*** Liang__ has joined #openstack-glance | 01:10 | |
*** gyee has quit IRC | 02:04 | |
*** coreycb has quit IRC | 02:36 | |
*** mnasiadka has quit IRC | 02:36 | |
*** coreycb has joined #openstack-glance | 02:38 | |
*** mnasiadka has joined #openstack-glance | 02:38 | |
*** rcernin has quit IRC | 03:38 | |
*** rcernin has joined #openstack-glance | 03:46 | |
*** udesale has joined #openstack-glance | 03:52 | |
*** m75abrams has joined #openstack-glance | 04:07 | |
*** evrardjp has quit IRC | 04:33 | |
*** evrardjp has joined #openstack-glance | 04:33 | |
*** ratailor has joined #openstack-glance | 05:00 | |
*** rcernin has quit IRC | 07:07 | |
*** rcernin has joined #openstack-glance | 07:08 | |
*** bhagyashris|pto is now known as bhagyashris | 07:16 | |
*** rcernin has quit IRC | 07:23 | |
*** tosky has joined #openstack-glance | 07:28 | |
*** k_mouza has joined #openstack-glance | 07:29 | |
*** bhagyashris is now known as bhagyashris|lunc | 07:29 | |
*** bhagyashris|lunc is now known as bhagyashris | 08:37 | |
*** rchurch has quit IRC | 08:40 | |
*** rchurch has joined #openstack-glance | 08:43 | |
*** tkajinam has quit IRC | 08:55 | |
*** priteau has joined #openstack-glance | 09:08 | |
*** priteau has quit IRC | 09:54 | |
*** Liang__ has quit IRC | 10:01 | |
*** mugsie has quit IRC | 10:07 | |
*** mugsie has joined #openstack-glance | 10:10 | |
openstackgerrit | Abhishek Kekane proposed openstack/glance master: Fix race condition in copy image operation https://review.opendev.org/737596 | 10:21 |
---|---|---|
openstackgerrit | Abhishek Kekane proposed openstack/glance master: Fix race condition in copy image operation https://review.opendev.org/737596 | 10:46 |
abhishekk | dansmith, ^ | 10:47 |
*** udesale_ has joined #openstack-glance | 11:04 | |
*** udesale has quit IRC | 11:06 | |
*** bhagyashris is now known as bhagyashris|brb | 11:32 | |
*** gregwork has quit IRC | 12:00 | |
*** mordred has quit IRC | 12:04 | |
*** bhagyashris|brb is now known as bhagyashris | 12:10 | |
*** mordred has joined #openstack-glance | 12:14 | |
*** ratailor has quit IRC | 12:22 | |
*** priteau has joined #openstack-glance | 12:45 | |
openstackgerrit | Sean McGinnis proposed openstack/glance master: Make test-setup.sh compatible with mysql8 https://review.opendev.org/738615 | 13:16 |
*** jdillaman has joined #openstack-glance | 13:45 | |
abhishekk | jokke, o/ | 13:57 |
*** m75abrams has left #openstack-glance | 14:03 | |
dansmith | abhishekk: are you going to flesh out the patch in the middle to fix the images/ tests? | 14:05 |
abhishekk | dansmith, I guess I will add that one line change in my race condition patch | 14:07 |
dansmith | instead of just fix the tests? | 14:07 |
abhishekk | with TODO to revisit the FakeImage tests | 14:07 |
abhishekk | dansmith, ? | 14:08 |
dansmith | it would be better to not import that hack line, and instead replace id with image_id and fix any tests that pass id= to FakeImage() at least right? | 14:11 |
dansmith | I'll just do if it if you're not going to | 14:11 |
abhishekk | dansmith, will do that | 14:11 |
dansmith | A comment on that FakeImage that it should be combined with a common fixture or something would be good too, but at least making it right would be an improvement | 14:12 |
dansmith | okayt | 14:12 |
abhishekk | it will not take much time to fix it, I will give it a try tonight and will post fully working patch by tomorrow | 14:13 |
dansmith | okay | 14:14 |
dansmith | got the ack from smcginnis yesterday on the auth approach, still waiting for rosmaita before I go much further on that | 14:15 |
abhishekk | ack | 14:15 |
rosmaita | dansmith: what's the review url? | 14:16 |
dansmith | rosmaita: https://review.opendev.org/#/c/738103/ | 14:16 |
rosmaita | ty | 14:16 |
dansmith | subtle nudging is hard on IRC :) | 14:16 |
rosmaita | that was fairly subtle | 14:16 |
dansmith | well, un-subtle in that you get a ding when I say your name.. would be more subtle if I could walk by your cube and whisper "review mah patch" :D | 14:17 |
smcginnis | :) | 14:17 |
rosmaita | that wouldn't be subtle, it would be creepy | 14:20 |
dansmith | heh | 14:27 |
rosmaita | dansmith: just want to make sure i understand the use-case -- we're going to introduce a policy to allow user A to copy an image owned by user B; B continues to own the image,and the problem is that A is not allowed to modify B's image to set the new location, so we need to elevate to an admin context to make that change and also to handle writing the progress-tracking image metadata | 14:41 |
jokke | abhishekk: \o | 14:42 |
abhishekk | dansmith, jokke To use fixture instead of FakeImage will be lot of work, ATM I will change id to image_id in FakeImage with TODO comment and will submit another patch to get rid of FakeImage class | 14:44 |
dansmith | rosmaita: among other limitations, yes that's the gist | 14:45 |
rosmaita | ok, thanks | 14:45 |
dansmith | rosmaita: in reality it'll be more like "allow users to copy public images to other stores so they can use them locally" | 14:45 |
dansmith | i.e. images that aren't charged to anyone | 14:46 |
rosmaita | yeah, that makes me feel better | 14:46 |
jokke | abhishekk: yeah, I'm fine with that as long as we're testing with something that at least looks like what would be coming from the code :P | 14:46 |
dansmith | or on a cloud where things aren't charged at all | 14:46 |
dansmith | abhishekk: yes, I just at least think that FakeImage needs fixing in the short term | 14:47 |
rosmaita | yeah, the operator wants public images available, but doesn't want to pre-stage them in every possible location, so this functionality will allow end-users to stage them on-demand | 14:47 |
rosmaita | that sounds like a good idea, actually | 14:47 |
abhishekk | dansmith, jokke ack | 14:47 |
*** k_mouza has quit IRC | 14:47 | |
jokke | rosmaita: and even more, it allows Nova to stage them on-demand when the user wants to boot something that is not available on the site it was requested | 14:48 |
dansmith | right, that's the use-case here.. nova doing this to pre-optimize the location of the image when booting on a remote deployment with its own ceph cluster | 14:48 |
dansmith | rosmaita: i.e. https://specs.openstack.org/openstack/nova-specs/specs/victoria/approved/rbd-glance-multistore.html | 14:48 |
rosmaita | dansmith: ty, that is the missing link | 14:49 |
dansmith | rosmaita: now when I walk by your cube, I'll just whisper ".....the missing link...." | 14:53 |
rosmaita | :D | 14:53 |
*** k_mouza has joined #openstack-glance | 14:56 | |
rosmaita | dansmith: https://review.opendev.org/#/c/738103/1 looks like a good start | 14:59 |
dansmith | rosmaita: cool, thanks | 15:00 |
openstackgerrit | Abhishek Kekane proposed openstack/glance master: Fix race condition in copy image operation https://review.opendev.org/737596 | 15:06 |
* abhishekk dinner break | 15:36 | |
*** gyee has joined #openstack-glance | 15:38 | |
* tosky use the break to remind people about https://review.opendev.org/#/c/733023/ - which will be backported to ussuri and then train once it gets merged | 15:50 | |
*** udesale_ has quit IRC | 16:39 | |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Deprecation cleanout Registry and related https://review.opendev.org/738671 | 16:49 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Removal of 'enable_v2_api' https://review.opendev.org/738672 | 16:49 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Cleanup remove api v1 and registry code https://review.opendev.org/738673 | 16:49 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Update sample configs post deprecation removals https://review.opendev.org/738674 | 16:49 |
jokke | rosmaita, smcginnis, abhishekk: ^^ ;) | 16:50 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Don't include plugins on 'copy-image' import https://review.opendev.org/738675 | 16:50 |
jokke | abhishekk, dansmith: ^^ | 16:50 |
dansmith | ah cool | 16:51 |
abhishekk | jokke, ack | 16:52 |
dansmith | abhishekk: jokke: is there an example of running a whole flow in the unit tests I can look at? I haven't gone looking, but figure you can point me if there is or warn me if it's hard for some reason | 17:03 |
tosky | if it's more complicated than testing a method or so, shouldn't that be a functional test? | 17:04 |
abhishekk | dansmith AFAIK, there is no test which asserts the entire import flow | 17:05 |
jokke | dansmith: tosky: yeah I'd assume we haven't even tried to get that under unittests ... would expect mocking about the whole namespace for that | 17:05 |
abhishekk | tosky, we have functional tests for copy image operation and import image in multiple stores | 17:05 |
dansmith | oh is there a functional test for it? | 17:06 |
abhishekk | dansmith, we have functional tests for copy image operation and import image in multiple stores | 17:06 |
tosky | I just know there are functional tests - my comment was more about the proper place for this kind of test | 17:06 |
* tosky is just passing by | 17:06 | |
dansmith | abhishekk: can you point me? I don't see anything "import" or "copy" related in the functional filename list at least | 17:07 |
abhishekk | but to add functional tests for conversion or decompression plugin we actually need to download qcow2 or compressed image which is difficult | 17:07 |
abhishekk | dansmith, ack, give me a minute | 17:08 |
dansmith | yeah mocking out the actual operations is fine, I just wanted to run the whole flow with suitable mocks | 17:08 |
dansmith | I already found things that aren't covered in the unit tests because I broke them and didn't get a failure | 17:08 |
dansmith | in the import flow | 17:09 |
openstackgerrit | Dan Smith proposed openstack/glance master: WIP: Admin import action wrapper https://review.opendev.org/738103 | 17:10 |
dansmith | abhishekk: before you disappear for the day, can you skim this fleshed-out import operation? https://review.opendev.org/#/c/738103/2/glance/async_/flows/api_image_import.py | 17:11 |
dansmith | I haven't tested it other than to make sure the existing import unit tests work, | 17:11 |
dansmith | but just thought I'd get a quick approval of all the things I have to change in the import flow (or at least, that I think I'll need to change) | 17:12 |
abhishekk | dansmith, ack | 17:13 |
abhishekk | dansmith, http://paste.openstack.org/show/795392/ | 17:14 |
dansmith | ah okay | 17:15 |
abhishekk | these are the functional tests we have around import operations | 17:15 |
dansmith | thanks | 17:15 |
abhishekk | If you have a look around test, https://github.com/openstack/glance/blob/master/glance/tests/functional/v2/test_images.py#L5402 | 17:16 |
abhishekk | this fails sometimes due to race condition, need more specific logic | 17:16 |
dansmith | the whole import process is full of race conditions, although I'm not sure if or which of those would manifest in functional tests | 17:17 |
*** k_mouza has quit IRC | 17:18 | |
dansmith | hmm, functional tests seem to try to create things in my real system var: | 17:23 |
dansmith | 2020-06-30 10:18:27.316 5298 ERROR glance_store._drivers.filesystem [-] Unable to create datadir: /var/lib/glance/os_glance_tasks_store: PermissionError: [Errno 13] Permission denied: '/var/lib/glance' | 17:23 |
dansmith | is that going to cause the test to fail? | 17:24 |
jokke | dansmith: did not fail on that in my system and this user has no permissions to do that either | 17:26 |
dansmith | yeah, I backed up to master and it does pass.. not sure why, but.. :) | 17:27 |
abhishekk | dansmith, actionwrapper patch looks good, though we might have another bug which we need to fix or take care in your patch | 17:29 |
abhishekk | https://review.opendev.org/#/c/738103/2/glance/async_/flows/api_image_import.py@238 | 17:29 |
abhishekk | Here we also need to set image-status as queued (IMO), if all_stores_must_succeed is False and and we are importing image in 5 stores, as soon as image is uploaded to 1st store image status will be set to active | 17:30 |
dansmith | okay, IMHO that should be fixed before this patch so that this is just moving stuff and not fixing things in the process | 17:32 |
abhishekk | now imagine if this operation fails while importing image to any other store, it will go to above line, remove image locations from all stores, removes checksum, hashing, size etc but image status will still remian active | 17:32 |
dansmith | yep | 17:32 |
abhishekk | also at some places we pass from_state attribute to image_repo.save method which needs to be added in your action class method as well | 17:37 |
dansmith | abhishekk: I'm always asserting the image state hasn't changed in every case.. certainly that's desirable right? | 17:37 |
dansmith | see L103 | 17:37 |
abhishekk | aha :) | 17:38 |
dansmith | I guess maybe we need to stash the previous state though, for cases where the state is changing as part of the import | 17:38 |
dansmith | haven't gotten that far yet | 17:38 |
abhishekk | probably | 17:41 |
dansmith | abhishekk: so, with a couple tweaks since the version you're looking at, all the functional tests pass unchanged | 17:43 |
dansmith | which hopefully means that all works and hasn't broken anything | 17:43 |
abhishekk | great | 17:44 |
jokke | dansmith: sounds super promising at least | 17:49 |
dansmith | hopefully | 17:52 |
abhishekk | dansmith, are you planning to push new revision or after some time? | 17:54 |
dansmith | abhishekk: after I flesh out tests | 17:54 |
abhishekk | dansmith, ack | 17:54 |
dansmith | not that it matters, but while I'm thinking of it.. I'll be out this Friday for our annual try-to-blow-ourselves-up celebration, | 17:56 |
dansmith | so will try to have as much up before then, but will be out Friday | 17:56 |
abhishekk | dansmith, ack, happy holidays | 17:56 |
* abhishekk signing out for the day | 18:06 | |
*** jmlowe has quit IRC | 18:19 | |
-openstackstatus- NOTICE: Due to a flood of connections from random prefixes, we have temporarily blocked all AS4837 (China Unicom) source addresses from access to the Git service at opendev.org while we investigate further options. | 18:22 | |
*** jmlowe has joined #openstack-glance | 18:25 | |
*** ianw_pto is now known as ianw | 19:04 | |
dansmith | jokke: is this going to prevent me from functional'y testing with a different tenant than owns an image? | 19:34 |
dansmith | self.api_server.deployment_flavor = 'noauth' | 19:34 |
dansmith | I would expect that to just not validate the token but still get the tenant from the headers, | 19:35 |
jokke | dansmith: let me look into it. I can't recall using/touching that for years | 19:36 |
dansmith | but X-Tenant-Id in the headers seems to be ignored | 19:36 |
dansmith | there are tests that try to do things against a shared image that seem to honor that header... | 19:39 |
dansmith | ah | 19:42 |
dansmith | I think there must be some magic | 19:42 |
dansmith | https://github.com/openstack/glance/blob/master/glance/tests/functional/v2/test_images.py#L3947-L3950 | 19:42 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Deprecation cleanout Registry and related https://review.opendev.org/738671 | 19:43 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Removal of 'enable_v2_api' https://review.opendev.org/738672 | 19:43 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Cleanup remove api v1 and registry code https://review.opendev.org/738673 | 19:43 |
openstackgerrit | Erno Kuvaja proposed openstack/glance master: Update sample configs post deprecation removals https://review.opendev.org/738674 | 19:43 |
dansmith | okay yeah, I got it | 19:50 |
jokke | cool | 20:02 |
openstackgerrit | Merged openstack/glance master: Use grenade-multinode instead of the custom legacy job https://review.opendev.org/733023 | 20:30 |
openstackgerrit | Luigi Toscano proposed openstack/glance stable/ussuri: Use grenade-multinode instead of the custom legacy job https://review.opendev.org/738693 | 20:40 |
*** priteau has quit IRC | 20:52 | |
openstackgerrit | Christopher Stone proposed openstack/glance_store master: [WIP] Return endpoint set in swift_store_endpoint for single tenant https://review.opendev.org/738700 | 21:32 |
openstackgerrit | Christopher Stone proposed openstack/glance_store master: [WIP] Return endpoint set in swift_store_endpoint for single tenant https://review.opendev.org/738700 | 21:35 |
openstackgerrit | Christopher Stone proposed openstack/glance_store master: [DNM] Use swift_store_endpoint if set with Keystone v3 https://review.opendev.org/738700 | 21:39 |
openstackgerrit | Christopher Stone proposed openstack/glance_store master: [DNM] Use swift_store_endpoint if set with Keystone v3 https://review.opendev.org/738700 | 21:43 |
openstackgerrit | Dan Smith proposed openstack/glance master: Check authorization before import for image https://review.opendev.org/737548 | 21:44 |
openstackgerrit | Dan Smith proposed openstack/glance master: Add a test to replicate the owner-required behavior of copy-image https://review.opendev.org/738101 | 21:44 |
openstackgerrit | Dan Smith proposed openstack/glance master: WIP: Refactor TaskFactory and Executor to take an admin ImageRepo https://review.opendev.org/738102 | 21:44 |
openstackgerrit | Dan Smith proposed openstack/glance master: Make import task capable of running as admin on behalf of user https://review.opendev.org/738103 | 21:44 |
openstackgerrit | Dan Smith proposed openstack/glance master: Refactor common auth token code in images test https://review.opendev.org/738701 | 21:44 |
openstackgerrit | Dan Smith proposed openstack/glance master: Add a functional test for non-owned image copying https://review.opendev.org/738702 | 21:44 |
openstackgerrit | Dan Smith proposed openstack/glance master: Add a policy knob for allowing non-owned image copying https://review.opendev.org/738703 | 21:44 |
dansmith | kaboom | 21:44 |
dansmith | rosmaita: gerrit decided to rebase this for me, but it and the next patch are unchanged: https://review.opendev.org/#/c/737548/ | 21:48 |
dansmith | if you have a sec to re-+W it | 21:49 |
rosmaita | sure | 21:50 |
dansmith | thanks | 21:50 |
openstackgerrit | Christopher Stone proposed openstack/glance_store master: [DNM] Use swift_store_endpoint if set with Keystone v3 https://review.opendev.org/738704 | 21:51 |
*** rcernin has joined #openstack-glance | 22:38 | |
*** tkajinam has joined #openstack-glance | 22:48 | |
*** rcernin has quit IRC | 22:48 | |
*** rcernin has joined #openstack-glance | 22:49 | |
*** tosky has quit IRC | 22:57 | |
openstackgerrit | Christopher Stone proposed openstack/glance_store master: [DNM] Use swift_store_endpoint if set with Keystone v3 https://review.opendev.org/738704 | 23:31 |
openstackgerrit | Christopher Stone proposed openstack/glance_store master: [DNM] Use swift_store_endpoint if set with Keystone v3 https://review.opendev.org/738704 | 23:59 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!