Wednesday, 2021-03-10

*** vishalmanchanda has quit IRC00:05
*** brinzhang has joined #openstack-cinder00:22
openstackgerritMerged openstack/python-cinderclient master: Add MV 3.63 to the max supported version
*** tosky has quit IRC00:36
openstackgerritAlan Bishop proposed openstack/python-cinderclient master: Bump API max version to 3.64
*** martinkennelly has quit IRC00:41
rm_workThanks for the review on ! The release note has been added as requested.00:55
rm_workClient releases are going out this week -- do folks think it might be feasible to get that merged in time, or would we need to wait for another release to roll around?00:56
*** sapd1 has quit IRC01:05
*** caiqilong has joined #openstack-cinder01:08
*** dviroel_ has joined #openstack-cinder01:31
*** dviroel has quit IRC01:38
*** dviroel_ is now known as dviroel01:39
*** hamalq has quit IRC02:14
openstackgerritBrian Rosmaita proposed openstack/cinder master: Use os-brick 4.3.0
openstackgerritMerged openstack/cinder master: [Pure] Add missing FC host personality support
*** udesale has joined #openstack-cinder03:29
*** rratnaka has joined #openstack-cinder03:40
*** rratnaka14 has joined #openstack-cinder03:40
*** knomura has joined #openstack-cinder04:27
*** rratnaka has quit IRC04:28
*** carloss has quit IRC04:29
*** GirishChilukuri_ has joined #openstack-cinder04:37
*** zenkuro has joined #openstack-cinder04:47
*** knomura has quit IRC04:53
*** vishalmanchanda has joined #openstack-cinder04:59
*** ajitha has joined #openstack-cinder05:11
ajitha@rosmaita : Thank you for your comments. Nimble driver is reporting within HPE Storage CI as iscsi-driver-nimble.05:13
*** m75abrams has joined #openstack-cinder05:16
openstackgerritMerged openstack/cinder master: Hitachi: Trace REST API input/output logs
*** zenkuro has quit IRC06:00
*** zenkuro has joined #openstack-cinder06:01
openstackgerritOpenStack Proposal Bot proposed openstack/cinder master: Imported Translations from Zanata
*** whoami-rajat has joined #openstack-cinder06:23
*** e0ne has joined #openstack-cinder06:31
*** e0ne has quit IRC06:31
*** manoj_katari has joined #openstack-cinder06:41
*** enriquetaso has quit IRC06:43
*** venkatakrishnath has joined #openstack-cinder07:02
*** knomura has joined #openstack-cinder07:05
*** knomura has quit IRC07:09
*** ajitha has quit IRC07:21
*** e0ne has joined #openstack-cinder07:22
*** e0ne has quit IRC07:27
*** ociuhandu has joined #openstack-cinder07:45
*** ociuhandu has quit IRC07:51
*** rratnaka14 has quit IRC07:58
*** ociuhandu has joined #openstack-cinder08:05
*** lpetrut has joined #openstack-cinder08:06
*** rratnaka has joined #openstack-cinder08:08
*** ociuhandu has quit IRC08:16
*** ociuhandu has joined #openstack-cinder08:17
*** ociuhandu has quit IRC08:17
*** ociuhandu has joined #openstack-cinder08:18
*** brinzhang_ has joined #openstack-cinder08:20
*** brinzhang has quit IRC08:23
*** ociuhandu has quit IRC08:23
*** rpittau|afk is now known as rpittau08:27
*** ociuhandu has joined #openstack-cinder08:28
*** sfernand has quit IRC08:36
*** tosky has joined #openstack-cinder08:46
*** MrClayPole has quit IRC08:47
*** MrClayPole has joined #openstack-cinder08:48
*** jbernard has quit IRC08:58
*** manoj_katari has quit IRC09:03
*** jbernard has joined #openstack-cinder09:10
openstackgerritXuan Yandong proposed openstack/cinder master: Restrict manage to encrypt volume type
*** sapd1 has joined #openstack-cinder09:16
openstackgerritLee Yarwood proposed openstack/cinder-tempest-plugin master: Set CINDER_ISCSI_HELPER explicitly for tgtadm job
*** martinkennelly has joined #openstack-cinder09:47
openstackgerritLuigi Toscano proposed openstack/cinder-tempest-plugin master: DNM test a change
*** rcernin has quit IRC09:54
*** knomura has joined #openstack-cinder10:10
lyarwoodtosky: sorry didn't see that, just updated the commit in my change and rechecked yours10:10
toskybtw, the cinder weekly meeting will happen before those are merged anyway10:13
toskyor maybe not, let's see what the gates say10:13
lyarwoodtosky: ack, what time is the meeting?10:14
* lyarwood checks the wiki10:14
*** knomura has quit IRC10:14
toskylyarwood: 14:00 UTC - I've added a point about this to the agenda10:14
lyarwoodtosky: I'm going to struggle to make the meeting today tbh, have some childcare issues between ~14:00 and ~16:00 UTC to take care of10:15
toskydid you mention you are still seeing issues with lioadm, though?10:15
lyarwoodno, lioadm didn't show any issues10:16
lyarwoodI've just not been able to reproduce the volume detach failure with tgtadm10:16
lyarwoodI've reproduced the encrypted volume failure where the path given by os-brick can point to another volume entirely10:16
lyarwoodbut that's specific to encrypted volumes as we use the /dev/disk/by-id paths for these volumes10:17
lyarwoodand again that was with tgtadm10:17
lyarwoodbut given that behaviour I'm pretty sure the use of multiple tgtadm but multiple c-vol's is going to cause other issues10:17
lyarwoodurgh sorry10:18
lyarwood`use of tgtadm by multiple c-vol's`10:18
* lyarwood grabs a coffee and then adds the above to the pad for the meeting10:19
*** ociuhandu has quit IRC10:23
*** ociuhandu has joined #openstack-cinder10:24
*** ociuhandu has quit IRC10:31
*** ociuhandu has joined #openstack-cinder10:31
*** e0ne has joined #openstack-cinder10:35
*** masterpe has quit IRC10:36
*** carloss has joined #openstack-cinder10:39
*** masterpe has joined #openstack-cinder10:45
openstackgerritGorka Eguileor proposed openstack/cinder master: Remove nested quota leftovers
whoami-rajatthanks geguileo  ^11:09
geguileowhoami-rajat: no problem  :-)11:10
geguileoI'm looking into quotas to try to fix them11:10
geguileoSo I'm doing some cleanup to make it easier11:10
whoami-rajatgeguileo:  great, i will also put up a patch to validate project with keystone in quotas, here's the base patch
whoami-rajatbasically refactoring some methods out of quotas (hope it doesn't conflict with your patch)11:12
geguileodoesn't look like it does11:13
*** sapd1 has quit IRC11:21
*** udesale_ has joined #openstack-cinder11:24
openstackgerritLee Yarwood proposed openstack/devstack-plugin-ceph master: tempest: Use a more appropriate name for the skip list
geguileowhoami-rajat: why are you removing some tests from
geguileowhoami-rajat: for example test_get_project_keystoneclient_v3_with_subtree11:28
*** udesale has quit IRC11:28
geguileoor test__filter_domain_id_from_parents_domain_as_parent11:28
openstackgerritVenkata krishna Thumu proposed openstack/cinder master: [SVF] Set volume IOPS based on volume size
whoami-rajatgeguileo: yes, they were part of getting project hierarchy which IIRC i confirmed with lance isn't used anymore11:38
whoami-rajatalso i think that feature isn't complete from keystone side11:38
geguileowhoami-rajat: but you left the code that checked that11:38
*** venkatakrishnath has quit IRC11:38
geguileoif we leave the code, we should leave the tests11:39
geguileowhoami-rajat: you didn't remove any code, right? you only moved the methods11:39
whoami-rajatgeguileo: the code was removed with nested quota removal11:39
geguileowhoami-rajat: oh, I see you just changed the code, you removed get_project_hierarchy11:41
whoami-rajatoh wait, i think I'm mixing up things11:41
whoami-rajatyes, i changed get_project_hierarchy to get_project11:41
whoami-rajatand the methods dependent on get_project_hierarchy were removed with the nested quota driver11:42
geguileowhoami-rajat: then you reaaaaaally need to explain that11:42
geguileoa refactor is basically doing the same thing11:42
geguileoif you change things in the same patch it should be explained (I think)11:43
whoami-rajatyes you're right, it was initially a refactor only patch then i changed things and forgot to update commit msg11:43
whoami-rajati will update it11:43
whoami-rajatthanks for mentioning it11:44
*** ociuhandu has quit IRC12:12
*** ociuhandu has joined #openstack-cinder12:12
*** ociuhandu has quit IRC12:12
*** ociuhandu has joined #openstack-cinder12:13
*** ociuhandu has quit IRC12:18
*** manoj_katari has joined #openstack-cinder12:22
*** ociuhandu has joined #openstack-cinder12:44
*** zenkuro has quit IRC12:45
*** ociuhandu has quit IRC12:48
*** venkatakrishnath has joined #openstack-cinder12:51
*** ociuhandu has joined #openstack-cinder12:51
*** venkatakrishnath has quit IRC12:53
*** rratnaka has quit IRC12:59
*** ociuhandu has quit IRC13:02
*** ociuhandu has joined #openstack-cinder13:05
*** manoj_katari has quit IRC13:17
*** ociuhandu has quit IRC13:19
*** ociuhandu has joined #openstack-cinder13:20
*** rratnaka has joined #openstack-cinder13:20
*** brinzhang_ has quit IRC13:23
*** brinzhang_ has joined #openstack-cinder13:24
*** ajitha has joined #openstack-cinder13:24
*** ociuhandu has quit IRC13:25
*** ociuhandu has joined #openstack-cinder13:27
*** ociuhandu has quit IRC13:33
*** sapd1 has joined #openstack-cinder13:38
eharneyeasy patch to review:
*** enriquetaso has joined #openstack-cinder13:46
*** manoj_katari has joined #openstack-cinder13:49
*** ociuhandu has joined #openstack-cinder13:49
*** venkatakrishnath has joined #openstack-cinder13:50
manoj_katari@Core reviewers: could you please review these minor changes in IBM SVF driver waiting for merge from a long time.13:56
rosmaitaCourtesy reminder: Cinder meeting in #openstack-meeting-alt at 1400 UTC14:00
rosmaitajungleboyj rosmaita smcginnis tosky whoami-rajat m5z e0ne geguileo eharney walshh_ jbernard lseki sfernand rajinir enriquetaso hemna ^^14:00
*** ociuhandu has quit IRC14:01
*** ociuhandu has joined #openstack-cinder14:01
*** sfernand has joined #openstack-cinder14:05
*** jrollen is now known as jroll14:13
*** felipe_rodrigues has joined #openstack-cinder14:14
openstackgerritVenkata krishna Thumu proposed openstack/cinder master: [SVF] Set volume IOPS based on volume size
openstackgerritLance Bragstad proposed openstack/cinder master: DNM: Configure tempest-integrated-storage to use new policy defaults
openstackgerritLance Bragstad proposed openstack/cinder master: DNM: Configure tempest-integrated-storage to use new policy defaults
openstackgerritLance Bragstad proposed openstack/cinder master: DNM: Configure tempest-integrated-storage to use new policy defaults
*** fnordahl has quit IRC14:42
lbragstadrosmaita ^ that's another approach we can use to increase confidence in the default policies while folks continue working on the protection tests in ctp14:42
openstackgerritLance Bragstad proposed openstack/cinder master: DNM: Configure tempest-integrated-storage to use new policy defaults
openstackgerritMerged openstack/python-brick-cinderclient-ext master: remove unicode from code
*** ociuhandu has quit IRC14:54
*** ociuhandu has joined #openstack-cinder14:54
eharneyi'm not sure whether is showing that data in the right place regardless of the policy considerations15:01
*** TusharTgite has joined #openstack-cinder15:01
rosmaitahemna: i think when you introduce the policy, you make it restrictive, and then it's up to the operator to open it up15:02
hemnaeharney yah it's odd, but it is how the default admin_metadata options are added15:02
* enriquetaso looking for a coffee 15:02
eharneynot sure what you mean, but it's not obvious to me that admin metadata should show up when looking at volumes here15:03
hemnathere are 2 admin_metadata fields that are added to the metadata output in the API
*** ociuhandu has quit IRC15:03
hemnaas an admin only15:03
eharneywe have a lot of different metadata fields, this kind of thing would benefit from a spec15:03
*** ociuhandu has joined #openstack-cinder15:03
enriquetasoOK, I'm going to start the meeting to discuss the bugs now. :)15:03
hemnayah a spec would be nice, but I'm just hacking around to figure some things out and start a discussion15:04
rosmaitahemna: and possibly making that list a config option15:04
rosmaitaunless we are talking about allowing write access also15:04
hemnathe way it works now, is it shows those 2 fields as part of metadata for a volume, which is odd15:04
rosmaitahemna: this is a good topic for the PTG15:04
hemnaI think separating out admin_metadata and allowing drivers to write stuff in it would be useful.   also making that only viewable by admins.15:05
hemnaok, I'm needing something sooner :P15:05
*** knomura has joined #openstack-cinder15:05
*** zzzeek has quit IRC15:05
hemnaanyway, can we talk about this after the bug discussion that enriquetaso wants to start?15:05
rosmaitatoo late for wallaby, i think :)15:05
eharneyIMO changes for fundamental API things like volume show and volume metadata really need a spec15:05
hemnarosmaita, that's fine.   I'll patch our cinder anyway internally15:05
rosmaitaeharney: ++15:06
hemnaeharney agreed, this is just hacking and trying to figure out the right want to do this.15:06
*** zzzeek has joined #openstack-cinder15:06
enriquetasohemna, thanks, sorry for interrupting your discussion15:06
*** TusharTgite has quit IRC15:06
rosmaitaenriquetaso: when you have some time, you can put up a patch to make this an "official" meeting15:06
hemnaenriquetaso np.15:06
rosmaitasomething like this one:
enriquetasorosmaita, sure, I'll do it15:07
enriquetaso#startmeeting bug-report-cinder15:07
openstackMeeting started Wed Mar 10 15:07:27 2021 UTC and is due to finish in 60 minutes.  The chair is enriquetaso. Information about MeetBot at
openstackUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:07
*** openstack changes topic to " (Meeting topic: bug-report-cinder)"15:07
openstackThe meeting name has been set to 'bug_report_cinder'15:07
enriquetasoI need to think of a more original and cool name for the next meeting.15:07
enriquetasoI couldn't fully prepare the etherpard for today's but I have some bugs to discuss:15:07
enriquetaso#topic #bug_1:"Cinder ignores reader role conventions in default policies"15:07
*** openstack changes topic to "#bug_1:"Cinder ignores reader role conventions in default policies" (Meeting topic: bug-report-cinder)"15:07
hemnaeharney, I'll work on a draft spec15:08
openstackLaunchpad bug 1917795 in Cinder "Cinder ignores reader role conventions in default policies" [Undecided,New]15:08
enriquetasoIn keystone, if I grant someone the reader role on a project [0], they're able to make writable changes in cinder.15:08
enriquetasoOpening this bug to track work for cinder to consume keystone's default read-only `reader` role.15:08
enriquetasoI am worried about doing this as we approach RC time but I think we want this for W, right?15:08
*** m75abrams has quit IRC15:08
rosmaitawell, we'll have to see15:08
rosmaitathis is a bug, so not subject to FF15:09
rosmaitaand it is a real bug, i think15:09
enriquetasoah OK, so I'll set medium importance15:09
enriquetaso#topic bug_2: "Cinder request to glance does not support TLS"15:10
*** openstack changes topic to "bug_2: "Cinder request to glance does not support TLS" (Meeting topic: bug-report-cinder)"15:10
*** knomura has quit IRC15:10
openstackLaunchpad bug 1917797 in Cinder "Cinder request to glance does not support TLS" [Undecided,In progress]15:10
enriquetasothe code does not send cert/key certificates to keystoneauth so if the glance API require TLS the request will fail15:10
eharneythat doesn't sound right15:10
eharneyor at least, needs a lot more detail15:10
enriquetasoThere's a patch for this so far:15:10
enriquetasoso, the bug it's not completed15:11
eharneyso is this really about mTLS again?15:11
eharneyi don't think we can go adding support for security related features without the bugs/patches at least having a clear description of what's going on15:12
enriquetasoas you mention last time this bug assumes that we support strict mTLS from cinderclient which i don't know is a reasonable assumption but worth fix it.15:12
rosmaitawell, i find it concerning that someone who wants to work on this is not being precise about what they are talking about15:12
eharneyme too15:13
eharneybecause it's not clear if we or they understand what the requirements actually are15:13
rosmaitai am going to request a spec15:14
enriquetasoso this could be more a topic for the next PTG if the reported would like to prepare it?15:14
enriquetasooh spec sounds cool15:14
enriquetasook, next one15:14
enriquetaso#topic bug_3:  "Volume backup timeout for large volumes"15:14
*** openstack changes topic to "bug_3: "Volume backup timeout for large volumes" (Meeting topic: bug-report-cinder)"15:14
openstackLaunchpad bug 1918119 in Cinder "Volume backup timeout for large volumes" [Undecided,In progress] - Assigned to kiran pawar (kiranpawar89)15:14
*** __ministry1 has joined #openstack-cinder15:14
enriquetasoThere's a patch for this15:14
enriquetasoUpdate from today's meeting earlier: we are going to push it to Xena15:14
enriquetasojust to double check ^15:15
eharneyi don't think this patch is even going in the right direction the longer i think about it15:15
eharneyboth Gorka and i left some feedback15:15
enriquetasocool, thanks!15:15
enriquetasoLast one15:15
enriquetaso#topic bug_4: "Cinder-backup progress notification has incorrect percentage."15:16
*** openstack changes topic to "bug_4: "Cinder-backup progress notification has incorrect percentage." (Meeting topic: bug-report-cinder)"15:16
openstackLaunchpad bug 1918102 in Cinder "Cinder-backup progress notification has incorrect percentage." [Undecided,New] - Assigned to Jon Cui (czl389)15:16
enriquetasoIn the process of volume backup, Jon Cui found the percentage number was wrong for progress notification.15:16
enriquetasoI wonder if this should be treat as medium o high importance bug.15:16
rosmaitado we give a real percentage or just make one up?15:16
eharneyprobably not high15:16
enriquetasogood question15:17
rosmaitaTIL: ctrl-L in my IRC client completely wipes out the buffer15:17
hemna++++ backup_percent is 1.5832483768463135e-0515:17
eharneythe calculation looks wrong15:17
eharneydividing number of blocks by volume size, those aren't even the same units15:18
rosmaitahemna: sounds like one  of your 2TB backups15:18
hemnarosmaita yah I think it's related to our backup setup.15:18
hemnawe are suffering pretty bad for backups taking ages right now15:18
eharneywe should just mark that as confirmed but it's medium prio at best15:18
eharney(probably low prio given that presumably it's been like that for ages with nobody noticing)15:19
enriquetasosounds good to me15:19
enriquetaso#topic open discussion15:20
*** openstack changes topic to "open discussion (Meeting topic: bug-report-cinder)"15:20
hemnaram usage peaked around 3G for the parallel backup w/ that patch15:20
enriquetasoNot sure if we need a open discussion15:20
eharneyone thing i've wondered about re: bug meetings15:20
hemnawe also have an issue with expiring tokens for the backups taking that long15:20
hemnawhich kills the backup15:21
eharneywe have a process for looking at incoming new bugs, do we need to periodically look at ones that have been stuck in in-progress for a long time?15:21
enriquetasohemna, do you have a link?15:21
openstackgerritSimon Dodsley proposed openstack/cinder stable/victoria: Pure Storage: check volumename length does not exceed maximum
eharneyi think that expiration problem no longer happens if you configure things to use the correct style of keystone tokens etc15:21
hemnaenriquetaso no, it's internal15:21
hemnaeharney correct style ?15:22
enriquetasoeharney, i think it's a good concern, I should probably start looking to older and stuck bugs too15:22
enriquetasothe problem is: how to do the report of 'stuck in in-progress' bugs, send an email?15:23
*** ociuhandu has quit IRC15:23
hemnarosmaita I just checked out cinder.conf and we have teh send_service_user_token = true15:24
*** ociuhandu has joined #openstack-cinder15:24
rosmaitado you have the username, project, etc for the service user?15:24
enriquetaso#action enriquetaso: think a good way to report the 'stuck in in-progress for a long time' bugs15:25
rosmaitaenriquetaso: short term, we could look them over at this meeting and see if they're still relevant15:26
*** ociuhandu has quit IRC15:26
*** ociuhandu has joined #openstack-cinder15:26
enriquetasothat's all I have for today's meeting15:27
rosmaitathanks, enriquetaso15:27
*** openstack changes topic to "The Block Storage Project | |"15:28
openstackMeeting ended Wed Mar 10 15:28:08 2021 UTC.  Information about MeetBot at . (v 0.1.4)15:28
openstackMinutes (text):
rosmaitahemna: there are some troubleshooting tips in that doc, but it could also be a bug :(15:28
enriquetasoI think i ended the meeting too soon15:29
rosmaitano, i don't think the bug has been filed yet!15:29
rosmaitaenriquetaso: this extra meeting is a good idea, we really don't have time during the regular meeting15:30
*** __ministry1 has quit IRC15:30
hemnarosmaita I think I found the issue :)15:31
rosmaitatell me it's not a bug!15:31
hemnaallow_expired_window defaults to 2 days in keystone15:31
hemnafor whatever reason ours set set to 8 hours15:31
hemnaso, it's not a bug....from what I can tell15:32
hemnaso our 20 hour long backups would see that problem from what I can tell.15:32
hemnaso I have another issue that I found yesterday that I didn't file a bug for yet.15:33
rosmaitai think you have a good case to your security team to get that increased15:33
rosmaitathey may not realize that the expired token must be accompanied by a good one, so it's not like you're opening the floodgates15:34
hemnayah I'm notifying the keystone guys in our org15:34
hemnaat init_host time() in the volume manager, the volume manager calls ensure_export() for each volume inuse and available15:35
hemnaseveral drivers have a model_update returned from that call15:35
hemnabut the volume manage ignores it15:35
hemnaour reference driver lvm returns a model update15:36
openstackgerritSimon Dodsley proposed openstack/cinder stable/victoria: Pure Storage FlashArray: Add active/active support
hemnathere are a few more, but I haven't done an extensive search15:36
hemnaworking on a patch for that15:37
openstackgerritSimon Dodsley proposed openstack/cinder master: Add QoS Suport for Pure Storage
hemnaI'll file a bug for it15:42
*** lpetrut has quit IRC15:44
*** ociuhandu has quit IRC15:46
*** ociuhandu has joined #openstack-cinder15:46
*** ociuhandu has quit IRC15:49
*** rratnaka has quit IRC15:49
*** ociuhandu has joined #openstack-cinder15:49
*** ociuhandu has quit IRC16:15
rosmaitaenriquetaso: LGTM!16:16
*** udesale_ has quit IRC16:18
*** ociuhandu has joined #openstack-cinder16:24
*** ociuhandu has quit IRC16:24
*** ociuhandu has joined #openstack-cinder16:25
*** ociuhandu has quit IRC16:29
openstackgerritJay Bryant proposed openstack/python-cinderclient master: client: Stop logging request-id twice in DEBUG
jungleboyjrosmaita:  Looking at cinderclient patches.  Have merged several.  Saw this one:
jungleboyjWhat are your thoughts here?16:32
jungleboyjI can see the reason for the change, but can also see why it was made that way from the start.16:33
rosmaitajungleboyj: yeha, i wasn't sure, just wanted to get eyes on it16:34
*** ociuhandu has joined #openstack-cinder16:35
rosmaitai guess it's not true that "there's nothing they can do about it", the user can request --os-volume-api-version that is the max supported by that server after the first message, and then the messages would go away16:36
*** ociuhandu has quit IRC16:37
*** ociuhandu has joined #openstack-cinder16:37
*** takamatsu has quit IRC16:37
rosmaitai think i agree with your intuition that we don't want to silently change something on the user16:37
rosmaitai downgraded that patch to "easy, but possibly controversial" on the etherpad16:38
openstackgerritVenkata krishna Thumu proposed openstack/cinder master: [SVF] Set volume IOPS based on volume size
GirishChilukuri_@Core reviewers: could you please review these minor changes waiting for second +2 to merge.16:42
jungleboyjrosmaita:  Ok.  Thank you.  That was my thought as well.  I am going to add a comment to the patch accordingly.16:46
rosmaitajungleboyj: sounds good, no sense rushing it16:46
jungleboyjI weighed in.  Will see what others think.16:48
*** venkatakrishnath has quit IRC16:50
*** ociuhandu has quit IRC16:51
openstackgerritFelix Huettner proposed openstack/cinder master: Comply with nas_secure_file_operations for qemu-img info
*** venkatakrishnath has joined #openstack-cinder16:58
openstackgerritWalt proposed openstack/cinder master: ensure_export model_update fix
*** ociuhandu has joined #openstack-cinder17:00
*** rpittau is now known as rpittau|afk17:05
*** GirishChilukuri_ has quit IRC17:09
*** sapd1 has quit IRC17:34
*** takamatsu has joined #openstack-cinder17:38
*** akrpan-pure has joined #openstack-cinder17:40
*** manoj_katari has quit IRC17:52
openstackgerritMerged openstack/python-cinderclient master: Remove more python2 compat code
openstackgerritMerged openstack/python-cinderclient master: Remove install unnecessary packages
openstackgerritMerged openstack/python-cinderclient master: Changed minversion in tox to 3.18.0
*** venkatakrishnath has quit IRC18:02
*** takamatsu has quit IRC18:10
*** akrpan-pure has quit IRC18:22
openstackgerritLuigi Toscano proposed openstack/python-cinderclient master: Use TOX_CONSTRAINTS_FILE
*** tbarron has joined #openstack-cinder18:37
*** e0ne has quit IRC18:48
*** ociuhandu has quit IRC19:12
*** ociuhandu has joined #openstack-cinder19:12
*** ociuhandu has quit IRC19:19
*** ociuhandu has joined #openstack-cinder19:20
*** ociuhandu has quit IRC19:22
*** ociuhandu has joined #openstack-cinder19:23
*** brinzhang0 has joined #openstack-cinder19:33
*** ociuhandu_ has joined #openstack-cinder19:34
*** tosky_ has joined #openstack-cinder19:37
*** benj_- has joined #openstack-cinder19:37
*** dosaboy_ has joined #openstack-cinder19:38
*** whoami-rajat has quit IRC19:40
*** ociuhandu has quit IRC19:42
*** brinzhang_ has quit IRC19:42
*** tosky has quit IRC19:42
*** jv_ has quit IRC19:42
*** benj_ has quit IRC19:42
*** dosaboy has quit IRC19:42
*** Underknowledge has quit IRC19:42
*** pck has quit IRC19:42
*** Underknowledge has joined #openstack-cinder19:43
*** irclogbot_1 has quit IRC19:44
*** jv_ has joined #openstack-cinder19:49
rosmaitasfernand: around?19:49
*** ajitha has quit IRC19:54
openstackgerritEric Harney proposed openstack/python-cinderclient master: Add flake8-import-order extension
eharneythis needed a rebase after another patch landed, if someone would like to re-review ^20:05
eharney(was approved but failed pep8)20:05
rosmaitaeharney: ack20:06
rosmaitahey everyone, here is the promised wallaby features review etherpad:
openstackgerrithemna proposed openstack/cinder-specs master: Add provider_metadata spec
hemnarosmaita ^^20:07
openstackgerritMerged openstack/python-cinderclient master: Doc: Functional Tests in python-cinderclient
hemnahopefully that will shed a little bit of light on what I'm trying to do20:08
hemnabasically, the vmware driver sucks and we have issues with volumes in vcenter20:08
hemnaand need a way to add some metadata to store in cinder to help us move those volumes around inside of vcenter20:09
hemnaI can hack our cinder to allow me to shove some data into the volume_admin_metadata table, or write a spec and some code to do it upstream in a provider_metadata table20:09
hemnathe crux is the admin user needs to see that data from volume fetch API calls.20:10
hemnaI figured adding an upstream feature is 'better' in the long run20:10
hemnaso others can use it. if need be.20:10
rosmaitayeah, i think it is probably something that would be useful20:11
hemnawe basically allow drivers today to shove vendor specific data into the cinder db20:11
hemnabut it's all over the place20:11
hemnakind of a mess really20:11
hemnaa few drivers compact several backend fields into the provider_id for extraction later20:12
hemnapretty much exactly backend metadata on a volume20:12
hemnaI don't think it would take much effort to make this work, as the majority of the lifting is already in place, with all the existing model_update dict return vars supported by the volume manager.20:13
hemnajust have to add the new db table, update the API to return the fields and save/load the fields into the table from the volume object.20:13
rosmaitaso you see this as mostly a read-only API?  the data is written by the drivers?20:14
eharneyi'm adding comments to the spec, but, this is not the first time we've added a field to solve a problem like this20:15
eharneyi think we need to understand how the current ones work before adding yet another20:15
*** tosky_ is now known as tosky20:16
hemna it's not a field, it's a table to allow storing multiple fields20:18
hemnaotherwise I'll just jam it into the volume_admin_metadata table and alter our API20:18
hemnaI need to store multiple fields20:18
eharneyi mean, we already use some of those fields to store multiple items20:18
hemnayes, I called those out20:19
hemnaprovider_id, provider_location, etc.20:19
hemnaI have several uuids I need to store per volume20:19
hemnaoverloading provider_* doesn't work/make sense20:19
eharneyok so we have nothing that explains how to use those and how they differ from each other, or admin_metadata exactly, and we also have driver_initiator_data (not mentioned in the spec) added for similar reasons20:21
eharneyi don't really see how just adding another table is going to help things, we can't just make the mess bigger20:21
eharneywe need some coherent idea of how drivers are supposed to use all of these and how they work20:22
hemnathat's outside the scope of my spec20:23
eharney"jamming" it into the volume_admin_metadata table might be the right answer20:23
hemnaif you want that, then I'll just abandon this and move on20:23
eharneyi don't really think it is20:23
hemnamy spec isn't altering how any driver uses the existing fields.  I'm not trying to rewrite the world.20:24
hemnaif that's what is expected, then I'm out20:24
*** takamatsu has joined #openstack-cinder20:25
hemnaI believe provider_* crap was added due to resistance to adding provider based metadata during the whole replication debacle20:26
hemnaI'm not interested in touching any of that! :P20:26
hemnalist of drivers that return a model_update from ensure_export lvm, solidfire, nimble, netapp, linstordrv, ibm, hpe 3par20:47
eharneyi think LVM doesn't, as i noted20:48
hemnavmware will shortly as well, and it's needed.20:48
hemnawell lvm does return from a call to the target driver20:48
eharneywhich returns nothing AFAICT20:49
*** tosky has quit IRC20:49
hemnaand lvm is the reference implementation, so the volume manager should handle it20:49
*** tosky has joined #openstack-cinder20:49
hemnathe rest of the drivers do return actual values20:49
eharneyor maybe the fact that it returns anything is just a bug and it's not supposed to be doing that20:49
eharneybut why are they returning values?  what needs updating in that situation?20:49
hemnait's the latest information about the export on the storage array20:50
hemnait's a bug that it's ignored20:50
eharneyit's not a bug if the Cinder API is that those calls aren't supposed to change information on the backend20:50
eharneythat's why i'm wondering what is actually changing20:51
hemnacinder is ignoring the updated information about the exports, that's the bug20:52
hemnaall of those drivers are returning an update and are asking the volume manager to save it.   it's not.20:52
hemnathat's a bug20:52
eharneywhat i'm asking is, what is in the update?20:53
hemnasome drivers are asking to update the provider_* fields20:54
hemnasome are updating the chap information20:54
hemnait's different for every backend20:54
hemnathe vmware driver needs to be able to return this as well.   I'm working on a patch against the vmware driver which is related to this and the metadata spec20:55
*** ociuhandu_ has quit IRC20:59
eharneyis it reasonable for a driver to change chap information on ensure_export()?20:59
eharneywhy is that a thing that even happens?20:59
eharneythe fact that we have been ignoring such things seems to indicate that maybe it's not required...20:59
*** ociuhandu has joined #openstack-cinder21:12
hemnaits a bug21:14
hemnajust because it's been ignored doesn't mean it's not a bug21:14
openstackgerrithemna proposed openstack/cinder-specs master: Add provider_metadata spec
*** martinkennelly has quit IRC21:20
*** NobodyCam has quit IRC21:39
*** NobodyCam has joined #openstack-cinder21:40
*** ociuhandu has quit IRC21:42
*** TheJulia has quit IRC21:45
*** TheJulia has joined #openstack-cinder21:45
*** johnsom has quit IRC21:46
*** johnsom has joined #openstack-cinder21:46
*** enriquetaso has quit IRC21:53
hemnaeharney I'm curious why we even have an volume_admin_metadata vs volume_metadata (with an admin bit) table22:01
hemnaI suppose we could update the volume_metadata table and add an admin bit22:01
hemnathen migrate all volume_admin_metadata fields into there.22:02
hemnaand then add provider_metadata dict into that same table with the admin bit set22:02
hemnathe only downside to that, is that it would be ideal to separate those out during volume index/show c-api calls22:02
hemnaso the volumes dict coming back from the rest api would have metadata and provider_metadata fields22:03
openstackgerritMerged openstack/python-cinderclient master: Bump API max version to 3.64
eharneyit looks like it was added because someone wanted a place to stash info related to read-only volume attachments.  not sure what else it might be used for now22:05
hemnavolume create API accepts metadata as well22:05
hemnawhich I believe is user facing key/values stuffed into the volume_metadata table22:06
hemnathe provider_metadata is just metadata coming from the driver/backend22:06
*** vishalmanchanda has quit IRC22:08
*** ociuhandu has joined #openstack-cinder22:10
*** zzzeek has quit IRC22:16
*** zzzeek has joined #openstack-cinder22:18
*** ociuhandu has quit IRC22:18
openstackgerritBrian Rosmaita proposed openstack/python-cinderclient master: Update requirements for wallaby release
*** rcernin has joined #openstack-cinder22:27
*** GirishChilukuri_ has joined #openstack-cinder22:58
GirishChilukuri_@Core reviewers: could you please review these minor changes waiting for second +2 to merge.23:00
*** GirishChilukuri_ has quit IRC23:06
*** GirishChilukuri_ has joined #openstack-cinder23:11
*** GirishChilukuri_ has quit IRC23:18
*** GirishChilukuri_ has joined #openstack-cinder23:30
*** rpittau|afk has quit IRC23:37
*** rpittau|afk has joined #openstack-cinder23:37
*** GirishChilukuri_ has quit IRC23:47

Generated by 2.17.2 by Marius Gedminas - find it at!