Friday, 2015-11-20

*** EinstCrazy has joined #openstack-cinder00:01
*** akshai has quit IRC00:01
*** markvoelker has quit IRC00:01
*** xyang1 has quit IRC00:03
*** bapalm has quit IRC00:03
*** bapalm has joined #openstack-cinder00:04
*** ntpttr has joined #openstack-cinder00:09
*** mc_nair has quit IRC00:11
*** EinstCrazy has quit IRC00:12
*** zhangjn has quit IRC00:15
openstackgerritNate Potter proposed openstack/cinder: Show qos_specs_id based on policy  https://review.openstack.org/24165900:15
*** rs007 has joined #openstack-cinder00:19
*** ntpttr has quit IRC00:19
*** davechen1 has left #openstack-cinder00:20
*** edtubill has joined #openstack-cinder00:26
*** setmason_ has joined #openstack-cinder00:28
*** daneyon_ has joined #openstack-cinder00:29
*** setmason has quit IRC00:30
*** daneyon has quit IRC00:32
*** jamielennox|away is now known as jamielennox00:32
*** setmason_ has quit IRC00:33
*** edtubill has quit IRC00:36
openstackgerritEdwin Wang proposed openstack/cinder: FlashSystem reports error in _find_host_exhaustive()  https://review.openstack.org/24754600:38
*** Splitpeas has quit IRC00:52
*** salv-orl_ has joined #openstack-cinder00:55
*** salv-orlando has quit IRC00:58
*** apoorvad has quit IRC00:58
*** rs007 has quit IRC00:58
*** jaypipes has quit IRC00:59
*** EinstCrazy has joined #openstack-cinder01:01
*** leeantho has quit IRC01:03
*** zhangjn has joined #openstack-cinder01:05
*** houming has joined #openstack-cinder01:06
*** zhenguo has joined #openstack-cinder01:08
*** flip214 has quit IRC01:09
*** houming has quit IRC01:10
*** flip214 has joined #openstack-cinder01:11
*** flip214 has quit IRC01:11
*** flip214 has joined #openstack-cinder01:11
*** garthb has quit IRC01:16
*** bswartz has joined #openstack-cinder01:17
*** markvoelker has joined #openstack-cinder01:17
*** markvoelker has quit IRC01:22
*** setmason has joined #openstack-cinder01:30
*** wilson-liu has quit IRC01:31
*** wilson-liu has joined #openstack-cinder01:32
*** Lee1092 has joined #openstack-cinder01:32
*** chhavi has joined #openstack-cinder01:37
*** mriedem has joined #openstack-cinder01:41
*** markvoelker has joined #openstack-cinder01:50
openstackgerritLisaLi proposed openstack/cinder: Delete unused codes in rbd.retype  https://review.openstack.org/24446001:54
openstackgerritMerged openstack/os-brick: Updated from global requirements  https://review.openstack.org/24610201:56
*** apoorvad has joined #openstack-cinder01:58
openstackgerritZhang Ni proposed openstack/cinder-specs: Add local tgt to lvm When the instance is in the same host with LVM volume, it's better to use local dev ( eg. /dev/ubuntu50-vg2/volume-9531dcba-16c5-4f1f-8b61-f0043d0b6a15) of lvm volume for high performence. So this blueprint will add a local tgt for LV  https://review.openstack.org/24788002:03
*** apoorvad has quit IRC02:03
mriedemanyone know tomoki sekiyama?02:03
jgriffithmriedem: yes02:03
openstackgerritThang Pham proposed openstack/cinder: Update replication to use versionedobjects  https://review.openstack.org/23890602:04
jgriffithmriedem: just taking a poll? :)02:04
mriedemjgriffith: wondering if they are still working on https://review.openstack.org/#/c/196482/02:05
jgriffithhrmm02:05
mriedemdidn't know irc nick02:05
jgriffithHe's tsekiyama02:06
jgriffithI usually see him around in the mornings02:06
jgriffithbut I must admit I haven't caught up with him here in quite a while02:06
jgriffithmriedem: I'll keep an eye out for him02:07
jgriffithmriedem: else maybe we should reassign that work02:07
mriedemthat's fine, i just asked in the change02:07
mriedemi'm going through old reviews and looking for abandons02:07
jgriffithmriedem: good deal02:08
openstackgerritZhang Ni proposed openstack/cinder-specs: Add local tgt to lvm  https://review.openstack.org/24788002:10
*** bswartz has quit IRC02:11
*** jwcroppe has quit IRC02:12
*** _amrith_ is now known as amrith02:14
*** jerrygb has quit IRC02:14
*** apoorvad has joined #openstack-cinder02:16
*** Yogi1 has joined #openstack-cinder02:18
*** bswartz has joined #openstack-cinder02:19
*** jerrygb has joined #openstack-cinder02:24
*** apoorvad has quit IRC02:30
*** haomaiwang has joined #openstack-cinder02:32
*** dalgaaf has quit IRC02:32
lixiaoy1mriedem: ping02:35
lixiaoy1mriedem: As both Cinder and Nova need nova/volume/encryptors, I am adding this part to os-brick02:36
*** haomaiwang has quit IRC02:37
*** Yogi1 has quit IRC02:37
mriedemlixiaoy1: are you referring to https://review.openstack.org/#/c/196482/ ?02:37
lixiaoy1please have a look at https://review.openstack.org/#/c/247372/02:37
*** gouthamr has joined #openstack-cinder02:38
mriedemoh boy02:38
mriedemso that needs a blueprint i think02:38
mriedemjgriffith: ^02:38
mriedemsmcginnis: ^02:38
mriedemhemna: ^02:38
*** haomaiwang has joined #openstack-cinder02:38
smcginnismriedem: Agreed02:38
mriedemspecifically b/c before nova is going to consume that, nova has to integrate the privsep stuff02:38
mriedemfor rootwrap filters02:38
mriedemthat came up in the nova meeting today02:39
*** dims_ has quit IRC02:39
smcginnismriedem: Are you going to privsep already?02:39
lixiaoy1I asked Sean, and he thinks this is just moving to os-brick. no need bp02:39
smcginnisWell..02:39
smcginnisMoving where things are, but a bp might be good to make sure all users (cinder and nove) understand what's going on.02:40
mriedemsmcginnis: sdague said in the nova meeting today that nova will have a bp for tracking the privsep work that has apparently landed in oslo but hasn't been released yet02:40
mriedemwe have to get going on that earlier than later probably02:40
mriedemotherwise we're going to have to keep duplicate rootwrap filters02:41
mriedem*duplicating02:41
smcginnismriedem: Ah, cool. I've been taking a wait and see approach.02:41
mriedemmikal was going to follow up with gus on the bp work in nova02:41
smcginnismriedem: I'm a little nervous picking it up right away before anyone has worked it out.02:41
mriedemearly adopters :)02:41
smcginnismriedem: I'll let Nova do that. :D02:41
mriedemheh02:41
mriedemyou bastard!02:41
smcginnishehe02:42
lixiaoy1smcginnis: you mean let Nova maintain rootwrap filters02:42
lixiaoy1?02:42
smcginnislixiaoy1: No, apparently the plan is to get rid of rootwrap faster than I thought would happen.02:42
smcginnismriedem: You'd like lixiaoy1 to submit a bp on the nova side, right?02:43
lixiaoy1I have a bp in cinder02:43
smcginnislixiaoy1: Right, but nova changes will be required as well, right?02:44
lixiaoy1early, I mean no spec is wroten02:44
smcginnisIIRC, the plan is to have both nova and cinder use the encryptors from os-brick rather than copy them between both, right?02:44
lixiaoy1I will put encryptors in os-brick, and later nova will be changed to call02:44
lixiaoy1yes02:45
smcginnislixiaoy1: So I think getting them in os-brick - you are all set.02:45
mriedemnova will need a compainion bp yes02:45
smcginnislixiaoy1: But in order to update nova to use them, you will need a bp there for that piece of work.02:46
lixiaoy1this is bp in cinder: https://blueprints.launchpad.net/cinder/+spec/improve-encrypted-volume02:46
mriedembut i think the nova team should have input on the blueprint/spec for this, i.e. why it's needed and what the plan is02:46
smcginnismriedem: OK, cool.02:46
mriedemalso noting that os-brick has historically really been only the pieces of nova that have been in the libvirt driver in nova,02:46
mriedemthe volume encryptor stuff is not necessarily libvirt specific02:46
smcginnislixiaoy1: Can you submit one now, even if things aren't quite ready to make any changes there yet?02:46
lixiaoy1ok02:47
*** elonden has joined #openstack-cinder02:47
mriedemalthough i feel like there are assumptions made that they are used together (since no other virt driver supports encrypted volumes from what i can tell)02:47
smcginnismriedem: Yeah, this may be a slight expansion of scope.02:47
mriedemposting this to the -dev list would be good too as a heads up for nova people, this is the first i've heard of it02:47
mriedembbiab02:47
smcginnismriedem: It's kind of another rootwrap/copy the same stuff everywhere kind of situation.02:47
smcginnismriedem: Good idea re: -dev.02:48
smcginnislixiaoy1: Can you post something to the mailing list giving an overview of what is being doen.02:48
smcginnis?02:48
lixiaoy1ok, I will do it02:48
smcginnislixiaoy1: Put [Cinder] and [Nova] in the subject line so both teams pay attention to it.02:48
smcginnislixiaoy1: Thanks!02:48
lixiaoy1yes02:49
lixiaoy1welcome02:49
smcginnislixiaoy1: I think it makes sense, but mriedem is right, we should probably make sure everyone is aware and understands what's being proposed.02:49
lixiaoy1I will02:49
smcginnisAwesome, thank you.02:50
lixiaoy1my pleasure02:50
*** sghanekar_ has quit IRC02:53
*** chenying has quit IRC02:55
openstackgerritMitsuhiro Tanino proposed openstack/cinder: Robustify writing iscsi target persistence file  https://review.openstack.org/22794302:55
*** houming has joined #openstack-cinder02:56
openstackgerritMitsuhiro Tanino proposed openstack/cinder: Robustify writing iscsi target persistence file  https://review.openstack.org/24788602:59
*** gouthamr has quit IRC03:00
*** gouthamr has joined #openstack-cinder03:01
*** rushiagr_away is now known as rushiagr03:02
*** jdurgin1 has quit IRC03:03
openstackgerritMitsuhiro Tanino proposed openstack/cinder: [WIP] Robustify writing iscsi target persistence file  https://review.openstack.org/22794303:09
*** bardia has quit IRC03:12
smcginnishemna: Question about bug 1487169...03:15
openstackbug 1487169 in os-brick "multipath -l can be very slow " [Medium,In progress] https://launchpad.net/bugs/1487169 - Assigned to Walt Boring (walter-boring)03:15
smcginnishemna: You got a patch in that said it was a partial fix.03:15
smcginnishemna: Wondering what else we can do there?03:15
*** setmason_ has joined #openstack-cinder03:16
*** setmason has quit IRC03:19
*** setmason_ is now known as setmason03:19
*** edmondsw has quit IRC03:19
*** dims has joined #openstack-cinder03:21
*** amrith is now known as _amrith_03:22
*** gcb has quit IRC03:35
*** gcb has joined #openstack-cinder03:35
*** setmason has quit IRC03:44
*** chhavi has quit IRC03:44
*** setmason has joined #openstack-cinder03:45
openstackgerritThang Pham proposed openstack/cinder: Register RPC and object versions  https://review.openstack.org/20970103:50
*** thangp has joined #openstack-cinder03:51
*** thangp has quit IRC03:53
*** Thelo has quit IRC03:59
*** mriedem has quit IRC03:59
*** yusuke has left #openstack-cinder04:02
*** yusuke has joined #openstack-cinder04:03
*** Thelo has joined #openstack-cinder04:05
*** dave-mccowan has quit IRC04:07
*** swebb has quit IRC04:07
*** Thelo has quit IRC04:08
*** EmilienM has quit IRC04:08
*** EmilienM has joined #openstack-cinder04:10
*** Thelo has joined #openstack-cinder04:10
*** zhangjn has quit IRC04:14
*** swebb has joined #openstack-cinder04:14
*** Thelo has quit IRC04:16
*** Thelo has joined #openstack-cinder04:16
*** gouthamr has quit IRC04:22
*** mc_nair has joined #openstack-cinder04:23
*** asselin has quit IRC04:25
*** asselin has joined #openstack-cinder04:25
*** zhangjn has joined #openstack-cinder04:29
*** houming has quit IRC04:30
*** dims has quit IRC04:36
*** rushiagr is now known as rushiagr_away04:44
*** zhangjn has quit IRC04:45
*** zhonghua-lee has quit IRC04:56
*** zhonghua-lee has joined #openstack-cinder04:56
yusukeHi all, excuse me. Do you know alternative url or web site of following url? Although I know how to setup Cinder CI for FibreChannel Protocol, this link looks dead....05:10
yusukehttps://git.openstack.org/cgit/stackforge/third-party-ci-tools/tree/provisioning_scripts/fibre_channel05:10
yusuke(from https://wiki.openstack.org/wiki/Cinder/tested-3rdParty-drivers)05:10
*** zhangjn has joined #openstack-cinder05:11
*** zhangjn has quit IRC05:12
openstackgerritxing-yang proposed openstack/cinder: Add a GateDriver  https://review.openstack.org/24495605:13
*** houming has joined #openstack-cinder05:14
*** zhangjn has joined #openstack-cinder05:18
*** mc_nair has quit IRC05:21
*** houming has quit IRC05:22
yusukes/Although I know how to/Although I want to know how to/05:23
*** jerrygb has quit IRC05:29
*** rushiagr_away is now known as rushiagr05:32
openstackgerritMerged openstack/cinder: VNX: Fix metadata get overriden issue  https://review.openstack.org/24623805:32
*** stevemar_ has quit IRC05:34
*** stevemar_ has joined #openstack-cinder05:35
*** stevemar_ has quit IRC05:38
*** haomaiwang has quit IRC05:46
openstackgerritLisaLi proposed openstack/os-brick: Copy encryptors from Nova to os-brick  https://review.openstack.org/24737205:50
openstackgerritCedric Zhuang proposed openstack/cinder: VNX: Get volume type from location throws error  https://review.openstack.org/24623005:51
openstackgerritLisaLi proposed openstack/os-brick: Copy encryptors from Nova to os-brick  https://review.openstack.org/24737205:51
*** CORTEX\krfa has quit IRC05:57
*** haomaiwang has joined #openstack-cinder05:59
*** shausy has joined #openstack-cinder06:01
openstackgerritMerged openstack/cinder: Remove .mailmap file  https://review.openstack.org/24767606:02
yusukeI'm sorry. I found alternative one, thanks.06:07
*** jkraj has joined #openstack-cinder06:08
*** lpetrut has joined #openstack-cinder06:10
*** smoriya is now known as smoriya_afk06:11
*** haomaiwang has quit IRC06:14
openstackgerritPeter Wang proposed openstack/cinder: VNX: Fix failure in SnapCopy feature  https://review.openstack.org/24301706:15
*** haomaiwa_ has joined #openstack-cinder06:18
openstackgerritwanghao proposed openstack/cinder-specs: Improvement to query consistency group detail  https://review.openstack.org/24299806:21
*** rcernin has joined #openstack-cinder06:30
*** jerrygb has joined #openstack-cinder06:30
*** jerrygb has quit IRC06:36
openstackgerritPeter Wang proposed openstack/cinder: VNX: Fix failure in SnapCopy feature  https://review.openstack.org/24301706:38
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Imported Translations from Zanata  https://review.openstack.org/24794306:39
*** elonden has left #openstack-cinder06:39
*** haomaiwa_ has quit IRC06:43
*** rushiagr is now known as rushiagr_away06:52
*** haomaiwang has joined #openstack-cinder06:53
*** salv-orlando has joined #openstack-cinder06:54
*** salv-orl_ has quit IRC06:58
*** haomaiwang has quit IRC07:01
*** haomaiwang has joined #openstack-cinder07:02
openstackgerritzhangsong proposed openstack/cinder: Fix the bug of can't get the desired image info  https://review.openstack.org/24524107:06
*** bluex has joined #openstack-cinder07:07
openstackgerritzhangsong proposed openstack/cinder: Fix the bug of can't get the desired image info  https://review.openstack.org/24524107:13
*** nkrinner has joined #openstack-cinder07:23
*** haomaiwang has quit IRC07:25
*** haomaiwang has joined #openstack-cinder07:26
*** anshul has joined #openstack-cinder07:28
*** Dhaval has joined #openstack-cinder07:30
*** ankit_ag has joined #openstack-cinder07:37
*** chenying_ has quit IRC07:40
*** alexschm has joined #openstack-cinder07:40
*** anshul has quit IRC07:40
*** anshul has joined #openstack-cinder07:40
*** zhangjn has quit IRC07:47
*** vgridnev has joined #openstack-cinder07:52
*** haomaiwang has quit IRC07:59
*** haomaiwang has joined #openstack-cinder08:02
DhavalI have installed Openstack on single machine using devstack08:07
DhavalI am trying to develop sample third party driver for cinder for POC purpose08:07
DhavalBut not sure where should I start from08:08
DhavalCan anyone please guide me about procedure to register sample cinder driver with openstack08:09
*** egonzalez has joined #openstack-cinder08:11
dulekDhaval: Maybe this will be helpful: https://wiki.openstack.org/wiki/Cinder/how-to-contribute-a-driver08:15
Dhavaldulek : Thanks for sharing link08:18
*** zhangjn has joined #openstack-cinder08:18
DhavalI am actually trying to implement sample cinder driver which could add/delete volumes just like LVM driver08:21
DhavalI even modified cinder.conf file as follows and restarted cinder-api, cinder-scheduler and cinder-volume services08:22
Dhavalfollowing contents were added in cinder.conf file:08:22
Dhavalenabled_backends = lvmdriver-1,lvmdriver-208:22
Dhaval[lvmdriver-2] volume_group = stack-volumes-lvmdriver-2 volume_driver = cinder.volume.drivers.lvm.LVMVolumeDriver volume_backend_name = lvmdriver-2                                                                                                                               26,1          Bot08:23
DhavalEven after restarting the cinder services; I am unable to see lvmdriver-2 in cinder service-list output08:23
openstackgerritFalk Reimann proposed openstack/cinder: Don't add project_id to auth_url in SwiftBkpDriver  https://review.openstack.org/24796408:24
dulekDhaval: Take a look into the logs of c-vol. If driver was unable to be initialized, then c-vol won't report it's presence.08:24
Dhavalokay08:24
dulekDhaval: LVM is uninitialized when for example volume_group doesn't exist.08:24
dulekDhaval: You may also let DevStack configure everything for you.08:25
dulekDhaval: Just add this to local.conf:08:25
dulekDhaval: CINDER_ENABLED_BACKENDS=${CINDER_ENABLED_BACKENDS:-lvm:lvmdriver-1,lvm:lvmdriver-2}08:25
Dhavaldulek : Okay, let me try it out08:26
*** rushiagr_away is now known as rushiagr08:26
*** zhangjn has quit IRC08:32
*** lpetrut has quit IRC08:35
*** houming has joined #openstack-cinder08:35
*** zhangjn has joined #openstack-cinder08:36
*** vgridnev has quit IRC08:40
*** shausy has quit IRC08:41
*** shausy has joined #openstack-cinder08:42
*** gcb has quit IRC08:43
*** gcb has joined #openstack-cinder08:44
openstackgerritFalk Reimann proposed openstack/cinder: Adds Auth Version 3 support in Swift Backup Driver  https://review.openstack.org/24797708:51
*** zhangjn has quit IRC08:51
*** houming has quit IRC08:53
*** setmason has quit IRC08:53
*** markus_z has joined #openstack-cinder08:54
*** nkrinner has quit IRC08:58
*** nkrinner has joined #openstack-cinder08:58
*** haomaiwang has quit IRC09:01
*** smoriya_afk has quit IRC09:01
*** haomaiwang has joined #openstack-cinder09:01
*** houming has joined #openstack-cinder09:04
*** ndipanov has joined #openstack-cinder09:06
*** nkrinner has quit IRC09:10
*** houming has quit IRC09:12
*** nkrinner has joined #openstack-cinder09:14
*** egonzalez is now known as egonzalez9009:17
*** egonzalez90 is now known as Guest2418509:17
*** houming has joined #openstack-cinder09:18
*** Guest24185 has quit IRC09:18
*** egonzalez has joined #openstack-cinder09:18
smcginnisyusuke: Please feel free to edit that wiki page with the correct link. Thanks!09:18
*** chhavi has joined #openstack-cinder09:19
yusukeOK, Thanks!09:19
smcginnisyusuke: Thanks for pointing it out!09:20
*** haomaiwang has quit IRC09:25
*** haomaiwang has joined #openstack-cinder09:27
Dhavaldulek : Thanks a lot for helping out. I am now able to see both lvmdriver-1 and lvmdriver-2 in cinder service-list output09:27
Dhavaldulek : had to create volume group for successful initialization09:27
dulekDhaval: You're welcomed.09:28
*** vgridnev has joined #openstack-cinder09:29
*** markvoelker has quit IRC09:30
*** e0ne has joined #openstack-cinder09:31
*** zhangjn has joined #openstack-cinder09:31
*** sgotliv has joined #openstack-cinder09:34
*** haomaiwang has quit IRC09:37
*** wilson-liu has quit IRC09:38
*** wilson-liu has joined #openstack-cinder09:39
*** haomaiwang has joined #openstack-cinder09:42
*** zhangjn has quit IRC09:44
*** jistr has joined #openstack-cinder09:44
*** jordanP has joined #openstack-cinder09:49
*** EinstCrazy has quit IRC09:49
openstackgerritTobias Urdin proposed openstack/cinder: Correct small misses in debug output for cinder-volume-usage-audit. Fixes variables not being included in output. Changed according to comments.  https://review.openstack.org/24734409:53
flip214help, please. with the LVM driver the "os-vol-host-attr:host" volume attribute includes the host, service, and backend; with my own driver only the host is listed.09:53
flip214I believe this is the reason for initialize_connection() (etc.) failing.09:54
flip214what is my driver missing that the LVM driver does, so that the "os-vol-host-attr:host" attribute gets filled in correctly?09:54
flip214The "class Volume_host_attribute" in cinder/api/contrib/volume_host_attribute.py should be used for all backends, as I understand it.09:55
*** zhenguo has quit IRC10:01
*** lpetrut has joined #openstack-cinder10:01
openstackgerritSzymon Wr√≥blewski proposed openstack/cinder: Tooz locks  https://review.openstack.org/18353710:04
*** aix has joined #openstack-cinder10:07
*** haomaiwang has quit IRC10:10
openstackgerritMichal Dulko proposed openstack/cinder: Add metadata aliases to Volume object  https://review.openstack.org/24776310:12
openstackgerritMichal Dulko proposed openstack/cinder: Recognize extra fields in CinderObjectDictCompat  https://review.openstack.org/24776210:12
*** lpetrut1 has joined #openstack-cinder10:13
*** lpetrut has quit IRC10:13
*** lpetrut1 is now known as lpetrut10:13
*** haomaiwang has joined #openstack-cinder10:16
*** gcb has quit IRC10:28
*** haomaiwang has quit IRC10:28
*** haomaiwang has joined #openstack-cinder10:28
*** gcb has joined #openstack-cinder10:28
*** markvoelker has joined #openstack-cinder10:31
*** markvoelker has quit IRC10:36
*** Rekha has joined #openstack-cinder10:37
Rekhahi all10:37
Rekhawhen volume creation from image is getting failed. when i checked in cinder-volume.log i could find this "TRACE cinder.volume.manager ImageCopyFailure: Failed to copy image to volume: Unable to find a Fibre Channel volume device."10:40
Rekhacan anybody help me out in resolving the issue...10:41
*** Dhaval has quit IRC10:41
Rekhacan anybody point out the reason for failure?10:41
openstackgerritFalk Reimann proposed openstack/cinder: Adds Auth Version 3 support in Swift Backup Driver  https://review.openstack.org/24797710:42
*** e0ne has quit IRC10:43
*** ankit_ag has quit IRC10:46
*** rushiagr is now known as rushiagr_away10:46
*** ankit_ag has joined #openstack-cinder10:47
*** lprice1 has joined #openstack-cinder10:49
*** lprice has quit IRC10:49
*** e0ne has joined #openstack-cinder10:51
*** EinstCrazy has joined #openstack-cinder10:54
*** lpetrut has quit IRC10:58
*** lpetrut has joined #openstack-cinder10:59
openstackgerritMerged openstack/cinder: Imported Translations from Zanata  https://review.openstack.org/24794311:00
*** egonzalez has quit IRC11:00
openstackgerritMerged openstack/cinder: Manage existing: fix volume object saving  https://review.openstack.org/24776911:03
*** houming has quit IRC11:05
*** haomaiwang has quit IRC11:08
nikeshmRekha: what type of setup are yu using?11:16
nikeshmopenstack or devstack11:16
nikeshmwhat volume driver are you using11:17
openstackgerritMitsuhiro Tanino proposed openstack/cinder: Robustify writing iscsi target persistence file  https://review.openstack.org/24788611:20
*** rushiagr_away is now known as rushiagr11:23
*** dims has joined #openstack-cinder11:24
*** salv-orlando has quit IRC11:28
*** salv-orlando has joined #openstack-cinder11:29
*** rushiagr is now known as rushiagr_away11:30
*** ociuhandu has quit IRC11:34
*** yusuke has quit IRC11:34
openstackgerritMichal Dulko proposed openstack/cinder: Eager load columns in volume_get_active_by_window  https://review.openstack.org/24803511:41
*** salv-orlando has quit IRC11:45
*** markus_z is now known as markus_z_meeting11:58
*** bluex has quit IRC11:58
*** houming has joined #openstack-cinder11:58
*** bluex has joined #openstack-cinder12:00
*** zhangjn has joined #openstack-cinder12:02
*** zhangjn has quit IRC12:02
*** zhangjn has joined #openstack-cinder12:03
*** haomaiwa_ has joined #openstack-cinder12:05
Rekha@nikeshm openstack setup and hp3parfcdriver12:12
*** _amrith_ is now known as amrith12:14
*** salv-orlando has joined #openstack-cinder12:17
*** e0ne has quit IRC12:19
*** e0ne has joined #openstack-cinder12:19
*** houming has quit IRC12:20
*** houming has joined #openstack-cinder12:20
nikeshmRekha : then i think some packages might be missing12:23
nikeshmregarding FC12:24
nikeshmon the node12:24
nikeshmbetter check hp3parfcdriver guide12:24
*** houming has left #openstack-cinder12:25
Rekha@nikeshm on wic node you are talking about...you men on storage node12:25
*** delatte has quit IRC12:25
Rekha@nikeshm *you mean on storage node12:25
nikeshmRekha: ya i think so12:28
nikeshmor may be on controller node12:29
*** alejandrito has joined #openstack-cinder12:30
nikeshmRekha : not sure better wait for hp guys hemna DuncanT kmartin12:30
Rekha@nikeshm thank you12:31
nikeshmRekha: are you able to create FC volumes and attach FC volumes12:31
*** markvoelker has joined #openstack-cinder12:32
Rekhai able to create empty fc volumes but not able to attach this created volume to instance12:33
nikeshmRekha: thats the issue12:33
Rekha@nikeshm but on same 3par storage array i able to create ISCSI volume from image and attach to instance also12:34
*** houming has joined #openstack-cinder12:34
Rekha@nikeshm how this issue can be resolved..12:35
nikeshmRekha: why the error are you getting when try to attach FC volumes12:36
nikeshmwhat12:36
*** aix has quit IRC12:36
nikeshmplease check nova-compute logs12:37
nikeshmany Traceback12:37
*** markvoelker has quit IRC12:37
*** raildo-afk is now known as raildo12:39
nikeshmRekha: please paste Traceback in compute logs while attaching FC volume in a pastebin and share to us12:40
nikeshmsmcginnis: how is parris, be careful there12:41
*** bluex has quit IRC12:45
*** delatte has joined #openstack-cinder12:46
Rekha@nikeshm I am not getting any information in nova logs12:47
nikeshmRekha: are you checking on compute node12:47
nikeshmi mean where nova-compute service is running12:48
*** delatte has quit IRC12:49
*** cdelatte has joined #openstack-cinder12:49
Rekha@nikeshm sorry i was checking on the controller node12:50
*** salv-orl_ has joined #openstack-cinder12:54
*** jwcroppe has joined #openstack-cinder12:55
openstackgerritFalk Reimann proposed openstack/cinder: Adds Auth Version 3 support in Swift Backup Driver  https://review.openstack.org/24797712:56
*** salv-orlando has quit IRC12:57
Rekha@nikeshm http://paste.openstack.org/show/479583/ traceback is pasted here13:00
*** jaypipes has joined #openstack-cinder13:00
*** delattec has joined #openstack-cinder13:02
nikeshmRekha: hmm. is any other?13:02
openstackgerritFalk Reimann proposed openstack/cinder: Adds Auth Version 3 support in Swift Backup Driver  https://review.openstack.org/24797713:02
nikeshmTracebacks13:03
nikeshmpresents13:03
*** cdelatte has quit IRC13:05
Rekha@nikeshm http://paste.openstack.org/show/479584/13:06
*** houming has quit IRC13:07
*** salv-orl_ has quit IRC13:09
*** salv-orlando has joined #openstack-cinder13:09
dulekAnyone aware on the status of HPE CI?13:12
*** gouthamr has joined #openstack-cinder13:13
*** setmason has joined #openstack-cinder13:15
*** salv-orlando has quit IRC13:16
nikeshmRekha: hmm better wait for hemna kmartin DuncanT13:16
*** salv-orlando has joined #openstack-cinder13:16
*** mkoderer has quit IRC13:18
*** shausy has quit IRC13:21
*** mkoderer has joined #openstack-cinder13:22
*** jkraj has quit IRC13:24
Rekha@nikeshm okay13:25
*** jkraj has joined #openstack-cinder13:25
*** jkraj has quit IRC13:26
*** aix has joined #openstack-cinder13:30
*** dave-mccowan has joined #openstack-cinder13:30
*** eikke has joined #openstack-cinder13:33
eikkewhat's the procedure to remove things from rootwrap filters? As in, how can you be sure not to remove things others depend on?13:34
openstackgerritGorka Eguileor proposed openstack/cinder: Add atomic conditional updates to objects  https://review.openstack.org/20583413:34
openstackgerritGorka Eguileor proposed openstack/cinder: Move get_by_id to CinderObject  https://review.openstack.org/21801213:34
openstackgerritGorka Eguileor proposed openstack/cinder: Improve metadata update operations  https://review.openstack.org/21637613:34
openstackgerritGorka Eguileor proposed openstack/cinder: Remove API races from attach and detach methods  https://review.openstack.org/21637713:34
*** markvoelker has joined #openstack-cinder13:35
*** akerr has joined #openstack-cinder13:35
*** ociuhandu has joined #openstack-cinder13:38
*** sborkows has joined #openstack-cinder13:40
*** bswartz has quit IRC13:41
*** amrith is now known as _amrith_13:43
*** porrua has joined #openstack-cinder13:46
*** sgotliv has quit IRC13:47
*** ankit_ag has quit IRC13:50
*** dustins has joined #openstack-cinder13:52
*** salv-orlando has quit IRC13:52
*** salv-orl_ has joined #openstack-cinder13:52
*** jaypipes has quit IRC13:59
*** [1]Thelo has joined #openstack-cinder13:59
*** Thelo has quit IRC14:02
*** [1]Thelo is now known as Thelo14:02
*** smoriya_ has quit IRC14:08
*** stevemar_ has joined #openstack-cinder14:13
*** bswartz has joined #openstack-cinder14:15
*** eharney has joined #openstack-cinder14:15
*** mriedem has joined #openstack-cinder14:16
*** salv-orl_ has quit IRC14:17
lyarwoodhemna: looks like our `DIE DIE DIE` spam in https://review.openstack.org/#/c/238838/ finally worked :)14:17
gouthamrlyarwood: for a random lurker, what is DIE DIE DIE all about?14:20
*** salv-orlando has joined #openstack-cinder14:21
*** jerrygb has joined #openstack-cinder14:21
openstackgerritEric Harney proposed openstack/cinder: Add check_uptodate.sh --checkopts to "pep8"  https://review.openstack.org/24779614:25
jgriffithgouthamr: apparantly it's top secret14:27
jgriffith;)14:27
jgriffithgouthamr: honestly I'm not quite sure either suspect it's CI recheck related but unclear14:28
gouthamrjgriffith: hah. wanted to join the spamming... :P14:30
jgriffithgouthamr: :)14:30
*** merooney has joined #openstack-cinder14:31
scottdagouthamr: People are always asking each other in this channel "How do I initiate a recheck on your CI system?" and someone mentioned that it would be great if you could do this by saying "DIE DIE DIE". It was a joke.14:31
scottdaBut there is probably some value in having a single command for this, to avoid having to ask each vendor the syntax.14:32
jgriffithscottda: oh, really?14:32
scottda"DIE DIE DIE" is a bit violent for my taste.14:32
jgriffithscottda: umm... what happened to "recheck <ci-name>"14:32
scottdajgriffith: I don't know. It just came up one day and I think it was patrickeast how changed his CI to allow the DIEDIEDIE to work.14:33
jgriffithscottda: and so that means everybody that listens for DDD does a recheck needed or not?14:33
jgriffithhuh14:33
jgriffithscottda: well alrighty then :)14:33
jgriffithscottda: thanks for the explanation :)14:33
scottdano, it's still just a joke.14:33
gouthamrjgriffith: too mainstream. We don't want mainstream. We're the new kids on the block. 8)14:34
jgriffithha14:34
scottdaPersonally, I think you storage vendors are all crazy.14:35
*** akshai has joined #openstack-cinder14:35
*** jgregor has joined #openstack-cinder14:35
gouthamrscottda: :D okay, I take that as a compliment. :)14:36
scottda:)14:36
scottdageguileo: dulek Hi. Is there an etherpad that gathers all the info, specs, and patches for Active-Active HA?14:37
jgriffithscottda: LOL14:37
*** dansmith is now known as superdan14:41
*** baumann has joined #openstack-cinder14:41
*** lprice has joined #openstack-cinder14:46
*** xyang1 has joined #openstack-cinder14:46
*** aix has quit IRC14:47
*** lprice1 has quit IRC14:47
*** aix has joined #openstack-cinder14:49
lyarwoodgouthamr: haha sorry went AFK, I honestly didn't know the background for `DIE DIE DIE`, I just saw hemna using it on my change and asked :)14:51
dulekscottda: Not the etherpad, but you may find related patches here: https://review.openstack.org/#/q/topic:bp/cinder-volume-active-active-support,n,z14:51
dulekscottda, hemna: Speaking of CIs - what's up with HPE one?14:52
*** xek has quit IRC14:55
*** nkrinner has quit IRC14:55
*** sgundur has joined #openstack-cinder14:56
scottdadulek: Thanks for the pointer to patches.14:56
scottdadulek: I don't know about HPE CI. I'm just a Helion distribution guy. HPE Storage is another world.....14:57
dulekscottda: Ah, sorry for mistaking that then.14:58
scottdadulek: I'm wondering if there's a way to help move these HA/AA specs and patches along better? I think it's hard for reviewers to figure out what to look at just from that list of patches.14:58
*** markus_z_meeting is now known as markus_z14:59
dulekThere's a main one here: https://review.openstack.org/#/c/232599/ It explains all the subsequent specs.14:59
dulekMaybe an etherpad would work, I don't know. I always felt that etherpads tend to get updated.15:00
duleks/updated/outdated15:00
scottdaYeah, that kinda works. But still doesn't give any ordering to help reviewers. It can kind of be figured out with dependencies.15:00
scottdaI'm just thinking that at this rate, we won't have HA/AA in Mitaka. Spec approval is pretty slow, and then there are the patches...15:00
*** boris-42 has joined #openstack-cinder15:01
scottdaThat's 8 specs on that list ^^^ !15:02
*** kfarr has joined #openstack-cinder15:02
*** breitz has joined #openstack-cinder15:03
dulekI'm a little behind on geguileo's updates in specs. I'll try to sync up and maybe come up with something.15:03
asselindulek, hpe ci seems to be failing lefthand-iscsi-driver-master-client-pip-vsa125http-dsvm, otherwise looks ok to me15:03
scottdaYeah, I'm not sure of the answer. I've  teammates working on testing the patches, and they were pretty confused about the dependency chain.15:04
*** _amrith_ is now known as amrith15:05
dulekscottda: It's more like specs reviews work now, not much implementation. I was focusing on upgrades for M-1, I'll try to work more on A/A next week.15:06
gouthamrlyarwood: Nice. Thanks :)15:07
scottdacool. I'm mostly out next week, but I'll attend the Cinder meeting Wed. Maybe one of us can add info to the spec priority page and nag people to review: https://etherpad.openstack.org/p/mitaka-cinder-spec-review-tracking15:08
*** jasondotstar is now known as jasondotstar_afk15:08
*** stevemar_ has quit IRC15:10
*** e0ne has quit IRC15:11
*** haomaiwa_ has quit IRC15:12
*** krtaylor has joined #openstack-cinder15:16
*** lpabon has joined #openstack-cinder15:21
openstackgerritEdwin Wang proposed openstack/cinder: FlashSystem reports error in _find_host_exhaustive()  https://review.openstack.org/24754615:22
*** jerrygb has quit IRC15:22
*** jerrygb has joined #openstack-cinder15:23
*** kfarr has quit IRC15:23
*** setmason has quit IRC15:24
*** vgridnev has quit IRC15:26
*** kfarr has joined #openstack-cinder15:26
*** crose has joined #openstack-cinder15:30
*** mc_nair has joined #openstack-cinder15:32
patrickeastjgriffith: haha yea so that die die die thing was a joke, but kind a of symptom of folks confused by how to recheck since not all the ci's respond to recheck <ci name>15:32
jgriffithpatrickeast: got it :)15:32
patrickeastPlus lots, including Jenkins get triggered by that format15:33
jgriffithpatrickeast: a number of folks seem to have been having fun putting it in comments15:33
*** alexschm has quit IRC15:33
*** e0ne has joined #openstack-cinder15:33
patrickeastLol yea I added it as an extra way to do it, zuul makes it trivial to add more trigger regexs15:33
jgriffith:)15:34
jgriffithI'll just stick with "recheck solidfire" for now15:34
patrickeastSeems to be a hit though... Gets more action than the real recheck syntax now15:34
jgriffith:)15:34
*** rcernin has quit IRC15:34
jgriffithpatrickeast: yeah, it's QUITE popular with all the kids these days15:34
jgriffith:)15:34
*** thangp has joined #openstack-cinder15:35
tbarronso I dunno but I've been told that 'recheck <backend>-ci' can trigger more than just the CI for that backend.  True?15:36
tbarronpatrickeast, jgriffith ^^^: I think that's why we do 'run netapp-ci'15:37
jgriffithtbarron: apparantly so can "DDD" :015:37
tbarronjgriffith: yeah :-)15:37
jgriffithtbarron: actually it's "recheck-<ciName>"  so Jenkins doesn't catch it15:37
jgriffithtbarron: yes, you're correrct, if there's whitespace after the "recheck" it will grab it15:38
jgriffithtbarron: anywho... DDD seems popular and seems to work so whatevs :)15:38
tbarronjgriffith: kk, I guess we did 'run ...' just to reduce the probability of people putting a space in there.15:38
jgriffithtbarron: yeah, that was another bike-shedding argument one day a while back15:39
jgriffithtbarron: I'd prefer not to re-live :)15:39
tbarronjgriffith: 'nuf said15:40
jgriffithtbarron: :)15:40
*** rhagarty_ has quit IRC15:40
*** ntpttr has joined #openstack-cinder15:40
*** rhagarty_ has joined #openstack-cinder15:40
*** wN has quit IRC15:44
openstackgerritMichal Dulko proposed openstack/cinder-specs: Add Tooz locks to support A/A HA spec  https://review.openstack.org/20261515:46
*** davechen has joined #openstack-cinder15:48
geguileoscottda: dulek In theory this should have them15:49
geguileoBut I'll check now that it is up to date and maybe structure it a little bit better15:49
*** wN has joined #openstack-cinder15:50
*** wN has joined #openstack-cinder15:50
*** stevemar_ has joined #openstack-cinder15:52
scottdageguileo: Which should have them?15:53
geguileoUuups, https://blueprints.launchpad.net/cinder/+spec/cinder-volume-active-active-support15:54
geguileoUpdating it now15:54
scottdaok, cool.15:54
geguileoBut I'm updating it now15:54
*** e0ne_ has joined #openstack-cinder15:57
*** e0ne has quit IRC15:58
*** svasheka has quit IRC15:58
*** stevemar_ has quit IRC15:58
*** stevemar_ has joined #openstack-cinder15:59
*** rcernin has joined #openstack-cinder16:00
*** rcernin is now known as rcernin|dinner16:00
*** sborkows has quit IRC16:03
*** svasheka has joined #openstack-cinder16:05
*** setmason has joined #openstack-cinder16:08
kmartinnikeshm, are you still have issues?16:12
*** daneyon_ has quit IRC16:13
*** daneyon has joined #openstack-cinder16:13
kmartinnikeshm, could be one of a few problems, either you don't have the required FC packages, http://docs.openstack.org/liberty/config-reference/content/fibrechannel.html on your compute node or your SAN fabric is not zoned correctly. Are you using the fibre channel zone manager?16:17
*** stevemar_ has quit IRC16:26
*** baumann has quit IRC16:29
*** jdurgin1 has joined #openstack-cinder16:31
*** eharney has quit IRC16:33
openstackgerritScott DAngelo proposed openstack/python-cinderclient: api-microversion support for python-cinderclient  https://review.openstack.org/24816316:33
*** stevemar_ has joined #openstack-cinder16:39
*** rcernin|dinner is now known as rcernin16:42
*** amrith is now known as _amrith_16:42
*** wilson has joined #openstack-cinder16:42
*** wilson is now known as Guest321116:43
*** stevemar_ has quit IRC16:44
*** dims_ has joined #openstack-cinder16:44
*** dims has quit IRC16:44
*** stevemar_ has joined #openstack-cinder16:45
*** wilson-liu has quit IRC16:45
duleke0ne_: Coverage for cinder/image/image_utils.py : 94%16:47
e0ne_dulek: hi. not bad:)16:47
duleke0ne_: That's why I haven't noticed, but yes, I was curious. ;)16:47
e0ne_dulek: thanks for review! I'll update patch with lost tests16:48
*** kfarr has quit IRC16:55
*** e0ne_ has quit IRC16:58
*** stevemar_ has quit IRC16:59
*** stevemar_ has joined #openstack-cinder17:00
*** chhavi has quit IRC17:00
dulekxyang1: Hi! You're rechecking EMC VNX because it failed on volume_metadata aliases patch, or just to trigger it?17:00
xyang1dulek: I was just triggering it to test your patch17:00
xyang1dulek: but the other patch is already merged so this won't really test your change17:01
dulekxyang1: Yes, that's why I've asked for testing. But don't bother, through the weekend I'll post a DO-NOT-MERGE change dependent on mine that's removing Peter's fix and trigger VNX CI there. This should make me sure on my patch. :)17:03
*** eharney has joined #openstack-cinder17:03
xyang1dulek: sure, that is way to do it:)17:03
*** stevemar_ has quit IRC17:04
xyang1dulek: I have asked Peter to test your patch but he is in. different time zone17:04
*** leeantho has joined #openstack-cinder17:07
*** sghanekar_ has joined #openstack-cinder17:09
*** jordanP has quit IRC17:09
*** jordanP has joined #openstack-cinder17:10
openstackgerritMichal Dulko proposed openstack/cinder: Fix debug output for cinder-volume-usage-audit  https://review.openstack.org/24734417:13
*** baumann has joined #openstack-cinder17:17
openstackgerritFalk Reimann proposed openstack/cinder: Add Auth Version 3 support in Swift Backup Driver  https://review.openstack.org/24797717:20
openstackgerritFergal Mc Carthy proposed openstack/cinder: Refactor SRB Unittest LVM command handlers.  https://review.openstack.org/24818717:21
*** wilson has joined #openstack-cinder17:26
*** wilson is now known as Guest4937017:26
*** rhedlind has joined #openstack-cinder17:27
*** baumann has quit IRC17:28
*** wilson-l has joined #openstack-cinder17:29
*** Guest3211 has quit IRC17:29
*** jdurgin1 has quit IRC17:29
openstackgerritNate Potter proposed openstack/cinder: Move quota checks from manager to API  https://review.openstack.org/24367317:32
*** Guest49370 has quit IRC17:32
*** edtubill has joined #openstack-cinder17:32
openstackgerritRyan McNair proposed openstack/cinder: Deprecate *_multipath_enabled flag for IBM drivers  https://review.openstack.org/24474917:37
*** zhangjn has quit IRC17:51
*** _amrith_ is now known as amrith17:52
*** crose has quit IRC17:54
*** ericksonsantos has quit IRC17:58
*** gouthamr has quit IRC18:00
*** jistr has quit IRC18:01
smcginnisdulek, scottda: This would be a great place to start capturing more HA links and info: https://etherpad.openstack.org/p/mitaka-cinder-spec-review-tracking18:03
smcginnisMaybe not everything, but at basis for where to start.18:03
*** jordanP has quit IRC18:03
hemnasmcginnis, not sure anyone in Cinder is aware of this change in Nova, but FYI: https://review.openstack.org/#/c/13585418:03
hemnaLooks like Nova is trying to move the iSCSI volume support into qemu18:04
hemnawhich doesn't support multipath at all18:04
hemna:(18:04
smcginnisHuh, that's an interesting approach. :[18:04
*** dims has joined #openstack-cinder18:05
smcginnisAnd it's not even supported by the qemu version in ubuntu.18:05
hemnayah18:05
hemnabah18:05
hemnasure would have been nice to have someone from Cinder review this before it merged18:06
hemna:(18:06
*** dims_ has quit IRC18:06
*** jordanP has joined #openstack-cinder18:06
smcginnisYeah, kind of surprised we're just finding out about it.18:06
*** apoorvad has joined #openstack-cinder18:06
hemnaback...in Feb.18:06
hemnad'oh18:06
*** jordanP has quit IRC18:06
smcginnisWhere's our nova liason!! :)18:06
hemnadangit18:06
smcginnisBut that was quite a while ago.18:06
hemnayah18:06
hemnasmcginnis, did you see the mailing list post about the encryptors patch in os-brick ?18:07
smcginnishemna: Yep, I had asked (well, Matt R did first) to have something posted there.18:07
hemnaDaniel seems to want to move Nova into doing all attachments directly w/ qemu18:07
smcginnishemna: We had discussed in reviews here, but nova should know what we're doing.18:07
* smcginnis thinks and vice versa :/18:08
hemnaI'm not sure qemu can handle all of the h/w initiators that os-brick supports18:08
hemnaand there is no FC support18:08
hemnaand there is no multipath support18:08
hemna:(18:08
smcginnishemna: I don't get it. What is the advantage of using qemu? Seems severely limited with no big win from what I see.18:08
hemnahttps://specs.openstack.org/openstack/nova-specs/specs/kilo/implemented/qemu-built-in-iscsi-initiator.html18:09
hemnaI also found that18:09
hemnakinda hints at some of their 'issues'18:09
hemnabut I'm not sure I'm buying it18:09
*** aix has quit IRC18:10
smcginnisOh! So the volume isn't attached to the nova host. Directly to the guest VM?18:10
smcginnisI'm still confused.18:10
tbarronunless I'm missing something, there are backends that won't work right without multipath support ...18:10
smcginnistbarron: It definitely could cause problems.18:11
*** e0ne has joined #openstack-cinder18:11
hemnawell it's attached to qemu18:11
hemnathey create a virsh/qemu/kvm pool18:11
smcginnisEven with one initiator, on ours we can have multiple front end ports configured that would result in multiple paths.18:11
hemnaand hand that to the guest.18:11
tbarronwe have backends that use multipath for controller failover.  And another backend that load-balances over multiple paths.18:11
hemnathis goes against the idea of multipath entirely18:11
smcginnishemna: So to get to qemu, isn't that exposing it to the host? You would still hit a LUN limit.18:12
tbarronimage copies etc. have problems unless the 'use_multipath' options are configured in nova libvirt.18:12
hemnayah I think so18:12
smcginnisWell... maybe the problem will fix itself because anyone who tries to use this will have problems.18:13
smcginnisSo they won't use it and it will go away.18:13
smcginnis:]18:13
hemnalol18:13
tbarronyeah, it's a nova.conf option, not default.18:13
*** mriedem has quit IRC18:13
hemnaqemu has no concept of FC18:14
tbarronand maybe it has its uses18:14
smcginnistbarron: Well, that's good news at least that it's not default. That could really screw up some deployments.18:14
smcginnistbarron: Maybe. Limited though.18:14
smcginnisI still don't see the benefit.18:14
hemnayah I'm grasping at it myself18:14
hemnayou lose all h/w initiator support18:15
*** harlowja_at_home has joined #openstack-cinder18:15
hemnaand force qemu to do all the iSCSI work, which is less than ideal.18:15
smcginnismriedem isn't online. I wonder if he knows anything about this. He's pretty good on the storage side of nova.18:15
hemnasmcginnis, I don't see daniel either.  berrange ?18:16
smcginnishemna: Right, why use a well established and featureful iSCSI initiator. Throw it all on qemu.18:16
smcginnisscottda: Maybe something you can bring up with nova, Mr. Liason? :)18:16
hemnaoh maybe it's danpb ?18:17
xyang1the doc says volume_drivers = iscsi=nova.virt.libvirt.volume.LibvirtNetVolumeDriver18:17
hemnaxyang1, yah, and they modified the netvolumedriver to support iscsi18:17
hemnabut only for single path18:17
xyang1:(18:17
hemnaand cinder drivers are going to get the initiator information from the host18:17
hemnawhich has it's iqn and multipath=True/False18:18
*** mriedem has joined #openstack-cinder18:18
hemnaso some backends are going to export as if it's multipath enabled18:18
hemnabut only get 1 path connected.18:18
hemnathat's bad mmmkay18:18
smcginnisProblem Description: The VMs on a compute node will be stopped when an iSCSI volume is attached or detached to a VM and it causes the host OS crash.18:19
smcginnisThere are some fundamental misconceptions here, I think.18:19
hemnayah I saw that too and thought...what?!18:20
hemnathat simply doesn't happen18:20
smcginnisThat whole list under Problem Description has me confused.18:20
xyang1wonder what backend they tested with to come up with that problem description18:21
hemnalvm ?18:21
smcginnismriedem: You're here now!18:21
smcginnismriedem: Don't suppose you've seen any of the scrollback?18:21
mriedemnope18:21
mriedembeen at lunch for the last hour18:21
smcginnismriedem: We're a bit puzzled by https://specs.openstack.org/openstack/nova-specs/specs/kilo/implemented/qemu-built-in-iscsi-initiator.html18:22
smcginnismriedem: Wondering if you have any innsight on it.18:22
smcginnisWas about to ask on -nova, but since you're back from lunch... :)18:22
*** e0ne has quit IRC18:22
hemnaooh18:23
hemnathe other downside is that extending an attached volume will never work in that scenario18:23
mriedemsmcginnis: puzzled how?18:23
smcginnismriedem: Main points being, it seems to have some fundamental issues and unclear advantage as to why.18:23
mriedemi remember pointing out long after it was in tree that the dependency story sucks18:23
smcginnismriedem: Everything under Problem Description doesn't appear to be accurate.18:23
mriedemyou have to build your own qemu18:24
smcginnisYa, there's that too.18:24
hemnamriedem, so what brings this up is the whole ML discussion about the encryptors support18:24
smcginnisOr hey, just rewrite qemu and compile it into your kernel while you're at it.18:24
smcginnisSorry. :)18:24
*** markus_z has quit IRC18:24
hemnait looks like Daniel wants to move all volume attaches to this mechanism eventually18:24
hemnathere are some serious downsides to doing this18:25
*** bardia has joined #openstack-cinder18:25
hemna1) no h/w initiator support.  2) No multipath support 3) no extending attached volumes support 4) no fibre channel support 5) no vendor based transports support18:25
smcginnisLooks like different discussion going on in -nova at the moment.18:26
*** edtubill has quit IRC18:27
hemnaI wonder if multiattach would even work as wel18:27
mriedemok i'm not up to date on the ML thread yet either18:27
smcginnishemna: Doesn't look like it.18:27
hemnamriedem, what you don't read the ML at 3am?18:27
hemna:P18:27
mriedemwhat was wrong with the spec again?18:27
mriedemi do, just haven't gotten to that one yet18:27
mriedemlots of plates spinning18:27
smcginnismriedem: I think hemna summed it up well a few lines back.18:28
mriedemi'm all for ripping crappy stuff out of nova18:28
mriedembtw18:28
smcginnismriedem: There are definite "shortcomings" with this approach.18:28
mriedemwhich is "1) no h/w initiator support.  2) No multipath support 3) no extending attached volumes support 4) no fibre channel support 5) no vendor based transports support"18:28
smcginnismriedem: And questionable motivation for needing it.18:28
mriedem?18:28
smcginnismriedem: Yeah, those are some of the issues with it.18:28
mriedemi seem to remember the reason for that spec was to just keep some gorp out of the host to make debugging easier18:29
hemnahttps://github.com/openstack/nova/blob/master/nova/virt/libvirt/volume/net.py#L5618:29
hemnaso it looks like that code made it into the split out libvirt drivers18:29
hemnaso it's still there.18:29
mriedemand possibly some performance thing, but i never reviewed that before it went in18:29
hemnafwiw18:29
mriedemit's in os-brick you mean?18:29
hemnano18:29
hemnathe https://review.openstack.org/#/c/13585418:29
hemnathat guy landed back in Feb.18:30
hemnaand it's still in the 'new' libvirt/volume/net.py code18:30
mriedemhttps://review.openstack.org/#/c/135854/18:30
mriedemyeah,18:30
hemnaso I guess there is 2 issues here really18:30
hemnathe existing patch https://review.openstack.org/#/c/13585418:30
mriedemit even says no multipath in the commit message18:30
hemnaand 2) the direction that "Nova" wants to go in the future wrt volume attaches18:31
mriedem"3) no extending attached volumes support" wasn't a thing when this was added to nova right?18:31
hemnamriedem, correct18:31
mriedem(4) could be added for fibrechannel i'd think?18:31
hemnaI bring it up now, because I'm working on os-brick code to add that ability18:31
smcginnismriedem: Yeah, not really at the time.18:31
hemnamriedem, https://review.openstack.org/#/c/243730/18:31
smcginnismriedem: But it prevents adding support for that. And possibly other betterness in the future.18:32
hemnathat's my WIP that I'm testing now to add the ability to resize attached18:32
mriedemfwiw, the if statements in https://review.openstack.org/#/c/135854/19/nova/virt/libvirt/volume.py make me said18:32
mriedem*sad18:32
mriedemif that helps18:32
hemnaerr https://review.openstack.org/#/c/243776/18:32
hemnathat one...not the spec :P18:32
hemnamriedem, yah I saw that as well. that seems like a painful way to go, long term18:33
hemnaeventually everything will be a big if/then/else18:33
mriedemok, so what i'm hearing is this very niche thing that you have to build your owen qemu for (on trusty at least), and with little gain (makes mgmt / debug easier for ops?), is blocking cinder from doing things it wants to do with volume resize18:34
*** harlowja_at_home has quit IRC18:34
mriedemplus it doesn't support multipath and fibre18:34
hemnacorrect18:34
*** e0ne has joined #openstack-cinder18:34
smcginnismriedem: Yeah. And we otherwise generally thing it's a Bad Thing to do for end users.18:35
hemnano idea how you'd be able to debug failed attaches/detaches18:35
smcginniss/thing/think/18:35
mriedembad thing b/c of not knowing how to debug attach/detach failures?18:35
mriedemor the multipath/fibre limitations?18:35
mriedemand/or18:36
hemnayes :)18:36
smcginnisD: All of the abobe18:36
* smcginnis can't type tonight. Must be the wine.18:36
mriedemtonight?18:36
hemna:P18:36
mriedemwhere are you?18:36
smcginnismriedem: Paris :)18:37
mriedemoh18:37
*** gouthamr has joined #openstack-cinder18:37
smcginnisDoing code reviews on my anniversary. Good thing my wife is jetlagged. :)18:37
mriedem\o/18:37
mriedemso you guys have raised this all in the ML and danpb said it's not a problem and you're crazy and i should just go read the ML?18:37
openstackgerritNate Potter proposed openstack/cinder: Move quota checks from manager to API  https://review.openstack.org/24367318:38
smcginnismriedem: Was thinking that's probably our next step.18:38
mriedemoh, well if you haven't raised it in the ML yet, please do18:38
smcginnismriedem: Was just curious if you had any extra insight, cuz we certainly didn't.18:38
mriedemi'm definitely not an expert on this specific thing18:38
hemnano I haven't made a post on the ML about it yet18:38
mriedemno, like i said, i stumbled on this spec long after it was merged,18:38
hemnajust discovered it like 15 minutes ago18:38
mriedemand i was raising some hell about the 'bring your own qemu' limitatoin18:38
smcginnismriedem: Yeah, thanks for you're input though.18:38
smcginnismriedem: Just wasn't sure if there was someting we were missing.18:39
mriedemred hat probably doesn't have a problem with that part, i'm assuming OSP provides that18:39
smcginnisGood question. eharney, any idea if this all works on RH?18:39
*** edmondsw has joined #openstack-cinder18:39
smcginniseharney: https://specs.openstack.org/openstack/nova-specs/specs/kilo/implemented/qemu-built-in-iscsi-initiator.html18:40
smcginniseharney: If you weren't following along at home. :)18:40
hemnaheh, I'll have to find out if it works on Helion.18:40
eharneyumm18:40
smcginnisI guess if it doesn't work on anyone's distro, there's probably not much to worry about for now. :)18:40
smcginniseharney: :)18:40
eharneyi'm not really sure if that's something we support18:41
hemnaI'm more concerned w/ the direction that danpb wants to move Nova.  it seems like a major step backwards IMHO18:41
smcginnisThe spec doesn't call out which version of qemu.18:41
*** jbernard has quit IRC18:44
openstackgerritMerged openstack/cinder: fast8: Skip git rm'd files  https://review.openstack.org/24779318:45
scottdasmcginnis: Sorry, I was not involved when that spec merged ...18:45
smcginnisscottda: Weak excuse.18:45
scottdaI don't think there was a Nova liason at the time :)18:45
smcginnisHah! :)18:45
eharneyseems like it's not totally uncharted territory, qemu has used rbd and gluster that way for a long time18:46
smcginnisscottda: Yeah, didn't realize at first it was so long ago.18:46
smcginniseharney: Oh! Didn't realize there was built in support for those.18:46
*** stevemar_ has joined #openstack-cinder18:46
scottdaI agree with hemna . This particular feature is probably not being used much, but it's not a good direction to go.18:46
hemnaeharney, yah they have, but moving all transports into qemu seems like a major mistake IMHO18:47
smcginniseharney: Don't suppose you have any idea the advantage of qemu handling that vs the host OS?18:47
eharneyfor rbd/gluster it's usually about performance scalability, not really sure in the case of iSCSI18:48
*** stevemar_ has quit IRC18:48
eharneyand, well, for rbd it's just the easiest way to implement it anyway18:48
smcginniseharney: It's more performant having qemu handle it?18:48
*** stevemar_ has joined #openstack-cinder18:48
eharneysmcginnis: that's what i hear, but i have no idea how that translates to this case18:49
hemnasounds like a potential for a bottleneck to me18:49
smcginniseharney: Interesting data point at least. Thanks!18:49
*** geguileo has quit IRC18:52
scottdaI think I'd rather debug iscsi attach errors at the OS level than at the libvirt level.18:53
hemnascottda, +1000018:53
mriedemdanpb did mention something about being able to tweak performance with qemu native stuff,18:53
mriedemin a spec asking for rbd encryption support in nova18:53
mriedemlet me dig that up18:53
mriedemhttps://review.openstack.org/#/c/239798/3/specs/mitaka/approved/rbd-volume-encryption.rst18:54
mriedembasically dan is -1 on that since it uses dm-crypt (more dm-crypt usage)18:54
mriedemand he wants to wait for native qemu encryption support18:54
mriedemwhich is like N at the earliest18:54
*** salv-orl_ has joined #openstack-cinder18:54
mriedemif you guys have comments on that spec too, please provide input18:54
hemnaman, I can only see the problems we'll have with this18:56
hemnacinder uses 1 mechanism to do attaches/encryption and nova uses another18:56
openstackgerritMerged openstack/os-brick: Parse FCoE sysfs device paths  https://review.openstack.org/23883818:56
smcginnisWhat could go wrong?18:56
*** e0ne has quit IRC18:57
*** salv-orlando has quit IRC18:57
hemnaso FWIW, I put a -2 on the os-brick encryptors patch for the time being.18:58
hemnahttps://review.openstack.org/#/c/247372/18:58
hemnauntil we hash this out18:58
hemnadanpb already -1'd it for the same reasons he posted in the ML.18:58
hemnaugh19:00
hemnasmcginnis, https://review.openstack.org/#/c/247372/19:00
hemnamore discussion happening there.19:00
hemnawe need to has this out.19:00
hemnasounds like, danpb thinks cinder shouldn't ever do encryption.19:01
hemnawhich precludes doing copy volume <--> image operations.19:01
hemnaor backup19:01
*** setmason has quit IRC19:02
scottdaOr Cinder attach without Nova19:02
hemnaheh19:02
hemnaand that19:02
smcginnisHmm, securing against cinder being able to do any encryption/decryption.19:03
smcginnisCuz we'd be the vulnerable point?19:03
hemnaevidently, Cinder will compromise the storage, according to him.19:03
*** setmason has joined #openstack-cinder19:03
* hemna has a sad19:03
smcginnisIf Cinder is compromised, I think there are much bigger issues. But that might just be me.19:04
hemnasmcginnis, +119:04
hemnathis isn't what I had planned for my Friday19:04
hemnaheh19:04
smcginnis:(19:05
smcginnisFWIW, I think your last comment on there is exactly the point.19:05
hemnathanks.19:06
hemnaDuncanT, ping19:06
hemnaDuncanT, replied to the ML post, but wasn't specific to the direction of attaches, but more direct to the encryptors code.19:06
*** lpetrut has quit IRC19:06
hemnawe need to craft a response about the direction and get our heads aligned with Nova on this.19:06
hemnaas this is a vast departure from the current situation and has major issues IMHO.19:07
scottdaDuncan likely won't be back to work until Sunday, FYI. It's his weekend.19:07
hemnascottda, ah yah good poing19:08
hemnapoint19:08
*** Lee1092 has quit IRC19:09
*** timcl has quit IRC19:10
scottdaso, hemna is right about  getting our heads aligned and crafting a response, but the logistics are tricky....19:10
scottdaWith some people on weekend, some folks living it up in Paris, and Thanksgiving Holiday week coming up...19:11
hemnayah19:11
*** jordanP has joined #openstack-cinder19:11
hemnaI'll be around Monday, Tuesday.19:11
*** jordanP has quit IRC19:12
scottdaIt seems that some things are clear: Cinder is not going to remove any existing encryption/decrypt support, as it is necessary.19:12
smcginnisHah! I'll be around Tuesday for sure.19:12
smcginnisscottda: +119:12
*** dustins has quit IRC19:12
hemnayah, that assertion about Cinder not doing encryption is unrealistic.19:13
*** dustins has joined #openstack-cinder19:13
*** stevemar_ has quit IRC19:13
scottdaSo is the real question about what can be moved into Brick, and whether it should?19:13
smcginnisI can see the desire for it. In general it probably is a good thing to limit decrypt points as much as possible.19:13
smcginnisBut that's just not the reality.19:13
*** mc_nair has quit IRC19:13
hemnathe only way to do it, is to completely change how we do the copy volume <--> image operations entirely19:14
hemnaand spin up worker VMs19:14
hemnathat then fire up a cinder-data service inside the vm19:14
hemnaand attach the volume to the qemu host.19:14
hemnathat seems like a nightmare to me19:14
*** jordanP has joined #openstack-cinder19:14
hemnagood luck supporting FC19:14
scottdaThere's still work in-flight to have the ability to attach/detach without nova19:14
smcginnisOi. :(19:15
smcginnisscottda: Right, this would completely preclude the ability to do non-Nova stuff with encryption.19:15
hemnaI haven't looked at the encryptor code in the brick patch yet.19:15
*** stevemar_ has joined #openstack-cinder19:16
hemnait may be fine, or it may be borked.19:16
smcginnisWell, wife finally waking up. Better stop working now. :)19:16
hemnabut the idea in general seems ok.19:16
hemna:)19:16
*** gouthamr has quit IRC19:16
scottdasmcginnis: Have a good one.19:17
scottdahemna: I need to go eat breakfast. Let's put our heads together later. Maybe others will catch up on the subject in scroll back and weigh in.19:18
tbarronsmcginnis: enjoy the weekend!19:18
hemnascottda, ok sounds good19:18
tbarronscottda, hemna: the idea that the storage service (cinder, not various backends) is more likely to be compromised than nova seems to me a bit weird.19:20
tbarronI get an argument that the attack surface should be made as small as possible though  - but not smaller than possible :-) and the copy to/from image counter seems pretty decisive.19:21
hemnatbarron, yah, I can't see how a c-vol host is more vulnerable than a n-cpu host19:21
tbarronan n-cpu host with guests whom from a security standpoint should be assumed to be hostile.19:22
*** jdurgin has quit IRC19:22
hemnayup19:22
hemnaand cinder only has that volume for a 'short' period of time19:22
hemnait's transient19:22
tbarronnow if somene wanted to implement a general data-transfer service and limit the use of the keys to that service ...19:23
*** jdurgin has joined #openstack-cinder19:25
hemnabut that's another point of access19:25
hemnawhich is danpb's argument that it should only ever be n-cpu19:25
*** baumann has joined #openstack-cinder19:25
tbarronyup19:27
openstackgerritGerald McBrearty proposed openstack/cinder: add force flag on first vdisk host mapping call  https://review.openstack.org/24822519:28
tbarronI wonder if when he says "a core aim .. is to provide against a compromised storage service" he's conflating cinder code and storage backend service?19:28
scottdac-vol is only accessible via admin API, and any sane deployment has it on a separate control plane network19:29
scottdaAnd storage backend is even further isolated.19:29
hemnatbarron, yah I'm not exactly sure what he's implying19:29
hemnascottda, c-vol is also accessible to an operator19:29
tbarroncloud administrators might not trust backend storage administrators, but I don't see cloud nova admins not trusting cloud storage admins, as it were19:29
scottdatrue. but not to an end user19:29
hemnascottda, correct.19:29
scottdaThe operator has access to the nova nodes as well.19:30
hemnait's a good goal to have19:30
hemnabut, in this case unrealistic.19:30
tbarronthe operator can get at the guest's memory if they really want to.19:30
openstackgerritAlex O'Rourke proposed openstack/cinder: LeftHand: Remove self.db assignment  https://review.openstack.org/24822819:30
scottdaIn our public cloud, Cinder team must have access to the Nova nodes. There's no way to debug and deal with incidents without access.19:31
scottdaSeems like this is a problem that is better solved with smart key management.19:31
tbarronscottda: there you go, messing up a good argument with real world facts.19:31
scottdaI respect security experts. But they are kinda like lawyers...find a possible flaw and shut things down just in case. Not realistic.19:32
tbarronyeah, I've had to debug/support customers who are inside an airgap.  Not fun telling them what to grep for and read back over the phone.19:34
* tbarron leaves for a meeting19:34
*** gouthamr has joined #openstack-cinder19:35
hemnaOpenStack has key management service already19:37
hemnaok I posted a quick reply on the ML19:37
hemnawrt to the encryptors stuffs.19:37
*** jdurgin has quit IRC19:41
scottdaCool19:42
tbarronhemna: well said.19:42
*** daneyon_ has joined #openstack-cinder19:44
*** mc_nair has joined #openstack-cinder19:44
*** timcl has joined #openstack-cinder19:45
*** daneyon has quit IRC19:48
*** ociuhandu has quit IRC19:49
*** eharney has quit IRC19:57
*** Guest37770 has quit IRC19:57
*** lpabon has quit IRC20:03
*** setmason has quit IRC20:07
*** sghanekar__ has joined #openstack-cinder20:11
*** sghanekar_ has quit IRC20:15
*** apoorvad has quit IRC20:21
*** sghanekar__ has quit IRC20:22
*** dims has quit IRC20:23
*** thangp has quit IRC20:26
*** setmason has joined #openstack-cinder20:27
*** daneyon has joined #openstack-cinder20:30
*** sghanekar__ has joined #openstack-cinder20:31
*** wilson has joined #openstack-cinder20:32
*** daneyon_ has quit IRC20:33
*** wilson is now known as Guest5750820:33
*** mriedem has quit IRC20:33
*** setmason_ has joined #openstack-cinder20:35
*** eharney has joined #openstack-cinder20:36
*** wilson-l has quit IRC20:36
*** setmason has quit IRC20:37
*** setmason_ is now known as setmason20:37
*** porrua has quit IRC20:40
*** akerr has quit IRC20:41
*** merooney has quit IRC20:42
openstackgerritScott DAngelo proposed openstack/python-cinderclient: api-microversion support for python-cinderclient  https://review.openstack.org/24816320:48
*** timcl has left #openstack-cinder20:49
*** setmason has quit IRC20:50
*** jdurgin has joined #openstack-cinder20:52
*** jgregor has quit IRC20:59
*** openstackstatus has quit IRC21:02
*** openstack has joined #openstack-cinder21:04
nikeshmkmartin: thanks, actually Rekha was facing issue21:08
nikeshmRekha: are you online21:09
nikeshmRekha: you can see chat history for solution of your problem21:10
nikeshmsmcginnis : whats up21:11
*** daneyon_ has joined #openstack-cinder21:12
*** jgregor has joined #openstack-cinder21:13
Swansonnikeshm, I'm pretty sure smcginnis is out on the town with his wife for their anniversary.21:13
*** marcusvrn_ has quit IRC21:15
nikeshmSwanson: nice, but i heard he is in Paris, so better be careful due to recent activities in Paris21:15
*** daneyon has quit IRC21:16
nikeshm:)21:16
openstackgerritGerald McBrearty proposed openstack/cinder: add config option to control flash copy rate  https://review.openstack.org/24825021:16
*** ociuhandu has joined #openstack-cinder21:20
*** apoorvad has joined #openstack-cinder21:21
*** apoorvad has quit IRC21:27
*** jbernard has joined #openstack-cinder21:30
*** stevemar_ has quit IRC21:31
*** dustins has quit IRC21:33
*** sghanekar__ has quit IRC21:35
*** amrith is now known as _amrith_21:39
*** _amrith_ is now known as amrith21:39
*** crose has joined #openstack-cinder21:39
openstackgerritxing-yang proposed openstack/cinder: Add a GateDriver  https://review.openstack.org/24495621:40
*** jordanP has quit IRC21:41
*** delattec has quit IRC21:44
hemnagatedriver!21:46
*** mriedem has joined #openstack-cinder21:46
* hemna waits for someone to file a bug against the GateDriver not supporting data consistency.......21:48
xyang1hemna: I know it is going to come:)21:49
hemnahehe yah21:49
hemnarename it to the DieDieDieDriver21:49
hemnaj/k21:50
*** gouthamr has quit IRC21:50
xyang1hemna: :)21:50
*** rcernin has quit IRC21:50
hemna:)21:50
*** IlyaG has joined #openstack-cinder21:53
*** amrith is now known as _amrith_21:56
*** dave-mccowan has quit IRC22:03
*** davechen has left #openstack-cinder22:06
*** bswartz has quit IRC22:08
*** sgundur has quit IRC22:11
*** edtubill has joined #openstack-cinder22:18
*** sghanekar__ has joined #openstack-cinder22:36
*** baumann has quit IRC22:38
*** crose has quit IRC22:42
*** jwcroppe has quit IRC22:44
*** jgregor has quit IRC22:49
*** tpeoples has quit IRC22:49
*** simondodsley_ has quit IRC22:50
*** mc_nair has quit IRC22:52
*** ntpttr has left #openstack-cinder22:52
*** Guest57508 has quit IRC22:52
*** Guest57508 has joined #openstack-cinder22:53
*** krtaylor has quit IRC22:55
*** stevemar has joined #openstack-cinder23:03
*** xyang1 has quit IRC23:03
*** stevemar_ has joined #openstack-cinder23:04
*** stevemar_ has quit IRC23:04
*** krtaylor has joined #openstack-cinder23:08
*** wilson has joined #openstack-cinder23:12
*** wilson is now known as Guest8112623:12
*** Guest57508 has quit IRC23:14
*** jwcroppe has joined #openstack-cinder23:14
*** edtubill has quit IRC23:19
*** changbl has quit IRC23:23
*** jwcroppe has quit IRC23:23
*** changbl has joined #openstack-cinder23:27
*** alejandrito has quit IRC23:29
*** changbl has quit IRC23:33
*** markvoelker has quit IRC23:34
*** adelia has joined #openstack-cinder23:38
*** adelia_ has joined #openstack-cinder23:38
*** jwcroppe has joined #openstack-cinder23:39
*** wilson-l has joined #openstack-cinder23:41
*** adelia has quit IRC23:43
*** adelia_ has quit IRC23:43
*** Guest81126 has quit IRC23:44
*** [1]Thelo has joined #openstack-cinder23:59

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!