Monday, 2015-03-16

*** jcru has joined #openstack-cinder00:25
*** aix has quit IRC00:26
*** jcru has quit IRC00:28
*** jcru has joined #openstack-cinder00:28
*** jcru has quit IRC00:29
*** aix has joined #openstack-cinder00:30
*** annashen has joined #openstack-cinder00:36
*** annashen has quit IRC00:41
*** xyang has quit IRC00:48
*** vilobhmm has joined #openstack-cinder00:52
*** topshare has joined #openstack-cinder01:04
*** Longgeek has joined #openstack-cinder01:08
*** liusheng has quit IRC01:13
*** Longgeek has quit IRC01:15
*** alexpilotti has quit IRC01:19
*** harlowja_at_home has joined #openstack-cinder01:21
*** mtanino has joined #openstack-cinder01:25
openstackgerritLiu Xinguo proposed openstack/cinder: Huawei driver check before associating LUN to a LUN group
*** vilobhmm has quit IRC01:27
*** lcurtis has joined #openstack-cinder01:30
openstackgerritLiu Xinguo proposed openstack/cinder: Huawei driver fix problems under multipath
*** Yogi1 has joined #openstack-cinder01:34
*** Lee1092 has joined #openstack-cinder01:40
*** harlowja_at_home has quit IRC01:50
*** annashen has joined #openstack-cinder01:54
*** annashen has quit IRC01:58
*** annashen has joined #openstack-cinder01:59
*** xyang has joined #openstack-cinder02:01
*** ho has joined #openstack-cinder02:02
*** Yogi1 has quit IRC02:08
*** Administrator__ has quit IRC02:14
*** Administrator__ has joined #openstack-cinder02:14
*** asselin_ has quit IRC02:23
*** julim has quit IRC02:24
*** takedakn has joined #openstack-cinder02:49
openstackgerritVincent Hou proposed openstack/cinder: Delete the temporary volume if migration fails
*** jungleboyj_ has joined #openstack-cinder02:53
*** takedakn has quit IRC02:57
openstackgerritVincent Hou proposed openstack/cinder-specs: Enhance list operations with the additional keys and next link
*** coolsvap|afk is now known as coolsvap02:58
*** annashen has quit IRC03:05
*** Longgeek has joined #openstack-cinder03:11
openstackgerritJeegn Chen proposed openstack/cinder: More error handling on VNX migration failure.
*** Longgeek has quit IRC03:13
*** Longgeek has joined #openstack-cinder03:14
*** topshare_ has joined #openstack-cinder03:18
*** Longgeek_ has joined #openstack-cinder03:19
*** topshare has quit IRC03:20
*** Longgeek has quit IRC03:20
*** lcurtis has quit IRC03:25
*** buster_604 has joined #openstack-cinder03:31
*** takedakn has joined #openstack-cinder03:35
*** topshare_ has quit IRC03:36
*** topshare has joined #openstack-cinder03:37
openstackgerritwanghao proposed openstack/cinder: Implement function of import/export snapshots
openstackgerritTina Tang proposed openstack/cinder: Attach/detach batch processing in VNX driver
*** Longgeek has joined #openstack-cinder03:41
*** Ilja has joined #openstack-cinder03:43
*** Longgeek_ has quit IRC03:44
*** takedakn has quit IRC03:48
*** haomaiwang has joined #openstack-cinder03:49
*** shyama has joined #openstack-cinder04:09
*** Longgeek has quit IRC04:11
*** Longgeek has joined #openstack-cinder04:11
*** topshare has quit IRC04:20
*** vincent_hou has joined #openstack-cinder04:22
vincent_houmtanino: Ping.04:22
mtaninovincent_hou: hi04:22
mtaninovincent_hou: May be previous comment is confusiong.04:23
mtaninovincent_hou: simple question. Can you delete the dest volume using "cinder delete vol-id" after the migration will be failed?04:24
mtaninovincent_hou: I think we can't.04:24
vincent_houOK. If we still have dest volume in the DB. It is already across two steps.04:25
vincent_houThe first is migration fails, then we delete the dest volume. The second is deleting this dest volume also fails.04:26
vincent_houWhen the first step happened, then we reach rpcapi.delete_volume(ctxt, volume)04:27
mtaninovincent_hou: yup04:27
vincent_houIf rpcapi.delete_volume(ctxt, volume) also fails, then we will have this dest volume still in DB.04:27
mtaninovincent_hou: Also we can see the volume using "cinder list" right?04:28
*** Ilja has quit IRC04:28
mtaninovincent_hou: In this situation can you delete the volume using cinder delete?04:28
vincent_houNo, because rpcapi.delete_volume(ctxt, volume) is already trying to do that and it fails.04:29
mtaninovincent_hou: OK. After trying to solve the root cause of deletion failure such as "occupied by other devices",04:30
vincent_houI guess one possible solution is to clean the DB record for it.04:31
mtaninovincent_hou: user wnat to delete the volume again.04:31
vincent_houRight, it is quite confusing.04:32
mtaninovincent_hou: So I want to clear the migration_status at Line 510 or Line 517 when volume deletion is failed04:32
vincent_houThe dest volume is still in DB, but user cannot delete ot even force delete it.04:32
mtaninovincent_hou: yes yes04:33
vincent_houFor this issue, I think clearing the migration_status is not enough.04:33
mtaninovincent_hou: Yes. both solving the root cause of deletion failure and clean up of migration_status are needed?04:34
mtaninovincent_hou: but we can't solve the root cause via Cinder.04:35
vincent_houHow about doing a check in delete_volume? First I need to know it is a dest volume or not. If it is a dest volume, then we need to clear the DB anyway.04:35
*** [1]Thelo has joined #openstack-cinder04:35
*** topshare has joined #openstack-cinder04:36
mtaninovincent_hou: Please see line 522 and 523. I added flags for cheking migration source, dest volumes04:36
mtaninovincent_hou: you can use "is_migrating_dest" flag04:37
vincent_houIf we cannot remove the dest volume from the backend, then we cannot force to delete it from the backend.04:37
mtaninovincent_hou: so please move these flags around Line 494 or someting and use it.04:38
*** Thelo has quit IRC04:38
*** [1]Thelo is now known as Thelo04:38
*** bkopilov has quit IRC04:39
*** harry has joined #openstack-cinder04:39
harryis e0ne here?04:40
harryping e0ne04:40
harry@e0ne  I am harry04:40
mtaninovincent_hou: good night, zzz04:40
*** mtanino is now known as mtanino_away04:40
harryOh, let's contact him tomorrow ;-)04:41
vincent_houmtanino: Thank you. take care.04:41
harrye0ne: are you here?04:42
harryping thingee04:43
harrythingee: are you here?04:43
openstackgerritMike Perez proposed openstack/cinder: Revert "Datera's export to avoid deprecated keys"
*** annashen has joined #openstack-cinder04:47
*** changbl has quit IRC04:48
*** bkopilov has joined #openstack-cinder04:55
*** xyang has quit IRC04:57
*** xyang has joined #openstack-cinder04:57
*** BharatK has joined #openstack-cinder05:00
*** annashen has quit IRC05:03
openstackgerritVincent Hou proposed openstack/cinder: Delete the temporary volume if migration fails
openstackgerritVincent Hou proposed openstack/cinder: Delete the temporary volume if migration fails
openstackgerritVincent Hou proposed openstack/cinder: Delete the temporary volume if migration fails
*** anuragpalsule has joined #openstack-cinder05:23
openstackgerritVincent Hou proposed openstack/cinder: Delete the temporary volume if migration fails
*** Ilja has joined #openstack-cinder05:28
openstackgerritVincent Hou proposed openstack/cinder: Delete the temporary volume if migration fails
*** Mandell has joined #openstack-cinder05:52
*** vincent_hou has quit IRC05:55
*** changbl has joined #openstack-cinder05:58
*** lpetrut has joined #openstack-cinder06:01
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Imported Translations from Transifex
*** nlevinki has joined #openstack-cinder06:12
*** vilobhmm has joined #openstack-cinder06:13
*** Mandell has quit IRC06:17
*** nshaikh has joined #openstack-cinder06:17
openstackgerritPranali Deore proposed openstack/cinder: Mocked utils.execute for broken tests
*** afazekas has joined #openstack-cinder06:23
*** pradipta has joined #openstack-cinder06:25
*** Administrator__ has quit IRC06:26
*** Mandell has joined #openstack-cinder06:26
*** Administrator__ has joined #openstack-cinder06:26
*** xyang has quit IRC06:33
*** nkrinner has joined #openstack-cinder06:42
*** abehl has joined #openstack-cinder06:47
*** vilobhmm has quit IRC06:49
*** Longgeek_ has joined #openstack-cinder06:52
*** Longgeek has quit IRC06:52
*** ronis has joined #openstack-cinder07:11
*** pratyu has joined #openstack-cinder07:13
*** pcaruana has quit IRC07:14
*** kaisers has joined #openstack-cinder07:24
*** TobiasE has joined #openstack-cinder07:26
*** mtanino_away has quit IRC07:28
*** e0ne has joined #openstack-cinder07:34
*** sgotliv has quit IRC07:36
*** e0ne has quit IRC07:39
*** rwsu has joined #openstack-cinder07:43
*** Miouge has joined #openstack-cinder07:44
*** liusheng has joined #openstack-cinder07:44
*** markus_z has joined #openstack-cinder08:08
*** jordanP has joined #openstack-cinder08:09
*** dulek has joined #openstack-cinder08:16
*** asselin_ has joined #openstack-cinder08:18
*** pradipta has quit IRC08:22
*** asselin_ has quit IRC08:23
*** lpetrut has quit IRC08:28
*** Miouge has quit IRC08:29
*** Miouge has joined #openstack-cinder08:32
*** ndipanov has joined #openstack-cinder08:33
*** Miouge has quit IRC08:36
*** nellysmitt has joined #openstack-cinder08:40
openstackgerritMichal Dulko proposed openstack/cinder: Backup object
openstackgerritTina Tang proposed openstack/cinder: Attach/detach batch processing in VNX driver
*** Miouge has joined #openstack-cinder08:44
*** topshare_ has joined #openstack-cinder08:53
*** Miouge has quit IRC08:54
*** topshare has quit IRC08:55
*** lpetrut has joined #openstack-cinder08:56
*** lpetrut has quit IRC08:57
*** nlevinki has quit IRC09:01
*** Mandell has quit IRC09:01
*** jistr has joined #openstack-cinder09:11
*** harry has quit IRC09:14
*** harry has joined #openstack-cinder09:14
*** nlevinki has joined #openstack-cinder09:16
*** deepakcs has joined #openstack-cinder09:22
*** sgotliv has joined #openstack-cinder09:26
deepakcsDuncanT, ping, reg. the discussion u were having with BharatK09:26
deepakcsDuncanT, I don't think its a good idea to assume user knows about fallocate (context: <DuncanT> choices = ['qcow2', 'sparse', 'dd', 'fallocate'])09:27
deepakcsDuncanT, or even know about which is better (dd Vs fallocate etc)09:27
deepakcsDuncanT, I was thinking of something like this:09:27
deepakcsvolume_format = ['qcow2', 'raw'] and provisioning_type = ['thick', 'thin'] - generic and easy to understand09:28
deepakcsDuncanT, ^^ what do you think ?09:28
DuncanTdeepakcs: Can you have thin and thick provisioned qcow2 in any meaningful way?09:30
*** harry has quit IRC09:31
deepakcsDuncanT, good point, we can ignore the prov_type for qcow2 (as we only support thin there, and it doesn't make sense to have thick)09:31
deepakcsDuncanT,  but for raw {thin, thick} we can use fallocate for thick and fall back to 'dd' if fallocate doesn't work for some FS reason, thin would be to use sparse09:32
DuncanTdeepakcs: I'm not sure which is clearer. Is there a reason you'd want to use dd rather than fallocate?09:32
DuncanTTrying fallocate, waiting for it to fail the using dd is likely to produce messy logs and bad performance, I'd have thought?09:32
deepakcsDuncanT, for some (rare?) cases where FS doesn't support fallocate, for eg: glusterfs fuse didn't support in 3.4 but it does in 3.6 and we would want to support both09:32
deepakcsDuncanT, and user doesn't have to worry whether it works or not, either way we give him thick volume09:33
deepakcsDuncanT, we can control the logs (just put a warn, don't dump stack)09:33
deepakcsDuncanT, one other way could be choice = ['qcow2', 'raw-thin', 'raw-thick']09:34
DuncanTI think I'd prefer the explicit choice for now. Warnings are supposed to be things admins can fix, so we definitely don't want warning.09:35
deepakcsDuncanT, choice = ['qcow2', 'raw-thin', 'raw-thick'] isn't explicit you mean ? I am not sure if dd and fallocate are good to be exposed Vs thin and thick ?09:36
DuncanTdeepakcs: If 'thick' was one choice, I'd agree, but ti isn't, it still leaves a choice between dd and falocate.09:37
deepakcsDuncanT, Hmm ok , my concern there is that just choosing dd/fallocate doesn't really tell the format type (raw) Vs choosign qcow2, there should be some symettry, no ?09:39
DuncanTdeepakcs: I'm far more concerned with correctness than symmetry here. Until we don't support any filesystems that don't support fallocate, we need to be able to force dd, and even then we might want to keep it for debugging09:40
DuncanTI think one option with all of the choices is clearer than any alternative presented09:41
*** marcusvrn1 has joined #openstack-cinder09:42
jordanPsorry for jumping in late: is the choice presented to end users or sysadmins ?09:42
jordanPif it's for sysadmin, imo we can assume they know the difference between dd and fallocate09:42
DuncanTjordanP: Admins, and I agree if it was tennant facing we'd need to clear it up09:43
deepakcsjordanP, its cinder.conf, mostly sysadmins, but making it friendlier is what i was aiming for, but DuncanT's point is good too09:43
deepakcsDuncanT, ok so 1 option with multiple explicit choices is what we go with then.09:44
jordanPright, though I am not sure to see the difference between 'sparse' and 'fallocate' myself :)09:44
DuncanTdeepakcs: Thanks09:44
deepakcsDuncanT, and what about back compat, i know u said u don't know whats the right way there, but BharatK says there is deprecate option provided in oslo.cfg, we should use that then ?09:45
DuncanTjordanP: It is fallocate .v. dd - fallocate is substantially faster and places less load on the system, for filesystems that support it09:45
deepakcsjordanP, and sparse is thinp , fallocate is thick (from FS perspective)09:46
jordanPah ok, yeah, got it thanks09:46
DuncanTdeepakcs: You should use the deprecate option, yes. The hard bit is figuring out if the new type option is specified or not, and if not how to emulate the old behaviour. I'm sure it can be done, but it is probably fiddly (and should have its own unit test(s) making all of the combinations clear)09:47
deepakcsDuncanT, " if the new type option is specified or not" - if new_option not in cfg , then its not specified, otherwise it is.. didn't get u on this one09:48
DuncanTdeepakcs: Config options have a default value09:48
DuncanTdeepakcs: Other than setting the default to none or something, and translating that in the code, I'm not sure how to tell it isn't set. It might be easy, if so, great. I don't know09:49
deepakcsDuncanT, I see ur point now, yeah, setting new_option default to some invalid value could be the only way out09:50
deepakcsDuncanT, unless oslo has a way to tell whther the option is literally specified in the conf or not :)09:51
DuncanTdeepakcs: I've no idea, sorry. All I know is that it is a tricky area.09:51
DuncanTdeepakcs: Might be worth approaching the oslo_cfg maintainers for advice?09:52
DuncanTdeepakcs: If you get a good solution, please let me know09:52
deepakcsDuncanT, sure, and thanks for your suggestions09:54
*** topshare_ has quit IRC09:55
deepakcsDuncanT, i know that in manila project they wanted to implement something similar (a config option that _must_ be specified by the admin), IIRC they set the default to None and used the same trick09:55
*** topshare has joined #openstack-cinder09:56
*** sseverson has quit IRC10:00
*** sseverson has joined #openstack-cinder10:01
*** Maike has joined #openstack-cinder10:06
*** nlevinki has quit IRC10:09
*** Ilja1 has joined #openstack-cinder10:12
*** Ilja has quit IRC10:13
*** sgotliv has quit IRC10:14
*** sgotliv has joined #openstack-cinder10:14
*** topshare has quit IRC10:15
*** coolsvap is now known as coolsvap|afk10:19
*** topshare has joined #openstack-cinder10:23
*** nlevinki has joined #openstack-cinder10:23
*** ho has quit IRC10:24
*** Administrator__ has quit IRC10:24
*** sgotliv has quit IRC10:25
*** Ilja1 has quit IRC10:29
*** topshare has quit IRC10:32
*** Ilja has joined #openstack-cinder10:37
*** rakesh_mishra__ has joined #openstack-cinder10:40
*** rakesh_mishra__ has quit IRC10:41
*** rakesh_mishra__ has joined #openstack-cinder10:42
rakesh_mishra__can any one tell me when i create the snapshot where it stores in the file system10:45
rakesh_mishra__Like images are store in the below location10:45
rakesh_mishra__this is the volume-snapshot10:45
*** topshare has joined #openstack-cinder10:47
*** topshare has quit IRC10:47
*** topshare has joined #openstack-cinder10:48
*** avishay has joined #openstack-cinder10:49
*** rakesh_mishra_ has joined #openstack-cinder10:51
*** rakesh_mishra__ has quit IRC10:51
DuncanTrakesh_mishra_: If you're using the default cinder driver (LVM) and the default devstack config (loopback) then they are stored on the cinder looback file10:56
DuncanT"sudo losetup -a" should list it10:56
rakesh_mishra_DuncanT: yes i am using deafult config file and lvm driver11:05
*** haomaiwang has quit IRC11:06
rakesh_mishra_DuncanT: it showing below output when i run the command11:08
rakesh_mishra_/dev/loop0: [0801]:404258 (/opt/stack/data/stack-volumes-default-backing-file)11:08
rakesh_mishra_/dev/loop1: [0801]:404677 (/opt/stack/data/stack-volumes-lvmdriver-1-backing-file)11:08
*** nlevinki has quit IRC11:09
*** takedakn has joined #openstack-cinder11:14
*** takedakn has quit IRC11:18
*** takedakn has joined #openstack-cinder11:18
*** aix has quit IRC11:21
DuncanTCan you post up /etc/cinder/cinder.conf please? Us paste.openstack.org11:22
*** timcl has joined #openstack-cinder11:24
*** rakesh_mishra_ has quit IRC11:24
*** rakesh_mishra_ has joined #openstack-cinder11:24
*** topshare has quit IRC11:26
*** sgotliv has joined #openstack-cinder11:31
*** IanGovett has joined #openstack-cinder11:31
rakesh_mishra_DuncanT: yes one minute11:34
DuncanTOk, your snapshots and volumes are all in /opt/stack/data/stack-volumes-lvmdriver-1-backing-file11:37
rakesh_mishra_but this is the single file11:37
rakesh_mishra_i have created so many snapshots11:37
DuncanTYes. It is an LVM device11:37
DuncanTThey all go on the same pv or pvs. Devstack by default sets up one11:38
DuncanTOn a 'real' deployment, you'd use one or more dedicated spindles most likely11:38
*** rushiagr_away is now known as rushiagr11:39
DuncanTPysical volume. It's LVM terminology, google should find you an LVM primer11:40
DuncanTPhysical volume, sorry11:40
rakesh_mishra_but i have created a volume of 1GB so it should take 1GB space instead of update the dictionary in stack-volumes-lvmdriver-1-backing-file11:41
*** rakesh_mishra_ has quit IRC11:42
*** rakesh_mishra_ has joined #openstack-cinder11:42
rakesh_mishra_but i have created a volume of 1GB so it should take 1GB space instead of update the dictionary in stack-volumes-lvmdriver-1-backing-file11:43
*** pratyu has quit IRC11:44
*** nlevinki has joined #openstack-cinder11:45
*** rakesh_mishra__ has joined #openstack-cinder11:48
*** rakesh_mishra_ has quit IRC11:50
*** jaypipes has joined #openstack-cinder11:51
*** rakesh_mishra_ has joined #openstack-cinder11:52
DuncanTLVM doesn't work like that, and you've got thin provisioning enabled11:54
DuncanTPlease go and find an LVM primer11:54
*** rakesh_mishra__ has quit IRC11:54
*** rushiagr is now known as rushiagr_away11:55
*** rakesh_mishra_ has quit IRC11:55
*** rakesh_mishra_ has joined #openstack-cinder11:55
*** TobiasE1 has joined #openstack-cinder12:00
*** rushiagr_away is now known as rushiagr12:01
*** jcru has joined #openstack-cinder12:01
*** jcru has quit IRC12:01
*** TobiasE has quit IRC12:01
*** jcru has joined #openstack-cinder12:02
*** nlevinki has quit IRC12:02
*** rakesh_mishra_ has quit IRC12:03
*** rakesh_mishra_ has joined #openstack-cinder12:03
*** deepakcs has quit IRC12:07
*** nlevinki has joined #openstack-cinder12:15
*** bswartz has quit IRC12:26
*** alexpilotti has joined #openstack-cinder12:27
*** rakesh_mishra_ has quit IRC12:32
*** porrua has joined #openstack-cinder12:42
*** lpetrut has joined #openstack-cinder12:47
*** openstackgerrit has quit IRC12:50
*** openstackgerrit has joined #openstack-cinder12:50
*** fifieldt has joined #openstack-cinder12:51
*** akerr has joined #openstack-cinder12:53
*** avishay has quit IRC13:00
*** bswartz has joined #openstack-cinder13:01
*** Yogi1 has joined #openstack-cinder13:02
*** Mandell has joined #openstack-cinder13:02
*** Mandell has quit IRC13:07
*** julim has joined #openstack-cinder13:09
*** aix has joined #openstack-cinder13:10
nikesh_vedamsi am using this local.conf for setting up devstack in my CI and running tox -e all -- volume | tee -a console.log.out   fro cinder volume tests for tgt but around 14 tests are failing13:11
*** dustins has joined #openstack-cinder13:13
nikesh_vedamsany idea13:13
nikesh_vedamsis any thing missing in local.conf13:13
*** TobiasE has joined #openstack-cinder13:18
*** TobiasE1 has quit IRC13:19
*** ganso_ has joined #openstack-cinder13:21
*** topshare has joined #openstack-cinder13:22
*** topshare has quit IRC13:22
*** topshare has joined #openstack-cinder13:23
*** IanGovett has quit IRC13:24
openstackgerritabhiram moturi proposed openstack/cinder: Enhance unit tests for zfssa drivers
*** timcl has quit IRC13:25
openstackgerritAnkit Agrawal proposed openstack/cinder: Fix missing translations for log messages
*** sandywalsh has joined #openstack-cinder13:31
*** IanGovett has joined #openstack-cinder13:32
*** scottda has joined #openstack-cinder13:34
*** shyama has quit IRC13:36
*** scottda has quit IRC13:36
*** jungleboyj_ has quit IRC13:37
*** mriedem has joined #openstack-cinder13:37
openstackgerritAnkit Agrawal proposed openstack/cinder: Fix missing translations for log messages
*** anuragpalsule has quit IRC13:38
*** timcl has joined #openstack-cinder13:41
*** scottda has joined #openstack-cinder13:42
openstackgerritAnkit Agrawal proposed openstack/cinder: Fix missing translations for log messages
*** lpabon has joined #openstack-cinder13:50
*** scottda has quit IRC13:52
*** rakesh_mishra_ has joined #openstack-cinder13:54
*** takedakn has quit IRC14:09
*** asselin_ has joined #openstack-cinder14:15
*** rakesh_mishra_ has quit IRC14:15
*** jecarey has joined #openstack-cinder14:21
*** bkopilov has quit IRC14:22
*** mwichmann has joined #openstack-cinder14:22
*** abehl has quit IRC14:22
*** topshare has quit IRC14:24
*** jungleboyj_ has joined #openstack-cinder14:24
*** BharatK has quit IRC14:26
*** thangp has joined #openstack-cinder14:26
*** jungleboyj_ is now known as jungleboyj14:26
*** Mandell has joined #openstack-cinder14:28
*** scottda has joined #openstack-cinder14:32
*** bkopilov has joined #openstack-cinder14:33
*** scottda has quit IRC14:35
*** [1]Thelo has joined #openstack-cinder14:37
*** Yogi11 has joined #openstack-cinder14:39
*** Thelo has quit IRC14:39
*** [1]Thelo is now known as Thelo14:39
*** Yogi1 has quit IRC14:42
*** rakesh_mishra_ has joined #openstack-cinder14:43
*** hemnafk is now known as hemna14:49
*** Maike has quit IRC14:50
*** nshaikh has quit IRC14:58
*** patrickeast has joined #openstack-cinder15:01
*** rushiagr is now known as rushiagr_away15:02
*** Maike has joined #openstack-cinder15:03
*** Maike has quit IRC15:03
*** Maike has joined #openstack-cinder15:04
hemnawinston-d_zZZ, ping15:10
*** topshare has joined #openstack-cinder15:11
*** david-lyle_afk is now known as david-lyle15:15
*** topshare has quit IRC15:16
*** scottda has joined #openstack-cinder15:16
*** scottda has quit IRC15:17
*** porrua has quit IRC15:17
*** dannywilson has joined #openstack-cinder15:18
*** anuragpalsule has joined #openstack-cinder15:20
*** anuragpalsule has quit IRC15:20
*** anuragpalsule has joined #openstack-cinder15:21
*** rmesta has joined #openstack-cinder15:21
*** tsekiyama has joined #openstack-cinder15:22
*** rushiagr_away is now known as rushiagr15:30
*** _cjones_ has joined #openstack-cinder15:31
*** rmesta has left #openstack-cinder15:33
*** scottda has joined #openstack-cinder15:33
*** scottda has quit IRC15:35
*** Longgeek_ has quit IRC15:36
*** Yogi11 has quit IRC15:37
*** Mandell has quit IRC15:38
openstackgerritMichal Dulko proposed openstack/cinder: Remove unused file locks on c-vol service startup
*** Yogi1 has joined #openstack-cinder15:42
SwansonAre the 9 million "Can not find policy directory: policy.d" littering the c-api log any kind of problem?15:42
DuncanTSwanson: A problem? Not unless you've got policy in a policy.d directory you're expecting to get used. It is a bug in the packaging though - whatever install c-api should create an empty policy.d directory15:44
hemnaSwanson, It would be a problem if it were 10million.   9million is fine.15:44
hemnaDuncanT, if you have a minute, can you go over again for me ?15:45
hemnaI'd like to get that one over with15:45
SwansonThanks!  Just noticed a lot of those messages.15:45
DuncanThemna: Sure15:47
openstackgerritRonen Mesonzhnik proposed openstack/cinder: Provide backup service that uses ProtecTIER as the backend
*** _cjones_ has quit IRC15:48
kaisersjgriffith: Hi! Is it ok for you to follow up now regarding the ci issue 3rdparty meeting ?15:49
*** rushiagr is now known as rushiagr_away15:53
SwansonEvery time I try to create an 11th volume I get this error "ERROR: Request Entity Too Large (HTTP 413) (Request-ID: req-a333fd96-c58e-42f8-9733-7f981d1edaf8)"15:53
SwansonI presume this isn't a quota thing as I've upped my quota.15:53
jgriffithkaisers: sure15:54
jgriffithkaisers: sorry I didn't catc exactly what you're running in to15:54
kaiserskaisers: np15:56
*** harlowja_at_home has joined #openstack-cinder15:56
jgriffithkaisers: gimmie the run down and maybe I can help :)15:56
kaisersBasically some of the feedbacks to gerrit do not turn up in the gerrit ui15:56
jgriffithkaisers: ahh15:56
jgriffithkaisers: failing issues?15:56
kaisersjgriffith: yep15:56
jgriffithkaisers: yeah... when's the last pull you made from the repo?15:57
kaisersmost stuff still fails :-(15:57
kaisersjgriffith: friday?15:57
jgriffithkaisers: so up until the new year I didn't report failures15:57
kaisersjgriffith: i think15:57
jgriffithkaisers: oh LOL15:57
jgriffithwell then that's different15:57
kaisersjgriffith: :)15:57
jgriffithkaisers: some show, and some don't?  That's odd15:57
kaisersjgriffith: i forked some weeks ago and did fiddle a bit with it15:57
*** kfox1111 has joined #openstack-cinder15:57
jgriffithkaisers: do you have the branch you're using in github somewhere?15:58
kaisersjgriffith: the casusbelli fork is mine15:58
*** anuragpalsule has quit IRC15:59
jgriffithkaisers: link?15:59
kaisersjgriffith: but i didn't do too many changes, primarily added logging config by variables15:59
*** BharatK has joined #openstack-cinder15:59
kaisersjgriffith: oh, sorry15:59
kaisersjgriffith: #link
jgriffithsure there's a way to find it but i'm not that smart :)16:00
*** rakesh_mishra_ has quit IRC16:00
kaisersjgriffith: sry, page is open most of the time in me browser and never thought it might be different for the rest of the world. :)16:00
jgriffithkaisers: haha.. I do that all the time16:01
*** nkrinner has quit IRC16:01
*** gary-smith_ has quit IRC16:01
kaisersjgriffith: Most of my adaptions are done at installation time by different ansible scripts that install sos-ci with my configuration settings on our ci server16:01
*** gary-smith_ has joined #openstack-cinder16:01
*** nlevinki has quit IRC16:02
*** rushiagr_away is now known as rushiagr16:02
*** gary-smith_ has quit IRC16:03
*** gary-smith_ has joined #openstack-cinder16:04
jgriffithkaisers: checkout your subject forming: ie around line 8516:04
nikesh_vedamsjgriffith: after switching to kvm,now performance is very good of SOSCI,but tempest tests were failing so i tried with tgt but tempest is failing onn tgt also,using this for local.conf16:04
kaisersjgriffith: ok16:05
jgriffithkaisers: actually that all looks pretty good16:05
smcgindianikesh_vedams: Publishing your logs might make it easier to see what's going on.16:06
kaisersjgriffith: did not touch much, only added the failure subject line16:06
jgriffithkaisers: so here's what I have done in the past:16:06
kaisersjgriffith: not added but adopted to your variable based subject setting16:06
jgriffithkaisers: add a logger in there that dumps the same data to a local log file16:06
jgriffiththen go through and see if you can find the missing pieces and why16:06
jgriffithalso add a check to that send notification16:06
jgriffithsee what falls out16:06
jgriffithI haven't had issues with that myself, but who know16:07
kaisersjgriffith: ok. Did you see the paste link in the meeting chat?16:07
nikesh_vedamssmcgindia :
*** Maike has quit IRC16:07
nikesh_vedamssmcgindia : did you move to india ,earlier you irc name was smcginnis right :)16:08
kaisersjgriffith: i just want to add: From the log ouput that string looks ok. Maybe you see something ?16:08
kaisersjgriffith: i mean only the output at line 7 or 12 in
*** rmesta has joined #openstack-cinder16:09
kaisersjgriffith: that pretty much contains the data posted.16:09
smcgindianikesh_vedams: In India for the week, so decided to change it for now. About ready to get some sleep. :)16:09
kaisersjgriffith: i'll see into reading std.err/out to see if that returns something...16:09
smcgindianikesh_vedams: It looks like you are getting LVM errors.16:09
*** scottda has joined #openstack-cinder16:10
smcgindianikesh_vedams: Search for "Traceback" here:
nikesh_vedamssmcgindia:which city?16:10
smcgindianikesh_vedams: Search for "Details" in your console.log.out. They are timing out.16:10
*** dulek has quit IRC16:11
smcgindianikesh_vedams: In Bangalore.16:11
nikesh_vedamssmcgindia: i m in hyderabad16:11
smcgindianikesh_vedams: Ah, wouldn't mind visiting there some day. I've only seen BLR.16:11
*** openstackgerrit has quit IRC16:11
smcgindianikesh_vedams: From the Detail output of the failed tests you can get the associated volumes.16:12
*** openstackgerrit has joined #openstack-cinder16:12
smcgindianikesh_vedams: You can then use that ID to search in your c-vol.log.16:12
*** scottda has quit IRC16:12
smcgindianikesh_vedams: You can trace through there and see what operations were being performed.16:12
smcgindianikesh_vedams: You can also get the request ID "req-xxxxxx" and search on that to get all log messages within the current operation.16:13
*** rmesta has left #openstack-cinder16:13
smcgindianikesh_vedams: Maybe not all LVMs fault. Would this be related to your work:16:13
smcgindianikesh_vedams:  ISCSITargetRemoveFailed: Failed to remove iscsi target for volume16:13
smcgindianikesh_vedams: File "/opt/stack/cinder/cinder/volume/targets/", line 362, in remove_iscsi_target16:14
*** Ilja has quit IRC16:14
nikesh_vedamssmcgindia:ok but is local.conf good16:15
smcgindianikesh_vedams: As far as I can tell. You would probably have bigger issues if not. Maybe.16:15
nikesh_vedamssmcgindia: because tgt setup is also failing16:16
*** aix has quit IRC16:16
smcgindianikesh_vedams: Are the resulting conf files (cinder.conf, etc) as you would expect for your setup?16:16
*** patrickeast has quit IRC16:17
*** btran has joined #openstack-cinder16:17
*** nlevinki has joined #openstack-cinder16:18
smcgindianikesh_vedams: Sorry, I don't have any experience with that part of things.16:18
openstackgerritThang Pham proposed openstack/cinder: Switch get_all_snapshots to use objects
openstackgerritThang Pham proposed openstack/cinder: Create unit tests for volume objects
openstackgerritThang Pham proposed openstack/cinder: Create unit tests for volume objects
*** TobiasE has quit IRC16:27
*** markstur has quit IRC16:28
*** nellysmitt has quit IRC16:30
*** dflorea has joined #openstack-cinder16:30
*** anuragpalsule has joined #openstack-cinder16:31
openstackgerritThang Pham proposed openstack/cinder: Switch to oslo_versionedobjects
openstackgerritThang Pham proposed openstack/cinder: Raise exception for invalid mock assert calls
*** sandywalsh has quit IRC16:36
*** Adriano_ has joined #openstack-cinder16:39
*** Apoorva has joined #openstack-cinder16:40
*** scottda has joined #openstack-cinder16:40
*** scottda has quit IRC16:42
openstackgerritThang Pham proposed openstack/cinder: Switch to oslo_versionedobjects
*** nlevinki has quit IRC16:45
*** markus_z has quit IRC16:52
*** Yogi11 has joined #openstack-cinder16:54
*** openstackgerrit has quit IRC16:54
*** ronis has quit IRC16:54
*** openstackgerrit has joined #openstack-cinder16:55
*** Yogi1 has quit IRC16:55
*** leeantho has joined #openstack-cinder16:57
*** vilobhmm has joined #openstack-cinder16:58
*** nellysmitt has joined #openstack-cinder16:58
*** mtanino has joined #openstack-cinder16:59
*** harlowja_at_home has quit IRC17:00
*** annashen has joined #openstack-cinder17:00
*** anuragpalsule has quit IRC17:05
*** thingee has joined #openstack-cinder17:05
marcusvrn1asselin: hi17:08
asselinmarcusvrn1, hi17:09
marcusvrn1asselin: if you have some free time, could you review my change( it's about the snapshot bug fix that was impacting my CI...and I would like to put all snapshots tests back again as soon as possible, because more changes to conclude the migration to snapshots object are coming and if the tests are excluded, I will not know if some change will break my driver.17:14
*** patrickeast has joined #openstack-cinder17:15
*** cbader has joined #openstack-cinder17:16
*** cbader02 has joined #openstack-cinder17:16
thingeemarcusvrn1: happy to see the ci reporting, good work to your team :)17:16
asselinmarcusvrn1, looking17:17
marcusvrn1thingee: thanks! :)17:18
marcusvrn1asselin: thanks too17:18
*** erlon has joined #openstack-cinder17:19
*** markstur has joined #openstack-cinder17:19
*** cbader02 has quit IRC17:21
*** Mandell has joined #openstack-cinder17:24
asselinmarcusvrn1, nice work on the ci system17:25
*** timcl has quit IRC17:25
*** BharatK has quit IRC17:26
*** BharatK has joined #openstack-cinder17:26
*** harlowja has joined #openstack-cinder17:28
patrickeastthingee: speaking of ci’s… just a quick update on the Pure Storage CI system, we’ve had it offline for the last week or so while moving things around in our lab (adding more physical machines to it, better networking, etc)17:28
*** timcl has joined #openstack-cinder17:29
patrickeastthingee: it should be back up by the 19th, but it something happens, hypothetically, what happens?17:29
marcusvrn1asselin: thanks! do you know who can give approval to this change? maybe thingee?17:30
asselinthingee, hemna jgriffith are cores:
marcusvrn1asselin: It's really hard to keep CI working... hehehe17:31
marcusvrn1jgriffith: thanks17:32
jgriffithmarcusvrn1: sure17:32
asselinyes it is...lots of interesting things come up17:33
*** BharatK has quit IRC17:35
thingeepatrickeast: we'll visit it on the 19th, just keep me updated17:35
thingeejgriffith: hitting this issue on my ci
openstackLaunchpad bug 1432490 in tempest "TestEncryptedCinderVolumes cryptsetup name is too long" [Undecided,New]17:36
thingeejgriffith: not really sure who to blame :)17:36
*** rushiagr is now known as rushiagr_away17:36
thingeejgriffith: I'm having an internal change be made to not pass the volume name in the iqn as a work around, but annoying17:36
jgriffiththingee: yeah, I was reading that.  So that's specific to your system not accepting the longer name though17:36
jgriffiththingee: right?17:36
thingeejgriffith: no cryptsetup says it'17:37
thingees too long17:37
thingeeDatera is passing the volume name in the iqn17:37
jgriffithhmm... why isn't anybody else hitting it?17:37
thingeewhich is causing it to be too long imo17:37
jgriffiththingee: hmm... lemme look17:38
*** rushil has joined #openstack-cinder17:38
thingeein case you missed the stack trace17:38
openstackLaunchpad bug 1432490 in tempest "TestEncryptedCinderVolumes cryptsetup name is too long" [Undecided,New]17:38
jgriffiththingee: I see the stack trace17:38
jgriffiththingee: not finding it in other CI setups though17:39
jgriffithat least not yet17:39
thingeemtreinish: I think this might just be something that needs to be worked out in my backend and or cryptsetup.17:39
thingeejgriffith: yeah, pretty sure not passing the volume name in would help17:39
mtreinishthingee: ok, that's what I was thinking too when I looked at the bug17:39
thingeejgriffith: the device name is just too long
thingeejgriffith: volume name in this case from tempest being OpenStack-TestEncryptedCinderVolumes-676292884:0117:40
thingeeI think just not adding that to the iqn would fix it17:40
thingeefor now17:40
DuncanTVolume names can be up to 255 characters long...17:41
jgriffiththingee: so what I'm saying is the combination of your iqn length etc is what seems to push it over the limit for crypt17:41
thingeejgriffith: yup, I'm saying the same too17:41
jgriffiththingee: also the dsvm-full logs from the gate don't show this trace that I found in ER17:41
jgriffiththingee: ahh... ok :)17:41
thingeejgriffith: how cinder constructs a device + iqn17:42
thingeedevice name*17:42
*** angela-s has joined #openstack-cinder17:42
thingeeDuncanT: yeah, but with the way cinder/brick constructs the device name, it ends being longer17:42
DuncanTDo we know how long dmcrypt can take?17:43
thingeemtreinish, jgriffith, DuncanT: so technically, this bug can be triggered with a longer volume name17:43
thingeemtreinish: DuncanT saying the volume name can be 255 chars.17:43
thingeemtreinish: if the backend is passing the volume name in the iqn I guess17:44
thingeeforgot about that part17:44
jgriffiththingee: mtreinish DuncanT FWIW, what I'm getting at is there's a bunch of extraneous 'stuff' in the datera iqn prefix17:44
*** jordanP has quit IRC17:44
*** afazekas has quit IRC17:44
jgriffiththingee: mtreinish DuncanT no need for IP address IMO17:44
jgriffiththingee: mtreinish DuncanT also you really shouldn't use Name anyway17:45
jgriffiththingee: mtreinish DuncanT names aren't unique17:45
thingeejgriffith: that's the device name that comes from brick17:45
DuncanTjgriffith: Agreed. We have the id for uniqueness17:45
thingeejgriffith: /dev/disk-by-path/ip-<portal>-<datera-iqn>-lun-017:45
thingeejgriffith: weird, datera iqn is
jgriffiththingee: oh.. yeah, I see what you're saying17:46
jgriffith^^ LVM17:46
*** jungleboyj has quit IRC17:46
thingeejgriffith, DuncanT:
jgriffiththingee: If you want my advice (and you probably don't) I would not use volume-name17:47
jgriffithor the display name at any rate17:48
hemnabrick issues ?17:48
hemnaor ?17:48
jgriffithuse Cinder's unique identifiers17:48
hemnathat looks like a target iqn ?17:48
openstackLaunchpad bug 1432490 in tempest "TestEncryptedCinderVolumes cryptsetup name is too long" [Undecided,New]17:48
*** jungleboyj has joined #openstack-cinder17:49
hemnaah yah, that's the iqn that comes from the driver17:49
hemnabrick constructs the /dev/disk/by-path entry to look for it showing up17:49
thingeehemna: not, that's the device name that comes from brick which contains the iqn from the driver17:49
hemnayah, saying the same thing17:49
thingeesort of :)17:49
hemnabrick constructs that path string based upon the target portal iqn information coming from the driver17:50
hemnaas that's what the kernel will dynamically populate in /dev/disk/by-path17:50
*** jistr has quit IRC17:51
jgriffithhemna: thingee DuncanT Added a comment to the bug since the discussion here is just repetitive and confusing to some :)17:51
hemnaso cryptsetup hates that long string ?17:51
thingeejgriffith: pretty sure uuid from cinder would reach a limit too17:51
thingeejgriffith: if it was in the iqn17:51
jgriffiththingee: nope17:51
jgriffiththingee: not if you cut that ridiculously long display-name out of it17:51
jgriffiththingee: otherwise everybody would be failing for the same thing17:51
jgriffiththingee: the name 'datera" isn't exactly *long* :)17:52
thingeejgriffith: ah you're right17:52
hemnaOpenStack-TestEncryptedCinderVolumes-676292884:01:sn:aef6a6f1cd84768f   that's a bit longer than a uuid17:52
jgriffiththingee: it happens once in a while... it's actually that I just get lucky sometimes17:52
jgriffiththingee: which is better than smart :)17:52
thingeehemna: technically it's just OpenStack-TestEncryptedCinderVolumes-67629288417:53
*** BharatK has joined #openstack-cinder17:53
thingeejgriffith: cool thanks, I'll suggest that internally17:53
thingeeI'll just close out the bug for now17:53
jgriffiththingee: if it's not doable shout, we can come up with an algorithm pretty easy to condense that17:53
jgriffiththingee: nahh... you should leave the bug open IMO17:53
hemnaanother way to fix this17:54
hemnais to have cryptsetup use the actual /dev/SDX device17:54
jgriffithhemna: NOOOOOO17:54
hemnawhich is linked to the /dev/disk/by-path entry17:54
thingeemtreinish: updated the bug17:54
jgriffithhemna: so FWIW there's funny stuff that goes on with the sdx files17:54
mtreinishthingee: cool, thanks17:54
jgriffithhemna: that I can't convince anybody of17:54
hemnaeducate me17:55
jgriffithhemna: yeah, like sometimes they "move" or get "remapped"17:55
hemnaah hence the need for /dev/mapper crap17:55
hemnayah there is that17:55
jgriffithhemna: so it's "ok" to do that sort of thing if everyone/everything is cognisant17:56
hemnabut in the case of crypsetup, it's simply initializing the crypto crap on the raw device isn't it?17:56
hemnaso if it moves, then it should be still ok?17:56
hemnano idea really, just guessing17:56
jgriffithin theory :)17:56
jgriffithbut when sdx files get unmapped and remapped things sometimes go a bit "wonky"17:56
*** Apoorva has quit IRC17:57
hemnatotally agree with that17:57
jgriffithbesides, why change the structure of how the cryptsetup stuff works when it's a simple fix on the other end?17:57
jgriffithhemna: I dunno, maybe I'm paranoid.  Suppose if folks want to go that route it might be fine17:57
hemnaI dunno, was just thinking out of the box a bit17:57
jgriffithhemna: yeah... and a month ago I would've said... "cool"17:58
hemnathen it doesn't really matter what the iqn pattern is for cinder volumes.17:58
jgriffithhemna: but after looking at the detach/rescue test I'm terrified17:58
hemnabut that length might pose issues elsewhere as well17:58
*** ronis has joined #openstack-cinder17:59
openstackgerrityatin proposed openstack/python-cinderclient: cinder list now prints dash '-' when data is None
*** jdurgin has joined #openstack-cinder18:02
thingeehemna, jgriffith: so wanted to talk about multi-attach again. I still don't agree with the current workflow and wanted to raise it on the ops ML since I already started initial discussions at the ops meetup, which i raised before things were merged.18:02
hemnaI'm not sure what the issue is really18:03
hemnaif this is the attachment_id thing18:03
jgriffiththingee: FWIW I did make comments regarding workflow, but I wanted the work to move forward and not get stuck for another release18:03
hemnano one is using cinder to do detaches directly.18:03
hemnanova volume-detach <nova uuid> <cinder uuid>18:03
thingeehemna, jgriffith: I would like to propose following a format that morganfainberg told me about in keystone and that's listing the API as experimental. Meaning use at your own risk. We can make backwards compat changes.18:03
hemnaoperators shouldn't even care18:03
hemnaand don't really.18:04
jgriffiththingee: ?18:04
hemnaI don't think that makes sense to me.18:04
thingeehemna: why?18:04
thingeehemna: when you introduce looking for an attachment_id, that's changing the flow18:05
hemnabecause no one is going to call cinder directly to detach a volume.18:05
hemnait simply won't work.18:05
jgriffithhemna: thingee sorry... can you guys back up a second?18:05
hemnayou have to coordinate w/ nova for anything meaningful to happen.18:05
thingeehemna: doesn't matter. we always assume someone is doing something weird with the api18:05
jgriffithhemna: thingee I think the two of you may have knowledge that I don't (like what you're talking about) :)18:05
hemnaoperators aren't always right.18:06
hemnathis is the most flexible way to ensure the correct attachment is detached.18:06
* jgriffith still doesn't actually know what's being discussed so keeping quiet18:06
thingeejgriffith: hemna please correct me if I'm wrong on any of this, and I really hope I am. So if you have multiple attachments going on, you have to now look up the instance, the volume you want to detach, then look through a list of attachment ids, for one that matches the attachment you want to detach18:07
thingeejgriffith, hemna: then detach with that attachment id18:07
hemnathingee, you already have to lookup the instance you want to detach18:07
thingeejgriffith: hemna mentioned we follow this format because of how vmware does things18:07
thingeehemna: yup18:07
hemnanova volume-detach <instance uuid> <cinder volume>18:07
thingeeI think the attachment id thing is not necessary18:07
hemnaand if we are just talking about cinder18:08
hemnathen you STILL have to know the instance uuid you want to detach18:08
hemnaso you aren't gaining anything18:08
hemnathe volume has a list of attachments18:08
jgriffithsorry... I'm confused18:08
hemnayou can't just say, hey cinder, detach18:08
hemnadetach what from what ?18:08
hemnayou have to know the instance uuid anyway18:08
hemnathe new api makes it clean18:08
jgriffithyou don't like "nova volume-detach <instance-uuid> <volume-uuid>"?18:08
jgriffithhow else would you do it?18:08
hemnaI don't get it.18:09
*** rushil_ has joined #openstack-cinder18:09
thingeejgriffith: I don'18:09
thingeet have a problem with that18:09
*** BharatK has quit IRC18:09
thingeethat's not what I'm talking about18:09
hemnajgriffith, so in the existing multiattach workflow18:10
hemnato detach a volume18:10
hemnafrom a specific instance18:10
thingeev1/<uuid>/os-admin-actions/os-force-detach body contains attachment_id18:10
hemnaor host18:10
hemnayou need to know the attachment id18:10
hemnaeach attachment for a volume has an attachment id18:10
thingeethis is tacked on. You already know the volume because it's in the attachment id18:10
jgriffithhemna: time out18:10
jgriffithlet's take one at a time for my sake :)18:10
jgriffiththingee: so having the attachment_id in the force18:10
hemnathat id references the specific entry in the volume_attachment table, which can be an instance_uuid or a hostname.18:10
jgriffiththingee: you could easily change that... add an optional instance_id18:11
thingeehemna: but you're passing a useless volume-uuid in the endpoint18:11
hemnaso, to detach a volume, you call detach with an optional attachment_id18:11
jgriffiththingee: do the lookup in Cinder18:11
thingeethis is just tacked on imo18:11
jgriffithif it's provided use it, else detach everything18:11
jgriffithok... I'll just let you guys handle it18:12
jgriffithI can't talk about multiple things at once... too confusing18:12
kaisersjgriffith: Solved the sos-ci issue we talked about, wanna know?18:12
kaisersjgriffith: oops18:12
kaisersjgriffith: not now :-D18:12
jgriffithkaisers: for sure!18:12
kaisersjgriffith: i'm working with openstack-dev/ci-sandbox project18:13
kaisersjgriffith: but down below in run_tempest.yml sos-ci expects the commit to come from the cinder project afaics18:14
thingeehemna: can you explain to me the workflow for multi-attach how you see it?18:14
kaisersjgriffith: so it fetches the wrong commit id18:14
jgriffithkaisers: Yup!18:14
kaisersjgriffith: will solve now...18:14
jgriffithkaisers: cool... makes sense18:14
hemnathingee, you fetch the volume.   it has a list of attachments.  find the attachment you want to detach, and call detach with that attachment id.18:14
hemnait's agnostic to hostname or instanceuuid.18:15
kaisersjgriffith: my working replies to gerrit probably came from tests that went wrong and where it did not to that lookup because stuff went bust beforehand...18:15
*** Mandell has quit IRC18:15
hemnaif the attachment id isn't passed, cinder tries it's best to comply.18:15
hemnaif there is 1 attachment, cinder will detach.   else, it complains that it doesn't know which one you want to detach, since you didn't tell it which one.18:15
jgriffithkaisers: interesting... some day I need to get some good error handling and unit tests in there18:16
jgriffithkaisers: and update the config info... it's painfully out of date18:16
kaisersjgriffith: Anyways, thanks for the hint earlier on!18:16
jgriffithkaisers: honestly I didn't think anybody else would use it :)18:16
*** sgotliv has quit IRC18:16
jgriffithkaisers: but... check out these stats for March:18:16
kaisersjgriffith: Never think things like that18:16
kaisersjgriffith: if it's out there, it is out there!!!18:16
*** IanGovett has quit IRC18:16
*** rushiagr_away is now known as rushiagr18:17
jgriffithkaisers: for March.  185/1118:17
*** IanGovett has joined #openstack-cinder18:17
jgriffithkaisers: that's 185 succesful sos-ci runs, 11 fails18:17
jgriffithkaisers: not bad numbers18:17
kaisersjgriffith: yep18:17
*** IanGovett has quit IRC18:18
jgriffithkaisers: and I don't have to "manage" it so to speak18:18
jgriffithkaisers: or hire somebody to manage it for me18:18
kaisersjgriffith: And i did invest quite some time in other approaches but from my perspective they were to complex. That's why i looked into it18:18
kaisersjgriffith: :)18:18
jgriffithkaisers: yeah, that's why I wrote it.  Anyway... glad to hear it's useful for ya18:19
jgriffithkaisers: FWIW I gladly accept pull requests :)18:19
kaisersjgriffith: yep, thanks & read you later18:19
jgriffithkaisers: have a good one!18:19
kaisersjgriffith: i just removed some i made weeks ago18:19
kaisersjgriffith: yeah, one should be up, right18:19
*** paul-- has quit IRC18:20
kaisersmight cherry pick one or two others, have to see if i can disentangle them from my stuff18:20
kaisersjgriffith: nope, nonsense, currently there is none up from me, sry18:20
jgriffithno worries18:20
kaisersok, I'll let loose the beast on cinder now18:21
*** Mandell has joined #openstack-cinder18:21
kaisersjgriffith: for testing :)18:21
*** jungleboyj has quit IRC18:21
*** jungleboyj has joined #openstack-cinder18:22
*** BharatK has joined #openstack-cinder18:23
*** dflorea has quit IRC18:24
*** dflorea has joined #openstack-cinder18:25
*** jecarey has quit IRC18:25
*** dflorea has quit IRC18:27
*** dflorea has joined #openstack-cinder18:27
openstackgerritTomoki Sekiyama proposed openstack/python-cinderclient: Implement reset-state for attach_status and migration_status
mtaninojungleboyj: ping,18:28
*** paul-- has joined #openstack-cinder18:29
*** timcl1 has joined #openstack-cinder18:29
*** timcl has quit IRC18:30
mtaninojungleboyj: Can we push this ( ) to juno/stable? we need one more +2 from someone?18:30
*** Yogi11 has quit IRC18:30
openstackgerritThang Pham proposed openstack/cinder: Complete switch to snapshot objects
*** dflorea has quit IRC18:32
*** IanGovett has joined #openstack-cinder18:32
openstackgerritThang Pham proposed openstack/cinder: Complete switch to snapshot objects
*** Maike has joined #openstack-cinder18:34
*** Mandell has quit IRC18:34
*** Yogi1 has joined #openstack-cinder18:35
*** timcl1 has quit IRC18:35
thingeehemna: so you pass the attachment_id to nova volume-detach?18:36
*** timcl has joined #openstack-cinder18:36
*** jungleboyj has quit IRC18:38
*** jungleboyj has joined #openstack-cinder18:39
*** emagana has joined #openstack-cinder18:41
hemnanikesh_vedams, when you migrate a volume, it potentially turns into a retype18:43
openstackgerritMitsuhiro Tanino proposed openstack/cinder: Make unused iscsi_num_targets, iser_num_targets configs as deprecated
*** annashen has quit IRC18:44
openstackgerritWalter A. Boring IV (hemna) proposed openstack/python-cinderclient: Update to change name for volume type client
*** annashen has joined #openstack-cinder18:45
*** annashen has quit IRC18:55
*** afazekas has joined #openstack-cinder18:57
*** Apoorva has joined #openstack-cinder18:57
*** annashen has joined #openstack-cinder18:58
nikesh_vedamshemna:hmm bit confuse with these two sentences "DuncanT -> Migrate - move between backends of the same volume type"  and "when you migrate a volume, it potentially turns into a retype"  :)18:59
thingeehemna: so you pass the attachment_id to nova volume-detach?19:01
hemnathingee, no19:01
hemnanova volume-detach <instance uuid> <volume uuid>19:01
hemnanova does the rest.19:01
kmartinnikesh_vedams, I can help you, what do you not understand?19:01
thingeehemna: ok good19:01
hemnathingee, part of the nova work is to have nova get the attachment_id from the attachments in the volume and pass the right attachment_id along.19:02
*** Apoorva has quit IRC19:02
SwansonSo I up quota to 100 volumes.  I create 10 volumes just fine.  The 11 fails with a Quota exceeded (10 volumes already consumed) in c-api.19:03
SwansonBug or is there something else I need to do to get the quota over 10?19:03
hemnathingee, also, if the attachment_id isn't passed, cinder tries to detach.   if there is only 1 attachment, it will detach.19:05
hemnaif there are more than 1, it balks.19:05
hemnathingee, jgriffith suggested at one point, we could have it simply untattach all, if no attachment_id is passed.19:05
hemnathingee, for now I opted to be 'safe', and not do that.19:06
hemnawe can revisit in L if you like.19:06
*** BharatK has quit IRC19:07
jgriffithhemna: ahh... yeah, I noted that in the comments at one point19:07
jgriffithhemna: honestly I also think that getting the code merged at this point was more important19:07
jgriffithhemna: I'd rather adjust some details later if possible19:07
jgriffithhemna: don't want to block this "again"19:07
jgriffithjust don't let anybody implement it yet, and since it's not in Nova that shouldn't be a problem19:08
*** lpetrut has quit IRC19:08
jgriffithlunch... bbl19:09
thingeejgriffith: so that was the point of marking it as expermenital now. If we decide later we want to change something about the API, there was a note in the doc string and release notes about it being experimental19:09
hemnajgriffith, +1   and the scheduler check is in as well.  unless a driver reports multiattach: true as a capability, the scheduler will puke on create 'no host found'.19:09
thingeejgriffith: it's what keystone folks are doing now for new APIs19:09
jgriffiththingee: sure, but really it's irrelevant for the most part I think since it can't be used anyway19:09
jgriffiththingee: cool19:09
thingeejgriffith: I always assume folks will use things.19:09
thingeejgriffith: even if I don't mean them to19:10
jgriffiththingee: fair19:10
hemnathingee, how do you mark the API as experimental ?19:10
thingeehemna: morganfainberg just told me in the doc string of the code itself, release notes and any api ref doc on it (which we won't have yet)19:10
morganfainbergthingee, we aslo include hints in our JSON Home doc19:11
*** annashen has quit IRC19:11
thingeejson home doc?19:11
*** annashen has joined #openstack-cinder19:11
morganfainbergif you do a get on the keystone root / you can get json home doc outlineing our APIs19:11
thingeemorganfainberg: nice19:12
hemnathingee, url ?19:12
hemnaI can add a comment block about it as a follow up patch if you like, if that's all it is.19:12
thingeehemna: url to what?19:13
thingeean example?19:13
hemnathingee, yes please19:13
thingeemorganfainberg: got an example?19:13
thingeefor hemna19:13
*** afazekas has quit IRC19:13
*** vilobhmm has quit IRC19:14
*** vilobhmm has joined #openstack-cinder19:15
hemnamorganfainberg, the json_home.py19:15
hemnanot sure how that translates to cinder's contrib code19:15
morganfainberghemna, hm. let me find an example19:15
hemnacinder's detach isn't experimental19:16
hemnabut passing in the attachment_id is. :P19:16
hemnathingee, what would you like me to do ?19:16
morganfainbergright. so i'd just mark in documentation and code as such.19:16
hemnajust comment in the code/19:16
morganfainbergdoesn't map to a whole API.19:16
*** leeantho has quit IRC19:21
*** afazekas has joined #openstack-cinder19:21
*** HoloIRCUser has joined #openstack-cinder19:21
*** HoloIRCUser has quit IRC19:21
thingeehemna: as morganfainberg mentioned, I'm fine with the doc just mentioning the attachment_id piece for multi-attach.19:22
thingeehemna: I can do the same for release notes later if you agree.19:22
thingeehemna: do we still have time to do the brick cut?19:22
thingeehemna: things are less hectic for me to help with that now19:22
hemnaat this point, it would be pushing it.19:23
hemnawe need brick to get a release to pypi19:23
hemnathen global requirements change19:23
hemnathen my cinder patch taken off of -2 and WIP19:23
hemnalots of moving parts19:23
hemnafwiw, there are no outstanding os-brick patches19:24
thingeejgriffith: on sos-ci, I worked on a thing that uses pushover notifications to do push notifications to ios/android. Pretty small change, not sure if you're interested in introducing every use case in the project, but thought I would ask before I did a PR.19:25
hemnais there a way to filter all cinder patches to see if anyone is touching cinder/brick/* ?19:25
*** annashen has quit IRC19:25
*** crose has joined #openstack-cinder19:26
thingeehemna: in gerrit you can do search for file:^Regex19:27
*** annashen has joined #openstack-cinder19:28
tsekiyamahemna: project:openstack/cinder file:^cinder/brick/.* status:open19:28
thingeetsekiyama: :)19:28
*** timcl1 has joined #openstack-cinder19:29
*** bswartz has quit IRC19:30
*** afazekas has quit IRC19:30
*** annashen has quit IRC19:31
*** timcl has quit IRC19:32
*** annashen has joined #openstack-cinder19:32
*** dflorea has joined #openstack-cinder19:32
*** crose has quit IRC19:34
*** annashen has quit IRC19:37
*** annashen has joined #openstack-cinder19:37
*** dflorea has quit IRC19:37
*** mriedem has quit IRC19:39
*** rushiagr is now known as rushiagr_away19:42
*** enterprisedc has quit IRC19:42
thingeebbl lunch19:43
*** leeantho has joined #openstack-cinder19:43
*** mriedem has joined #openstack-cinder19:47
*** alexpilotti has quit IRC19:50
*** Apoorva has joined #openstack-cinder19:50
*** Mandell has joined #openstack-cinder19:51
*** sgotliv has joined #openstack-cinder19:52
*** xyang1 has joined #openstack-cinder19:53
*** xyang has joined #openstack-cinder19:54
*** nellysmitt has quit IRC19:54
*** afazekas has joined #openstack-cinder19:57
*** r-daneel has joined #openstack-cinder19:58
jbernardthangp: heya20:00
jbernardthangp: i believe you're working on cinder objects, no?20:01
thangpjbernard: yup20:01
jbernardthangp: im getting a strange import error maybe you can explain20:01
jbernardthangp: from brick/initiator/connector im attempting to import a volume driver20:03
thangpjbernard: it's probably coming from
thangpjbernard: 1 sec20:05
jbernardthangp: i see this when starting cinder-volume20:05
jbernardthangp: sure20:05
thangpjbernard: ah ok...the traceback is coming from the object serializer20:06
*** dflorea has joined #openstack-cinder20:06
*** rushiagr_away is now known as rushiagr20:07
thangpjbernard: what is the value of CONF.volume_api_class?20:07
jbernardthangp: at which point?20:08
thangp  File "/vagrant/cinder/cinder/volume/", line 27, in <module>20:08
thangp    API = importutils.import_class(CONF.volume_api_class)20:08
jbernardthangp: ahh, one sec20:08
thangpi'm just trying to trace back20:08
jbernard(Pdb) print CONF.volume_api_class20:09
*** sgotliv has quit IRC20:10
thangpjbernard: did you make any changes to cinder/brick/initiator/
jbernarda few ;)20:10
jbernardwhich all seem to be okay, except the imports20:11
thangpjbernard: could you share, at least the imports20:11
jbernardso 'from cinder.volume.drivers.rbd import RBDDriver'20:11
jbernardthat is when i begin to see this traceback20:11
thangpjbernard: that's it? hmm...20:11
jbernardthangp: i can paste the entire diff if you need20:12
jbernardthangp: or the whole patchset, let me know what you need20:12
thangplet me look in inder.volume.drivers.rbd20:12
thangpjbernard: sure20:12
hemnathingee, tsekiyama thank you20:12
jbernardthangp: could a circular import cause such a thing?20:13
thangpjbernard: maybe, but i'm not sure where20:14
thangpjbernard: cinder/volume/ isnt doing anything special with base20:14
*** sgotliv has joined #openstack-cinder20:14
*** afazekas has quit IRC20:15
*** ndipanov has quit IRC20:16
*** afazekas has joined #openstack-cinder20:16
*** lpabon has quit IRC20:16
*** alexpilotti has joined #openstack-cinder20:19
*** timcl1 has quit IRC20:19
*** jungleboyj has quit IRC20:21
*** rushil__ has joined #openstack-cinder20:21
*** timcl has joined #openstack-cinder20:22
*** jungleboyj has joined #openstack-cinder20:22
*** afazekas has quit IRC20:23
*** rushil has quit IRC20:23
thangpjbernard: try removing the import - from cinder.objects import base as objects_base and instead of using objects_base, using objects.base in cinder/cinder/volume/api.py20:24
openstackgerritVipin Balachandran proposed openstack/cinder: VMware: Fail immediately for images in a container
*** lpetrut has joined #openstack-cinder20:25
*** timcl has quit IRC20:26
*** jungleboyj has quit IRC20:27
*** rushiagr is now known as rushiagr_away20:27
*** jungleboyj has joined #openstack-cinder20:28
thangpjbernard: so I tried that myself, after a couple of changes, it now complains with File "/opt/stack/cinder/cinder/volume/", line 45, in <module>20:28
thangp    from cinder import utils ImportError: cannot import name utils20:28
thangpjbernard: there's something odd with importing RBDriver in connector.py20:29
openstackgerritRushil Chugh proposed openstack/cinder: Fixing mount when state_path is configured with a final '/'
jbernardthangp: yeah, im seeing similar results20:30
*** julim has quit IRC20:31
*** jungleboyj has quit IRC20:33
*** jungleboyj has joined #openstack-cinder20:33
openstackgerritRushil Chugh proposed openstack/cinder: Fixing mount when state_path is configured with a final '/'
*** afazekas has joined #openstack-cinder20:38
*** e0ne has joined #openstack-cinder20:39
*** annashen has quit IRC20:40
thangpjbernard: can you try to import RBDDriver in InitiatorConnector.__init__ - it worked for me20:42
thangpjbernard: I think there is a circular dependency somewhere20:42
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
*** jcru has quit IRC20:44
jbernardthangp: yep, one sec20:44
*** jcru has joined #openstack-cinder20:46
*** afazekas has quit IRC20:47
jbernardthangp: you mean import at the class level?20:49
thangpjbernard: yup20:49
jbernardthangp: would that be allowed?20:49
*** afazekas has joined #openstack-cinder20:49
*** e0ne has quit IRC20:49
thangpjbernard: depends on what you're doing :)20:49
thingeerhe00: hey are the ci stuff going? have you been getting the help you need?20:49
thingeehey how are*20:50
jbernardthangp: ok :) well at least it will unblock my current path20:50
jbernardthangp: thanks for your time20:50
thangpjbernard: but I think there are some dependencies that is pulling that is not there yet20:50
openstackgerritOpenStack Proposal Bot proposed openstack/python-cinderclient: Updated from global requirements
*** Yogi1 has quit IRC20:50
thangpjbernard: np20:50
*** annashen has joined #openstack-cinder20:51
jbernardthangp: importing within the class still breaks for me20:53
thangpjbernard: ^%20:54
jbernardthangp: will that support 'from X import Y' syntax as well?20:55
*** dustins has quit IRC20:56
thangpyeah, just do __import__('x.y')20:56
*** jaypipes has quit IRC20:56
*** jaypipes has joined #openstack-cinder20:57
jbernardthangp: oh man, that's pretty sneaky20:57
jbernardthangp: ;)20:57
thangpjbernard: :)20:57
jbernardthangp: also, i worry that kittens will be killed when i save this buffer20:57
*** boichev has quit IRC20:58
*** harlowja is now known as harlowja_away20:58
thangpjbernard: try just __import__('cinder.volume.drivers.rbd')21:01
*** dflorea has quit IRC21:01
thangpjbernard: that should pull everything in rdb21:01
jbernardthangp: ahh, that works21:02
thangpjbernard: cool21:02
jbernardthangp: i think, one sec ;)21:02
jbernardthangp: so, i should be able to access RBDDriver as 'rbd.RBDDriver', right?21:03
thangpjbernard: yup21:03
jbernardthangp: assuming the latest import21:03
*** Mandell has quit IRC21:03
mtaninohemna: thank you for the review :)21:05
hemnamtanino, np :)21:05
thangpjbernard: oh, __import__('cinder.volume.drivers.rbd', globals=globals())21:06
thangpjbernard: or just assign a var to it21:07
thangpjbernard: then just use it, e.g. rdb = __import__()  rdb.RDBDriver()21:07
jbernardthangp: ok, i need to read about that21:08
jbernardthangp: thanks again!21:08
thangpjbernard: np21:08
*** akerr has quit IRC21:10
*** annashen has quit IRC21:12
hemnasmcgindia, so there was no way around the \ in ?21:12
*** tellesnobrega has quit IRC21:13
smcgindiahemna: Unfortunately no. :[21:14
smcgindiahemna: Things go boom without it when trying to use ().21:15
hemnasmcgindia, dang ok.  not a big deal.  it's more readable than it was previously anyway21:15
*** rushil__ has quit IRC21:15
hemnaI was about to suggest using ()21:15
smcgindiahemna: I think so. And safer.21:15
smcgindiahemna: ;)21:15
hemnabut if it pukes....bleh21:15
smcgindiahemna: I think if you look back in the comments I have a paste of some tests I did there.21:15
*** thangp has quit IRC21:16
hemnaheh Eric complained about it too.21:16
smcgindiahemna: Yeah, I knew it was going to happen so I put the note in the commit.21:16
*** afazekas has quit IRC21:16
hemnaI'm kew w/ it.21:16
smcgindiahemna: But of course got the comments anyway. :)21:16
hemnathanks man21:16
smcgindiahemna: Thanks for looking at it!21:16
*** tellesnobrega has joined #openstack-cinder21:18
*** Mandell has joined #openstack-cinder21:18
*** rushil_ has quit IRC21:19
*** dflorea has joined #openstack-cinder21:22
*** lpetrut has quit IRC21:23
*** jungleboyj has quit IRC21:23
*** afazekas has joined #openstack-cinder21:23
*** afazekas has quit IRC21:29
*** annashen has joined #openstack-cinder21:30
*** hemna has quit IRC21:32
*** mriedem is now known as mriedem_away21:33
*** jcru has quit IRC21:33
*** bswartz has joined #openstack-cinder21:42
*** sgotliv has quit IRC21:42
*** hemna has joined #openstack-cinder21:43
*** hemna has quit IRC21:47
*** hemna has joined #openstack-cinder21:52
*** dflorea has quit IRC21:59
*** Lee1092 has quit IRC22:00
*** ronis has quit IRC22:03
*** dflorea has joined #openstack-cinder22:04
*** Maike has quit IRC22:05
*** harlowja_away is now known as harlowja22:11
*** annashen has quit IRC22:14
*** annashen has joined #openstack-cinder22:15
*** dflorea has quit IRC22:18
*** dflorea_ has joined #openstack-cinder22:21
*** Mandell has quit IRC22:24
*** Mandell has joined #openstack-cinder22:29
openstackgerritRichard Hedlind proposed openstack/cinder: Decrement remaining retries after failed REST call.
openstackgerritWalter A. Boring IV (hemna) proposed openstack/cinder: WIP: Remove Brick from cinder codebase
hemnajust rebasing22:37
*** xyang has quit IRC22:42
openstackgerritxing-yang proposed openstack/cinder: Don't override extra specs with config in VMAX
*** tsekiyam_ has joined #openstack-cinder22:52
*** tsekiyama has quit IRC22:55
*** sgotliv has joined #openstack-cinder22:56
*** dflorea_ has quit IRC22:57
*** tsekiyam_ has quit IRC22:57
*** dflorea has joined #openstack-cinder23:01
*** dflorea has quit IRC23:02
*** dflorea has joined #openstack-cinder23:03
*** patrickeast_ has joined #openstack-cinder23:03
*** patrickeast has quit IRC23:04
*** patrickeast_ is now known as patrickeast23:04
*** dflorea has quit IRC23:07
*** sgotliv has quit IRC23:10
*** hemna is now known as hemnafk23:13
*** dflorea has joined #openstack-cinder23:13
*** ganso_ has quit IRC23:17
*** r-daneel has quit IRC23:19
*** lnr has joined #openstack-cinder23:30
*** lnr has left #openstack-cinder23:31
*** jaypipes has quit IRC23:41
*** r-daneel has joined #openstack-cinder23:42
*** r-daneel has quit IRC23:47
*** dannywilson has quit IRC23:51
*** alexpilotti has quit IRC23:57
*** ho has joined #openstack-cinder23:58

Generated by 2.14.0 by Marius Gedminas - find it at!