Thursday, 2015-07-02

*** tsekiyam_ has joined #openstack-cinder00:02
*** Apoorva has quit IRC00:03
*** hemna is now known as hemnafk00:04
*** tsekiyama has quit IRC00:06
*** mtanino has quit IRC00:06
*** tsekiyam_ has quit IRC00:07
*** garthb_ has quit IRC00:07
*** bswartz has quit IRC00:18
*** Rockyg has quit IRC00:29
*** heyun has joined #openstack-cinder00:30
*** albertom has quit IRC00:32
*** Yogi1 has quit IRC00:34
*** vokt has quit IRC00:35
*** leeantho has quit IRC00:37
*** albertom has joined #openstack-cinder00:40
*** adurbin_ has quit IRC00:46
openstackgerritxing-yang proposed openstack/os-brick: Add connector driver for the ScaleIO cinder driver
*** xyang has quit IRC00:48
*** annashen has joined #openstack-cinder00:53
*** jdurgin has quit IRC00:56
*** annashen has quit IRC00:58
*** albertom has quit IRC01:02
*** dannywilson has quit IRC01:02
*** jasondotstar has joined #openstack-cinder01:02
*** blmartin has joined #openstack-cinder01:07
openstackgerritLi Yingjun proposed openstack/cinder: Validate maximum limit for quota
*** patrickeast has quit IRC01:09
*** albertom has joined #openstack-cinder01:09
openstackgerritLiu Xinguo proposed openstack/cinder: Add iSCSI multipath support for Huawei driver
*** laudo has quit IRC01:17
*** dims_ has quit IRC01:19
*** jwcroppe_ has quit IRC01:23
*** blmartin has quit IRC01:23
*** jasondotstar has quit IRC01:27
*** annegentle has quit IRC01:28
*** alau2 has quit IRC01:32
*** xyang has joined #openstack-cinder01:35
*** dims has joined #openstack-cinder01:39
*** ociuhandu has quit IRC01:46
*** mutoulbj has joined #openstack-cinder01:49
*** mutoulbj has quit IRC01:49
*** annashen has joined #openstack-cinder01:54
*** marcusvrn has joined #openstack-cinder01:55
*** blmartin has joined #openstack-cinder01:55
*** annashen has quit IRC01:58
openstackgerritliusheng proposed openstack/cinder: Switch to oslo.reports
*** jwcroppe has joined #openstack-cinder02:01
*** mutoulbj has joined #openstack-cinder02:14
*** jwang_ has joined #openstack-cinder02:18
*** marcusvrn has quit IRC02:19
openstackgerritVincent Hou proposed openstack/cinder: Implement the update_migrated_volume for the drivers
*** jwang__ has quit IRC02:21
*** rajinir has quit IRC02:25
*** xyang1 has quit IRC02:26
*** blmartin has quit IRC02:26
*** rajinir has joined #openstack-cinder02:27
*** rajinir has quit IRC02:29
*** rajinir has joined #openstack-cinder02:30
*** jaypipes has quit IRC02:30
*** bill_az has quit IRC02:33
*** Yogi1 has joined #openstack-cinder02:35
*** winston-d_zZZ is now known as winston-d02:36
*** blmartin has joined #openstack-cinder02:40
openstackgerritwanghao proposed openstack/cinder: Incremental backup improvements for L
openstackgerritKuo-tung Kao proposed openstack/python-cinderclient: Add more details for replication
*** blmartin has quit IRC02:50
*** blmartin has joined #openstack-cinder02:51
*** annashen has joined #openstack-cinder02:55
*** annashen has quit IRC02:59
openstackgerritTeruaki Ishizaki proposed openstack/cinder: Sheepdog: Add class for dog command executor
*** davechen has joined #openstack-cinder03:12
*** davechen1 has joined #openstack-cinder03:14
*** blmartin has quit IRC03:14
*** davechen has quit IRC03:17
*** Yogi1 has quit IRC03:18
*** dims has quit IRC03:22
*** harlowja_at_home has joined #openstack-cinder03:24
*** BharatK has joined #openstack-cinder03:41
*** coolsvap|away is now known as coolsvap03:45
*** sgotliv has joined #openstack-cinder03:52
*** annashen has joined #openstack-cinder03:56
*** annashen has quit IRC04:00
*** BharatK has quit IRC04:04
*** harlowja_at_home has quit IRC04:05
*** xyang has quit IRC04:11
*** BharatK has joined #openstack-cinder04:17
*** DericHorn-HP has joined #openstack-cinder04:19
*** heyun has quit IRC04:20
*** heyun has joined #openstack-cinder04:21
*** DericHorn-HP has quit IRC04:23
*** davechen1 has quit IRC04:27
openstackgerritVincent Hou proposed openstack/cinder-specs: Volume migration improvement for Liberty version
*** davechen has joined #openstack-cinder04:27
*** vincent_hou has joined #openstack-cinder04:28
vincent_houHi cinder folks, I have updated the spec for the volume migration improvement: Thank you for your comments.04:29
*** vilobhmm has joined #openstack-cinder04:29
*** ianbrown has joined #openstack-cinder04:30
*** yamada-h has joined #openstack-cinder04:41
*** rajinir has quit IRC04:43
*** delattec has joined #openstack-cinder04:43
*** jwcroppe_ has joined #openstack-cinder04:43
*** cdelatte has quit IRC04:45
*** jwcroppe has quit IRC04:46
*** nkrinner has joined #openstack-cinder04:54
*** BharatK has quit IRC04:54
*** harlowja_at_home has joined #openstack-cinder04:54
*** dobson has quit IRC04:55
*** annashen has joined #openstack-cinder04:56
*** amit213 has joined #openstack-cinder05:00
*** annashen has quit IRC05:01
*** dobson has joined #openstack-cinder05:02
*** BharatK has joined #openstack-cinder05:08
*** deepakcs has joined #openstack-cinder05:11
*** BharatK has quit IRC05:12
*** vilobhmm has quit IRC05:13
*** vilobhmm has joined #openstack-cinder05:13
*** shausy has joined #openstack-cinder05:15
*** DericHorn-HP has joined #openstack-cinder05:18
*** yamada-h_ has joined #openstack-cinder05:21
*** yamada-h has quit IRC05:24
*** harlowja_at_home has quit IRC05:27
*** BharatK has joined #openstack-cinder05:29
*** coolsvap is now known as coolsvap|away05:37
*** coolsvap|away is now known as coolsvap05:38
*** nihilifer has joined #openstack-cinder05:40
*** DericHorn-HP has quit IRC05:45
*** Longgeek has joined #openstack-cinder05:52
*** deepakcs has quit IRC05:53
*** deepakcs has joined #openstack-cinder05:53
*** coolsvap is now known as coolsvap|away05:54
*** annashen has joined #openstack-cinder05:57
*** annashen has quit IRC06:02
*** coolsvap|away is now known as coolsvap06:02
*** Vikash_cz has joined #openstack-cinder06:04
*** anshul has joined #openstack-cinder06:13
*** vincent_hou_ has joined #openstack-cinder06:15
openstackgerritwanghao proposed openstack/cinder: Incremental backup improvements for L
*** coolsvap is now known as coolsvap|away06:16
*** vincent_hou has quit IRC06:17
*** vincent_hou_ is now known as vincent_hou06:17
openstackgerritTeruaki Ishizaki proposed openstack/cinder: Sheepdog: change create and delete operation
*** ianbrown has quit IRC06:29
*** belmoreira has joined #openstack-cinder06:44
*** vilobhmm has quit IRC06:45
*** belmoreira has quit IRC06:47
*** annashen has joined #openstack-cinder06:58
*** belmoreira has joined #openstack-cinder06:58
*** belmoreira has quit IRC06:58
*** belmoreira has joined #openstack-cinder06:59
*** annashen has quit IRC07:03
*** fifieldt has quit IRC07:09
*** yamada-h_ has quit IRC07:09
*** yamada-h has joined #openstack-cinder07:12
*** coolsvap|away has quit IRC07:20
openstackgerritMichal Dulko proposed openstack/cinder: Service object
*** coolsvap|away has joined #openstack-cinder07:23
*** aix has joined #openstack-cinder07:24
*** mutoulbj_ has joined #openstack-cinder07:32
*** mutoulbj has quit IRC07:34
*** yamada-h has quit IRC07:44
*** jistr has joined #openstack-cinder07:48
*** pck has quit IRC07:50
*** pck has joined #openstack-cinder07:50
davechenwinston-d: zhiteng, ping?07:50
openstackgerritLin Yang proposed openstack/python-cinderclient: Fix typo in comment message
davechenDuncanT: ping?07:52
DuncanTdavechen: hi07:53
davechenhi, Duncan :)07:53
davechennice to see you are still online.07:53
davechenah, can you help to review that patch again?07:54
davechen, this is link07:55
davechenSean and winston review that patch and give some comments recently.07:56
openstackgerritLin Yang proposed openstack/python-cinderclient: Improve error message when cinder quota exceeded
davechenSeems Sean are happy with that patch, he has voted +2 on that patch couple of days ago.07:57
*** bluex-pl has joined #openstack-cinder07:57
*** bluex-pl has quit IRC07:58
*** bluex-pl has joined #openstack-cinder07:58
*** annashen has joined #openstack-cinder07:59
*** afazekas has joined #openstack-cinder08:01
*** annashen has quit IRC08:04
*** jordanP has joined #openstack-cinder08:06
*** dosaboy has quit IRC08:10
*** dosaboy has joined #openstack-cinder08:11
*** dosaboy has quit IRC08:12
*** rwsu has quit IRC08:13
*** rwsu has joined #openstack-cinder08:16
openstackgerritwanghao proposed openstack/cinder: Incremental backup improvements for L
*** jamielennox is now known as jamielennox|away08:20
*** dosaboy has joined #openstack-cinder08:21
*** msno has joined #openstack-cinder08:28
msnocan i have enabled_backends = lvm-1,netapp_iscsi in cinder.conf to support multi backend08:29
msnois it supported or should i just have enabled_backends = netapp_iscsi to get netapp iscsi working08:30
msnoby default it has enabled_backends = lvm-108:30
*** vincent_hou_ has joined #openstack-cinder08:37
msnois there a netapp expert here in the room08:38
*** vincent_hou has quit IRC08:39
*** vincent_hou_ is now known as vincent_hou08:39
*** jamielennox|away is now known as jamielennox08:41
*** ankit_ag has joined #openstack-cinder08:43
*** laughterwym has quit IRC08:49
*** aix has quit IRC08:50
*** Longgeek has quit IRC08:52
*** Longgeek has joined #openstack-cinder08:53
*** afazekas has quit IRC08:54
openstackgerritGaurang Tapase proposed openstack/cinder: Adding NFS support to the GPFS Driver
*** vincent_hou has quit IRC08:59
*** annashen has joined #openstack-cinder09:00
DuncanTmsno: If you don't want to use the LVM backend as well, then take it out. If you want to use both, then that is supported09:01
msnois this the right forum to ask cinder conf issues wrt netapp storage09:01
DuncanTmsno: Yes, it is the right place to ask, however people can be slow to reply - we are all working on other things09:02
msnook.. when i give both .. my tempest test fails on test_volume_types.VolumeTypesV1Test.test_create_get_delete_volume_with_volume_type_and_extra_specs09:02
*** ndipanov has joined #openstack-cinder09:02
msnowhen i take out netapp_iscsi this test passes09:02
DuncanTWhat about with just netapp_iscsi?09:04
*** annashen has quit IRC09:04
msnosame result test_volume_types.VolumeTypesV1Test.test_create_get_delete_volume_with_volume_type_and_extra_specs fails09:06
msnoso i guess.. netapp_iscsi is not able to pass that extra_specs checks09:07
*** afazekas has joined #openstack-cinder09:09
*** gaurangt has joined #openstack-cinder09:10
openstackgerritPradeep Sathasivam proposed openstack/cinder: Adds friendly zone name support
*** BharatK has quit IRC09:16
msnoanyone has experience in working with netapp storage + openstack can give me a pointer09:17
*** openstackgerrit has quit IRC09:19
*** openstackgerrit has joined #openstack-cinder09:20
*** jasondotstar has joined #openstack-cinder09:22
*** afazekas has quit IRC09:24
*** dims has joined #openstack-cinder09:24
*** BharatK has joined #openstack-cinder09:28
*** aix has joined #openstack-cinder09:34
*** e0ne has joined #openstack-cinder09:35
*** lixiaoy1 has quit IRC09:39
*** afazekas has joined #openstack-cinder09:39
*** e0ne is now known as e0ne_09:46
*** BharatK has quit IRC09:46
*** davechen has left #openstack-cinder09:55
*** Swanson has quit IRC09:57
*** Swanson has joined #openstack-cinder09:58
*** dtynan has quit IRC10:00
*** dtynan has joined #openstack-cinder10:00
*** annashen has joined #openstack-cinder10:01
*** zerda has joined #openstack-cinder10:03
*** shyama has quit IRC10:03
*** yamada-h has joined #openstack-cinder10:04
*** e0ne_ is now known as e0ne10:05
*** annashen has quit IRC10:06
*** yamada-h has quit IRC10:08
*** ociuhandu has joined #openstack-cinder10:12
*** BharatK has joined #openstack-cinder10:14
*** IanGovett has joined #openstack-cinder10:16
*** ociuhandu has quit IRC10:19
*** mutoulbj_ has quit IRC10:19
*** mutoulbj has joined #openstack-cinder10:20
*** laughterwym has joined #openstack-cinder10:21
*** mutoulbj has quit IRC10:25
*** ociuhandu has joined #openstack-cinder10:27
*** BharatK has quit IRC10:34
*** BharatK has joined #openstack-cinder10:34
*** rushiagr_away is now known as rushiagr10:42
*** shyama has joined #openstack-cinder10:42
*** lpetrut has joined #openstack-cinder10:43
*** hakimo has joined #openstack-cinder10:55
*** annashen has joined #openstack-cinder11:02
openstackgerritTeruaki Ishizaki proposed openstack/cinder: Sheepdog: change create and delete operation
*** heyun has quit IRC11:04
*** ronis has joined #openstack-cinder11:05
*** annashen has quit IRC11:06
*** lpetrut has quit IRC11:08
*** gaurangt has left #openstack-cinder11:11
openstackgerritTeruaki Ishizaki proposed openstack/cinder: Sheepdog: change create and delete operation
*** BharatK has quit IRC11:14
*** BharatK has joined #openstack-cinder11:14
*** mutoulbj has joined #openstack-cinder11:14
*** Apoorva has joined #openstack-cinder11:19
geguileoDuncanT: ping11:22
DuncanTgeguileo: hi11:23
geguileoDuncanT: Hi11:23
geguileoDuncanT: Yesterday you mentioned that we accept combinations of commands that queue on the lock on volume manager11:23
geguileoDuncanT: I don't know anything about that  XD11:23
geguileoDuncanT: Could you point me to where in the code can I see an example, please?11:24
DuncanTgeguileo: It is emergent behaviour. e.g. do a long running operation that locks on the volume, then delete the volume. Both operations complete in order11:26
geguileoOk, I see what you mean11:27
geguileoYes, the lock on the volume will make the delete wait11:27
DuncanTAnd there are client scripts based around that behaviour11:27
geguileoOk, I thought you meant we actually had a queue in Cinder and I was surprised  :)11:27
*** lpetrut has joined #openstack-cinder11:28
geguileoThen since we cannot break API contract11:28
geguileoOur options are use DLM or create new API, right?11:29
DuncanTgeguileo: Exactly :-(11:29
geguileoOk, that sucks but it's logical11:30
*** deepakcs has quit IRC11:30
geguileohemnafk: Was working on removing locks from the manager, right?11:30
geguileoSo how is he going to remove them?11:31
geguileoSince we cannot remove them without changing behavior?11:31
*** lpabon has joined #openstack-cinder11:33
openstackgerritVictor Stinner proposed openstack/cinder: Port image/ to Python 3
*** e0ne is now known as e0ne_11:34
geguileoDuncanT: ^ (not the patch, but my last comment ;))11:35
*** e0ne_ is now known as e0ne11:36
DuncanTgeguileo: There not yet any good answer to that question11:36
geguileoDuncanT: Ok, I'll bother him on that one  :)11:36
geguileoDuncanT: Thanks11:38
DuncanTYou're welcome11:38
*** Lee1092 has joined #openstack-cinder11:39
*** Apoorva has quit IRC11:42
openstackgerritVictor Stinner proposed openstack/cinder: Fix Python 3 issues in cmd
*** marcusvrn has joined #openstack-cinder11:44
*** jasondotstar has quit IRC11:50
*** Apoorva has joined #openstack-cinder11:55
*** zhenguo has quit IRC12:01
*** ganso_ has joined #openstack-cinder12:02
*** annashen has joined #openstack-cinder12:02
*** e0ne is now known as e0ne_12:06
*** annashen has quit IRC12:07
*** dtynan has quit IRC12:07
*** dtynan has joined #openstack-cinder12:08
*** shyama_ has joined #openstack-cinder12:09
openstackgerritAndrey Pavlov proposed openstack/cinder: Avoid race condition at snapshot deletion stage
*** shyama has quit IRC12:09
*** shyama_ is now known as shyama12:10
*** p_deore has joined #openstack-cinder12:11
*** e0ne_ is now known as e0ne12:11
openstackgerritwanghao proposed openstack/cinder: Validate value when user update quota
openstackgerritAndrey Pavlov proposed openstack/cinder: Avoid race condition at snapshot deletion stage
openstackgerritAndrey Pavlov proposed openstack/cinder: Avoid race condition at snapshot deletion stage
*** bill_az has joined #openstack-cinder12:17
*** julim has joined #openstack-cinder12:22
*** marcusvrn has quit IRC12:23
*** zerda has quit IRC12:24
*** marcusvrn has joined #openstack-cinder12:25
*** julim has quit IRC12:27
*** Apoorva has quit IRC12:29
*** jaypipes has joined #openstack-cinder12:32
*** annegentle has joined #openstack-cinder12:34
*** BharatK has quit IRC12:35
*** BharatK has joined #openstack-cinder12:36
*** jasondotstar has joined #openstack-cinder12:37
*** nkrinner has quit IRC12:38
*** edmondsw has joined #openstack-cinder12:42
*** marcusvrn has quit IRC12:43
*** lpetrut has left #openstack-cinder12:46
*** lpetrut has joined #openstack-cinder12:46
openstackgerritAndrey Pavlov proposed openstack/cinder: Avoid race condition at snapshot deletion stage
*** amoturi has quit IRC12:53
*** amoturi has joined #openstack-cinder12:53
*** gaurangt has joined #openstack-cinder12:57
*** annashen has joined #openstack-cinder13:03
*** mutoulbj has quit IRC13:04
*** marcusvrn has joined #openstack-cinder13:05
*** annashen has quit IRC13:08
*** julim has joined #openstack-cinder13:15
*** zhenguo has joined #openstack-cinder13:16
*** ankit_ag has quit IRC13:23
*** thangp has joined #openstack-cinder13:23
*** mriedem has joined #openstack-cinder13:25
*** rmesta has joined #openstack-cinder13:26
*** e0ne is now known as e0ne_13:29
*** mutoulbj has joined #openstack-cinder13:35
*** r-daneel has joined #openstack-cinder13:36
*** rmesta has left #openstack-cinder13:38
*** e0ne_ is now known as e0ne13:43
*** nihilifer has quit IRC13:45
*** fifieldt has joined #openstack-cinder13:53
*** amoturi_ has joined #openstack-cinder14:00
*** Apoorva has joined #openstack-cinder14:03
*** amoturi has quit IRC14:04
*** annegentle has quit IRC14:04
*** annashen has joined #openstack-cinder14:04
*** lcurtis has joined #openstack-cinder14:04
*** anshul has quit IRC14:08
*** blmartin has joined #openstack-cinder14:08
*** bluex-pl has quit IRC14:09
*** annashen has quit IRC14:09
*** Apoorva has quit IRC14:11
*** rmesta has joined #openstack-cinder14:14
*** Zhongjun has joined #openstack-cinder14:16
openstackgerritVincent Hou proposed openstack/cinder: Implement the update_migrated_volume for the drivers
*** rmesta has left #openstack-cinder14:16
*** BharatK has quit IRC14:16
*** dims has quit IRC14:19
openstackgerritAdriano Freires Rosso proposed openstack/cinder: Adds manage/unmanage methods for HNAS drivers
*** aix has quit IRC14:21
openstackgerritDave Chen proposed openstack/cinder: Role based properties protection
openstackgerritDave Chen proposed openstack/cinder: Policies based properties protection
*** MentalRay has joined #openstack-cinder14:22
*** jungleboyj has joined #openstack-cinder14:22
*** BharatK has joined #openstack-cinder14:24
openstackgerritAdriano Freires Rosso proposed openstack/cinder: Adds manage/unmanage methods for HNAS drivers
*** erlon has joined #openstack-cinder14:27
*** Vikash_cz has quit IRC14:28
*** eharney has quit IRC14:30
*** e0ne is now known as e0ne_14:30
openstackgerritDaniel Tadrzak proposed openstack/cinder: CGSnapshot Object
openstackgerritDaniel Tadrzak proposed openstack/cinder: ConsistencyGroup Object
*** e0ne_ is now known as e0ne14:32
jungleboyjjgriffith: You around?14:35
*** dims has joined #openstack-cinder14:38
openstackgerritIvan Kolodyazhny proposed openstack/cinder: Pass volume size GB to copy_volume function
*** annegentle has joined #openstack-cinder14:41
openstackgerritIvan Kolodyazhny proposed openstack/cinder: Pass volume size in GB to copy_volume function
*** dims has quit IRC14:44
*** dims has joined #openstack-cinder14:49
*** mutoulbj has quit IRC14:51
*** mutoulbj has joined #openstack-cinder14:52
*** mtanino has joined #openstack-cinder14:52
*** ronis has quit IRC14:56
*** xyang has joined #openstack-cinder14:59
*** bswartz has joined #openstack-cinder15:00
*** annashen has joined #openstack-cinder15:05
*** dims has quit IRC15:06
*** jskinner has joined #openstack-cinder15:06
*** Guest42148 has joined #openstack-cinder15:07
*** afazekas has quit IRC15:08
*** Guest42148 is now known as dims_15:09
*** annashen has quit IRC15:10
*** belmoreira has quit IRC15:10
*** amoturi_ has quit IRC15:14
*** asselin_ has joined #openstack-cinder15:14
*** harlowja_at_home has joined #openstack-cinder15:14
*** MentalRay has quit IRC15:15
*** DericHorn-HP has joined #openstack-cinder15:18
*** chlong has quit IRC15:23
*** hemna has joined #openstack-cinder15:26
*** hemna has quit IRC15:29
*** hemna has joined #openstack-cinder15:30
*** hemna has quit IRC15:30
*** tpsilva has joined #openstack-cinder15:36
*** rakesh_mishra has joined #openstack-cinder15:44
rakesh_mishracan any one help me please,15:45
*** marcusvrn has quit IRC15:45
rakesh_mishrai want to block cinder API's how can i do it?15:46
smcginnisrakesh_mishra: Not sure I understand what you mean. Can you elaborate?15:46
rakesh_mishrai want to disable some API15:46
smcginnisrakesh_mishra: Only some API calls but still have others accessible?15:47
e0nerakesh_mishra: did you try policy.json?15:47
rakesh_mishralike cinder list should not work but cinder create should work15:47
rakesh_mishra"volume:create": "!"15:48
rakesh_mishrait disable the volume create api15:48
*** laughterwym has quit IRC15:49
rakesh_mishrabut in def get_all(self, context, marker=None, limit=None, sort_keys=None,15:49
rakesh_mishra                sort_dirs=None, filters=None, viewable_admin_meta=False):15:49
rakesh_mishra        check_policy(context, 'get_all')15:49
rakesh_mishrawe are checking the policy , that's why it doesn't allow the request15:50
*** vokt has joined #openstack-cinder15:50
*** bswartz has quit IRC15:50
rakesh_mishrabut i don't want to change the source code for other API where we are not using check_policy15:51
rakesh_mishrabut i want to disable the API15:51
rakesh_mishra@smcginnis, @e0ne : is there any way , i can give a list of api which should be support and rest of the API should disable15:53
rakesh_mishraor is there any way to disable the API from keystone?15:54
*** vokt has quit IRC16:00
*** asselin_ has quit IRC16:02
*** jistr has quit IRC16:02
*** dannywilson has joined #openstack-cinder16:03
*** annashen has joined #openstack-cinder16:06
*** leeantho has joined #openstack-cinder16:07
smcginnisrakesh_mishra: Sorry, pulled away.16:08
*** __afazekas is now known as afazekas16:08
smcginnisrakesh_mishra: Other than setting policy I don't know of another way.16:08
smcginnisrakesh_mishra: I suppose you can customize the code if you really really needed to, but... ick.16:09
smcginnisSorry, gotta run again.16:09
*** p_deore has quit IRC16:09
*** annashen has quit IRC16:10
*** leeantho has quit IRC16:11
*** ndipanov has quit IRC16:12
*** MentalRay has joined #openstack-cinder16:13
*** dannywilson has quit IRC16:14
*** bill_az has quit IRC16:15
*** _cjones_ has joined #openstack-cinder16:18
*** jordanP has quit IRC16:19
*** alau2 has joined #openstack-cinder16:21
*** annashen has joined #openstack-cinder16:26
openstackgerritGorka Eguileor proposed openstack/cinder: Fix backup metadata import missing fields
openstackgerritGorka Eguileor proposed openstack/cinder: Fix saving tz aware datetimes in Versioned Objects
*** eharney has joined #openstack-cinder16:31
*** MentalRay has quit IRC16:31
haypohi. do you plan to release a new version of os-brick including hemnafk fixes for Python3?16:32
e0neeharney: hi. did you have a time to review my patch?16:34
eharneye0ne: not yet16:34
e0neok, thanks16:35
*** hemnafk is now known as hemna16:38
*** MentalRay has joined #openstack-cinder16:38
nikeshmhi, i am going back to INDIA today, so will be available on monday IST, in case any query, please send me email :)16:40
*** garthb has joined #openstack-cinder16:42
*** annashen has quit IRC16:42
*** annashen has joined #openstack-cinder16:43
*** annashen has quit IRC16:43
*** annashen has joined #openstack-cinder16:44
nikeshmrakesh_mishra : better to use "" for posting code  or contents of a file or traceback etc :)16:45
rmstarhi guys.  anyone here for a quick question? :)16:47
e0nermstar: just ask it:)16:49
*** tsekiyama has joined #openstack-cinder16:50
*** msno has quit IRC16:51
*** shausy has quit IRC16:53
*** storagesujai has joined #openstack-cinder16:54
*** patrickeast has joined #openstack-cinder16:55
*** ronis has joined #openstack-cinder16:59
*** rakesh_mishra has quit IRC16:59
*** BharatK has quit IRC17:01
*** annashen has quit IRC17:02
*** annashen has joined #openstack-cinder17:03
*** blmartin has quit IRC17:03
*** blmartin has joined #openstack-cinder17:03
*** annashen has quit IRC17:08
*** Zhongjun has quit IRC17:08
*** gaurangt1 has joined #openstack-cinder17:08
*** leeantho has joined #openstack-cinder17:10
*** garthb has quit IRC17:10
*** gaurangt has quit IRC17:10
*** Adriano__ has joined #openstack-cinder17:11
*** e0ne has quit IRC17:11
* thingee is so behind on irc backlog17:13
thingeemy favorite from yesterday's discussion on DLM and other random stuff was...17:13
thingee"nova cell is optional, but people get tricked into using it and hard to get away from it."17:13
harlowja_at_homethingee, where did u read that17:17
* harlowja_at_home i still honestly believe cells is just a hack-fix for bigger fundamental issues17:19
*** jasondotstar is now known as jasondotstar|afk17:19
thingeejgriffith: still catching up from yesterday's discussion... if we can write a sort-of-simple locks library ;)17:23
*** BharatK has joined #openstack-cinder17:23
thingeeharlowja_at_home: in this channel yesterday at 17:35 utc17:23
thingee17:36 rather17:23
harlowja_at_homethingee, ya, found it, interesting :-p17:24
harlowja_at_homesort-of-simple-locks library :-P17:24
harlowja_at_homesounds like tooz, ha17:24
harlowja_at_homehow much simpler u want it :-P17:24
patrickeastsort of simpler?17:24
harlowja_at_homewith a side of veggies17:24
thingeeI want options man. I want just in middle simple.17:25
harlowja_at_homehow would u like your french fries sir17:25
harlowja_at_homeis the poriddge just right, or to hot17:25
thingeeplain, but a side of wiiiiiiine please17:25
harlowja_at_homeok, coming up17:26
*** MentalRay has quit IRC17:26
Adriano__Hi guys... Is there any policy about reusing code from patches that aren't merged yet??17:26
geguileoAdriano__: You mean your patch depends on that other patch?17:27
patrickeastdependant patches are A-ok, duplicating the code… less good17:27
harlowja_at_homeoh ya, jgriffith pulled out the lipstick on a pig, nice17:27
Adriano__no.. I just reused a part of the code17:28
Adriano__but got a -1 asking about copyright issues or something like this17:28
*** bswartz has joined #openstack-cinder17:29
geguileoAdriano__: No copyright issues17:29
*** eharney has quit IRC17:29
geguileoAdriano__: AFAIK17:30
Adriano__geguileo: tks17:30
patrickeastyea i mean, you could make it a dependant patch so you both don’t add in that util change, otherwise someone would get a conflict17:30
harlowja_at_home' 'cinder++' blue sky codebase'17:30
harlowja_at_homeuh oh17:30
harlowja_at_homeman this is a good one thingee :-P17:30
geguileoAdriano__: Usually do what patrickeast says17:30
geguileoAdriano__: We don't want the same code repeated all over the place17:31
*** garthb has joined #openstack-cinder17:31
Adriano__geguileo: I see17:31
Adriano__geguileo: but I needed to implement complementary functions in our driver and the common code is only in cinder/utils now17:34
geguileoAdriano__: You mean that you only need some stuff from cinder/utils that the other patch has?17:35
Adriano__geguileo: would it be dependant only because this cinder/utils?17:35
Adriano__geguileo: yes17:35
Adriano__geguileo: and it's replicated in my patch17:35
*** anshul has joined #openstack-cinder17:35
geguileoAdriano__: You should add the other as dependent, unless we are talking of 3 lines of code17:36
*** anshul has quit IRC17:36
geguileoAdriano__: You are getting "credit" for his code, and imagine your patch merges before his does17:36
Adriano__geguileo: hmmm.. I see17:37
geguileoAdriano__: He will have a conflict, and imagine how you would feel if you got a conflict in your patch with somebody else's patch and you see that the conflict is your own code in other patch ;-)17:37
Adriano__geguileo: Actually, it seems that the other patch is stopped for some time17:37
Adriano__geguileo: But I agree17:37
geguileoAdriano__: Let me see it17:37
Adriano__geguileo: I will do that17:38
*** storagesujai has quit IRC17:38
geguileoAdriano__: Ok, 3 months without any update on the patch17:39
*** albertom has quit IRC17:39
*** annashen has joined #openstack-cinder17:39
*** garthb has quit IRC17:39
geguileoAdriano__: Don't depend on that patch  XD17:40
geguileoAdriano__: If you want to be polite drop him a line as a comment in his patch letting him know you'll be using that code17:40
Adriano__ geguileo17:40
geguileoAdriano__: np17:40
Adriano__ geguileo: Anyway, I'll try to contact someone involved in the other patch to talk about later :)17:41
geguileoAdriano__: Extra polite, that's always good  ;-)17:43
geguileohemna: ping - removing manager locks17:43
harlowja_at_homehemna, +1 on the 'ing' stuff, that'd really help17:44
*** jasondotstar|afk has quit IRC17:45
harlowja_at_homei mean make a cinder version of and start using it...17:45
*** eharney has joined #openstack-cinder17:45
harlowja_at_homeget the 80% benefit, andddd then fix another 5% and ...17:45
*** albertom has joined #openstack-cinder17:47
geguileohemna: Got a minute?17:47
geguileohemna: You are working on removing the locks from the manager, right?17:47
hemnaI'm not currently17:48
hemnanova has to change first17:48
geguileohemna: I know, but I was told you were working on that  ;-)17:48
hemnawe had done research on doing it last release, and found that it broke Nova badly.17:49
hemnaso Nova has to change in every call to Cinder to expect a VolumeIsBusy exception and gracefully handle that.17:49
geguileoOk, so you haven't kept working on it17:49
hemnathen Cinder's API can change to do the 'ing' checks on every call, and return VolumeIsBusy17:49
hemnanah, I haven't had time to17:49
hemnaI've been slammed with os-brick work in Nova, and various other things17:50
geguileoBut then you would change the API contract17:50
hemnait has to change17:50
geguileoYeah, I've been seening lots of os-brick stauff17:50
geguileohemna: But rules say that we cannot change the API that way, right?17:50
geguileoXD XD XD17:51
hemnabut anyway, we need a nova-spec laying it out17:51
hemnaand then a BP17:51
hemnaand then a patch17:51
hemnago through the normal process17:51
geguileoThose rules  ;-)17:51
geguileoSo our options as far as I can see are use DLM locks :-(    or new API version  :''-(17:52
hemnaI'm not so sure about that17:53
hemnait's just catching an exception17:53
geguileoBecause right now with those locks some clients may be queuing operations17:53
hemnawhich nova can get exceptions coming back already17:53
hemnatimeouts, etc.17:53
hemnait's just catching a specific one17:54
geguileoYes, I see the nova fix17:54
hemnait's a grey area IMHO17:54
geguileoAnd cinderclient can also be changed17:54
geguileoBut other clients could be queuing operations based on current API behavior, right?17:54
geguileoOr scripts using cinderclient could be working based on this behaviour as well17:55
* harlowja_at_home really VolumeIsBusy (imho) should of been there from the start (its really impossible to guarantee 100% of the time that a volume will not be busy)17:55
hemnaharlowja_at_home, +117:55
geguileoharlowja_at_home: The problem is that now we have "contracts" we must honor17:55
geguileoWe cannot go around changing things because they were not done as they should in the first time17:56
hemnaI dunno if it's a contract change though17:56
harlowja_at_homegeguileo, understood, i get that17:56
harlowja_at_homeisn't there a http code for this, one of those standard ones17:56
geguileohemna: I think it is a contract change17:57
patrickeastasselin: oo just saw emails from the hp ci for the fc-passthrough, thats awesome!17:57
harlowja_at_homemaybe next time, say in contract, 'we reserve the right to use any http code that we didn't use'17:57
geguileoSince most of the functionallity is not documented we rely on implicit contracts17:57
geguileoA lot of the time17:57
harlowja_at_homeand then have little signature line that u get notarized +17:57
geguileoSo that would be one of those implicit contracts :(17:57
openstackgerritJohn Griffith proposed openstack/python-cinderclient: Revert "Enable version discovery"
harlowja_at_homeya, implict stuff sucks17:58
geguileoharlowja_at_home: But we don't love documenting either  XD17:58
harlowja_at_homeand thats how u end up with perl, lol17:58
geguileoXD XD XD17:59
geguileohemna: So to summarize, we must decide if that implicit contract should be honored or not17:59
geguileoI think I'll send an email asking people to vote on this17:59
geguileoSo we can really know how to go from here18:00
*** ronis has quit IRC18:02
*** logan2 has quit IRC18:03
rmstarok.  thanks eOne :D18:05
rmstari am installing Juno, and I am at the part where i am installing storage node.  After going through all the steps, should i see a new partition for the LVM?18:05
*** harlowja_at_home has quit IRC18:06
thingeegeguileo: you might want to sync up with winston-d... he's currently tasked with the nova changes to expect VolumeIsBusy.18:08
thingeegeguileo: he has been busy though, so it would be good if someone can make that move forward18:09
* thingee finds bug18:09
geguileothingee: Ok, I'll ping him to see how he's doing with that18:09
geguileothingee: But I think I should send the email to confirm that we agree on the API change on Cinder side18:10
geguileothingee: It would be terrible if he keeps working on that and then we decide that we won't do the Cinder change because it breaks API contract18:10
*** MentalRay has joined #openstack-cinder18:11
*** ronis has joined #openstack-cinder18:16
*** gaurangt1 has quit IRC18:16
*** annashen has quit IRC18:19
geguileowinston-d: ping - VolumeIsBusy stuff18:19
*** annashen has joined #openstack-cinder18:20
asselinpatrickeast, yeah! I have an intern learning ci and set everything up18:20
asselinhe's not on freenode yet, but will be hopefully today18:20
*** BharatK has quit IRC18:20
asselinpatrickeast, ok I'll review the patch now....thanks for being patient. :)18:22
patrickeastasselin: haha, no worries, i’m happy to see that stuff getting tested18:23
patrickeastasselin: i think so far i’m the only one who has needed those changes anyway18:23
*** mutoulbj has quit IRC18:24
*** annashen has quit IRC18:24
hemnageguileo, so I think step 1) is to create a nova-spec18:24
hemnadetail out the changes18:24
hemnathen do a ML post asking for feedback and discussion and point to the nova-spec18:25
hemnaand detail out why it's needed, and how Cinder is broken, because of the lack of it, etc.18:25
hemnaI can help if you like.18:25
hemnaI'd be championing this one if I had time18:25
*** rushiagr is now known as rushiagr_away18:26
geguileohemna: But the nova spec would made no sense if we are against changing Cinder API contract, right?18:27
hemnawrite the spec, get the discussion started and buyin by all parties interested18:27
hemnathat's how I'd approach it18:27
hemnawe'll get feedback on the contract in the ML and the spec itself.18:28
hemnaand figure out how to proceed18:28
geguileohemna: So we work on the nova-specs, winston keeps working on the patch and then we decide we don't want to change the Cinder side...18:28
hemnaif we don't start with the detailed spec, then there is no way we can move forward.18:28
hemnawe have to change the cinder side IMO18:28
geguileohemna: But the problem right now is not the contract with Nova18:28
*** earlephilhower has quit IRC18:28
geguileohemna: Is with other clients outside our control18:28
hemnathe entire point of the nova changes is so we can change the cinder side and remove the locks18:29
hemnawe can't stay broken forever because of 'rules'18:29
geguileohemna: But here are people who thing that we cannot change Cinder that way18:29
geguileohemna: Anyway, I'll send the email with just the part of changing Cinder18:29
geguileoAnd start working on the specs, which will take me longer18:30
hemnadon't send the email about the Cinder change without first discussing the nova changes.18:30
hemnaor it makes no sense.18:30
hemnaanyway, I've said my $0.02.18:30
geguileohemna: You mean discuss the nova changes in a spec? With winston? With nova guys?18:31
hemnaall of the above18:31
hemnaI've stated my case pretty clearly.  not sure where the confusion is.18:31
geguileohemna: Ok, I'll talk with winston to see where he's at18:31
*** r-daneel has quit IRC18:32
geguileohemna: Anyway, Nova specs are closed for L (afaik), so that's a no go for this cycle18:33
*** zhenguo has quit IRC18:33
hemnathe spec's purpose is to detail and layout the changes18:33
hemnaat this point, who cares if it can't make it in L18:34
hemnaat least the spec is available for folks to reference and discuss.18:34
*** shyama has quit IRC18:38
*** Longgeek has quit IRC18:43
*** Longgeek has joined #openstack-cinder18:43
*** ronis has quit IRC18:44
*** e0ne has joined #openstack-cinder18:46
*** julim has quit IRC18:47
*** logan2 has joined #openstack-cinder18:49
*** r-daneel has joined #openstack-cinder18:50
*** dannywilson has joined #openstack-cinder18:50
*** VManiac16 has joined #openstack-cinder18:53
*** blmartin has quit IRC18:54
*** Longgeek has quit IRC18:54
*** dannywilson has quit IRC18:56
*** dannywilson has joined #openstack-cinder18:56
tbarronAdriano__: just take what you need from that other patch
tbarronAdriano__: Don't make your patch dependent on it.18:58
tbarronAdriano__: it is blocked waiting on NFS snapshots, which may still be a while.18:59
tbarronAdriano__: we went on and implemented manage/unmanage directly in NetApp NFS drivers instead of in generic NFS18:59
tbarronbecause of this blockage.18:59
tbarronWhen NFS snapshots get implmemted in the generic driver we can re-visit, implement manage-unmanage in generic NFS driver, and remove some code from our driver - and presumably yours too.19:00
*** marcusvrn has joined #openstack-cinder19:01
*** Longgeek has joined #openstack-cinder19:02
Adriano__tbarron: Ok.. Tks for the info :)19:02
tbarronmsno: just catching up on backlog here.  NetApp is closed this week so we're slow responding, but ...19:02
tbarronAdriano__: np19:03
marcusvrnhi all, an oslo change broke my driver...I posted a patch to fix it, if someone can review it to fix the driver and turn the CI on again, would be great:
tbarronmsno: you *can* certainly run with enabled_backends = lvm-1,netapp_iscsi in cinder.conf to support multi backend19:03
tbarronmsno: for volume_type tempest tests modify this line in /etc/tempest;19:04
*** sp4wnr0ot_ has joined #openstack-cinder19:05
tbarronmwno: # Backend vendor to target when creating volume types (string value)19:05
tbarronmwno: vendor_name = NetApp19:05
tbarronmsno: ^^19:05
e0nemarcusvrn: did your CI find this issue?19:06
*** tbarron is now known as tbarron_afk19:07
e0nemarcusvrn: imo, it would be good to have such issue tested by 3rd-party CI and unit-tests19:07
marcusvrne0ne: yes, actually after the oslo update it stopped19:07
marcusvrnI ran the CI on this patch19:08
e0nemarcusvrn: stopped? i thought it should just fail on every commit19:08
marcusvrnit's the HBSD2 CI19:08
marcusvrne0ne: sry,'s failling on every commit19:08
e0nemarcusvrn: thanks19:10
*** Longgeek has quit IRC19:14
*** marcusvrn_ has joined #openstack-cinder19:14
*** lpetrut has quit IRC19:16
*** marcusvrn has quit IRC19:17
*** marcusvrn_ is now known as marcusvrn19:17
*** annashen has joined #openstack-cinder19:20
*** nikeshm has quit IRC19:20
*** annashen has quit IRC19:25
*** lpetrut has joined #openstack-cinder19:31
*** delatte has joined #openstack-cinder19:34
*** lpetrut has quit IRC19:34
jgriffiththingee: thingee jgriffith: in reply to your unit test for check methods implemented by drivers, wouldn't CI catch this from ABC eventually? The volume manager would be unable to load due a method not being implemented.19:37
jgriffiththingee: oops...NM19:37
*** delattec has quit IRC19:37
*** Lee1092 has quit IRC19:45
*** igor___ has joined #openstack-cinder19:48
*** ociuhandu has quit IRC19:49
*** jasondotstar|afk has joined #openstack-cinder19:49
*** DericHorn-HP has quit IRC19:53
*** mtanino has quit IRC19:54
*** e0ne has quit IRC19:55
*** tbarron_afk is now known as tbarron19:57
*** marcusvrn1 has joined #openstack-cinder19:59
*** sp4wnr0ot_ has quit IRC20:01
eharneyLIO CI proposed:
*** marcusvrn1 has quit IRC20:09
*** boris-42 has quit IRC20:12
*** thangp has quit IRC20:13
*** edmondsw has quit IRC20:14
*** Longgeek has joined #openstack-cinder20:14
*** dtynan has quit IRC20:15
*** e0ne has joined #openstack-cinder20:15
*** dtynan has joined #openstack-cinder20:15
*** hodos has joined #openstack-cinder20:17
*** mtanino has joined #openstack-cinder20:18
*** annashen has joined #openstack-cinder20:21
*** jasondotstar|afk has quit IRC20:22
*** annashen has quit IRC20:26
hemnaeharney, nice20:30
*** jasondotstar|afk has joined #openstack-cinder20:34
*** tpsilva has quit IRC20:35
*** reed has quit IRC20:37
*** reed_ is now known as reed20:37
*** reed has quit IRC20:38
*** reed has joined #openstack-cinder20:38
*** reed_ has joined #openstack-cinder20:38
*** jwcroppe has joined #openstack-cinder20:48
*** jskinner has quit IRC20:49
*** jwcroppe_ has quit IRC20:50
*** jwcroppe_ has joined #openstack-cinder20:51
*** jwcroppe has quit IRC20:53
*** jwcroppe has joined #openstack-cinder20:54
*** jwcroppe_ has quit IRC20:56
*** jskinner has joined #openstack-cinder20:56
*** DericHorn-HP has joined #openstack-cinder20:56
*** jskinner has quit IRC21:02
*** annashen has joined #openstack-cinder21:04
*** jwcroppe_ has joined #openstack-cinder21:08
*** Rockyg has joined #openstack-cinder21:08
*** DericHorn-HP has quit IRC21:08
*** annegentle has quit IRC21:08
*** jwcroppe has quit IRC21:09
*** r-daneel has quit IRC21:11
*** xyang has quit IRC21:12
*** jasondotstar|afk has quit IRC21:17
*** tbarron is now known as tbarron_afk21:17
*** ganso_ has quit IRC21:18
*** Apoorva has joined #openstack-cinder21:18
*** Longgeek has quit IRC21:20
*** e0ne has quit IRC21:20
*** logan2 has quit IRC21:31
*** Apoorva has quit IRC21:31
*** Rockyg has quit IRC21:31
*** DericHorn-HP has joined #openstack-cinder21:33
*** lpabon has quit IRC21:36
*** mriedem has quit IRC21:36
*** DericHorn-HP has quit IRC21:40
*** boris-42 has joined #openstack-cinder21:49
*** logan2 has joined #openstack-cinder21:56
*** ianbrown has joined #openstack-cinder22:01
*** r-daneel has joined #openstack-cinder22:04
*** rwsu has quit IRC22:05
*** annegentle has joined #openstack-cinder22:07
*** pv_ has joined #openstack-cinder22:11
pv_i have done basic testing on my local vm for the cinder driver I am writing22:11
pv_and everything seems to be working with our backend22:12
pv_how do i now test my driver code with the upperlevel cinder client?22:12
pv_if thats even possible22:12
patrickeastpv_: im not 100% sure what you mean by upperlevel cinder client, but if you mean the python-cinderclient you can do that with devstack, just put your driver code in there and configure it to use your backend22:14
pv_hell yes thats exactly what i meant22:15
pv_ill look into devstack22:15
*** eharney has quit IRC22:15
*** Longgeek has joined #openstack-cinder22:17
*** chlong has joined #openstack-cinder22:18
thingeejgriffith: been 90 something days now22:18
hemnayah +A that shit22:19
hemnawant me to pull the trigger ?22:20
thingeeI also reached out to them a while back
jgriffiththingee: you get to run interference when I get the nasty email from DuncanT22:20
jgriffiththingee: about that spec...22:21
jgriffiththingee: :)22:21
thingeejgriffith: yup!22:21
jgriffiththingee: so is the plan to just clean up the report_stats in the drivers and expose it via API call?22:21
jgriffiththingee: IMHO that would be the right way to go22:22
jgriffiththingee: ok22:22
*** DericHorn-HP has joined #openstack-cinder22:24
*** annegentle has quit IRC22:24
thingeejgriffith: so are we going to thinking of non-standard capabilities as the extra specs that get passed into the volume driver?22:24
thingeeand standard capabilities would be what the scheduler can filter on?22:24
*** annegentle has joined #openstack-cinder22:24
jgriffiththingee: I guess so, but the thing is the scheduler CAN filter on anything that's reported in that dict22:25
*** Longgeek has quit IRC22:25
jgriffiththingee: if it's in the capabilities dict from the stats call it's filterable via an extra-spec key22:25
jgriffiththingee: know what I mean?22:25
thingeeok, so if we don't include them in the report stats... or is capabilities a separate report to the scheduler..22:25
jgriffiththingee: so report_stats is what goes to the scheduler22:26
jgriffiththingee: ie it "is" the capabilities22:26
jgriffiththingee: I don't know how we ended up with two names there22:26
thingeeoh I see by "clean up the report_stats" ... you didn't mean remove capabilities there completely22:27
thingeemy mistake22:27
*** ociuhandu has joined #openstack-cinder22:27
jgriffithnahh, I just meant rather than reinvent everything just start with exposing what we already have22:27
thingeeso non-standard capabilities still makes it a mess IMO, but that's left up to the driver. maybe we're ok with that though22:27
jgriffiththingee: yeah, I'm just thinking if we expose the update_stats or whatever we want to call if for now that's at least fwd progress22:28
jgriffithand *should* be non-controversial22:28
thingeejgriffith: like this already does? ... except being a particular backend22:29
*** annegentle has quit IRC22:30
jgriffiththingee: maybe the same... I was actually saying just retrieve and return:
jgriffithso you'd just leverage all the existing stuff the scheduler uses today22:31
jgriffithmaybe do some fancy formatting on it, and spit it out as proper json to the end-user22:31
thingeeis this as a first step? are we still going to do the standards and non-standards?22:33
jgriffiththingee: well, I'm not as interested in that as some, but yes... I would call it a "first" step22:33
jgriffiththingee: just seems relatively straight forward22:33
jgriffiththingee: and if people want "more stuff" they can easily add it there22:34
jgriffiththingee: we could derive standard based on what common across everyrthing there22:34
jgriffithwhich kinda makes sense to me22:34
thingeeso the formatting then22:34
thingeewe had different ideas there before.22:35
thingeeand people bring up this graffiti thing22:35
thingeeHere's what I want... I want people to be able to know what the extra spec keys are from a backend they have deployed. The driver gets that information how they choose.22:36
jgriffiththingee: well... I'm not really a fan of what's going on with all of that.  But if people want to work on it and provide something sane I'm certainly not going to block it22:36
jgriffiththingee: so I see what you're saying22:36
*** annashen has quit IRC22:36
*** briancurtin has quit IRC22:36
jgriffiththingee: but I still argue that for anything other than the standard/common stuff, vendors should write a doc22:37
jgriffiththingee: but we could certainly make it work IMO22:37
jgriffiththingee: so just append the "vendor-unique" stuff into it's own dict22:37
*** DericHorn-HP has quit IRC22:37
jgriffiththingee: and that means the work/effort is up to the vendor interested in exposing the info22:37
*** andymccr has quit IRC22:38
jgriffiththingee: in other words they need to setup the call in their driver to collect the data and append it to the update_stats call (if it's not there already)22:38
thingeeare no longer worried about the amount of data we're sending?22:38
thingeeare we*22:38
jgriffithWell, that's the thing...22:39
jgriffithwe're already sending the stuff in that update_stats call every 60 seconds22:39
*** briancurtin has joined #openstack-cinder22:39
jgriffithadding additional vendor-unique stuff "when asked" isn't a big deal I don't think22:39
jgriffiththingee: do you want me to show you a prototype of what I'm talking about?22:39
thingeeI see, a flag to ask for the appended data to report stats.22:39
thingeethat are vendor unique22:40
thingeehere's an idea...why not let the manager store in memory the last set of capabilities. if that changes, report it up with the next stats update?22:41
*** dannywilson has quit IRC22:41
*** erlon has quit IRC22:41
*** rhefner has quit IRC22:41
*** dannywilson has joined #openstack-cinder22:42
*** amit213 has quit IRC22:42
*** MentalRay has quit IRC22:43
*** rwall has quit IRC22:43
*** erlon has joined #openstack-cinder22:44
*** andymccr has joined #openstack-cinder22:44
*** lcurtis has quit IRC22:44
*** rhefner has joined #openstack-cinder22:44
*** rwall has joined #openstack-cinder22:44
*** VManiac16 has quit IRC22:50
thingeerhe00_: are you around?22:50
*** annashen has joined #openstack-cinder22:53
*** mtanino has quit IRC22:59
*** erlon has quit IRC23:00
*** tsekiyama has quit IRC23:01
thingeejgriffith: ^ in case you didn't see my earlier comments, but yeah I would be happy to see your prototype23:01
*** tsekiyama has joined #openstack-cinder23:01
*** annashen has quit IRC23:02
*** dannywilson has quit IRC23:03
*** r-daneel has quit IRC23:09
*** mtanino has joined #openstack-cinder23:13
*** BharatK has joined #openstack-cinder23:14
thingeejgriffith: unrelated , my long running juno setup has almost 7000 lock files in the cinder state directory =/23:16
jgriffiththingee: sigh23:16
jgriffiththat's just embarassing IMO23:16
patrickeast:o where are they all coming from?23:18
jgriffithpatrickeast: everybody that creates a lock file in their driver, and those in the manager23:19
jgriffithor anywhere else for that matter23:19
patrickeastwell yea, but those should be cleaned up in normal use, right?23:19
jgriffithpatrickeast: people are great at creating lock files, but they don't know how to delete them when they're done23:19
jgriffithI was told by duncan and others you "can't"23:19
jgriffithit was a big long argument in IRC when I filed a bug against it23:20
patrickeastso maybe i should have asked this *before* putting locks in my driver… but i assumed the lock decorator did the right thing, does it not?23:20
jgriffithI think I'll just write a periodic that deletes any older than 24 hours23:20
jgriffithpatrickeast: you might want to verify that23:20
jgriffithpatrickeast: nova IIRC cleans up its files when it's done23:20
patrickeasthmm thats not good23:21
jgriffithpatrickeast: indeed, it's not good at all23:21
*** hodos has quit IRC23:21
mtreinishpatrickeast: if it's the oslo.concurrency lock decorator with an external lock it should delete the lock file when the function exits23:23
jgriffithmtreinish: Yeah, in Juno we had a bunch of folks add some home-grown lock decos IIRC23:23
patrickeastmtreinish: ah perfect, so as long as it exits the function its ok23:23
patrickeastwhat happens if the service is killed mid-function?23:23
jgriffithbut I could be wrong on that23:24
*** pv_ has quit IRC23:27
mtreinishpatrickeast: if the process stops midfunction then the lock file will probably be left dangling23:27
mtreinishbecause the rm will never be called23:27
mtreinishbut that'll be like if something crashes23:27
mtreinishjgriffith: hmm, yeah that seems weird23:27
openstackgerritPatrick East proposed openstack/cinder: WIP generic image cache
mtreinishI'm also confused what you guys are doing if you're making thousands of lock files23:28
mtreinishare there that many potential inter-process conflicts23:29
openstackgerritPatrick East proposed openstack/cinder: WIP generic image cache
patrickeastmtreinish: not sure about the other vendors, but we have some issues around creating/modifying config stuff on our array in parallel so some volume operations have to be locked23:30
patrickeastmtreinish: so if someone was hammering cinder with those type of commands it could build up pretty quick23:31
mtreinishpatrickeast: sure but shouldn't that just be 1 lock for config operations (or a couple for different types)23:31
patrickeastmtreinish: ah yea, good point23:31
patrickeasti wonder if something is generating a lock name from a volume id or something?23:32
*** alexpilotti has joined #openstack-cinder23:32
jgriffithmtreinish: IMO NO, locks are very heavily abused by drivers writers and others23:32
jgriffithmtreinish: but it turns out I apparantly didn't realize that a bunch of vendors backend devices can't do simultaneous api calls to a resource23:33
jgriffithwhich seems ludicrous to me23:33
patrickeastjgriffith: speaking of scary things, i got non-public images to work with the cache… but it required some uh… unnatural acts23:34
jgriffithpatrickeast: LOL23:34
mtreinishheh ok, whatever. I'm still confused, because if you need to serialize api calls to your expensive san because it can't handle multiple requests that still seems like 1 lock to me23:35
patrickeastcheck out the diff between the last two patches on that POC review23:35
mtreinishbut I'll let it go23:35
jgriffithI'm still trying to convince our IT dept to log on to the switch and figure out how the darn iSCSI port got blocked23:35
*** lixiaoy1 has joined #openstack-cinder23:35
jgriffiththey keep coming back "we don't block anything"23:35
jgriffithoi vai23:35
*** IanGovett has quit IRC23:35
patrickeastmtreinish: yea i agree, something else must be going on to get thousands of lock files23:35
jgriffithmtreinish: no.. you're absolutely correct IMO23:35
*** chlong has quit IRC23:36
*** alexpilotti has quit IRC23:38
mtreinishjgriffith: hehe, ok23:38
*** lixiaoy11 has joined #openstack-cinder23:40
mtreinishjgriffith: ah, now I see what you mean, you've got things like:
thingeejgriffith: this juno setup was using lvm23:41
mtreinishif they're reinventing things like that I'm sure it's leaky23:41
*** lixiaoy1 has quit IRC23:41
*** Adriano__ has quit IRC23:42
*** yamada-h has joined #openstack-cinder23:42
jgriffiththingee: hmm... lvm driver has no locks... but people have been sprinkling them all over through the manager and API code23:42
jgriffithmaybe other places as well23:42
thingeeyeah, was just saying23:43
jgriffithany-who... mtreinish yes... like that one :(23:43
thingeejgriffith: do you have time for us to go back to capabilities? we never finished23:43
jgriffiththingee: here's your culprit I think:
mtreinishjgriffith: ah, yeah that would do it, a per volume external lock23:44
mtreinishthat'll create a lot of file noise23:44
jgriffithmtreinish: yup23:44
thingeejgriffith: yup matches what I'm seeing in the state directory23:44
jgriffithshame shame23:44
*** alexpilotti has joined #openstack-cinder23:44
jgriffiththingee: beside, IMO they're of little value in most of those places anyway23:44
jgriffithwhich is highly annoying23:45
jgriffiththingee: kinda.. I have about 5 minutes until I get dagger eyes from my wife23:45
patrickeastare those the infamous manager locks i keep hearning about when HA topics come up?23:45
thingeejgriffith: not getting in the middle of that.23:45
jgriffithpatrickeast: yes, and those are what some folks would like to create "more" of23:45
jgriffiththingee: LOL23:45
jgriffithsmart man!23:45
thingeejgriffith: so from earlier23:46
thingee15:40 <thingee> I see, a flag to ask for the appended data to report stats.23:46
thingee15:40 <thingee> that are vendor unique23:46
thingee15:41 <thingee> here's an idea...why not let the manager store in memory the last set of capabilities. if that changes, report it up with the next stats update?23:46
*** alexpilotti has quit IRC23:46
*** yamada-h has quit IRC23:47
*** jskinner has joined #openstack-cinder23:47
jgriffiththingee: yeah... a flag that's set by another API call even23:47
jgriffiththingee: so the "in mem and update" might be an ok idea as well23:47
jgriffiththingee: but we already poll the stats call every 60 seconds anyway23:48
*** zhenguo has joined #openstack-cinder23:48
thingeejgriffith: so then when does the scheduler get the capabilities with your approach? I would assume this flag would be a cast directly to the backend, which the scheduler wouldn't know about23:49
jgriffiththingee: so for the vendor unique stuff correct, I wouldn't provide it to the scheduler23:50
thingeebackend in my last message meant c-vol service,whoops23:50
jgriffiththingee: but you just suggested storing it in memory23:50
jgriffithso I dunno :)23:50
thingeewell yeah, earlier I though we were saying the scheduler could filter on these.23:50
thingeeso in order to do that...23:50
thingeewe need to tell the scheduler every now and then.23:51
thingeeand in order to determine when we need to would be, on start up, and if the driver thinks the capabilities changed.23:51
jgriffiththingee: so if you want the scheduler to filter on them they MUST be in that stats update23:51
jgriffiththingee: so that's an impl detail as far as I'm concerned23:52
jgriffiththingee: and it's left to the driver23:52
jgriffithI don't give a rip how it does it frankly :)23:52
thingeejgriffith: let me try again23:52
thingeejgriffith: in order to filter on capabilities we need them in the stats. We don't want to send these all the time because of the size with the appended unique stuff. How the driver gets the information to surface up to the manager is not what we're concerned with. What I was thinking is we could enforce at the manager level when it should send it for the23:55
thingeeappended information in the stats. If the manager keeps in memory at startup what the capabilities were that came from the driver, it can from there on out not send them along with the stats update. If the driver reports something different for capabilities, the manager can compare and recognize it has to send up the capabilities with the stats update.23:55
thingeejgriffith: but I agree, we could just not enforce anything in the manager and let drivers do what they want.23:56
jgriffiththingee: so I'm "ok" with the memory idea maybe...honestly I'm not sure about the concern over size (but I don't know what monster someone is going to build with this)23:58
*** laughterwym has joined #openstack-cinder23:58
jgriffiththingee: I know *some* of this has to be periodic for the capacity filter obviously23:58
jgriffiththingee: but yes a number of things could be just on init and stored23:58
jgriffiththingee: and even updated by driver when need be23:59
thingeejgriffith: I'm unsure of the size issue too. I'm not familiar with message queues in this regard.23:59
thingeejgriffith: this was just raised in the midcycle meetup by DuncanT I believe23:59
jgriffiththingee: it's probably ok... just "stupid"23:59
jgriffithcould be23:59

Generated by 2.14.0 by Marius Gedminas - find it at!