| *** gouthamr has joined #openstack-manila | 00:01 | |
| *** qeelee has joined #openstack-manila | 00:39 | |
| *** ameade has quit IRC | 01:10 | |
| zengyingzhe | Hi all, please take a look at this patch https://review.openstack.org/#/c/331586/, thanks. | 01:11 |
|---|---|---|
| *** scottda has quit IRC | 01:11 | |
| *** mtanino has joined #openstack-manila | 01:16 | |
| *** yangyapeng has joined #openstack-manila | 01:23 | |
| *** scottda has joined #openstack-manila | 01:23 | |
| *** xiaoyang has joined #openstack-manila | 01:23 | |
| *** magic has joined #openstack-manila | 01:24 | |
| *** yangyapeng has quit IRC | 01:24 | |
| *** yangyapeng has joined #openstack-manila | 01:24 | |
| *** magic is now known as Guest47074 | 01:25 | |
| *** ameade has joined #openstack-manila | 01:25 | |
| *** xiaoyang has quit IRC | 01:28 | |
| *** Guest47074 has quit IRC | 01:28 | |
| openstackgerrit | Nam Nguyen Hoai proposed openstack/manila: TrivialFix: Remove cfg import unused https://review.openstack.org/360960 | 01:28 |
| *** yangyape_ has joined #openstack-manila | 01:31 | |
| *** yangyapeng has quit IRC | 01:35 | |
| *** kaisers_ has joined #openstack-manila | 01:36 | |
| *** kaisers_ has quit IRC | 01:41 | |
| *** wangqun has joined #openstack-manila | 01:47 | |
| *** vbellur has joined #openstack-manila | 02:32 | |
| *** kaisers_ has joined #openstack-manila | 03:26 | |
| *** kaisers_ has quit IRC | 03:30 | |
| *** rhefner has joined #openstack-manila | 03:42 | |
| openstackgerrit | Yingzhe Zeng proposed openstack/manila: Fix test bugs for replication CI https://review.openstack.org/361042 | 03:47 |
| *** gouthamr has quit IRC | 04:01 | |
| *** chlong has quit IRC | 04:12 | |
| *** chlong has joined #openstack-manila | 04:25 | |
| *** xiaoyang has joined #openstack-manila | 04:52 | |
| *** magic has joined #openstack-manila | 04:58 | |
| *** magic is now known as Guest98474 | 04:59 | |
| *** xiaoyang has quit IRC | 05:02 | |
| *** kaisers_ has joined #openstack-manila | 05:14 | |
| *** mtanino has quit IRC | 05:15 | |
| *** kaisers_ has quit IRC | 05:19 | |
| *** shausy has joined #openstack-manila | 05:34 | |
| *** sandanar has joined #openstack-manila | 05:47 | |
| *** lpetrut has joined #openstack-manila | 05:58 | |
| openstackgerrit | Bhagyashri Shewale proposed openstack/python-manilaclient: Replace functions 'Dict.get' and 'del' with 'Dict.pop' https://review.openstack.org/361858 | 06:09 |
| *** sandanar has quit IRC | 06:12 | |
| *** akapil has joined #openstack-manila | 06:14 | |
| *** nkrinner_afk is now known as nkrinner | 06:22 | |
| openstackgerrit | gecong proposed openstack/manila: Add Python 3.5 classifier and venv https://review.openstack.org/361880 | 06:24 |
| *** kaisers_ has joined #openstack-manila | 06:45 | |
| *** dsariel has joined #openstack-manila | 06:56 | |
| *** akapil has quit IRC | 06:57 | |
| *** pcaruana has joined #openstack-manila | 07:00 | |
| *** qeelee has quit IRC | 07:00 | |
| *** qeelee has joined #openstack-manila | 07:02 | |
| *** lpetrut has quit IRC | 07:02 | |
| toabctl | tbarron, bswartz, ganso, xyang1: Regarding this FFE - I have the same opinion as cknight which is 'no'. For the same reasonas as cknight | 07:21 |
| openstackgerrit | Peter Wang proposed openstack/manila: VNX: Use job for NFS share creation https://review.openstack.org/359567 | 07:25 |
| Guest98474 | manila | 07:37 |
| *** Guest98474 has quit IRC | 07:37 | |
| *** xiaoyang has joined #openstack-manila | 07:38 | |
| *** tovchinnikova has joined #openstack-manila | 07:39 | |
| *** xiaoyang has quit IRC | 07:42 | |
| *** xiaoyang has joined #openstack-manila | 07:42 | |
| xiaoyang | how to use manila in public cloud? | 07:43 |
| *** aovchinnikov has joined #openstack-manila | 07:49 | |
| *** qeelee has quit IRC | 07:50 | |
| dgonzalez | ganso, bswartz: Is the gate stable again after https://review.openstack.org/#/c/361413/ ? | 07:53 |
| *** daidv has joined #openstack-manila | 08:00 | |
| *** dsariel has quit IRC | 08:09 | |
| *** dsariel has joined #openstack-manila | 08:09 | |
| *** daidv has quit IRC | 08:13 | |
| *** daidv has joined #openstack-manila | 08:13 | |
| *** daidv has quit IRC | 08:19 | |
| *** qeelee has joined #openstack-manila | 08:25 | |
| *** akapil has joined #openstack-manila | 08:37 | |
| *** qeelee has quit IRC | 08:48 | |
| *** lpetrut has joined #openstack-manila | 08:57 | |
| *** qeelee has joined #openstack-manila | 09:05 | |
| *** senk has joined #openstack-manila | 09:17 | |
| *** qeelee has quit IRC | 09:33 | |
| *** qeelee has joined #openstack-manila | 09:49 | |
| *** jcsp has joined #openstack-manila | 09:50 | |
| *** zigo_ is now known as zigo | 09:56 | |
| *** yangyape_ has quit IRC | 10:08 | |
| *** shausy has quit IRC | 10:08 | |
| *** shausy has joined #openstack-manila | 10:08 | |
| *** qeelee has quit IRC | 10:37 | |
| *** shausy has quit IRC | 10:37 | |
| *** shausy has joined #openstack-manila | 10:38 | |
| *** qeelee has joined #openstack-manila | 10:39 | |
| *** qeelee has quit IRC | 10:39 | |
| *** akapil has quit IRC | 10:41 | |
| *** akapil has joined #openstack-manila | 10:42 | |
| *** wangqun has quit IRC | 10:44 | |
| *** gaurangt has joined #openstack-manila | 11:05 | |
| *** yangyapeng has joined #openstack-manila | 11:16 | |
| *** alyson_ has joined #openstack-manila | 11:21 | |
| *** ganso has joined #openstack-manila | 11:49 | |
| *** gouthamr has joined #openstack-manila | 12:06 | |
| *** tpsilva has joined #openstack-manila | 12:14 | |
| ganso | gouthamr: ping | 12:21 |
| gouthamr | ganso: Good morning! | 12:24 |
| ganso | gouthamr: Morning gouthamr! :) | 12:25 |
| ganso | gouthamr: I liked your idea about including the task_state in the migration_get_progress response | 12:25 |
| *** magic has joined #openstack-manila | 12:26 | |
| *** magic is now known as Guest94048 | 12:26 | |
| gouthamr | ganso: yes.. i was wondering if removing 'total' from the progress would make sense too; because that seems to be the confusion? | 12:26 |
| ganso | gouthamr: but ideally, all those scenario at line 1059 in https://review.openstack.org/#/c/332267/69/manila/share/api.py should be changed to non-error responses | 12:27 |
| ganso | gouthamr: it is confusing because the current_file_progress is not shown | 12:27 |
| gouthamr | ganso: hmmm.. how will that help? | 12:27 |
| *** xiaoyang has quit IRC | 12:28 | |
| ganso | gouthamr: instead of saying "migration of share has already completed" as an error response, we would show "total_progress: 100, task_state: migration_success" | 12:28 |
| *** xiaoyang has joined #openstack-manila | 12:28 | |
| gouthamr | ganso: oh.. | 12:29 |
| ganso | gouthamr: I prefer if those scenarios had progress hardcoded | 12:29 |
| ganso | gouthamr: I think the current approach we have today shows a clear message | 12:29 |
| gouthamr | ganso: so, when phase1 is done and API request is made to migration_get_progress: we'll not take the call to the backends? | 12:30 |
| ganso | gouthamr: basically it "translates" the task_state into a friendlier message... but it is an error response | 12:30 |
| ganso | gouthamr: no, I don't think it should be done | 12:30 |
| ganso | gouthamr: although you said "it is on drivers" | 12:30 |
| ganso | gouthamr: the user should not care that the migration_get_progress call is failing because the driver has not implemented that | 12:31 |
| gouthamr | ganso: yes.. i still think drivers can handle that.. and probably have a mechanism for the fallback case to do the same | 12:31 |
| gouthamr | ganso: we could make it a design requirement... ? | 12:31 |
| gouthamr | ganso: i don't know why drivers will not be able to get progress after phase1 has been completed.. | 12:31 |
| *** Guest94048 has quit IRC | 12:31 | |
| gouthamr | ganso: any scenario that i'm missing? | 12:32 |
| ganso | gouthamr: when phase1 is completed, the driver may clean up the replication/data migration jobs | 12:33 |
| ganso | gouthamr: if driver is using private storage to store the job progress, that informaiton may be cleaned up | 12:33 |
| gouthamr | ganso: afaics, migration_start - spawn a new share in the destination, initiate some data copy asynchronously | continue - keep checking on progress, try to fix any erros data copy if you can, raise on failure | migration_complete - switchover to your destination, and cleanup | 12:34 |
| ganso | gouthamr: the moment when "migration_continue" returns True is when driver cleans up phase1 | 12:34 |
| gouthamr | ganso: i don't agree | 12:35 |
| ganso | gouthamr: it hasn't deleted the source share yet | 12:35 |
| gouthamr | ganso: drivers shouldn't cleanup before migration_complete | 12:35 |
| gouthamr | ganso: or even perform a switchover.. | 12:35 |
| ganso | gouthamr: not a switchover, or delete the source share | 12:36 |
| ganso | gouthamr: but in the storage backend, the replication/data migration job may be finished, and it does not need to be stored or tracked anymore | 12:36 |
| ganso | gouthamr: so I can see a scenario where the driver, when invoked by migration_get_progress, asks the backend what the job progress is... when phase1 is done, there is no such job anymore | 12:37 |
| gouthamr | ganso: hmm, in that case, if you are unable to get the status, couldn't you try...except and look at the task_state and not reraise? | 12:37 |
| *** timcl has joined #openstack-manila | 12:37 | |
| ganso | gouthamr: so when invoking migration_get_progress in this case, drivers will need to do something a bit different | 12:37 |
| ganso | gouthamr: yes, could | 12:37 |
| ganso | gouthamr: the progress can be returned and derived from other data | 12:37 |
| gouthamr | ganso: i don't suspect a lot of backends may have that limitation.. maybe you can do that in the manager? | 12:38 |
| ganso | gouthamr: but I prefer to have the approach we have to day, show a friendly message that the phase1 is done, than invoking the driver and requiring an additional complexity | 12:38 |
| gouthamr | ganso: if phase1 is complete, and the driver/fallback code raises an exception in get_progress, set the progress yourself? | 12:38 |
| gouthamr | ganso: i'm thinking of a time when you'll let drivers return more information :) | 12:38 |
| ganso | gouthamr: yes, that was what I derived from your suggestion | 12:38 |
| gouthamr | ganso: currently, i'm trying to log some useful information... the goal was to let administrators make informed decisions using that information.. | 12:40 |
| ganso | gouthamr: yes I will allow drivers to return more information (common information among several drivers) in migration_get_progress when task_state is 'driver_migration_in_progress" :) | 12:40 |
| ganso | gouthamr: what kind of information/ | 12:40 |
| *** akapil has quit IRC | 12:41 | |
| *** akapil has joined #openstack-manila | 12:41 | |
| gouthamr | ganso: drivers can have different flavors of data movement.. so, sometimes the information necessary may be different.. but at the least, i wanted to send back a status, progress as a percentage, estimated time of completion, and a message.. | 12:42 |
| gouthamr | :D | 12:42 |
| gouthamr | overkill, i know. so, i am logging all of that.. | 12:42 |
| ganso | gouthamr: not overkill, it is good info... I am just concerned if we could get all drivers to do that :D | 12:42 |
| gouthamr | ganso: no, probably not.. and it defeats abstraction if they have different levels/kinds of information unless they have single vendor deployments.. | 12:43 |
| ganso | gouthamr: that's the point, show only what everyone can do so it is included in the abstraction | 12:44 |
| *** akapil has quit IRC | 12:44 | |
| *** akapil has joined #openstack-manila | 12:44 | |
| ganso | gouthamr: btw, line 906 in https://review.openstack.org/#/c/328431/43/manila/share/manager.py | 12:44 |
| ganso | gouthamr: I did not change anything, got confused in your comment | 12:44 |
| gouthamr | ganso: i ruined your weekend, didn't i :P | 12:45 |
| ganso | gouthamr: lol no | 12:45 |
| ganso | gouthamr: fortunately, I was free this weekend | 12:45 |
| openstackgerrit | Daniel Gonzalez Nothnagel proposed openstack/manila: Add multi-segment support https://review.openstack.org/277731 | 12:45 |
| gouthamr | ganso: oh. that, i guess you can ignore that.. i typed out a long message a few lines above that on something and later discarded it.. | 12:45 |
| ganso | gouthamr: if I had any appointments then I wouldn't be able to get everything done by this morning | 12:45 |
| ganso | gouthamr: ok cool | 12:46 |
| ganso | gouthamr: did you see my response to your migration-cancel suggestion? | 12:46 |
| gouthamr | ganso: "to be able to always cancel, I would need to add task_state checks at each step checkpoint" - didn't understand this portion.. i made a couple of changes in my env.. could there be more? | 12:47 |
| ganso | gouthamr: have you taken a look at share backup patch? | 12:48 |
| gouthamr | ganso: yes.. i believe that canceling should only be possible during phase1, after phase1 before completion | 12:48 |
| gouthamr | ganso: not during completion.. | 12:48 |
| gouthamr | ganso: no i haven't.. | 12:48 |
| ganso | gouthamr: hummm what you just said is how it is today... let me grab you a link to show you what I am talking about as "checkpoints" | 12:49 |
| gouthamr | ganso: If it has an APIImpact (i'm sure it does), i need to get on it sometime soon. | 12:49 |
| ganso | gouthamr: so you're the APIImpact guy? | 12:49 |
| ganso | gouthamr: :P | 12:49 |
| gouthamr | ganso: i've been called many names on this channel :P | 12:49 |
| ganso | gouthamr: lol | 12:50 |
| ganso | gouthamr: https://review.openstack.org/#/c/343980/9/manila/data/manager.py@380 | 12:50 |
| gouthamr | ganso: migration_canceling ? | 12:52 |
| ganso | gouthamr: that's what I meant by "checkpoint" | 12:53 |
| gouthamr | ganso: ^ as a busy task state will help you avoid any races | 12:53 |
| ganso | gouthamr: so to be able to cancel migration at any point, not only while it is copying, I need to add a check like that at each step | 12:53 |
| ganso | gouthamr: before creating a share server in the destination, before invoking the driver, before deleting the source instance in migration_complete | 12:54 |
| ganso | gouthamr: to check if the user has cancelled the migration, so I don't proceed | 12:55 |
| ganso | gouthamr: so *the code does not proceed | 12:55 |
| ganso | gouthamr: that's what zhongjun_ is doing | 12:55 |
| ganso | gouthamr: but if so, if you have JUST checked that variable, and then the user cancels, it is going to cancel only at the next checkpoint | 12:56 |
| ganso | gouthamr: this is an additional complexity compared to allowing it to be cancellable only while it is copying, and telling the driver to stop it, as it is today's design | 12:56 |
| gouthamr | ganso: admin should be able to cancel only when the task state is 'migration_driver_in_progress' or 'data_copying_in_progress' or 'data_copying_completed' or 'migration_driver_phase1_done' | 12:56 |
| ganso | gouthamr: it would be an additional mechanism | 12:56 |
| *** tovchinnikova has quit IRC | 12:57 | |
| ganso | gouthamr: oh you want to cancel when phase1 is done? like if admin regrets doing the migration and now that phase1 is done he wants to cancel, but he cannot do so anymore he could only invoke migration-complete, is that it? | 12:57 |
| gouthamr | ganso: i meant allow cancelation when in the middle of phase1 or after, before waiting for completion.. | 12:58 |
| bswartz | dgonzalez: it seems yes | 12:58 |
| gouthamr | ganso: yes.. its entirely possible that data copy takes 10 hours and the admin goes home; comes back and decides not to go ahead with the completion.. | 12:58 |
| ganso | gouthamr: oh, I get it now lol | 12:59 |
| ganso | gouthamr: great I like it :) | 12:59 |
| gouthamr | ganso: the alternative currently is to force-delete the share instance and go out of band of manila nd make changes.. | 12:59 |
| ganso | gouthamr: hummm | 12:59 |
| ganso | gouthamr: ok I will include in the next PS for the bugfix patch | 13:00 |
| gouthamr | ganso: nice. sure thing; i'm ready with my +1 whenever you are, lol | 13:00 |
| ganso | gouthamr: ok :) | 13:00 |
| ganso | gouthamr: about your API comments in the newton-migration-improvements patch | 13:00 |
| gouthamr | ganso: microversioning? | 13:01 |
| ganso | gouthamr: I've never seen this before: "?skip_optimized_migration=null" | 13:01 |
| ganso | gouthamr: I mean, not in OpenStack | 13:01 |
| gouthamr | ganso: keyword args in queries need not have values | 13:01 |
| ganso | gouthamr: I thought it accepted only POST | 13:01 |
| gouthamr | ganso: wait, this is in the body.. | 13:02 |
| ganso | gouthamr: yes | 13:02 |
| ganso | gouthamr: always body | 13:02 |
| gouthamr | ganso: still, it's perfectly valid to have null | 13:02 |
| *** pcaruana has quit IRC | 13:02 | |
| gouthamr | ganso: it's a BadRequest | 13:02 |
| ganso | gouthamr: yes, the code is handling that | 13:03 |
| ganso | gouthamr: if not found, it is False by default: skip_optimized_migration = params.get('skip_optimized_migration', False) | 13:03 |
| ganso | gouthamr: then, if found, it is converted to boolean: skip_optimized_migration = strutils.bool_from_string(skip_optimized_migration, strict=True) | 13:03 |
| ganso | gouthamr: are you saying we should accept it being null? | 13:04 |
| ganso | gouthamr: in the body: {'migration_start': {'skip_optimized_migration': None} } ? | 13:04 |
| gouthamr | ganso: try this: params = {'skip_optimized_migration': None} ; you should see a alueError(msg) | 13:04 |
| ganso | gouthamr: yes | 13:04 |
| gouthamr | ValueError | 13:04 |
| ganso | gouthamr: it captures ValueError | 13:05 |
| gouthamr | ganso: no, i'm saying.. separate the two things.. skip_optimized_migration = params.get('skip_optimized_migration') or False | 13:05 |
| ganso | gouthamr: I see | 13:06 |
| ganso | gouthamr: then it will accept the None and change it to False using the code you suggested | 13:07 |
| ganso | gouthamr: cool | 13:07 |
| gouthamr | ganso: yes | 13:07 |
| ganso | gouthamr: thanks :) | 13:07 |
| gouthamr | ganso: no problem.. | 13:07 |
| ganso | gouthamr: lastly, about the policy thing | 13:07 |
| ganso | bswartz: ping | 13:07 |
| ganso | gouthamr: so I am not sure if this idea I had is really crazy or not | 13:08 |
| ganso | gouthamr: about disabling migration individually | 13:08 |
| gouthamr | ganso: really crazy, yeah. | 13:08 |
| gouthamr | :P | 13:08 |
| ganso | gouthamr: why do you think it is crazy? | 13:09 |
| gouthamr | ganso: i don't think that level of granularity is expected; if its a really highly secure cloud, why would they run non-secure applications alongside secure ones? | 13:09 |
| ganso | gouthamr: let's say company X has a cloud that it uses for its customers, and internally | 13:10 |
| ganso | gouthamr: so company X would not do a migration that may not preserve-metadata or even because it handles users data because they have a contract saying they cannot do that | 13:11 |
| ganso | gouthamr: but they would do it with their own data internally | 13:11 |
| gouthamr | ganso: you already have a config option to disable fallback migration | 13:12 |
| ganso | gouthamr: yes, at this moment, it disables for all backends, in this case I agree that the policy is better | 13:12 |
| ganso | gouthamr: but I am considering the case of disabling selectively | 13:13 |
| ganso | bswartz: do you consider valid the above scenario? ^ | 13:13 |
| gouthamr | ganso: if the company cares about this heterogeneity, they would buy storage that have efficient data movement technology? | 13:13 |
| gouthamr | ganso: hmmm, so you're suggesting an opt in share/driver.py to disable migrations on a backend.. | 13:15 |
| ganso | gouthamr: yes. And disabling in the API would also disable for optimized migration as well | 13:15 |
| gouthamr | ganso: would this opt cover migration both to and from cases? | 13:16 |
| ganso | gouthamr: it has to start from the source backend | 13:16 |
| ganso | gouthamr: it would cover only from | 13:16 |
| gouthamr | ganso: why not 'to'.. you shouldn't be able to take your data to a backend and then not come back from it when your cloud's operational.. | 13:17 |
| *** pcaruana has joined #openstack-manila | 13:17 | |
| gouthamr | ganso: i may be wrong, but i recall a midcycle discussion where cknight was in favor of something like this | 13:18 |
| ganso | gouthamr: well admin should know what he is doing :P | 13:18 |
| ganso | gouthamr: in favor of blocking migration on specific backends? | 13:19 |
| gouthamr | ganso: i *think*, he's not here yet to confirm | 13:19 |
| ganso | gouthamr: but for other features that use migration, I need to make sure it does not go to backends that it is not allowed to come back, yes | 13:19 |
| *** xyang1 has joined #openstack-manila | 13:20 | |
| gouthamr | ganso: yes.. that's true.. so maybe not do this until some use case arises? :) | 13:20 |
| *** nkrinner has quit IRC | 13:21 | |
| ganso | gouthamr: I don't know... the impacts of disabling fallback migration per backend and disabling the migration API are different | 13:22 |
| bswartz | ganso: pong | 13:22 |
| ganso | gouthamr: I should talk to bswartz and mkoderer, mkoderer__ to understand this better | 13:22 |
| ganso | bswartz: could you please share your opinion on the above case? | 13:23 |
| bswartz | I'm unclear on the case | 13:23 |
| bswartz | someone wants to completely disable lossy migrations? | 13:24 |
| *** nkrinner has joined #openstack-manila | 13:24 | |
| ganso | bswartz: yes, in migration's session in Austin, mkoderer raised this concern | 13:24 |
| bswartz | what is the concern? that someone will forget to send the --preserve-metadata flag? | 13:25 |
| ganso | bswartz: I believe he said he has customers that either want to disable migration entirely, or want to disable fallback migration | 13:25 |
| ganso | bswartz: preserve-metadata is True by default | 13:25 |
| ganso | bswartz: I think maybe because os user agreements saying that user data should not be handled by the cloud | 13:25 |
| ganso | bswartz: *there may be user agreements | 13:26 |
| bswartz | I don't agree with this | 13:26 |
| bswartz | An API is an API | 13:26 |
| bswartz | if you choose not to use parts of it, that's up to you | 13:26 |
| bswartz | if you don't trust yourself (to not use certain features) then you can build a layer on top of the API and use that | 13:27 |
| ganso | bswartz: thing is, in the future, share-retype may use it and the user may not be fully aware that it will do a migration | 13:27 |
| bswartz | I think I understand the concern but there has to be a better way to deal with it | 13:27 |
| ganso | bswartz: I already have those switches implemented in the code, do you think I should remove them | 13:28 |
| ganso | ? | 13:28 |
| bswartz | ganso: that's ridiculous -- we're concerned that a user won't understand how to use a feature that doesn't even exist yet? | 13:28 |
| *** rooneym has joined #openstack-manila | 13:28 | |
| bswartz | when we add the feature we'll explain how it works -- people who fail to understand it do so at their own peril | 13:29 |
| bswartz | also I think --preserve-metadata should default to false probably, but that's another topic | 13:29 |
| *** porrua has joined #openstack-manila | 13:29 | |
| ganso | bswartz: I think that preserve-metadata gives enough control over the API, it will prevent fallback migration from running | 13:31 |
| ganso | bswartz: the switches are mostly redundant | 13:31 |
| ganso | bswartz: but I added because it was requested | 13:31 |
| ganso | bswartz: so I guess I will just remove them | 13:31 |
| *** nkrinner has quit IRC | 13:31 | |
| ganso | bswartz: and if someone is concerned, they can remove access through policy and would not require any code change, as gouthamr suggested | 13:32 |
| *** hoonetorg has quit IRC | 13:33 | |
| *** gaurangt has left #openstack-manila | 13:34 | |
| *** eharney has joined #openstack-manila | 13:37 | |
| *** dustins has joined #openstack-manila | 13:40 | |
| *** nkrinner has joined #openstack-manila | 13:44 | |
| *** pcaruana has quit IRC | 13:55 | |
| bswartz | which will you remove? | 13:56 |
| bswartz | is there some policy.json stuff to control the API behavior? | 13:57 |
| gouthamr | bswartz: you can turn off all the APIs or some of the APIs.. or we can make an API specific for fallback migration and control it with policy | 13:58 |
| *** shausy has quit IRC | 13:58 | |
| bswartz | yes but what are you going to remove? | 13:58 |
| gouthamr | bswartz: oh.. ganso was talking about two config opts.. disable_fallback_share_migration and disable_share_migration | 13:58 |
| gouthamr | bswartz: https://review.openstack.org/#/c/328431/43/manila/share/api.py | 13:59 |
| *** akerr has joined #openstack-manila | 14:00 | |
| gouthamr | bswartz: and https://review.openstack.org/#/c/328431/43/manila/share/manager.py | 14:00 |
| *** aovchinnikov has quit IRC | 14:02 | |
| *** senk has quit IRC | 14:04 | |
| *** tpatzig_ has joined #openstack-manila | 14:04 | |
| *** marcusvrn_ has quit IRC | 14:04 | |
| *** tpatzig has quit IRC | 14:06 | |
| *** mkoderer has quit IRC | 14:06 | |
| *** mtanino has joined #openstack-manila | 14:07 | |
| *** mkoderer has joined #openstack-manila | 14:08 | |
| *** pcaruana has joined #openstack-manila | 14:09 | |
| *** ksumit has joined #openstack-manila | 14:15 | |
| *** dsariel has quit IRC | 14:23 | |
| *** yangyapeng has quit IRC | 14:24 | |
| *** ksumit has quit IRC | 14:34 | |
| bswartz | gouthamr: I can see the argument for such config opts but they would make more sense as a policy thing than a config opt | 14:44 |
| bswartz | i.e. I'd rather see them go into policy.json than manila.conf | 14:44 |
| *** kaisers_ has quit IRC | 14:46 | |
| *** eharney has quit IRC | 14:59 | |
| gouthamr | bswartz: +1 i agree.. | 14:59 |
| dustins | bswartz: ping | 15:00 |
| bswartz | dustins: pong | 15:01 |
| dustins | bswartz: Hey, I was wondering if I'd need an FFE for landing this patch for updating our Tempest tests to use the new stable clients https://review.openstack.org/#/c/334596/19 | 15:02 |
| dustins | I chatted with andreaf last week about getting this in, and it looks like the main hurtle is the missing stable IdentityClient interface that the scenario tests need | 15:03 |
| bswartz | dustins: no | 15:03 |
| bswartz | QA improvements are not subject to FF | 15:03 |
| dustins | \o/ | 15:03 |
| dustins | That's good to know | 15:03 |
| bswartz | so we can merge it anytime (except for the quiet periods before the RCs/final release) | 15:03 |
| dustins | Right, that's understandable | 15:04 |
| dustins | Don't wanna bust up the gate when everyone's shoving to get through it :D | 15:04 |
| bswartz | I wish everyone would work on QA improvements between FF and release, but that will never happen | 15:04 |
| dustins | Not with that attitude :P | 15:05 |
| dustins | I wish I had more time to work on it myself, if I'm honest | 15:05 |
| bswartz | true, I need to improve my attitude | 15:07 |
| bswartz | I'm so glad everyone is going to work on QA improvements between FF and release! | 15:08 |
| dustins | There we go :D | 15:14 |
| *** hoonetorg has joined #openstack-manila | 15:16 | |
| *** eharney has joined #openstack-manila | 15:30 | |
| *** porrua has quit IRC | 15:34 | |
| *** pcaruana has quit IRC | 15:43 | |
| *** porrua has joined #openstack-manila | 15:57 | |
| *** david-lyle_ is now known as david-lyle | 16:01 | |
| *** nkrinner is now known as nkrinner_afk | 16:05 | |
| *** sandanar has joined #openstack-manila | 16:17 | |
| openstackgerrit | Alyson proposed openstack/manila: Fix Manila HNAS driver managing a share twice https://review.openstack.org/355906 | 16:21 |
| *** kaisers_ has joined #openstack-manila | 16:35 | |
| *** vbellur has quit IRC | 16:35 | |
| *** lpetrut has quit IRC | 16:38 | |
| *** kaisers_ has quit IRC | 16:40 | |
| openstackgerrit | Ben Swartzlander proposed openstack/manila: [ZFSonLinux] Add share migration support https://review.openstack.org/353417 | 16:43 |
| bswartz | ^ this is a simple rebase, didn't actually fix the driver | 16:43 |
| gouthamr | ganso: ping | 17:05 |
| ganso | gouthamr: pong | 17:05 |
| gouthamr | ganso: so 'inactive' can be removed as a valid share state? | 17:05 |
| ganso | gouthamr: yes | 17:05 |
| gouthamr | ganso: nice.. need an update in models.py.. | 17:05 |
| ganso | gouthamr: really? been doing that since liberty | 17:06 |
| ganso | gouthamr: update where? | 17:06 |
| gouthamr | ganso: ? i noticed the removal in your driver fixes patch | 17:06 |
| gouthamr | https://github.com/openstack/manila/blob/addb65225d015e5489ab1898d78011380c2be4e6/manila/db/sqlalchemy/models.py#L796 | 17:07 |
| ganso | gouthamr: that's for share server | 17:09 |
| *** sandanar has quit IRC | 17:09 | |
| ganso | gouthamr: the share instance model does not have that enum | 17:09 |
| gouthamr | ganso: ah.. yes, you're right but the unit tests and the API have it: https://github.com/openstack/manila/blob/addb65225d015e5489ab1898d78011380c2be4e6/manila/share/api.py#L683 | 17:11 |
| gouthamr | https://github.com/openstack/manila/blob/bdea762769f9e5a69f5e690c18a0d93fc017de61/manila/tests/db/sqlalchemy/test_models.py#L55 | 17:11 |
| ganso | gouthamr: yes, they do... what does need updating? | 17:12 |
| gouthamr | ganso: asking for it to be removed.. | 17:12 |
| ganso | gouthamr: I still need it to suport INACTIVE | 17:13 |
| ganso | gouthamr: let me get you a link | 17:13 |
| gouthamr | ganso: oh... | 17:13 |
| *** aovchinnikov has joined #openstack-manila | 17:14 | |
| ganso | gouthamr: https://review.openstack.org/#/c/332267/70/manila/share/manager.py@1170 | 17:15 |
| gouthamr | ganso: oh, yes.. didn't notice that | 17:17 |
| gouthamr | thanks.. | 17:17 |
| gouthamr | ganso: did you see the comment on teh rpc call? | 17:17 |
| ganso | gouthamr: no, haven't seen | 17:19 |
| gouthamr | ganso: https://review.openstack.org/#/c/332267/ <-- | 17:19 |
| ganso | gouthamr: so it creates another safe thread | 17:19 |
| ganso | gouthamr: and the init_host can proceed to do other stuff | 17:20 |
| ganso | gouthamr: and either way, that will be removed in the newton-migration-improvements patch | 17:20 |
| gouthamr | ganso: :P | 17:20 |
| gouthamr | ganso: okay.. it is getting hard to see these changes. that patch needs to merge | 17:21 |
| ganso | gouthamr: yes | 17:21 |
| ganso | gouthamr: Ideally I should improve tempest tests to include the cancel scenario I just added | 17:26 |
| gouthamr | ganso: okay.. | 17:27 |
| ganso | gouthamr: code worked testing manually | 17:27 |
| *** lpetrut has joined #openstack-manila | 17:28 | |
| openstackgerrit | Alyson proposed openstack/manila: Fix Manila HNAS driver managing a share twice https://review.openstack.org/355906 | 17:32 |
| *** eharney_ has joined #openstack-manila | 17:49 | |
| *** eharney has quit IRC | 17:51 | |
| *** eharney_ is now known as eharney | 17:55 | |
| *** akapil_ has joined #openstack-manila | 18:03 | |
| *** akapil has quit IRC | 18:06 | |
| *** lpetrut has quit IRC | 18:15 | |
| *** cknight has joined #openstack-manila | 18:20 | |
| *** aovchinnikov has quit IRC | 18:21 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Fix Share Migration improper behavior for drivers https://review.openstack.org/332267 | 18:22 |
| *** kaisers_ has joined #openstack-manila | 18:24 | |
| *** vbellur has joined #openstack-manila | 18:25 | |
| openstackgerrit | Merged openstack/manila-ui: Remove incorrect docstrings https://review.openstack.org/340843 | 18:27 |
| *** kaisers_ has quit IRC | 18:29 | |
| *** Suyi_ has joined #openstack-manila | 18:31 | |
| *** senk has joined #openstack-manila | 18:45 | |
| *** zhugaoxiao has quit IRC | 19:16 | |
| *** zhugaoxiao has joined #openstack-manila | 19:16 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Share migration Newton improvements https://review.openstack.org/328431 | 19:16 |
| openstackgerrit | Ben Swartzlander proposed openstack/manila: [ZFSonLinux] Add share migration support https://review.openstack.org/353417 | 19:29 |
| *** akapil_ has quit IRC | 19:31 | |
| *** kaisers_ has joined #openstack-manila | 20:03 | |
| *** darrenc has quit IRC | 20:05 | |
| gouthamr | bswartz xyang1: would appreciate your review on https://review.openstack.org/#/c/332267 -> a bugfix that is the base of a series of migration patches.. I've tested the fixes and they work against a driver optimized migration.. | 20:05 |
| *** darrenc has joined #openstack-manila | 20:07 | |
| bswartz | gouthamr: I *am* reviewing that | 20:08 |
| gouthamr | bswartz: nice.. thank you. | 20:08 |
| *** porrua has quit IRC | 20:09 | |
| *** lpetrut has joined #openstack-manila | 20:11 | |
| *** Suyi_ has quit IRC | 20:17 | |
| openstackgerrit | Jay Mehta proposed openstack/manila: HPE 3PAR driver pool support https://review.openstack.org/329552 | 20:29 |
| *** Suyi_ has joined #openstack-manila | 20:34 | |
| *** senk has quit IRC | 20:46 | |
| *** cknight has quit IRC | 20:52 | |
| openstackgerrit | Ben Swartzlander proposed openstack/manila: [ZFSonLinux] Add share migration support https://review.openstack.org/353417 | 20:52 |
| *** akerr has quit IRC | 20:53 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Fix Share Migration improper behavior for drivers https://review.openstack.org/332267 | 20:56 |
| *** gouthamr has quit IRC | 21:02 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Fix Share Migration improper behavior for drivers https://review.openstack.org/332267 | 21:19 |
| *** dustins has quit IRC | 21:31 | |
| *** eharney has quit IRC | 21:34 | |
| *** rooneym has quit IRC | 21:39 | |
| openstackgerrit | Jay Mehta proposed openstack/manila: HPE 3PAR driver pool support https://review.openstack.org/329552 | 22:03 |
| *** assassin has quit IRC | 22:12 | |
| *** cknight has joined #openstack-manila | 22:25 | |
| *** cknight has quit IRC | 22:32 | |
| *** xyang1 has quit IRC | 22:41 | |
| *** timcl has quit IRC | 22:43 | |
| openstackgerrit | Ivan Chavero proposed openstack/puppet-manila: Add more info for deprecated parameters documentation. https://review.openstack.org/362482 | 22:49 |
| *** lpetrut has quit IRC | 23:04 | |
| *** gouthamr has joined #openstack-manila | 23:07 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Share migration Newton improvements https://review.openstack.org/328431 | 23:10 |
| *** tpsilva has quit IRC | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!