Thursday, 2015-06-25

*** salv-orlando has joined #openstack-cinder00:05
*** david-ly_ has joined #openstack-cinder00:06
*** jwcroppe has quit IRC00:07
*** jwcroppe has joined #openstack-cinder00:07
*** akerr_ has quit IRC00:08
*** david-lyle has quit IRC00:08
*** jwcroppe has quit IRC00:10
*** akerr has joined #openstack-cinder00:10
*** garthb_ has quit IRC00:11
*** garthb has quit IRC00:11
*** salv-orlando has quit IRC00:12
*** Longgeek has joined #openstack-cinder00:12
*** jwcroppe has joined #openstack-cinder00:13
*** dims_ has joined #openstack-cinder00:14
*** esker has quit IRC00:15
*** edmondsw has joined #openstack-cinder00:16
*** gouthamr has joined #openstack-cinder00:16
*** dims has quit IRC00:16
*** edmondsw_ has joined #openstack-cinder00:16
*** edmondsw_ has quit IRC00:17
*** leeantho has quit IRC00:17
harlowjaanyone know if has/is being fixed ?00:20
harlowjamight be something new (?)00:21
*** salv-orlando has joined #openstack-cinder00:21
*** Longgeek has quit IRC00:23
*** esker has joined #openstack-cinder00:25
*** salv-orlando has quit IRC00:27
*** rushiagr_away is now known as rushiagr00:32
*** rwsu has quit IRC00:38
*** boris-42 has quit IRC00:42
mtaninoharlowja: oslo.service is imported today. so I guess it is new.
*** mriedem has quit IRC00:45
harlowjamtanino interesting, wonder how that passed/merged :-/00:46
harlowjamagically somehow, ha00:47
mtaninohmm, what kind of magic happened....00:47
*** jwcroppe_ has joined #openstack-cinder00:47
*** edmondsw has quit IRC00:47
* harlowja just was wondering how happened, shouldn't 193951 not have merged if that was possible00:48
*** diogogmt has quit IRC00:48
*** r-daneel has quit IRC00:48
mtaninoCan you reproduce it in your devstack environment?00:49
*** rushiagr is now known as rushiagr_away00:49
*** jwcroppe has quit IRC00:49
harlowjadon't really have to, can just follow the code :)00:52
*** zhenguo has joined #openstack-cinder00:52
*** tobe has joined #openstack-cinder00:56
openstackgerritJoshua Harlow proposed openstack/cinder: Ensure 'WSGIService' derives from oslo_service base class
*** Rockyg has joined #openstack-cinder01:05
*** smoriya has joined #openstack-cinder01:06
*** DericHorn-HP has joined #openstack-cinder01:16
mtaninoThis is not happen in my devstack environment... hmm01:18
winston-dmtanino, harlowja: i am not seeing the issue in my devstack01:23
harlowjahmmm, odd01:24
harlowjawhats up with that01:24
mtaninomee too.01:24
mtaninobut your proposed code seems fix the root cause.01:24
* harlowja that should be triggered, are u guys using the installed oslo.service ?01:25
*** laughterwym_ has joined #openstack-cinder01:25
*** Lee1092 has joined #openstack-cinder01:26
*** Yogi11 has joined #openstack-cinder01:27
harlowjawhich eventually gets called by ...01:27
harlowjaand all that01:27
mtaninoyes, oslo.service 0.1.0 is installed in my environment.01:27
harlowjawhich comes from
*** laughterwym has quit IRC01:29
*** davechen_away is now known as davechen01:31
*** DericHorn-HP has quit IRC01:31
harlowjawell anywho, seems like the right fix anyway01:34
harlowjaeven if somehow isn't being hit for u guys01:34
harlowjalikely would of anyway (if change caused that)01:34
*** david-ly_ is now known as david-lyle01:36
*** Yogi11 has quit IRC01:37
mtaninoahh, using master branch oslo.service causes error which you mentioned.01:40
*** _cjones_ has quit IRC01:41
mtaninoharlowja: And using your patch, the problem was solved :)01:41
mtaninoI guess same issue will be happen at nova?01:42
winston-dso CI is not using pip for oslo_service?01:45
*** dims_ has quit IRC01:45
mtaninofmm, this did not happen nova...01:48
mtaninoah, they still use nova.openstack.common.service01:49
*** davechen is now known as davechen_afk01:50
*** dims has joined #openstack-cinder01:50
mtaninowinston-d: not sure but using pip package(0.1.0), that error did not happen.01:51
winston-dmtanino: yeah, that's what I am seeing in my devstack01:53
mtaninomaster branch is 0.1.1.dev12 and this version causes that error in my devstack environment.01:53
mtaninoso if you uninstall pip version oslo.service and "python install" for oslo.service you will hit the error.01:54
harlowjawinston-d so what i think is happening is that hasn't propagated out yet01:54
harlowjaand the is an early warning system01:54
harlowjathat would of eventually hit everyone i guess01:54
harlowja*or will with the next oslo.service release01:54
winston-dharlowja: i thought it won't hit us until next oslo.service release01:55
harlowjaright, so it might not be an immediate fix needed, but it doesn't hurt01:56
*** dontalton has quit IRC01:56
harlowja*and avoids the pain in the near-future01:56
winston-dharlowja: your fix is good fix, no doubt about that.01:57
winston-dharlowja: but if you look at c-sch service code, it should have hit the issue:
earlephilhowerIs anyone else's CI nodepool dying today on a missing oslo_context during
winston-dneed to run for a meeting01:59
patrickeastearlephilhower: just checked, and yes, we are hitting that02:03
patrickeastearlephilhower: our image failed to build last night from it02:03
*** dims has quit IRC02:04
mtaninowinston-d: harlowja I think scheduler inherit Class service in oslo.service and the service class inherits the class Servicebase.02:04
mtaninoso the problem does not happen.02:04
akerrpatrickeast: how long ago?  Our latest image was built 22.8 hours ago (so almost due for a rebuild)02:05
earlephilhowerpatrickeast, Thanks for the confirmation!  I just moved to a new (stable I hope) host and can't get a new nodepool image, didn't change the VMs at all.   May steal one from old dead system and hack nodepool DB.02:06
patrickeastakerr: mmm roughly 15 hours ago, it tried to run at 2am pst02:06
akerrgoing to try and kick one off now02:07
patrickeastlets maybe go ask in -infra if they have seen this02:09
*** Yogi1 has joined #openstack-cinder02:09
*** dims has joined #openstack-cinder02:10
*** gouthamr has quit IRC02:10
*** DericHorn-HP has joined #openstack-cinder02:11
*** DericHorn-HP has quit IRC02:12
*** DericHorn-HP has joined #openstack-cinder02:14
*** Yogi11 has joined #openstack-cinder02:14
openstackgerritXiaolei Hu proposed openstack/cinder: Split FC driver code into Common and FC
*** DericHorn-HP has quit IRC02:17
*** Yogi1 has quit IRC02:17
*** gouthamr has joined #openstack-cinder02:21
*** Rockyg has quit IRC02:22
*** patrickeast has quit IRC02:22
openstackgerritXiaolei Hu proposed openstack/cinder: Split FC driver code into Common and FC
*** akerr has quit IRC02:25
*** salv-orl_ has joined #openstack-cinder02:29
*** bkopilov has quit IRC02:33
*** salv-orl_ has quit IRC02:34
*** laughterwym__ has joined #openstack-cinder02:35
*** laughterwym_ has quit IRC02:36
openstackgerritXi Yang proposed openstack/cinder: Provide snap copy feature in EMC VNX Cinder driver
*** diogogmt has joined #openstack-cinder02:40
*** dguryanov has quit IRC02:46
*** IanGovett has quit IRC02:49
*** gouthamr has quit IRC02:50
*** gouthamr has joined #openstack-cinder02:51
*** dims has quit IRC02:58
*** harlowja_at_home has joined #openstack-cinder03:01
*** akerr has joined #openstack-cinder03:02
*** akerr has quit IRC03:08
openstackgerritQian Gao proposed openstack/cinder: Storwize: add the missing stops in the end of the messages
*** Tross has quit IRC03:12
*** Tross has joined #openstack-cinder03:12
*** amit213 has quit IRC03:12
*** zul has quit IRC03:13
*** amit213 has joined #openstack-cinder03:13
*** gouthamr_ has joined #openstack-cinder03:16
*** zul has joined #openstack-cinder03:17
*** gouthamr has quit IRC03:20
*** Longgeek has joined #openstack-cinder03:20
*** Longgeek has quit IRC03:21
*** Longgeek has joined #openstack-cinder03:21
*** tobe has quit IRC03:26
*** mtanino has quit IRC03:31
*** Yogi11 has quit IRC03:35
*** gouthamr_ has quit IRC03:36
*** dannywil_ has joined #openstack-cinder03:36
*** diogogmt has quit IRC03:38
*** tobe has joined #openstack-cinder03:39
*** dannywilson has quit IRC03:40
*** dannywil_ has quit IRC03:40
*** bkopilov has joined #openstack-cinder03:54
*** coolsvap|away is now known as coolsvap04:04
*** heyun has joined #openstack-cinder04:07
*** vilobhmm has joined #openstack-cinder04:14
*** heyun1 has joined #openstack-cinder04:17
*** esker has quit IRC04:18
*** esker has joined #openstack-cinder04:18
*** hemna has joined #openstack-cinder04:19
*** MJ_ has joined #openstack-cinder04:19
MJ_unable to attach volume to instance in openstack kilo04:20
*** heyun has quit IRC04:20
MJ_debugging gives me following outpuy04:20
MJ_DEBUG (shell:914) The server has either erred or is incapable of performing the requested operation. (HTTP 500) (Request-ID: req-0fe16b68-0efa-408d-a8b5-c21513c78dea) Traceback (most recent call last):   File "/usr/lib/python2.7/dist-packages/novaclient/", line 911, in main     OpenStackComputeShell().main(argv)   File "/usr/lib/python2.7/dist-packages/novaclient/", line 838, in main     args.func(self.cs, args04:20
hemnalooks like a nova api issue04:21
MJ_how to resolve it?04:21
MJ_I am able to launch a vm so i guess issue with nova - cinder communication04:22
*** esker has quit IRC04:23
openstackgerritEdwin Wang proposed openstack/cinder: Split FC driver code into Common and FC
*** DericHorn-HP has joined #openstack-cinder04:28
*** laughterwym__ has quit IRC04:30
*** laughterwym has joined #openstack-cinder04:30
*** laughterwym_ has joined #openstack-cinder04:30
*** laughterwym_ has quit IRC04:31
*** yamada-h has joined #openstack-cinder04:31
*** shausy has joined #openstack-cinder04:33
*** annashen has joined #openstack-cinder04:36
laughterwymhemna: ping?04:36
laughterwymHemma: We proposed a change that split our old FlashSystem FC driver into common and FC part. So IMO it's not a new FC driver.04:37
*** Longgeek has quit IRC04:37
*** Longgeek has joined #openstack-cinder04:37
hemnait looks like a new FC driver to me04:38
hemnais there an FC CI for it already in place?04:38
laughterwymIt's running from Kilo.04:39
*** hemna has quit IRC04:40
laughterwymhemna: The original FC driver is merged in Kilo. This proposal is to break it into common part and then the new iSCSI driver can inherit from the common part.04:41
*** salv-orlando has joined #openstack-cinder04:41
*** hemna has joined #openstack-cinder04:42
hemnaok I'll check the CI results later04:42
hemnaI see something that resembles an FC based driver in the original common file04:42
laughterwymhemna: iSCSI driver has been merged in L-1. And you can see this a small change that is split from old
hemnawell, it's not a small change04:43
hemnaI had to look at the removed file to see the original FC code04:43
hemnavery confusing files04:43
hemnalooks like the refactor cleans it up a bit though :)04:44
*** Longgeek has quit IRC04:44
laughterwymhemna: Yes. It was confused before. So we'd like to make it clear - three parts: common, FC and iSCSI.04:44
*** shyama has joined #openstack-cinder04:45
*** MJ_ has quit IRC04:45
*** salv-orlando has quit IRC04:47
*** hemna has quit IRC05:05
*** annashen has quit IRC05:05
*** dannywilson has joined #openstack-cinder05:07
*** BharatK has joined #openstack-cinder05:12
*** Longgeek has joined #openstack-cinder05:17
*** esker has joined #openstack-cinder05:19
*** vilobhmm has quit IRC05:26
*** dannywilson has quit IRC05:30
*** dannywilson has joined #openstack-cinder05:31
*** esker has quit IRC05:32
*** shausy has quit IRC05:33
*** dannywilson has quit IRC05:35
*** nkrinner has joined #openstack-cinder05:39
*** nlevinki has joined #openstack-cinder05:41
*** markvoelker has quit IRC05:43
*** eduardo has quit IRC05:49
*** esker has joined #openstack-cinder05:59
*** laughterwym_ has joined #openstack-cinder06:02
*** DericHorn-HP has quit IRC06:04
*** ericksonsantos has quit IRC06:05
*** laughterwym has quit IRC06:05
*** BharatK has quit IRC06:07
*** agarciam has joined #openstack-cinder06:10
*** tobe has quit IRC06:14
*** changbl_ has quit IRC06:15
*** esker has quit IRC06:15
*** BharatK has joined #openstack-cinder06:19
*** deepakcs has joined #openstack-cinder06:21
*** nlevinki has quit IRC06:25
*** tobe has joined #openstack-cinder06:33
openstackgerritTeruaki Ishizaki proposed openstack/cinder: sheepdog: Add class for dog command executor
*** mkoderer_cloud has joined #openstack-cinder06:39
*** ian_brown has quit IRC06:40
*** esker has joined #openstack-cinder06:41
*** markvoelker has joined #openstack-cinder06:43
*** markvoelker has quit IRC06:49
*** belmoreira has joined #openstack-cinder06:53
*** salv-orlando has joined #openstack-cinder06:54
*** salv-orlando has quit IRC06:56
*** salv-orlando has joined #openstack-cinder06:56
*** esker has quit IRC06:56
*** aix has joined #openstack-cinder06:56
*** markus_z has joined #openstack-cinder06:57
*** lennyb has quit IRC07:03
*** root_ has joined #openstack-cinder07:07
*** root_ has quit IRC07:09
*** lennyb has joined #openstack-cinder07:12
*** anshul has joined #openstack-cinder07:14
*** lan has quit IRC07:20
*** lan has joined #openstack-cinder07:20
*** esker has joined #openstack-cinder07:23
openstackgerritTaoBai proposed openstack/cinder: test
*** esker has quit IRC07:38
*** BharatK has quit IRC07:39
*** lpetrut has joined #openstack-cinder07:41
*** jordanP has joined #openstack-cinder07:42
*** tobe has quit IRC07:48
*** BharatK has joined #openstack-cinder07:51
*** BharatK has quit IRC07:58
*** BharatK has joined #openstack-cinder07:59
openstackgerritTeruaki Ishizaki proposed openstack/cinder: sheepdog: Add class for dog command executor
*** tobe has joined #openstack-cinder08:04
*** esker has joined #openstack-cinder08:04
*** ndipanov has joined #openstack-cinder08:06
*** harlowja_still_a has joined #openstack-cinder08:10
aarefievwinston-d: hi, are you around?08:13
*** harlowja_at_home has quit IRC08:14
aarefievwinston-d: could you look at , small fix with resolving host with pool part08:16
*** coolsvap is now known as coolsvap|away08:16
*** deepakcs has quit IRC08:17
*** coolsvap|away is now known as coolsvap08:19
*** esker has quit IRC08:20
*** e0ne has joined #openstack-cinder08:32
*** markvoelker has joined #openstack-cinder08:32
*** e0ne has quit IRC08:37
*** markvoelker has quit IRC08:37
*** shyama has quit IRC08:37
*** haypo has joined #openstack-cinder08:38
*** laughterwym_ has quit IRC08:42
*** BharatK has quit IRC08:45
*** esker has joined #openstack-cinder08:46
*** yamada-h has quit IRC08:51
winston-daarefiev: sure08:52
winston-daarefiev: looking08:52
aarefievwinston-d: thank you08:53
*** salv-orl_ has joined #openstack-cinder08:53
winston-daarefiev: done08:54
aarefievwinston-d: you are fast :) , thanks08:55
*** salv-orlando has quit IRC08:57
*** salv-orl_ has quit IRC08:58
*** esker has quit IRC09:01
*** e0ne has joined #openstack-cinder09:10
*** shyama has joined #openstack-cinder09:14
*** e0ne is now known as e0ne_09:16
openstackgerritSergey Vilgelm proposed openstack/os-brick: Switch to oslo.service
openstackgerritTaoBai proposed openstack/cinder: Storwize driver report capability for multiattach
*** e0ne_ has quit IRC09:22
*** esker has joined #openstack-cinder09:28
*** lpetrut has quit IRC09:28
*** e0ne has joined #openstack-cinder09:33
*** laughterwym has joined #openstack-cinder09:36
*** BharatK has joined #openstack-cinder09:40
openstackgerritPranali Deore proposed openstack/cinder: Validate string, integer limit for input parameter
*** esker has quit IRC09:44
*** alexpilotti has joined #openstack-cinder09:45
*** atuvenie has joined #openstack-cinder09:50
openstackgerritMichal Dulko proposed openstack/python-cinderclient: Add set_management_url to cinderclient.client
*** salv-orlando has joined #openstack-cinder09:51
*** lixiaoy1 has quit IRC09:52
*** laughterwym_ has joined #openstack-cinder09:53
*** nlevinki has joined #openstack-cinder09:53
*** salv-orl_ has joined #openstack-cinder09:56
*** laughterwym has quit IRC09:56
*** lpetrut has joined #openstack-cinder09:57
*** salv-orlando has quit IRC09:58
*** salv-orl_ has quit IRC09:59
*** salv-orlando has joined #openstack-cinder09:59
*** salv-orl_ has joined #openstack-cinder10:01
*** salv-orlando has quit IRC10:03
aarefievdulek: hi, I've answered on your question
*** edmondsw has joined #openstack-cinder10:06
*** edmondsw has quit IRC10:06
*** edmondsw has joined #openstack-cinder10:07
*** yamada-h has joined #openstack-cinder10:09
*** esker has joined #openstack-cinder10:11
*** dims has joined #openstack-cinder10:14
*** BharatK has quit IRC10:15
*** husanu4 has joined #openstack-cinder10:19
*** markvoelker has joined #openstack-cinder10:21
openstackgerritGorka Eguileor proposed openstack/cinder: Add pagination to snapshots
*** husanu4 has quit IRC10:25
*** markvoelker has quit IRC10:25
*** husanu3 has joined #openstack-cinder10:26
*** husanu3 has quit IRC10:28
*** shausy has joined #openstack-cinder10:28
*** esker has quit IRC10:29
*** husanu3 has joined #openstack-cinder10:29
shausyHi Im facing problem with attaching volume to instance  [ Driver - Netapp iSCSI]10:30
shausyfollowing is my error10:30
shausy  ERROR nova.virt.block_device [req-21a265d0-1bc0-4bab-86f2-f2b6d167727b ef123e34625a466794faabc52fe0eda9 7e7a2c9f4bef469882f6808641dd9eab] [instance: a1364e74-9b40-46e6-aafc-a6a51a9caa1b] Driver failed to attach volume 310814f5-e564-46d1-a5aa-e064f3cb4e07 at /dev/vdb10:30
shausyAny help on this ?10:30
*** husanu3 has quit IRC10:31
*** e0ne is now known as e0ne_10:31
*** jasondot_ has joined #openstack-cinder10:32
*** e0ne_ has quit IRC10:36
*** jasondot_ is now known as jasondotstar10:37
openstackgerritSasikanth Eda proposed openstack/cinder: GPFS volume encryption-at-rest support
*** yamada-h_ has joined #openstack-cinder10:43
dulekaarefiev: Good point, I'll happily +1 the spec once tbarron's comments will be addressed.10:46
*** yamada-h has quit IRC10:47
*** juzuluag has quit IRC10:47
aarefievdulek: ok, thanks, I'm working om impl, I'll try to push it today and update the spec10:47
*** heyun1 has quit IRC10:52
*** esker has joined #openstack-cinder10:54
*** e0ne has joined #openstack-cinder10:55
*** IanGovett has joined #openstack-cinder10:59
*** timcl has joined #openstack-cinder11:00
openstackgerritDaniel Tadrzak proposed openstack/cinder: CGSnapshot Object
openstackgerritDaniel Tadrzak proposed openstack/cinder: ConsistencyGroup Object
*** dims has quit IRC11:02
*** breitz has quit IRC11:02
openstackgerritwanghao proposed openstack/cinder: Do not allow to modify access for public volume type
*** breitz has joined #openstack-cinder11:02
*** dims has joined #openstack-cinder11:05
*** shyama_ has joined #openstack-cinder11:06
*** shyama has quit IRC11:09
*** shyama_ is now known as shyama11:09
*** dims_ has joined #openstack-cinder11:10
*** dims has quit IRC11:11
*** tobe has quit IRC11:11
*** esker has quit IRC11:11
*** yamada-h_ has quit IRC11:14
*** salv-orl_ has quit IRC11:14
*** e0ne is now known as e0ne_11:15
*** dims_ has quit IRC11:16
openstackgerritDaniel Tadrzak proposed openstack/cinder: ConsistencyGroup Object
*** david-lyle has quit IRC11:17
*** dims has joined #openstack-cinder11:21
*** david-lyle has joined #openstack-cinder11:21
*** e0ne_ has quit IRC11:26
*** e0ne has joined #openstack-cinder11:28
*** ociuhandu has joined #openstack-cinder11:33
*** ericksonsantos has joined #openstack-cinder11:33
*** markvoelker has joined #openstack-cinder11:37
*** esker has joined #openstack-cinder11:37
*** EmilienM|off is now known as EmilienM11:38
*** ociuhandu_ has joined #openstack-cinder11:39
*** ociuhandu has quit IRC11:40
*** ociuhandu_ is now known as ociuhandu11:40
*** markvoelker has quit IRC11:41
*** sgotliv has quit IRC11:44
*** sgotliv has joined #openstack-cinder11:44
*** salv-orlando has joined #openstack-cinder11:49
*** jasondotstar has quit IRC11:50
*** marcusvrn has quit IRC11:53
*** esker has quit IRC11:54
geguileodulek: ping - Tooz to replace file locks11:57
dulekgeguileo: Hi!11:57
geguileodulek: Hi11:58
geguileodulek: I was wondering, have you thought about replacing current file locks with Tooz in parallel with the Nova/Cinder fixes?11:58
*** markvoelker has joined #openstack-cinder11:58
geguileodulek: I'm not talking about atomic state changes, just the file locks11:58
geguileodulek: For example that a snapshot is deleted while it's being used to create a volume11:59
dulekgeguileo: Ah, these kind of things...11:59
dulekgeguileo: I thought that such situation will be prevented by atomic state changes?12:00
dulekWe have something like that now:
geguileodulek: Not really, because in that case the snapshot would not change its state12:00
dulek(+dependent patch).12:00
dulekgeguileo: Ah, okay, you're right...12:00
geguileodulek: Yes I saw you patch, but maybe we should look at using it just to replace current file locks12:01
geguileodulek: And to include the missing lock you saw while working on volume create12:01
dulekgeguileo: Do you mean locks in volume.manager?12:02
geguileodulek: Since we should be possible to work in parallel with fixes to Cinder-Nova interaction12:02
dulekgeguileo: You mean that you're working on Nova error handling now?12:03
*** zhenguo has quit IRC12:04
dulekOkay, let me try to summarize what I know. ;)12:04
geguileodulek: No I'm not12:04
dulekI'm working on Nova->Cinder interactions now.12:04
geguileodulek: Ok, then I'll ask you some questions later  :)12:05
geguileodulek: about that nova-cinder12:05
dulekI've got a patch for attach almost ready, hoping to get it in review today.12:05
geguileodulek: For Nova or Cinder?12:05
dulekI'm probably not addressing all the issues, but at least Nova doesn't blow up when Cinder returns an error.12:05
dulekFor Nova.12:05
geguileodulek: Nice!12:06
dulekSzymon from my team is trying to reproduce some issues with locks in drivers. Right now we're stressing c-vol with GlusterFS driver working in A/A to see if things will get nasty (without distributed locks).12:07
geguileodulek: Ok12:08
dulekSo another activity you propose is to add locking for resources we're using (for example creating volume from snapshot, or snapshot from volume).12:09
geguileodulek: Wouldn't it be easier for a first step to replace utils.synchronized that use external with Tooz?12:09
*** agarciam has quit IRC12:09
geguileodulek: Yes, basically replacing external synchronized everywhere12:09
geguileodulek: Default can be to file locks (same as it is now)12:09
geguileodulek: And easily changed to a DLM12:10
dulekYeah, that's how our Tooz patch is supposed to work - default to files.12:10
dulekSo I'm not sure if we're actually locking the snapshot anywhere when creating a volume from it.12:10
dulekDo we?12:10
geguileodulek: There are some missing places12:11
geguileoI think you found one the other day in Volume Create12:11
geguileoAnd there's sure to be some more12:11
dulekOkay, this would require some thinking how to approach that probably...12:11
geguileoSo I think we could retake your patch and put it up to date12:11
dulekWhich one? Can you link it?12:11
*** marcusvrn has joined #openstack-cinder12:12
geguileoThen create another one istead of the POC: Tooz locks demo12:12
geguileoThat addresses all calls to synchronized with external12:12
geguileoAnd we would be no worse than we are now regarding Cinder-Nova interaction12:12
dulekSo you mean this one:
geguileoThen we can look for missing places where we have to add locks12:13
geguileodulek: Yes, retake that one12:13
dulekI'm not even sure why it has -1 from Jenkins...12:14
geguileodulek: Don't know12:14
geguileoIn parallel to finding those missing locks we can look into making sure all drivers have the right kind of locks (external if they need)12:15
dulekAh, requirements... Seems like CI failure.12:15
dulekOkay, I agree with you.12:16
geguileoYou are working on Cinder-Nova interactions now12:16
dulekFeel free to revive the patch and base your work on it. :)12:17
geguileoAre you going to keep working on that?12:17
dulekIf that's what you've asked actually. :P12:17
dulekI would love to stop it. :D It's a mess.12:17
geguileodulek: I was actually asking (1) if you think it's a good idea12:17
geguileodulek: (2) If you were already working on it or if you were going to12:17
dulekSo for last 2 weeks we tried to prepare tests.12:18
geguileodulek: Ok, so you'll keep working on N-C12:18
geguileoWhat kind of tests?12:18
dulekYeah, DuncanT had a little time for that.12:18
dulekSteps to reproduce issues without DLM - we've talked about that at the summit.12:18
geguileoYes, I know, but I mean: Rally? Tempest? Manual?12:19
dulekRight now we're trying to Rally.12:19
*** bswartz has quit IRC12:19
dulekIf it won't work we'll probably stop - it eats too much time. ;)12:19
geguileoSo no success up to now? :(12:20
*** esker has joined #openstack-cinder12:21
dulekYup, I'm not even sure what for are these locks in GlusterFS and RemoteFS.12:21
dulekhemnafk said that FC drivers brake without locks, but we don't have any FC gear.12:22
geguileoOk, because I was going to start working on tests to break Cinder and didn't want to redo work12:22
dulekSo were looking only on tests to prove that DLM is needed for locks *in drivers*.12:22
dulekWe haven't looked at any more.12:22
geguileoBy the way, the patch you'll be pushing today (if I leave you alone) is not about fixing Issue2 ( but just not breaking Nova on Cinder exceptions, right?12:24
dulekMore like:
openstackLaunchpad bug 1458958 in OpenStack Compute (nova) "Exceptions from Cinder detach volume API not handled" [Undecided,In progress] - Assigned to Vilobh Meshram (vilobhmm)12:26
dulekIt's mentioned on Issue2, but isn't the root case I think12:26
*** rakesh_hs has joined #openstack-cinder12:26
*** coolsvap is now known as coolsvap|away12:26
geguileoOk, but that was just another thing I was going to ask hemnafk if we could do12:27
geguileoSplit between fixing things that are broken and things that prevent us from changing Cinder state changes12:27
dulekGood point.12:28
dulekSo I think my work will enable atomic state changes.12:28
rakesh_hsHi Guys, could you please review
dulekBecause Nova won't blow up if attach returns some exception.12:29
dulekThis means you can return VolumeBusy12:29
dulekAnd Nova will be ready to accept new request to attach volume.12:29
geguileodulek: That's fantastic!12:29
*** akerr has joined #openstack-cinder12:29
dulekRight now it requires manual DB cleanup to recover from such issues.12:30
dulekBut definitely - I'm not sure. ;)12:30
dulekIt's probably best to ask hemnafk and others once I'll have the patch in Gerrit.12:30
geguileodulek: Maybe we should discuss with hemnafk what other changes need to be done just to allow us to keep working on Cinder side....12:30
dulekYou're right. :)12:31
geguileodulek: Ok, I'll leave you to finish you patch and then we'll open the discussion, ok? ;-)12:31
*** dims has quit IRC12:31
*** dims has joined #openstack-cinder12:31
dulekI have time till around 6 PM, lets see if we'll be able to catch anyone by then.12:32
*** bkopilov has quit IRC12:32
*** kjnelson has joined #openstack-cinder12:32
geguileodulek: It doesn't have to be today12:32
dulekOkay. Sounds like we have a plan. :)12:33
geguileodulek: We can start reviewing your patch and then open the discussion tomorrow or next week12:33
geguileodulek: Cool!12:33
*** coolsvap|away is now known as coolsvap12:35
*** david-ly_ has joined #openstack-cinder12:35
*** resker has joined #openstack-cinder12:36
*** coolsvap is now known as coolsvap|away12:36
*** esker has quit IRC12:38
*** geguileo has quit IRC12:38
*** david-lyle has quit IRC12:39
*** geguileo has joined #openstack-cinder12:43
*** jasondotstar has joined #openstack-cinder12:46
smcginnishaypo: You around?12:47
hayposmcginnis: hi12:47
smcginnishaypo: Hey! So I was watching the py34 tests yesterday.12:48
smcginnishaypo: Didn't see any issues.12:48
smcginnishaypo: So for me at least I am more comfortable moving that to voting now.12:48
smcginnishaypo: Just seemed a little too quick yesterday. :)12:48
*** lpetrut1 has joined #openstack-cinder12:49
hayposmcginnis: good :) for nova, i had to check the previous week. but failures were only real py3 bugs because a patch introduced py3 regressions12:49
haypo(a patch under review)12:49
hayposmcginnis: we can wait one week, there is no urgency12:49
smcginnishaypo: Yeah, it's a balance I think between risking something non-py3 getting in and having gate issues.12:50
*** lpetrut has quit IRC12:50
smcginnishaypo: But based on the ones I've looked at my concerns are met.12:50
*** husanu4 has joined #openstack-cinder12:50
smcginnisCan't speak for others of course.12:50
openstackgerritRakesh H S proposed openstack/cinder: check for encrypted volume type existance before deleting
*** husanu4 has quit IRC12:52
hayposmcginnis: i removed the WIP status of feel free to elaborate your point of view their with a vote ;)12:53
smcginnishaypo: Will do.12:53
*** tellesnobrega_ has joined #openstack-cinder12:55
*** resker has quit IRC12:55
*** esker has joined #openstack-cinder12:55
openstackgerritMichal Dulko proposed openstack/cinder: Check for encrypted volume type existence before deleting
hayposmcginnis: if the py34 starts failing, don't hesitate to harass me until it's fixed :)12:56
smcginnishaypo: Definitely will do! :)12:56
*** Ctina has joined #openstack-cinder12:56
*** husanu5 has joined #openstack-cinder12:56
openstackgerritMichal Dulko proposed openstack/cinder: Check for encrypted volume type existence before deleting
openstackgerritRakesh H S proposed openstack/cinder: check for existence of volume type encryption before deleting
*** bswartz has joined #openstack-cinder12:59
*** smoriya has quit IRC12:59
*** coolsvap|away is now known as coolsvap13:00
*** atuvenie_ has joined #openstack-cinder13:00
*** husanu5 has quit IRC13:00
*** julim has joined #openstack-cinder13:01
*** amoturi has joined #openstack-cinder13:03
*** atuvenie has quit IRC13:04
*** nlevinki has quit IRC13:04
*** e0ne is now known as e0ne_13:04
*** e0ne_ is now known as e0ne13:04
*** belmoreira has quit IRC13:05
*** afazekas has joined #openstack-cinder13:08
*** jecarey has joined #openstack-cinder13:10
*** nlevinki has joined #openstack-cinder13:12
openstackgerritForest Romain proposed openstack/cinder: Correct overquota error message
*** husanux2 has joined #openstack-cinder13:13
*** husanux2 has quit IRC13:13
*** Yogi1 has joined #openstack-cinder13:13
*** merooney has joined #openstack-cinder13:13
*** husanux3 has joined #openstack-cinder13:14
*** husanux3 has quit IRC13:18
*** husanux9 has joined #openstack-cinder13:20
*** e0ne is now known as e0ne_13:20
*** husanux9 has quit IRC13:21
*** coolsvap is now known as coolsvap|away13:21
*** e0ne_ is now known as e0ne13:22
*** jdurgin has joined #openstack-cinder13:24
*** rushil has joined #openstack-cinder13:25
*** husanux1 has joined #openstack-cinder13:25
*** salv-orl_ has joined #openstack-cinder13:27
openstackgerritMichal Dulko proposed openstack/python-cinderclient: Use shared shell arguments provided by Session
openstackgerritForest Romain proposed openstack/cinder: Correct overquota error message
*** husanux1 has quit IRC13:29
*** salv-orlando has quit IRC13:31
*** bluex-pl has joined #openstack-cinder13:31
*** mriedem has joined #openstack-cinder13:31
*** nkrinner has quit IRC13:32
*** rushil has quit IRC13:32
*** eharney has quit IRC13:35
*** marcusvrn1 has joined #openstack-cinder13:38
openstackgerritPranali Deore proposed openstack/cinder: Change generic NotFound to specific exception
*** rushil has joined #openstack-cinder13:38
*** gouthamr has joined #openstack-cinder13:38
*** marcusvrn has quit IRC13:39
*** dustins has joined #openstack-cinder13:40
*** jungleboyj has quit IRC13:40
*** changbl_ has joined #openstack-cinder13:40
*** julim has quit IRC13:44
*** thangp has joined #openstack-cinder13:48
*** rushil has quit IRC13:48
*** sjcherry has joined #openstack-cinder13:51
*** tellesnobrega_ has quit IRC13:51
*** ericksonsantos has quit IRC13:51
openstackgerritYuriy Nesenenko proposed openstack/cinder: Filter snapshots data on the DB side
*** merooney has quit IRC13:54
openstackgerritAnton Arefiev proposed openstack/cinder-specs: Service healthcheck
*** nllrte has joined #openstack-cinder13:56
nllrteAnyone know if Rackspace private cloud supports the EMC cinder driver?13:56
smcginnisnllrte: She's not online right now, but xyang might know that answer.13:57
*** earlephilhower has quit IRC13:57
*** xyang1 has joined #openstack-cinder13:58
nllrtethanks smcginnis13:58
*** rakesh_hs has quit IRC14:01
*** tellesnobrega has joined #openstack-cinder14:01
*** eharney has joined #openstack-cinder14:01
*** bkopilov has joined #openstack-cinder14:02
nllrtehi xyang1, are you aware whether or not Rackspace private cloud supports the EMC cinder driver?14:04
*** julim has joined #openstack-cinder14:04
*** rmesta has joined #openstack-cinder14:05
*** merooney has joined #openstack-cinder14:06
xyang1nllrte: There was certification on that long time ago, but I am not sure if it was done recently.  I can find out about it14:06
nllrteok thanks xyang114:07
*** r-daneel has joined #openstack-cinder14:09
openstackgerritEdwin Wang proposed openstack/cinder: Split FC driver code into Common and FC
*** rushil has joined #openstack-cinder14:11
*** tellesnobrega has quit IRC14:11
*** e0ne is now known as e0ne_14:13
*** tellesnobrega has joined #openstack-cinder14:14
*** rmesta has left #openstack-cinder14:15
*** ronis has joined #openstack-cinder14:16
*** akerr has quit IRC14:16
*** rushil has quit IRC14:21
*** jungleboyj has joined #openstack-cinder14:21
albertomhi aarefiev:can you review my patch again? I added the changes in your comment and test suite is passing with 0 failures14:23
openstackgerritSzymon Wróblewski proposed openstack/cinder: POC: Tooz locks demo
*** e0ne_ has quit IRC14:23
aarefievalbertom: sure, I'll look on it later14:23
*** tellesnobrega has quit IRC14:24
*** nlevinki has quit IRC14:24
smcginnisthingee, hemnafk, jungleboyj, jgriffith, xyang1, patrickeast: Any interest in doing another "Conversation" again at the next summit?14:25
xyang1smcginnis: Sounds good to me14:26
*** Zhongjun has joined #openstack-cinder14:26
smcginnisxyang1: Cool. I'll see who else is interested and submit a proposal.14:26
smcginnisDeadline is coming up kinda soon.14:26
jungleboyjsmcginnis: Works for me.14:26
jungleboyjsmcginnis: I have one other one that I need to get proposed.14:27
smcginnisjungleboyj: Awesome. At least three of us then. :)14:27
*** akerr has joined #openstack-cinder14:29
*** esker has quit IRC14:29
*** esker has joined #openstack-cinder14:30
*** marcusvrn has joined #openstack-cinder14:31
*** eharney has quit IRC14:33
xyang1nllrte: Hi, We didn't do any certification work with Rackspace recently, not sure about future plan yet14:33
*** vilobhmm has joined #openstack-cinder14:33
*** dtynan has quit IRC14:33
*** marcusvrn1 has quit IRC14:34
nllrteok, thanks xyang114:34
*** dtynan has joined #openstack-cinder14:34
*** esker has quit IRC14:35
*** e0ne has joined #openstack-cinder14:36
*** juzuluag has joined #openstack-cinder14:39
*** eharney has joined #openstack-cinder14:40
*** Tross has quit IRC14:40
*** rushil has joined #openstack-cinder14:47
*** MIDENN_ has quit IRC14:52
*** mdenny has joined #openstack-cinder14:53
*** kaisers_ has joined #openstack-cinder14:57
openstackgerritKazumasa Nomura proposed openstack/cinder: HP XP driver should call register_opts()
*** fifieldt has joined #openstack-cinder15:00
*** rakesh_mishra has joined #openstack-cinder15:00
dulekgeguileo: You may be interested - few minutes ago with bluex-pl we were able to cause problems when creating a lot of snapshots in parallel in GlusterFS driver15:00
*** david-ly_ is now known as david-lyle15:01
geguileodulek: Awesome!!15:01
geguileodulek: (in a way) ;-)15:01
dulekgeguileo: (when running two c-vol in A/A)15:01
*** fifieldt has quit IRC15:01
dulekgeguileo: We also confirmed that Tooz patch solves the problems. :)15:01
*** fifieldt has joined #openstack-cinder15:01
geguileodulek: Then you confirmed that it needed the lock15:01
geguileodulek: Thanks for the update15:02
dulekgeguileo: Yes, thing is - we have clear test that can confirm that we're solving the issue.15:02
geguileodulek: Is the test upstream so I can have a look?15:02
*** kaisers_0 has joined #openstack-cinder15:03
dulekgeguileo: Not yet, bluex-pl is working on some kind of "steps to reproduce" list, we'll post it in cinder-active-active-vol-service-issues once ready15:04
*** kaisers_ has quit IRC15:04
geguileodulek: Great15:04
*** mtanino has joined #openstack-cinder15:05
*** reed has quit IRC15:05
*** lpabon has joined #openstack-cinder15:05
openstackgerritYuriy Nesenenko proposed openstack/cinder: Filter snapshots data on the DB side
openstackgerritEdwin Wang proposed openstack/cinder: Separate FlashSystem FC and iSCSI common code
mtaninojungleboyj: hi,15:08
mtaninojungleboyj: My folks posted a patch for this bug.
openstackLaunchpad bug 1466198 in Cinder "tox -e genconfig fails" [Critical,In progress] - Assigned to Kazumasa Nomura (kazumasa-nomura-rx)15:09
*** med_` is now known as med_15:09
*** med_ is now known as med___15:09
mtaninojungleboyj: If you have a chance, please visit here.
dulekgeguileo: One more thing -
dulekgeguileo: Turns out change I was working on in Nova is already in review.15:10
*** reed has joined #openstack-cinder15:10
*** jasondotstar has quit IRC15:10
geguileodulek: I'll have a look at that one to understand what is blocking it15:11
aarefievtbarron: hi, I've mentioned about modified field in alternatives section, could you look?15:13
openstackgerritAnton Arefiev proposed openstack/cinder-specs: Service healthcheck
jungleboyjmtanino: Ok, will try to take a look shortly.15:14
*** rakesh_mishra has quit IRC15:14
tbarronaarefiev: looking, thanks for your patience15:15
*** mtanino_ has joined #openstack-cinder15:15
aarefievtbarron: np, thanks15:15
*** mtanino has quit IRC15:16
*** ronis has quit IRC15:16
*** Tross has joined #openstack-cinder15:16
*** edmondsw has quit IRC15:16
*** MentalRay has joined #openstack-cinder15:17
tbarronaarefiev: thank you.  My -1 is now gone.  Might end up doing +1 but I want to follow the discussion of the15:18
*** anshul has quit IRC15:18
*** dannywilson has joined #openstack-cinder15:18
tbarronoverloaded field and the design choice a bit.  If there is any discussion, maybe everyone else is15:18
tbarronfine with it.15:18
tbarronThe health check is a great idea and I very much appreciate your proposing it.15:19
*** merooney has quit IRC15:20
aarefievtbarron: thanks, I don't think using 'modified_at' will be a problem, it uses only on start/stop services, and I'm testing it now15:20
*** dannywil_ has joined #openstack-cinder15:20
*** dannywilson has quit IRC15:20
openstackgerritShay Halsband proposed openstack/cinder: XtremIO driver fix array snapshot problem
*** edmondsw has joined #openstack-cinder15:25
*** rakesh_mishra has joined #openstack-cinder15:26
*** rakesh_mishra has quit IRC15:26
*** rakesh_mishra has joined #openstack-cinder15:26
*** laughterwym_ has quit IRC15:27
*** DericHorn-HP has joined #openstack-cinder15:27
*** agarciam has joined #openstack-cinder15:27
*** mtanino has joined #openstack-cinder15:28
*** Apoorva has joined #openstack-cinder15:29
*** garthb_ has joined #openstack-cinder15:30
*** garthb has joined #openstack-cinder15:30
*** mtanino_ has quit IRC15:31
*** dtynan has quit IRC15:32
*** daneyon has joined #openstack-cinder15:33
*** dtynan has joined #openstack-cinder15:33
*** merooney has joined #openstack-cinder15:34
*** daneyon_ has joined #openstack-cinder15:34
*** e0ne is now known as e0ne_15:35
*** med___ is now known as med_15:36
*** med_ has quit IRC15:36
*** med_ has joined #openstack-cinder15:36
*** jdurgin has quit IRC15:37
*** e0ne_ is now known as e0ne15:37
*** daneyon has quit IRC15:37
*** kjnelson has quit IRC15:37
*** hemnafk is now known as hemna15:38
*** jungleboyj has quit IRC15:38
*** rakesh_mishra has quit IRC15:39
*** mtanino has quit IRC15:40
*** annegentle has joined #openstack-cinder15:40
*** merooney has quit IRC15:42
*** mkoderer_cloud has quit IRC15:44
*** jasondotstar has joined #openstack-cinder15:45
*** marcusvrn1 has joined #openstack-cinder15:47
*** marcusvrn has quit IRC15:47
*** rakesh_mishra has joined #openstack-cinder15:47
*** jungleboyj has joined #openstack-cinder15:47
*** vilobhmm has quit IRC15:51
*** timcl has quit IRC15:51
*** annegentle has quit IRC15:52
*** timcl has joined #openstack-cinder15:52
*** shyama has quit IRC15:52
*** annegentle has joined #openstack-cinder15:52
*** rakesh_mishra has quit IRC15:53
*** agarciam has quit IRC15:54
*** e0ne is now known as e0ne_15:56
*** kaisers_0 has quit IRC15:57
*** Longgeek has quit IRC15:59
*** atuvenie_ has quit IRC16:00
*** Longgeek has joined #openstack-cinder16:00
openstackgerritAmador Pahim proposed openstack/cinder: Do not accept qos name with only invisible characters
*** leeantho has joined #openstack-cinder16:01
*** lcurtis has joined #openstack-cinder16:05
*** rakesh_mishra has joined #openstack-cinder16:05
*** harlowja_still_a has quit IRC16:06
*** e0ne_ has quit IRC16:06
*** rushil has quit IRC16:11
*** e0ne has joined #openstack-cinder16:11
*** jordanP has quit IRC16:16
*** annegentle has quit IRC16:17
*** annashen has joined #openstack-cinder16:20
*** DericHorn-HP has quit IRC16:20
*** afazekas has quit IRC16:23
*** Tross has quit IRC16:23
openstackgerritAnton Arefiev proposed openstack/cinder: WIP: Implement service healthcheck tool
*** esker has joined #openstack-cinder16:25
*** esker has quit IRC16:25
*** esker has joined #openstack-cinder16:25
*** bluex-pl has quit IRC16:28
*** VManiac16 has joined #openstack-cinder16:29
*** bswartz has quit IRC16:29
*** erlon has joined #openstack-cinder16:29
*** markus_z has quit IRC16:29
*** atuvenie has joined #openstack-cinder16:30
*** Yogi1 has quit IRC16:34
*** shyama has joined #openstack-cinder16:39
*** ndipanov has quit IRC16:42
*** ronis has joined #openstack-cinder16:44
*** jwcroppe_ has quit IRC16:51
*** jwcroppe has joined #openstack-cinder16:51
*** rakesh_mishra has quit IRC16:52
*** nllrte has quit IRC16:52
*** rwsu has joined #openstack-cinder16:52
*** Yogi1 has joined #openstack-cinder16:53
*** jasondotstar has quit IRC16:55
*** jwcroppe has quit IRC16:56
*** _cjones_ has joined #openstack-cinder16:58
Swansonjungleboyj: Re 193334.  Thingee said on here somewhere that it didn't need a blueprint.17:01
jungleboyjSwanson: Thanks for confirming.  I trusted you guys.  :-)17:02
*** garthb_ has quit IRC17:03
*** garthb has quit IRC17:03
*** rushil has joined #openstack-cinder17:04
*** DericHorn-HP has joined #openstack-cinder17:05
*** esker has quit IRC17:05
*** hemna_ has joined #openstack-cinder17:06
*** garthb has joined #openstack-cinder17:07
*** garthb_ has joined #openstack-cinder17:07
*** jasondotstar has joined #openstack-cinder17:08
*** patrickeast has joined #openstack-cinder17:08
*** e0ne has quit IRC17:08
*** hemna_ has quit IRC17:09
*** _cjones_ has quit IRC17:17
*** _cjones_ has joined #openstack-cinder17:17
*** rushil_ has joined #openstack-cinder17:19
*** sseverson has quit IRC17:21
*** merooney has joined #openstack-cinder17:21
*** rushil has quit IRC17:22
*** shausy has quit IRC17:22
*** Ctina_ has joined #openstack-cinder17:31
*** jasondotstar has quit IRC17:34
*** marcusvrn1 has quit IRC17:34
*** Ctina has quit IRC17:35
smcginnisIs that the db exception one?17:36
*** Ctina_ has quit IRC17:36
smcginnisOops, wrong window. :)17:36
*** jasondotstar has joined #openstack-cinder17:39
*** jwcroppe has joined #openstack-cinder17:41
jungleboyjsmcginnis: Who on earth are you talking to?  ;-)17:44
eharneyi think he was asking "the cloud"17:48
*** Tross has joined #openstack-cinder17:48
*** nllrte has joined #openstack-cinder17:48
smcginniseharney: Follow up from yesterday. py34 tests were stable all day so I have my +1 to the patch to make it voting.17:49
smcginnisGot approved so it's going through gate right now.17:49
eharneyi wonder if we're all about to learn new things about py317:49
jungleboyjThat is good.17:49
smcginniseharney: No doubt.17:49
*** afazekas has joined #openstack-cinder17:49
openstackgerritAmador Pahim proposed openstack/cinder: Do not accept qos name with only invisible characters
patrickeastasselin: hey, have you run into any issues with building dib images the last day or so?17:53
*** marcusvrn has joined #openstack-cinder17:53
*** boris-42 has joined #openstack-cinder17:54
*** breitz has quit IRC17:54
*** afazekas has quit IRC17:55
*** markstur has quit IRC17:55
*** tellesnobrega has joined #openstack-cinder17:56
asselinpatrickeast, will look17:57
*** breitz has joined #openstack-cinder17:57
patrickeastasselin: i’m getting this error
patrickeastasselin: from…
patrickeastunsure if its just my system or if something is actually borked17:58
asselinpatrickeast, my last one was 4 days ago18:00
*** husanu0 has joined #openstack-cinder18:02
*** dontalton has joined #openstack-cinder18:02
*** bswartz has joined #openstack-cinder18:02
*** markstur has joined #openstack-cinder18:05
*** jasondotstar has quit IRC18:05
*** jwcroppe_ has joined #openstack-cinder18:08
*** gouthamr has quit IRC18:08
*** esker has joined #openstack-cinder18:08
*** gouthamr has joined #openstack-cinder18:08
asselinpatrickeast, building one now....18:08
*** jwcroppe_ has quit IRC18:09
*** jwcroppe_ has joined #openstack-cinder18:09
*** jwcroppe has quit IRC18:11
*** MentalRay has quit IRC18:11
openstackgerritEric Harney proposed openstack/cinder: Don't allow >4-byte Unicode characters in volume names
*** husanu0 has quit IRC18:14
asselinpatrickeast, but that file I override it to be empty18:16
*** merooney has quit IRC18:16
patrickeastasselin: oh? why do you do that?18:17
asselinpatrickeast, I must've add problems with it way back and didn't really need it18:17
asselinpatrickeast, do you know what it does?18:18
patrickeastasselin: not really :D18:19
patrickeastasselin: prepares the tempest testrepository cache thing?18:19
*** jasondotstar has joined #openstack-cinder18:19
openstackgerritAmador Pahim proposed openstack/cinder: Do not accept qos name with only invisible characters
patrickeast looks like it is creating subunit results from the logstash db?18:21
patrickeastnot sure why we would need that18:21
asselinpatrickeast, this is my error:
patrickeastasselin: hmm18:23
*** kjnelson has joined #openstack-cinder18:24
*** alyfon has joined #openstack-cinder18:26
*** alyfon has quit IRC18:26
*** alyfon has joined #openstack-cinder18:26
asselinpatrickeast, I seem to have a old build still running from a few days ago
alyfonhey there I was looking at the reference LVM driver and wondering why they didnt explicitly write an attach/detach volume method even though it18:27
alyfonit's part of the min requirements18:27
alyfonis it because the base volumedriver class has it implemented18:28
*** julim has quit IRC18:28
*** annashen has quit IRC18:28
*** shyama has quit IRC18:28
*** annegentle has joined #openstack-cinder18:28
*** julim has joined #openstack-cinder18:28
asselinpatrickeast, killed and restarted18:29
dannywil_alyfon: yes18:30
*** DericHorn-HP has quit IRC18:31
alyfonok and i know this may sound really stupid but I was wondering what the difference between the call back attach/detach volume methods and the _attach/_detach functions are18:31
*** krtaylor has quit IRC18:31
alyfonb/c the callback functions are empty with pass18:31
alyfonin driver.py18:31
*** timcl has quit IRC18:32
*** lpetrut1 has quit IRC18:36
openstackgerritKazumasa Nomura proposed openstack/cinder: HP XP driver should call register_opts()
dannywil_alyfon: are you talking about these functions?
dannywil_alyfon: I am not too familiar with what those are for18:37
alyfonyeah exactly18:38
*** ericksonsantos has joined #openstack-cinder18:38
dannywil_alyfon: have to step out for a bit, sorry I couldn't be more help18:38
alyfonthanks for the help so far18:38
*** Zhongjun has quit IRC18:38
*** Yogi1 has quit IRC18:39
*** Yogi1 has joined #openstack-cinder18:40
*** merooney has joined #openstack-cinder18:43
*** ericksonsantos has quit IRC18:43
*** dontalton has quit IRC18:44
*** annegentle has quit IRC18:45
*** jasondotstar has quit IRC18:47
*** ociuhandu has quit IRC18:48
*** timcl has joined #openstack-cinder18:48
*** julim has quit IRC18:52
*** r-daneel has quit IRC18:54
*** r-daneel has joined #openstack-cinder18:55
*** jwcroppe_ has quit IRC18:55
*** jwcroppe has joined #openstack-cinder18:56
*** ericksonsantos has joined #openstack-cinder18:56
*** jwcroppe has quit IRC18:57
*** jwcroppe has joined #openstack-cinder18:57
*** julim has joined #openstack-cinder18:59
*** MentalRay has joined #openstack-cinder19:02
*** julim has quit IRC19:02
*** julim has joined #openstack-cinder19:03
*** ericksonsantos has quit IRC19:03
*** jasondotstar has joined #openstack-cinder19:07
*** Svedrin has quit IRC19:09
*** tellesnobrega has quit IRC19:10
*** kjnelson has quit IRC19:10
*** Svedrin has joined #openstack-cinder19:11
*** merooney has quit IRC19:12
*** merooney has joined #openstack-cinder19:12
openstackgerritAndrey Pavlov proposed openstack/cinder: Avoid race condition at snapshot deletion stage
*** merooney has quit IRC19:13
*** Tross has quit IRC19:15
*** garthb_ has quit IRC19:15
*** garthb has quit IRC19:15
openstackgerritThang Pham proposed openstack/cinder: Conversion to volume object
openstackgerritThang Pham proposed openstack/cinder: WIP: RPC and object versioning
openstackgerritAndrey Pavlov proposed openstack/cinder: Small fix in doc string definition
*** jdurgin has joined #openstack-cinder19:24
*** merooney has joined #openstack-cinder19:27
*** annashen has joined #openstack-cinder19:29
*** tellesnobrega has joined #openstack-cinder19:29
*** merooney has quit IRC19:32
*** rushiagr_away is now known as rushiagr19:32
openstackgerritJay Bryant proposed openstack/cinder: Fix cinder.conf.sample generation
*** annashen has quit IRC19:34
openstackgerritJay Bryant proposed openstack/cinder: Fix cinder.conf.sample generation
*** tellesnobrega has quit IRC19:35
*** annashen has joined #openstack-cinder19:35
*** merooney has joined #openstack-cinder19:36
*** jdurgin has quit IRC19:37
*** merooney has quit IRC19:39
*** kjnelson has joined #openstack-cinder19:40
*** daneyon_ has quit IRC19:42
*** julim has quit IRC19:44
*** merooney has joined #openstack-cinder19:48
*** DericHorn-HP has joined #openstack-cinder19:50
*** lpetrut has joined #openstack-cinder19:52
openstackgerritAmador Pahim proposed openstack/cinder: Do not accept qos name with only invisible characters
*** esker has quit IRC19:54
eharneyjungleboyj: it seems like we should have a job/test somewhere that ensures that the sample conf actually can be generated19:54
eharney(though not requiring updates like we used to)19:54
*** kjnelson has quit IRC19:55
*** nllrte has quit IRC19:58
*** timcl has quit IRC19:58
*** juzuluag has quit IRC20:01
*** merooney has quit IRC20:01
*** mtanino has joined #openstack-cinder20:01
mtaninojungleboyj: ping20:03
jungleboyjmtanino: Was about to step out for a bit.  What's up?20:05
*** atuvenie has quit IRC20:05
mtaninojungleboyj: about the bug
openstackLaunchpad bug 1466198 in Cinder "tox -e genconfig fails" [Critical,In progress] - Assigned to Jay Bryant (jsbryant)20:06
jungleboyjmtanino: Yes.20:06
mtaninojungleboyj: my folks posted a patch
mtaninodo you plan to post someting to fix it?20:06
mtaninoI saw you took the bug again. so just asking :)20:07
jungleboyj*sigh*  I took the bug as I was working on it.20:08
jungleboyjI had talked to the driver owner and to Walt about this about the solution.20:08
openstackgerritEric Harney proposed openstack/cinder: Don't allow >4-byte Unicode chars in volume names
*** jecarey has quit IRC20:09
*** krtaylor has joined #openstack-cinder20:09
mtaninohmm, the patch was posted by driver owner.20:10
jungleboyjYeah, and this is changing more than it needs to.20:10
jungleboyjOh, that is the driver owner?  nestor ?20:10
mtaninoyes Kazumasa Nomura is driver owner20:11
mtaninoHitachi guy.20:11
jungleboyjOk.  Really wish he had talked to me since I was the bug owner.20:12
mtaninoAh, because he locates in Japan, he is not here now :(20:12
jungleboyjWasted a notable amount of time.20:12
*** Longgeek has quit IRC20:13
*** amoturi has quit IRC20:14
mtaninoSorry for about that :(20:14
jungleboyjI need to drop off now.20:15
mtaninojungleboyj: Thank you for taking time.20:15
jungleboyjOh, that was what you asked me to review earlier today mtanino20:16
jungleboyjWhich Ironically, I didn't look at because I was busy working on that and other fixes.20:16
jungleboyjOh well.20:16
mtaninojungleboyj: Yes, yes.20:17
*** gfarmerfr has joined #openstack-cinder20:17
mtaninojungleboyj: I'm joining Red Hat summit and sometimes lost wifi connection... sorry.20:17
*** thangp has quit IRC20:24
*** jungleboyj has quit IRC20:25
gfarmerfrHi everyone, has anyone already tested the Storwise SVC driver in conjuction with zonemanager for FC autozoning ?20:30
*** ronis has quit IRC20:31
*** marcusvrn has quit IRC20:34
*** rushil_ has quit IRC20:35
*** MentalRay has quit IRC20:36
*** jasondotstar has quit IRC20:39
*** jaypipes has quit IRC20:39
*** MentalRay has joined #openstack-cinder20:40
*** rushil has joined #openstack-cinder20:40
*** afazekas has joined #openstack-cinder20:41
openstackgerritAlex Meade proposed openstack/cinder: Add tracing facility for drivers
*** julim has joined #openstack-cinder20:47
patrickeastameade: oo metaclass magic20:50
patrickeastameade: thats pretty cool, only need the one annotation on the class now20:51
ameadeand you can still do the other way if you dont want every method wrapped20:52
patrickeastameade: does it end up being the last wrapper around methods? or the first one?20:54
ameadethe outer20:54
alyfonhey i asked this before but im having trouble fully understanding why most of the driver files im looking at (pure, emc_vmax, and even in the lvm reference driver) they don't explicily call an attach/detach volume method20:55
ameadepatrickeast: yessir, almost too tricky though20:55
alyfonand for example emc_vmax_fc doesnt derive its class from the volumedriver class that includes _attach_volume/_detach_volume only from the fibrechannel driver class20:56
patrickeastalyfon: speaking for the pure driver, we haven’t needed to, all we need is the stuff done in initialize_connection/terminate_connection20:56
alyfonohh ok that makes sense20:57
alyfonim guessing the emc driver also works in a similar way20:57
patrickeastyea probably20:57
alyfonthanks a lot20:58
*** garthb has joined #openstack-cinder20:59
*** garthb_ has joined #openstack-cinder20:59
*** garthb_ has quit IRC21:00
*** garthb_ has joined #openstack-cinder21:01
openstackgerritxing-yang proposed openstack/os-brick: Add connector driver for the ScaleIO cinder driver
*** garthb_ has quit IRC21:01
*** garthb has quit IRC21:01
openstackgerritxing-yang proposed openstack/cinder: EMC VMAX Create CG from CG Snapshot
*** garthb has joined #openstack-cinder21:01
*** garthb_ has joined #openstack-cinder21:02
*** dannywilson has joined #openstack-cinder21:02
hemnafinding more failures in nova live migration process.  bleh.21:02
*** patrickeast_ has joined #openstack-cinder21:03
*** annegentle has joined #openstack-cinder21:03
openstackgerritAlex Meade proposed openstack/cinder: Add tracing facility for drivers
openstackgerritAlex Meade proposed openstack/cinder: Add tracing facility for drivers
*** patrickeast has quit IRC21:05
*** patrickeast_ is now known as patrickeast21:05
*** dannywil_ has quit IRC21:05
*** e0ne has joined #openstack-cinder21:05
*** akerr has quit IRC21:07
*** DericHorn-HP has quit IRC21:08
*** DericHorn-HP has joined #openstack-cinder21:11
*** chlong has quit IRC21:11
openstackgerritEric Harney proposed openstack/cinder: Don't allow >3-byte Unicode chars in volume names
hemnaxyang1, so I'm not sure how your drivers react to having initialize_connection called multiple times for the same volume.21:13
hemnabut our drivers react differently.  Lefthand turns into a no op, but 3PAR creates an export for even initialize_connection call.21:13
*** KurtMartin has joined #openstack-cinder21:13
*** amoturi has joined #openstack-cinder21:13
xyang1hemna: I think it will just return success if it is already attached21:13
hemnaxyang1, not all drivers do that though.21:14
*** IanGovett has quit IRC21:14
hemnaxyang1, and I'm curious how multi-attach works then in that case for you guys :)21:14
hemnabut nova's assumptions of calling initialize_connection multiple times is bad21:14
hemnaxyang1, here
xyang1hemna: We will nave to check host and volume21:15
hemnabad mmmkay21:15
xyang1hemna: We have not tested multi attach yet21:15
hemnaxyang1, that refresh_connection_info was also the source of the lost devices21:15
hemnafor the bdm21:15
*** rushil has quit IRC21:15
hemnabecause it pounds over whatever the libvirt volume connect_volume driver returned for the connection_info, with what cinder's initialize_connection returns.21:16
hemnaalso bad21:16
xyang1hemna: :(21:16
hemnait's so badly broken, I want to puke.21:16
*** Yogi1 has quit IRC21:17
xyang1hemna: How is switching over to os-brick going?21:17
hemnaso, it's going well actually21:17
hemnawe are testing live migration with the nova os-brick patch21:17
hemnajust to make sure that it doesn't cause anything that doesn't already exist.21:18
hemnainternally I have a CI setup against the nova os-brick patch that tests against our hardware21:18
hemna3par iscsi, FC and Lefthand iSCSI21:18
xyang1hemna: Oh, I don't see instruction on os brick CI on the wiki21:18
hemnaasselin, can help you with that21:19
hemnaxyang1,  I had been rebasing your nova patch when I pushed a new os-brick patch up.21:19
xyang1hemna: We are setting it up too.  We had to shut down the normal CI to test os brick21:19
xyang1hemna: Thanks for the help21:19
*** IanGovett has joined #openstack-cinder21:20
*** rushil has joined #openstack-cinder21:20
*** Longgeek has joined #openstack-cinder21:21
*** chlong has joined #openstack-cinder21:23
*** sjcherry has quit IRC21:28
*** eharney has quit IRC21:28
hemnaso it looks like rescan_iscsi was added there21:28
hemnathat causes dangling iSCSI volumes during live migration21:29
xyang1hemna: Line 270?21:30
hemnabecause nova's compute manager's post_live_migration()21:30
xyang1hemna: There is a reason that was added though21:30
xyang1hemna: There is comment there21:30
xyang1hemna: I think it is moved, the rescan was at different location before21:31
hemnaso if there are more than 1 volumes on a vm during live migration, for iSCSI, you will most likely end up with dangling volumes on the source host when it's done21:31
hemnawhen iscsiadm --rescan is called, the volumes that were previously removed show back up.21:31
hemnabecause the compute manager doesn't call cinder's terminate_connection with every libvirt volume disconnect_volume call.21:32
hemnaso if there is 1 volume, it's ok.21:32
hemnabut if there are 2 or more21:32
xyang1hemna: Did you see the comments, the call was there at a different location21:32
hemnayou will end up with volumes left behind on the source.21:32
hemnayah, I see the comment21:33
hemnanot sure I understand the reason for the rescan though21:33
hemnaI think this is the same issue basically.21:33
hemnaso for iSCSI21:34
hemnaif we issue a delete volume from the host21:34
hemnaw/o an immediate call to cinder to terminate_connection21:34
hemnathat volume can show back up21:34
hemnaanyone that issues an iscsiadm --rescan21:34
hemnawhich can happen in parallel for volume attaches21:34
xyang1This is after terminate connection, no?21:35
hemnanot for live migration21:35
*** sseverson has joined #openstack-cinder21:36
xyang1I see21:36
*** rushil has quit IRC21:36
hemnalook at
*** mtanino has quit IRC21:38
hemnalibvirt/'s post_live_migration walks every volume in the BDM and calls disconnect_volume21:38
hemnaeach disconnect_volume call issues an iscsiadm --rescan21:38
hemnaterminate_connection needs to be called for every disconnect_volume call21:39
hemnaor iSCSI volumes will be orphaned on the source21:39
*** e0ne is now known as e0ne_21:39
*** Longgeek has quit IRC21:40
*** e0ne_ is now known as e0ne21:41
*** ian_brown has joined #openstack-cinder21:41
xyang1If you call terminate connection so early in live migration, the volume will be attached to no one in the middle of migration21:42
*** mriedem has quit IRC21:44
hemnathis is post_live_migration21:45
hemnaso this is after the destination has been stood up and volumes attached to it.21:45
hemnathis is all about tearing down the source side and cleaning up21:45
*** krtaylor has quit IRC21:47
*** ian_brown has quit IRC21:48
*** lpetrut has quit IRC21:50
*** julim has quit IRC21:51
hemnaxyang1, the driver.post_live_migration() walks through the list of volumes and calls disconnect_volume(), that returns and then immediately the compute manager calls terminate_connection on each volume.21:55
xyang1Are you suggesting to move rescan after terminate connection?21:56
hemnaI'm questioning why we need rescan at all21:57
*** annegentle has quit IRC21:58
xyang1I think if you don't rescan, the detached device won't go away21:59
hemnaI'm not sure about that21:59
jgriffithxyang1: +121:59
jgriffithIt's better to be explicit than implicit21:59
hemnathe device goes away in brick with the call to remove it from the kernel22:00
jgriffithbecause implicit isn't guaranteed to work22:00
jgriffithhemna: "your" devices, or everybody elses?22:00
*** chlong has quit IRC22:00
jgriffithhemna: and the bigger problem is it doesn't "always" work22:00
hemnathat removes the device entirely from the host.22:01
hemnaa rescan makes it come back22:01
*** dannywilson has quit IRC22:01
hemnawhich is what is happening in live migration22:01
jgriffithI guess I'm confused, what you're talking about22:01
jgriffithnova, brick, both, neither, something else22:02
hemnawell yah it's confusing22:02
hemnabut it's both really.22:02
hemnanova's compute manager post_live_migration gets called22:02
*** dannywilson has joined #openstack-cinder22:02
*** dannywilson has quit IRC22:02
jgriffithYeah, I know you have a LOT of issues with how the nova side deals with volumes, some of it I get, some of it I don't understand the problem at all22:02
*** dannywilson has joined #openstack-cinder22:03
hemnait calls the libvirt volume driver's post_live_migration(), which loops through each volume and calls the libvirt volume driver's disconnect_volume()22:03
hemnathen the compute manager calls terminate_connection for each volume.22:03
jgriffithlive-migration inparticular, I just tested again last week, seems to work fine22:03
*** rushiagr is now known as rushiagr_away22:03
hemnafor VMs that have more than 1 volume attached22:03
hemnathat leaves iSCSI volumes behind on the source.22:03
SwansonI can't help but think having 4 returns in a 16 statement function (4 of which are logs) will get me a sea of -1s when I push this.22:04
xyang1jgriffith: Are you testing single path or multipath?22:04
hemnabecause terminate_connection isn't called until after all libvirt volume disconnect_volume calls are done.22:04
*** Rockyg has joined #openstack-cinder22:04
jgriffithxyang1: oh... just multipath22:04
hemnaeach libvirt volume disconnect_volume issues an iscsi rescan at the very start22:04
xyang1jgriffith: I think that explained it22:04
jgriffithhemna: have you logged a bug, submitted a patch?22:04
xyang1jgriffith: We tested multipath as well and it worked22:04
jgriffithhemna: I'd be happy to review either or both22:04
hemnaso the 2nd disconnect_volume is called, it issues a rescan, and the volume it just removed shows back up.22:04
jgriffithxyang1: ooops... typo22:05
hemnaI haven't filed a bug yet22:05
hemnaI'm trying to figure out how to fix it :P22:05
jgriffithxyang1: I have NOT done anything with multi-path as of yet22:05
*** jwcroppe_ has joined #openstack-cinder22:05
xyang1jgriffith: oh22:05
jgriffithhemna: well, a bug might be really useful to write up the problem and help others understand what you're talking about22:05
xyang1jgriffith: hemna runs into problem with single path22:06
hemnawe are just spending cycles looking heavily at live migration and it's issues, and doing lots of testing.  this is the latest issue we found.22:06
hemnajgriffith, yah ok.22:06
jgriffithhemna: you need to log bugs... becuase you say "it's issues" but meanwhile there are a number of folks using it without issue22:07
jgriffithhemna: which concerns me22:07
*** jwcroppe has quit IRC22:07
hemnajgriffith, well, that's part of the problem I've been discovering actually22:07
hemnapart of the problem is the interaction between nova and cinder22:07
hemnaand also how some cinder driver's have implemented initialize_conneciton.22:07
jgriffithhemna: I'm aware, you keep saying that :)22:07
jgriffithhemna: I'm just pointing out that it would be good to get things written down so others can follow along and help out22:08
hemnaI documented some of it at the bottom of that22:08
jgriffithparticularly since there's a number of devices that seem to work fine and it would suck if you broke them :)22:08
hemnawell I didn't break them22:08
hemnait's a combination of things22:08
jgriffithhemna: slow down22:08
jgriffithhemna: read what I said please :)22:08
hemnaheh I did, I gave you a url :)22:09
jgriffithI'll repeat:  There are a number of drivers that WORK properly today22:09
hemnaanyway, I documented some of it in that etherpad of what we've been seeing.22:09
hemnayes I know22:09
jgriffithI'm concerned that changes made for 3Par could "break" those drivers22:09
hemnaand I understand why that is22:09
hemnaI understand why some backends work and others don't.22:09
hemnaI documented it in that etherpad22:10
hemna"Live migration broken for some cinder backends, and works for others:" section22:10
*** gouthamr has quit IRC22:10
hemnathat's issue 1) there22:10
jgriffithhemna: I know... but I'm pointing out that be sure "you're" not the one making incorrect assumptions22:11
hemnaI believe I have fixed issue 2) for FC based drivers already in my patch here:
hemnaI'm just documenting what I'm seeing, and looking at a few cinder drivers, it's broken for more than just the 3PAR.22:12
hemnalefthand 'works'22:13
*** KurtMartin has quit IRC22:14
*** KurtMartin has joined #openstack-cinder22:14
*** merooney has joined #openstack-cinder22:19
openstackgerritxing-yang proposed openstack/cinder: EMC VMAX Create CG from CG Snapshot
angela-sgood news for those interested, Brocade CI is finally posting to the VF and Friendly Zone Names code reviews22:24
xyang1angela-s: Nice!22:25
angela-sbad news is there are still about 16 failed tests22:25
*** e0ne has quit IRC22:26
openstackgerritxing-yang proposed openstack/cinder: EMC VMAX Create CG from CG Snapshot
angela-salso, PG&E is shutting power off to our campus for the next week for upgrades, so Brocade CI will be offline starting tomorrow until July 8.22:26
angela-sgot it working just in time to shut if off. :)22:27
*** annashen has quit IRC22:27
*** edmondsw has quit IRC22:27
angela-sthanks xyang122:28
xyang1angela-s: Np, glad it started reporting22:28
asselinangela-s, cool!22:29
*** KurtMartin has quit IRC22:30
*** krtaylor has joined #openstack-cinder22:30
*** dims has quit IRC22:30
*** garthb has quit IRC22:31
*** garthb_ has quit IRC22:31
*** dontalton has joined #openstack-cinder22:32
*** garthb has joined #openstack-cinder22:33
*** garthb_ has joined #openstack-cinder22:33
*** IlyaG has joined #openstack-cinder22:35
*** Longgeek has joined #openstack-cinder22:37
*** dims_ has joined #openstack-cinder22:38
*** gfarmerfr has quit IRC22:40
*** lcurtis has quit IRC22:41
*** Longgeek has quit IRC22:42
openstackgerritRyan Lucio proposed openstack/cinder: VMEM v7000: external-head storage pool and volume type selection
*** dustins_ has joined #openstack-cinder22:45
*** DericHorn-HP has quit IRC22:45
*** rwall has quit IRC22:45
*** dontalton has quit IRC22:45
*** dhellmann has quit IRC22:46
*** dhellmann has joined #openstack-cinder22:46
*** dustins has quit IRC22:47
*** rwall has joined #openstack-cinder22:48
*** merooney has quit IRC22:55
*** annegentle has joined #openstack-cinder22:59
*** mriedem has joined #openstack-cinder23:02
patrickeastjgriffith: hey, if you are around/have a few min, i have a couple questions for stuff i’m running into with the generic image cache that you probably know23:03
*** annegentle has quit IRC23:05
*** mriedem has quit IRC23:06
*** mriedem has joined #openstack-cinder23:06
jgriffithpatrickeast: sure, what's up23:13
*** dustins_ has quit IRC23:14
*** zhenguo has joined #openstack-cinder23:18
*** markvoelker has quit IRC23:24
*** garthb has quit IRC23:24
*** garthb_ has quit IRC23:24
*** garthb has joined #openstack-cinder23:25
*** garthb_ has joined #openstack-cinder23:25
*** aorourke has quit IRC23:27
*** aorourke has joined #openstack-cinder23:27
*** MentalRay has quit IRC23:35
patrickeastjgriffith: sry got side tracked, still around?23:36
*** garthb has quit IRC23:37
*** garthb_ has quit IRC23:37
jgriffithpatrickeast: sure23:38
patrickeastjgriffith: cool, so first thing is for the size of the cached volumes, in your driver you are using the virtual_size, and maybe i’m missing something, but wouldn’t the min_size property be better?23:39
patrickeastjgriffith: or is the virtual_size actually the smaller one?23:39
patrickeastjgriffith: s/min_size/min_disk/23:39
jgriffithpatrickeast: min_size doesn't work :(23:40
patrickeastjgriffith: oh23:40
jgriffithpatrickeast: trying to remember why LOL23:40
patrickeastjgriffith: haha, well i see that sometimes it may not be set23:40
patrickeastlike uhh23:40
* patrickeast pulls up github links23:40
jgriffithpatrickeast: so IIRC min_disk has a different meaning though23:41
patrickeastjgriffith: so maybe better question would be, whats virtual_size then?23:41
patrickeastjgriffith: thats like the actual image size uncompressed?23:42
jgriffithpatrickeast: so the main reason I went with that is I didn't want to mess around with anything Glance had stored in the image23:42
jgriffithpatrickeast: in our case things like "0" don't do us any good23:42
jgriffithpatrickeast: but if you change that it screws up the ephemeral case23:42
jgriffithpatrickeast: virtual size is safe because it's metadata and for the most part not used23:42
patrickeastjgriffith: gotcha23:42
jgriffithpatrickeast: also, virtual_size is the expanded/virtual size reported back from the Instance23:42
patrickeastjgriffith: so then my next question, does that mean a user would have to go set it manually on each image?23:43
jgriffithpatrickeast: so in our case, we're kinda making it "non" virtual23:43
patrickeastor is that automatic23:43
patrickeastahh ok, thats where the virtual size comes from23:43
jgriffithYes, and Admin has to do it23:43
patrickeastok cool23:43
jgriffithI have/had a comment in the code about that23:43
jgriffithpatrickeast: It would be cool if we automated that at some point23:43
jgriffithpatrickeast: but for now, I have it listed as a requirement23:43
patrickeastjgriffith: seems pretty reasonable to me23:44
jgriffithpatrickeast: if it's not set I just skip it and move on23:44
patrickeastso i’ll stick with virtual_size then23:44
jgriffithpatrickeast: yeah, there's probably a better way to do all of that part, but that was a quick and dirty method that didn't introduce problems and was explicit23:44
patrickeastjgriffith: ok, so last question, maybe more of a check to see if i’m going way off in the woods23:45
jgriffithpatrickeast: It was kinda surprising to me that by default we didn't seem to store any "real" size info in the image23:45
patrickeasti currently am calling back into the volume api to create the volume23:45
patrickeastbut i am going to need to skip the scheduler23:45
patrickeastand go straight to the host23:45
patrickeasti dunno how much cinder devs adhere to layering violations and what not23:45
patrickeastbut i’m going through like 8 million of them23:46
jgriffithpatrickeast: so personally I am not a big fan of that... but it might be necessary/ok23:46
*** esker has joined #openstack-cinder23:46
jgriffithpatrickeast: in my case I prefer calling the driver itself in the manager23:46
jgriffithpatrickeast: given you are already there and have the handle it seemed like an "ok" way to do it23:46
patrickeasti need to pass some kind of parameter that is like a don’t schedule and don’t try to create this volume from the cache, it really needs to download option23:46
*** esker has quit IRC23:46
patrickeastone sec i’ll put up a WIP review, maybe this will make more sense23:46
jgriffithpatrickeast: so you can do that.. but keeping it in the manager might be better IMHO23:47
jgriffithpatrickeast: ok23:47
patrickeastyea thats what im asking, it feels kind of wrong23:47
*** esker has joined #openstack-cinder23:47
jgriffithpatrickeast: so we have the mechanism to do what you want on the rpc side (ie snapshot, delete etc)23:47
jgriffithpatrickeast: cramming that into tflow might be interesting23:47
jgriffithpatrickeast: might be a piece of cake, I'm not sure TBH23:47
patrickeastjgriffith: yea the problem is that the quota and db stuff is on the other side of the rpc23:47
openstackgerritPatrick East proposed openstack/cinder: WIP generic image cache
jgriffithpatrickeast: ahh...hmm, interesting point23:48
patrickeastignore some of the ghetto pieces23:48
patrickeastits mostly a POC23:48
*** chlong has joined #openstack-cinder23:48
patrickeastit works right now23:48
patrickeastbut it doesn’t have a way to say ‘don’t use the cache’23:48
patrickeastso it will recursively fill up the cache with the same image until it errors, then continue on : /23:49
jgriffithpatrickeast: LOL23:51
jgriffithpatrickeast: why can't you just add a call to DB like I have in the driver?23:51
jgriffithpatrickeast: or do you mean you just haven't fixed that up yet23:51
patrickeastjgriffith: just haven’t fixed it yet23:51
*** erlon has quit IRC23:51
jgriffithpatrickeast: ahh... got ya23:51
patrickeasti also need to figure out the keystone internal tenant stuff.. but one problem at a time23:51
jgriffithpatrickeast: :)23:51
jgriffithpatrickeast: so I'll have to load it up to really understand it all, but looking through it, sure seems reasonable to me23:53
*** alyfon has quit IRC23:53
jgriffithpatrickeast: I'm unsure of putting the calls in tflow vs manager still23:53
*** laughterwym has joined #openstack-cinder23:53
jgriffithpatrickeast: but that's just kind of an impl detail more than anything else probably23:54
jgriffithpatrickeast: can't think of a real reason why it matters right now23:54
patrickeastjgriffith: yea i’m going to keep thinking on that, i don’t like that it needs to make so many extra rpc and glance calls now23:54
patrickeastdrop in the bucket compared with downloading an image from a cache hit23:54
patrickeastbut still doesn’t sit well23:54
jgriffithpatrickeast: yeah, that's the only ugly thing23:54
jgriffithpatrickeast: so one advantage of the manager is you can just do driver direct calls and such.  But then you're on that other side of the boundary like you said, which also causes problems with things like context in some cases23:55
patrickeastjgriffith: yea exactly, i was trying to find a nice balance of re-using code vs having to call in so high up in the stack23:56
*** smoriya has joined #openstack-cinder23:56
jgriffithpatrickeast: I'd love to see if we could have more reuse23:56
patrickeastjgriffith: the taskflow stuff makes it hard, i almost want to just do a few of the taska and sort of cheat to put a new volume entry in w/ quota and stuff23:56
patrickeastits silly to do all of the checks over again23:57
patrickeastbut they are built in to the flows23:57
jgriffithpatrickeast: yeah, that's one of the things I've never been a huge fan of23:57
*** esker has quit IRC23:57
jgriffithpatrickeast: it's a bit ironic, it's so dynamic about some things, yet completely monolithic about others23:57
*** Rockyg has quit IRC23:57
jgriffithpatrickeast: split personality :)23:58
patrickeastjgriffith: haha yea, i suspect thats a symptom of us being kind of half-in with it23:58
patrickeastanyway, i guess i’ll keep going with this direction and get it a bit more polished and tested23:59
patrickeastjgriffith: let me know if you have any more ideas on how to stick with just the manager, or if you have time to look through the code a bit more and see anything i need to change23:59

Generated by 2.14.0 by Marius Gedminas - find it at!