Thursday, 2017-01-05

« Wednesday, 2017-01-04 Index Friday, 2017-01-06 »
*** gouthamr has quit IRC00:04
*** tommylikehu_ has joined #openstack-manila00:11
*** tommylikehu_ has quit IRC00:15
*** absubram has quit IRC00:20
*** tommylikehu has joined #openstack-manila00:50
*** dsariel has quit IRC01:05
*** tuanluong has joined #openstack-manila01:12
*** tuanluong_ has joined #openstack-manila01:12
*** furlongm has quit IRC01:36
*** furlongm has joined #openstack-manila01:36
*** xinyanzhang has quit IRC02:03
*** xinyanzhang has joined #openstack-manila02:06
*** wlhc has joined #openstack-manila02:18
openstackgerritzhongjun proposed openstack/manila: Enable IPv6 in manila(Implement in generic driver)  https://review.openstack.org/41616202:42
*** wlhc_ has joined #openstack-manila02:44
openstackgerritRyan Liang proposed openstack/manila: Unity/VNX Driver: Rename driver options  https://review.openstack.org/41507902:45
*** wlhc has quit IRC02:47
openstackgerritBen Swartzlander proposed openstack/manila: Modify test filter to run all backend tests  https://review.openstack.org/41681802:52
*** harlowja has quit IRC03:05
*** rhagarty has quit IRC03:05
*** kaisers1 has quit IRC03:05
*** rhagarty has joined #openstack-manila03:06
*** kaisers1 has joined #openstack-manila03:06
openstackgerritzhongjun proposed openstack/manila: Enable IPv6 in manila(network plugins and drivers)  https://review.openstack.org/40677603:07
*** gouthamr has joined #openstack-manila03:20
*** wlhc_ has quit IRC03:28
*** wlhc_ has joined #openstack-manila03:28
*** bswartz_ has joined #openstack-manila04:35
*** bswartz_ has quit IRC04:35
*** furlongm has quit IRC04:35
*** furlongm has joined #openstack-manila04:36
*** gouthamr has quit IRC04:43
openstackgerritMerged openstack/manila: Migration Data Check fixes  https://review.openstack.org/41669006:13
*** gouthamr has joined #openstack-manila06:14
*** breitz_ has quit IRC06:40
*** nkrinner_afk is now known as nkrinner06:40
*** breitz has joined #openstack-manila06:41
*** pcaruana has joined #openstack-manila07:05
openstackgerritmelissaml proposed openstack/python-manilaclient: Removes unnecessary utf-8 encoding  https://review.openstack.org/41686607:08
*** lpetrut has joined #openstack-manila07:16
openstackgerritzhongjun proposed openstack/manila: Enable IPv6 in manila(documentation)  https://review.openstack.org/41687007:16
*** jprovazn has joined #openstack-manila07:21
*** openstackgerrit has quit IRC07:50
*** gouthamr has quit IRC07:53
*** zhugaoxiao has quit IRC07:54
*** gouthamr has joined #openstack-manila07:54
*** zhugaoxiao has joined #openstack-manila07:54
*** lpetrut has quit IRC07:58
*** gouthamr has quit IRC08:06
*** dsariel has joined #openstack-manila08:47
*** igajsin has quit IRC08:48
*** wlhc_ has quit IRC08:56
*** a-pugachev has joined #openstack-manila09:20
*** openstackgerrit has joined #openstack-manila09:36
openstackgerritzhongjun proposed openstack/manila: Enable IPv6 in manila(network plugins and drivers)  https://review.openstack.org/40677609:36
*** lpetrut has joined #openstack-manila09:48
*** ganso has joined #openstack-manila10:01
*** openstackstatus has quit IRC10:13
*** openstack has joined #openstack-manila10:17
*** furlongm has quit IRC10:48
*** furlongm_ has joined #openstack-manila10:48
*** furlongm has joined #openstack-manila11:12
gansobswartz, tbarron: ping11:13
*** furlongm_ has quit IRC11:13
*** lpetrut has quit IRC11:29
openstackgerritVitaliy Levitski proposed openstack/manila: Add MapR-FS native driver  https://review.openstack.org/35003511:30
*** lpetrut has joined #openstack-manila11:31
openstackgerritVitaliy Levitski proposed openstack/manila: Add MapR-FS native driver  https://review.openstack.org/35003511:54
*** raissa has quit IRC11:55
*** kaisers has quit IRC11:55
*** csaba has quit IRC11:55
*** smcginnis has quit IRC11:55
*** smcginnis has joined #openstack-manila11:55
*** raissa_ has joined #openstack-manila11:56
*** kaisers has joined #openstack-manila11:56
*** csaba has joined #openstack-manila11:56
*** JoseMello has joined #openstack-manila12:05
openstackgerritMerged openstack/python-manilaclient: Removes unnecessary utf-8 encoding  https://review.openstack.org/41686612:06
openstackgerritJan Provaznik proposed openstack/puppet-manila: Allow setting default_share_type config option  https://review.openstack.org/41695612:12
*** tuanluong_ has quit IRC12:14
*** tuanluong has quit IRC12:14
*** catintheroof has quit IRC12:38
*** catintheroof has joined #openstack-manila12:44
*** jcsp has quit IRC12:54
*** jcsp has joined #openstack-manila12:55
*** catintheroof has quit IRC12:57
*** catintheroof has joined #openstack-manila12:58
*** catintheroof has quit IRC13:02
*** catintheroof has joined #openstack-manila13:04
*** catintheroof has quit IRC13:11
*** tpatzig_ has joined #openstack-manila13:25
*** carthaca_ has joined #openstack-manila13:25
*** dgonzalez_ has joined #openstack-manila13:25
*** david_1 has joined #openstack-manila13:25
*** carthaca_ has quit IRC13:27
*** tpatzig_ has quit IRC13:27
*** dgonzalez_ has quit IRC13:27
*** david_1 has quit IRC13:27
*** jprovazn has quit IRC13:31
*** tommylikehu_ has joined #openstack-manila13:35
*** tommylikehu_ has quit IRC13:35
*** tommylikehu_ has joined #openstack-manila13:35
tommylikehu_ping ganso13:36
tommylikehu_ganso: could you take some time on this patch? https://review.openstack.org/#/c/366614/13:47
*** catintheroof has joined #openstack-manila13:47
*** cdelatte has joined #openstack-manila13:53
*** jprovazn has joined #openstack-manila13:54
*** porrua has joined #openstack-manila14:00
*** xyang1 has joined #openstack-manila14:18
*** gaurangt has joined #openstack-manila14:24
bswartzganso: pong14:27
*** gaurangt has left #openstack-manila14:30
*** dustins has joined #openstack-manila14:35
*** eharney has joined #openstack-manila14:37
gansobswartz: Hi Ben14:39
bswartz.o/14:39
gansobswartz: I would like to talk about something one of my coworkers found, a possible bug14:39
gansobswartz: I remember that we had a similar discussion in the past, but I don't remember the conclusion14:39
bswartzis it something for the weekly meeting/14:40
gansobswartz: I don't think so, sounds simple14:40
bswartzor something we can discuss now14:40
bswartzok14:40
tommylikehu_bswartz: real ganso comes :)14:40
gansobswartz: basically, if a tenant A creates a share, should a tenant B be able to obtain any details of that share if it knows its ID?14:40
gansobswartz: if he/she* knows its ID14:41
bswartzganso: if it's marked public, then yes, otherwise no14:41
gansobswartz: it is not marked public14:41
bswartzthe randomness of the UUIDs is not a security mechanism14:41
bswartzalthough it does help14:42
gansobswartz: if that's the case, is this a critical security flaw?14:42
gansobswartz: so, is manila committed to not allowing a tenant to find resources of other tenants by querying the API with different UUIDs?14:43
gansobswartz: or, if the resource UUID leaks somehow, are we committed to not allowing any operation to be performed on a share that does not belong to that tenant?14:43
bswartzganso: those 2 question sound like the same thing14:45
bswartzI would say yes to both14:45
gansobswartz: ok, so we have this kind of bug, and it seems like a security issue to me14:46
*** absubram has joined #openstack-manila14:58
bswartzganso cknight toabctl gouthamr: # openstack-meeting-alt15:02
gansobswartz: thanks15:03
*** chlong has joined #openstack-manila15:05
*** cknight has joined #openstack-manila15:08
*** alyson_ has joined #openstack-manila15:11
*** sapcc-bot has quit IRC15:12
*** sapcc-bot has joined #openstack-manila15:13
*** catintheroof has quit IRC15:32
*** nkrinner is now known as nkrinner_afk15:59
bswartztbarron: over here16:01
tbarronhi16:01
tbarronso what I'm trying to figure out is16:01
bswartztbarron: so it seems as vponomaryov set it up, "backend" is the tests that actually exercise the backends and are the ones we really want to run for each different driver16:01
*** timcl has quit IRC16:01
bswartz"api" are the tests that could run with manila-share not even started16:02
tbarronok16:02
tbarronok16:02
bswartzthe funny one is "api_with_backend" which I was wrong about16:02
bswartzvponomaryov used that for tests that require manila-share to be running, but don't actually exercise anything interesting in the driver, and thus the dummy driver is perfectly suitable for those tests16:03
tbarronok, so only "backend" should be running always16:03
bswartzwe should probably rename that tag, of make it much clearer what it means16:03
openstackgerritMauricio Lima proposed openstack/puppet-manila: Add support to configure Hitachi NAS Platform driver with Manila shares.  https://review.openstack.org/39575916:03
tbarronbut you are thinking maybe more tests need to be included in "backend" - got it16:03
bswartztbarron: well in akerr's case, he was adding new tests, and definitely mistagged his new tests because he didn't understand the tags16:04
bswartzI'm not sure how many existing tests are miss-tagged -- it could be zero16:04
tbarronk16:04
dustinsbswartz: Is there a list of possible tags and what they correspond to?16:05
bswartzmaybe we only have 50 tests worth running on each driver job16:05
bswartzdustins: https://github.com/openstack/manila/blob/master/manila_tempest_tests/tests/api/base.py#L4616:05
bswartztbarron: this is where the scenario tests come in16:05
tommylikehu_hope you can have some time on this simple patch: https://review.openstack.org/#/c/366614/  :)16:06
dustinsbswartz: So we do, but it's in source and not really clear as to what differentiates one from another16:06
*** tommylikehu_ has quit IRC16:06
bswartzscenario tests are like backend tests insofar as they actually tests that the backend works, but traditionally we haven't had many scenario tests16:06
bswartzdustins: true16:06
bswartzvponomaryov has been working to expand backend test coverage by adding more scenario tests, which is exactly the right way to do it, as long as CI systems run those tests too16:07
bswartzif we find any tests tagged incorrectly let's file bugs and fix tem16:07
bswartzthem16:07
tbarronk16:08
bswartztechnically a bug isn't needed to fix a test16:08
dustinsAnd perhaps add something to the devref to explain which tags to use where16:08
bswartzbut it can help with tracking the work16:08
bswartzdustins: +116:08
bswartzyeah I actually discussed this with vponomaryov last year and it was still unclear to me, so people who don't know will definitely be confused16:08
gansobswartz: is Valeriy on vacation?16:08
dustinsSince looking at the API with backend tag makes my brain hurt16:08
bswartzganso: until 10 Jan16:09
tbarronbswartz: BTW, CI coverage recently looks pretty poor http://ci-watch.tintri.com/project?project=manila16:09
gansobswartz: ok thanks16:09
bswartztbarron: we have a plan to address that -- just need to spend time on it16:09
tbarronbswartz: sure, understood16:10
*** catintheroof has joined #openstack-manila16:32
*** jprovazn has quit IRC16:32
*** porrua has quit IRC16:45
*** gouthamr has joined #openstack-manila16:46
*** pcaruana has quit IRC16:58
openstackgerritMerged openstack/manila: Trivial fix LOG.exception issues  https://review.openstack.org/41268817:07
*** porrua has joined #openstack-manila17:16
*** porrua_ has joined #openstack-manila17:23
*** porrua has quit IRC17:26
bswartztbarron: this is what I have in mind: https://github.com/torvalds/linux/blob/master/MAINTAINERS#L8517:31
tbarronbswartz: I like it.17:32
*** lpetrut has quit IRC17:37
*** a-pugachev has quit IRC17:48
openstackgerritGoutham Pacha Ravi proposed openstack/manila: NetApp cDOT: Add Intra-vserver migration support  https://review.openstack.org/41378317:54
openstackgerritMauricio Lima proposed openstack/puppet-manila: Add support to configure Hitachi NAS Platform driver with Manila shares.  https://review.openstack.org/39575917:56
openstackgerritRodrigo Barbieri proposed openstack/manila: Mountable snapshots scenario tests  https://review.openstack.org/41200117:58
openstackgerritMiriam Yumi proposed openstack/manila: Add mountable snapshots support  https://review.openstack.org/34552618:02
openstackgerritRodrigo Barbieri proposed openstack/manila: Mountable snapshots scenario tests  https://review.openstack.org/41200118:09
gouthamrganso: any issue with the HSP/HNAS CI? Asking because of: https://review.openstack.org/#/c/414180/18:14
gansogouthamr: Hi Goutham! It has been fixed today, I will trigger rechecks18:14
gouthamrganso: nice, thank you..18:15
openstackgerritMiriam Yumi proposed openstack/manila: Add mountable snapshots support  https://review.openstack.org/34552618:21
openstackgerritClinton Knight proposed openstack/manila: Implement share revert to snapshot  https://review.openstack.org/34050218:31
*** dustins has quit IRC18:33
*** dustins has joined #openstack-manila18:34
cknightganso: ^^  Thanks for pointing out those comments I missed.18:34
gansocknight: reading18:35
openstackgerritTom Barron proposed openstack/manila: Remove NovaNetworkPlugin  https://review.openstack.org/41045018:36
cknightganso: There's definitely a corner case around reverting shares that have been shrunk, but I don't think we can do anything to prevent that.  At least I detect the change in size and update the quotas.18:37
gansocknight: I think this only happens because snapshots are allowed to be managed without a proper size18:39
gansocknight: from your side I believe there is nothing you can do, the design of managed_snapshots should be improved to address this IMO18:40
gansocknight: regarding the access rules, there are drivers that are not able to read the backend access rules. I don't remember which ones from the top of my mind.18:40
cknightganso: Did you have a suggestion for how to address that?  I haven't audited the drivers for which ones fallback to the share size when managing snapshots.18:41
cknightganso: I can peek at that now.18:41
cknightganso: For the access rule issue, it would have to be a driver that can't read & reapply its own rules, *and* cannot preserve rules while reverting to a snapshot.  That's gotta be a small list, if any.18:42
gansocknight: From top  of my mind I remember Huawei driver does that, I think there was another one I don't remember now18:43
cknightganso: I guess I'm not terribly inclined to accommodate such a limited backend.18:43
gansocknight: since the problem is unrelated to your patch, just aggravated by it, I would suggest leaving it open to be fixed when Huawei guys face this issue, if they ever implement revert-to-snapshot18:44
cknightganso: Agreed.  I look at guys like the Ceph team who made changes to their filesystem to work better with Manila, which suggests other products could do the same if it's really important to them to support a feature.18:45
bswartzwe can't get dragged down by the lowest common denominator18:46
gansocknight: there are easy ways to fix it, my driver has the same issue, but our approach was to require the size, instead of just get the share's18:47
cknightganso: good idea.  A solution could be to remove the fallback size logic and make it required.18:48
gansocknight: not requiring it makes it more likely for the admin to manage a snapshot without specifying a size because he/she does not know the risks18:48
gansocknight: but for backends that can detect the size (I think NetApp does that), it is an unecessary requirement18:49
cknightganso: What I meant was that it is required by the share manager.  The driver could figure it out (as NetApp does), or require it as an extra parameter if necessary.18:50
openstackgerritMiriam Yumi proposed openstack/manila: Add mountable snapshots support  https://review.openstack.org/34552618:51
gansocknight: oh yes. That would be great18:51
* bswartz scrolls back and reads18:52
bswartzare you talking about how to compute the snapshot size at snapshot creation time or at revert time?18:53
gansobswartz: at snapshot_manage time18:53
bswartzganso: why can't it default to the share size at that time?18:53
cknightganso: Just 4 drivers manage snapshots now: ZFSonLinux, Generic, NetApp, HNAS.18:53
gansobswartz: because the share could have been shrunk18:54
gansocknight: not Huawei's? o_O18:54
bswartzyou mean manage the share, then shrink the share, then manage the snapshot?18:54
bswartzI thought that wasn't allowed18:54
gansocknight: manila/share/drivers/huawei/huawei_nas.py:189, manila/share/drivers/huawei/v3/connection.py:954, manila/share/drivers/huawei/base.py:8118:55
cknightganso: You're right.  I wonder why the devref doc doesn't list it.18:56
gansobswartz: the share may have been shrunk before being managed, and a previously snapshot may exist18:56
bswartzganso: you're saying that the storage device might have its own idea of how big the share and it snapshots are, and they might not agree?18:57
bswartzbecause in NetApp land snapshots have no size18:58
bswartzin fact the "size" of a share is really just a quota for the max amount of data you can write before we return ENOSPC18:59
bswartzand that size is totally unrelated to how many bytes we actually stored19:02
bswartzbecause compression/dedupe/efficient snapshots can all reduce the amount of space required to store a user's data19:02
markstur_so much voodoo  :)19:03
gansobswartz: for HNAS it is a quota as well19:03
gansobswartz: but it gets messed up when we try to apply a quota that is lower than the actual amount of data19:03
bswartzmarkstur: this was voodoo 15 years ago -- I'm shocked that not everything works this way now19:03
bswartzganso: yeah I'm pretty sure we don't allow that either19:04
markstur_I'm just going to pretend I'm too young to feel the same way19:04
gansobswartz: that scenario would be that the storage had already a share, a snapshot taken, and shrunk, then the admin manages the share with the current size, and then manages the old snapshot, and uses it ro revert19:04
markstur_FYI -- the HPE 3PAR can shrink the quota to less than the share size, but then it won't let you add a byte to it.19:05
gansobswartz: *to revert19:05
bswartzwhere things get complicated on NetApp is that we account for space in such a way that snapshot data subtracts from your available space, which means you can run out of space a lot faster than you'd expect19:06
markstur_You kind of keep your squaters rights to the space you have.  Not perfect for accounting, but works nicely otherwise19:06
bswartzganso: yeah we can't get into that situation because the attempt to shrink smaller than the snapshot size would have failed19:06
markstur_bswartz: See I was kind of skeptical about that zero size snapshot but I'm trying not to throw stones19:06
bswartzmarkstur: it's all a matter of how you account for used space19:07
bswartzmarkstur: if your system is able to deduplicate data across tenants then there is no right answer for who is "consuming" the space on disk19:08
markstur_Yeah and my mention of accounting ^ was meaning that if it was really really important that no one was using more than their shrunken size then that squatter thing could theoretically bite someone.19:08
gansomarkstur_: if your snapshot is efficient, then if has just been taken, the actual size is 0, despite having loads of data... so you would need to calculate it properly19:08
gansobswartz: the manila API blocks shrinking if share has a snapshot with bigger size? I have never tried that19:09
bswartzganso: no our storage controlled would have failed the resize operation19:10
markstur_ganso: Sure snapshots can be zero data size, but not always19:10
markstur_although if the space comes out of the share itself then we wouldn't want to double count it19:10
gansobswartz: oh ok19:10
bswartzthe way snapshots work on NetApp is that initially they consume zero space, because the bytes are 100% shared with the active file system -- every time you delete or overwrite data though the snapshot size increases19:11
bswartzbecause the snapshot grows every time you delete a file, you can never get in the situation where the share size is less than the sum of all of the snapshots plus the active file system19:14
*** harlowja has joined #openstack-manila19:16
*** dsariel has quit IRC19:19
*** lpetrut has joined #openstack-manila19:19
*** gouthamr has quit IRC19:39
*** gouthamr has joined #openstack-manila19:40
*** dsariel has joined #openstack-manila19:44
*** mtanino has joined #openstack-manila19:57
*** chlong has quit IRC20:03
*** cknight has quit IRC20:38
*** catintheroof has quit IRC20:45
*** eharney has quit IRC20:53
openstackgerritTom Barron proposed openstack/manila: Remove NovaNetworkPlugin  https://review.openstack.org/41045020:53
*** porrua_ has quit IRC21:01
*** porrua has joined #openstack-manila21:02
*** alyson_ has quit IRC21:07
openstackgerritTom Barron proposed openstack/manila: Remove NovaNetworkPlugin  https://review.openstack.org/41045021:13
*** porrua has quit IRC21:16
*** cdelatte has quit IRC21:19
*** JoseMello has quit IRC21:25
*** sage_ has quit IRC21:32
*** sage_ has joined #openstack-manila21:34
*** eharney has joined #openstack-manila21:41
*** darrenc is now known as darrenc_afk22:07
openstackgerritMerged openstack/manila: NetApp: set proper broadcast domain for IPspace  https://review.openstack.org/40604222:08
*** darrenc_afk is now known as darrenc22:25
openstackgerritMerged openstack/manila: [api-ref] Refactor share instance export locations API documentation  https://review.openstack.org/41511422:47
*** tommylikehu_ has joined #openstack-manila22:54
*** tommylikehu_ has quit IRC22:57
*** lpetrut has quit IRC22:58
*** xyang1 has quit IRC23:06
*** tommylikehu_ has joined #openstack-manila23:08
*** tommylikehu_ has quit IRC23:12
*** gouthamr has quit IRC23:21
*** gouthamr has joined #openstack-manila23:22
*** absubram has quit IRC23:28
openstackgerritTom Barron proposed openstack/manila: Support python 3.5 in tox  https://review.openstack.org/41724523:48
*** ganso has quit IRC23:54
*** cknight has joined #openstack-manila23:57
« Wednesday, 2017-01-04 Index Friday, 2017-01-06 »

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