16:23
*** portdirect has quit IRC16:24
hemnarunning "multipath -ll" on a box should yield the entire list of multipath devices known.16:24
hemnaif it doesn't, multipath is in a bad state16:24
*** cknight has quit IRC16:24
jgriffithjohnthetubaguy: what if I didn't condense the initialize and attach, but created the attachment_id  on the actual attach and stored all the stateful stuff like I'm doing16:24
jgriffithjohnthetubaguy: so I wouldn't change much on the Nova side, except adding the attachment_id to the bdm object16:25
hemnasudipto, and that patch you listed, which is already in os-brick, won't make any difference16:25
jgriffithjohnthetubaguy: so then whatever crazy mechanisms the "special" drivers are using to combat this problem today still works?16:25
sudiptohemna, ok16:26
jgriffithjohnthetubaguy: I'm still unclear that this isn't a seperate/existing problem already16:26
sudiptowhen i attach the device though with the scsi-* pattern it does appear to be fine inside the guest...16:26
johnthetubaguyjgriffith: possibly, but I would rather we go down the "what we would do if we start over", as we are so close, then see how it maps back to what we have already16:26
johnthetubaguyjgriffith: all the problems we are fixing here are generally existing today16:27
*** rlrossit has joined #openstack-cinder16:27
sudiptoand i wonder how i can debug the multi path thing - any clues?16:27
hemnasudipto, sure, but it's not a multipath device.16:27
jgriffithjohnthetubaguy: yeah, so I'd definitely be going the direction I am now if I was starting from scratch16:27
hemnamultipath isn't working16:27
hemnaI can't say that enough.16:27
jgriffithjohnthetubaguy: and drivers would have actually followed the reference16:27
sudiptohemna, ok got it...16:27
hemnasudipto, you can stop the daemon and run it manually in non-daemon mode16:27
sudiptohemna, ok16:28
hemnaand it will puke out everything16:28
johnthetubaguyjgriffith: so drivers had to be different due to technological limitations of what they do, as I understand it, they only support like 10-20 connections, etc16:28
jgriffithjohnthetubaguy: ok... so I'll change the create_attachment so it leaves it as "attaching" and add an "ack" call back16:28
hemnamultipathd -v3 -d16:28
jgriffithjohnthetubaguy: it sort of sucks for non-nova consumers though :(16:28
johnthetubaguyjgriffith: they have the same problems right?16:28
jgriffithjohnthetubaguy: nah, that's not the biggest problem.  The bigger problem is we ended up with an interesting split between how people interpretted the attach-flow16:29
hemnajgriffith, so the odd thing about that is, that's exactly what we have today16:29
sudiptohemna, ok - it says wwid xxx is not in wwid file.16:29
openstackgerritTom Swanson proposed openstack/cinder: WIP: Dell SC: Adding additional async task logging  https://review.openstack.org/37784416:29
hemnajgriffith, we already have that mechanism with initialize_connection and then the follow up with attach16:29
jgriffithhemna: that was kinda my point yeah :(  It's slightly different, but mostly the same16:29
johnthetubaguyhemna: it does feel like we have come full circle here16:29
scottdajgriffith: Sorry if I'm not remebering what's in your spec, but weren't you going to implement an is_safe_to_disconnect() call that the driver could answer if they are "special"?16:29
jgriffithscottda: yeah16:29
hemnacinder has exported the volume16:30
hemnashould cinder just say....it's exported16:30
scottdaNova would have to change to use that call, but that seems like a solution to this issue.16:30
hemnaand if someone wants to say...yah I see it, they can.16:30
*** jgregor has quit IRC16:30
jgriffithhemna: cinder isn't actually exporting it either though :)16:30
hemnadoes it really affect cinder's follow on actions to the volume at all?16:30
jgriffithhemna: cinder is just a control plane16:30
hemnawell, the backend has exported it16:30
*** leeantho has joined #openstack-cinder16:31
hemnabut as far as cinder is concerned you can take actions on that volume after it's exported from the backend16:31
*** ekarlso_ has quit IRC16:31
hemnacinder has done it's job of making sure the backend has exported the volume16:31
hemnanova just hasn't seen it yet, or ran discovery on it's host.16:31
jgriffithhemna: so we go back to "initialize_connection" is the export, and attach should've just been a db update :)16:31
hemnathe only thing left to do is ack the export16:32
jgriffithhemna: I still say this isn't Cinder's problem :)16:32
hemnait's not really16:32
*** rlrossit has quit IRC16:32
scottdaAnd perhaps there's need for a 'connected state for array <-> node connection16:32
*** akapil has joined #openstack-cinder16:32
hemnathe only caveat is what cinder will allow from an API perspective on that volume after the export has been done.16:32
jgriffithscottda: oh lord... I'm out :)16:32
*** jordanP has quit IRC16:32
hemnaright now we leave it in attaching state, which means there are no other actions allowed16:32
scottdaI think I agree with jgriffith  that it's not cinder's problem. WE just export , but we cannot control whether it's connected to a host, or attached to a VM16:33
hemnascottda, that's what I'm getting at16:33
hemnabut cinder currently doesn't allow any other actions on that volume unless it gets that ack16:33
*** zhongjun_ has quit IRC16:33
hemnawe could alter this16:34
johnthetubaguyso I updated my spec (https://review.openstack.org/373203) with the extra ack call16:34
hemnato say, the volume is exported and the attachment just hasn't been acked16:34
johnthetubaguyI think the only different thing in that spec now is keeping a per attachment state machine16:34
*** asselin__ has joined #openstack-cinder16:34
hemnajohnthetubaguy, the current volume_attachment table has a state16:34
*** kaisers__ has joined #openstack-cinder16:35
*** david-lyle_ has joined #openstack-cinder16:35
*** tongli_ has joined #openstack-cinder16:35
johnthetubaguyhemna: does it match what I have put in the spec? it may just be different names16:35
sudipto_hemna, one final thing - if the multi path method is able to find a scsi-* device - things just seem to work fine...16:35
hemnasudipto_, but you aren't using a multipath device.16:35
*** rooneym_ has joined #openstack-cinder16:36
hemnawhich means you don't get any benefits from multipathing16:36
sudipto_hemna, agreed.16:36
hemnathat's a failure IMHO16:36
*** zhenguo_ has joined #openstack-cinder16:36
*** xek_ has joined #openstack-cinder16:36
*** rooneym has quit IRC16:36
*** zhenguo has quit IRC16:36
*** ildikov has quit IRC16:36
*** haplo37__ has quit IRC16:36
*** kambiz has quit IRC16:36
*** unrahul has quit IRC16:36
*** patrickeast has quit IRC16:36
*** kong has quit IRC16:36
*** pots has quit IRC16:36
*** alaski has quit IRC16:36
*** fmccrthy has quit IRC16:36
*** rooneym_ is now known as rooneym16:36
*** mtanin___ has joined #openstack-cinder16:36
*** dansmith has quit IRC16:36
*** mpjetta_ has quit IRC16:37
*** cknight has joined #openstack-cinder16:37
*** aswadr_ has quit IRC16:37
*** tbarron has quit IRC16:37
*** pckizer_ has quit IRC16:37
*** arch-nemesis has quit IRC16:37
*** porrua has quit IRC16:37
*** nicolasbock has quit IRC16:37
*** snarwade has quit IRC16:37
*** cebruns_ has joined #openstack-cinder16:37
*** Kimmo__ has joined #openstack-cinder16:37
*** narayrak has quit IRC16:37
*** salv-orlando has quit IRC16:37
*** Kimmo_ has quit IRC16:37
*** sticker_ has joined #openstack-cinder16:37
*** lpetrut has quit IRC16:37
*** david-lyle has quit IRC16:37
*** kragniz has quit IRC16:37
*** akerr_ has joined #openstack-cinder16:37
*** sticker_ is now known as 17WAAQNWW16:37
hemnajgriffith, does it make any sense to change the state management of the volume ?16:37
*** mriedem_meeting has quit IRC16:37
*** serverascode has quit IRC16:37
*** pkoniszewski has quit IRC16:37
*** stevemar has quit IRC16:37
*** zengyingzhe has joined #openstack-cinder16:37
*** rhedlind has joined #openstack-cinder16:37
*** nkrinner_afk has quit IRC16:37
*** mfisch has quit IRC16:37
*** stevemar has joined #openstack-cinder16:37
*** tbarron_ has joined #openstack-cinder16:37
*** sticker_ has joined #openstack-cinder16:37
*** bluex has quit IRC16:37
*** cebruns has quit IRC16:37
*** med_ has quit IRC16:37
hemnathat after create_attachment is called, set the state of the volume to in-use16:37
*** nicolasbock has joined #openstack-cinder16:37
*** bkopilov has quit IRC16:37
*** gcb has quit IRC16:37
*** cebruns__ has quit IRC16:37
*** jdillaman has quit IRC16:37
*** pleia2 has quit IRC16:37
*** huanan has quit IRC16:37
*** alaski_ has joined #openstack-cinder16:37
*** kaisers_ has quit IRC16:37
*** krtaylor has quit IRC16:37
*** pckizer has joined #openstack-cinder16:37
*** merooney has quit IRC16:37
*** rhe00 has quit IRC16:37
*** dosaboy has quit IRC16:37
*** mikal has quit IRC16:37
*** jgriffith has quit IRC16:37
*** lennyb has quit IRC16:37
*** jungleboyj has quit IRC16:37
*** hoonetorg has quit IRC16:37
*** mnaser has quit IRC16:37
*** huanan_L has joined #openstack-cinder16:37
*** tbarron_ is now known as tbarron16:37
*** pckizer has quit IRC16:37
*** pckizer has joined #openstack-cinder16:37
hemnaif nova fails to discover, then nova calls cinder again to terminate16:37
*** dmellado has quit IRC16:37
*** porrua_ has joined #openstack-cinder16:38
*** raunak has quit IRC16:38
*** mtanino has quit IRC16:38
*** tongli has quit IRC16:38
*** mdovgal has quit IRC16:38
*** anteaya has quit IRC16:38
*** zengyingzhe__ has quit IRC16:38
*** sticker has quit IRC16:38
*** asselin_ has quit IRC16:38
*** nkrinner_afk has joined #openstack-cinder16:38
*** GB21 has quit IRC16:38
*** Julien-zte has quit IRC16:38
*** thingee has quit IRC16:38
hemnacinder doesn't really care if nova hasn't seen/discovered yet.16:38
*** alaski_ is now known as alaski16:38
*** crose has quit IRC16:38
*** ccesario has quit IRC16:38
*** sileht has quit IRC16:38
hemnaI guess it might introduce a race though16:38
*** eharney has quit IRC16:38
*** gouthamr has quit IRC16:38
*** nherciu_ has quit IRC16:38
*** sudipto has quit IRC16:38
*** jidar has quit IRC16:38
*** krtaylor has joined #openstack-cinder16:38
*** anteaya has joined #openstack-cinder16:38
*** kambiz has joined #openstack-cinder16:38
*** kaisers1 has joined #openstack-cinder16:38
*** cknight1 has quit IRC16:38
*** zzzeek has quit IRC16:38
*** Wi1d_ has quit IRC16:38
*** akerr has quit IRC16:38
*** kaisers has quit IRC16:38
*** lpetrut1 has joined #openstack-cinder16:38
*** pkoniszewski has joined #openstack-cinder16:38
*** xek has quit IRC16:38
*** lennyb has joined #openstack-cinder16:38
*** dmellado has joined #openstack-cinder16:38
hemnabetween the time that cinder sets the state to in-use and nova calls terminate, we could get another action being taken on the volume.16:38
*** links has quit IRC16:38
*** wxy__ has quit IRC16:38
*** apopovych has quit IRC16:38
*** dgonzalez has quit IRC16:38
*** lpetrut1 is now known as lpetrut16:38
*** hoonetorg has joined #openstack-cinder16:38
*** diablo_rojo has quit IRC16:38
*** cebruns has joined #openstack-cinder16:38
*** timcl has quit IRC16:38
hemnawe'd have to modify the API to allow terminate on that volume to be called.16:38
*** xinli has quit IRC16:38
hemnagets us into an odd place16:39
*** pleia2 has joined #openstack-cinder16:39
hemna(sorry I'm just thinking outloud)16:39
*** sudipto has joined #openstack-cinder16:39
*** jgriffith has joined #openstack-cinder16:39
*** Julien-zte has joined #openstack-cinder16:39
*** nherciu_ has joined #openstack-cinder16:39
*** arch-nemesis has joined #openstack-cinder16:39
jgriffithjohnthetubaguy: sorry... got bounced16:39
jgriffithjohnthetubaguy: ok, we'll go with that for now then16:39
*** 17WAAQNWW has quit IRC16:40
*** dansmith has joined #openstack-cinder16:40
*** snarwade has joined #openstack-cinder16:40
johnthetubaguyjgriffith: so I updated the spec with my current preference, I think: https://review.openstack.org/37320316:40
*** zhenguo_ is now known as zhenguo16:40
*** kragniz1 has joined #openstack-cinder16:40
*** haplo37 has joined #openstack-cinder16:40
jgriffithjohnthetubaguy: seems like an ok comporomise, and I have an idea to keep it from being problematic for consumers other than Nova that don't need the extra step16:40
openstackgerritmichaeldovgal proposed openstack/cinder: Add if host's cinder-volume service is disabled validation  https://review.openstack.org/37788616:40
openstackgerritMerged openstack/os-brick: Docstrings should not start with a space  https://review.openstack.org/37726516:40
jgriffithjohnthetubaguy: yeah, I read the update, looks good to me16:41
*** ircuser-1 has quit IRC16:41
*** Apoorva has joined #openstack-cinder16:41
*** gouthamr has joined #openstack-cinder16:41
jgriffithjohnthetubaguy: thanks for your patience, I can be slow sometimes :)16:41
*** jidar has joined #openstack-cinder16:41
*** DuncanT has quit IRC16:41
johnthetubaguyjgriffith: mind meld over IRC isn't that efficient16:41
*** ducttape_ has joined #openstack-cinder16:42
*** mfisch has joined #openstack-cinder16:42
jgriffithjohnthetubaguy: true16:42
*** mfisch has quit IRC16:42
*** mfisch has joined #openstack-cinder16:42
johnthetubaguyjgriffith: so there was one more case, os-brick fails to detach16:42
*** mikal has joined #openstack-cinder16:42
johnthetubaguyits tempting to have an ERROR state for an attachment for that case, but it feels like overkil16:43
*** ildikov has joined #openstack-cinder16:44
*** dosaboy has joined #openstack-cinder16:44
*** wilson_liu has joined #openstack-cinder16:45
*** dgonzalez has joined #openstack-cinder16:45
*** Apoorva_ has joined #openstack-cinder16:45
*** rhe00 has joined #openstack-cinder16:45
*** cknight1 has joined #openstack-cinder16:45
scottdaWell, as johnthetubaguy pointed out, the problem is that we mask the truth. We could be more explicit around volume state: exported, connected, in-use.16:45
*** sudswas__ has joined #openstack-cinder16:46
*** zengyingzhe_ has joined #openstack-cinder16:46
*** huanan_M has joined #openstack-cinder16:46
hemnascottda, +116:46
*** Guest66676 has quit IRC16:46
*** grumpycatt has quit IRC16:46
*** grumpycatt has joined #openstack-cinder16:46
*** kaisers has joined #openstack-cinder16:46
*** cburgess_ has joined #openstack-cinder16:46
*** nherciu__ has joined #openstack-cinder16:47
*** lennyb has quit IRC16:47
*** ChubYann has quit IRC16:47
johnthetubaguyscottda: hemna: my spec tries to do most of that by adding extra state into the attachments, I am curious if I am missing any now16:47
*** rhagarty_ has joined #openstack-cinder16:47
*** rhagarty_ has quit IRC16:47
*** rhagarty_ has joined #openstack-cinder16:47
*** rhedlind has quit IRC16:47
*** cburgess has quit IRC16:47
*** pkoraca has quit IRC16:47
*** mtanino has joined #openstack-cinder16:47
*** porrua has joined #openstack-cinder16:47
*** yangyapeng has joined #openstack-cinder16:47
*** sudipto has quit IRC16:47
*** pckizer has quit IRC16:47
*** Guest96871 has quit IRC16:47
*** mtanino has quit IRC16:47
*** mtanino has joined #openstack-cinder16:47
hemnaok I'll go read the update :)16:47
*** mfisch` has joined #openstack-cinder16:47
*** tbarron has quit IRC16:47
*** zengyingzhe has quit IRC16:47
*** sdague has quit IRC16:47
*** zhengyin has quit IRC16:47
*** pocketprotector has quit IRC16:47
*** nherciu_ has quit IRC16:47
*** jgriffith has quit IRC16:47
*** tbarron has joined #openstack-cinder16:47
*** dmellado_ has joined #openstack-cinder16:47
*** jidar_ has joined #openstack-cinder16:47
*** kaisers2 has quit IRC16:47
*** ducttape_ has quit IRC16:48
*** lpetrut1 has joined #openstack-cinder16:48
*** gouthamr has quit IRC16:48
*** cebruns has quit IRC16:48
*** sticker_ has quit IRC16:48
*** briancurtin has quit IRC16:48
*** Apoorva has quit IRC16:48
*** huanan_L has quit IRC16:48
*** mtanin___ has quit IRC16:48
*** jamielennox has quit IRC16:48
*** vkmc has quit IRC16:48
*** Wi1d has joined #openstack-cinder16:48
*** zzzeek has joined #openstack-cinder16:48
*** dansmith has quit IRC16:48
*** lpetrut has quit IRC16:48
*** mfisch has quit IRC16:48
*** kragniz1 has quit IRC16:48
*** yangyape_ has quit IRC16:48
*** rhagarty__ has quit IRC16:48
*** rmk` has joined #openstack-cinder16:48
*** dansmith has joined #openstack-cinder16:48
*** gouthamr_ has joined #openstack-cinder16:48
*** vkmc- has joined #openstack-cinder16:48
*** dmellado has quit IRC16:48
*** logan- has quit IRC16:48
*** rmk has quit IRC16:48
*** anteaya has quit IRC16:48
*** jidar has quit IRC16:48
*** porrua_ has quit IRC16:48
*** pckizer has joined #openstack-cinder16:48
*** pckizer has quit IRC16:48
*** pckizer has joined #openstack-cinder16:48
*** med_ has joined #openstack-cinder16:48
*** lpetrut1 is now known as lpetrut16:48
*** mpjetta has joined #openstack-cinder16:48
*** rmk` is now known as rmk16:48
*** rmk has quit IRC16:48
*** rmk has joined #openstack-cinder16:48
*** dansmith is now known as Guest374716:48
*** cebruns has joined #openstack-cinder16:48
*** vkmc- is now known as vkmc16:48
*** vkmc has quit IRC16:48
*** vkmc has joined #openstack-cinder16:48
*** med_ is now known as Guest4406416:48
*** fmccrthy has joined #openstack-cinder16:48
*** jamielennox has joined #openstack-cinder16:48
*** bkopilov has joined #openstack-cinder16:49
*** cknight has quit IRC16:49
*** winston-d has joined #openstack-cinder16:49
*** jidar_ is now known as jidar16:49
*** jgriffith has joined #openstack-cinder16:49
*** Guest3203 is now known as melwitt16:49
*** timcl has joined #openstack-cinder16:49
*** lennyb has joined #openstack-cinder16:50
*** GB21 has joined #openstack-cinder16:50
*** cdelatte has quit IRC16:50
*** eharney has joined #openstack-cinder16:50
*** ccesario has joined #openstack-cinder16:50
*** anteaya has joined #openstack-cinder16:50
*** cdelatte has joined #openstack-cinder16:51
*** kragniz1 has joined #openstack-cinder16:51
*** gcb has joined #openstack-cinder16:51
*** crose has joined #openstack-cinder16:51
*** sileht has joined #openstack-cinder16:51
*** ChubYann has joined #openstack-cinder16:51
*** jklare has quit IRC16:51
*** diablo_rojo has joined #openstack-cinder16:51
*** links has joined #openstack-cinder16:52
*** bluex has joined #openstack-cinder16:52
*** AndyWojo has quit IRC16:52
*** zhengyin has joined #openstack-cinder16:52
*** jdillaman has joined #openstack-cinder16:52
*** Guest3747 is now known as dansmith16:52
*** pocketprotector has joined #openstack-cinder16:53
*** logan- has joined #openstack-cinder16:54
*** patrickeast has joined #openstack-cinder16:54
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements  https://review.openstack.org/37742616:54
*** e0ne has quit IRC16:56
*** wxy__ has joined #openstack-cinder16:56
*** raunak has joined #openstack-cinder16:57
*** mnaser has joined #openstack-cinder16:57
*** Guest66666 has joined #openstack-cinder16:57
*** jklare has joined #openstack-cinder16:58
*** sdague has joined #openstack-cinder16:58
*** mriedem has joined #openstack-cinder16:58
*** karthikp- has joined #openstack-cinder16:59
*** salv-orlando has joined #openstack-cinder16:59
*** rajinir has quit IRC17:00
*** akapil has quit IRC17:00
*** ducttape_ has joined #openstack-cinder17:01
*** Apoorva has joined #openstack-cinder17:01
openstackgerritOpenStack Proposal Bot proposed openstack/os-brick: Updated from global requirements  https://review.openstack.org/37377117:01
*** unrahul has joined #openstack-cinder17:02
openstackgerritOpenStack Proposal Bot proposed openstack/python-cinderclient: Updated from global requirements  https://review.openstack.org/37613117:02
*** ircuser-1 has joined #openstack-cinder17:03
*** akapil has joined #openstack-cinder17:03
scottdaWell, at some point "special" drivers need to know if there's already an connection to the NOva host for a given volume, or possibly for any volume. This is the info needed for a is_safe_to_disconnect call. I think it's better to be explicit that to hack around to figure that out. But just my opinion.17:04
*** Apoorva_ has quit IRC17:04
johnthetubaguyscottda: so the initialize_detaching call is the is_safe_to_disconnect call17:06
*** ducttape_ has quit IRC17:06
*** frontrunner has joined #openstack-cinder17:06
johnthetubaguyscottda: so I guess its a hack, but we would know if volume x has been connected on host y, with that proposed attachment logic17:06
johnthetubaguyscottda: basically I am leaving it os-brick and cinder to sort out its own interface for that, at least I think I am...17:08
johnthetubaguyscottda: I should think on that more17:08
*** nherciu_ has joined #openstack-cinder17:08
*** timcl1 has joined #openstack-cinder17:09
*** Apoorva_ has joined #openstack-cinder17:10
*** jidar_ has joined #openstack-cinder17:10
*** jgriffith_ has joined #openstack-cinder17:10
*** cdelatte has quit IRC17:10
*** timcl has quit IRC17:10
*** kaisers has quit IRC17:10
*** jidar has quit IRC17:10
*** ccesario has quit IRC17:10
*** GB21 has quit IRC17:10
*** Guest44064 has quit IRC17:10
*** kaisers2 has joined #openstack-cinder17:10
*** nherciu__ has quit IRC17:11
*** jidar_ is now known as jidar17:11
*** mpjetta has quit IRC17:11
*** jgriffith has quit IRC17:11
*** unrahul has quit IRC17:11
*** logan- has quit IRC17:11
*** stevemar has quit IRC17:11
*** stevemar has joined #openstack-cinder17:11
*** ircuser-1 has quit IRC17:11
*** porrua_ has joined #openstack-cinder17:11
*** delattec has joined #openstack-cinder17:11
*** apopovych has joined #openstack-cinder17:11
*** raunak_ has joined #openstack-cinder17:11
*** mtanino__ has joined #openstack-cinder17:11
*** sdague has quit IRC17:11
*** patrickeast has quit IRC17:11
*** dmellado_ has quit IRC17:11
*** alaski has quit IRC17:11
*** apopovych has quit IRC17:11
*** apopovych has joined #openstack-cinder17:11
*** mtanino__ has quit IRC17:11
*** mtanino__ has joined #openstack-cinder17:11
*** lpetrut1 has joined #openstack-cinder17:11
*** mnaser has quit IRC17:11
*** rhagarty_ has quit IRC17:11
*** GB21 has joined #openstack-cinder17:11
*** frontrunner has quit IRC17:11
*** pocketprotector has quit IRC17:11
*** eharney has quit IRC17:11
*** bkopilov has quit IRC17:11
*** tbarron has quit IRC17:11
*** porrua has quit IRC17:11
*** mtanino has quit IRC17:11
*** kambiz has quit IRC17:11
*** dansmith has quit IRC17:11
*** links has quit IRC17:11
*** gouthamr_ has quit IRC17:11
*** Julien-zte has quit IRC17:12
*** lpetrut has quit IRC17:12
*** jamielennox has quit IRC17:12
*** Wi1d has quit IRC17:12
*** pleia2 has quit IRC17:12
*** cebruns has quit IRC17:12
*** Guest66666 has quit IRC17:12
*** wxy__ has quit IRC17:12
*** lpetrut1 is now known as lpetrut17:12
*** mpjetta has joined #openstack-cinder17:12
*** eharney has joined #openstack-cinder17:12
*** akapil has quit IRC17:12
*** jklare has quit IRC17:12
*** jklare has joined #openstack-cinder17:12
*** winston-d has quit IRC17:12
*** Wi1d_ has joined #openstack-cinder17:12
*** kambiz_ has joined #openstack-cinder17:12
*** bkopilov_ has joined #openstack-cinder17:12
*** gouthamr has joined #openstack-cinder17:12
*** jgriffith_ is now known as jgriffith17:12
*** tbarron has joined #openstack-cinder17:12
*** pocketprotector has joined #openstack-cinder17:12
*** pocketprotector has quit IRC17:12
*** pocketprotector has joined #openstack-cinder17:12
*** Apoorva has quit IRC17:12
*** raunak has quit IRC17:12
*** raunak_ is now known as raunak17:12
*** ircuser-1 has joined #openstack-cinder17:12
*** jraju has joined #openstack-cinder17:12
*** ccesario has joined #openstack-cinder17:12
*** dmellado has joined #openstack-cinder17:12
*** cebruns has joined #openstack-cinder17:12
*** logan- has joined #openstack-cinder17:12
*** kambiz_ is now known as kambiz17:12
*** dansmith has joined #openstack-cinder17:13
*** sdague has joined #openstack-cinder17:13
*** Julien-zte has joined #openstack-cinder17:13
*** frontrunner has joined #openstack-cinder17:14
*** alaski has joined #openstack-cinder17:14
*** jamielennox has joined #openstack-cinder17:14
*** kragniz1 is now known as kragniz17:15
*** pleia2 has joined #openstack-cinder17:15
scottdajohnthetubaguy: ok17:15
*** winston-d has joined #openstack-cinder17:16
*** Guest66666 has joined #openstack-cinder17:17
*** harlowja has joined #openstack-cinder17:17
*** ircuser-1 has quit IRC17:17
*** patrickeast has joined #openstack-cinder17:17
patrickeastis freenode having problems? i keep getting disconnected :(17:18
*** ircuser-1 has joined #openstack-cinder17:18
*** porrua has joined #openstack-cinder17:18
*** gouthamr_ has joined #openstack-cinder17:19
*** mtanino has joined #openstack-cinder17:19
scottdaIRC is pretty flaky today....17:20
*** cdelatte has joined #openstack-cinder17:20
*** raunak_ has joined #openstack-cinder17:20
openstackgerritKarthik Prabhu Vinod proposed openstack/cinder: Switch service capabilities to ovo  https://review.openstack.org/31904017:20
openstackgerritKarthik Prabhu Vinod proposed openstack/cinder: Refactoring total_capacity_gb & free_capavity_gb in capabilities  https://review.openstack.org/37497117:20
*** grumpycatt has quit IRC17:21
*** grumpycatt has joined #openstack-cinder17:21
*** dmellado_ has joined #openstack-cinder17:21
*** haplo37_ has quit IRC17:21
*** mikal_ has joined #openstack-cinder17:21
*** lpetrut has quit IRC17:21
*** dmellado has quit IRC17:21
*** stevemar has quit IRC17:21
*** swamireddy_ has quit IRC17:21
*** delattec has quit IRC17:21
*** stevemar has joined #openstack-cinder17:21
patrickeastscottda: yea, irccloud was down… back to using ancient means of communication17:21
*** tbarron has quit IRC17:21
*** mtanino__ has quit IRC17:21
*** sileht has quit IRC17:21
*** diegows has quit IRC17:21
*** harlowja_ has joined #openstack-cinder17:22
*** Wi1d has joined #openstack-cinder17:22
*** Guest66676 has joined #openstack-cinder17:22
*** sage_ has quit IRC17:22
*** zzzeek has quit IRC17:22
*** krtaylor has quit IRC17:22
*** harlowja has quit IRC17:22
*** pleia2 has quit IRC17:22
*** logan- has quit IRC17:22
*** raunak has quit IRC17:22
*** porrua_ has quit IRC17:22
*** zhengyin has quit IRC17:22
*** raunak_ is now known as raunak17:22
patrickeastah yea https://twitter.com/freenodestaff/status/78081327673321062917:22
*** kambiz has quit IRC17:22
*** bluex has quit IRC17:22
*** jklare has quit IRC17:22
openstackgerritKarthik Prabhu Vinod proposed openstack/cinder: Refactoring total_capacity_gb & free_capacity_gb in capabilities  https://review.openstack.org/37497117:22
*** tbarron has joined #openstack-cinder17:22
*** frontrunner has quit IRC17:22
*** mikal has quit IRC17:22
*** kambiz has joined #openstack-cinder17:22
*** Guest66666 has quit IRC17:22
*** bkopilov_ has quit IRC17:22
*** gouthamr has quit IRC17:22
*** Wi1d_ has quit IRC17:22
*** frontrunner has joined #openstack-cinder17:22
*** salv-orl_ has joined #openstack-cinder17:22
*** sage_ has joined #openstack-cinder17:22
*** jklare has joined #openstack-cinder17:22
*** krtaylor has joined #openstack-cinder17:22
*** pleia2 has joined #openstack-cinder17:23
*** haplo37_ has joined #openstack-cinder17:23
*** bkopilov_ has joined #openstack-cinder17:23
*** serverascode has joined #openstack-cinder17:23
*** alonma has joined #openstack-cinder17:23
*** pkoraca has joined #openstack-cinder17:23
*** shyama has joined #openstack-cinder17:23
*** zzzeek has joined #openstack-cinder17:24
*** sileht has joined #openstack-cinder17:24
*** david-lyle_ is now known as david-lyle17:24
*** diablo_rojo has quit IRC17:25
*** bluex has joined #openstack-cinder17:25
*** salv-orlando has quit IRC17:25
*** haplo37| has joined #openstack-cinder17:25
scottdapatrickeast: Yes! IT's not just you.17:25
*** swamireddy has joined #openstack-cinder17:26
*** rmk` has joined #openstack-cinder17:27
*** zhengyin has joined #openstack-cinder17:27
*** GB21 has quit IRC17:27
*** haplo37 has quit IRC17:28
*** alonma has quit IRC17:28
*** logan- has joined #openstack-cinder17:28
*** salv-orlando has joined #openstack-cinder17:31
*** frontrunner2 has joined #openstack-cinder17:32
*** lennyb has quit IRC17:32
*** rmk has quit IRC17:32
*** rmk` is now known as rmk17:32
*** rmk has quit IRC17:32
*** rmk has joined #openstack-cinder17:32
*** apopovych has quit IRC17:32
*** stevemar has quit IRC17:32
*** mtreinish has quit IRC17:33
*** unrahul has joined #openstack-cinder17:33
*** unrahul has quit IRC17:33
*** unrahul has joined #openstack-cinder17:33
*** logan- has quit IRC17:33
*** sage_ has quit IRC17:33
*** tbarron has quit IRC17:33
*** harlowja_ has quit IRC17:33
*** sdague has quit IRC17:33
*** tbarron has joined #openstack-cinder17:33
*** dillaman has joined #openstack-cinder17:33
*** frontrunner has quit IRC17:33
*** jklare_ has joined #openstack-cinder17:33
*** shyama has quit IRC17:33
*** jklare has quit IRC17:34
*** salv-orl_ has quit IRC17:34
*** sileht has quit IRC17:34
*** jdillaman has quit IRC17:34
*** stevemar has joined #openstack-cinder17:34
*** sileht has joined #openstack-cinder17:34
*** harlowja has joined #openstack-cinder17:34
*** sage_ has joined #openstack-cinder17:34
*** rmcall has joined #openstack-cinder17:35
*** shyama has joined #openstack-cinder17:35
*** sdague has joined #openstack-cinder17:35
jgriffithpatrickeast: I tihnk it's freenode17:36
jgriffithpatrickeast: i've been bouncing on and off all morning17:36
*** mnaser has joined #openstack-cinder17:37
*** diegows has joined #openstack-cinder17:37
*** logan- has joined #openstack-cinder17:38
*** aswadr_ has joined #openstack-cinder17:39
*** mtreinish has joined #openstack-cinder17:40
*** lennyb has joined #openstack-cinder17:40
*** rmcall has quit IRC17:41
SwansonArgle bargle17:41
*** Apoorva_ has quit IRC17:41
*** harlowja has quit IRC17:42
*** sudipto_ has quit IRC17:42
*** sudswas__ has quit IRC17:42
*** Apoorva has joined #openstack-cinder17:42
*** med_ has joined #openstack-cinder17:44
*** med_ is now known as Guest7927817:45
*** kong has joined #openstack-cinder17:46
*** shyama has quit IRC17:46
*** harlowja has joined #openstack-cinder17:47
*** DuncanT has joined #openstack-cinder17:47
openstackgerritErlon R. Cruz proposed openstack/cinder: Fix Cinder migrate_volume_completion  https://review.openstack.org/37794517:48
*** mvk has joined #openstack-cinder17:51
openstackgerritKarthik Prabhu Vinod proposed openstack/cinder: Switch service capabilities to ovo  https://review.openstack.org/31904017:52
*** wxy__ has joined #openstack-cinder17:52
*** rajinir has joined #openstack-cinder17:52
*** jraju has quit IRC17:54
*** apopovych has joined #openstack-cinder17:56
*** AndyWojo has joined #openstack-cinder17:57
*** aswadr_ has quit IRC17:58
*** briancurtin has joined #openstack-cinder18:00
*** rlrossit has joined #openstack-cinder18:01
*** rlrossit has quit IRC18:03
jgriffithSwanson: Mugles Bugles18:08
jgriffithso intersting thing about objects...18:08
jgriffithhow is it that cinderclient "knows" how to transform a response to an object ?18:08
*** kambiz_ has joined #openstack-cinder18:09
*** aspiers has quit IRC18:09
*** grumpycatt has quit IRC18:10
*** grumpycatt has joined #openstack-cinder18:10
*** harlowja has quit IRC18:10
*** jklare_ has quit IRC18:10
*** haplo37| has quit IRC18:10
*** pkoraca has quit IRC18:10
*** serverascode has quit IRC18:10
*** mtreinish_ has joined #openstack-cinder18:10
*** patrickeast_ has joined #openstack-cinder18:10
*** haplo37- has joined #openstack-cinder18:10
*** patrickeast_ has quit IRC18:10
*** patrickeast_ has joined #openstack-cinder18:10
*** yangyape_ has joined #openstack-cinder18:10
*** mvk has quit IRC18:10
*** Guest79278 has quit IRC18:10
*** lennyb has quit IRC18:10
*** sage_ has quit IRC18:10
*** yangyapeng has quit IRC18:10
*** mvk has joined #openstack-cinder18:11
jgriffithmy cinder/api/v3/attachments.py looks like this:  https://gist.github.com/j-griffith/8798ff8d4af4c9d0c3c7f209f53cca6d18:11
*** sdague has quit IRC18:11
*** wxy__ has quit IRC18:11
*** kambiz has quit IRC18:11
*** mtreinish has quit IRC18:11
*** mtreinish_ is now known as mtreinish18:11
*** diegows has quit IRC18:11
*** sileht has quit IRC18:11
*** diegows has joined #openstack-cinder18:11
*** jklare has joined #openstack-cinder18:11
jgriffithbut the data returned from cinderclient is the attachment object ???18:11
*** kambiz_ is now known as kambiz18:11
*** sage_ has joined #openstack-cinder18:12
jgriffithoh.. hehe, clever!!!18:12
*** serverascode has joined #openstack-cinder18:12
*** stevemar_ has joined #openstack-cinder18:12
*** sdague has joined #openstack-cinder18:13
*** sileht has joined #openstack-cinder18:13
*** lennyb has joined #openstack-cinder18:13
*** aspiers has joined #openstack-cinder18:15
*** chris_morrell has joined #openstack-cinder18:15
*** stevemar_ is now known as stevemar__18:17
*** stevemar__ has quit IRC18:18
*** e0ne has joined #openstack-cinder18:18
*** bardia has joined #openstack-cinder18:19
*** stevemar____ has joined #openstack-cinder18:20
*** patrickeast_ has quit IRC18:20
*** chris_morrell has quit IRC18:21
*** diegows has quit IRC18:21
*** diegows has joined #openstack-cinder18:21
*** sileht has quit IRC18:21
*** akapil has joined #openstack-cinder18:21
*** sbezverk_ has joined #openstack-cinder18:22
sbezverk_jgriffith: ping18:23
*** gershon__ has joined #openstack-cinder18:23
*** harlowja has joined #openstack-cinder18:24
*** sileht has joined #openstack-cinder18:24
*** pkoraca has joined #openstack-cinder18:26
*** patrickeast_ has joined #openstack-cinder18:28
*** akapil has quit IRC18:28
*** wxy__ has joined #openstack-cinder18:29
*** dillaman has quit IRC18:30
*** diegows has quit IRC18:30
*** dillaman has joined #openstack-cinder18:30
*** salv-orlando has quit IRC18:35
*** sdake has quit IRC18:35
*** Rajini_Ram_ has joined #openstack-cinder18:36
*** cknight1 has quit IRC18:37
*** logan- has quit IRC18:39
*** thingee has joined #openstack-cinder18:41
*** diegows has joined #openstack-cinder18:42
*** logan- has joined #openstack-cinder18:42
*** med_ has joined #openstack-cinder18:43
*** med_ is now known as Guest4087618:44
*** sdake has joined #openstack-cinder18:49
*** thingee has quit IRC18:49
*** viks has joined #openstack-cinder18:49
*** mtanino_ has joined #openstack-cinder18:50
vikshi, i need some help in a check-in18:50
viksI want to check-in under cinder and cinderclient but both are different repositories18:50
vikshow can I make a single checkin under one bug which is for these 2 repositories18:51
viksor am I missing something here?18:51
hemnayou can't18:51
hemna2 different repos = 2 different commits.18:51
hemna2 different reviews.18:51
*** Apoorva has quit IRC18:52
*** mtani____ has joined #openstack-cinder18:52
*** mtanino has quit IRC18:52
vikswhat about the bug...should I open 2 different bugs18:52
*** Apoorva has joined #openstack-cinder18:52
hemnaif they both contribute to fixing the same issue, no18:52
hemnaviks, https://wiki.openstack.org/wiki/GitCommitMessages18:52
viksbut then shold I have 2 chekcin under the same bug18:52
viksI mean 2 different commit ids18:53
*** lpetrut has joined #openstack-cinder18:53
vikswill that be OK?18:53
hemnaread ^^18:53
*** raunak has quit IRC18:54
*** Cibo has joined #openstack-cinder18:54
*** raunak has joined #openstack-cinder18:55
*** mtanino_ has quit IRC18:55
*** raunak has quit IRC18:55
*** raunak has joined #openstack-cinder18:56
*** DuncanT has quit IRC18:58
*** porrua has quit IRC18:58
*** DuncanT has joined #openstack-cinder19:00
*** stevemar____ has quit IRC19:02
*** sdake has quit IRC19:10
*** lpetrut has quit IRC19:11
*** haplo37- has quit IRC19:11
*** haplo37- has joined #openstack-cinder19:12
*** lpetrut1 has joined #openstack-cinder19:12
*** zzzeek has quit IRC19:12
*** zzzeek has joined #openstack-cinder19:12
*** lpetrut1 is now known as lpetrut19:12
*** salv-orlando has joined #openstack-cinder19:13
*** rlrossit has joined #openstack-cinder19:13
*** patrickeast has quit IRC19:14
*** patrickeast_ is now known as patrickeast19:14
*** erlon_ has joined #openstack-cinder19:18
*** salv-orlando has quit IRC19:18
*** xyang1 has joined #openstack-cinder19:18
patrickeastyay irccloud is alive again for me19:20
patrickeasthemna: jgriffith: not sure if my pings actually got sent earlier, i've got questions about attach/detach if you guys have a few min19:20
hemnaabout to head to lunch...19:21
patrickeastso uh, i think there is a race condition that breaks things with setups where you have multiple luns per iscsi session19:21
*** Rajini_Ram_ has quit IRC19:21
patrickeastponder this while at lunch19:21
patrickeastwe run the disconnect code in os-brick, and determine that we shouldn't run the iscsi logout since there are other volumes19:22
jgriffithpatrickeast: sounds like this might be similar to the issue johnthetubaguy is concerned about19:22
*** akapil has joined #openstack-cinder19:22
hemnaif you iscsi logout with other volumes there, they will be gone19:23
hemnaok lunch....19:25
*** logan- has quit IRC19:26
*** wxy__ has quit IRC19:26
*** serverascode has quit IRC19:27
*** briancurtin has quit IRC19:27
*** stevemar has quit IRC19:27
patrickeastno problemo, everythings good, and we do the scsi device deletes/flush everything, etc19:27
patrickeastwe release the lock in os-brick, and someone grabs it to do an iscsi attachment *before* terminate connection is called for cinder19:27
patrickeastos-brick happily does a iscsi rescan19:27
patrickeastand those paths we deleted come right back :(19:27
patrickeastlooking at the nova code and how we do the disconnect -> terminate connection flow doing them as a isolated operation seems... uh.. tricky19:27
patrickeastjgriffith: my assumption is that this isn't a problem for something like lvm, since we always logout of the iscsi session, right?19:27
patrickeastwhich is done inside the os-brick lock19:27
*** sileht has quit IRC19:27
*** unrahul has quit IRC19:27
*** yohoffman has quit IRC19:27
patrickeastsort of end result being that are devices left behind on the initiator that no one ever cleans up19:27
jgriffithpatrickeast: correct19:27
*** Guest40876 has quit IRC19:27
*** dillaman has quit IRC19:27
*** rlrossit has quit IRC19:27
*** aspiers has quit IRC19:27
*** sage_ has quit IRC19:27
*** tbarron has quit IRC19:27
*** mtani____ has quit IRC19:27
patrickeastjgriffith: maybe, i was looking through the eavesdrop from the conversation this morning19:27
jgriffithpatrickeast: it's the backends that share a single connection for multiple volumes19:27
patrickeastnot 100% i followed19:27
*** xyang1 has quit IRC19:27
*** lpetrut has quit IRC19:27
*** lpetrut has joined #openstack-cinder19:27
*** serverascode has joined #openstack-cinder19:28
*** tbarron has joined #openstack-cinder19:28
*** akapil has quit IRC19:28
patrickeastbut similar at least19:28
jgriffithpatrickeast: oh... the rescan thing, that's been brought up before I thought19:28
*** yohoffman has joined #openstack-cinder19:28
patrickeasthemna: right, we shouldn't have logged out19:28
*** dillaman has joined #openstack-cinder19:28
*** sage_ has joined #openstack-cinder19:28
patrickeastderp, i think i got dropped for a sec there19:29
openstackgerritMerged openstack/cinder: Updated from global requirements  https://review.openstack.org/37742619:29
*** aspiers has joined #openstack-cinder19:29
*** patrickeast_ has joined #openstack-cinder19:30
* patrickeast_ grumbles about irc19:30
patrickeast_might have lost some history19:30
*** stevemar has joined #openstack-cinder19:30
jgriffithpatrickeast_: better than carrier pigeons19:31
patrickeast_hemna: last thing i saw was your comment about not logging out when other volumes are there, and thats totally fine, that part of this is working correctly19:31
*** jroll has quit IRC19:31
*** alonma has joined #openstack-cinder19:32
*** patrickeast__ has joined #openstack-cinder19:32
*** lpetrut1 has joined #openstack-cinder19:34
*** logan- has joined #openstack-cinder19:34
*** freerunner has quit IRC19:35
*** mkoderer__ has quit IRC19:35
*** Lehtu has quit IRC19:35
*** tonyb has quit IRC19:35
*** jamespd_ has quit IRC19:35
*** patrickeast has quit IRC19:35
*** pkoraca has quit IRC19:35
*** patrickeast__ is now known as patrickeast19:35
*** aspiers has quit IRC19:35
*** sage_ has quit IRC19:35
*** serverascode has quit IRC19:35
*** DuncanT has quit IRC19:35
*** rajinir has quit IRC19:35
*** swamireddy has quit IRC19:35
*** pleia2 has quit IRC19:35
*** rha has quit IRC19:35
*** zul has quit IRC19:35
*** jamespd has joined #openstack-cinder19:36
*** tbarron has quit IRC19:36
*** erlon_ has quit IRC19:36
*** mtreinish has quit IRC19:36
*** AndyWojo has quit IRC19:36
*** krtaylor has quit IRC19:36
*** vkmc has quit IRC19:36
*** lpetrut has quit IRC19:36
*** stevemar has quit IRC19:36
*** patrickeast_ has quit IRC19:36
*** zzzeek has quit IRC19:36
*** bhagyashri has quit IRC19:36
*** tonyb_ has joined #openstack-cinder19:36
*** tbarron has joined #openstack-cinder19:36
*** mtreinish has joined #openstack-cinder19:36
*** lpetrut1 is now known as lpetrut19:36
*** pleia2 has joined #openstack-cinder19:36
*** alonma has quit IRC19:36
*** stevemar has joined #openstack-cinder19:36
*** sage__ has joined #openstack-cinder19:36
*** sileht has joined #openstack-cinder19:37
*** freerunner has joined #openstack-cinder19:37
*** aspiers has joined #openstack-cinder19:37
*** vkmc has joined #openstack-cinder19:37
*** jroll has joined #openstack-cinder19:37
*** jroll has quit IRC19:37
*** jroll has joined #openstack-cinder19:37
*** rlrossit has joined #openstack-cinder19:37
*** krtaylor has joined #openstack-cinder19:37
*** bhagyashri has joined #openstack-cinder19:38
*** erlon_ has joined #openstack-cinder19:38
*** mkoderer__ has joined #openstack-cinder19:39
*** rha has joined #openstack-cinder19:39
*** rha has joined #openstack-cinder19:39
*** swamireddy has joined #openstack-cinder19:40
*** alonma has joined #openstack-cinder19:40
*** Lehtu has joined #openstack-cinder19:40
*** rlrossit_ has joined #openstack-cinder19:41
*** med_ has joined #openstack-cinder19:43
*** med_ is now known as Guest4578919:43
*** kfox1111 has quit IRC19:44
*** krtaylor has quit IRC19:44
*** jroll has quit IRC19:44
*** stevemar has quit IRC19:44
*** rha has quit IRC19:44
*** krtaylor has joined #openstack-cinder19:44
*** haplo37- has quit IRC19:44
*** swamireddy has quit IRC19:45
*** aspiers has quit IRC19:45
*** jamespd has quit IRC19:45
*** mdavidson has quit IRC19:45
*** salv-orlando has joined #openstack-cinder19:45
*** bhagyashri has quit IRC19:45
*** vkmc has quit IRC19:45
*** sage__ has quit IRC19:45
*** rlrossit has quit IRC19:45
*** alonma has quit IRC19:45
*** mkoderer__ has quit IRC19:45
*** dillaman has quit IRC19:45
*** jamespd_ has joined #openstack-cinder19:45
*** mkoderer__ has joined #openstack-cinder19:45
*** dillaman has joined #openstack-cinder19:45
*** rha has joined #openstack-cinder19:46
*** rha has joined #openstack-cinder19:46
*** vkmc has joined #openstack-cinder19:46
*** jroll has joined #openstack-cinder19:46
*** jroll has quit IRC19:46
*** jroll has joined #openstack-cinder19:46
*** sage__ has joined #openstack-cinder19:46
*** xyang has joined #openstack-cinder19:47
*** aspiers has joined #openstack-cinder19:47
*** haplo37- has joined #openstack-cinder19:47
*** patrickeast has quit IRC19:48
*** alonma has joined #openstack-cinder19:48
*** jdurgin has quit IRC19:48
*** swamireddy has joined #openstack-cinder19:49
*** kfox1111 has joined #openstack-cinder19:49
*** zul has joined #openstack-cinder19:49
*** patrickeast has joined #openstack-cinder19:49
*** xyang has quit IRC19:51
patrickeastcan’t keep connected19:51
*** xyang has joined #openstack-cinder19:51
patrickeastjgriffith: so that scenario where potentailly we need to be careful about cleanup and really like worst case could even need locking between teh os-brick stuff and terminate_connection is something that concerns me19:52
*** rlrossit_ has quit IRC19:52
patrickeastjgriffith: i figure since we’re (you’re) rewriting things we should try and figure out at least a strategy to handle it19:52
*** alonma has quit IRC19:52
*** rlrossit has joined #openstack-cinder19:52
jgriffithpatrickeast: yeah19:52
*** zzzeek has joined #openstack-cinder19:52
*** stevemar has joined #openstack-cinder19:52
jgriffithpatrickeast: so I'm working right now on changing the method to by default require an ack before setting in-use19:53
patrickeastjgriffith: sort of high level view ideally we are smarter and after we do a rescan we can be like “hey, you were deleted, go away”19:53
patrickeastjgriffith: ack from the backend?19:53
*** DuncanT has joined #openstack-cinder19:54
*** Apoorva has quit IRC19:54
jgriffithpatrickeast: well, so this is the part you can help me with :)  Since I don't have experience with a backend that behaves this way19:54
*** unrahul has joined #openstack-cinder19:54
jgriffithpatrickeast: my thought was that this is something that's magically being handled with initialize_connection currently today?19:54
jgriffithpatrickeast: if that's not the case, then it's an existing issue?19:54
patrickeastjgriffith: well so for the most part this stuff is pretty well figured out19:55
*** alonmarx has joined #openstack-cinder19:55
jgriffithpatrickeast: haha... yes, but where?19:55
patrickeastjgriffith: but this rescan before terminate connection is an issue today19:55
*** xyang has quit IRC19:55
*** alonma has joined #openstack-cinder19:55
patrickeastjgriffith: sort of like the whole process needs to be synchronized… well thats a bad term, orchestrated? better between the os-brick disconnect and terminate_connection19:55
jgriffithpatrickeast: oh,  sorry... I assumed you were talking about the other case; the "brick connection" failed while there were two attaches in flight19:55
*** rajinir has joined #openstack-cinder19:56
patrickeastjgriffith: thats also a problem19:56
jgriffithpatrickeast: we could do that if brick wasn't and independent thing19:56
jgriffithpatrickeast: well...19:56
sbezverk_jgriffith: ping19:56
jgriffithpatrickeast: so I *think* the idea of saying that we don't go to "in-use" or "avaialble" until AFTER we get an ack back from nova would help woulnd't it?19:57
jgriffithsbezverk_: pong19:57
jgriffithpatrickeast: so the default states are attaching and detaching... and we treat those as hard locks19:57
johnthetubaguyjgriffith: heh, thats a familiar discussioin19:57
jgriffithpatrickeast: although if the backend doesn't know how to deal with that then I'm kinda sunk :)19:57
sbezverk_jgriffith: that VG leak fix, I saw it went to rc2, but did it go to master?19:57
patrickeastjgriffith: that helps but for a different problem, the state change or volume operations isn’t such a big deal19:57
jgriffithjohnthetubaguy: indeed!  dejavous19:57
*** briancurtin has joined #openstack-cinder19:58
jgriffithjohnthetubaguy: it seems that it's an existing problem :(  patrickeast has been fortunate enough to encounter it on the detach side of things19:58
jgriffithpatrickeast: well, you could use that as a cheap lock19:58
patrickeastjgriffith: yea but its like locking at the wrong granularity, i think19:58
jgriffithpatrickeast: but it requires the driver to check the status of any/all volumes on that host19:58
*** jdurgin has joined #openstack-cinder19:59
jgriffithpatrickeast: I won't argue that19:59
patrickeastjgriffith: needs to be like on an iscsi session level19:59
patrickeastjgriffith: it might be OK though, need to think about it more19:59
jgriffithpatrickeast: so if you look at the table I introduced, it has connector info including iqn19:59
jgriffithpatrickeast: seems to me that a backend could be intelligent enough to inspect that info when doing initialize or attach terminate/detach and do the right thing20:00
patrickeastjgriffith: oh yea, thats true20:00
patrickeastjgriffith: but the problem is n-cpu+os-brick20:00
patrickeastjgriffith: so like if it does a disconnect volume with os-brick20:00
jgriffithpatrickeast: the goal was to have that detail available on the Cinder side to be smarter.  Yeah, the brick rescan thing is a whole different ball of wax20:00
patrickeastjgriffith: and then attach volume (different volumes)20:00
*** alonma has quit IRC20:00
patrickeastjgriffith: before it calls terminate connection for the first one (thats the trick)20:01
jgriffithpatrickeast: but those are problems in brick since day-1 I think20:01
patrickeastjgriffith: oh yea totally20:01
patrickeastjgriffith: i guess i’m looking at this as our chance to kinda take another look at what were doing and maybe adjust20:01
jgriffithpatrickeast: I don't know how to fix them now that brick is outside.  Unless the stuff I'm working on for nova-vol-agents pans out20:01
patrickeastjgriffith: not just cinder <-> nova but the brick calls too20:01
*** diablo_rojo has joined #openstack-cinder20:02
jgriffithpatrickeast: yeah, I hear ya20:02
jgriffithpatrickeast: and most definitely smart20:02
patrickeastjgriffith: oo nova-vol-agents?20:02
jgriffithsbezverk_: I am paying attention kinda :)20:02
jgriffithpatrickeast: yeah20:02
patrickeastjgriffith: i was thinking about that c-data agent again… let it do all the work20:02
patrickeastjgriffith: but mebbe thats the same idea20:02
jgriffithpatrickeast: I"m workign on a blueprint in my spare time20:02
jgriffithpatrickeast: so this is back to the local volumes problem and the original brick proposal20:02
*** serverascode has joined #openstack-cinder20:03
*** alonma has joined #openstack-cinder20:03
jgriffithpatrickeast: two birds with one stone20:03
sbezverk_jgriffith: np I see you have hot discussion. Appreciate if you could ping me when you have spare second. by some reason I still see the issue with the latest master..20:03
patrickeastjgriffith: yea totally, if we just had a piece of cinder on each compute node things get a little bit easier to coordinate20:03
jgriffithpatrickeast: said agent (vol, data... hippo) can coordinate a number of things for us20:03
jgriffithsbezverk_: will do, thanks20:03
*** jungleboyj has joined #openstack-cinder20:03
jgriffithpatrickeast: it allows us to do some really interesting things.  I'm not getting on my soap box on this again though.  Instead I'm going to code it this time :)20:04
jgriffithpatrickeast: https://wiki.openstack.org/wiki/CinderBrick20:04
patrickeastjgriffith: the interesting thing is we can, theoretically, almost just drop a c-vol on every single compute node with the promises of our shiny A-A HA future20:04
jgriffithpatrickeast: that's obviously just the beginning, add in the parts you're talking about as well20:04
jgriffithpatrickeast: :)20:05
jgriffithpatrickeast: so the prototype I started last week-end does exactly that20:05
*** xyang has joined #openstack-cinder20:05
jgriffithpatrickeast: but it also takes my "detached" lvm driver for compute nodes that you don't actually want to serve volumes from20:05
patrickeastjgriffith: for just local lvm vols?20:06
jgriffithpatrickeast: but, the good thing is at the end of the day it gets rid of custom block-device drivers, lvm sharing and provides a coordination capability20:06
*** pkoraca has joined #openstack-cinder20:06
patrickeastjgriffith: i like the sound of that20:06
*** AndyWojo has joined #openstack-cinder20:07
jgriffithpatrickeast: so the way I have it mapped out is you *can* have a cinder VG on the compute node.  If you specify to the scheduler you want a vol on host-X to attach to an Instance on host-X then is just skps doing target creation20:07
jgriffithpatrickeast: if on the other hand you're not on the same host, it attaches a target and exports it the way we do currently20:07
jgriffithpatrickeast: that also means you can live-migrate and do all the the things you do today if you need/want to20:07
jgriffithpatrickeast: I'm trying to slowly get us back to having a consistent API :)20:08
*** alonma has quit IRC20:08
jgriffithpatrickeast: but I'm sure there will be some new whiz-bang features added this release that will torpedo that anyway :)20:08
patrickeastjgriffith: consistent? whaaa?20:08
*** kaisers1 has quit IRC20:08
jgriffithpatrickeast: yeah I know right :)20:08
jgriffithpatrickeast: anyway... back to our earlier convo :)20:09
jgriffithpatrickeast: let me get this first iteration of the new flow working again20:09
*** kaisers has joined #openstack-cinder20:09
jgriffithpatrickeast: then we can pick that apart a bit and see what we can improve20:09
patrickeastjgriffith: sounds good20:09
jgriffithsbezverk_: what's up?20:09
*** patrickeast_ has joined #openstack-cinder20:13
*** sdake has joined #openstack-cinder20:13
sbezverk_jgriffith: I just build new cinder-volume image based on latest master and I still hit that issue20:14
sbezverk_I checked the bug and it says fix is in rc220:14
sbezverk_I thought it would make to master first20:15
*** julim has joined #openstack-cinder20:15
jgriffithsbezverk_: hmm, it certainly should have20:16
jgriffithsbezverk_: https://review.openstack.org/#/c/365167/20:16
*** julim has quit IRC20:16
jgriffithsbezverk_: just yesterday20:16
jgriffithsbezverk_: I don't think it's been backported as of yet20:16
*** e0ne has quit IRC20:18
jgriffithsbezverk_: nope, it hasn't but I'll propose that right now20:18
*** e0ne has joined #openstack-cinder20:18
*** wilson_liu has quit IRC20:19
*** huanan_M has quit IRC20:19
sbezverk_jgriffith: thank you. appreciate it..20:20
jgriffithsbezverk_: https://review.openstack.org/#/c/378013/20:20
*** wilson has joined #openstack-cinder20:20
jgriffithsbezverk_: no problem, glad you reminded me :)20:20
*** rlrossit_ has joined #openstack-cinder20:20
sbezverk_jgriffith: do you think it will take anoher month of so to get backported ;-) to master ?20:20
*** wilson is now known as Guest7430720:20
*** patrickeast__ has joined #openstack-cinder20:21
*** rlrossit_ has quit IRC20:21
*** huanan has joined #openstack-cinder20:21
jgriffithLOL... it should be in master now; it should make it into Newton shortly20:21
*** logan- has quit IRC20:21
*** pkoniszewski has quit IRC20:21
*** jschwarz has quit IRC20:21
*** e0ne_ has joined #openstack-cinder20:22
*** unrahul has quit IRC20:22
*** e0ne_ has quit IRC20:22
jgriffithsbezverk_: I'll get smcginnis and patrickeast__ to give it a vote and I'll +a it20:22
*** flip214 has quit IRC20:22
*** _alastor_ has quit IRC20:22
*** j_king has quit IRC20:22
smcginnisjgriffith: Looking...20:22
jgriffithsbezverk_: just waiting for jenkins to finish20:22
patrickeast__jgriffith: link?20:22
*** asselin has quit IRC20:22
*** mkoderer___ has joined #openstack-cinder20:22
jgriffithpatrickeast__: https://review.openstack.org/#/c/378013/20:22
patrickeast__oh dis one? https://review.openstack.org/#/c/378013/120:22
*** pkoraca has quit IRC20:22
*** rajinir has quit IRC20:22
*** DuncanT has quit IRC20:22
*** swamireddy has quit IRC20:23
*** xyang has quit IRC20:23
*** jungleboyj has quit IRC20:23
jgriffithyeah, still needs to run through gate20:23
jgriffithbut it's a straight backport20:23
patrickeast__oh man, i thought we backported it already20:23
jgriffithpatrickeast: yeah, me too :(20:23
*** sdake_ has joined #openstack-cinder20:23
*** zzzeek has quit IRC20:23
*** sage__ has quit IRC20:23
*** jroll has quit IRC20:23
*** vkmc has quit IRC20:23
smcginnisjgriffith: Hmm, we should probably have had a DocImpact tag on that one.20:23
jgriffithpatrickeast: but I had some interference getting it to land in master that slowed things down a bit20:23
*** j_king has joined #openstack-cinder20:23
smcginnisOh well20:23
jgriffithsmcginnis: garrrr20:23
*** jroll has joined #openstack-cinder20:23
*** jroll has quit IRC20:23
*** jroll has joined #openstack-cinder20:23
*** patrickeast has quit IRC20:23
*** mkoderer__ has quit IRC20:24
*** e0ne has quit IRC20:24
*** asselin_ has joined #openstack-cinder20:24
*** sdake has quit IRC20:24
*** rlrossit has quit IRC20:24
*** zul has quit IRC20:24
*** aspiers has quit IRC20:24
*** jungleboyj_ has joined #openstack-cinder20:24
smcginnisAt least the release note is good.20:24
*** patrickeast__ is now known as patrickeast20:24
jgriffithsmcginnis: You want me to update the commit on the backport at least?20:24
*** _alastor_ has joined #openstack-cinder20:24
smcginnisAnd it's targeted to a specific case20:24
*** vkmc has joined #openstack-cinder20:24
jgriffithsmcginnis: yeah, I mean how much docs do we needs :)20:24
*** gouthamr_ has quit IRC20:24
*** pkoniszewski has joined #openstack-cinder20:24
smcginnisjgriffith: I guess I'm fine either way.20:24
*** sage__ has joined #openstack-cinder20:24
*** gouthamr has joined #openstack-cinder20:24
*** Apoorva has joined #openstack-cinder20:25
*** aspiers has joined #openstack-cinder20:25
*** swamireddy has joined #openstack-cinder20:25
hemnapatrickeast, back20:26
jgriffithsmcginnis: I updated20:26
*** wxy__ has joined #openstack-cinder20:26
smcginnisjgriffith: Cool, thanks.20:26
*** logan- has joined #openstack-cinder20:26
*** jschwarz has joined #openstack-cinder20:26
patrickeasthemna: :D how much scrollback do you have?20:27
*** flip214 has joined #openstack-cinder20:27
*** flip214 has quit IRC20:27
*** flip214 has joined #openstack-cinder20:27
*** sam64_ has quit IRC20:27
smcginnisWould be awesome if folks would stop DoSing freenode. :[20:27
patrickeastsmcginnis: +120:27
*** sam64 has joined #openstack-cinder20:27
jgriffithsmcginnis: +A20:27
*** zzzeek has joined #openstack-cinder20:27
jgriffithsmcginnis: patrickeast can't we all just get along20:28
smcginnisMaybe someone's not happy that their patch didn't land. :D20:28
jgriffithsmcginnis: huh... well that's a thing I guess20:29
sbezverk_jgriffith: when it is merged it will be in cinder's tarball right away or tomorrow?20:29
hemnapatrickeast, so, after a disconnect_volume call, and a cinder terminate_connection call, we potentially get another os_brick.connect_volume call that induces an iscsi rescan?20:29
hemnawhich would then bring those volumes back20:30
*** viks has quit IRC20:30
patrickeasthemna: so the order is like disconnect_volume(A), connect_volume(B), rescan, terminate_connection(A)20:30
*** Apoorva_ has joined #openstack-cinder20:30
hemnaso I think the only way we can be safe there, is to change the order of operations20:30
hemnacall terminate_connection(a), disconnect_volume(a)20:31
patrickeasthemna: is that safe? what about flushing and all that?20:31
hemnacinder will remove the export from the backend first.20:31
patrickeasti guess its like disconnect from vm, flush paths, terminate connection, disconnect volume?20:31
hemnaI think the volume is already removed from the instance at that point20:31
hemnabefore calling brick20:32
patrickeastit is when we call disconnect volume20:32
*** ntpttr has quit IRC20:32
hemnawe could add a flush io api to brick20:32
patrickeastso the problem (or part of it) is looking through where we do terminate connection its like in a different layer than where the disconnect happens20:32
hemnato force any transient io out20:32
*** AndyWojo has quit IRC20:32
*** jschwarz has quit IRC20:32
*** serverascode has quit IRC20:32
*** aspiers has quit IRC20:32
*** briancurtin has quit IRC20:32
*** guitarzan has quit IRC20:32
openstackgerritErlon R. Cruz proposed openstack/cinder: Fix Cinder migrate_volume_completion  https://review.openstack.org/37794520:32
patrickeastits in like the nova virt driver destroy iirc20:32
patrickeastand the terminate comes from the like manager later on after the destroy20:32
*** jamespd_ has quit IRC20:33
*** jamespd has joined #openstack-cinder20:33
*** Apoorva has quit IRC20:33
*** vkmc has quit IRC20:33
*** jroll has quit IRC20:33
*** tbarron has quit IRC20:33
patrickeastso we’d need to rearrange some stuff over there20:33
*** jschwarz has joined #openstack-cinder20:33
*** tbarron has joined #openstack-cinder20:33
*** flip214 has quit IRC20:33
*** guitarzan has joined #openstack-cinder20:33
patrickeastjgriffith: which i guess is where your new changes for the nova side of things comes in :D20:33
*** flip214 has joined #openstack-cinder20:33
*** flip214 has quit IRC20:33
*** flip214 has joined #openstack-cinder20:33
*** ntpttr has joined #openstack-cinder20:33
jgriffithpatrickeast: :)20:34
*** jroll has joined #openstack-cinder20:34
*** jroll has quit IRC20:34
*** jroll has joined #openstack-cinder20:34
jgriffithpatrickeast: hemna that doesn't cause problems for *something* does it?20:34
patrickeastoh no doubt20:34
*** vkmc has joined #openstack-cinder20:34
* patrickeast glances at the remotefs backends20:34
hemnayah I'm not sure actually20:35
patrickeastjgriffith: as much as i don’t want to double down on snowflakes, it might be something where we could give hints to nova (or whoever) on how they should disconnect20:35
patrickeastjgriffith: at a more fine grained level than rbd, or iscsi20:36
*** aspiers has joined #openstack-cinder20:36
jgriffithpatrickeast: sigh.... but frankly, isn't that what brick is supposed to do for us?20:36
patrickeastjgriffith: probably20:36
jgriffithpatrickeast: abstract out the customization for things like initiatlize/terminate20:36
patrickeastjgriffith: i guess it comes down to extending what brick does then20:36
patrickeastjgriffith: if we put the terminate_connection calls and stuff into it, then yea totally20:37
jgriffithpatrickeast: so create a new driver_type for brick like "iscsi-shared" or something20:37
hemnaso one thing is that os-brick does a blind rescan20:37
*** zul has joined #openstack-cinder20:37
hemnaregardless of the # of sessions it has20:37
jgriffithhemna: don't do that :)20:37
hemnait's been that way for ages20:37
hemnawe could somehow get the session id for the target_portal in question20:37
hemnaand only rescan that20:37
patrickeasthemna: well, even if we target them to a specific session if there are multiple luns sharing one its the “right” thing to do, but same problem20:37
jgriffithhemna: I know, I'm pretty sure this has come up before... like last year or something20:37
hemnabut that still leaves the race for that portal20:37
*** rhagarty_ has joined #openstack-cinder20:38
patrickeastwhat i’m wondering is if maybe we just say screw it, assume we can’t cleanly disconnect20:38
patrickeastand look more at our attachment code to see if we can make it work better if it starts on a borked system20:38
*** jklare has quit IRC20:38
patrickeastin theory if brick had a list of paths it knew 100% were not valid anymore20:39
patrickeastit can just rescan, and clean them up again20:39
*** DuncanT has joined #openstack-cinder20:39
patrickeastlike, as a sysadmin attaching volumes20:39
hemnabrick currently is stateless20:39
patrickeastif i did a rescan and saw them come back thats how i would fix it20:39
hemnathe problem is that the next attach, those paths could be valid20:39
patrickeasthemna: yep, which is a problem20:39
hemnaand we don't want to nuke em20:40
patrickeasthemna: hmm true20:40
patrickeasthemna: i wonder if we can check the scsi id on them or something to see how valid they are20:40
hemnawhat to do in the case where someone is attaching that same volume to another instance20:41
hemnathe wwn would be valid20:41
patrickeastcry a little on the inside?20:41
*** unrahul has joined #openstack-cinder20:41
patrickeasti guess its possible some backends would have another attachment for the same volume?20:42
patrickeastbut why wouldn’t they just be the same /dev/sdX’s20:42
hemnawhen we do connect_volume, I return the scsi wwn along with everything else.20:42
*** jklare has joined #openstack-cinder20:42
hemnaudev controls what /dev/sdX is created20:42
hemnaand that's not guaranteed to be the same value20:42
smcginnispatrickeast: If it's the same volume over the same path it would be the same sdX.20:43
hemnahence the entire reason for using scsi_id in the linux scsi subsystem20:43
*** rajinir has joined #openstack-cinder20:43
*** eharney has quit IRC20:43
*** serverascode has joined #openstack-cinder20:43
smcginnisBut if it's over a different path it would end up different.20:43
hemnasmcginnis, not if another volume came along20:43
hemnabetween the disconnect and attach20:43
patrickeastmaybe i give too much credit to udev and team20:43
hemnaudev is going to use the next avail20:43
*** zengyingzhe_ has quit IRC20:44
hemnathis is the entire reason for using scsi wwn's20:44
smcginnispatrickeast: Oh, different host. Sorry, read through too quick.20:44
smcginnisYeah, then no guarantee of sameness.20:44
*** patrickeast_ has quit IRC20:44
*** asselin has joined #openstack-cinder20:44
smcginnisI was thinking multiple attaches on the same host.20:44
*** Anticimex has quit IRC20:44
hemnaage old linux problem20:44
hemnafor mountpoints etc.20:44
*** jamespd_ has joined #openstack-cinder20:44
patrickeastsmcginnis: no wait, your right it is the same host20:44
*** rhagarty__ has joined #openstack-cinder20:44
*** jdillaman1 has joined #openstack-cinder20:44
*** frakt_ has quit IRC20:44
*** zengyingzhe_ has joined #openstack-cinder20:44
patrickeasti think the problem is that the same volume could be presented as different sdx devices if attached again20:45
*** jamespd has quit IRC20:45
*** dillaman has quit IRC20:45
*** guitarzan has quit IRC20:45
*** jungleboyj_ has quit IRC20:45
*** asselin_ has quit IRC20:45
*** diablo_rojo has quit IRC20:45
smcginnispatrickeast: OK, then if you're adding another attachment for the same volume on the same host, it should be the same. Assuming your storage works like mine with the same targets.20:45
*** jungleboyj__ has joined #openstack-cinder20:45
smcginnispatrickeast: Shouldn't be.20:45
patrickeastsmcginnis: yep, thats that i was thinking20:45
* hemna wonders if creating a python lib in front of libiscsi would help anything20:45
patrickeastbut i guess i can’t speak for all backends20:45
*** diablo_rojo has joined #openstack-cinder20:45
smcginnispatrickeast: I think for us it's a noop basically, right? For mine if I'm asked to attach a volume to the same host it's already attached to I just return the existing info.20:46
smcginnishemna: I thought I saw the start of a library like that.20:46
*** DuncanT has quit IRC20:46
*** swamireddy has quit IRC20:46
*** sdake_ has quit IRC20:46
*** jamielennox has quit IRC20:46
*** rhagarty_ has quit IRC20:46
*** vkmc has quit IRC20:46
*** jroll has quit IRC20:46
*** tbarron has quit IRC20:46
*** sage__ has quit IRC20:46
*** rha has quit IRC20:46
*** krtaylor has quit IRC20:46
*** Anticime1 has joined #openstack-cinder20:46
*** guitarza1 has joined #openstack-cinder20:46
*** tbarron has joined #openstack-cinder20:46
*** jroll has joined #openstack-cinder20:46
*** sage__ has joined #openstack-cinder20:46
patrickeasthemna: yea i was poking around a little bit at something like that, or python bindings for like libdevmapper20:46
*** krtaylor has joined #openstack-cinder20:46
*** jroll is now known as Guest4994520:46
patrickeastif we can cut out some of the hacky cli tricks we have to do it would be (maybe) easier to manage some of this stuff20:47
hemnapatrickeast, yah I think that might be worth hacking at20:47
hemnamaybe in my next job.....20:47
*** sdake has joined #openstack-cinder20:47
*** briancurtin has joined #openstack-cinder20:47
*** Guest49945 is now known as jroll|dupe20:47
*** jroll|dupe has quit IRC20:47
*** jroll|dupe has joined #openstack-cinder20:47
*** rha has joined #openstack-cinder20:48
hemna2 years old.....20:48
*** swamireddy has joined #openstack-cinder20:48
smcginnishemna: Don't know about all the "swig" business, but could be a start.20:49
*** vkmc has joined #openstack-cinder20:50
*** xyang has joined #openstack-cinder20:50
*** jamielennox has joined #openstack-cinder20:51
*** xyang has quit IRC20:51
*** pkoraca has joined #openstack-cinder20:53
hemnapatrickeast, https://github.com/openstack/os-brick/blob/master/os_brick/initiator/connectors/iscsi.py#L410-L41220:53
*** AndyWojo has joined #openstack-cinder20:53
hemnathat comment is....suspect20:53
jgriffithsmcginnis: from somebody that did a lot of swig in a past life, "DON"T DO IT"20:53
*** xyang has joined #openstack-cinder20:53
hemnaand the funny thing is, rescan looks like it happens twice20:53
smcginnisjgriffith: Never had to touch it.20:54
jgriffithsmcginnis: awesome idea/concept, and works great for the more trivial cases20:54
openstackgerritXinli Guan proposed openstack/cinder: qos_specs: Correcting the qos_specs format  https://review.openstack.org/37802820:54
*** jroll|dupe is now known as jroll20:54
jgriffithsmcginnis: it gets kinda gnarly depending on how much OO you have in your domain20:54
jgriffithsmcginnis: hard to debug20:55
*** frakt has joined #openstack-cinder20:55
smcginnisHah: "for the most trivial cases" <- probably not a good fit for us. We don't do anything trivial. :D20:55
*** DuncanT has joined #openstack-cinder20:57
hemnapatrickeast, I'm going to remove line 41220:58
*** manjeets has joined #openstack-cinder20:59
*** patrickeast_ has joined #openstack-cinder20:59
*** Apoorva has joined #openstack-cinder21:02
*** akapil has joined #openstack-cinder21:02
*** eharney has joined #openstack-cinder21:04
openstackgerritWalter A. Boring IV (hemna) proposed openstack/os-brick: Remove the duplicate calls to rescan  https://review.openstack.org/37803321:05
hemnapatrickeast, ^^21:05
*** lpetrut has quit IRC21:05
*** Apoorva_ has quit IRC21:05
openstackgerritWalter A. Boring IV (hemna) proposed openstack/cinder: 3PAR use OSLO versioned Object for volume  https://review.openstack.org/33914721:05
hemnaso that just gets us to a single rescan call21:06
hemnaso that issue still exists21:06
* hemna wonders if we just removed iscsi_rescan entirely ?21:07
patrickeasti think its needed21:08
patrickeastat least when i tested21:08
*** xyang has quit IRC21:14
karthikpPlease could somebody review this : https://review.openstack.org/#/c/332480/  ??21:14
hemnapatrickeast, https://github.com/open-iscsi/open-iscsi/blob/master/usr/iscsiadm.c#L756-L77821:14
*** xyang has joined #openstack-cinder21:15
*** patrickeast has quit IRC21:15
*** patrickeast_ is now known as patrickeast21:16
*** xyang has quit IRC21:17
*** akapil has quit IRC21:20
*** sdake has quit IRC21:22
*** kmartin has joined #openstack-cinder21:22
hemnaok well that was fruitless21:22
hemnapatrickeast, smcginnis http://paste.openstack.org/show/583169/21:23
smcginnishemna: Ship it.21:23
*** mikal_ is now known as mikal21:29
*** cknight has joined #openstack-cinder21:32
*** gouthamr has quit IRC21:33
*** cknight1 has joined #openstack-cinder21:35
patrickeastoo this stackviz thing is pretty cool21:36
patrickeasti wonder how long jenkins has been doing this21:37
smcginnispatrickeast: That's part of jenkins now?21:37
hemnastackviz ?21:38
*** cknight has quit IRC21:38
smcginnishemna: https://github.com/openstack/stackviz21:38
smcginnisCool idea21:39
hemnaanalyze the performance of devstack setup ?21:40
*** Apoorva has quit IRC21:44
patrickeastsmcginnis: hemna: yea http://logs.openstack.org/86/366686/4/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/8f04b42/logs/stackviz/#/stdin/timeline21:45
*** Apoorva has joined #openstack-cinder21:45
patrickeasti added in the nodepool element for it, and now my ci does it too21:45
hemnaah nice21:45
patrickeastwas amazed at how much it just kinda worked without any config http://openstack-logs.purestorage.com/86/366686/4/check/PureISCSIDriver-tempest-dsvm-trusty-mn-multipath/79db9e4/logs/stackviz/#/stdin21:46
patrickeastnot sure just how helpful it is, but sure is a pretty graph21:46
smcginnispatrickeast: Would be good if you could drill down into more details.21:46
patrickeastsmcginnis: yea i guess for now its just tempest logs21:48
*** tonyb_ is now known as tonyb21:56
*** xyang has joined #openstack-cinder21:57
*** nherciu_ has quit IRC21:58
sbezverk_jgriffith: sorry to bug you, that config parameter to ignore message, which section it goes under?21:59
*** kaisers2 has quit IRC21:59
*** kaisers1 has joined #openstack-cinder21:59
jgriffithsbezverk_: needs to go under the driver section22:00
jgriffithsbezverk_: which I guess is kind of annoying with multiple LVM backends22:00
*** sdague has quit IRC22:02
patrickeastif only we had some sort of default section22:03
openstackgerritKarthik Prabhu Vinod proposed openstack/cinder: Fix for resource field length in database for quota tables  https://review.openstack.org/37517922:08
patrickeasthemna: found a new fun trick while testing out those changes for os-brick to stop rescanning things22:10
patrickeasthemna: somehow by the end of tempest i had 1 iscsi session still up22:11
*** alonma has joined #openstack-cinder22:11
hemnastill up ?22:11
patrickeastthe other 3 were logged out22:11
patrickeastyea uhh one sec22:12
*** mriedem has quit IRC22:12
*** ccesario has quit IRC22:12
*** haplo37- has quit IRC22:12
jgriffithpatrickeast: lol22:13
*** delattec has joined #openstack-cinder22:13
patrickeastjgriffith: it like never ends22:13
*** xyang has quit IRC22:13
jgriffithpatrickeast: the gift that keeps on giving :)22:14
jgriffithpatrickeast: but didn't somebody figure out a fix for that?22:14
*** haplo37- has joined #openstack-cinder22:14
patrickeastjgriffith: at least once before22:14
patrickeastjgriffith: and i'm sure we'll be fixing it again22:15
*** salv-orlando has quit IRC22:15
*** alonma has quit IRC22:15
patrickeasthemna: so like, that one session looks perfectly happy22:15
*** cdelatte has quit IRC22:16
*** alonma has joined #openstack-cinder22:18
*** thingee has joined #openstack-cinder22:22
*** Guest45789 is now known as med_22:22
*** Julien-zte has quit IRC22:22
*** med_ has quit IRC22:23
*** med_ has joined #openstack-cinder22:23
*** alonma has quit IRC22:23
*** salv-orlando has joined #openstack-cinder22:23
openstackgerritJohn Griffith proposed openstack/python-cinderclient: Add new attach/detach API calls  https://review.openstack.org/32740922:23
openstackgerritJohn Griffith proposed openstack/cinder: Add new attach/detach API calls  https://review.openstack.org/32740822:23
*** ccesario has joined #openstack-cinder22:24
*** mtanino has joined #openstack-cinder22:29
*** rooneym has quit IRC22:29
*** diablo_rojo has quit IRC22:32
*** jungleboyj__ has quit IRC22:37
*** salv-orlando has quit IRC22:46
*** frontrunner2 has quit IRC22:47
*** akerr_ has quit IRC22:48
*** sbezverk has joined #openstack-cinder22:50
*** sbezverk_ has quit IRC22:54
*** sdake has joined #openstack-cinder22:56
*** sdake_ has joined #openstack-cinder22:58
*** cknight1 has quit IRC23:00
*** nicolasbock has quit IRC23:01
*** sdake has quit IRC23:01
*** nicolasbock has joined #openstack-cinder23:02
*** sticker_ has joined #openstack-cinder23:10
*** sticker_ is now known as sticker23:11
*** markvoelker has quit IRC23:19
*** lprice has joined #openstack-cinder23:28
*** lprice1 has quit IRC23:28
*** leeantho has quit IRC23:33
*** gershon__ has quit IRC23:37
*** ducttape_ has joined #openstack-cinder23:41
*** ducttape_ has quit IRC23:44
*** EinstCrazy has joined #openstack-cinder23:47
*** cdelatte has joined #openstack-cinder23:48
*** EinstCrazy has quit IRC23:48
*** mtanino has quit IRC23:49
*** frontrunner has joined #openstack-cinder23:49
*** gershon_ has quit IRC23:50
*** delattec has quit IRC23:50
*** gershon has quit IRC23:51
*** salv-orlando has joined #openstack-cinder23:53
*** ducttape_ has joined #openstack-cinder23:57
*** salv-orlando has quit IRC23:58

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