| *** haomaiwa_ has quit IRC | 00:01 | |
| *** haomaiwang has joined #openstack-manila | 00:01 | |
| *** ayma has quit IRC | 00:06 | |
| *** mtanino has quit IRC | 00:11 | |
| *** xyang1 has quit IRC | 00:11 | |
| *** jasonsb has quit IRC | 00:11 | |
| *** jasonsb has joined #openstack-manila | 00:11 | |
| *** lpabon has joined #openstack-manila | 00:20 | |
| *** EinstCrazy has quit IRC | 00:25 | |
| *** lpabon has quit IRC | 00:40 | |
| *** haomaiwang has quit IRC | 00:45 | |
| *** bwolfe has joined #openstack-manila | 00:52 | |
| *** bwolfe has quit IRC | 00:56 | |
| *** EinstCrazy has joined #openstack-manila | 01:01 | |
| openstackgerrit | Anthony Lee proposed openstack/manila: Refactor HP 3PAR share driver to now be HPE https://review.openstack.org/240353 | 01:03 |
|---|---|---|
| *** vbellur has quit IRC | 01:16 | |
| *** bwolfe has joined #openstack-manila | 01:21 | |
| *** bwolfe has quit IRC | 01:22 | |
| *** a_ta has quit IRC | 01:23 | |
| *** bwolfe has joined #openstack-manila | 01:23 | |
| *** a_ta has joined #openstack-manila | 01:24 | |
| *** bwolfe has quit IRC | 01:27 | |
| *** a_ta has quit IRC | 01:28 | |
| *** vbellur has joined #openstack-manila | 01:31 | |
| *** bwolfe has joined #openstack-manila | 01:35 | |
| *** bwolfe has quit IRC | 01:46 | |
| *** Yogi has joined #openstack-manila | 01:48 | |
| *** bwolfe has joined #openstack-manila | 01:49 | |
| *** Yogi2 has joined #openstack-manila | 01:52 | |
| *** Yogi has quit IRC | 01:54 | |
| *** bwolfe has quit IRC | 01:56 | |
| *** jasonsb has quit IRC | 02:02 | |
| *** breitz1 has joined #openstack-manila | 02:08 | |
| *** bwolfe has joined #openstack-manila | 02:11 | |
| *** breitz has quit IRC | 02:11 | |
| *** breitz2 has joined #openstack-manila | 02:11 | |
| *** breitz1 has quit IRC | 02:13 | |
| *** bwolfe has quit IRC | 02:14 | |
| *** bwolfe has joined #openstack-manila | 02:15 | |
| *** Yogi2 has quit IRC | 02:51 | |
| *** haomaiwang has joined #openstack-manila | 02:55 | |
| *** bwolfe has quit IRC | 02:59 | |
| *** bwolfe has joined #openstack-manila | 03:00 | |
| *** haomaiwang has quit IRC | 03:00 | |
| *** bwolfe has quit IRC | 03:08 | |
| *** haomaiwang has joined #openstack-manila | 03:19 | |
| *** tobe has joined #openstack-manila | 03:27 | |
| *** jasonsb has joined #openstack-manila | 03:27 | |
| *** jasonsb has quit IRC | 03:34 | |
| *** a_ta has joined #openstack-manila | 03:35 | |
| *** a_ta has quit IRC | 03:39 | |
| *** yangyapeng has joined #openstack-manila | 03:39 | |
| *** shausy has joined #openstack-manila | 03:43 | |
| *** vbellur has quit IRC | 03:45 | |
| *** haomaiwang has quit IRC | 03:48 | |
| *** shausy has quit IRC | 03:55 | |
| *** shausy has joined #openstack-manila | 03:55 | |
| openstackgerrit | jay.xu proposed openstack/manila: EMC VNX Manila Driver Refactoring https://review.openstack.org/237490 | 04:06 |
| *** bwolfe has joined #openstack-manila | 04:12 | |
| *** haomaiwa_ has joined #openstack-manila | 04:32 | |
| *** bwolfe has quit IRC | 04:35 | |
| *** jasonsb has joined #openstack-manila | 04:38 | |
| *** gouthamr has quit IRC | 04:50 | |
| *** haomaiwa_ has quit IRC | 05:01 | |
| *** haomaiwa_ has joined #openstack-manila | 05:01 | |
| *** vbellur has joined #openstack-manila | 05:20 | |
| *** martyturner has joined #openstack-manila | 05:35 | |
| *** vbellur has quit IRC | 05:36 | |
| *** martyturner has quit IRC | 05:40 | |
| *** haomaiwa_ has quit IRC | 05:44 | |
| *** a_ta has joined #openstack-manila | 05:54 | |
| *** vbellur has joined #openstack-manila | 05:55 | |
| *** a_ta has quit IRC | 06:01 | |
| *** shausy has quit IRC | 06:12 | |
| *** deepakcs has joined #openstack-manila | 06:21 | |
| *** shausy has joined #openstack-manila | 06:28 | |
| *** vbellur has quit IRC | 06:29 | |
| *** nidhimittalhada has joined #openstack-manila | 06:31 | |
| *** vbellur has joined #openstack-manila | 06:43 | |
| *** shausy has quit IRC | 07:05 | |
| *** shausy has joined #openstack-manila | 07:08 | |
| *** Zhongjun has joined #openstack-manila | 07:09 | |
| *** sgotliv has joined #openstack-manila | 07:10 | |
| *** haomaiwang has joined #openstack-manila | 07:11 | |
| *** nkrinner has joined #openstack-manila | 07:40 | |
| *** jvarlamova has quit IRC | 07:53 | |
| *** zigo has quit IRC | 07:53 | |
| *** zigo has joined #openstack-manila | 07:56 | |
| *** a_ta has joined #openstack-manila | 07:57 | |
| *** hodos has quit IRC | 08:00 | |
| *** haomaiwang has quit IRC | 08:01 | |
| *** haomaiwang has joined #openstack-manila | 08:01 | |
| *** a_ta has quit IRC | 08:05 | |
| *** jvarlamova has joined #openstack-manila | 08:13 | |
| *** openstackgerrit has quit IRC | 08:16 | |
| *** openstackgerrit has joined #openstack-manila | 08:16 | |
| *** openstack has joined #openstack-manila | 08:34 | |
| *** EinstCrazy has quit IRC | 08:45 | |
| *** nidhimittalhada has quit IRC | 08:47 | |
| *** EinstCrazy has joined #openstack-manila | 08:50 | |
| *** a_ta has joined #openstack-manila | 08:57 | |
| openstackgerrit | Silvan Kaiser proposed openstack/manila: Add share id to Quobyte backend volume name https://review.openstack.org/241671 | 08:58 |
| *** haomaiwang has quit IRC | 09:01 | |
| *** a_ta has quit IRC | 09:01 | |
| *** shausy has quit IRC | 09:06 | |
| *** haomaiwa_ has joined #openstack-manila | 09:07 | |
| *** Zhongjun has quit IRC | 09:17 | |
| *** gaurangt has joined #openstack-manila | 09:26 | |
| *** ganso has joined #openstack-manila | 09:27 | |
| *** haomaiwa_ has quit IRC | 09:46 | |
| *** haomaiwang has joined #openstack-manila | 09:46 | |
| *** haomaiwang has quit IRC | 10:01 | |
| *** haomaiwang has joined #openstack-manila | 10:01 | |
| *** haomaiwang has quit IRC | 10:04 | |
| *** yangyapeng has quit IRC | 10:14 | |
| *** EinstCrazy has quit IRC | 10:20 | |
| openstackgerrit | Silvan Kaiser proposed openstack/manila: Add share id to Quobyte backend volume name https://review.openstack.org/241671 | 10:45 |
| *** EinstCrazy has joined #openstack-manila | 10:48 | |
| *** a_ta has joined #openstack-manila | 10:58 | |
| *** a_ta has quit IRC | 11:02 | |
| *** gouthamr has joined #openstack-manila | 11:05 | |
| *** u_glide1 has quit IRC | 11:17 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Prevent Share operations during share migration https://review.openstack.org/233052 | 11:25 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Sync Manila Tempest plugin with latest Tempest https://review.openstack.org/242002 | 11:29 |
| *** u_glide has joined #openstack-manila | 11:30 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Sync Manila Tempest plugin with latest Tempest https://review.openstack.org/242002 | 11:32 |
| *** gouthamr_ has joined #openstack-manila | 11:34 | |
| gouthamr_ | vponomaryov : ping | 11:36 |
| vponomaryov | gouthamr_: pong | 11:36 |
| gouthamr_ | hi vponomaryov : Regarding https://review.openstack.org/#/c/242002/ | 11:36 |
| vponomaryov | yes? | 11:37 |
| gouthamr_ | vponomaryov : the references to isolated creds seem to be in a few more places.. | 11:37 |
| *** gouthamr has quit IRC | 11:38 | |
| gouthamr_ | vponomaryov : atleast in name.. we're calling the method 'get_client_with_isolated_creds' | 11:38 |
| vponomaryov | gouthamr_: it is manila plugin naming only | 11:38 |
| gouthamr_ | vponomaryov : so, you could refactor the name to say 'get_client_with_dynamic_creds': | 11:38 |
| vponomaryov | gouthamr_: why? | 11:39 |
| gouthamr_ | vponomaryov : because its confusing :) | 11:39 |
| gouthamr_ | vponomaryov : a teammate had a bugfix.. i guess it's still on our internal review system | 11:40 |
| vponomaryov | gouthamr_: why confusing and what bugfix? | 11:41 |
| vponomaryov | gouthamr_: in scope of Manila it was isolated creds and still so | 11:42 |
| *** lpabon has joined #openstack-manila | 11:42 | |
| vponomaryov | gouthamr_: also, dynamic does not mean isolated, and Manila requires exactly isolated | 11:43 |
| vponomaryov | gouthamr_: for tests that use it | 11:43 |
| gouthamr_ | vponomaryov : oh... didn't know that. thanks for clarifying | 11:44 |
| vponomaryov | gouthamr_: returning to internal bugfix... | 11:44 |
| gouthamr_ | vponomaryov : teammate ran into this bug yesterday and had prepared a bugfix to push today. | 11:45 |
| vponomaryov | gouthamr_: what did you mean? | 11:45 |
| vponomaryov | gouthamr_: please, file bugs in launchpad when you find something | 11:45 |
| gouthamr_ | vponomaryov : Yep.. | 11:46 |
| gouthamr_ | vponomaryov : https://bugs.launchpad.net/manila/+bug/1513105 | 11:46 |
| openstack | Launchpad bug 1513105 in Manila "Manila tempest tests are failing due to change in isolated_creds file name" [Undecided,In progress] - Assigned to Yogesh (ykshirsa) | 11:46 |
| *** lpabon has quit IRC | 11:48 | |
| *** lpabon has joined #openstack-manila | 11:48 | |
| vponomaryov | gouthamr_: https://bugs.launchpad.net/manila/+bug/1513105 describes known incompatibility of manila plugin with latest tempest, but not the reason of CI jobs failure that use not latest Tempest | 11:52 |
| openstack | Launchpad bug 1513105 in Manila "Manila tempest tests are failing due to change in isolated_creds file name" [Undecided,In progress] - Assigned to Yogesh (ykshirsa) | 11:52 |
| vponomaryov | gouthamr_: so, to fix our gates, we could hack requirements file for tempest | 11:53 |
| vponomaryov | gouthamr_: and sync it | 11:53 |
| *** cfey has quit IRC | 11:55 | |
| gouthamr_ | vponomaryov: We can get that bugfix in asap... I think it was fixing some multi-svm test/s as well that were affected by this change. | 11:56 |
| *** cfey has joined #openstack-manila | 11:56 | |
| vponomaryov | gouthamr_: when it is expected to be uploaded to CI? | 11:56 |
| gouthamr_ | vponomaryov: Dynamic credential manager is expecting another kwarg, admin_role; if you don't provide it, it will not initialize the admin client. So, msvm tests were failing downstream.. I am not sure if these tests run upstream on the manila gate at all.. | 11:57 |
| *** a_ta has joined #openstack-manila | 11:57 | |
| gouthamr_ | vponomaryov: I should expect early today, it's about 7AM EST now. I'll shoot off a mail. | 11:58 |
| vponomaryov | gouthamr_: why you work so early then? )) | 12:01 |
| *** a_ta has quit IRC | 12:01 | |
| *** deepakcs has quit IRC | 12:14 | |
| *** gouthamr_ has quit IRC | 12:20 | |
| *** 17SADZP6R has joined #openstack-manila | 12:27 | |
| *** timcl has joined #openstack-manila | 12:29 | |
| openstackgerrit | Silvan Kaiser proposed openstack/manila: Add share id to Quobyte backend volume name https://review.openstack.org/241671 | 12:32 |
| *** porrua has joined #openstack-manila | 12:41 | |
| *** dzamboni has joined #openstack-manila | 12:44 | |
| *** Yogi has joined #openstack-manila | 12:51 | |
| *** 17SADZP6R has quit IRC | 13:01 | |
| *** haomaiwang has joined #openstack-manila | 13:01 | |
| *** tobe has quit IRC | 13:03 | |
| *** Yogi has quit IRC | 13:03 | |
| openstackgerrit | Julia Varlamova proposed openstack/manila: Add LVM driver https://review.openstack.org/232970 | 13:05 |
| *** vbellur has quit IRC | 13:09 | |
| *** tobe has joined #openstack-manila | 13:10 | |
| *** tobe has quit IRC | 13:14 | |
| *** tobe has joined #openstack-manila | 13:14 | |
| *** tobe has quit IRC | 13:19 | |
| *** tobe has joined #openstack-manila | 13:20 | |
| *** tobe has quit IRC | 13:25 | |
| openstackgerrit | Emilien Macchi proposed openstack/puppet-manila: [DO NOT MERGE] test with pure RDO Liberty testing repo (stage CBS repos) https://review.openstack.org/242045 | 13:27 |
| *** sghatty__ has joined #openstack-manila | 13:30 | |
| *** bswartz has quit IRC | 13:35 | |
| *** xyang1 has joined #openstack-manila | 13:57 | |
| *** a_ta has joined #openstack-manila | 13:58 | |
| *** gouthamr has joined #openstack-manila | 13:59 | |
| *** haomaiwang has quit IRC | 14:01 | |
| *** haomaiwa_ has joined #openstack-manila | 14:01 | |
| *** bswartz has joined #openstack-manila | 14:01 | |
| *** ktolstoy has joined #openstack-manila | 14:02 | |
| *** akerr has joined #openstack-manila | 14:02 | |
| *** a_ta has quit IRC | 14:03 | |
| vponomaryov | cfouts: ping | 14:03 |
| *** lpetrut has joined #openstack-manila | 14:07 | |
| gouthamr | vponomaryov: (replying to a message from 2 hrs ago), I'm still getting used to daylight savings .. haha | 14:08 |
| *** david-lyle has joined #openstack-manila | 14:08 | |
| *** martyturner has joined #openstack-manila | 14:10 | |
| *** ktolstoy has quit IRC | 14:10 | |
| *** akshai has joined #openstack-manila | 14:21 | |
| jcsp | is there any kind of helper script for generating DB migrations? | 14:24 |
| jcsp | I'm guessing that the usual way is to run up a devstack instance, and then use "alembic revision --autogenerate" to compare a models.py with the populated DB | 14:24 |
| jcsp | which feels a bit heavyweight | 14:24 |
| *** absubram has joined #openstack-manila | 14:26 | |
| *** Yogi has joined #openstack-manila | 14:26 | |
| *** gaurangt has left #openstack-manila | 14:27 | |
| *** vkmc has joined #openstack-manila | 14:28 | |
| *** Yogi has quit IRC | 14:29 | |
| *** Yogi has joined #openstack-manila | 14:30 | |
| *** mtanino has joined #openstack-manila | 14:31 | |
| *** dustins has joined #openstack-manila | 14:39 | |
| *** gregsfortytwo has quit IRC | 14:45 | |
| *** cfouts has quit IRC | 14:46 | |
| *** gregsfortytwo has joined #openstack-manila | 14:47 | |
| *** neochin has joined #openstack-manila | 14:48 | |
| *** neochin has quit IRC | 14:48 | |
| *** gnarld_ has joined #openstack-manila | 14:48 | |
| *** gouthamr has quit IRC | 14:49 | |
| *** Zhongjun has joined #openstack-manila | 14:56 | |
| *** markstur_ has joined #openstack-manila | 14:57 | |
| *** a_ta has joined #openstack-manila | 14:57 | |
| *** zhongjun2 has joined #openstack-manila | 14:58 | |
| gnarld_ | vponomaryov: pong | 15:00 |
| *** haomaiwa_ has quit IRC | 15:01 | |
| *** cknight has joined #openstack-manila | 15:01 | |
| vponomaryov | jcsp: manila-manage db revision | 15:01 |
| vponomaryov | gnarld_: Chuck? | 15:01 |
| *** haomaiwang has joined #openstack-manila | 15:01 | |
| *** gouthamr has joined #openstack-manila | 15:01 | |
| *** gnarld_ is now known as cfouts | 15:01 | |
| vponomaryov | jcsp: it is exactly for generating new migration. To run migration is used "manila-manage db sync" | 15:02 |
| *** a_ta has quit IRC | 15:02 | |
| cfouts | vponomaryov: yes, it is cfouts. the nick cfouts must be a zombie client that hasn't timed out yet | 15:02 |
| vponomaryov | cfouts: according to https://review.openstack.org/#/c/229142/, I have rebased it partially and did not find appropriate decorator for python methods with versioning | 15:03 |
| vponomaryov | cfouts: api_versioning.wraps dedicated to shell commands, according to docstring and logic | 15:03 |
| cfouts | is that due to the merge conflict? I haven't pushed up that change yet | 15:05 |
| vponomaryov | cfouts: no, there is no such decorator for "python" methods | 15:05 |
| *** david-lyle has quit IRC | 15:06 | |
| jcsp | vponomaryov: aha, thanks. I'm getting an empty migration though. what I'm trying is dropping the db, creating it, running "db sync", then editing models.py and then running "db revision". | 15:06 |
| jcsp | I'm definitely doing something wrong but not sure what :-) | 15:06 |
| jcsp | I tried passing --autogenerate | 15:06 |
| cfouts | vponomaryov: @wraps annotation was tested and there are unit tests that were ported over from the Nova that also test it. | 15:06 |
| jcsp | and it gave me " Can't proceed with --autogenerate option; environment script /opt/stack/manila/manila/db/migrations/alembic/env.py does not provide a MetaData object to the context." | 15:06 |
| *** neochin has joined #openstack-manila | 15:06 | |
| cfouts | vponomaryov: do you have api_versions imported in shell.py? | 15:06 |
| *** a_ta has joined #openstack-manila | 15:07 | |
| *** neochin has quit IRC | 15:07 | |
| vponomaryov | cfouts: "wraps" expects "APIVersion" class instance | 15:07 |
| vponomaryov | cfouts: do you have example? | 15:07 |
| vponomaryov | cfouts: how to decorate python methods? | 15:07 |
| vponomaryov | cfouts: with versions | 15:07 |
| vponomaryov | cfouts: it is not written in your commit | 15:07 |
| *** neochin has joined #openstack-manila | 15:07 | |
| cfouts | https://review.openstack.org/#/c/229142/1/manilaclient/tests/unit/test_api_versions.py | 15:09 |
| cfouts | vponomaryov: starting at line 155 | 15:09 |
| cfouts | I can add an example to the commit | 15:10 |
| *** a_ta has quit IRC | 15:10 | |
| vponomaryov | cfouts: i get - http://paste.openstack.org/show/478106/ usign "wraps" decorator | 15:11 |
| *** a_ta has joined #openstack-manila | 15:13 | |
| vponomaryov | cfouts: so, did you try this decorator on lab? not in unit tests? | 15:13 |
| cfouts | vponomaryov: yes, we tested it internally. Otherwise I wouldn't have been allowed to push it upstream | 15:14 |
| *** neochin has quit IRC | 15:15 | |
| vponomaryov | cfouts: just tested your original commit , same situation | 15:23 |
| vponomaryov | cfouts: if a use this wrapper, it fails with mentioned error | 15:23 |
| vponomaryov | s/a/I/ | 15:26 |
| *** leo_ has joined #openstack-manila | 15:30 | |
| cfouts | vponomaryov: looking | 15:34 |
| vponomaryov | cfouts: change on top of your commit - http://paste.openstack.org/show/478107/ | 15:35 |
| vponomaryov | cfouts: so, here I expected No supported version, and did not reach it | 15:35 |
| *** juzuluag has joined #openstack-manila | 15:38 | |
| cfouts | vponomaryov: I pulled the patch from upstream and it worked for me | 15:43 |
| cfouts | http://paste.openstack.org/show/478110/ | 15:43 |
| vponomaryov | cfouts: upstream commit does not use this wrapper | 15:44 |
| *** leo_ has quit IRC | 15:44 | |
| vponomaryov | cfouts: did you make two lines addon as in my example? | 15:44 |
| cfouts | I only tested this in v2/shell.py | 15:45 |
| vponomaryov | correct | 15:46 |
| cfouts | ok, that is the part that was probably unclear. | 15:46 |
| vponomaryov | call of v2 method with this wrapper fails | 15:46 |
| cfouts | Nova only uses the @wraps annotation in the shell.py | 15:46 |
| cfouts | it depends on being able to build a list of all the functions when the client is created | 15:47 |
| cfouts | if the annotation is used at a lower level then this is a bigger problem to deal with and doesn't fit what Nova implemented | 15:47 |
| vponomaryov | cfouts: why do you make such stress on what Nova has implemented? | 15:48 |
| *** neochin has joined #openstack-manila | 15:48 | |
| cfouts | vponomaryov: because that was the intent when bswartz and others wanted to implement microversions | 15:49 |
| cfouts | vponomaryov: port the code that Nova has | 15:49 |
| cfouts | vponomaryov: cinder is also looking at implementing microversions in this way | 15:49 |
| cfouts | vponomaryov: I believe magnum and ironic are also doing this | 15:50 |
| vponomaryov | cfouts: not allowing versioning of python methods? | 15:50 |
| vponomaryov | cfouts: can you please elaborate why it is much bigger problem on lower level? | 15:51 |
| *** ayma has joined #openstack-manila | 15:52 | |
| vponomaryov | cfouts: for example, microversions support on server side was ported from Nova too. And both, Nova and Manila have following bug - https://bugs.launchpad.net/manila/+bug/1512403 | 15:55 |
| openstack | Launchpad bug 1512403 in Manila "API actions can not be versioned" [Undecided,New] | 15:55 |
| cfouts | vponomaryov: sure, there are a couple of reasons. The first would be that not all of the @api_versions.wraps annotations are in one location. This would make it more difficult to maintain from a client perspective. | 15:55 |
| *** bwolfe has joined #openstack-manila | 15:56 | |
| cfouts | vponomaryov: a second reason would be if you want to go from using services.py to services2.py but keep services.py around. You could manage this from shell.py more easily than again having @api_versions.wraps in multiple files | 15:57 |
| cfouts | vponomaryov: the third thing is that the way wraps works is adding all of the versioned methods in shell.py to _VERSIONED_METHOD_MAP in api_versions. in this way as soon as a shell command is called but a matching version is not found we can return an error | 16:00 |
| cfouts | vponomaryov: the logic to support this in multiple lower level files would be more complicated | 16:00 |
| *** haomaiwang has quit IRC | 16:01 | |
| *** markstur_ has left #openstack-manila | 16:01 | |
| bswartz | jasonsb: over here | 16:01 |
| *** haomaiwang has joined #openstack-manila | 16:01 | |
| bswartz | jasonsb: yeah I think you misread the slides | 16:01 |
| bswartz | jasonsb: I don't see potential for malice involving ganesha | 16:01 |
| jasonsb | sure | 16:01 |
| *** cknight has quit IRC | 16:01 | |
| *** cknight has joined #openstack-manila | 16:02 | |
| jasonsb | but if there is a proxy of any kind to facilitate automation then somebody could go after it | 16:02 |
| vponomaryov | cfouts: ok, if so, then its usage will produce big duplication | 16:02 |
| jasonsb | i think that is the concern for containers because of the nature of the shared kernel side | 16:02 |
| jasonsb | should this be a concern of manila is another matter | 16:03 |
| jasonsb | in any case i will continue to join the magnum meeting and see if there is any requirements guidance which would be of use to manila | 16:04 |
| u_glide | actually attacks are possible, but only in case when we have vulnerable LXC/Kernel + Ganesha | 16:04 |
| *** sghatty__ is now known as sghatty_ | 16:04 | |
| jasonsb | oh | 16:05 |
| jasonsb | u_glide: i have example for cinder | 16:05 |
| jasonsb | adrian: imagine this | 16:05 |
| cfouts | vponomaryov: so take do_list for example. Maybe there is one new change per release to do_list that creates a new microversion. And we never deprecate anything. Then after a couple of years there could be more than 5 do_list methods. Is this what you are thinking? | 16:05 |
| jasonsb | adrian: we allow cinder volumes to be mounted, and accessed by the host | 16:06 |
| jasonsb | adrian: the host mounts the filesystem, causing execution within kernel space | 16:06 |
| jasonsb | adrian: evil user supplies a hacked filesystem designed to take advantage of kernel space execution | 16:06 |
| jasonsb | adrian: leads to compromise of the host | 16:06 |
| bswartz | jasonsb: oh I see they're worried about a different kind of exploit | 16:06 |
| cfouts | vponomaryov: I'd rather be able to sanely deprecate support in the client but I know bswartz has voiced the opposite | 16:06 |
| jasonsb | adrian: so one way to mitigate this is for filesystems to not be supplied by users | 16:06 |
| bswartz | they're worried about the container equivalent of a hypervisor-breakout exploit | 16:07 |
| jasonsb | yes | 16:07 |
| jasonsb | without automation, i think manila is immune | 16:07 |
| vponomaryov | cfouts: right, manilaclient should support wide range of microversions | 16:07 |
| jasonsb | with automation maybe you would introduce an attack surface | 16:07 |
| vponomaryov | cfouts: to be able to be used for old server solutions | 16:08 |
| vponomaryov | cfouts: and yes, each new implementation of "do_list" is HUGE amount of code | 16:08 |
| bswartz | jasonsb: I guess I don't see why the introduction of a mount would allow users do anything different from writing to whatever local file storage their container gives them | 16:09 |
| vponomaryov | cfouts: CLI arguments definition for "do_list" use about 130 lines of code | 16:09 |
| jasonsb | i'm kind of dubious too | 16:09 |
| jasonsb | kernel is not hardened for multi-tenant containers | 16:09 |
| jasonsb | so it seems this a bit forward looking stuff | 16:09 |
| cfouts | vponomaryov: there is one other reason for doing it in shell.py that I forgot. You have to be able to return the appropriate help for that method | 16:09 |
| jasonsb | but i haven't been following the cinder oriented conversation, so i need to listen more | 16:10 |
| cfouts | vponomaryov: I don't disagree with anything you say btw | 16:10 |
| bswartz | jasonsb: yeah I agree -- containers are either secure for multitenancy or they aren't -- that's not our problem | 16:10 |
| jasonsb | if adrian wants to use manila in production at rackspace then it is interesting to hear his concerns | 16:11 |
| jasonsb | so i'll do that. | 16:11 |
| jasonsb | its an interesting use-case | 16:11 |
| jasonsb | (he didn't say he wanted too, but i think the door is open) | 16:12 |
| *** zhongjun2 has quit IRC | 16:14 | |
| *** bwolfe has quit IRC | 16:15 | |
| vponomaryov | cfouts: should be better solution. Current one is duplication-madness | 16:15 |
| vponomaryov | cfouts: just imagine. For update of microversion here - https://review.openstack.org/#/c/240220/ will be need only update of URLs and action names, and now, to be able to to do such small things we will be required to implement hure amount of code duplication for no reason | 16:17 |
| vponomaryov | s/hure/huge/ | 16:18 |
| cfouts | vponomaryov: if you don't do it at that level then how do you pass back the help text for the appropriate version? | 16:19 |
| vponomaryov | cfouts: nothing restricts us to provide help command considering all versions | 16:19 |
| cfouts | vponomaryov: that would be very confusing to the end user I'd think | 16:20 |
| vponomaryov | cfouts: If I want know all of them for some API, I would not want to perform lots of separate help requests | 16:20 |
| cfouts | vponomaryov: as an end user I'd only want the arguments that work for a particular version. I would not want to sort through all of the possible options to determine which ones I can use | 16:21 |
| cfouts | vponomaryov: need to go for now. Would like cknight and bswartz to comment on this since they've seen the code already | 16:21 |
| *** neochin has quit IRC | 16:26 | |
| *** bwolfe has joined #openstack-manila | 16:29 | |
| *** csaba1 has joined #openstack-manila | 16:32 | |
| *** csaba has quit IRC | 16:33 | |
| *** csaba1 has quit IRC | 16:37 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Added driver minimum requirements and features doc https://review.openstack.org/235914 | 16:39 |
| bswartz | jasonsb: how could it be safe to use containers at rackspace (obviously multitenant) if containers aren't secure from breakout exploits? | 16:39 |
| jasonsb | bswartz: good question. i think you would have to run them inside KVM. i will inquire | 16:41 |
| *** Yogi has quit IRC | 16:42 | |
| bswartz | okay so the idea is that even if you break out from your container you're still trapped in a hypervisor shared with nobody | 16:47 |
| bswartz | err trapped in a guest VM | 16:47 |
| *** rhagarty has quit IRC | 16:48 | |
| bswartz | I think the utility of containers is dramatically reduced if they have to be run inside VMs for security | 16:48 |
| jasonsb | yes. if you try to provide a service of containers-on-bare-metal then this is trickier | 16:48 |
| jasonsb | even if single tenant you may have people running around your datacenter who own your bare-metal kernel | 16:48 |
| *** rhagarty has joined #openstack-manila | 16:49 | |
| *** akerr_ has joined #openstack-manila | 16:49 | |
| jasonsb | i'm not sure how this is supposed to work. i need to learn more | 16:49 |
| *** csaba1 has joined #openstack-manila | 16:51 | |
| *** rhagarty_ has joined #openstack-manila | 16:52 | |
| *** ameade has quit IRC | 16:53 | |
| *** rhagarty has quit IRC | 16:53 | |
| *** rhagarty_ has quit IRC | 16:53 | |
| *** ameade has joined #openstack-manila | 16:57 | |
| *** u_glide1 has joined #openstack-manila | 16:57 | |
| *** dhellmann_ has joined #openstack-manila | 17:00 | |
| *** haomaiwang has quit IRC | 17:01 | |
| *** haomaiwang has joined #openstack-manila | 17:01 | |
| ganso | bswartz: hey, remember we talked about removing task_state field? | 17:03 |
| *** akerr has quit IRC | 17:04 | |
| *** bswartz has quit IRC | 17:04 | |
| *** u_glide has quit IRC | 17:04 | |
| *** dhellmann has quit IRC | 17:04 | |
| *** dhellmann_ is now known as dhellmann | 17:05 | |
| *** bswartz has joined #openstack-manila | 17:07 | |
| *** rebase has joined #openstack-manila | 17:08 | |
| ganso | bswartz: ping | 17:09 |
| *** nkrinner has quit IRC | 17:10 | |
| gouthamr | bswartz: https://wiki.openstack.org/wiki/Manila/design/manila-mitaka-data-replication updated. | 17:10 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: Sync Manila Tempest plugin with latest Tempest https://review.openstack.org/242002 | 17:11 |
| *** lpabon has quit IRC | 17:11 | |
| bswartz | ganso: hey I'm here until I finish my lunch | 17:12 |
| * bswartz wants to know why his connection to freenode is dropping and coming back | 17:13 | |
| ganso | bswartz: ok so, there is a distinction in purpose when having a task_state field. Currently when migration fails, the errors are cleaned-up whenever possible and we revert status back to available, while the task_state field shows "migration_error". If "migration_error" was on status field, this status would be blocking for all other operations. We had a | 17:15 |
| ganso | discussion back then if we would like to have the share operational when a migration fails or leave it on an inoperable error state | 17:15 |
| bswartz | ganso: that was before we had instances though | 17:15 |
| bswartz | now we can put the original instance back into the available state and leave the failed instance in migration failure state | 17:16 |
| bswartz | we just need to make sure that the presentation logic doesn't expose the migration failure state to the end user when he looks at the whole share | 17:17 |
| *** bwolfe has quit IRC | 17:18 | |
| ganso | bswartz: humm good point... but if migration fails before having a second instance created, or when we have optimized migration code running but failed, then we still have one instance and no place to put "migration_error" status | 17:18 |
| bswartz | write it to the log maybe? | 17:18 |
| ganso | bswartz: there may be no way to prevent that, because share status reads instance status whehn there is only one instance | 17:18 |
| bswartz | prevent what? | 17:19 |
| *** vbellur has joined #openstack-manila | 17:19 | |
| ganso | bswartz: the information from being shown to the user | 17:19 |
| ganso | bswartz: I also do not think the log should be the only source of information of what state the share really is or what happened to migration | 17:20 |
| ganso | bswartz: Cinder guys had this talk back in Vancouver, that there was confusion about a migration succeeded or migration error situation because there was no status to show it | 17:20 |
| ganso | bswartz: that was another reason why I added task_state | 17:20 |
| bswartz | we write the code so we can prevent whatever we want | 17:21 |
| bswartz | in the case of a failed migration there should always be 2 instances -- the original (unchanged) one and the new (failed) one | 17:21 |
| bswartz | the whole point is not to damage the original instance so after a failure you can at least go back to the way things were | 17:22 |
| ganso | bswartz: that would be the case only in fallback migration alternative | 17:22 |
| bswartz | I'm confused about the issue with cinder -- they have task_state already -- but they don't use it? | 17:22 |
| bswartz | ganso: I feel that some changes are needed to the optimized migration path | 17:23 |
| ganso | bswartz: I am not sure how they are using it now, after Vancouver | 17:23 |
| ganso | bswartz: hummm what changes? | 17:23 |
| bswartz | one thing that's missing right now is a 2-phase migration -- phase 1 would be nondisruptive copy the data over while the source is still writable (long time) -- phase 2 would be disruptive unmount and remount new location (brief) | 17:24 |
| ganso | bswartz: they have migration_status field | 17:24 |
| bswartz | without an API to move from phase 1 to phase 2 the time when the disruptive part of the migration hits is not controllable | 17:25 |
| ganso | bswartz: the optimized migration code path probably can do phase 1 without problems, when migration is finished, it would probably drop the old export location path and return the model update for the new export location | 17:26 |
| openstackgerrit | Anthony Lee proposed openstack/manila: Refactor HP 3PAR share driver to now be HPE https://review.openstack.org/240353 | 17:27 |
| ganso | bswartz: I am not sure I understand why it should be controllable at all | 17:27 |
| bswartz | ganso: yeah but users are going to want to stay in phase 1 until they chose to move to phase 2 | 17:27 |
| *** Zhongjun has quit IRC | 17:27 | |
| ganso | bswartz: oh I see | 17:27 |
| bswartz | consider migrating a 50 TB share over a slow network link | 17:27 |
| bswartz | it may take a week to get the data copied | 17:28 |
| bswartz | once the data is on the other side, you may want to schedule downtime to actually complete the migration | 17:28 |
| bswartz | (downtime of the tenant's application, not of the whole cloud) | 17:28 |
| ganso | bswartz: I see, that is an interesting use case | 17:29 |
| ganso | bswartz: I wrote it down... but so, we drifted a bit away from the status field subject | 17:30 |
| bswartz | yeah so where I was going with that is | 17:31 |
| ganso | bswartz: there are situations where migration fails and we do not have 2 instances, in the fallback migration code path | 17:31 |
| bswartz | I think some more changes are needed at the API level for at least the optimized case, and perhaps the general case, to do 2-phase migrations | 17:31 |
| bswartz | and if we're making changes we should consider possibly forcing optimized migration to use share instances just like fallback migrations | 17:31 |
| ganso | bswartz: if we force that, it could possible unoptimize migration in the backend | 17:32 |
| *** lpetrut has quit IRC | 17:33 | |
| bswartz | well I'm flexible about the approach | 17:33 |
| bswartz | but having a row in the database (other than the original share instance) to track the state of the migration seems like a big win | 17:34 |
| ganso | bswartz: better a row than a column, right? | 17:34 |
| bswartz | I haven't given much thought to what it would looks like to do an optimized migration using multiple share instances | 17:34 |
| bswartz | well I would like to have as much consistency as possible in the way migrations are done | 17:35 |
| bswartz | we don't want to expose end users to different semantics depends on what they have | 17:35 |
| bswartz | depending | 17:35 |
| ganso | bswartz: ok so, users should not see "migration_error" status... and as opposed to Cinder guys, we are not interested in having a "migration_success" status as well, since that would not be possible at the end of migration when we are back with 1 instance | 17:36 |
| bswartz | yeah the goal is for users not to observe any changes when possible | 17:37 |
| ganso | bswartz: also, just to clarify, did you mean that phase 1 => phase 2 is a manual process? | 17:38 |
| bswartz | well I meant that it should be another API call | 17:38 |
| bswartz | I'm also open to the possibility that what we really need is a whole new table to track long-running jobs | 17:38 |
| bswartz | we haven't had any deep discussion on the data copy service yet, but one of my concerns there is state tracking of operations expected to take multiple minutes, hours, or days | 17:39 |
| bswartz | possibly the right thing to do is to create a new table to track those operations | 17:39 |
| ganso | bswartz: right... task_state field was intended for jobs... but it seems like we do not have many jobs scenarios | 17:39 |
| bswartz | if we go down that path then any migration status could go into that table | 17:40 |
| bswartz | this needs a design though | 17:40 |
| ganso | bswartz: yes... it is a very good idea if there are more than one job scenario... but currently we do not know of any besides migration | 17:41 |
| bswartz | ganso: how about backup? | 17:41 |
| bswartz | it's not a proposed feature yet, but my vision of a manila backup service is that we tar up all your data and upload it to an object store | 17:41 |
| bswartz | that would be a long running operation | 17:42 |
| tbarron | +1 | 17:42 |
| ganso | bswartz: yes, seems like a good one... the data copy service would probably up for it as well :) | 17:42 |
| bswartz | tbarron: do you have a keyword pounce set for "backup"? lol | 17:43 |
| gouthamr | hi ganso, bswartz: sorry for butting in, just a heads up.. one of the unanswered questions in DR is migrating a share that has replicas... would appreciate your thoughts on that. | 17:43 |
| bswartz | gouthamr: we need to make it unsupported | 17:43 |
| tbarron | bswartz: no, I actually watch the manila channel, just don't say much. | 17:43 |
| bswartz | and then design support for it later | 17:43 |
| bswartz | tbarron: I'm just joking -- I know you're invested in backup though | 17:43 |
| ganso | gouthamr: we discussed that in Tokyo, well... we would make it unsupported for now | 17:44 |
| gouthamr | bswartz: prevent migration of a share that has replicas? | 17:44 |
| ganso | gouthamr: yes | 17:44 |
| gouthamr | ganso: nice.. will need a code update. Thanks. | 17:45 |
| tbarron | bswartz: and in a vision of elasticaly provisioned data copy services | 17:45 |
| ganso | gouthamr: but the future idea was that the fallback migration code would migrate the active replica and along with the share type, it would replicate it in the destination backend... the unsorted out part is AZ configuration for the destination backend | 17:46 |
| tbarron | bswartz: and separation of control and data plane capabilities, and doing some things in manila before cinder because it should be easier there - closer to a green field | 17:46 |
| gouthamr | ganso: Doable but nasty. | 17:46 |
| bswartz | gouthamr: it's not nasty -- it's literally the only alternative | 17:47 |
| bswartz | gouthamr: unless you implement optimized migration | 17:48 |
| ganso | gouthamr: the fallback migration itself is a bit nasty from a certain point of view... but there are not many better alternatives when we are dealing with backends from different vendors | 17:48 |
| gouthamr | true.. | 17:49 |
| bswartz | ganso: +1 | 17:49 |
| bswartz | in the real world we hope that fallback migration would not get used except for very rare circumstances like replacing one vendor with another perhaps | 17:50 |
| bswartz | the hope is that everyone implements optimized migrations | 17:50 |
| ganso | yes... sometimes it feels a bit like reinventing the wheel... as ameade has been lately more used to | 17:51 |
| *** marcusvrn_ has joined #openstack-manila | 17:59 | |
| *** haomaiwang has quit IRC | 18:01 | |
| *** haomaiwang has joined #openstack-manila | 18:01 | |
| *** timcl has quit IRC | 18:07 | |
| bswartz | I really wish we had a live migration story like cinder does | 18:09 |
| bswartz | but the philosophy here is that even a disruptive migration is better than no migration | 18:10 |
| bswartz | okay I have to drop offline for a bit brb | 18:10 |
| *** bswartz has quit IRC | 18:10 | |
| *** JoseMello has joined #openstack-manila | 18:22 | |
| *** haomaiwa_ has joined #openstack-manila | 18:27 | |
| *** haomaiwang has quit IRC | 18:28 | |
| *** Yogi has joined #openstack-manila | 18:30 | |
| *** alyson_ has quit IRC | 18:34 | |
| *** jasonsb has quit IRC | 18:38 | |
| *** a_ta has quit IRC | 18:47 | |
| *** a_ta has joined #openstack-manila | 18:48 | |
| *** a_ta has quit IRC | 18:52 | |
| *** bswartz has joined #openstack-manila | 18:59 | |
| *** ChanServ changes topic to "Mitaka-1 is Dec 3rd" | 19:00 | |
| *** haomaiwa_ has quit IRC | 19:01 | |
| *** haomaiwang has joined #openstack-manila | 19:01 | |
| *** dustins has quit IRC | 19:08 | |
| *** gouthamr has quit IRC | 19:08 | |
| *** a_ta has joined #openstack-manila | 19:14 | |
| *** rhagarty has joined #openstack-manila | 19:21 | |
| *** jasonsb has joined #openstack-manila | 19:24 | |
| *** vbellur has quit IRC | 19:25 | |
| *** gouthamr has joined #openstack-manila | 19:31 | |
| *** gouthamr_ has joined #openstack-manila | 19:33 | |
| *** gouthamr has quit IRC | 19:36 | |
| *** dustins has joined #openstack-manila | 19:51 | |
| *** martyturner has quit IRC | 19:56 | |
| *** cknight has quit IRC | 19:56 | |
| *** martyturner has joined #openstack-manila | 19:57 | |
| *** cknight has joined #openstack-manila | 19:59 | |
| *** haomaiwang has quit IRC | 20:01 | |
| *** haomaiwang has joined #openstack-manila | 20:01 | |
| *** rhagarty has quit IRC | 20:05 | |
| *** gouthamr_ has quit IRC | 20:23 | |
| *** rhagarty has joined #openstack-manila | 20:26 | |
| *** gouthamr has joined #openstack-manila | 20:37 | |
| *** sgotliv has quit IRC | 20:53 | |
| *** cknight has quit IRC | 20:56 | |
| *** lpabon has joined #openstack-manila | 20:56 | |
| *** a_ta has quit IRC | 20:58 | |
| *** Yogi2 has joined #openstack-manila | 20:59 | |
| *** haomaiwang has quit IRC | 21:01 | |
| *** haomaiwang has joined #openstack-manila | 21:01 | |
| *** Yogi has quit IRC | 21:02 | |
| *** dzamboni has quit IRC | 21:03 | |
| *** martyturner has quit IRC | 21:04 | |
| *** martyturner has joined #openstack-manila | 21:06 | |
| *** timcl has joined #openstack-manila | 21:07 | |
| *** rhagarty_ has joined #openstack-manila | 21:12 | |
| *** rhagarty has quit IRC | 21:13 | |
| *** a_ta has joined #openstack-manila | 21:19 | |
| *** timcl has quit IRC | 21:21 | |
| openstackgerrit | Anthony Lee proposed openstack/manila: Refactor HP 3PAR share driver to now be HPE https://review.openstack.org/240353 | 21:22 |
| *** ayma has quit IRC | 21:27 | |
| *** ayma has joined #openstack-manila | 21:28 | |
| *** a_ta has quit IRC | 21:30 | |
| *** a_ta has joined #openstack-manila | 21:32 | |
| *** Yogi2 has quit IRC | 21:35 | |
| *** akerr_ has quit IRC | 21:36 | |
| *** a_ta has quit IRC | 21:41 | |
| *** a_ta has joined #openstack-manila | 21:42 | |
| *** absubram has quit IRC | 21:44 | |
| *** gouthamr_ has joined #openstack-manila | 21:48 | |
| *** gouthamr has quit IRC | 21:51 | |
| *** JoseMello has quit IRC | 21:52 | |
| *** lpabon has quit IRC | 22:00 | |
| *** haomaiwang has quit IRC | 22:01 | |
| *** haomaiwang has joined #openstack-manila | 22:01 | |
| *** gouthamr_ has quit IRC | 22:02 | |
| *** ganso has quit IRC | 22:08 | |
| *** bswartz has quit IRC | 22:09 | |
| *** a_ta has quit IRC | 22:13 | |
| *** a_ta has joined #openstack-manila | 22:14 | |
| *** ayma has quit IRC | 22:14 | |
| *** akshai has quit IRC | 22:15 | |
| *** akshai has joined #openstack-manila | 22:16 | |
| openstackgerrit | Anthony Lee proposed openstack/manila: Refactor HP 3PAR share driver to now be HPE https://review.openstack.org/240353 | 22:18 |
| *** a_ta has quit IRC | 22:18 | |
| *** porrua has quit IRC | 22:25 | |
| *** dustins has quit IRC | 22:26 | |
| *** a_ta has joined #openstack-manila | 22:27 | |
| *** ayma has joined #openstack-manila | 22:27 | |
| *** bswartz has joined #openstack-manila | 22:29 | |
| *** ayma has quit IRC | 22:32 | |
| *** dgonzalez has quit IRC | 22:51 | |
| *** martyturner has quit IRC | 22:52 | |
| *** haomaiwang has quit IRC | 23:01 | |
| *** haomaiwa_ has joined #openstack-manila | 23:01 | |
| *** dgonzalez has joined #openstack-manila | 23:04 | |
| *** dgonzalez has quit IRC | 23:06 | |
| *** dgonzalez has joined #openstack-manila | 23:07 | |
| *** dgonzalez has quit IRC | 23:12 | |
| *** dgonzalez has joined #openstack-manila | 23:12 | |
| *** a_ta has quit IRC | 23:30 | |
| *** gouthamr has joined #openstack-manila | 23:46 | |
| *** ameade has quit IRC | 23:54 | |
| *** ameade has joined #openstack-manila | 23:55 | |
| *** jasonsb_ has joined #openstack-manila | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!