Friday, 2015-10-02

*** edmondsw has quit IRC00:01
*** garthb has quit IRC00:02
*** diogogmt has quit IRC00:03
*** tsekiyam_ has joined #openstack-cinder00:09
*** mtanino has quit IRC00:12
*** tsekiyama has quit IRC00:13
*** tsekiyam_ has quit IRC00:14
*** dsariel_ has quit IRC00:21
*** chlong has joined #openstack-cinder00:22
*** _cjones_ has quit IRC00:23
*** bill_az has quit IRC00:25
*** markvoelker has joined #openstack-cinder00:30
*** jwcroppe has joined #openstack-cinder00:32
*** markvoelker has quit IRC00:35
*** daneyon has joined #openstack-cinder00:40
*** daneyon_ has joined #openstack-cinder00:42
*** daneyon has quit IRC00:46
*** IlyaG has quit IRC00:59
*** daneyon has joined #openstack-cinder01:02
*** EinstCrazy has joined #openstack-cinder01:02
*** daneyon__ has joined #openstack-cinder01:03
*** daneyon_ has quit IRC01:04
*** daneyon has quit IRC01:07
*** leeantho has quit IRC01:08
*** EinstCrazy has quit IRC01:12
*** amit213 has quit IRC01:24
*** apoorvad has quit IRC01:24
*** amit213 has joined #openstack-cinder01:24
*** Lee1092 has joined #openstack-cinder01:31
*** apoorvad has joined #openstack-cinder01:31
*** dims has quit IRC01:34
*** mudassirlatif has quit IRC01:34
*** daneyon__ has quit IRC01:38
*** gouthamr has joined #openstack-cinder01:46
*** EinstCrazy has joined #openstack-cinder01:46
*** gouthamr_ has quit IRC01:49
*** buster_604 has quit IRC02:02
*** jwcroppe has quit IRC02:13
openstackgerritApoorva Deshpande proposed openstack/cinder: Snapshot GET API to return provider_id
*** dave-mccowan has quit IRC02:14
*** EinstCrazy has quit IRC02:14
*** apoorvad_ has joined #openstack-cinder02:15
*** apoorvad has quit IRC02:18
*** apoorvad_ has quit IRC02:19
*** chlong has quit IRC02:27
*** markvoelker has joined #openstack-cinder02:32
*** martyturner has joined #openstack-cinder02:33
*** markvoelker has quit IRC02:36
*** buster_604 has joined #openstack-cinder02:37
*** martyturner has quit IRC02:37
*** bill_az_ has quit IRC02:40
*** chlong has joined #openstack-cinder02:42
*** r-daneel has quit IRC02:44
*** wN has quit IRC02:48
*** buster_604 has quit IRC02:49
*** stevemar has joined #openstack-cinder02:54
*** wN has joined #openstack-cinder02:54
*** wN has joined #openstack-cinder02:54
*** edtubill has joined #openstack-cinder02:58
*** IlyaG has joined #openstack-cinder02:58
*** stevemar has quit IRC02:59
*** IlyaG has quit IRC03:02
*** changbl has joined #openstack-cinder03:03
*** edtubill has quit IRC03:04
*** changbl has quit IRC03:08
*** IlyaG has joined #openstack-cinder03:08
*** mriedem has quit IRC03:15
*** IlyaG has quit IRC03:21
*** stevemar_ has joined #openstack-cinder03:28
*** gouthamr has quit IRC03:30
*** links has joined #openstack-cinder03:31
*** julim has quit IRC03:33
*** julim has joined #openstack-cinder03:34
*** setmason has joined #openstack-cinder03:35
*** stevemar_ has quit IRC03:41
*** stevemar has joined #openstack-cinder03:42
*** salv-orlando has joined #openstack-cinder03:43
*** salv-orl_ has quit IRC03:45
*** bnemec has quit IRC04:02
*** stevemar_ has joined #openstack-cinder04:07
*** setmason_ has joined #openstack-cinder04:08
*** stevemar has quit IRC04:09
*** setmason has quit IRC04:11
*** setmason has joined #openstack-cinder04:12
*** setmason_ has quit IRC04:15
*** markvoelker has joined #openstack-cinder04:33
*** stevemar has joined #openstack-cinder04:34
*** stevemar_ has quit IRC04:34
*** stevemar has quit IRC04:39
*** apoorvad has joined #openstack-cinder04:48
*** stevemar has joined #openstack-cinder04:50
openstackgerritApoorva Deshpande proposed openstack/cinder: Snapshot GET API to return provider_id
*** setmason has quit IRC04:54
*** jwcroppe has joined #openstack-cinder04:58
*** jwcroppe has quit IRC05:02
*** stevemar has quit IRC05:05
*** stevemar has joined #openstack-cinder05:06
*** mragupat has joined #openstack-cinder05:06
*** markvoelker has quit IRC05:07
*** setmason has joined #openstack-cinder05:09
*** stevemar has quit IRC05:10
*** mudassirlatif has joined #openstack-cinder05:11
*** apoorvad has quit IRC05:24
*** nkrinner has joined #openstack-cinder05:27
*** e0ne has joined #openstack-cinder05:45
*** setmason has quit IRC05:48
*** stevemar has joined #openstack-cinder05:50
*** stevemar_ has joined #openstack-cinder05:51
*** stevemar has quit IRC05:55
*** e0ne has quit IRC05:56
*** setmason has joined #openstack-cinder06:18
*** nkrinner has quit IRC06:22
*** nkrinner has joined #openstack-cinder06:22
*** setmason has quit IRC06:24
*** mragupat has quit IRC06:30
*** wilson1 has quit IRC06:31
*** wilson1 has joined #openstack-cinder06:33
*** medberry has quit IRC06:39
openstackgerrityogeshprasad proposed openstack/cinder: Retype support for CloudByte iSCSI cinder driver
*** med_ has joined #openstack-cinder06:45
*** med_ is now known as Guest4345806:46
*** setmason has joined #openstack-cinder06:56
openstackgerritVictor Stinner proposed openstack/cinder: Fix volume throttling to Python 3
*** lpetrut has joined #openstack-cinder07:03
*** ronis has joined #openstack-cinder07:21
*** mudassirlatif has quit IRC07:22
*** chlong has quit IRC07:31
*** haomaiwang has joined #openstack-cinder07:44
*** mudassirlatif has joined #openstack-cinder07:45
*** openstackgerrit has quit IRC07:46
*** openstackgerrit has joined #openstack-cinder07:47
*** jordanP has joined #openstack-cinder07:47
openstackgerritVictor Stinner proposed openstack/cinder: Fix various Python 3 issues
*** bauzas is now known as bauwser07:56
*** haypo has joined #openstack-cinder07:57
*** haomaiwang has quit IRC08:01
*** haomaiwa_ has joined #openstack-cinder08:01
*** stevemar_ has quit IRC08:01
*** stevemar has joined #openstack-cinder08:02
*** lpetrut has quit IRC08:03
*** stevemar has quit IRC08:06
*** bluex-pl has joined #openstack-cinder08:11
*** bluex-pl has quit IRC08:11
*** bluex-pl has joined #openstack-cinder08:12
*** lpetrut has joined #openstack-cinder08:13
*** setmason has quit IRC08:16
openstackgerritSzymon Wróblewski proposed openstack/cinder: CGSnapshot Object
openstackgerritSzymon Wróblewski proposed openstack/cinder: Cleanup for cinder tests with CGSnapshot
*** harlowja has quit IRC08:19
*** jseiler has quit IRC08:19
*** jseiler has joined #openstack-cinder08:20
*** jwcroppe has joined #openstack-cinder08:23
*** jistr has joined #openstack-cinder08:24
openstackgerritVictor Stinner proposed openstack/cinder: Port scheduler host manager to Python 3
*** vgridnev has quit IRC08:31
*** zhonghua-lee has quit IRC08:39
*** jwcroppe has quit IRC08:40
*** alexschm has joined #openstack-cinder08:44
*** links has quit IRC08:50
*** aix has joined #openstack-cinder08:54
*** mudassirlatif has quit IRC08:56
*** e0ne has joined #openstack-cinder08:57
*** haomaiwa_ has quit IRC09:01
*** haomaiwang has joined #openstack-cinder09:01
*** markvoelker has joined #openstack-cinder09:04
*** markvoelker has quit IRC09:09
*** lpetrut has quit IRC09:10
*** openstack has joined #openstack-cinder09:22
*** openstackstatus has joined #openstack-cinder09:22
*** ChanServ sets mode: +v openstackstatus09:22
*** alexpilotti has quit IRC09:22
*** e0ne has quit IRC09:23
*** vgridnev has joined #openstack-cinder09:23
*** ndipanov has joined #openstack-cinder09:23
*** e0ne has joined #openstack-cinder09:31
*** vgridnev has quit IRC09:31
*** vgridnev_ has joined #openstack-cinder09:31
*** salv-orlando has quit IRC09:33
e0nezigo: hi! do you still need help with config generation?09:37
zigoe0ne: I did a hack which fixes my CI, but it's not satisfying at all, so yes.09:37
*** lprice has quit IRC09:37
e0nezigo: I hope, you resolved all issues today, just want to confirnm09:37
zigoe0ne: enabled_backends is still not present by default, so I'm using wc -l / tail / cat / echo to add it in my debian/rules file.09:37
*** lprice has joined #openstack-cinder09:38
e0nezigo: enabled_backends is not required if you have only one backend09:38
zigoe0ne: Then, I've been told in my bug report that the [lvm] section isn't present by default in the generated cinder.conf.sample. I don't think that's normal !!!09:38
zigoe0ne: Well, the generator should be showing absolutely *all* config options.09:38
zigoIt's like that for every project in OpenStack.09:39
zigoThere's no point in generating the config file if it's not exhaustive.09:39
*** strony-zhang has quit IRC09:42
e0nezigo: we've got mutually options in configs. config generator should know about them09:44
*** dims__ has joined #openstack-cinder09:44
*** haomaiwang has quit IRC10:01
*** haomaiwang has joined #openstack-cinder10:01
openstackgerritAbhishek Shrivastava proposed openstack/cinder: Setup error check & minor bug fix in CloudByte
*** daemontool__ has joined #openstack-cinder10:07
*** daemontool_ has quit IRC10:07
*** ollie1 has left #openstack-cinder10:08
*** pcaruana has joined #openstack-cinder10:13
*** IanGovett has joined #openstack-cinder10:17
*** e0ne has quit IRC10:21
*** e0ne has joined #openstack-cinder10:34
*** jaypipes has joined #openstack-cinder10:36
*** jaypipes is now known as leakypipes10:38
*** ndipanov is now known as ndipanoff10:47
*** ig0r_ has joined #openstack-cinder10:55
*** ig0r_ has quit IRC10:56
*** e0ne has quit IRC10:57
*** takedakn has joined #openstack-cinder10:59
*** haomaiwang has quit IRC11:01
*** haomaiwang has joined #openstack-cinder11:01
*** markvoelker has joined #openstack-cinder11:05
*** markvoelker has quit IRC11:09
*** kevincarr1991 has joined #openstack-cinder11:13
*** salv-orlando has joined #openstack-cinder11:15
*** delatte has joined #openstack-cinder11:16
*** cdelatte has joined #openstack-cinder11:16
*** salv-orlando has quit IRC11:20
*** kevincarr1991 has quit IRC11:24
*** kevincarr1991 has joined #openstack-cinder11:27
kevincarr1991I have installed Cinder on my controller node and used the nfs driver to attach the service to a nfs share. Cinder attaches to the nfs and I am able to create volumes, but I can't attach the volumes to instances using the dashboard11:29
*** salv-orlando has joined #openstack-cinder11:32
*** dzamboni has joined #openstack-cinder11:36
*** julim has quit IRC11:39
*** julim has joined #openstack-cinder11:44
*** lpetrut has quit IRC11:46
*** pcaruana has quit IRC11:47
*** haomaiwang has quit IRC12:01
*** e0ne has joined #openstack-cinder12:01
*** haomaiwang has joined #openstack-cinder12:01
*** zul has quit IRC12:01
*** zul has joined #openstack-cinder12:04
*** lpetrut has joined #openstack-cinder12:05
*** markvoelker has joined #openstack-cinder12:05
openstackgerritKeiichi KII proposed openstack/os-brick: iSCSI Stop calling multipath command line
*** markvoelker has quit IRC12:09
*** markvoelker has joined #openstack-cinder12:09
*** haomaiwang has quit IRC12:10
*** 20WAB697A has joined #openstack-cinder12:11
*** martyturner has joined #openstack-cinder12:11
*** ociuhandu has joined #openstack-cinder12:13
*** 20WAB697A has quit IRC12:25
*** ociuhandu has quit IRC12:27
*** delatte has quit IRC12:36
*** cdelatte has quit IRC12:36
*** edmondsw has joined #openstack-cinder12:37
*** delatte has joined #openstack-cinder12:37
*** cdelatte has joined #openstack-cinder12:37
*** Yogi1 has joined #openstack-cinder12:38
kevincarr1991I have installed Cinder on my controller node. I have Cinder attached to a nfs share. I am able to create volumes but I cannot attach the volumes to any instances. I have tried to attach volumes using the dashboard and cli and both state that they work successfully, but the instances wont have a new volume attached12:40
*** porrua has joined #openstack-cinder12:41
*** CBADER has quit IRC12:41
*** kevincarr1991 has quit IRC12:42
*** vgridnev_ has quit IRC12:43
*** akerr has joined #openstack-cinder12:44
*** vgridnev has joined #openstack-cinder12:44
haypohum, one more time someone asks me to change how to link my Python 3 patches to the Python 3 blueprint with a -1 vote12:44
haypoI'm using the syntax "Blueprint cinder-python3" as I did for my previous +30 changes already merged in Cinder12:44
haypobut Kendall Nelson asked me to write "Partial-Implements: blueprint cinder-python3"12:44
haypocan we please agree on one syntax and stop nitpicking on this stupid link?12:45
dims__haypo: we have to start charging for -1's :) $1 per -112:45
haypo("Partial-Implements: blueprint cinder-python3" and "Blueprint cinder-python3" are technically the same, both add the change to the blueprint when a new change is submitted)12:45
*** leitan has joined #openstack-cinder12:45
haypodims__: +1 for $1 per -112:45
*** diablo_rojo has quit IRC12:46
*** Yogi1 has quit IRC12:46
*** CBADER has joined #openstack-cinder12:46
haypoit's so annoying that i even wrote an email about this line of the commit message:12:46
haypo[openstack-dev] [cinder] How to link a change to the completed cinder-python3 blueprint?12:46
haypo... and i didn't get any reply12:46
hayposo what?12:46
*** bluex-pl has quit IRC12:49
*** eharney has joined #openstack-cinder12:50
*** kevincarr1991 has joined #openstack-cinder12:51
*** leitan has quit IRC12:55
*** leitan has joined #openstack-cinder12:55
*** dustins has joined #openstack-cinder12:57
*** diablo_rojo has joined #openstack-cinder12:58
*** martyturner has quit IRC12:59
*** salv-orlando has quit IRC12:59
*** dave-mccowan has joined #openstack-cinder13:00
*** timcl has joined #openstack-cinder13:02
*** leitan has quit IRC13:02
*** martyturner has joined #openstack-cinder13:02
*** jseiler has quit IRC13:10
*** xyang1 has joined #openstack-cinder13:10
e0nehaypo: partial implements looks better, but I won't -1 if it's only "Blueprint bp-name"13:11
*** leitan has joined #openstack-cinder13:12
*** Yogi1 has joined #openstack-cinder13:13
haypoe0ne: would you be brave enough to write it in a comment at ? thanks :)13:14
e0nehaypo: sure, I'll do13:14
e0nehaypo: probably, I have to post to ML too13:15
openstackgerritAdriano Freires Rosso proposed openstack/cinder: HNAS iSCSI manage does not work with spaces
haypoe0ne: cool :)13:15
*** raildo-afk is now known as raildo13:16
*** merooney has joined #openstack-cinder13:18
*** gouthamr has joined #openstack-cinder13:19
smcginniszigo, e0ne: I think there is a misunderstanding on the config sample output. There should not be an [lvm] section in that file.13:22
*** martyturner has quit IRC13:22
zigosmcginnis: I think *you* are not understanding what other projects are doing. Take for example Nova: they do have a [vmware] section, even though not everyone uses vmware. They do have absolutely *all* of the options shown in the .sample file.13:23
zigoWhy should Cinder be different?13:23
zigoIs it that adding the [lvm] section breaks something if we don't use lvm?13:23
smcginniszigo: It's not. [lvm] would be whatever you name your backend.13:23
zigosmcginnis: How do our users guess the options for lvm then?13:24
smcginnisAnd all the options are in there, including enabled_backends from what I've seen from several tests.13:24
smcginnisAll the lvm options are in there.13:24
zigoAh? Where?13:24
diablo_rojosmcginnis: zigo The reason there is no lvm section being created is becuase when those options get registered, they aren't being registed with a group.13:24
smcginnisWhether you name that backend lvm or not is up to you.13:24
*** jgregor has joined #openstack-cinder13:24
zigoOk, got it.13:25
*** haomaiwa_ has joined #openstack-cinder13:25
smcginnisRight, they are not a group. The examples shown in the docs are for a backend named lvm.13:25
zigosmcginnis: Now, how would users know all of the options of [backend-foo] ?13:25
smcginnisMaybe the doc should be updated to be named [my-lvm] or something to make it obvious.13:25
zigoYup, sounds reasonable.13:25
zigo[lvm-foo] :P13:26
*** mriedem has joined #openstack-cinder13:26
zigosmcginnis: I still insist that I don't have the enabled_backends option anywhere. Sorry to be a pain here, but I think it's important to fix it for every downstream distro, especially that maybe, it hides the fact that some other options may be also not there.13:27
smcginniszigo: No problem, that part would be legitimate.13:27
smcginniszigo: I've not been able to reproduce that. I always get enabled_backends in every place I've tried it.13:28
zigosmcginnis: You only tried with devstack, right?13:28
zigoUsing tox...13:28
smcginnisOnly using tox, but that should really matter if you are passing in from_tox and running in the correct root directory.13:29
smcginnisdiablo_rojo: I'm not lying, am I? ^13:29
diablo_rojosmcginnis: I believe that is correct.13:29
smcginnisLine 544 is my latest test.13:30
smcginniszigo: What's the line count you get? wc -l etc/cinder/cinder.conf.sample13:31
diablo_rojosmcginnis: zigo The reason tox was made to be the only way to run it was because people were running it in other ways from other directories and it was missing options. Tox was decided as being the only way it should be run to guarantee that all options would be picked up.13:31
zigosmcginnis: 805 lines13:32
openstackgerritGorka Eguileor proposed openstack/cinder: Detach volume on device validation failure
diablo_rojozigo: It should have over 300013:32
smcginniszigo: OK, something definitely wrong there. I get over 3000.13:32
zigodiablo_rojo: I understand, but that's not downstream distribution friendly...13:32
zigodiablo_rojo: I can't run tox when building the Debian package, as there's no virtual env involved.13:33
*** jungleboyj has joined #openstack-cinder13:33
zigosmcginnis: Could you, as a temporary hack fix, send me the file?13:33
zigoIt's problematic too, as it wont be generated with the build-depends that I am using in Debian though ...13:33
smcginniszigo: Sure, I can send you my output.13:34
zigoSo it can only be considered as a *very* temporary hack.13:34
smcginnisLet me see how paste.os.o likes 3000 lines. ;)13:34
zigosmcginnis: Is it even possible to past 3000 lines there?13:34
smcginnisI wasn't aware distributors were packaging that any more. I seem to remember a long discussion of getting rid of it.13:34
zigootherwise, zigo@debian.org13:34
smcginniszigo: OK, that's probably better. On its way.13:35
zigosmcginnis: Ubuntu people don't care. *I* do care having a correct conf file in Debian that works by default.13:35
eharneysmcginnis: we package sample conf too13:35
jungleboyjHowdy all, sorry I am just jumping in now zigo and smcginnis13:35
zigojungleboyj: No prob.13:35
smcginniseharney: Oh good, I was just wondering about RH.13:35
smcginniseharney: How do you guys handle it?13:35
zigoeharney: Are you working for Red Hat and doing Cinder packaging?13:36
jungleboyjSo, did we land on the fact that we were running on an old file?13:36
eharneysmcginnis: manually at this point, like zigo referred to, running things like tox and having it install deps in our build environments is completely disallowed, so tox-only is not really a whole solution13:37
*** setmason has joined #openstack-cinder13:37
*** haomaiwa_ has quit IRC13:37
zigoeharney: Thanks for also explaining it! :)13:38
eharneysmcginnis: IMO we should just be checking in a sample conf at the end of the release upstream and shipping it w/ Cinder13:38
*** haomaiwang has joined #openstack-cinder13:38
zigoeharney: That's the wrong answer to the problem.13:38
smcginniseharney: That does sound more reasonable.13:38
zigoeharney: The .sample file *must* be generated from within the dependencies that you have in the OS.13:38
zigoOtherwise, you may risk to have a wrong file.13:38
eharneywe already have code that ensures the sample is generated successfully, we just don't ship it13:38
smcginniszigo: We only generate from within cinder.13:38
eharney(as part of the gate)13:39
*** nkrinner has quit IRC13:39
zigoFor example, what if one module (let's say, oslo.messaging) adds an option, and the version in pypi is different from the one packaged in the distro? Well, I'll tell you what: the sample config file will simply be wrong.13:39
zigoWhich is why it isn't recommended to store the config file in upstream git.13:39
zigoDownstream distros *must* be allowed to build them.13:40
jungleboyjSo, if we need to have the file out there for the deployers do we need to add that to the distribution and add a pep8 check for when it needs to be updated as the result of a change.  That shouldn't happen very often.13:40
jungleboyjzigo: You are right.13:40
eharneyi don't want to build a sample file downstream, because i'm not going to package random dependencies for drivers we don't support, and without doing that, i can't build one w/o tox13:40
eharneyat least, can't build a full sample file13:40
zigoeharney: I have a different view.13:41
* smcginnis wonders if we need a packager working session13:41
zigoI do want to generate it, because I want the sample file to include only drivers that I support.13:41
zigoI don't care whatever driver that isn't maintained within Debian.13:41
eharneyzigo: well, that's more straightforward to solve13:41
jungleboyjsmcginnis: We probably do.  I am glad we are having this discussion.13:41
smcginnisMe too13:41
smcginnisI wasn't aware of any of this.13:42
eharneyzigo: in that case, you just need to get a generate_sample fixed up that runs w/o tox, which shouldn't be that bad... though i don't know the details of all the new dynamic conf stuff13:42
zigosmcginnis: It's been a long time I'm thinking about doing a session about what's going on downstream, and how we expect upstream stuff to behave...13:42
jungleboyjsmcginnis: I wondered if we might hit something like this but had hoped not.13:42
smcginnisAFAIK, as long as you pass in13:42
eharneysmcginnis: distribution build environments are highly controlled, have tracked dependencies, don't allow downloads from the internet during build time, etc13:42
smcginnis"from_tox" and run from the right dir, it should be fine.13:42
zigoeharney: The current issue is that it's broken for me right now.13:42
eharneysmcginnis: it's a whole different world from all of this "just pip install everything" theory upstream13:42
jungleboyjeharney: So, that is what I just proposed above.13:43
zigosmcginnis: Well, it "should" be fine, but it isn't ... :(13:43
*** lprice1 has joined #openstack-cinder13:43
jungleboyjIf we need generate_sample to work without tox, we need to ship an file.13:43
eharneyjungleboyj: i think your proposal makes sense, just don't know the ins and outs of it13:43
zigojungleboyj: Why do you think so?13:43
jungleboyjThe part that zigo is worried about (other libraries) is handled outside of the file.13:43
*** haomaiwang has quit IRC13:43
zigojungleboyj: Ah, good then!13:44
zigo(and, yes of course, stupid me...)13:44
jungleboyjzigo: the oslo-config-generator looks at a separate library for the dependencies.  Right diablo_rojo13:44
*** haomaiwang has joined #openstack-cinder13:44
diablo_rojojungleboyj: correct13:44
*** lprice has quit IRC13:45
eharneyall of this was not an issue when we did "you must update the sample conf at review time when changing options"13:45
*** martyturner has joined #openstack-cinder13:46
jungleboyjSo, what I think diablo_rojo and I could do is add the file to the tree.  Then add a pep8 check like we used to have that will generate the file and verify there are  no changes.  If there are changes it will flag that so the person committing the code knows they need to also update that file.13:46
smcginnisjungleboyj: Ah, I like.13:46
jungleboyjThat should be a rare occasion as that file only changes if a whole new group of options is added.13:46
jungleboyjsmcginnis: What is eating at me is the fact that I feel like there is a reason we didn't do that.13:46
smcginnisAs long as we have some way to enforce it's being done and it's not up to someone to review and realize it's missed.13:46
smcginnisjungleboyj: Yeah...13:46
jungleboyjsmcginnis: Right.13:46
smcginnisBut I think the intent of the generation was to take out user error in missing that.13:47
smcginnisAs long as we can protect against that, I guess I'm fine.13:47
jungleboyjIt isn't going to break the gate  like it used to because we are only checking the opts file which is only based on Cinder code.13:47
jungleboyjIf the developer doesn't run pep8 before pushing up, that isn't our fault.13:48
smcginnisIt's still one more manual step (update the file) but it's guarded.13:48
smcginnisWe have enough missed pepp8 before pushing. ;)13:48
diablo_rojosmcginnis: That we do ;)13:48
jungleboyjYeah, not ideal, but it appears that we can't go fully automated.13:49
*** baumann has joined #openstack-cinder13:49
jungleboyjWant our deployers happy.  :-)13:49
* eharney needs to go study this thing again13:49
jungleboyjdiablo_rojo: Now has a 'pep8!?!' post it note on her computer from jungleboyj13:49
jungleboyjeharney: Look away, look away!13:49
jungleboyjeharney: Just trust us.  :-)13:49
smcginnisAt least thanks to eharney's quick8 patch you can install a hook!13:50
eharneyjungleboyj: upgrade the note to "fast8" :)13:50
*** stevemar has joined #openstack-cinder13:50
jungleboyjHuh?  What is fast8?13:50
*** stevemar has quit IRC13:50
smcginnisThe best thing sinced sliced bread. :)13:50
* smcginnis can't spell today13:50
eharneyjungleboyj: "tox -e fast8"13:50
jungleboyjMmmm, raisin bread?  I am hungry.13:50
jungleboyjOk, I will have to check that out.13:51
smcginnisjungleboyj: Automation!! ^^13:51
diablo_rojosmcginnis: yay for automation!13:51
jungleboyjdiablo_rojo: Oh the irony of that statement at the moment.  :-)13:52
smcginnisSo none of this actually helps deployers now for liberty though. :\13:52
smcginnisjungleboyj: LOL13:52
eharneysmcginnis: i think i'll turn that on and see how life goes trying to use that hook13:52
*** jwcroppe has joined #openstack-cinder13:52
smcginniseharney: Yeah, it would be great if it works well and is commonly used.13:52
smcginnisCould save a lot of time.13:52
kaisersHey Cinders! Scheduling Q: Can the instance_locality_filter be used during nova boot? i.e. can i use create volume from image and boot instance in conjunction with instance_locality_filter in the cinder scheduler in order to achieve booting from an image stored on a local volume?13:53
eharneysmcginnis: i kind of suspect i'm going to end up wanting it as pre-push instead of pre-commit13:54
jungleboyjsmcginnis: We could backport the fix?13:54
* jungleboyj says uneasily13:54
smcginniseharney: True. At least we will have a sample showing how. Should be easy enough to modify. Unless you think I should change that to be the default?13:54
smcginnisjungleboyj: Yeah, not sure.13:54
jungleboyjThe stable maintainer in me says it adds a file and I should say now, but it is for the conf file which seems ok.  We aren't going to break things worse.13:54
eharneysmcginnis: dunno, i'll play with it13:54
* jungleboyj needs to resist the urge to Ninja merge.13:55
*** mriedem1 has joined #openstack-cinder13:55
*** alejandrito has joined #openstack-cinder13:55
smcginnisjungleboyj: Well, I get an out here because I'm not a stable maintainer, so I'll leave that to you. :P13:55
*** mriedem has quit IRC13:55
jungleboyjsmcginnis: Chicken.13:56
eharneysmcginnis: as a side note on stable things... i'm still wondering what's up w/ liberty-rc213:56
jungleboyjsmcginnis: eharney zigo So, are we in agreement that we should add the file to our tree?  That we should then run a check to verify that the code is up to date as part of pep8?13:56
*** merooney has quit IRC13:56
jungleboyjeharney: Me too!13:57
smcginnisjungleboyj: Seriously though, if it is just a matter of adding back in an opt file that's fine. But wouldn't it require a bigger change?13:57
zigojungleboyj: I'm fine with this solution, yes.13:57
eharneyjungleboyj: probably in agreement, i'll make sure once i look at a patch and see how it works13:57
* jungleboyj figured zigo would be. :-)13:57
zigoSorry, I couldn't follow some of the discussion, reading backlog.13:57
jungleboyjeharney: Fair enough.13:57
smcginniseharney: I haven't heard anything lately, but I believe RC2 is imminent.13:57
eharneysmcginnis: just wondering since we have some must-land patches13:58
smcginnisI definitely will also make it a point to have some downstream discussions in Tokyo.13:58
jungleboyjsmcginnis: It is going to be checking in a file and adding a diff between the existing and a new one into pep8.13:58
smcginniseharney: I think we should probably make sure thingee is aware of those.13:58
smcginnis"adding a diff between the existing and a new one" << bigger than it sounds? That's my concern.13:59
jungleboyjsmcginnis: eharney zigo I need to drive into the office.  diablo_rojo and I will get something up hopefully this morning to look at.13:59
jungleboyjsmcginnis: Yee have little faith.  ;-)13:59
smcginnisEverything sounds easy until you dig into it. ;)14:00
jungleboyj:-p  Anyway, gotta drive in.  Back online in a bit.14:00
zigoThanks a lot everyone.14:00
smcginniszigo: Thanks for bringing more of the context to our attention.14:00
*** mriedem1 is now known as mriedem_meeting14:00
smcginniszigo: We obvious need to get a little more educated here.14:00
*** haomaiwang has quit IRC14:01
eharneysmcginnis: there are probably some packaging-related discussions happening in Tokyo already, just not sure what/where14:01
smcginniseharney: I'll check the schedule for them once it's up.14:02
smcginniseharney: Hopefully we don't have conflicts.14:02
* smcginnis feels like the summit could be a few days longer to fit everything in14:02
*** stevemar has joined #openstack-cinder14:02
*** lcurtis has joined #openstack-cinder14:03
e0nesmcginnis: did you mean few weeks?14:03
smcginnise0ne: No doubt. :)14:04
*** haomaiwa_ has joined #openstack-cinder14:04
*** lcurtis has quit IRC14:04
*** jungleboyj has quit IRC14:04
*** lcurtis has joined #openstack-cinder14:04
*** merooney has joined #openstack-cinder14:05
*** stevemar_ has joined #openstack-cinder14:05
*** mdenny has joined #openstack-cinder14:06
*** stevemar has quit IRC14:06
dulekgeguileo: I've seen you've answered "done" to my comments on 205834, but you probably forgot to execute git-review. ;)14:07
*** takedakn1 has joined #openstack-cinder14:08
geguileodulek: I didn't really forgot to do it...  I wanted to see if I can improve something else14:09
*** takedakn has quit IRC14:09
*** takedakn1 is now known as takedakn14:09
dulekgeguileo: :)14:09
geguileoAnd I don't want you guys having to review every little change  XD14:09
*** haomaiwa_ has quit IRC14:12
*** ociuhandu has joined #openstack-cinder14:12
*** haomaiwa_ has joined #openstack-cinder14:13
*** dims__ has quit IRC14:14
*** dims__ has joined #openstack-cinder14:15
*** bswartz has quit IRC14:18
*** erhudy has joined #openstack-cinder14:19
*** ronis has quit IRC14:23
*** takedakn has quit IRC14:26
*** thangp has joined #openstack-cinder14:28
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
*** kevincarr1991 has quit IRC14:30
*** haomaiwa_ has quit IRC14:33
smcginnise0ne: Thanks for the ML list post re: XML API. I'll wait to see if we get any complaints, then approve the spec.14:33
*** timcl has quit IRC14:34
*** kevincarr1991 has joined #openstack-cinder14:35
*** setmason has quit IRC14:35
*** haomaiwang has joined #openstack-cinder14:35
smcginnistbarron: Sorry, I know that's a weird one. I don't really like it myself.14:36
*** martyturner has quit IRC14:36
*** bswartz has joined #openstack-cinder14:36
e0nesmcginnis: np. I'm totally agree that we need get feedback first14:37
tbarronsmcginnis: well, you can make it not puke with "with", you just have to use the right parens for continuation14:37
tbarronsmcginnis: that said, I think I agree that the rule shouldn't be enforced there14:37
*** timcl has joined #openstack-cinder14:37
smcginnistbarron: I could have sworn I tried that when getting rid of contextlib.nested...14:37
tbarronsmcginnis: cause it's kinda ugly14:38
tbarronsmcginnis: well, I typed that form into a python 2.7 interpreter and it liked it14:38
tbarronjust a moment ago14:38
smcginnistbarron: Oh, awesome. Good, maybe if we clean up we can fix it.14:38
smcginnistbarron: We still have a ton of places using \.14:39
smcginnisAt least it's mostly just in tests.14:39
tbarronall that said, I was pushing the point a bit because I personally am often in yuriy's spot and if there are corner-cases where we14:39
tbarrondon't apply rules, then I wonder whether the rules actually make sense to enforce in the first place14:39
*** haomaiwang has quit IRC14:40
tbarronseems like a lot of energy goes into that stuff that might be better channeled14:40
smcginnisThat's certainly true. :\14:40
tbarronI've been doing -1s for commit message line lengths, as I received those myself and thought it was to be enforced.14:40
nikeshmhi,one question,i know answer, just for confirmation, currently most of the cinder backup drivers support incremental feature from kilo onwards, can we add this feature in juno?14:40
tbarronbut in the larger openstack community, there was just a big discussion of that, leaning the other way.14:41
smcginnistbarron: Yeah, I had done that too. Don't know if you saw the LONG ML thread about that. :)14:41
smcginnistbarron: I'm mixed on that.14:41
tbarronsmcginnis: yes, that is the thread I'm referring to./14:41
smcginnisIf it's a guideline then I think it should be enforced.14:41
tbarronsmcginnis: if we had crisper review standards it would save everybody time.14:41
smcginnisBut if it's just a recommendation if you feel like following it, maybe we shouldn't have it.14:41
smcginnisIt sends mixed messages if some -1 on it and some don't.14:42
smcginnistbarron: Agree14:42
*** tsekiyama has joined #openstack-cinder14:43
*** stevemar_ is now known as stevemar14:43
*** dansmith is now known as superdan14:44
smcginnisnikeshm: Juno is security fixes only at this point.14:47
smcginnisnikeshm: You can certainly backport it yourself and offer it as a custom option for your customers.14:47
smcginnisnikeshm: But not officially in tree.14:47
*** timcl1 has joined #openstack-cinder14:53
nikeshmsmcginnis: thanks14:53
*** ociuhandu has quit IRC14:53
*** timcl has quit IRC14:54
*** mtanino has joined #openstack-cinder14:57
*** mriedem_meeting is now known as mriedem14:59
*** edtubill has joined #openstack-cinder14:59
*** ociuhandu has joined #openstack-cinder15:01
*** setmason has joined #openstack-cinder15:02
*** stlbigdog has joined #openstack-cinder15:04
*** takedakn has joined #openstack-cinder15:04
*** lpetrut has quit IRC15:06
openstackgerritxing-yang proposed openstack/cinder: VMAX Truncate Storage Group Name
*** chlong has joined #openstack-cinder15:10
*** lpetrut has joined #openstack-cinder15:13
*** alexschm has quit IRC15:24
*** takedakn has quit IRC15:24
bswartzeharney: ping15:25
eharneybswartz: hi15:25
*** hemnafk is now known as hemna15:27
scottdaI've a change to /etc/cinder/api-paste.ini for my WIP cinder-api-microversions and it needs to be manually copied over from the cinder repo on Devstack in order to get c-api to start....15:28
scottdaThis is causing Grenade to fail in starting Cinder API.15:29
scottdaDoes anyone know how to fix this?15:29
*** martyturner has joined #openstack-cinder15:29
scottdai.e. What needs to get done to make the Grenade tests copy over the change to /etc/cinder/api-paste.ini ?15:29
*** jdurgin1 has joined #openstack-cinder15:30
*** vgridnev has quit IRC15:31
bswartzsmcginnis: ping15:33
*** Yogi1 has quit IRC15:35
*** lpetrut has quit IRC15:36
*** ronis has joined #openstack-cinder15:37
scottdaIs there a way to make upgrade perform the copy of api-paste.ini?15:37
*** setmason has quit IRC15:38
dulekscottda: No. You need to write from-liberty script./15:40
dulekscottda: Something similar to that.15:40
dulekscottda: To get that accepted you need both a PTL approval and a release note entry linked.15:41
scottdadulek: Thanks. That helps.15:42
openstackgerritVictor Stinner proposed openstack/cinder: Port netapp SSC Cmode to Python 3
scottdaso, does that mean Grenade would pass with this script? i.e. does that script get run by Grenade?15:42
openstackgerritVictor Stinner proposed openstack/cinder: Port test_netapp to Python 3
*** setmason has joined #openstack-cinder15:44
dulekscottda: Yes, that's the way it works.15:44
*** harlowja_at_home has joined #openstack-cinder15:44
scottdaOK, cool.15:45
dulekscottda: Grenade runs from-sth while upgrading sth->foo.15:45
dulekscottda: It's annoying you need all that stuff to make a simple change into config files…15:45
scottdaYeah, that's OK. At this point, I'm trying to get Jenkins to pass, so your pointer is a big help.15:47
openstackgerritVictor Stinner proposed openstack/cinder: Port test_quobyte to Python 3
openstackgerritVictor Stinner proposed openstack/cinder: Port test_netapp to Python 3
e0neare anyone looking on failed pytho34 jobs?15:48
e0neI didn't found filed bug15:48
e0nehaypo: ^^15:48
openstackgerritVictor Stinner proposed openstack/cinder: Port netapp SSC Cmode to Python 3
e0neI would like to take a look if needed15:48
haypoe0ne: what is the problem?15:48
haypoe0ne: oh, i always saw it once15:50
haypoe0ne: it may be related to hash randomization15:50
*** tsekiyama has quit IRC15:50
haypoe0ne: the problem is that i was unable to reproduce the bug locally in .tox/py34/ venv15:50
e0nehaypo: I can reproduce it:)15:51
*** lpetrut has joined #openstack-cinder15:51
jgriffithmgagne: ping15:51
haypoe0ne: how? do you run "tox -e py34"? or only test_volume?15:52
haypoe0ne: i'm running python -bb -m cinder.tests.unit.test_volume in the py34 venv15:53
e0nehaypo: "tox -e py34", ubuntu, python 3.4.015:53
haypoe0ne: can you please try my testtools commmand?15:54
haypoe0ne: tox sets the PYTHONHASHSEED to 015:54
e0nehaypo: sure, one moment15:54
xyang1e0ne: are you talking about this one
openstackLaunchpad bug 1501745 in Cinder "gate-cinder-python34 failed" [High,Confirmed] - Assigned to Eric Harney (eharney)15:55
e0nexyang1: yes15:56
xyang1e0ne: eharney made one fix yesterday15:56
eharneyit didn't work15:56
xyang1eharney: oh15:56
e0nehaypo: your command works15:58
haypoe0ne: oh, it is supposed to fail :-p16:00
e0nehaypo: :(16:00
smcginnisLooks like we will have a Liberty RC2 cut early next week.16:01
smcginnisPlease tag anything that should go back with liberty-backport-potential.16:01
haypoe0ne: lol16:01
xyang1smcginnis: so we can backport now?16:01
smcginnisxyang1: Only if it makes sense.16:01
smcginnisIsolated and safe changes only I think.16:01
haypoe0ne: i'm now "fuzzing" the test with various random seeds...16:01
xyang1smcginnis: sure16:02
eharneysmcginnis: not liberty-rc-potential?16:02
smcginnisxyang1: I don't think we want to dump too much in rc2, but if it's already in master and we're going to backport anyway, nows the time.16:02
smcginniseharney: Oh, or that I guess. :)16:02
e0nehaypo: I removed everything except in tox.ini. and now it works:(16:03
haypoe0ne: hum, maybe it's an eventlet trap16:05
eharneye0ne: haypo: so this current bug looks basically identical to the one fixed here:
eharneywhich was just badly-structured code sharing things it shouldn't have been16:07
eharneybut i'm not sure why it's happening again16:07
*** harlowja_at_home has quit IRC16:07
haypoe0ne: hum, maybe it's an eventlet trap16:08
e0neeharney: yes, it looks similar16:08
*** lpetrut has quit IRC16:08
*** lpetrut has joined #openstack-cinder16:08
haypoe0ne: looks good to me16:09
haypoe0ne: it cannot be exactly the same bug, since fake_notifier was modified16:09
*** jdurgin1 has quit IRC16:09
eharneyit may be that the CG test has a race where sometimes it finishes before "volume.create.end" notifications are fired and sometimes it doesn't16:09
*** jordanP has quit IRC16:09
eharneyin the case of the previous bug, those notifications were coming from unrelated tests16:10
haypoeharney: that makes sense, i don't see why we get a backup here16:10
haypoit also explains why the test doesn't fail when it's run alone16:11
*** jordanP has joined #openstack-cinder16:11
*** salv-orlando has joined #openstack-cinder16:11
eharneyhaypo: the previous bug also never showed up until we turned on concurrency in the gate16:11
xyang1eharney: the CG test along should not have race, but I can take a look16:12
eharneyxyang1: thanks16:14
*** apoorvad has joined #openstack-cinder16:14
*** kevincarr1991 has quit IRC16:14
eharneyhaypo: do we have any way to make unit tests run in a more randomized order?16:16
haypoeharney: right now, tox.ini uses a sequential list, but i don't know if they are run in this exact order16:16
eharneyhaypo: they aren't, i was just hoping i could purposefully shuffle them around more to try to shake out problems16:17
haypoeharney: is the low-level stuff under testr, it doesn't have a wide choice of options16:17
e0neeharney:, line #144 - does it mock per test case all one mock for all test cases?16:18
haypoe0ne: if you are able to reproduce the failure, try to remove test_backup_* in tox.ini16:19
*** leeantho has joined #openstack-cinder16:19
haypoe0ne: i expect that the backup notifications come from these tests :)16:19
haypoi'm now running tox -e py34 instead of testtools16:19
e0nehaypo: sounds reasonable:)16:19
eharneyhaypo: are those failures logged somewhere?  the ones i saw in the gate were different notifications i think16:20
eharneye0ne: well... setUp is run per-test16:20
*** vgridnev has joined #openstack-cinder16:21
haypoeharney: you can play with, search for  project:"openstack/cinder" for example16:21
haypoi don't know how to use this tool :-p16:21
haypoask on #openstack-infra16:21
eharneyhaypo: yeah, i've used it a bit16:22
openstackgerritLisaLi proposed openstack/cinder: Copy encryptors from Nova to Cinder
*** mudassirlatif has joined #openstack-cinder16:23
hemnaeharney, ping16:23
eharneyhemna: hey16:24
hemnathis might be unrelated, but I'm seeing strange failures in my os-brick patch related to lio16:24
haypohere we got MismatchError: 'consistencygroup.create.start' != 'volume.create.end'16:24
hemna2 rechecks = same error16:24
*** salv-orlando has quit IRC16:24
krotscheckwilson1: There's a comment on for you.16:24
eharneyhemna: looks like the same thing16:24
krotscheckAlso, in case anyone wants to help look at a CORS patch :D16:25
eharneyit also always seems to be that test and the same notifications... let me go write a query for this16:25
hemnaI've not seen this before16:25
hemnaglanceclient.exc.ServiceUnavailable: ServiceUnavailable (HTTP 503)16:25
hemnaalso that16:25
*** julim has quit IRC16:25
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
e0neeharney: yes, but does mock returns different objects for tests?16:26
*** julim has joined #openstack-cinder16:26
e0neeharney: in othe words: is notifiere mocked only once or each time for every test?16:27
*** _cjones_ has joined #openstack-cinder16:27
eharneye0ne: it's supposed to be different objects16:27
e0nehemna: hi. I've got few questions about brickclient and attch volume. will you be able here in ~2 hours?16:27
e0neeharney: need to check it16:27
hemnae0ne, yah16:28
e0nehemna: cool16:28
eharneye0ne: well it did eliminate a rampant gate failure at the time...16:28
hemnae0ne, I haven't had time to play with it yet16:28
hemnafound some replication issues yesterday that I patched and got those in16:28
hemnawell in gerrit at least16:28
hemnabut am seeing odd cinder py34 failures now16:28
e0neI need to head home. I'll be online in ~2 hours16:28
e0nehemna: I've got questions about user expirience of this tool. what is expectes and what is not16:29
hemnae0ne, ok16:29
*** timcl1 has quit IRC16:30
haypoi started "tox -e py34" on a fresh cinder clone: it pass...16:30
haypoi'm still unable to reproduce the issue on fedora 22 with python 3.4.216:30
mgagnejgriffith: pong16:31
*** e0ne has quit IRC16:31
haypothe workaround is to disable test_volume on python 3.4 (remove it from tox.ini)16:31
*** diogogmt has joined #openstack-cinder16:32
haypoi opened  test_volume randomly fails on Python 3.416:35
openstackLaunchpad bug 1502229 in Cinder "test_volume randomly fails on Python 3.4" [Undecided,New]16:35
*** yuriy_n17 has quit IRC16:35
*** timcl has joined #openstack-cinder16:36
*** jordanP has quit IRC16:36
*** ociuhandu has quit IRC16:39
xyang1eharney: are the notifications not isolated per test?16:40
eharneyxyang1: the fix we made previously was to isolate them per test16:40
xyang1eharney: ok, i don't know how volume can be created before cg16:41
eharneyxyang1: there's enough complexity in here that maybe they aren't isolated properly anymore16:42
eharneyi don't really know how to explain this yet :/16:42
xyang1eharney: I wonder if we should just check if the event happened, but not check the exact order16:43
apoorvadjgriffith: Thanks for the insights on 150197616:43
eharneyxyang1: i dunno, i think the idea of how these tests work is the right thing... something just isn't implemented correctly16:43
*** kevincarr1991 has joined #openstack-cinder16:44
*** setmason has quit IRC16:47
*** diogogmt_ has joined #openstack-cinder16:48
*** dims__ is now known as dimsum__16:49
haypook, i found a search for Kibana: build_name:"gate-cinder-python34" AND build_status:"FAILURE" AND test_create_consistencygroup_from_src16:49
haypotype it in
*** diogogmt has quit IRC16:50
xyang1eharney: if you send notification #1 first and immediately send #2, is it possible that #2 will be received before #116:51
eharneyxyang1: some of the failures indicate that the wrong total number of notifications is present16:51
xyang1eharney: so missing notifications16:52
eharneyxyang1: extra, actually16:52
eharneythis was one:
*** baumann has quit IRC16:52
*** diogogmt_ has quit IRC16:53
*** julim has quit IRC16:53
xyang1I see16:54
hayposorry guys, i have to go. ping me next week if you still hit the issue16:54
haypoyou can take notes in my bug report:
openstackLaunchpad bug 1502229 in Cinder "test_volume randomly fails on Python 3.4" [Undecided,New]16:54
*** julim has joined #openstack-cinder16:54
*** kevincarr1991 has quit IRC16:56
*** diablo_rojo has quit IRC16:56
*** jgregor has quit IRC16:56
*** jgregor has joined #openstack-cinder16:58
eharney is to help debug these failures if anyone would like to review it16:59
xyang1eharney: in this case, backup failed, so it should not send notifications. failed to obtain backup success notification from server.17:00
*** diogogmt has joined #openstack-cinder17:00
xyang1eharney: in that log, backup failed for tsm driver, so no notification should be sent out17:01
eharneyxyang1: which run was that?17:01
*** harlowja has joined #openstack-cinder17:01
xyang1eharney: the one you just posted, copy and paste does not work for me17:02
*** jgregor has quit IRC17:02
eharneyxyang1: i see that message17:02
xyang1eharney: ok17:02
eharneyxyang1: not really sure if that's related or not, though?17:03
xyang1eharney: i don't know why backup failed there, I just noticed that notification should not be expected17:04
eharneyxyang1: well there are tsm tests that purposefully fail backups i think17:05
xyang1eharney: Ok, I haven't looked at their tests yet17:05
*** daneyon has joined #openstack-cinder17:06
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
*** lpetrut has quit IRC17:08
*** daneyon_ has joined #openstack-cinder17:10
*** stlbigdog has quit IRC17:10
*** jistr has quit IRC17:10
*** daneyon has quit IRC17:12
*** garthb has joined #openstack-cinder17:13
*** ociuhandu has joined #openstack-cinder17:14
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
*** daneyon_ has quit IRC17:17
openstackgerritOpenStack Proposal Bot proposed openstack/python-cinderclient: Updated from global requirements
*** ociuhandu has quit IRC17:22
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
bswartzxyang1: ping17:26
xyang1bswartz: hi17:26
*** mriedem has quit IRC17:28
*** garthb has quit IRC17:29
*** daemontool__ has quit IRC17:29
*** daemontool__ has joined #openstack-cinder17:30
*** mriedem has joined #openstack-cinder17:32
*** pv_ has joined #openstack-cinder17:32
pv_is there any way we can store storage backend specific data in the sql database cinder uses? it would eliminate the need for multiple api calls and stuff for us17:33
jgriffithpv_: you can do some things like provider_id17:34
jgriffithbut depends on what specifically you're looking at17:34
hemnajgriffith, didn't we add a generic table to store credentials ?17:34
hemnaI guess it was specific to storing chap credentials for pure's driver17:35
apoorvadpv_: snapshots and volume tables can store provider_id17:35
hemnaI think that was it no ?17:35
jgriffithhemna: yes17:36
jgriffithhemna: but again I don't know specifically what pv_ is asking :)17:36
hemnayah true17:36
pv_ahh i see so it's specific data that is supported17:36
hemnait's not uber 'generic'17:36
jgriffithpv_: no17:37
jgriffithhemna: no17:37
jgriffithpv_: what kind of data do you want to store?17:37
*** ChanServ has quit IRC17:37
jgriffithIs it initiator/connection data?17:37
jgriffithis it mapping Cinder resource ID's to your internal ID's?17:38
hemnapv_, that's for data that you want to store and have passed into initialize_connection time though17:38
jgriffithor is it something completely different?17:38
hemnait's not a general purpose store for driver 'data'17:38
pv_the issue im running into with the way our backend API is set up is  to find a specific snapshot, volume or anything without having its ID i have to iterate through a list object of everything before finding the thing i need17:38
tbarronwe talked about the pure chap case at midcycle Austin and said it would be great to have general purpose driver storage17:38
tbarronbut it didn't happen17:38
hemnatbarron, yah, the fear I thought was it would be a dumping ground17:39
pv_i was talking to my boss and he told me to check with you guys to see if there was a way to upload that info into the sql data base that cinder uses instead of iteratively going through our API every time17:39
jgriffithpv_: hemna and that's what I thought, and why I suggested provider_id fields before out little detour through the db :)17:39
hemnaand drivers would just started talking to the db directly :(17:39
pv_they don't right, its the manager that talks to the db17:39
pv_volume manager17:39
hemnapv_, yah supposed to not talk to the db.17:39
tbarronhemna: but it should be a way to *keep* drivers from talking to the db, as they could return something to the manager17:39
hemnait's 'bad form'17:39
hemnabut a few drivers in certain cases have to currently.17:40
pv_yeah i remember seeing that somewhere in boldish text, the driver isn't supposed to make db calls17:40
*** ChanServ has joined #openstack-cinder17:40
*** sets mode: +o ChanServ17:40
hemnabut we'd like to prevent it if at all possible17:40
tbarronhemna: pv_: and if they "have" to today, if they had a way to return some private stuff to the manager, manager could stash it for them.17:40
jgriffithpv_:  when you create a volume or snapshot you have the ability to return provider_id in the model17:41
tbarronnote that I avoided use of the word that must not be said (starts with 'm') and said 'private stuff' instead17:41
jgriffithpv_: that is intended to be for example your devices internal ID of the resource17:41
hemnajgriffith, +117:41
jgriffithpv_: so that then when say a delete or whatever comes through you can just access that17:41
pv_thats awesome, thats what i need because our id is different than cinder's id scheme17:41
bswartzjgriffith: you mean provider_location?17:41
pv_i see thanks17:41
pv_ill definitely look into that17:41
bswartzor is provider_id different than provider_location?17:41
jgriffithpv_: it's not required, but it's a handy resource for some that can't do look ups very easily17:41
jgriffithbswartz: yes!17:42
pv_yeah definitely would help i think17:42
jgriffithit's VERY different17:42
pv_thanks guys17:42
jgriffithbswartz: which is why it has it's own column :)17:42
hemnapv_, and you need to be careful with that during volume migration17:42
*** rhagarty_ has quit IRC17:42
jgriffithbswartz: provider_location is the actual target info... like iscsi IP, iqn, chap etc17:43
jgriffithbswartz: first thought was to put provider id in there, but there's a risk of overflowing the 255 char limit :)17:44
bswartzjgriffith: do you recall when provider_id was added? is it new in liberty or kilo?17:45
apoorvadbswartz: Kilo17:45
jgriffithYeah... what apoorvad said17:45
bswartzthat's a good feature17:46
jgriffithbswartz: if you consider using it keep in mind there are some pretty big problems with upgrades17:46
bswartzwhat kind of problems?17:46
jgriffithbswartz: well, if you change your driver to just use that and dump your old lookups, you might be surprised17:46
jgriffithbswartz: after an upgrade for example (to the new driver) you won't have that info17:46
bswartzoh right17:46
jgriffithbswartz: unless you implement my little update mapping on init17:47
jgriffithbswartz: as of L there's a manager call to do that so that's good17:47
jgriffithbswartz: honestly I found that it's actually easier/more robust in my case to just keep doing the lookups17:47
bswartzyeah I would hope that would go without saying that volumes which existed at upgrade time wouldn17:47
bswartzwouldn't have a value in that column17:47
*** anish__ is now known as anish17:47
jgriffithbswartz: yeah17:48
jgriffithbswartz: there's a number of little corner cases17:48
jgriffithbswartz: the other is things like a failure on create17:48
jgriffiththe provider_id field never gets set17:48
jgriffithbut depending on where/when the failure occured you may have a resource on the backend17:48
bswartzoh we have a strategy to address that kind of problem in manila17:49
bswartzin manila the driver can attach a model update to the exception it throws17:49
jgriffithIn my case after looking at a number of corner cases like that it was honestly easier/safer to just keep doing the lookups the way I'd always done17:49
jgriffithUntil I rewrite my driver to V3 :)17:49
jgriffithwhich means by by backports... so I'm avoiding it17:49
bswartzbut yeah with no model updates the error state volume will be hard to fix17:50
jgriffithyeah, it's easy to just add a "hard lookup on delete if provider-id == None" or something17:50
jgriffithbut anyway... just little things that in looking at other peoples code they didn't think of17:51
bswartzyeah backports suck17:51
jgriffithbswartz: indeed... I'd love to rewrite and make a much cleaner/optimized version of my driver17:51
jgriffithbut then I loose the ability to do "things"17:51
jgriffithIf my driver were out of tree I could write something super elegant/cool :)17:52
bswartzif only there weren't any damn customers....17:52
jgriffithMake an "uber" driver that works for all OpenStack versions17:52
jgriffithLOL... yeah, the price of success right :)17:52
jgriffithbswartz: my latest experiment is a type of micro-versions for drivers17:53
jgriffithbswartz: but the only way I can conceivably see that ever working is by introducing a manager for my own driver17:53
jgriffithbswartz: because I know better than to think I'll get everybody to agree/implement it :)17:54
bswartzI don't think that out-of-tree drivers is the panacea some people think it is17:55
bswartzyou missed the lively discussion on that topic in ft. collins17:55
bswartzyou're right that it would make backports less painful, but it would make compatibility with future cinders much harder17:56
jgriffithnahh... makes EVERYTHING easier :)17:58
bswartzalso nothing prevents you from maintaining and out-of-tree driver in addition to your upstream driver -- netapp does this
jgriffithbswartz: indeed17:58
smcginnisbswartz: Isn't that a maintenance headache?17:59
bswartzsmcginnis: you have no idea17:59
tbarronsmcginnis: not for bswartz. he's doing manila :-)17:59
tbarronsmcginnis: ++18:00
bswartzbut it's the only choice if you want to backport to releases that no longer exist upstream18:00
smcginnistbarron: Hah!18:00
tbarronyeah, we have to do it b/c we have to support customers more than 6-12 months back18:00
* akerr still thinks OpenStack should start implementing LTS releases18:01
*** e0ne has joined #openstack-cinder18:01
bswartzakerr: I started a ML thread on that and got shot down18:01
tbarronit would be a big investment by the community, but we at least would benefit.  probably helion and rhel osp too, right?18:02
tbarronnow we all have to maintain backports and fixes downstream.18:02
e0nehaypo, eharney: did I miss anything with py34?18:03
tbarronsmcginnis: does dell or dell/compellent have to do this?18:03
*** Lee1092 has quit IRC18:03
smcginnistbarron: Not yet at least. I won't be surprised if we end up in the same situation though.18:03
eharneye0ne: i don't think so18:04
akerrbswartz: :(18:04
e0neeharney: :(.18:04
e0neeharney: I will take a look on it this weekend or Monday it issue will be still reproducible18:05
*** delatte has quit IRC18:06
*** cdelatte has quit IRC18:06
*** gouthamr has quit IRC18:06
tbarronOn a different subject, I've recently had occasion to work with backup to swift with ssl/tls enabled.  The ssl part wasn't what we were18:06
*** garthb has joined #openstack-cinder18:06
tbarrontesting.  We couldn't turn it off. And there were certificate issues that were painful.18:07
xyang1e0ne: do you want to merge this to help debug the problem
e0nexyang1: looking on it18:07
*** ociuhandu has joined #openstack-cinder18:07
tbarronI noticed that glance has an option (like curl -k) to run with https and ignore ssl cert issues.18:07
tbarronSeems to me that this might be useful in cinder backup, if only for test purposes.  Certainly off by default.18:08
tbarronAnyone see problems with implementing this?18:08
smcginnistbarron: Seems reasonable if it is off by default.18:08
tbarronsmcginnis: thanks for the feedback18:09
e0neeharney: done18:09
e0nehemna: are you still around?18:09
e0nexyang1: done18:09
xyang1e0ne: thanks18:10
hemnae0ne, yuh18:11
e0nehemna: so, I've got iscsi PoC and tried to implement rbd attach18:11
e0nehemna: as you said in the comments "user should not may know what protocol do we use"18:12
hemnais that the latest ?18:12
e0nehemna: yes, rbd part is not ready18:12
hemnaso I think os-brick needs something as well to make sure this works best18:12
e0nehemna: TBH, nothing new, I only moved all code out of cinderclient18:12
hemnathe client needs a way to test to see which connectors will even work on the host.18:13
hemnawe kinda know18:13
hemnare: FC HBA's need to exist and at least 1 listed as active18:13
hemnafor a PoC, we can live w/o it for the time being18:13
e0nehemna: good point, noted18:14
hemnaI had plans on looking at that for M18:14
e0nehemna: this client it mu highest priority at the moment18:15
hemnaand adding a new connector API that can be called to test to see if it think it's "enabled" on the host18:15
e0nehemna: my quation is: for different protocols, we need to install some tools: open-iscsi, ceph-common, etc18:15
e0nehemna: how should we handle it?18:16
hemnaand then add get_enabled_connectors() or something and return a dict of enabled/disabled connectors18:16
hemnae0ne, yup18:16
hemnasome connectors need tools installed, drivers enabled, etc18:16
hemnaFC needs at least 1 active HBA18:16
*** Yogi1 has joined #openstack-cinder18:16
hemnathe vendor specific connectors need their apps installed, etc18:16
e0nehemna: yea, this is a question. we need to document it and let user to install needed software18:17
hemnaso my plan was to add a new API to the connector to ask if it's viable/enabled something like that18:17
hemnathen each connector could do it's own test for it's required things18:17
e0nehemna: we can't get it works out of the box w/o additional tools:(18:17
hemnathat's another good point18:18
hemnashould update the documentation to the lib to say what's needed for each connector to work18:18
*** Yogi11 has joined #openstack-cinder18:18
e0nehemna: ageed18:18
e0nehemna: I'm glad that we are on the same page about this usability issue18:19
hemnaso for now, I'd ignore the testing of the connector for the PoC18:19
e0nehemna: will you have some time later to implement attachement using FC?18:19
e0nehemna: I didn't have env to test it18:19
hemnae0ne, yah I can try it18:20
hemnaprobably should add a call to reserve_volume18:20
e0nehemna: cool18:20
*** Yogi1 has quit IRC18:21
e0nehemna: noted18:21
*** daemontool__ has quit IRC18:22
e0neok, will do, thanks18:22
hemnaand try blocks around that18:22
hemnaexpecting exceptions18:22
hemnaetc etc18:22
e0nehemna: sure18:23
e0nehemna: it's all in my TODO list. I'll add it to get github18:23
e0nehemna: thank you for the feedback. I'll update this poc on Monday18:24
hemnaI'll see if I can do a test18:24
*** julim has quit IRC18:25
e0nehemna: I've tested it only throw iscsi. rbd stuff is almost ready. will be ready early next week, Mon or Tue18:25
e0nehemna: please, ping me if you'll get any feedback18:26
e0neI'm going to take a break until for the weekend18:26
e0nehave a good Friday!18:27
smcginnise0ne: Have a good one!18:27
*** e0ne has quit IRC18:28
*** thangp has quit IRC18:29
*** julim has joined #openstack-cinder18:30
*** salv-orlando has joined #openstack-cinder18:31
*** timcl has quit IRC18:33
*** gouthamr has joined #openstack-cinder18:37
*** salv-orlando has quit IRC18:47
*** timcl has joined #openstack-cinder18:47
*** e0ne has joined #openstack-cinder18:48
*** dzamboni has quit IRC18:52
*** cfriesen has joined #openstack-cinder18:56
cfriesenwhen using ceph/rbd for cinder volumes, is it safe to set volume_clear=none ?  (i.e. will ceph return zeros if the user tries to access parts of their volume that they haven't written to, or will they get the previous contents of the disk at that location?)18:58
cfriesenfrom what I've read RBD uses thin provisioning, which would seem to imply that the above is true.19:00
*** bswartz has quit IRC19:00
eharneycfriesen: the rbd driver doesn't use that option19:01
eharneycfriesen: deletions are handled within ceph, new volumes won't get old data19:02
cfrieseneharney: you mean the "volume_clear" option?  so when using ceph we don't actually zero out the volumes first even if it's set to volume_clear=zero?19:03
eharneycfriesen: correct, only the LVM and block_device drivers use that option to actually do any wiping19:03
eharneycfriesen: and that's just because those really need it, in general, storage backends are assumed to handle this sufficiently for us19:04
*** marcusvrn_ has quit IRC19:05
*** julim has quit IRC19:05
*** julim has joined #openstack-cinder19:05
cfrieseneharney: okay, thanks for the clarification.  Just to confirm, if using LVM with thin provisioning then does it automatically ignore the volume_clear setting or do I have to explicitly set it to none?19:06
eharneycfriesen: it ignores it19:06
cfrieseneharney: how refreshingly logical. :)19:06
*** cfriesen has left #openstack-cinder19:07
*** e0ne has quit IRC19:08
*** erlon has quit IRC19:08
*** wolsen has quit IRC19:10
*** julim has quit IRC19:10
*** julim has joined #openstack-cinder19:11
*** merooney has quit IRC19:12
*** daemontool_ has joined #openstack-cinder19:14
*** julim_ has joined #openstack-cinder19:16
*** wolsen has joined #openstack-cinder19:16
*** tsekiyama has joined #openstack-cinder19:16
openstackgerritNate Potter proposed openstack/cinder: Added more options while uploading volume as image
*** julim has quit IRC19:17
*** e0ne has joined #openstack-cinder19:22
*** subscope has joined #openstack-cinder19:24
*** subscope_ has joined #openstack-cinder19:25
*** marcusvrn_ has joined #openstack-cinder19:26
*** martyturner has quit IRC19:33
*** stevemar has quit IRC19:35
*** david-lyle has quit IRC19:35
*** akerr has quit IRC19:35
*** merooney has joined #openstack-cinder19:35
*** david-lyle has joined #openstack-cinder19:35
*** merooney has quit IRC19:37
*** porrua has quit IRC19:40
*** e0ne has quit IRC19:40
openstackgerritNate Potter proposed openstack/python-cinderclient: Add options when uploading images to Glance
*** Yogi11 has quit IRC19:45
*** david-ly_ has joined #openstack-cinder19:45
openstackgerritNate Potter proposed openstack/python-cinderclient: Add options when uploading images to Glance
*** david-lyle has quit IRC19:49
*** david-ly_ has quit IRC19:49
*** apoorvad has quit IRC19:59
*** apoorvad has joined #openstack-cinder20:12
*** tsekiyama has quit IRC20:12
*** akerr has joined #openstack-cinder20:16
*** tsekiyama has joined #openstack-cinder20:17
nikeshmis there any good link for how to use ceph cluster in devstack20:22
nikeshmi mean external ceph cluster20:23
nikeshmnot devstack created ceph cluster20:23
*** david-lyle has joined #openstack-cinder20:24
*** akerr has quit IRC20:26
*** bauwser is now known as bauzas20:28
openstackgerritNate Potter proposed openstack/python-cinderclient: Add options when uploading images to Glance
*** leakypipes has quit IRC20:28
*** gouthamr has quit IRC20:30
*** ntpttr has joined #openstack-cinder20:32
*** timcl has quit IRC20:34
*** Yogi1 has joined #openstack-cinder20:38
openstackgerritScott DAngelo proposed openstack/cinder: cinder-api-microversions code
*** julim_ has quit IRC20:47
*** tsekiyama has quit IRC20:47
*** julim has joined #openstack-cinder20:48
*** setmason has joined #openstack-cinder20:52
*** julim has quit IRC20:53
*** julim has joined #openstack-cinder20:53
*** gouthamr has joined #openstack-cinder20:53
*** ronis has quit IRC20:56
*** raildo is now known as raildo-afk20:56
*** julim has quit IRC20:58
*** tsekiyama has joined #openstack-cinder20:58
*** tsekiyam_ has joined #openstack-cinder21:03
*** dustins has quit IRC21:03
*** mtanino has quit IRC21:04
*** bswartz has joined #openstack-cinder21:05
*** Yogi1 has quit IRC21:06
*** ntpttr has quit IRC21:06
*** tsekiyama has quit IRC21:07
*** tsekiyam_ has quit IRC21:07
*** stevemar has joined #openstack-cinder21:12
openstackgerritxing-yang proposed openstack/cinder: VMAX Target iSCSI IP Address
*** ociuhandu has quit IRC21:18
*** salv-orlando has joined #openstack-cinder21:19
*** gman-tx has joined #openstack-cinder21:19
*** kevincarr1991 has joined #openstack-cinder21:25
*** kevincarr1991 has quit IRC21:26
*** diogogmt_ has joined #openstack-cinder21:26
*** salv-orlando has quit IRC21:27
*** diogogmt has quit IRC21:29
*** diogogmt_ is now known as diogogmt21:29
*** netapp has joined #openstack-cinder21:32
*** CBADER has quit IRC21:41
*** bnemec has joined #openstack-cinder21:44
*** jungleboyj has joined #openstack-cinder21:44
*** marcusvrn_ has quit IRC21:45
*** edtubill has quit IRC21:47
*** mriedem has quit IRC21:51
*** subscope_ has quit IRC21:56
*** xyang1 has quit IRC22:03
*** bnemec has quit IRC22:06
*** ntpttr has joined #openstack-cinder22:07
*** dave-mccowan has quit IRC22:08
*** dimsum__ has quit IRC22:09
*** gouthamr has quit IRC22:10
*** gman-tx has left #openstack-cinder22:11
*** stevemar has quit IRC22:14
*** stevemar has joined #openstack-cinder22:14
*** stevemar has quit IRC22:18
*** gman-tx has joined #openstack-cinder22:19
*** gman-tx has left #openstack-cinder22:25
*** lcurtis has quit IRC22:26
*** jwcroppe has quit IRC22:26
*** ntpttr has quit IRC22:28
*** subscope has quit IRC22:30
*** julim has joined #openstack-cinder22:34
openstackgerritJay Bryant proposed openstack/cinder: Remove unused 'deprecated' option from auth_strategy opt
*** david-lyle has quit IRC22:45
*** edmondsw has quit IRC22:46
*** david-lyle has joined #openstack-cinder22:47
*** jasondotstar is now known as jasondotstar|afk22:48
*** gouthamr has joined #openstack-cinder22:52
*** setmason has quit IRC22:53
*** IanGovett has quit IRC22:55
*** jwcroppe has joined #openstack-cinder22:58
*** mdenny has quit IRC22:59
*** jwcroppe has quit IRC22:59
*** jwcroppe has joined #openstack-cinder23:00
*** dimsum__ has joined #openstack-cinder23:02
*** jwcroppe has quit IRC23:05
*** eharney has quit IRC23:09
*** julim has quit IRC23:10
*** alejandrito has quit IRC23:10
*** julim has joined #openstack-cinder23:11
*** leitan has quit IRC23:15
*** julim has quit IRC23:17
*** julim has joined #openstack-cinder23:18
*** setmason has joined #openstack-cinder23:22
*** julim has quit IRC23:23
*** gouthamr has quit IRC23:24
*** julim has joined #openstack-cinder23:24
*** hemna is now known as hemnafk23:24
*** julim has quit IRC23:28
*** wN has quit IRC23:32
*** garthb has quit IRC23:34
*** wN has joined #openstack-cinder23:34
*** wN has joined #openstack-cinder23:34
*** smcginnis has quit IRC23:41
*** Swanson has quit IRC23:42
*** dimsum__ has quit IRC23:46
*** erhudy has quit IRC23:49
*** diogogmt_ has joined #openstack-cinder23:51
*** pv_ has quit IRC23:51
*** diogogmt has quit IRC23:52
*** diogogmt_ is now known as diogogmt23:52
*** julim has joined #openstack-cinder23:54
*** diogogmt has quit IRC23:54
*** diogogmt has joined #openstack-cinder23:56
*** setmason has quit IRC23:56

Generated by 2.14.0 by Marius Gedminas - find it at!