Wednesday, 2018-08-15

*** ganso has quit IRC00:07
*** dklyle has quit IRC00:20
*** Nel1x has joined #openstack-cinder00:31
*** brinzhang has joined #openstack-cinder00:47
*** sapd1 has joined #openstack-cinder01:15
*** zhaochao has joined #openstack-cinder01:16
openstackgerritMerged openstack/cinder master: NetApp SolidFire: Fix NetApp SolidFire SSL option
zuljungleboyj: sorry i probably should have asked my question on irc01:28
jungleboyjzul: what was the question?01:29
zuljungleboyj: about the retries01:32
zuljungleboyj: doh...nm....i suck :)01:32
openstackgerritTommyLike proposed openstack/cinder master: Update doc for get-pools API
*** tommylikehu has joined #openstack-cinder01:36
openstackgerritTommyLike proposed openstack/cinder master: Update doc for get-pools API
*** hemna_ has quit IRC01:59
*** hemna_ has joined #openstack-cinder02:09
*** openstack has joined #openstack-cinder02:37
*** ChanServ sets mode: +o openstack02:37
*** cmart has joined #openstack-cinder02:51
*** cmart has quit IRC03:01
*** Nel1x has quit IRC03:11
*** hemna_ has quit IRC03:15
*** cmart has joined #openstack-cinder03:32
*** cmart has quit IRC03:41
*** gnufied has quit IRC03:58
*** gnufied has joined #openstack-cinder04:00
*** eharney has quit IRC04:15
*** gnufied has quit IRC04:26
whoami-rajatHi jungleboyj , smcginnis , i'm confused about Nible Storage CI behaviour here, initially it was failing both tests and currently passing one, i checked out other patches as well, some have both failing some have both passing, and the LOGS aren't available as well.04:26
whoami-rajatThis is the recent behaviour i saw here
whoami-rajatFAILURE and UNSTABLE.04:26
*** eharney has joined #openstack-cinder04:27
*** hemna_ has joined #openstack-cinder05:12
*** pcaruana has joined #openstack-cinder05:12
*** hemna_ has quit IRC05:17
*** e0ne has joined #openstack-cinder06:16
*** alexchadin has joined #openstack-cinder06:59
*** geguileo has quit IRC07:08
*** lennyb_ has joined #openstack-cinder07:34
*** lennyb_ is now known as lennyb07:35
*** lennyb has quit IRC07:35
openstackgerritBoxiang Zhu proposed openstack/python-cinderclient master: Use store_true action for attachment cli
*** rha has quit IRC08:34
*** lennyb has joined #openstack-cinder08:35
*** alexchadin has quit IRC09:00
*** josecastroleon has quit IRC09:00
*** josecastroleon has joined #openstack-cinder09:00
*** andreaf has quit IRC09:16
*** andreaf has joined #openstack-cinder09:17
*** alexchadin has joined #openstack-cinder09:17
*** dirk_ has joined #openstack-cinder09:17
*** dirk has quit IRC09:18
*** wiggin15 has quit IRC09:18
*** dirk_ is now known as dirk09:18
*** gkadam has joined #openstack-cinder09:30
*** rtjure has joined #openstack-cinder09:35
*** josecastroleon has quit IRC10:09
*** zul has quit IRC10:25
*** alexchadin has quit IRC10:59
*** sapd1 has quit IRC10:59
*** alexchadin has joined #openstack-cinder11:00
*** alexchadin has quit IRC11:00
*** alexchadin has joined #openstack-cinder11:01
*** alexchadin has quit IRC11:01
*** alexchadin has joined #openstack-cinder11:01
openstackgerritMatan Sabag proposed openstack/cinder master: Fixing security bug in scaleio driver
*** alexchadin has quit IRC11:02
*** pvc has joined #openstack-cinder11:02
pvchi guys11:02
pvcVolume size extension is not supported by the hypervisor.: ExtendVolumeNotSupported: Volume size extension is not supported by the hypervisor.11:03
*** alexchadin has joined #openstack-cinder11:33
*** alexchadin has quit IRC11:38
*** brinzhang has quit IRC11:39
*** alexchadin has joined #openstack-cinder11:56
*** eharney has quit IRC11:58
*** josecastroleon has joined #openstack-cinder12:00
*** abishop has joined #openstack-cinder12:08
*** gcb_ has joined #openstack-cinder12:19
*** alexchadin has quit IRC12:23
*** alexchadin has joined #openstack-cinder12:30
*** erlon has joined #openstack-cinder12:34
openstackgerritMatan Sabag proposed openstack/cinder master: Fixing security bug in scaleio driver
*** zul has joined #openstack-cinder12:43
*** josecastroleon has quit IRC12:54
*** josecastroleon has joined #openstack-cinder12:54
*** cmart has joined #openstack-cinder12:58
*** gnufied has joined #openstack-cinder13:00
*** eharney has joined #openstack-cinder13:09
*** mvkr has quit IRC13:10
openstackgerritMatan Sabag proposed openstack/cinder master: Fixing security bug in scaleio driver
*** markstur_ has joined #openstack-cinder13:18
openstackgerritSean McGinnis proposed openstack/cinder master: Fix api-ref title levels and index
mnaserhas anyone seen a weird issue where cinder-volume for some reason doesnt send heartbeats to rabbitmq which results in the connection being dropped, or something along those lines..13:28
*** luizbag has joined #openstack-cinder13:29
*** luizbag has quit IRC13:30
mnaser"closing AMQP connection <0.894.0> ( -> - cinder-volume:23478:a2a29ff1-a431-4db5-9d05-f3ffb68d431d): missed heartbeats from client, timeout: 60s"13:30
*** dklyle has joined #openstack-cinder13:31
mnaserand at the same time.. cinder "[b956eec0-15bb-4070-8a23-fd0f3c5e5a8e] AMQP server on is unreachable: <AMQPError: unknown error>. Trying again in 1 seconds."13:32
*** rambo_li has joined #openstack-cinder13:35
*** rambo_li has quit IRC13:38
*** mvkr has joined #openstack-cinder13:40
openstackgerritMichael McAleer proposed openstack/cinder master: VMAX Driver - Initiator retrieval short hostname fix
*** david-lyle has joined #openstack-cinder13:48
*** dklyle has quit IRC13:49
*** david-lyle has quit IRC13:49
*** david-lyle has joined #openstack-cinder13:49
KeithMnemonicjungleboyj: smcginnis is this ready for WF +1 now?
smcginnisKeithMnemonic: I'm not too comfortable with the versioning scheme there being the same as the now divergent later driver versions.14:16
smcginnisKeithMnemonic: It really would be better if there was some indication that this is different.14:16
smcginnisKeithMnemonic: It doens't need to conform to a full semver scheme. Just putting something like 2.0.13a or something to have an indication that this is a variant of 2.0.13 would be better IMO.14:17
*** erlon has quit IRC14:21
KeithMnemonicok i can do that14:21
KeithMnemonicis my "not included" for 2.0.12 ok?14:21
KeithMnemonici want to make sure that is clear14:22
openstackgerritChi Lo proposed openstack/cinder master: Add policy granularity to the encryption API
smcginnisKeithMnemonic: Yeah, that's probably fine to have that called out.14:24
*** erlon has joined #openstack-cinder14:24
openstackgerritKeith Berger proposed openstack/cinder stable/ocata: HPE Lefthand: Fixed cloning operation
KeithMnemoniceharney: i did not quite get that. 2.0.11 is lined up with master, it is only 2.13 that becomes special14:30
KeithMnemonicand whar happnes with 2.14, 2.15, 2.1614:30
eharneywell ocata is 2.0.11, right?14:31
KeithMnemonicthe ocata branch has 2.11, correct14:32
*** alexchadin has quit IRC14:32
eharneyso if we branched the version numbers off for additional fixes, it would split there, not at 2.0.1314:32
KeithMnemonicok but dont i need to reference the original version that change was in14:33
KeithMnemonicso someone looking can line it up14:34
eharneystandard versioning schemes would indicate that 2.0.13a includes everything that's in 2.0.13, but on ocata it doesn't14:35
KeithMnemonicso maybe 2.0.13_special?  the problem is if i do 2.0.11a, then what is the version for the next fix 2.0.11b?14:36
jungleboyjpots:  You around?14:37
eharneythe next fix would be 2.0.11b, yes14:37
KeithMnemonicok i can do that if that is the majority opinion ;-)14:37
KeithMnemonicso with that i remove the 2.0.12 "not included" entry?14:38
KeithMnemonicor just leave it at the end?14:39
eharneythe ¯\_(ツ)_/¯ answer would be to just not use version numbers at all because they aren't really needed in drivers14:39
*** zul has quit IRC14:39
eharneybut since we're already here...14:39
jungleboyjWhere is the tableflipping bot when you need it?14:40
smcginnisMaybe we should document expectations for driver version numbers for backports.14:40
jungleboyjsmcginnis:  I don't remember this being an issue in the past.14:40
openstackgerritHelen Walsh proposed openstack/cinder stable/queens: VMAX  Queens driver - incorrect condition for an empty list.
eharneytypically everyone just follows standard semantic version rules, or doesn't bump the version at all14:41
KeithMnemonicit looks funny leaving the 2.0.12 "not included entry" so i would just remove it if that is ok14:41
openstackgerritKeith Berger proposed openstack/cinder stable/ocata: HPE Lefthand: Fixed cloning operation
openstackgerritHelen Walsh proposed openstack/cinder stable/queens: VMAX Queens driver - incorrect condition for an empty list.
*** david-lyle is now known as dklyle15:01
*** zhaochao has quit IRC15:08
*** masayukig has joined #openstack-cinder15:09
jungleboyjsmcginnis:  eharney  Ok, not going to be the go between here.15:29
jungleboyjLets talk about config changes on backports.15:29
jungleboyjOk.  reading smcginnis  comments in the bug since I am naughty and haven't done that yet.15:31
jungleboyjeharney:  Why do you feel that your solution is not backwards incompatible?15:32
eharneythe goal is to support both the old and new option name15:33
eharneythe goal i had in mind when writing that patch, i mean15:33
smcginnisjungleboyj: It is backwards compatible.15:33
smcginnisThe only issue is that now we would have a config option that would be flagged as deprecated on older branches when it had not actually been deprecated.15:34
jungleboyjOk ...15:34
smcginnisHence my suggestion for the two step process to get the fix in in an easy way to backport, then get it deprecated.15:34
smcginnisWhich was followed by multiple comments asking why it was separated.15:34
eharneyi'm not sure that deprecated_name = flagged as deprecated?  (maybe, honestly not sure)15:35
jungleboyjsmcginnis:  I am sorry.  I hadn't see your comment.  I understand now.15:35
eharneydoesn't it just act as an alias?15:35
smcginnisI believe deprecated_name would cause the use of the old name to trigger warnings in the log that they are using a deprecated config option.15:35
smcginnisI can double check that though.15:35
jungleboyjsmcginnis:  And the concern is that suddenly people who have been using that in previous drivers will get a deprecation warning?15:37
smcginnisWe can't really retroactively deprecate something.15:38
eharneyi think the warning may only happen if you set "deprecated_for_removal"
eharneyi can go try some things to find out though15:39
smcginnisNot sure if this does it -
smcginnisIf not, then I think we are OK using deprecated_name.15:39
jungleboyjSo, the advantage of Eric's change is that people who have enabled this option automatically get coverage for both options.  Right?15:40
jungleboyjBut we are backporting a deprecation which seems kind-of wrong.15:40
eharneythat is the theory, i'm hoping it actually works that way15:40
eharneyi think deprecation is a misnomer for this case... it's a config parameter rename15:41
jungleboyjeharney: With the two option approach then the user will get the creation error until they add the new option or fix their backend config.15:42
smcginnisNo, same behavior with the two opt option.15:43
smcginnisIt just needs to check both opts instead of just one.15:44
eharneyok, just using deprecated_name does issue a warning at startup15:44
eharneyso what's the ideal case here that we actually want?15:45
jungleboyjNo comment.15:45
smcginnisWhat is the problem with the two step process I recommended?15:45
jungleboyjIt is fine.  It is the more correct way of doing this.15:46
*** mvkr has quit IRC15:46
jungleboyjI don't like adding more config options but more importantly we need to get this fixed.15:46
eharneyit looks like it works fine15:46
smcginnisYeah, it's kind of a hokey config dance, but it makes it clean to backport and gets us ready to remove the option going forward. We still end up where eharney's patch was going.15:48
eharneyi'm a bit unsure about the logic in the patch15:49
eharneyif you set allow_non_padded_volumes to True but allow_non_padded_thick_volumes to False, it passes the check15:50
jungleboyjOk.  So, if either of the options is set to trye it will say creation is safe.15:50
smcginniseharney: Hmm, yeah. Looks like that part needs to be a little smarter.15:50
smcginnisAnd of course the unit tests need to be addressed.15:51
jungleboyjAnd there needs to be a release note.15:51
*** ganso has joined #openstack-cinder15:51
eharneythe same concerns about compatibility for backports come up with changing the logic though :/15:51
eharney(i think)15:52
smcginnisThat part I think we are OK on since we are addressing a security issue and it will be called out in the release notes.15:52
*** itlinux has joined #openstack-cinder15:52
smcginnis"Why isn't this working now? Oh, because it was bad." :)15:52
eharneyif someone has written tooling to set the old option, and we require the new option to be set anyway, then we introduce a compat issue15:53
*** pvc_ has joined #openstack-cinder15:53
pvc_im currently using ceph backend for my voluems15:53
pvc_but when i extending it on a attached state it have an error15:53
eharneybut if we only allow one to be set -- then it's confusing what it means for thin volumes15:53
pvc_Hypervisor is not supported the volume extension15:53
eharneyonly require*15:53
pvc_any help guys?15:53
pvc_my instance cant get the new size of the ceph volume :(15:53
smcginniseharney: You're saying if someone has set tooling to allow them to do the unsecure way, now it changes that?15:54
pvc_@smcginnis hi can you help me15:54
*** FracKen has joined #openstack-cinder15:54
smcginnispvc_: A little busy at the moment.15:54
smcginnispvc_: But I don't think attached extend is supported with ceph.15:54
eharneyif we try to preserve compatibility, we end up in a situation where allow_non_padded_volumes would apply to thick volumes, even if the old allow_non_padded_thick_volumes option is set to False15:54
jungleboyjWith how it is now we don't have an issue but it is confusing as to what it really means.15:54
pvc_So sad to hear that :(15:55
eharneywhich i guess works, but is kind of confusing15:55
smcginniseharney: No, it should still be defaulted to not allow it even though that is a change in behavior. Since it's fixing a security issue.15:55
eharneyi agree they should default to False, the question is what happens when you only set one of the options15:56
jungleboyjsmcginnis:  That makes more sense to me.15:56
jungleboyjI think we should probably throw an error if only one is set so they are aware they had an issue.15:57
*** cmart has quit IRC15:57
pvc_anyone using ceph here as a volume?15:57
jungleboyjI realize that that kind-of goes against what I said earlier.15:57
eharneythat sounds a lot like a deprecation15:57
jungleboyjBut this is a security thing.15:57
jungleboyjjbernard:  ^^^ Can you address the Ceph question?15:58
pvc_Hi jbernard can help me?15:58
jungleboyjThe support matrix has Attached Extend as supported but maybe that is wrong?  If so please let me know so I can fix the matrix.15:58
pvc_are you referring to me jungleboyj?15:59
jungleboyjpvc_: Yes.15:59
pvc_when my cinder backend is LVM I can extend its size on a attached state15:59
eharney is in progress for ceph extend16:00
eharneyneeds testing/review16:00
pvc_but when im using ceph rbd volume backend, my instance cant get it even if the horizon change its size.16:00
jungleboyjOk, I should fix the existing infor for that then.16:01
pvc_i see eharney thank you for this16:01
eharneyso it seems like we may need a warning message if "allow_thick" is False but "allow" is True saying that this may be doing something unexpected?16:02
eharney(given the context that the original security notice said you could avoid this problem by using thin)16:02
smcginniseharney: That might be the best approach.16:02
eharneyand just keep the logic that's in the patch currently for how the options actually work?16:02
pvc_hi eharney is that review applicable for Kolla Openstack deployment? thank you for your help16:03
eharneypvc_: it's part of supporting it in cinder/nova16:11
pvc_thank you, so for now the on fly extension of ceph rbd is not yet supported?16:13
*** sayalilunkad has quit IRC16:16
*** sayalilunkad has joined #openstack-cinder16:20
*** cmart has joined #openstack-cinder16:30
*** sayalilunkad has quit IRC16:32
*** clayc has joined #openstack-cinder16:33
jungleboyjsmcginnis: eharney  I need to go get my boys from their Mom's.16:37
jungleboyjWe need to land on an answer for the config options though so we can resolve the issue.16:37
smcginnisI commented on the patch that they should add a release note.16:39
KeithMnemoniceharney: hopefully this does the trick
smcginnisIf we are good with the two step plan, maybe good to follow up on there since right now there's still some dissent as the last comments.16:39
smcginnisAnd the email thread.16:39
jungleboyjOk.  Will look like when I get back.16:42
*** pvc_ has quit IRC16:45
*** sayalilunkad has joined #openstack-cinder16:47
*** e0ne has quit IRC16:47
*** sayalilunkad has quit IRC16:51
*** sayalilunkad has joined #openstack-cinder16:52
*** openstackstatus has joined #openstack-cinder16:56
*** ChanServ sets mode: +v openstackstatus16:56
mnaserdoes anyone know if this job is actually running in cinder ci ?
openstackgerritHelen Walsh proposed openstack/cinder master: VMAX Driver - Initiator retrieval short hostname fix
mnaserbecause that change is def. failing in OSA17:05
*** gkadam has quit IRC17:05
mnaser with this stacktrace17:05
*** zul has joined #openstack-cinder17:08
mnaserlooks like its manifesting because of this failure --
*** sayalilunkad has quit IRC17:09
mnaserit looks like cinder_tempest_plugin is not actually ran anywhere?17:16
*** gkadam has joined #openstack-cinder17:20
*** eharney has quit IRC17:26
*** dustins has joined #openstack-cinder17:33
*** markstur_ has quit IRC17:34
*** cmart has quit IRC17:36
*** markstur has joined #openstack-cinder17:41
*** markstur has quit IRC17:45
smcginnismnaser: Hmm, that would be concerning. I thought it was being pulled in, but it would appear otherwise.17:46
mnasersmcginnis: i checked a few recent commits and it's def not being ran17:46
smcginnisOK, we must be missing something somewhere that pulls that in.17:46
mnaserrecent run, dont see it there so not sure if that is where it should be17:47
smcginnisLooks like we may need a devstack change to pull it in for Cinder that is dependent on the fix for whatever is failing right now.17:48
mnaserwell, i'd add it first and see if it fails at your side17:48
mnaserit could be an OSA specific thing, though i doubt it17:49
smcginnisYeah, I would blame Cinder first until proven otherwise.17:49
*** cmart has joined #openstack-cinder17:52
*** e0ne has joined #openstack-cinder17:56
*** mvkr has joined #openstack-cinder17:56
*** markstur has joined #openstack-cinder17:57
*** zul has quit IRC18:18
*** gkadam has quit IRC18:19
*** eharney has joined #openstack-cinder18:19
jungleboyjsmcginnis: eharney  I have put my $0.02 in the patch.  Really need Matan to address this and get a fix out there.18:40
*** ganso has quit IRC18:41
eharneyjungleboyj: so what changes still remain there?18:52
jungleboyjNeed a release note and I think they need to reconsider the logic for two config options eharney18:53
eharneysounds good18:53
*** markstur has quit IRC18:57
openstackgerritMatan Sabag proposed openstack/cinder master: Fixing security bug in scaleio driver
hemnaanyone mess with gerrit searching much ?19:05
smcginnishemna: A bit. What are you trying to do?19:07
hemnafind all driver patches19:07
smcginnismnaser: Doesn't look like any of the main projects get their tempest plugins pulled in to tempest-full. Just specific jobs.19:07
hemnafile:"^*drivers*" doesn't seem to worky19:08
smcginnishemna: Open ones?19:08
hemnaopen or closed really19:08
mnasersmcginnis: yeah, that's what i noticed from codesearch.o.o19:08
mnaserhemna:* seems to work?19:09
smcginnishemna: Or
hemnaI guess *drivers* wasn't good enough19:10
smcginnisFile instead of files for some reason.19:10
*** markstur has joined #openstack-cinder19:11
hemnasmcginnis, thanks man19:12
smcginnishemna: No problem. It's definitely not intuitive.19:12
*** markstur has quit IRC19:14
*** markstur has joined #openstack-cinder19:14
*** openstackgerrit has quit IRC19:19
itlinuxhello guys, back to the IRC channel.. strange thing in Pike, and maybe it is a bug.. but here it is.. I tried every combo and set a default location but it never holds.. so now I have an az-1 and az2 and default to az-2 but for some reasons it still go to nova az which I do not have enabled and it's off so if I just do cinder create 1 , it looks like it is looping and sometimes it does create the volume the 1st time sometimes the 2time and19:21
itlinux sometimes the 3rd time.. kinda frustr.. on this .. any suggestions on really what I should be checking since I restarted all cinder services and the pacemaker just in case after every change. Thanks19:21
jbernardpvc: i don't think so, but ill add it to my list for stein19:30
hemnahmm, finding this info is going to be harder than I thought.19:33
hemnaso I'm trying to find the total number of driver reviews per release vs non driver reviews19:33
hemnausing branch alone doesn't cut it19:33
hemnasince almost all new features and reviews were against master at one point.19:33
hemnamaybe reviews before the date when the stable/queens tag was added and after stable/ocata will give me the queens release ?19:38
*** openstackgerrit has joined #openstack-cinder19:41
openstackgerritSean McGinnis proposed openstack/cinder master: [2/2] Fix mock for sleep mocks in volume drivers tests
smcginnishemna: queens and pike19:42
*** _alastor_ has joined #openstack-cinder19:42
smcginnisThere might be an easier way to query directly in git.19:42
smcginnisSorry, need to drop for a minute.19:43
hemnaman git pull from wifi on a plane at 35,000' takes a while19:45
mnaserlatency is rough19:47
*** e0ne has quit IRC20:03
hemnaok that seems to be working, but the query only returns 500 results at a time.20:08
hemnaI really only want a total count, not the actual individual reviews20:09
*** pocketprotector has quit IRC20:24
*** pocketprotector has joined #openstack-cinder20:26
*** markstur has quit IRC20:27
*** markstur has joined #openstack-cinder20:29
*** markstur has quit IRC20:33
*** markstur has joined #openstack-cinder20:35
*** markstur has quit IRC20:40
jungleboyjhemna: What are you trying to get the data for?20:43
*** markstur has joined #openstack-cinder20:47
*** markstur has quit IRC20:52
*** markstur has joined #openstack-cinder20:56
*** markstur has quit IRC20:56
*** masayukig has quit IRC21:12
openstackgerritSean McGinnis proposed openstack/cinder master: ScaleIO: Disable volume creation without padding
*** rcernin has joined #openstack-cinder21:29
openstackgerritMerged openstack/python-brick-cinderclient-ext stable/rocky: Update UPPER_CONSTRAINTS_FILE for stable/rocky
*** KeithMnemonic has quit IRC21:44
*** erlon has quit IRC21:50
openstackgerritMerged openstack/cinder stable/ocata: HPE Lefthand: Fixed cloning operation
openstackgerritMerged openstack/cinder master: Update document for metadata show API
openstackgerritMerged openstack/cinder master: Default functional tests to use v3 API
openstackgerritMerged openstack/cinder stable/rocky: Update UPPER_CONSTRAINTS_FILE for stable/rocky
*** FracKen has left #openstack-cinder22:17
*** masayukig has joined #openstack-cinder22:18
*** itlinux has quit IRC22:22
*** cmart has quit IRC22:26
*** imacdonn has quit IRC22:38
*** imacdonn has joined #openstack-cinder22:38
*** erlon has joined #openstack-cinder22:52
openstackgerritSean McGinnis proposed openstack/cinder stable/pike: Fix _attachment_reserve to not allow attaching an invalid status volume
*** rcernin has quit IRC23:18
*** rcernin has joined #openstack-cinder23:19
openstackgerritMerged openstack/os-brick stable/rocky: Update UPPER_CONSTRAINTS_FILE for stable/rocky
*** dustins has quit IRC23:27
openstackgerritMerged openstack/python-cinderclient stable/rocky: Update .gitreview for stable/rocky
openstackgerritMerged openstack/python-cinderclient stable/rocky: Update UPPER_CONSTRAINTS_FILE for stable/rocky

Generated by 2.15.3 by Marius Gedminas - find it at!