Friday, 2016-02-19

*** vmorris has quit IRC00:00
DuncanTmc_nair: It's 2am for me now, but I'll take a look at a more sensible hour00:02
openstackgerritMerged openstack/os-brick: Include multipath -ll output in failed to parse warning
*** tobe has quit IRC00:06
*** sdague has quit IRC00:06
openstackgerritKurt Martin proposed openstack/cinder: 3PAR get host by WWN now handles mixed cases
*** Fdaisuke has joined #openstack-cinder00:12
*** suhani has joined #openstack-cinder00:12
*** suhani has quit IRC00:13
*** mriedem has joined #openstack-cinder00:13
mc_nairDuncanT: ah that'd be awesome :) I'm going to pop on later at a less sensible hour for me, so we can sync up then00:13
*** suhani has joined #openstack-cinder00:13
*** mylu has joined #openstack-cinder00:13
*** Abhijith has joined #openstack-cinder00:15
*** Abhijith has quit IRC00:16
*** suhani has quit IRC00:16
*** Abhijith has joined #openstack-cinder00:16
*** Suhani has joined #openstack-cinder00:16
*** Suhani has quit IRC00:19
AbhijithHey guys00:19
*** Suhani has joined #openstack-cinder00:19
Abhijithanyone here?00:19
*** mylu has quit IRC00:19
AbhijithI have some queries regarding contributing a new cinder driver00:19
hemnaAbhijith, hey00:20
*** diogogmt has quit IRC00:21
Abhijithjust to give you guys some background info.. I work for a company called Hedvig and we've developed a cinder driver and done all the testing at our end..00:21
*** mylu has joined #openstack-cinder00:21
Abhijithwe now want to contribute the driver to the openstack release00:21
AbhijithI found out that the deadline for contributing a driver is already past for the Mitaka release00:21
openstackgerritMerged openstack/cinder: Rtstool sets wrong exception message on save
hemnayah we are already past the Mitaka deadline for new drivers00:21
Abhijithso here's my question - is it possible to contribute a driver to the older releases of openstack?00:22
SwansonMitaka 1, right?00:22
hemnabut there is no reason why you can't submit the patch and work on getting CI up and running00:22
patrickeastits a good time to start on getting your CI system ready (if you don't have one) to get it in the N release00:22
SwansonYou can't get a new driver into an old release.00:22
hemnaAbhijith, no, they won't accept new code into already released versions00:22
*** mylu has quit IRC00:22
*** mylu has joined #openstack-cinder00:22
Abhijithokay, thanks..00:22
hemnaAbhijith, read that wiki url I just gave you.  that'll give you a good place to start00:23
Abhijithis the CI setup a prerequisite for posting a review? or can I post the review for the cinder driver and work on the CI setup in parallel?00:23
hemnaAbhijith, it's a prerequisite to getting the code to land.00:23
hemnaAbhijith, you can submit a review any time.00:23
patrickeastyou can do them in parallel, but i guarantee the ci setup will end up blocking the review00:23
patrickeastit takes much more time/effort00:23
SwansonIt will not merge until the CI is in place.00:24
Abhijith@patrickeast, alright, that makes sense00:24
kmartinpatrickeast, CI can be setup in day :)00:24
patrickeastkmartin: well, yea, if you already know how it works00:24
hemnaAbhijith, talk to this guy about it asselin00:24
openstackgerritMerged openstack/cinder: Roll back reservations quota in RPC if necessary
AbhijithTBH, we are having some troubles setting up the CI :)00:25
Abhijithso it might take more than a day for us00:25
patrickeastAbhijith: haha, don't sweat it, unfortunately they are a huge pain to setup00:25
SwansonAbhijith, if you only have some trouble you're doing a great job.00:25
Abhijithto put it politely, some trouble :)00:26
AbhijithI think I took far less time to develop the cinder driver00:26
patrickeastAbhijith: if you haven't already I recommend attending some of the thirdparty meetings they are a good resource for working through issues00:26
kmartinAbhijith, if your using common CI, many in here and openstack-infra can help00:26
patrickeastand lots of us have gone through it so we can probably help00:26
*** mylu has quit IRC00:28
Abhijithokay cool.. thanks guys..00:28
AbhijithI have been following these 2 links to setup CI00:28
Abhijithare these the right links to follow or do you guys recommend any other resources to setup CI?00:29
patrickeastAbhijith: ehh be careful with the joinfu one, its very out of date00:29
patrickeastAbhijith: the concepts and info about configuration for some of the stuff like zuul are still ok but thats about it00:30
hemnaI guess that's the original spec00:30
bardiaAbhijith, try this:
bardiaI setup our CI using the guide above. asselin and the infra team put it together00:31
asselinAbhijith, the other links are good background information, but use use it for more than that. is the 'official' new source that's being actively maintained.00:31
*** salv-orlando has quit IRC00:32
bardiaI can vouch that it works :)00:32
*** jamielennox|away is now known as jamielennox00:33
Abhijithokay cool.. let me give it a shot then..00:34
Abhijithif I run into any issues, I'll come back here and post my questions or join the next thridparty meeting00:34
Abhijiththanks for responding guys.. this was really helpful00:34
*** mylu has joined #openstack-cinder00:36
*** laughterwym has joined #openstack-cinder00:39
*** laughterwym has quit IRC00:39
*** laughterwym has joined #openstack-cinder00:40
mc_nairjgriffith: you still around?  I've had some time to digest your comments on more00:44
*** cknight has joined #openstack-cinder00:47
*** mylu has quit IRC00:49
*** cknight1 has joined #openstack-cinder00:49
*** qeelee has joined #openstack-cinder00:51
*** cknight has quit IRC00:52
*** IlyaG has joined #openstack-cinder00:59
*** yangyapeng has joined #openstack-cinder01:03
*** tobe has joined #openstack-cinder01:04
*** angela-s has quit IRC01:05
*** JoseMello has joined #openstack-cinder01:06
*** yangyapeng has quit IRC01:07
openstackgerritWilson Liu proposed openstack/cinder: Huawei: Manage volume fails due to lower array version
*** IlyaG has quit IRC01:11
*** ntpttr has quit IRC01:12
*** IlyaG has joined #openstack-cinder01:16
*** IlyaG has quit IRC01:16
*** Suyash has quit IRC01:17
*** diogogmt has joined #openstack-cinder01:18
lixiaoy1dulek: You are welcome to rework the patch
*** davechen has joined #openstack-cinder01:22
*** jroll has quit IRC01:22
*** mylu has joined #openstack-cinder01:23
*** jroll has joined #openstack-cinder01:24
*** mylu has quit IRC01:26
*** Thelo has quit IRC01:26
*** yangyapeng has joined #openstack-cinder01:27
*** leeantho has quit IRC01:28
*** EinstCrazy has joined #openstack-cinder01:28
*** laughterwym has quit IRC01:29
*** Thelo has joined #openstack-cinder01:29
*** laughterwym has joined #openstack-cinder01:29
*** lprice1 has quit IRC01:29
*** laughterwym has quit IRC01:30
*** laughterwym has joined #openstack-cinder01:30
*** mylu has joined #openstack-cinder01:31
*** yangyape_ has joined #openstack-cinder01:31
*** EinstCrazy has quit IRC01:32
*** yangyap__ has joined #openstack-cinder01:33
*** mylu has quit IRC01:33
*** yangyapeng has quit IRC01:34
*** EinstCrazy has joined #openstack-cinder01:35
*** yangyape_ has quit IRC01:36
*** baojg has joined #openstack-cinder01:48
lixiaoy1hemna: ping? May I talk to you about your comments in the patch add sheepdog support in os-brick?01:48
*** cheneydc has joined #openstack-cinder01:51
*** mudassirlatif has quit IRC01:53
*** chlong_ has joined #openstack-cinder01:56
*** IlyaG has joined #openstack-cinder01:56
*** tobe has quit IRC01:56
*** laughterwym has quit IRC01:57
*** laughterwym has joined #openstack-cinder01:57
*** IlyaG has quit IRC02:00
*** jamielennox is now known as jamielennox|away02:02
*** lprice has joined #openstack-cinder02:02
*** laughterwym has quit IRC02:02
*** winston-d_ has joined #openstack-cinder02:02
*** david-lyle_ has joined #openstack-cinder02:04
*** david-lyle_ is now known as david-lyle02:05
*** erlon has quit IRC02:06
*** qeelee has quit IRC02:10
*** qeelee has joined #openstack-cinder02:11
*** jamielennox|away is now known as jamielennox02:12
*** john5223 has joined #openstack-cinder02:14
*** aorourke__ has joined #openstack-cinder02:15
*** aorourke_ has quit IRC02:15
*** Abhijith has quit IRC02:15
*** aorourke__ has quit IRC02:16
*** aorourke__ has joined #openstack-cinder02:16
*** haomaiwang has joined #openstack-cinder02:16
openstackgerritXinXiaohui proposed openstack/cinder: Update detach_volume() with versionedobjects
openstackgerritXinXiaohui proposed openstack/cinder: Update attach_volume() with versionedobjects
openstackgerritXinXiaohui proposed openstack/cinder: Update test cases related to attach_volume and detach_volume API
*** bardia has quit IRC02:19
*** haomaiwang has quit IRC02:21
*** jamielennox is now known as jamielennox|away02:24
*** apoorvad has quit IRC02:25
*** jamielennox|away is now known as jamielennox02:25
*** jamielennox is now known as jamielennox|away02:26
*** dims has joined #openstack-cinder02:28
*** dims_ has quit IRC02:29
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
*** cknight1 has quit IRC02:32
*** haomaiwang has joined #openstack-cinder02:32
*** jamielennox|away is now known as jamielennox02:33
*** markvoelker_ has quit IRC02:34
*** jamielennox is now known as jamielennox|away02:34
*** jamielennox|away is now known as jamielennox02:35
*** jamielennox is now known as jamielennox|away02:36
*** haomaiwang has quit IRC02:37
*** Suhani has quit IRC02:37
*** apoorvad has joined #openstack-cinder02:37
*** cknight has joined #openstack-cinder02:37
*** laughterwym has joined #openstack-cinder02:42
*** qeelee_ has joined #openstack-cinder02:42
*** laughterwym has quit IRC02:43
*** qeelee has quit IRC02:43
*** qeelee_ is now known as qeelee02:43
*** laughterwym has joined #openstack-cinder02:47
*** haomaiwa_ has joined #openstack-cinder02:48
*** jamielennox|away is now known as jamielennox02:51
*** houming has joined #openstack-cinder02:55
*** apoorvad has quit IRC02:58
*** haomaiwa_ has quit IRC03:01
*** haomaiwa_ has joined #openstack-cinder03:01
*** JoseMello has quit IRC03:08
*** salv-orlando has joined #openstack-cinder03:09
openstackgerritMerged openstack/python-cinderclient: Fix return type in backup docstring
*** salv-orlando has quit IRC03:11
*** ponychou has quit IRC03:18
*** krotscheck_dr is now known as krotscheck_afk03:18
*** Lee1092 has joined #openstack-cinder03:18
*** esker has joined #openstack-cinder03:27
*** links has joined #openstack-cinder03:30
*** qeelee has quit IRC03:30
*** dims has quit IRC03:32
*** markvoelker has joined #openstack-cinder03:34
*** dims has joined #openstack-cinder03:34
*** dims has quit IRC03:37
*** markvoelker has quit IRC03:39
*** david-lyle_ has joined #openstack-cinder03:42
*** david-lyle has quit IRC03:45
*** baojg has quit IRC03:53
*** qeelee has joined #openstack-cinder04:00
*** david-lyle has joined #openstack-cinder04:00
*** guitarza1 is now known as guitarzan04:00
*** haomaiwa_ has quit IRC04:01
*** haomaiwang has joined #openstack-cinder04:01
*** david-lyle_ has quit IRC04:03
*** david_lyle__ has joined #openstack-cinder04:03
*** david-lyle has quit IRC04:04
openstackgerritMerged openstack/cinder: Coho volume stats update
*** jwcroppe has quit IRC04:07
*** david_lyle__ has quit IRC04:08
*** jwcroppe has joined #openstack-cinder04:08
*** qeelee has quit IRC04:10
*** jwcroppe has quit IRC04:13
*** qeelee has joined #openstack-cinder04:16
*** mriedem has quit IRC04:18
*** mudassirlatif has joined #openstack-cinder04:20
*** qeelee has quit IRC04:20
*** asselin__ has quit IRC04:30
*** asselin_ has joined #openstack-cinder04:30
*** qeelee has joined #openstack-cinder04:31
*** bardia has joined #openstack-cinder04:39
*** qeelee has quit IRC04:39
*** vivekd has joined #openstack-cinder04:47
*** salv-orlando has joined #openstack-cinder04:51
*** laughterwym has quit IRC04:53
*** bkumar has joined #openstack-cinder04:55
openstackgerritAngus Lees proposed openstack/os-brick: Trivial rootwrap -> privsep replacement
openstackgerritLisaLi proposed openstack/os-brick: Add sheepdog support
*** haomaiwang has quit IRC05:01
*** haomaiwang has joined #openstack-cinder05:01
openstackgerritLisaLi proposed openstack/cinder: DO NOT MERGE
*** Poornima has joined #openstack-cinder05:05
*** salv-orlando has quit IRC05:07
*** qeelee has joined #openstack-cinder05:09
*** daneyon_ has quit IRC05:17
*** daneyon has joined #openstack-cinder05:17
*** salv-orlando has joined #openstack-cinder05:22
openstackgerritjaveme proposed openstack/cinder: Remove RequestBodySizeLimiter from middleware
*** houming has quit IRC05:27
*** cknight has quit IRC05:27
*** salv-orlando has quit IRC05:28
*** dave-mccowan has quit IRC05:28
*** vivekd has quit IRC05:28
*** mtanino has quit IRC05:29
openstackgerritLisaLi proposed openstack/cinder: Handle OverQuota exception during volume transfer
openstackgerritjaveme proposed openstack/cinder: Remove RequestBodySizeLimiter from middleware
openstackgerritWang Bo proposed openstack/cinder: Fix python2,3 compatibility issues
*** laughterwym has joined #openstack-cinder05:35
*** markvoelker has joined #openstack-cinder05:36
*** sgotliv has joined #openstack-cinder05:36
*** rcernin has joined #openstack-cinder05:38
*** markvoelker has quit IRC05:40
*** baojg has joined #openstack-cinder05:45
*** bharathk has joined #openstack-cinder05:56
*** bkumar has quit IRC05:56
*** aorourke__ has quit IRC05:56
*** cheneydc has quit IRC05:57
*** cheneydc has joined #openstack-cinder05:57
*** qeelee has quit IRC05:58
*** haomaiwang has quit IRC06:01
*** haomaiwa_ has joined #openstack-cinder06:01
*** vivekd has joined #openstack-cinder06:02
*** mudassirlatif has quit IRC06:13
*** resker has joined #openstack-cinder06:13
openstackgerritKeiichi KII proposed openstack/os-brick: remove unneeded in_use flag checking for multipath
*** mudassirlatif has joined #openstack-cinder06:15
*** esker has quit IRC06:15
*** belmoreira has joined #openstack-cinder06:23
*** lpetrut has joined #openstack-cinder06:25
*** resker has quit IRC06:30
*** salv-orlando has joined #openstack-cinder06:34
*** cheneydc has quit IRC06:34
*** flip214_ is now known as flip21406:35
*** flip214 has joined #openstack-cinder06:36
*** salv-orlando has quit IRC06:37
openstackgerritDinesh Bhor proposed openstack/cinder: Fix 500 error if 'offset' is out of range
*** nkrinner has joined #openstack-cinder06:43
openstackgerritMatan Sabag proposed openstack/cinder: Changes in ScaleIO configurations options.
*** esker has joined #openstack-cinder06:53
*** EinstCra_ has joined #openstack-cinder06:57
*** yangyapeng has joined #openstack-cinder06:58
*** yangyap__ has quit IRC06:59
*** yangyapeng has quit IRC06:59
*** EinstCrazy has quit IRC06:59
*** yangyapeng has joined #openstack-cinder06:59
*** haomaiwa_ has quit IRC07:01
*** 20WAABHNQ has joined #openstack-cinder07:01
*** ankit_ag has joined #openstack-cinder07:18
*** sgotliv has quit IRC07:20
*** lpetrut has quit IRC07:21
*** yangyape_ has joined #openstack-cinder07:26
*** yangyapeng has quit IRC07:26
*** esker has quit IRC07:31
*** markvoelker has joined #openstack-cinder07:36
*** markvoelker has quit IRC07:40
*** yangyapeng has joined #openstack-cinder07:41
*** chlong_ has quit IRC07:41
*** winston-d_ has quit IRC07:42
*** yangyape_ has quit IRC07:42
*** yangyapeng has quit IRC07:46
*** e0ne has joined #openstack-cinder07:49
*** esker has joined #openstack-cinder07:50
*** EinstCra_ has quit IRC07:52
*** bardia has quit IRC07:56
*** abhishek_k has joined #openstack-cinder07:57
*** abhishek_k has quit IRC07:57
*** 20WAABHNQ has quit IRC08:01
*** haomaiwang has joined #openstack-cinder08:01
*** boris-42 has joined #openstack-cinder08:04
openstackgerritVivek Dhayaal proposed openstack/cinder: Removed potential races from volume update method
openstackgerritKazumasa Nomura proposed openstack/cinder: Refactor Hitachi Block Storage Driver (patch - 1/4)
openstackgerritKazumasa Nomura proposed openstack/cinder: Refactor Hitachi Block Storage Driver (patch - 2/4)
*** salv-orlando has joined #openstack-cinder08:22
openstackgerritKazumasa Nomura proposed openstack/cinder: Refactor Hitachi Block Storage Driver (patch - 3/4)
*** zhangjn has quit IRC08:23
openstackgerritKazumasa Nomura proposed openstack/cinder: Refactor Hitachi Block Storage Driver (patch - 4/4)
*** lpetrut has joined #openstack-cinder08:30
*** houming has joined #openstack-cinder08:31
*** esker has quit IRC08:31
*** salv-orlando has quit IRC08:31
*** zhangjn has joined #openstack-cinder08:32
*** teju has joined #openstack-cinder08:34
*** lpetrut has quit IRC08:35
*** ildikov has quit IRC08:35
mc_nairDuncanT: lemme know if you're around and have some time to discuss some quota stuff.  I'll be online for a bit longer08:36
openstackgerritLisaLi proposed openstack/cinder: Add update_host for backup in cinder-manager
*** laughterwym has quit IRC08:39
*** esker has joined #openstack-cinder08:41
*** annasort_ has joined #openstack-cinder08:45
*** e0ne_ has joined #openstack-cinder08:46
*** shyama_ has joined #openstack-cinder08:46
*** yhayashi has quit IRC08:47
*** e0ne has quit IRC08:48
*** annasort has quit IRC08:48
*** annasort_ is now known as annasort08:48
*** e0ne_ has quit IRC08:48
*** shyama has quit IRC08:50
*** shyama_ is now known as shyama08:50
*** arecknag has joined #openstack-cinder08:53
*** zhangjn has quit IRC09:00
*** haomaiwang has quit IRC09:01
*** haomaiwang has joined #openstack-cinder09:01
*** salv-orlando has joined #openstack-cinder09:03
*** haypo has joined #openstack-cinder09:08
haypohey, can a core reviewer please re-approved ? it was approved, but i had to fix a simple and obvious conflict on tests-py3.txt09:09
openstackgerrithuanan proposed openstack/cinder: Huawei: Check the QoS status before we use
*** zhangjn has joined #openstack-cinder09:12
openstackgerritMichal Dulko proposed openstack/cinder: Add SIGHUP handlers to reset RPC version pins
*** jordanP has joined #openstack-cinder09:20
*** esker has quit IRC09:30
*** esker has joined #openstack-cinder09:34
*** zhangjn has quit IRC09:35
*** markvoelker has joined #openstack-cinder09:37
openstackgerritMichal Dulko proposed openstack/cinder: Add SIGHUP handlers to reset RPC version pins
*** markvoelker has quit IRC09:42
*** zhangjn has joined #openstack-cinder09:44
*** baojg has quit IRC09:44
lixiaoy1dulek: Hi Michal09:45
*** baojg has joined #openstack-cinder09:45
*** zhangjn has quit IRC09:46
duleklixiaoy1: Hi!09:47
duleklixiaoy1: Can I have a quick question if you're here?09:47
lixiaoy1dulek: thank you for your support in scaling backup!09:47
lixiaoy1dulke: your help to review the codes09:47
duleklixiaoy1: Sure, I believe this is a very important feature in this release, so it was a pleasure. :)09:48
duleklixiaoy1: - do we need this line now?09:48
lixiaoy1dulek: no!09:48
lixiaoy1dulek: yesterday you said you were going to rework the upgrade patch about backup09:49
duleklixiaoy1: I was a little confused when rebasing my patch, it's cool we don't need it.09:49
duleklixiaoy1: Yes, I've already started. It turns out a little harder than I thought, but I think I'm making progress.09:50
duleklixiaoy1: The idea is to add a dummy backup RPC API version (1.3) that marks the beginning of scalable backups era.09:50
openstackgerritJordan Pittier proposed openstack/cinder: Fix race condition in RemoteFS create_snapshot_online
duleklixiaoy1: And then be able to do scheduling based on that.09:50
lixiaoy1dulek: I will review the code later09:51
lixiaoy1dulek:dulek:  Another questions about rpc_current_version, currently when service starts, it gets the minimum version.09:51
duleklixiaoy1: If we're running with RPC API pinned to something lower than 1.2 - we're scheduling the old way - to the node where the backup/host is.09:51
duleklixiaoy1: Yes, it gets it. What's the issue here?09:52
lixiaoy1dulek: what I concerned is: should we exclude the services which are not available?09:52
lixiaoy1dulek: for example, I have a volume service, and later I kill it and never use.09:52
duleklixiaoy1: Excellent question! I've got back and forth on that.09:53
lixiaoy1dulek: :)09:53
duleklixiaoy1: So my motivation to *not* exclude unavailable ones is that Cinder doesn't know if the service is gone for good or not.09:54
duleklixiaoy1: It may be just a small network issue that made 5-6 heartbeats to disappear, but service will be back in a moment.09:55
duleklixiaoy1: We don't know that.09:55
lixiaoy1dulek: yes, this is a tricky problem09:55
*** davechen has left #openstack-cinder09:55
*** zhangjn has joined #openstack-cinder09:55
*** zhangjn has quit IRC09:56
duleklixiaoy1: I prefer administrators to clean up their service list manually (we have "cinder-manage service remove" command) as they will know which services are really gone.09:56
lixiaoy1dulek: sorry I have to leave office now.09:57
duleklixiaoy1: The fact that you're asking about it means that we certainly need a release note explaining that. I'll make a note.09:57
lixiaoy1dulek: catch bus!09:57
duleklixiaoy1: Safe travel!09:57
lixiaoy1dulek: talk you later09:57
*** zhangjn has joined #openstack-cinder10:00
*** haomaiwang has quit IRC10:01
*** haomaiwang has joined #openstack-cinder10:01
*** ildikov has joined #openstack-cinder10:01
*** sgotliv has joined #openstack-cinder10:02
*** salv-orl_ has joined #openstack-cinder10:03
yuriy_n17geguileo: Hi! Very sorry for disturbing. Please review whenever you have free time. Thanks a lot for advance.10:05
*** salv-orlando has quit IRC10:06
*** zhangjn has quit IRC10:13
*** zhangjn has joined #openstack-cinder10:22
*** zhangjn has quit IRC10:23
*** houming has quit IRC10:24
*** ndipanov has joined #openstack-cinder10:27
*** zhangjn has joined #openstack-cinder10:27
*** sdague has joined #openstack-cinder10:28
*** zhangjn has quit IRC10:28
*** slunkad_ has joined #openstack-cinder10:29
*** tpsilva has joined #openstack-cinder10:31
*** zhangjn has joined #openstack-cinder10:34
*** slunkad_ has quit IRC10:36
*** ildikov has quit IRC10:43
*** ociuhandu has quit IRC10:47
*** salv-orl_ has quit IRC10:48
*** baojg has quit IRC10:58
*** baojg has joined #openstack-cinder10:58
*** haomaiwang has quit IRC11:01
*** haomaiwang has joined #openstack-cinder11:01
*** baojg has quit IRC11:03
*** haomaiwang has quit IRC11:08
yuriy_n17dulek: Hi! Very sorry for disturbing. Please review whenever you have free time. Thanks a lot for advance.11:14
dulekyuriy_n17: I'm not sure if it will be today, but I'll try to take a look.11:16
yuriy_n17dulek: Whenever you have free time. Thanks.11:16
*** vivekd has quit IRC11:21
*** smoriya_ has quit IRC11:22
*** vivekd has joined #openstack-cinder11:24
*** [1]Thelo has joined #openstack-cinder11:25
*** vivekd_ has joined #openstack-cinder11:26
*** Thelo has quit IRC11:28
*** [1]Thelo is now known as Thelo11:28
*** vivekd has quit IRC11:29
*** vivekd_ is now known as vivekd11:29
*** salv-orlando has joined #openstack-cinder11:29
*** JoseMello has joined #openstack-cinder11:30
*** esker has quit IRC11:31
*** Poornima has quit IRC11:31
*** vivekd has quit IRC11:35
*** bharathk has quit IRC11:36
*** markvoelker has joined #openstack-cinder11:38
*** markvoelker has quit IRC11:42
*** esker has joined #openstack-cinder11:43
*** sgotliv has quit IRC11:43
*** dave-mccowan has joined #openstack-cinder11:49
*** lixiaoy1 has quit IRC11:52
*** ociuhandu has joined #openstack-cinder12:03
*** chlong_ has joined #openstack-cinder12:11
*** cheneydc has joined #openstack-cinder12:17
*** teju has quit IRC12:19
*** laughterwym has joined #openstack-cinder12:22
*** Lee1092 has quit IRC12:22
*** cdelatte has joined #openstack-cinder12:22
*** cheneydc has quit IRC12:26
*** cheneydc has joined #openstack-cinder12:27
*** EinstCrazy has joined #openstack-cinder12:27
*** esker has quit IRC12:27
*** dims_ has joined #openstack-cinder12:29
*** raildo-afk is now known as raildo12:35
*** esker has joined #openstack-cinder12:36
*** raildo is now known as raildo-afk12:38
*** cheneydc has quit IRC12:39
*** raildo-afk is now known as raildo12:39
*** lpetrut has joined #openstack-cinder12:39
*** zhangjn has quit IRC12:46
*** lpetrut has quit IRC12:50
*** markvoelker has joined #openstack-cinder12:54
*** zhangjn has joined #openstack-cinder12:54
*** shyama has quit IRC12:55
*** shyama has joined #openstack-cinder12:56
*** markvoelker has quit IRC12:58
*** klkumar has joined #openstack-cinder13:01
*** markvoelker has joined #openstack-cinder13:04
*** krotscheck_afk is now known as krotscheck13:05
*** vivekd has joined #openstack-cinder13:07
*** mylu has joined #openstack-cinder13:08
*** ankit_ag has quit IRC13:09
*** mylu has quit IRC13:09
*** fthiagogv__ has quit IRC13:12
*** salv-orlando has quit IRC13:21
*** edmondsw has joined #openstack-cinder13:28
*** porrua has joined #openstack-cinder13:29
*** diablo_rojo has joined #openstack-cinder13:36
*** cbader has quit IRC13:40
*** jyoti-ranjan has joined #openstack-cinder13:42
*** salv-orlando has joined #openstack-cinder13:54
*** lpetrut has joined #openstack-cinder13:57
haypohey, here are two easy win for python 3, review please! & ;)13:58
hayposmcginnis: ^^ thanks for the +2, but why not reapproving ? the change was approved, i just had to rebase it in the meanwhile13:59
*** jyoti-ranjan has quit IRC13:59
*** akerr has joined #openstack-cinder13:59
smcginnishaypo: Well, even on rebases we _should_ be going for two +2's.14:00
smcginnisMaybe even especially on rebases.14:00
smcginnishaypo: But I think that one was pretty straightforward. If no one else bites before I get back to it I'll take a look again and reapprove.14:00
hayposmcginnis: ah? i didn't know. on oslo, if i see that a change was approved before, i reapprove it. maybe i should ask how other oslo dev handle this case14:01
*** diogogmt has quit IRC14:01
hayposmcginnis: CIs on Cinder are quite annoying :-(14:01
*** Julien-zte has joined #openstack-cinder14:01
smcginnishaypo: Hah! There's a true statement.14:01
duleksmcginnis: I have similar problem on - it was approved yesterday, but scalable backups got in before it and broke UT.14:01
smcginnisVery necessary and useful, but very annoying.14:01
hayposmcginnis: i had to wait for a specific CI for most of my pending changes, and each time the specific CI failed, multiple times...14:01
* dulek needs to switch from "works on master" to "works on master with every combination of things in zuul".14:02
*** fthiagogv has joined #openstack-cinder14:02
smcginnisdulek: ;) Added to my list.14:02
hayposmcginnis: for HGST, it looks like the CI is gone or doesn't run anymore!? my change the CI
hayposmcginnis: i sent an email to the owner14:02
*** diogogmt has joined #openstack-cinder14:02
*** cbader has joined #openstack-cinder14:02
hayposmcginnis: "Hah! There's a true statement." i don't expect any solution, it's just to complain :)14:03
smcginnisOK, good. Thanks for doing that. I think they've been having issues lately. That needs to get resolved.14:03
smcginnishaypo: Some times you just need to get it out, even if it doesn't change anything. ;)14:03
hayposmcginnis: hehe. i took the habit of working on many changes at the same time. it's rare that i have less than 15 pending changes. but for cinder, it's very excited, because in 5 changes, all unit tests will pass on py3!14:04
haypo(i checked again this morning, 0 failure, really)14:04
smcginnishaypo: I love it!14:05
smcginnishaypo: That's a great milestone to reach for us. Thanks for all your work on that.14:05
hayposmcginnis: "a great milestone to reach for us" yeah, for me too :) i ported glance, horizon & cinder in this cycle14:06
smcginnishaypo: Wow. You've really been busy!14:06
hayposmcginnis: only trove & swift are working hard to block my python 3 happyness :)14:06
smcginnishaypo: Really? I'm surprised by that.14:07
smcginnisI think we've been talking py3 at least since Atlanta.14:07
haypo(at least, i fixed the py3 gate and made it voting for swift)14:07
hayposmcginnis: ahah14:07
hayposmcginnis: come on14:07
*** baojg has joined #openstack-cinder14:07
*** rlrossit has joined #openstack-cinder14:08
hayposmcginnis: they really don't care. for swift, i'm bugging them like every week, i joiend some meetings on IRC, and at the end i joiend a physical meeting at tokyo :) it's just stuck14:08
smcginnishaypo: Hmm. Well, once everyone else has switched over maybe they'll see the need. That's unfortunate.14:09
smcginnisbbib - gotta grab some coffee to make it through all these tabs I've opened. ;)14:09
*** lpetrut has quit IRC14:12
*** gouthamr has joined #openstack-cinder14:13
*** jgregor has joined #openstack-cinder14:20
*** martyturner has joined #openstack-cinder14:20
*** zhangjn has quit IRC14:23
*** mriedem has joined #openstack-cinder14:23
mriedemcan anyone tell me which cinder 3rd party CIs run with multipath enabled?14:23
mriedemi thought there were a couple14:23
smcginnismriedem: I think Pure and at least some of the HPE ones.14:24
smcginnismriedem: I recall I started to update mine, but honestly can't remember if I ever finished it :)14:24
mriedemk, just checking logs to see if this warning shows up
mriedemi'll look at the purestorage ones14:25
smcginnisIt's too bad we don't have a central ELK for all CIs.14:25
*** jyoti-ranjan has joined #openstack-cinder14:25
mriedemyeah, logstash for 3rd party ci is something that would be very helpful14:26
*** esker has quit IRC14:26
*** links has quit IRC14:27
*** timcl has joined #openstack-cinder14:27
*** lprice has quit IRC14:29
mriedemHPE FC ci uses multipath14:29
mriedemoh snap
*** cknight has joined #openstack-cinder14:29
openstackLaunchpad bug 1547539 in os-brick "KeyError: 'multipath_id' when attaching a volume" [Undecided,New]14:30
mriedemhemna: ^14:30
smcginnismriedem: New one that you've uncovered with that additional logging? Or logging that's backing up a bug that was out there.14:31
mriedemlooks unrelated14:31
mriedemeasy fix14:32
mriedemthe logic is a bit wonky14:32
smcginnisJust about to say that. :)14:33
smcginnisThe easy part.14:33
*** zhangjn has joined #openstack-cinder14:33
mriedemi'll push up a fix14:33
smcginnismriedem: Awesome, thanks.14:33
scottdamriedem: Do you have an opinion on the issue of adding a new Cinder /v3 endpoint for using api-microversions?14:33
*** vivekd has quit IRC14:34
scottda(while I've got you here)14:34
smcginnisscottda: Surprised we haven't had more response on that ML thread.14:34
scottdabecause I was just going to talk to smcginnis about this.....14:34
scottdasmcginnis: yes14:34
mriedemrather than v2.1?14:34
*** baojg has quit IRC14:34
*** dustins has joined #openstack-cinder14:34
scottdaand I'm thinking I'll un-WIP the patch14:34
mriedemi guess i haven't read the thread14:34
*** dansmith is now known as superdan14:34
mriedemsdague: would probably be the best to answer14:34
scottdamriedem: yes. I don't see much difference if the endpoint changes, a new /2.1 vs a new /3.014:35
sdagueyeh, sorry, I was knee deep in this metadata bug yesterday. I can compose a response later today.14:35
scottdayes, sdague weighed in yesterday, cautioning that Nova had regrets about adding a new endpoint.14:35
scottdasdague: ok, thanks14:35
smcginnismriedem: If you're interested:
scottdaBut the momentum to getting this in is strong, and I think we're a bit loathe to push this out to Newton....14:36
scottdasmcginnis: That sound about right? I just spoke with DuncanT on this...14:36
smcginnisscottda: Yeah, I agree.14:36
scottdasmcginnis: Anyway, I'm going to un-WIP and get back to work on adding /v3 support to the cinderclient.14:37
*** diogogmt has quit IRC14:37
scottdaand maybe poke e0ne, dulek , etc to have a look at reviewing14:37
smcginnisGood plan.14:37
scottda(looks like e0ne is offline. Is he on holiday?)14:37
smcginnisNo official Ukraine holiday today that I know of.14:38
scottdaI think he lives in Moscow, no?14:38
smcginnisscottda: I didn't think so. I could be wrong though.14:39
scottdaor maybe that's just where he flies through...14:39
dulekscottda: Nope, it's Kharkov.14:39
duleksmcginnis, scottda: Every Friday is a little holiday. ;)14:39
smcginnisdulek: Hah! ;)14:39
dulekscottda: I can't promise, I'm swamped by getting scalable backups to be able to upgrade seamlessly.14:40
scottdaanyway, I get the gist of sdague comment in the grenade patch about not copying the entire api-paste.ini file for the changes, but does anyone have an example of a more elegant way? I can only think of a bunch of ugly sed/grep/awk stuff.....14:40
dulekscottda: But I need to do it one time.14:40
scottdadulek: no problem. There's 2 x +2 from Sean and Patrick and I think Duncan will follow as well...14:40
scottdasmcginnis: so it's mainly an issue of figuring out which cores plan on reviewing and getting them to do it. I tried to get some names at last week's meeting, but I don't know of anyone else besides Ivan, and diablo_rojo said she was interested.14:41
smcginnisscottda: Maybe you need to start offering cheesecake to get volunteers.14:42
smcginnisscottda: But really, once it's ready I think we need to ping folks here and make sure everyone knows.14:42
*** openstackgerrit_ has quit IRC14:43
*** zhangjn has quit IRC14:43
scottdaIt's friday, which means I make home-made pizza. I think the code name for microversions should be pepperoni pizza.14:43
scottdaso my offer is "pizza + beer"14:43
smcginnisHah, love it.14:43
scottdasmcginnis: It is ready now.14:43
smcginnisAll new features need to have food code names.14:44
smcginnisscottda: I'll try to load it up in a bit.14:44
scottdasmcginnis: dulek For help with testing you can try using this:
smcginnisscottda: Have you taken a look at the API-WG related specs?14:45
smcginnisscottda: It might be good to have your input on those.14:45
smcginnisscottda: Or at least a skim to make sure there's no glaring issues.14:46
*** jwcroppe has joined #openstack-cinder14:47
*** diogogmt has joined #openstack-cinder14:49
*** zhangjn has joined #openstack-cinder14:49
*** martyturner has quit IRC14:51
*** erlon has joined #openstack-cinder14:53
*** vivekd has joined #openstack-cinder14:53
scottdasmcginnis: will do14:55
*** cknight has quit IRC14:58
*** cknight has joined #openstack-cinder15:01
*** openstackgerrit_ has joined #openstack-cinder15:01
*** timcl1 has joined #openstack-cinder15:01
*** akerr has quit IRC15:03
*** timcl has quit IRC15:04
*** akerr has joined #openstack-cinder15:05
*** Julien-zte has quit IRC15:06
*** baumann has joined #openstack-cinder15:07
*** lprice has joined #openstack-cinder15:08
*** eharney has joined #openstack-cinder15:09
*** eharney has quit IRC15:10
*** eharney has joined #openstack-cinder15:10
*** al_indigo has quit IRC15:11
*** dave-mccowan has quit IRC15:11
*** vivekd has quit IRC15:12
*** nkrinner has quit IRC15:12
*** links has joined #openstack-cinder15:13
*** mtanino has joined #openstack-cinder15:13
*** dhellmann has quit IRC15:16
*** dhellmann has joined #openstack-cinder15:19
*** esker has joined #openstack-cinder15:20
*** ociuhandu has quit IRC15:22
*** zhangjn has quit IRC15:24
*** ociuhandu has joined #openstack-cinder15:24
openstackgerritMichal Dulko proposed openstack/cinder: Scalable backup service - Liberty compatibility
*** esker has quit IRC15:25
*** dave-mccowan has joined #openstack-cinder15:25
*** dims_ has quit IRC15:27
*** zhangjn has joined #openstack-cinder15:29
*** shyama has quit IRC15:29
*** jyoti-ranjan has quit IRC15:29
*** jyoti-ranjan has joined #openstack-cinder15:30
openstackgerritMichael Price proposed openstack/cinder: NetApp: E-Series remove snapshot limitations
openstackgerritMichael Price proposed openstack/cinder: NetApp: Add Consistency Group support for E-Series
*** alaski is now known as lascii15:33
*** martyturner has joined #openstack-cinder15:33
*** timcl1 has quit IRC15:33
*** aorourke_ has joined #openstack-cinder15:35
*** mvk has quit IRC15:36
*** zhangjn has quit IRC15:36
*** jungleboyj has joined #openstack-cinder15:38
*** zhangjn has joined #openstack-cinder15:41
*** sgotliv has joined #openstack-cinder15:44
*** sheel has joined #openstack-cinder15:46
*** timcl has joined #openstack-cinder15:46
*** lpetrut has joined #openstack-cinder15:48
jgriffithmc_nair: ping15:48
*** chris_morrell has joined #openstack-cinder15:48
*** fthiagogv has quit IRC15:49
*** jdurgin1 has joined #openstack-cinder15:49
*** fthiagogv has joined #openstack-cinder15:49
*** aorourke_ has quit IRC15:51
mc_nairjgriffith: hey hey15:51
*** chris_morrell has quit IRC15:52
jgriffithmc_nair: so what's up, you mentioned you had some thoughts on that review?15:52
*** annasort has quit IRC15:53
*** annasort has joined #openstack-cinder15:53
*** zhangjn has quit IRC15:54
*** xyang1 has joined #openstack-cinder15:55
openstackgerritAlex O'Rourke proposed openstack/cinder: WIP - 3PAR: Update replication to v2.1
*** fthiagogv has quit IRC15:56
*** lpetrut has quit IRC15:57
sdaguescottda: I just posted the long response on the ML15:58
mc_nairjgriffith: ah - forgot about that :) I'll just put my questions on that review - since it's not high priority or anything.  I was initially going to start yesterday which is why I pung but didn't end up getting to it anyway.15:58
scottdasdague: thanks15:58
*** fthiagogv has joined #openstack-cinder15:58
sdaguesorry I didn't sooner, just, you know, we discovered we broke the metadata server15:58
sdagueso I was a bit preoccupied15:58
*** zhangjn has joined #openstack-cinder15:58
jgriffithmc_nair: so all I was getting at with the comment was that if we want to do things like request verifications we should implement a generic way to do it.  As opposed to a bunch of one-off adds in the volume.api15:59
*** jyoti-ranjan has quit IRC15:59
*** jyoti_ranjan has joined #openstack-cinder15:59
scottdasdague: no  problem. While I have you here...15:59
scottdasdague: IN the grenade patch for upgrading api-past.ini
*** mriedem is now known as mriedem_meeting16:00
scottdaWhen you ask that the specific lines be added to the file, do you want to see some awk/sed stuff, or am I missing something?16:00
openstackgerritAlex O'Rourke proposed openstack/cinder: WIP - 3PAR: Update replication to v2.1
sdaguescottda: yeh16:01
sdagueI think iniset probably can be used16:01
scottdasdague: ok, thanks. Just wanted to check before I went down that path.16:01
sdaguealthough, I'm a little confused why it is needed. On the upgraded side the new code should work with the old config, because nothing is going to hit the new api yet16:02
mc_nairjgriffith: generic as in would look at the front of the key for a vol_type extra spec, decide if this is something we should be validating.  Then break off "min" and decide that should be a <=, and look at the next portion of the name "size" and grabs that prop from volume?16:03
*** coolsvap|away has quit IRC16:03
*** sgotliv has quit IRC16:03
*** salv-orlando has quit IRC16:06
*** klkumar has quit IRC16:06
sdaguescottda: what was actually failing without the new lines?16:07
scottdasdague:  The new lines were to add the v3 endpoints. I may be wrong that the are required...just a sec...16:08
jgriffithscottda: gimmie a shout when you're wrapped up with sdague; I think I misunderstood your plan with micro-versions16:08
*** crose has joined #openstack-cinder16:09
sdaguescottda: right, grenade is just testing that mitaka code works with liberty config16:09
sdaguegiven that these would be additive, I don't think there is any reason they'd be needed16:09
scottdasdague: So this is the change to api-paste.ini ..16:09
sdaguescottda: yep16:09
*** coolsvap|away has joined #openstack-cinder16:09
sdagueyeh, I don't think you need that to pass grenade16:10
scottdasdague: ok, thanks. Sorry about my grenade ignorance.16:10
sdagueunless something is really weird in how things are initialized on the cinder side16:10
scottdaOK, I'll change that dependency16:10
sdagueif you hit a real issue, let me know16:10
sdagueand we can work through it16:10
scottdasdague: thanks. I appreciate it.16:10
*** arecknag has quit IRC16:11
openstackgerritScott DAngelo proposed openstack/cinder: cinder-api-microversions code
openstackgerritWalter A. Boring IV (hemna) proposed openstack/os-brick: Fix setting the multipath_id
hemnamriedem_meeting, ^16:13
*** haypo has left #openstack-cinder16:13
*** arch-nemesis has joined #openstack-cinder16:13
mriedem_meetingsorry for not getting that up soon enough16:15
*** vivekd has joined #openstack-cinder16:15
mriedem_meetingi've had distractions this morning16:15
mriedem_meetingwould be good to have a unit test on that case since it slipped through16:15
*** jgregor has quit IRC16:15
openstackgerritKurt Martin proposed openstack/cinder: 3PAR get host by WWN now handles mixed cases
scottdajgriffith: Hi16:15
jgriffithscottda: hey16:16
hemnamriedem_meeting, np16:16
jgriffithscottda: so I think I missed something16:16
jgriffithscottda: I thought we talked about doing the versioning without introducing a V3 at all?16:16
scottdajgriffith: No, I think by the end of the conversation at the mid-cycle it was decided to add a /v3. I could be wrong here, I hadn't reviewed the video. But I think smcginnis might have just watched it...16:17
scottdajgriffith: And while we're on the subject, what's your opinion on new /v3 vs no-new-endpoint ?16:18
jgriffithscottda: ok, you're probably right.  But for some reason I thought we talked about just adding something like ' "id": "v2.1" ' etc16:18
jgriffithscottda: I'm not a fan of a V3 at all :(16:18
jgriffithscottda: I had hoped we could do this by just extending V216:18
jgriffithbut I don't want to torpedo all the work you've already done either16:19
scottdajgriffith: That would have made my life easier. AT this point, I've coded it both ways. I can see both sides of the issue...16:19
jgriffithscottda: :)16:19
scottdaand I'd like to see this get in one way or the other in Mitaka. I think many/most Cinder folks want that as well.16:19
*** lpetrut has joined #openstack-cinder16:20
jgriffithscottda: so I had envisioned/hoped and even thought the path we were on was saying like:  curl -g -i -X GET
jgriffithscottda: so I had envisioned/hoped and even thought the path we were on was saying like:  curl -g -i -X GET
jgriffithetc etc16:20
smcginnisjgriffith: I think we talked about that a bit at the midcycle.16:20
jgriffithscottda: so then old clients and such that aren't updated or don't care are still just using the default "v2.0" and everything still "just works"16:21
jgriffithsmcginnis: yeah, that's what I was remembering... but perhaps we diverged and I've forgotten :(16:21
scottdaYeah, but that means adding a new endpoint for each microversion change. Do we really want that?16:21
smcginnisjgriffith: IIRC, others haven't gone with the version in the URL, so a bit too late for us to.16:21
*** mriedem_meeting is now known as mriedem16:21
openstackgerritMerged openstack/cinder: Update quota when volume type renames
smcginnisjgriffith: The headers being the existing approach.16:21
scottdaand like smcginnis , the pattern I've chosen is similar to how Nova, Ironic, Manila have all done it...16:22
smcginnisjgriffith: So the big argument, at least in my mind, for /v3 is safety against unaware clients.16:22
scottdawith the HTTP header containing the info, i.e "OpenStack-Volume-microversion: 2.37"16:22
jgriffithscottda: I guess I thought that we could have a single endpoint and "cinder" could figure out the version number interally... maybe I'm crazy there16:22
smcginnisjgriffith: If a new, microversion aware, client tries to perform an action on an older non-microversion aware cinder.16:23
smcginnisjgriffith: In that case, the new call could have some special params.16:23
jgriffithsmcginnis: the call fails16:23
smcginnisjgriffith: That the old cinder doesn't know to look for.16:23
smcginnisjgriffith: So it does succeed.16:23
jgriffithsmcginnis: you have the same problem if you are mv aware and call an invalid version no?16:23
smcginnisjgriffith: But the client does actually get what they asked for.16:23
smcginnisjgriffith: No, in that case cinder knows about mv and can treat it correctly.16:24
jgriffithsmcginnis: ok... guess I don't get it :(16:24
smcginnisjgriffith: The issue is the older cinder's that don't know and therefore don't know to fail it.16:24
hemnalooks like others are seeing 503 errors talking to  :(16:24
jgriffithsmcginnis: scottda so what I had *thought* of was you actually have a "mv" number for each Cinder release16:24
smcginnisjgriffith: Liberty, as it is today, would just take the request and ignore extra params.16:24
jgriffithsmcginnis: scottda so everything that goes in M for example gets a "2.1" endpoint16:25
jgriffithsmcginnis: scottda everything in N gets a 2.2 endpoint and so on16:25
jgriffithmaybe that's a horrible idea16:25
*** mylu has joined #openstack-cinder16:25
jgriffithsmcginnis: scottda and when I say "everything" I mean everything that is an API change that needs a version change16:25
jgriffithsmcginnis: scottda so we tie things to Cinder release every 6 months16:26
scottdajgriffith: No, that's not the model other's use. Instead, the microversion is bumped for every patch that changes the api in a way that requires a new microversion. Because we never really know when the last patch will land before release.16:26
jgriffithscottda: well... are you sure the model others use is the best one for us to use :)16:26
scottdajgriffith: No, not sure of that at all. But there is a benefit to all of us doing it the same way.16:26
jgriffithscottda: so the model you describe is a bit complex IMO16:27
jgriffithscottda: but yeah, consistency might be a good thing16:27
jgriffithscottda: even if it's consistently complex/difficult :(16:28
jgriffithscottda: ok, I'll leave you to it then and won't make start a schism on this16:28
scottdaI can imagine that someday all/most services will have microversions. If we all do it the same, i.e. with an HTTP header requesting the version, it sure will be easier to client developers, users, etc.16:28
*** yuriy_n17 has quit IRC16:28
smcginnisjgriffith: This all just makes me really hesitant to land this in M.16:28
scottdajgriffith: While we're on the subject, do you care to review the patch?16:29
smcginnisIt seems like there is still a lot of concern, confusion, misconceptions, etc. around microversions.16:29
scottdajgriffith: As in, should we wait for you to review when it comes time to merge?16:29
jgriffithscottda: well, I did :)16:29
jgriffithscottda: I gave it a -1 for the reasons I outlined here16:29
jgriffithscottda: but I just cleared that, because it sounds like consistency is preferred16:30
scottdajgriffith: Sorry, needed to refresh my browser.16:30
jgriffithscottda: honestly the approach being taken is kinda confusing to me, so I'm not sure my review is that helpful.  I'm also pretty strongly opposed to V3 after the debacle with V216:31
jgriffithscottda: and like I said the other day... the "cp -R v2/ v3/" is just bad IMO16:31
scottdajgriffith: I've yanked out all those /v3/ unit tests16:31
jgriffithscottda: well, I don't mean just the unit tests though16:32
scottdajgriffith: And we went with your suggestion: new tests for /v3 (microversioned) fucntions would go in the new tests/unit/api/v3 dir16:32
jgriffithscottda: the V2 thing was kind of a learning experience in what not to do IMO16:32
smcginnisjgriffith: But given the header approach instead of the explicit /v2.1 /v2.2 approach - would you rather just extend /v216:32
jgriffithscottda: it created A LOT of confusion and A LOT of extra maintenance for no gain16:32
scottdaThere's not much new in /api/v3, just the that re-uses /v2 stuff16:32
jgriffithsmcginnis: yeah, extend FOR SURE16:32
jgriffithsmcginnis: although I have to admit I'm not completely clear on what the approach your taking is going to look like16:33
smcginnisjgriffith: So no concern that a new client with microversions will hit an old cinder and do Somethign Bad?16:33
jgriffithsmcginnis: scottda but it has to be better than a V316:33
jgriffithsmcginnis: well... that's the problem.  In my model it would hit it with an unsupported v number (or endpoint) and fail16:34
jgriffithno harm no foul16:34
jgriffithbut the important thing is that it would preserve backward compatability16:34
scottdajgriffith: That would happen if the request was for some new method that does not exist on the old server...16:34
*** timcl has quit IRC16:34
smcginnisjgriffith: Yeah, I do like the explicitivity (is that a word?) of that.16:34
jgriffithscottda: no... it's more than that16:34
smcginnisBut if we use headers, that's not an option.16:34
scottdasmcginnis: Right. Putting it in the HTTP header will be ignored by the old, unaware client16:35
jgriffithscottda: so if I have an old cinder version that only supports V2.0... and I try and send something that says "v2.1" in the endpoint, header or whatever it fails16:35
scottdajgriffith: That is true. With a new endpoint.16:35
*** belmoreira has quit IRC16:35
*** arch-nemesis has quit IRC16:35
smcginnisjgriffith: THe header will just be ignored I beleive.16:35
scottdajgriffith: So you are saying, add a new /v2.1 endpoint?16:35
jgriffithI guess I'm hung up still thinking my endpoint per-release approach is better :)16:36
scottdajgriffith: It may be better. But I think the villagers would be at our door with torches and pitchforks if we did that.16:36
jgriffithit just seems "easier" and cleaner... but I guess we need to either say that's "off the table" or not16:36
jgriffithso I don't waste your time on it16:36
jgriffithscottda: which villagers?  Cuz if you mean other OpenStack project villagers I'm not sure I care... I'm fine with blasting them with a fire hose16:37
*** shyama has joined #openstack-cinder16:37
jgriffithscottda: think of it from the deployer/consumer villagers view-point16:37
jgriffithrather clear/easy to understand, backward compatible and pretty well defined in terms of what version you get based on the Cinder release16:38
scottdajgriffith: Those are the villagers I meant. I'm not sure if they care, of if life is harder to use a new endpoint vs. changing the header to a new microversion. It might be the same.16:38
*** mudassirlatif has quit IRC16:38
scottdaBut some of the consumers of our API are other OS projects...16:38
jgriffithscottda: you know we don't support old clients and new cinders very well anyway.  Nor does Nova, Neutron etc16:39
smcginnisNew clients and old cinders are the issue though.16:39
jgriffithbut it's worse, because you get crappy things like "module not found" "can't do x on None object" etc etc16:39
jgriffithsmcginnis: and my argument there is that's easy to fix16:39
scottdajgriffith: Yes, but this is an attempt to support various old vs. new combinations.16:39
jgriffithsmcginnis: via the endpoint strategy16:40
jgriffithsmcginnis: scottda well... it's not even "easy to fix" it's just "not a problem"16:40
*** salv-orlando has joined #openstack-cinder16:40
scottdaSo there's an unfortunate timing issue here. IF we want this in Mitaka, we'll need to come to consensus pretty darn quickly...16:41
scottdaIf we don't care if it slips to Newton, we can have many more debates.16:41
jgriffithscottda: yeah, that's why I was kinda reluctant to start making noise here16:42
jgriffithscottda: would it be useful for me to write it up on ML and see what kinda feedback comes in?16:43
jgriffithscottda: or would you prefer I went back to replication and let you make progress :)16:43
smcginnisjgriffith: I would appreciate if you can respond to that thread.16:43
scottdajgriffith: I don't want to shut you down. But as you know from your experience, this can get pretty exhausting....16:43
jgriffithscottda: FWIW, I don't want to have "debates" I just want my solution :)16:44
jgriffithscottda: ok, I'll leave you to it then16:44
jgriffithscottda: sorry for the side-track16:44
scottdaI think OpenStack could benefit from a Benevolent  Dictator , the way Linux has Linus...16:45
jgriffithscottda: I nominate /me16:45
smcginnisWe have various dictators with varying levels of benevolence.16:45
scottdajgriffith: Fine with me. As long as it's someone.16:45
jgriffithscottda: :)16:45
jgriffithsmcginnis: LMAO16:45
scottdasmcginnis: That's the problem , democracy degenerating to a mob of dictators16:45
jgriffithscottda: smcginnis what's always funny to me about the title is there is NO benevolence when it comes to Linus16:46
guitarzanI was just going to ask that16:46
*** kvidvans has joined #openstack-cinder16:46
guitarzanjgriffith: can you be as benevolent as linus? :)16:46
scottdaOK, I'll spend the rest of my day cleaning up loose ends and enjoying some pre-spring Colorado weather....16:46
jgriffithscottda: smcginnis and frankly the linux kernel is NOTHING like the monster that OpenStack has become16:46
jgriffithscottda: no kidding huh!  What's up with these temps!16:47
*** arch-nemesis has joined #openstack-cinder16:47
jgriffithsmcginnis: 73 here yesterday16:47
smcginnisIt's even warm-ish in MN.16:47
scottdajgriffith: I think the linux kernel is cleaner because it has the Dictator. Whether Benevolent or not, it keeps things consistent and makes decisions more clear.16:47
*** mylu has quit IRC16:48
jgriffithscottda: well... we have smcginnis as the benevolent dictator in Cinder16:48
jgriffithwhich is all I'm really concerned about here16:48
openstackgerritMerged openstack/cinder: Fix last Python 3 issues in zonemanager
jgriffithpitch forks be damned, especially if they're aimed at smcginnis and not me ;)16:49
kvidvansHi Team, looking for a +1 workflow for
smcginnisI need to find a tall tower to lock myself in. Preferably stone so that can't be set on fire.16:50
jgriffithsmcginnis: LOL16:50
*** laughterwym has quit IRC16:51
*** martyturner has quit IRC16:52
scottdasmcginnis: So one final thing...I would need to update the client for /v3 and haven't done that yet. So I'd like a decision on all this pretty early next week.16:52
*** laughterwym has joined #openstack-cinder16:52
smcginnisscottda: Yeah, we need to decide quick.16:52
smcginnisscottda: I need to stew some more.16:52
openstackgerritWalter A. Boring IV (hemna) proposed openstack/os-brick: Fix setting the multipath_id
smcginnisscottda: At least some input trickling in on the ML post now.16:52
openstackgerritMerged openstack/cinder: Test middleware test_faults to Python 3
*** links has quit IRC16:53
*** laughterwym has quit IRC16:56
*** martyturner has joined #openstack-cinder16:57
*** rcernin has quit IRC16:58
*** e0ne has joined #openstack-cinder17:02
hemnapersonally, I don't like the idea of doing v317:02
hemnait's taken fordamn ever to get folks to use v217:02
hemnav3 simply won't be used for years, if the migration to v2 is any example17:02
*** stupidnic has left #openstack-cinder17:04
smcginnishemna: Very true.17:06
smcginnishemna: The protection from dumb clients is my only concern.17:06
hemnait basically means microversioned changes won't be used at all.  :(17:06
scottdaI agree as well. And we need Nova to be onboard to make some of the microversioned changes we'd like for things like multi-attach17:07
hemnaI think that's far worse in my opinion17:07
smcginnishemna: But with a lot of folks still using v1, but the time we get to real support we can have backported protection for k and L.17:07
scottdaAnd I don't care about dumb clients. IF you are dumb, you get to learn a lesson and hopefully become smarter.17:07
smcginnisWell, TBF, they have to make a change to use microversions in the first place.17:07
*** porrua has quit IRC17:07
hemnascottda, +117:07
smcginnisSo going to v2 with microversion or v3, it's a small difference.17:07
scottdasmcginnis: That's right. IF the client is choosing to use microversions they'd better be smart enough to deal with it. IF they are not, too bad.17:08
*** cknight has quit IRC17:08
*** mylu has joined #openstack-cinder17:08
*** esker has joined #openstack-cinder17:08
smcginnisscottda: I haven't looked. How much work would it be to backport a fix for microversion awareness just enough to reject calls if the header is present.17:09
smcginnisPretty trivial, right?17:09
*** sbezverk has joined #openstack-cinder17:10
scottdasmcginnis: I'm not positive, but it might be done easily with changes to api-paste.ini :17:10
*** baumann has quit IRC17:10
scottdaIf that stuff is enforced, there should be a way to have something like "latent_disallow_headers". But not too sure on that...17:10
smcginnisSo if it's not in latent_allow_headers it gets rejects?17:10
scottdaNo, I don't think so.17:11
scottdapatrickeast: pointed out that I should probably add the new stuff for the microverson headers, which I did...17:11
scottdaBut they are not allowed on an older server, and it does not reject the headers if you send it.17:11
sbezverkHello, anybody could help with LVM/iscsi related question?17:12
smcginnissbezverk: We can try. Shoot.17:12
scottdaThat is , they are not specifically allowed via that filter.17:12
sbezverkI need a bit of guidance in troubleshooting an issue I am seeing. I brought up iscsid with targetcli in a docker container and I would like to cinder to use if for volumes. cinder.conf has right configuration for  iscsi and lvm backend. When I create a volume I see all existing volume groups get locked up and a volume gets stuck in creating state..17:15
*** porrua has joined #openstack-cinder17:16
smcginnissbezverk: Anything in the c-vol log file?17:16
smcginnisSorry, late for something. If you can put any errors in I'm sure someone here can help.17:16
*** garthb has joined #openstack-cinder17:17
sbezverksmcginnis no much, I should probably enable debug, but check this out17:17
*** porrua has quit IRC17:21
hemnapatrickeast, pure's CI seems to be unhappy at the moment.  503's from17:21
*** leeantho has joined #openstack-cinder17:22
rhedlindBoth my CI tests have been failing for the last 14 hours. It hits an exception in os-brick complaining about an unsupported operand type. Still looking for clues as to what might be wrong. Not aware of any changes in our internal environment. Any ideas are welcome. Trying to figure out if any changes to os-brick went in about that time that would help me get to the bottom of this.
*** timcl has joined #openstack-cinder17:24
scottdasmcginnis: bswartz Ok, how crazy is this idea to solve the issue of new_microversion client taking to old, non-micro server (and not checking)...17:26
scottdaDon't allow adding new parameters to methods that currently exist. Isn't that the only scenario where there is a problem?17:26
scottdaInstead, add a new API call with the new parameter, and in the API code just make that a wrapper using the old method.17:26
scottdaNew client talking to old server would fail.17:26
bswartzscottda: that approach would take away most of the power of microversions to allow us to improve the API17:27
bswartzif you can't add to existing APIs you'll have the exact same problem you have without microversions17:27
SwansonAbhijith, if you only have some trouble you're doing a great job.17:27
scottdaOK, I said it might be crazy.17:28
openstackgerritSheel Rana proposed openstack/python-cinderclient: Extra 'u' in output of cinder cli commands
*** jdurgin1 has quit IRC17:28
*** esker has quit IRC17:29
scottdaBut am I correct that the only troublesome scenario is when we add a parameter to an existing API?17:29
scottdaI'm still not convinced that the caller would not have to deal with it anyway, i.e. try: new way else: old way17:30
bswartzscottda: any time we significantly redefine how an API works in such a way that an invocation of the new API could be mistaken for an invocation of the old API by a server that's not microversion aware17:30
*** rcernin has joined #openstack-cinder17:30
bswartzscottda: the thing confusing me here is -- why is simply changing /v2 to /v3 such a big deal?17:31
bswartzchanging that 1 character in the URL solves so many problems, yet there seems to be a lot of resistence to it17:31
scottdabswartz: See the ML thread for some reasons.17:32
bswartzI did read it -- and it wasn't evident17:32
bswartzI will read again17:32
jgriffithbswartz: well, there's a few reasons17:33
jgriffithbswartz: after living through V2 for over a year, that's one big argument why it should be avoided.  Nobody picked it up, most were unclear on why they would want it17:34
jgriffithbswartz: contributors were unclear where/why they should put things17:34
jgriffithbswartz: we ended up maintaining bug fixes in 2 places17:34
jgriffithbswartz: the code was almost an exact copy/paste with no real value/add/change17:34
*** shyama has quit IRC17:35
jgriffithbswartz: we still have some consumers using V117:35
bswartzthe last part seems like an implementation issue17:35
jgriffithbswartz: we had a slew of bugs when nova tried to move to V217:35
bswartzit should be possible for v1and v2 to invoke common code if the code is in fact identical17:35
jgriffithbswartz: well.. yeah, arguably much of it is impl issues17:35
*** dims has joined #openstack-cinder17:35
jgriffithbswartz: except the endpoint :)17:35
bswartzas far a clients continuing to use v1 -- that's their choice and I don't see why it's a problem17:35
jgriffithbswartz: maintenacne17:36
jgriffithmaintenance even17:36
jgriffithbug reports17:36
jgriffithetc etc17:36
*** chris_morrell has joined #openstack-cinder17:36
jgriffithbswartz: and when you add a feature in Cinder for Nova to consume and they're not on a version that supports it... well then your'e kinda screwed17:36
bswartzwell the goal of removing old compabitility layers is different from the goal of adding new versions to people who want to run the latest can do so without insanity17:36
jgriffithbswartz: ok17:37
bswartzAs nice as it would be to drop compatibility for old stuff, I've never advocated actually doing that17:37
jgriffithbswartz: we could discuss implementation insanity if you like ;)17:37
bswartzI advocate making the new version great, and keeping compatible with old versions through whatever tricks are necessary17:37
jgriffithbswartz: I'm certainly not saying that either.. never have17:37
jgriffithI don't think anybody is proposing that at all17:37
jgriffithbswartz: we're on the same page17:38
jgriffithbswartz: My point was that copy/paste thousands of lines of code is stoopid17:38
jgriffithbswartz: and unnecessary17:38
bswartzyes but that can be avoided IMO17:38
bswartzpython is magic after all17:38
jgriffithbswartz: well that's the crux of the discussion on going v2-->v3 really17:39
jgriffithbswartz: and yes, I mentioned that I think there are ways to do it via endpoints or internal endpoints branched off of V2.017:39
bswartzscottda: since you proposed a crazy idea, let me throw a crazy idea back at you17:39
jgriffithbut that's apparantly not the "community" way17:39
bswartzhow about we implement a /v3 endpoint, that interally is simply aliased to the v2 endpoint17:40
bswartzthe only difference between the 2 would be that if a client came in through the /v2 endpoint we would note that that client does not support microversions17:40
bswartzand if they came in through the v3 endpoint, we would require the microversion header and track the version17:41
jgriffithbswartz: seems like a nice compromise17:41
bswartzIIRC that's what manila does on the server side regarding v1 vs v217:41
scottdabswartz: I think the aliasing is done via the /v3/ :
scottdaSo that's kinda done, unless I'm missing something.17:42
bswartzthat brings me back to confusion about why adding v3 is considered a big deal17:43
bswartzits 1 character in a URL string and it's not a log of code to handle it17:43
scottdaIt's not really a big deal to me17:43
bswartznot a lot*'17:43
scottdaIt's implemented and ready to go. Ship it.17:43
patrickeasthemna: thanks for the heads up, just got in and saw the ci scoreboard... something must have broken in a new and exciting way since nagios has been silent :(17:45
*** jyoti_ranjan has quit IRC17:45
*** jyoti_ranjan has joined #openstack-cinder17:45
*** baumann has joined #openstack-cinder17:45
patrickeastscottda: fwiw those header changes in the api-paste were only for cors stuff, they shouldn't affect normal traffic to the api service17:45
scottdapatrickeast: OK , thanks.17:46
patrickeastscottda: would need different settings to block requests17:46
*** shyama has joined #openstack-cinder17:46
patrickeastscottda: also, reading scrollback, whats the tl;dr for micro-drama today?17:46
scottdaWell, backporting something like that would only help in a very limited number of cases (i.e. actually deploying the backport), so I don't think that gets us very far.17:47
scottdapatrickeast: ARe you up on the micro-drama from yesterday?17:47
scottdai.e. should we be adding a /v3 endpoint?17:47
scottdaToday is not much different17:47
patrickeastoh ok17:47
patrickeastyea i was following some of it and the ML thread17:48
patrickeastscottda: sounds like we are kind of going in circles a little...17:48
scottdaahhh..that's why I'm so dizzy17:48
sheelhi all17:49
sheelare we still supporting cinder client v1 api?17:49
openstackgerritYucong Feng proposed openstack/cinder: Mark oslo.vmware as optional dependency
scottdaHere's an idea, since you are a newly-minted core patrickeast . Just +2 A that sucker right now and be done with it. What could go wrong?17:50
*** mudassirlatif has joined #openstack-cinder17:50
*** haomaiwang has joined #openstack-cinder17:52
patrickeastscottda: looks like it just failed pep8 timing out cloning : /17:52
cfoutsscottda: the commit message there might be of some use to you. :)17:53
sheelsmcginnis: hi17:54
sheelsmcginnis: just for confirmation if we still support cinderclient V117:54
*** jyoti_ranjan has quit IRC17:54
*** jordanP has quit IRC17:56
sheelsmcginnis: some bug is reported for v1 "cinder --os-volume-api-version 1 list --limit 0", so confirming whether we should fix or can let it go.17:56
*** annasort has quit IRC17:56
*** annasort has joined #openstack-cinder17:56
*** haomaiwang has quit IRC17:56
sheeljgriffith: Hi, any input on ^^ :)17:57
jgriffithsheel: my guess would be if it's a bug and you want to fix it then do so17:58
*** jungleboyj has quit IRC17:59
sheeljgriffith: actually this is fixed in v2 but v1 is still having it. so wanted to confirm if we still have v1 support17:59
scottdacfouts: Thanks!18:00
scottdapatrickeast: That looks really weird to me.18:00
patrickeastscottda: the pep8 thing?18:01
sheeljgriffith: I am confirming on early hand because getting things reviewed takes lot of efforts than fixing :) and if this is rejected later in review because we do not care for V1, it hurts :(18:01
scottdapatrickeast: yes. Any idea what that is, I was just going to recheck18:01
jgriffithsheel: I hear ya.  If the change fixes something and doesn't break compat I'd +2/A it.  That's all I can tell ya18:01
patrickeastscottda: oh, yea i posted a recheck already, i've seen it a few times over the last few days18:01
patrickeastscottda: i think its just the new infra mirrors timing out18:02
scottdapatrickeast: cool, thanks.18:02
*** martyturner has quit IRC18:02
sheeljgriffith: hmm... that make sense.... its enough..18:03
sheeljgriffith: :)18:03
*** rlrossit has quit IRC18:03
*** markusz has quit IRC18:08
*** IlyaG has joined #openstack-cinder18:09
*** Suyash has joined #openstack-cinder18:12
*** vivekd has quit IRC18:16
*** chlong_ has quit IRC18:18
*** mylu has quit IRC18:20
*** dims_ has joined #openstack-cinder18:22
*** dims has quit IRC18:25
*** ccesario has quit IRC18:25
*** tonyb has quit IRC18:25
*** netapp has quit IRC18:25
*** krotscheck has quit IRC18:25
*** ByPasS has quit IRC18:25
*** mikal has quit IRC18:25
*** dobson has quit IRC18:25
*** mkoderer__ has quit IRC18:25
*** ByPasS has joined #openstack-cinder18:25
*** mikal has joined #openstack-cinder18:25
*** krotscheck has joined #openstack-cinder18:25
*** netapp has joined #openstack-cinder18:26
*** mkoderer__ has joined #openstack-cinder18:26
*** dobson has joined #openstack-cinder18:26
*** anish has quit IRC18:27
*** rhedlind has quit IRC18:27
*** hparekh has quit IRC18:27
*** hemna has quit IRC18:27
*** abhi has quit IRC18:27
*** lifeless has quit IRC18:27
*** gcb has quit IRC18:27
*** anish has joined #openstack-cinder18:27
*** hparekh has joined #openstack-cinder18:27
*** lifeless has joined #openstack-cinder18:27
*** david-lyle has joined #openstack-cinder18:27
*** gcb has joined #openstack-cinder18:27
*** hemna has joined #openstack-cinder18:27
*** ccesario has joined #openstack-cinder18:28
*** abhi has joined #openstack-cinder18:28
*** rhedlind has joined #openstack-cinder18:28
*** tonyb has joined #openstack-cinder18:31
*** diablo_rojo has quit IRC18:34
*** diablo_rojo has joined #openstack-cinder18:35
*** diablo_rojo has quit IRC18:35
*** apoorvad has joined #openstack-cinder18:36
*** lpetrut has quit IRC18:36
*** mriedem has quit IRC18:39
*** cknight has joined #openstack-cinder18:40
*** IlyaG has quit IRC18:41
*** mriedem has joined #openstack-cinder18:41
*** crose has quit IRC18:41
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Updated from global requirements
*** mudassirlatif has quit IRC18:41
*** IlyaG has joined #openstack-cinder18:42
*** chris_morrell has quit IRC18:45
smcginnissheel: Fixes for v1 were actively being blocked for some time because the plan was to get rid of it.18:47
smcginnissheel: That's no longer the case. So I would welcome any v1 fixes. :)18:47
sheelsmcginnis: oh thanks..18:48
smcginnissheel: Thanks for looking at it.18:48
sheelsmcginnis: :)18:48
*** haomaiwang has joined #openstack-cinder18:53
*** rlrossit has joined #openstack-cinder18:54
openstackgerritOpenStack Proposal Bot proposed openstack/os-brick: Updated from global requirements
*** EinstCrazy has quit IRC18:56
*** diablo_rojo has joined #openstack-cinder18:56
*** fthiagogv has quit IRC18:56
*** rcernin has quit IRC18:56
*** haomaiwang has quit IRC18:58
*** vkmc has joined #openstack-cinder18:59
*** vkmc has left #openstack-cinder19:01
*** bardia has joined #openstack-cinder19:05
*** sdague has quit IRC19:05
*** mylu has joined #openstack-cinder19:08
*** diablo_rojo has quit IRC19:08
*** shyama has quit IRC19:10
*** martyturner has joined #openstack-cinder19:10
openstackgerritMitsuhiro Tanino proposed openstack/cinder: [LVM] Restore target config during ensure_export
*** esker has joined #openstack-cinder19:15
*** erlon has quit IRC19:16
*** ildikov has joined #openstack-cinder19:18
*** e0ne has quit IRC19:18
*** porrua has joined #openstack-cinder19:20
*** e0ne has joined #openstack-cinder19:23
*** rcernin has joined #openstack-cinder19:23
*** diogogmt has quit IRC19:24
*** IlyaG has quit IRC19:25
*** e0ne has quit IRC19:25
*** diogogmt has joined #openstack-cinder19:25
*** sdague has joined #openstack-cinder19:33
*** laughterwym has joined #openstack-cinder19:33
SwansonThese cafeteria sandwiches are not good.  I didn't know chicken had hooves.19:33
*** ildikov has quit IRC19:35
*** mylu has quit IRC19:36
*** laughterwym has quit IRC19:38
*** diogogmt has quit IRC19:40
*** ociuhandu has quit IRC19:41
*** mylu has joined #openstack-cinder19:43
*** rcernin has quit IRC19:45
*** diogogmt has joined #openstack-cinder19:46
*** mylu has quit IRC19:48
*** kfarr has joined #openstack-cinder19:49
*** ChubYann has joined #openstack-cinder19:52
*** lpetrut has joined #openstack-cinder19:55
*** ChubYann has quit IRC19:58
*** ChubYann_ has joined #openstack-cinder19:58
*** baumann has quit IRC20:02
*** mudassirlatif has joined #openstack-cinder20:06
*** eharney has quit IRC20:18
*** baumann has joined #openstack-cinder20:21
*** timcl has quit IRC20:24
openstackgerritSean McGinnis proposed openstack/cinder: Prevent unsupported microversion API requests
smcginnisscottda: Interested in your thoughts. ^^20:24
smcginnisjgriffith: You too.20:24
smcginnisbswartz: Curious on what you have to say on that too. ^^20:25
smcginnisbswartz: Though I may regret asking. ;)20:25
*** porrua has quit IRC20:28
*** akerr has quit IRC20:30
*** dims_ has quit IRC20:31
*** mudassirlatif has quit IRC20:31
openstackgerritJohn Griffith proposed openstack/cinder: Replication v2.1 (Cheesecake)
openstackgerritJohn Griffith proposed openstack/python-cinderclient: Add replication v2.1 (cheesecake) calls
patrickeastsmcginnis: so would that go along with adding the v3 endpoint? or instead?20:35
*** diablo_rojo has joined #openstack-cinder20:35
smcginnispatrickeast: Instead. This would give at least some level of safety to be able to stay on /v2.20:36
smcginnispatrickeast: If we backport this now to Liberty and Kilo, theoretically by the time folks actually start using microversion...20:36
smcginnisand by the time we make an incompatible API change...20:36
jgriffithsmcginnis: I'm not sure I see why this wouldn't be good20:36
jgriffithsmcginnis: just don't know if I'm missing something else in the impl that impacts it20:37
smcginnisthen backports will have been released and folks will have updated their installs.20:37
patrickeastyea, the only part i'm not sure about it the whole waiting for folks to update their deployments20:37
patrickeastwhereas the v3 endpoint means anyone using a newer client with microversions just can't do anything without a v3 enabled deployment20:38
smcginnispatrickeast: I raised that on the ML discussion. Having a fix available is different than being protected.20:38
patrickeastbut it definitely wouldn't hurt anything to put it in20:38
patrickeastother than maybe false sense of having it 'fixed'20:38
smcginnispatrickeast: But folks seem to be hung up on /v3, so this is a bit of a compromise.20:38
*** IlyaG has joined #openstack-cinder20:39
cfoutspatrickeast: a newer client should be able to fall back to using v2 for deployments not using v3, right?20:40
patrickeastcfouts: yea, and as a newer microversion aware client they should know that microversions don't work on v220:41
*** haomaiwa_ has joined #openstack-cinder20:41
patrickeastit seems like (granted I'm probably missing something) that would be the safest approach to this20:41
scottdasmcginnis: That seems like added protection. But it won't help existing deploys, of course.20:42
smcginnisscottda: Yep20:42
patrickeastjgriffith: oo nice, so the api's for failover are ready to go on the service controller?20:42
smcginnisscottda: So if we do go the v3 route...20:42
scottdacfouts: I think people are concerned about poorly implemented clients, i.e. not using cinderclient.20:42
jgriffithpatrickeast: yup... and yes, after looking at it putting everything under services just seemed to make sense20:43
smcginnisscottda: Eventually we could switch directions and have the v2 ones routed to v3 as "microversion 2.0", right?20:43
patrickeastjgriffith: sweet, yea i agree with putting it there20:43
patrickeastjgriffith: i'll update my environment and play with it this afternoon20:43
jgriffithpatrickeast: i think there's good room for improvements in a lot of the code, but I have no desire to polish the rivet from now until next release20:43
jgriffithpatrickeast: cool, lemme know if you run into issues with it20:44
* smcginnis gets out the spellchecker20:44
jgriffithsmcginnis: LOL... careful, I ignored the type in your commit message :)20:44
smcginnisjgriffith: Doh!!20:44
*** eharney has joined #openstack-cinder20:44
scottdasmcginnis: Yes, I think so. I'd have to play around with that. In fact, a microversion is always required now and if you hit the /v1 or /v2 endpoint my code sets the microversion to 1.0 and 2.0 today.20:44
*** IlyaG has quit IRC20:45
smcginnisscottda: Cool. So we theorectically wouldn't have a maintainability issue like we have today between v1 and v2.20:45
smcginnisNo copy/pasted duplicate code to maintain.20:45
*** IlyaG has joined #openstack-cinder20:45
scottdaI think that's correct, yes.20:45
smcginnisSo really the only drawback on /v3 I'm really hearing is that it's taken forever for v1 clients to migrate to v2.20:46
*** haomaiwa_ has quit IRC20:46
smcginnisBut their going to have to make a change anyway to go to microversion usage, whether that is /v2 or /v320:46
jgriffithpatrickeast: crap!  I still need to fix the status_update thing I think20:46
scottdasmcginnis: Well I don't see much difference between migrating to a new /v3 vs. using a microversion header.20:46
smcginnisBut if they don't care, they can still keep on using /v2 and it won't make any difference to them.20:46
scottdasmcginnis: yup20:46
smcginnisscottda: Yeah, still noodling.20:47
scottdasmcginnis: It's really only for people that want to move on. If they don't want to have new API changes, that's fine.20:47
smcginnisscottda: If we go the v3 route I'll abandon that patch.20:47
smcginnisIf we do go the v2 route I'll get it proposed for backport ASAP and get new releases queued.20:47
scottdaI eagerly await your decision, Great Leader.20:48
smcginnisscottda: It is slightly simpler to maintain on our side sticking to /v2?20:48
scottdasmcginnis: Yes, I think that is true.20:49
scottdaBut If we can do the re-routing that you mentioned , it might not be much difference.20:49
scottdasmcginnis: Well, right now all /v3 stuff is routed to /v2. If we add something new, it would be under /v3/ but never exist in /v2 anyway. So I don't know if there would ever be code duplication.20:51
scottdasmcginnis: IF we need to fix something existing, it goes under /v2 (and gets routed to /v2 from /v3). For something new, it's in /v3 anyway.20:52
scottdaThis is how Manila did it. So perhaps bswartz xyang1 cfouts  could confirm that I have this right.20:53
*** apoorvad has quit IRC20:53
smcginnisscottda: I'm just wondering if there will be any developer confusion if they need to fix something and have to figure out which directory it's actually in.20:53
smcginnisEspecially for new folks.20:53
scottdasmcginnis: Well, at least it's only in 1 directory.20:54
scottdaI cannot imagine any new person looking at the cinder code and being confused.20:55
openstackgerritMitsuhiro Tanino proposed openstack/cinder: Permit volume type operations for policy authorized users
mtaninojgriffith: Hi, Could you revisit the LIO fix? Seems Oleg is waiting that the problem is fixed immediately.21:00
*** alyson_ has quit IRC21:02
xyang1scottda: you are right21:03
jgriffithmtanino: no problem, updated with my +2 again21:03
xyang1scottda: for common code, there is a MixIn class under v2 (v1 in manila)21:04
mtaninojgriffith: Thank you for handling quickly :)21:04
scottdaxyang1: cool, thanks21:04
xyang1scottda: there is version handling in views as well if we need to add new fields21:06
scottdaI do like how Manila has solved all the problems for us.21:09
*** e0ne has joined #openstack-cinder21:16
*** apoorvad has joined #openstack-cinder21:17
*** e0ne has quit IRC21:21
openstackgerritJohn Griffith proposed openstack/cinder: Remove useless get_replication_updates driver call
jgriffithpatrickeast: there ^^21:23
jgriffithpatrickeast: next up... abc's21:25
patrickeastjgriffith: yay!21:25
smcginnisjgriffith: I've been thinking about abc's lately.21:25
jgriffithbut that one may have to wait til Monday21:25
*** [1]Thelo has joined #openstack-cinder21:25
jgriffithsmcginnis: sorry for you21:25
smcginnisjgriffith: I've actually been thinking of not having abc's, so that's good. ;)21:26
smcginnisBut probably not something for a Friday afternoon.21:26
jgriffithsmcginnis: I have a compromise to propose21:26
jgriffithsmcginnis: I'll submit it next week and people can take it or leave it21:26
patrickeastsmcginnis: that might be the best time, grab a couple of beers and start reverting code21:26
jgriffithpatrickeast: :)21:26
* jgriffith think "shit.. they're on to me"21:26
smcginnisjgriffith: Well, I'll be interested to see what you have. Especially if it's along the lines of what I've been thinking of. ;)21:27
jgriffithsmcginnis: so it's a working version of what I proposed in the past21:27
jgriffithsmcginnis: skips all the class nonsense21:27
*** Thelo has quit IRC21:27
*** [1]Thelo is now known as Thelo21:28
jgriffithsmcginnis: uses a decorator only on the base methods that are required21:28
smcginnisDoes it quack like a duck?21:28
jgriffithsmcginnis: anything without is considered options21:28
smcginnisAh, I see.21:28
jgriffithsmcginnis: were you thinking abandon altogether?21:28
jgriffithsmcginnis: zope?21:28
smcginnisHah, nope zope. Thought about that for a bit, but not sure.21:29
jgriffithsmcginnis: hehe21:29
jgriffithsmcginnis: so FWIW my idea works like zope kinda... so what are you thinking?21:29
smcginnisJust learned something interesting things about how python handles isinstance() that could be useful.21:29
smcginnisMaybe I'll throw together something too.21:29
jgriffithsmcginnis: ahhhh21:29
smcginnisIt's magic.21:29
jgriffithsmcginnis: so isinstance magic on a sub-class might be interesting21:30
jgriffithOk, I'll eagerly await next week :)21:30
smcginnisI see two aspects to object heirarchy.21:30
smcginnis1) Enforce a contract for an interface that can be programmatically verified.21:30
*** garthb_ has joined #openstack-cinder21:30
smcginnis2) Reuse code so the implementation burden is less and we all use as much common well-tested code as possible.21:31
smcginnisI've mostly been looking at 1.21:31
smcginnisI know when I came in to this, just figuring out what methods I was supposed to implement and which ones I didn't need to was a big hurdle.21:31
jgriffithsmcginnis: yeah, 2 is awesome... but well sort of doesn't fit with our calls21:31
smcginnisMuch less what those methods were actually supposed to do.21:32
jgriffithsmcginnis: yeah, that's what I'd like to fix21:32
*** garthb has quit IRC21:33
*** jungleboyj has joined #openstack-cinder21:34
openstackgerritYucong Feng proposed openstack/cinder: Return all target_wwpns for FC storwize.
openstackgerritWalter A. Boring IV (hemna) proposed openstack/os-brick: Fix setting the multipath_id
*** edmondsw has quit IRC21:44
*** dave-mccowan has quit IRC21:44
*** timcl has joined #openstack-cinder21:45
*** baumann has quit IRC21:45
openstackgerritPatrick East proposed openstack/cinder: WIP: Update Pure replication to cheesecake
*** kvidvans has quit IRC21:46
*** tpsilva has quit IRC21:46
*** e0ne has joined #openstack-cinder21:47
*** timcl has quit IRC21:47
*** e0ne has quit IRC21:48
smcginnishemna: Is that os-brick change something we'll need to quick cut a 1.1.0 release for?21:50
hemnasmcginnis, yes21:51
smcginnishemna: Bleh21:51
hemnasmcginnis, it's causing FC tests to fail, we are going to update our CI to skip the failing tests until that patch lands and is released.21:51
smcginnishemna: So the combination of FC and multipath?21:51
smcginnisI guess I should probably get mine configured for multipathing some day. :]21:52
hemnawe need more CI tests runs against os-brick patches21:53
hemnathat would help catch these :(21:53
kmartinpatrickeast, not sure how Pures (patrickeast) are passing... he has multipath enabled in cinder and nova, he might be skipping them?21:53
openstackgerritJohn Griffith proposed openstack/cinder: Replication v2.1 (Cheesecake)
patrickeasthemna: yea it never broke mine21:54
patrickeastkmartin: ^21:54
jgriffithaorourke: ^^21:54
smcginnishemna: We should probably add instructions here to tell folks how to test os-brick patches:21:54
jgriffithsmcginnis: I'll pick your brain next week on release notes and add those in as a follow up21:54
aorourkejgriffith, looking through it right now :) downloaded the cinderclient patch as well21:54
jgriffithaorourke: the admin_actions calls are old cruft21:55
smcginnisjgriffith: Awesome. No worries, they can be added later.21:55
jgriffithaorourke: I was trying to keep parallel calls for IBM, but that's not really going to work out here21:55
smcginnisI believe we had commitment at the beginning of the cycle that the v1 stuff would be pulled out.21:56
jgriffithsmcginnis: yes, we did... so I'm counting on that21:56
hemnasmcginnis, +121:56
hemnasmcginnis, I also think it's time to start adding CI tests for connectors21:56
smcginnishemna: +121:57
*** dave-mccowan has joined #openstack-cinder21:57
hemnaand requiring them21:57
aorourkejgriffith, quick question from the few minutes of can i get the Replication Status field out of not-capable?21:58
smcginnishemna: We probably should.21:58
aorourkejgriffith, this all looks really good by the way21:58
*** adrianofr_ has quit IRC22:00
jgriffithaorourke: capability reporting (stats update)22:01
jgriffithaorourke: and thank you22:01
*** IlyaG has quit IRC22:03
*** salv-orl_ has joined #openstack-cinder22:03
jgriffithaorourke: oh... hehe, that may not be in there :)22:03
hemnascottda, so are we still sticking with v3 endpoint? :(22:03
hemnascottda, just looking at the review22:03
aorourkejgriffith, I was going to say...looking through it I cannot seem to find it22:03
jgriffithaorourke: wait one22:03
scottdahemna: That question is still up in the air. Waiting for some guidance from Great Leader.22:04
scottdahemna: I think it's looking like we will stick with /v3 and that you are OK to review.22:04
scottdahemna: You've seen this for some testing ideas?
hemnanot yet22:05
scottdaok. might make things easier22:05
*** salv-orlando has quit IRC22:06
*** ByPasS has quit IRC22:07
*** sheel has quit IRC22:07
*** martyt has joined #openstack-cinder22:09
smcginnisscottda: I'm leaning toward /v3, so yeah, definitely review it.22:10
*** annasort has quit IRC22:10
jgriffithWTF are extra_capabilities ?22:12
*** changbl has quit IRC22:12
jgriffithhemna: ok.. WTF are "special_specs"22:12
hemnaI'm just messing with you22:13
jgriffithhemna: not nice :)22:13
*** ByPasS has joined #openstack-cinder22:13
*** martyt has quit IRC22:17
*** gouthamr has quit IRC22:19
*** lascii is now known as alaski22:22
*** dustins has quit IRC22:25
*** IlyaG has joined #openstack-cinder22:26
jgriffithaorourke: there.. let's just use that for now22:28
jgriffithaorourke: we can get clever later22:28
*** garthb__ has joined #openstack-cinder22:29
*** haomaiwang has joined #openstack-cinder22:29
*** daneyon has quit IRC22:30
*** lprice has quit IRC22:30
*** mylu has joined #openstack-cinder22:31
*** garthb_ has quit IRC22:32
aorourkejgriffith, looks good to me22:32
* patrickeast might have missed something22:32
patrickeastjgriffith: why can't it just use the driver stats api?22:32
*** mylu has quit IRC22:32
*** mylu has joined #openstack-cinder22:32
openstackgerritYucong Feng proposed openstack/cinder: Return all target_wwpns for FC storwize.
*** haomaiwang has quit IRC22:35
*** dims has joined #openstack-cinder22:36
patrickeastjgriffith: fwiw i ask because for my driver right now I won't know if we have replication enabled until do_setup happens, which i think is later on from when that call happens22:36
*** mriedem has quit IRC22:37
*** JoseMello has quit IRC22:38
*** rlrossit has quit IRC22:39
*** changbl has joined #openstack-cinder22:40
jgriffithpatrickeast: I'm down with that22:41
jgriffithpatrickeast: just couldn't figure out where to shove that in :)22:41
*** annasort has joined #openstack-cinder22:41
openstackgerritWalter A. Boring IV (hemna) proposed openstack/os-brick: Fix setting the multipath_id
patrickeastjgriffith: cool, i think it will be easier that way too22:41
patrickeastjgriffith: less code for everyone22:42
jgriffithpatrickeast: throw up a gist or just add it to the patch in progress22:42
*** lpetrut has quit IRC22:45
patrickeastjgriffith: can do22:45
*** annasort has quit IRC22:45
*** IlyaG has quit IRC22:49
*** david-lyle has quit IRC22:50
*** david-lyle has joined #openstack-cinder22:51
*** ociuhandu has joined #openstack-cinder22:53
*** david-lyle has quit IRC22:54
*** david-lyle has joined #openstack-cinder22:54
*** kfarr has quit IRC22:55
*** sgotliv has joined #openstack-cinder22:56
patrickeastjgriffith: hokay, two options for init_host
*** martyt has joined #openstack-cinder23:03
patrickeastjgriffith: one of them uses that new driver api, the other doesn't need it, but lacks a way to get the distinction between disabled and not-capable23:03
patrickeastdunno how much we care about them23:03
jgriffithpatrickeast: cool.. lemme look23:04
jgriffithpatrickeast: so disabled/not-capable isn't important to me right now23:04
patrickeastjgriffith: ditto23:05
jgriffithpatrickeast: so without the driver method is good for me.23:05
patrickeastjgriffith: sounds good to me23:05
jgriffithpatrickeast: Oh... DERP23:05
*** mylu has quit IRC23:05
jgriffithpatrickeast: don't know why I was hung up on that23:06
jgriffithpatrickeast: I'm far too lazy to change the "not-capable" default at this point though :)23:06
jgriffithpatrickeast: but I think I should23:07
patrickeastjgriffith: maybe thats the default if stats doesn't contain replication_enabled key?23:07
jgriffithpatrickeast: yeah, that's how I have it now, but I can't think of why I wouldn't just use disabled for that23:07
*** martyt has quit IRC23:07
jgriffithpatrickeast: except that I thought it might be useful to know the difference between "supported" and just not configured, or configured incorrectly23:08
patrickeastjgriffith: yea, although for like the scheduler it won't care right?23:08
jgriffithpatrickeast: correct23:08
patrickeastjgriffith: we just look at enabled as a bool23:08
*** mylu has joined #openstack-cinder23:08
jgriffithpatrickeast: I'll roll with your suggestion and clean things out to just enabled/disabled I think23:09
patrickeastjgriffith: cool, works for me23:09
*** haomaiwang has joined #openstack-cinder23:09
jgriffithpatrickeast: thanks for that example23:09
jgriffithpatrickeast: haha.. my patch is using the wrong db field anyway23:10
patrickeastjgriffith: i was wondering if it was supposed to be the replication_status23:12
*** mudassirlatif has joined #openstack-cinder23:13
*** sgotliv has quit IRC23:13
jgriffithpatrickeast: yeah :)23:13
*** haomaiwang has quit IRC23:14
*** sbezverk has quit IRC23:16
*** sdague has quit IRC23:21
*** xyang1 has quit IRC23:22
*** ociuhandu has quit IRC23:23
*** mylu has quit IRC23:24
*** mudassirlatif has quit IRC23:25
mc_nairjgriffith: I have a throw-back of a quota question for you... was looking at - is there a particular reason you split up the usage into reserved/in_use/total?23:26
*** martyturner has quit IRC23:26
mc_nairjgriffith: as opposed to just adding to the usage on a reserve, and then subtracting if you rollback that reservation?23:27
jgriffithmc_nair: TBF that was like 4 years ago :)23:27
*** mylu has joined #openstack-cinder23:29
*** david-lyle_ has joined #openstack-cinder23:29
mc_nairFriday evening tradition, review arbitrary patch sets from 6 releases ago23:30
*** david-lyle has quit IRC23:30
*** mylu has quit IRC23:30
*** mylu has joined #openstack-cinder23:30
mc_nairjgriffith: I'm asking because I'm looking to enable a similar quota reserve system for the "allocated" portion of a limit. Right now I'm just hacking it by letting a Reservation point to a Usage or a Quota (where "allocated" lives), but "allocated" doesn't have a split for reserved/allocated.  I think it will work still though, unless there's some pitfall you were avoiding by doing that23:31
openstackgerritJohn Griffith proposed openstack/cinder: Replication v2.1 (Cheesecake)
jgriffithmc_nair: there was a reason that we had them seperated in the past23:32
jgriffithmc_nair: but I honestly don't remember what it was or if it's still valid23:33
openstackgerritMerged openstack/cinder: Split out NestedQuotas into a separate driver
mc_nairI'm taking that as "continue haphazardly forth" :)23:34
mc_nairjgriffith: if you happen to remember lemme know, otherwise I'm sure someone will shout it out on the review23:34
*** laughterwym has joined #openstack-cinder23:35
*** annasort has joined #openstack-cinder23:35
mc_nair - maybe that will help jog your memory23:35
mc_nairget you back in the mindset23:35
jgriffithmc_nair: LOL23:37
jgriffithmc_nair: lemme have a look23:38
smcginnisOh, I forgot about McAfee.23:38
jgriffithmc_nair: so at one point which actually used to have a concept of a "real" reserve23:38
jgriffithmc_nair: where you wouldn't create the resource, but "save" some allocation for it23:39
jgriffithmc_nair: I don't think it ever got finished or worked though23:39
*** mylu has quit IRC23:39
*** annasort has quit IRC23:39
*** laughterwym has quit IRC23:40
mc_nairjgriffith: ok - incomplete and broken, I already implemented that approach23:40
mc_nairanyway, I'm headed out.  I'll push on with this approach now, hopefully get together some better POC for the -1 stuff, and then you can tell me why it's a bad idea23:42
mc_nairjgriffith: thanks for the background on that23:43
openstackgerritPatrick East proposed openstack/cinder: WIP: Update Pure replication to cheesecake
*** diablo_rojo has quit IRC23:48
*** jwcroppe has quit IRC23:53
*** jwcroppe has joined #openstack-cinder23:54
*** jwcroppe has quit IRC23:58
*** martyt has joined #openstack-cinder23:59

Generated by 2.14.0 by Marius Gedminas - find it at!