| *** joeljwright has joined #openstack-swift | 00:03 | |
| *** joeljwright has quit IRC | 00:09 | |
| *** shakamunyi has joined #openstack-swift | 00:20 | |
| *** kevinc_ has quit IRC | 00:21 | |
| *** shakamunyi has quit IRC | 00:21 | |
| *** shakamunyi has joined #openstack-swift | 00:21 | |
| *** mmcardle has joined #openstack-swift | 00:25 | |
| *** shakamunyi has quit IRC | 00:25 | |
| *** shakamunyi has joined #openstack-swift | 00:26 | |
| *** tdasilva has left #openstack-swift | 00:27 | |
| *** shakamunyi has quit IRC | 00:29 | |
| *** mmcardle has quit IRC | 00:30 | |
| *** zackf has quit IRC | 00:31 | |
| *** matsuhashi has joined #openstack-swift | 00:31 | |
| openstackgerrit | A change was merged to openstack/swift: Super small code cleanup in Updater https://review.openstack.org/87424 | 00:37 |
|---|---|---|
| *** shri has quit IRC | 00:44 | |
| *** mlipchuk has joined #openstack-swift | 00:48 | |
| *** mwstorer has quit IRC | 00:59 | |
| *** joeljwright has joined #openstack-swift | 01:05 | |
| *** joeljwright has quit IRC | 01:09 | |
| *** mtreinish has quit IRC | 01:22 | |
| *** mtreinish_ has joined #openstack-swift | 01:22 | |
| *** mtreinish_ is now known as mtreinish | 01:22 | |
| *** mmcardle has joined #openstack-swift | 01:26 | |
| *** mmcardle has quit IRC | 01:31 | |
| *** nosnos has joined #openstack-swift | 01:33 | |
| *** saschpe has quit IRC | 01:37 | |
| *** saschpe has joined #openstack-swift | 01:40 | |
| *** joeljwright has joined #openstack-swift | 02:06 | |
| *** joeljwright has quit IRC | 02:10 | |
| *** fifieldt has joined #openstack-swift | 02:17 | |
| *** mmcardle has joined #openstack-swift | 02:27 | |
| *** zhiyan_ is now known as zhiyan | 02:29 | |
| *** mmcardle has quit IRC | 02:32 | |
| openstackgerrit | A change was merged to openstack/swift: Fix testcase test_print_db_info_metadata() https://review.openstack.org/87394 | 02:35 |
| *** lnxnut_ has joined #openstack-swift | 02:51 | |
| *** Longgeek_ has joined #openstack-swift | 02:51 | |
| *** lnxnut has quit IRC | 02:53 | |
| *** Longgeek_ has quit IRC | 02:57 | |
| peluse_ | portante: saw your note on the in process patch, will look at it some more tomorrow, thanks! | 03:03 |
| *** joeljwright has joined #openstack-swift | 03:07 | |
| *** Longgeek_ has joined #openstack-swift | 03:07 | |
| *** joeljwright has quit IRC | 03:12 | |
| openstackgerrit | Takashi Kajinami proposed a change to openstack/swift: Make container PUT and DELETE to allow asynchronous updating to account https://review.openstack.org/87882 | 03:13 |
| *** matsuhashi has quit IRC | 03:14 | |
| *** gyee has quit IRC | 03:23 | |
| *** nosnos has quit IRC | 03:27 | |
| *** mmcardle has joined #openstack-swift | 03:28 | |
| *** lnxnut_ has quit IRC | 03:33 | |
| portante | peluse_: great | 03:33 |
| *** mmcardle has quit IRC | 03:34 | |
| portante | I have some comments about 87101 but wanted to try something out first | 03:34 |
| *** bach has joined #openstack-swift | 03:39 | |
| *** Longgeek_ has quit IRC | 03:39 | |
| *** zhiyan is now known as zhiyan_ | 03:41 | |
| *** bach has quit IRC | 03:48 | |
| *** Longgeek_ has joined #openstack-swift | 03:49 | |
| *** bach has joined #openstack-swift | 03:53 | |
| *** Anju has joined #openstack-swift | 03:56 | |
| *** bach_ has joined #openstack-swift | 03:59 | |
| *** bach has quit IRC | 04:03 | |
| *** bach_ has quit IRC | 04:03 | |
| *** bach has joined #openstack-swift | 04:03 | |
| *** zackf has joined #openstack-swift | 04:05 | |
| *** zackf has quit IRC | 04:06 | |
| *** gvernik has joined #openstack-swift | 04:07 | |
| *** praveenkumar has joined #openstack-swift | 04:07 | |
| *** joeljwright has joined #openstack-swift | 04:08 | |
| *** zaitcev has quit IRC | 04:10 | |
| *** joeljwright has quit IRC | 04:12 | |
| *** gvernik has quit IRC | 04:14 | |
| *** d89 has joined #openstack-swift | 04:14 | |
| *** matsuhashi has joined #openstack-swift | 04:20 | |
| *** nosnos has joined #openstack-swift | 04:23 | |
| *** mmcardle has joined #openstack-swift | 04:30 | |
| *** mmcardle has quit IRC | 04:34 | |
| *** chandan_kumar has joined #openstack-swift | 04:35 | |
| *** chandan_kumar has quit IRC | 05:03 | |
| *** chandan_kumar has joined #openstack-swift | 05:03 | |
| *** mmcardle has joined #openstack-swift | 05:09 | |
| *** joeljwright has joined #openstack-swift | 05:09 | |
| *** mmcardle has quit IRC | 05:13 | |
| *** joeljwright has quit IRC | 05:14 | |
| *** mmcardle has joined #openstack-swift | 05:15 | |
| *** mmcardle1 has joined #openstack-swift | 05:17 | |
| *** mmcardle has quit IRC | 05:17 | |
| *** mmcardle1 has quit IRC | 05:21 | |
| *** baojg has joined #openstack-swift | 05:23 | |
| *** baojg has quit IRC | 05:23 | |
| *** mlipchuk has quit IRC | 05:24 | |
| *** baojg has joined #openstack-swift | 05:24 | |
| *** Longgeek_ has quit IRC | 05:38 | |
| *** zhiyan_ is now known as zhiyan | 05:41 | |
| *** baojg_ has joined #openstack-swift | 05:54 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Keep status_changed_at up-to-date with status changes. https://review.openstack.org/86714 | 05:54 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Merge container storage_policy_index values. https://review.openstack.org/86720 | 05:54 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Expose container info on deleted containers. https://review.openstack.org/86708 | 05:54 |
| *** Longgeek_ has joined #openstack-swift | 05:56 | |
| *** baojg has quit IRC | 05:56 | |
| *** baojg_ has quit IRC | 05:57 | |
| *** cheri has joined #openstack-swift | 05:58 | |
| *** baojg has joined #openstack-swift | 05:58 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Extend direct_client https://review.openstack.org/88144 | 06:00 |
| *** Longgeek_ has quit IRC | 06:00 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Enqueue misplaced objects in object-updater https://review.openstack.org/88145 | 06:00 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Enqueue misplaced objects in container-replicator https://review.openstack.org/88146 | 06:00 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add object-reconciler daemon https://review.openstack.org/85908 | 06:00 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Container Confict on obj update with wrong Storage-Policy-Index https://review.openstack.org/79731 | 06:00 |
| *** RockKuo_ has joined #openstack-swift | 06:07 | |
| *** RockKuo_Office has joined #openstack-swift | 06:07 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Convert POLICIES to list for random.choice https://review.openstack.org/88150 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: add write_fake_ring to test.unit https://review.openstack.org/88151 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Extend interface on InternalClient https://review.openstack.org/88152 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Enqueue misplaced objects in object-updater https://review.openstack.org/88145 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Keep status_changed_at up-to-date with status changes. https://review.openstack.org/86714 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Extend direct_client https://review.openstack.org/88144 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Enqueue misplaced objects in container-replicator https://review.openstack.org/88146 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add object-reconciler daemon https://review.openstack.org/85908 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Container Confict on obj update with wrong Storage-Policy-Index https://review.openstack.org/79731 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Merge container storage_policy_index values. https://review.openstack.org/86720 | 06:08 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Expose container info on deleted containers. https://review.openstack.org/86708 | 06:08 |
| *** baojg has quit IRC | 06:09 | |
| *** baojg has joined #openstack-swift | 06:09 | |
| *** joeljwright has joined #openstack-swift | 06:10 | |
| *** baojg has quit IRC | 06:11 | |
| *** baojg has joined #openstack-swift | 06:11 | |
| *** joeljwright has quit IRC | 06:14 | |
| *** praveenkumar has quit IRC | 06:15 | |
| *** nacim has joined #openstack-swift | 06:20 | |
| openstackgerrit | A change was merged to openstack/python-swiftclient: Makes use of requests.Session https://review.openstack.org/74444 | 06:27 |
| *** baojg has quit IRC | 06:29 | |
| *** baojg has joined #openstack-swift | 06:30 | |
| *** cheri has quit IRC | 06:33 | |
| *** praveenkumar has joined #openstack-swift | 06:33 | |
| *** baojg_ has joined #openstack-swift | 06:39 | |
| *** baojg has quit IRC | 06:42 | |
| *** bach has quit IRC | 06:54 | |
| *** joeljwright has joined #openstack-swift | 06:56 | |
| *** nacim has quit IRC | 07:05 | |
| *** ekarlso has quit IRC | 07:12 | |
| *** mattoliverau has joined #openstack-swift | 07:15 | |
| *** hurricanerix has quit IRC | 07:16 | |
| *** dfg has quit IRC | 07:17 | |
| *** dfg has joined #openstack-swift | 07:17 | |
| *** hurricanerix has joined #openstack-swift | 07:17 | |
| *** haomaiw__ has quit IRC | 07:26 | |
| *** haomaiwa_ has joined #openstack-swift | 07:27 | |
| openstackgerrit | Madhuri Kumari proposed a change to openstack/swift: Bypass committing pending records from .pending file https://review.openstack.org/85351 | 07:29 |
| *** haomai___ has joined #openstack-swift | 07:30 | |
| *** joeljwright1 has joined #openstack-swift | 07:31 | |
| *** joeljwright1 has quit IRC | 07:32 | |
| *** joeljwright1 has joined #openstack-swift | 07:32 | |
| *** haomaiwa_ has quit IRC | 07:33 | |
| *** joeljwright has quit IRC | 07:33 | |
| *** Longgeek_ has joined #openstack-swift | 07:41 | |
| *** Longgeek_ has quit IRC | 07:42 | |
| *** Longgeek_ has joined #openstack-swift | 07:43 | |
| *** Longgee__ has joined #openstack-swift | 07:45 | |
| *** Longgeek_ has quit IRC | 07:48 | |
| openstackgerrit | Yuan Zhou proposed a change to openstack/swift: Fix delete versioning objects when previous is expired https://review.openstack.org/88204 | 07:56 |
| openstackgerrit | Yuan Zhou proposed a change to openstack/swift: Fix delete versioning objects when previous is expired https://review.openstack.org/88204 | 07:59 |
| *** RockKuo_Office has quit IRC | 08:00 | |
| *** _bluev has joined #openstack-swift | 08:01 | |
| _bluev | morninb | 08:02 |
| _bluev | morning... | 08:02 |
| *** nacim has joined #openstack-swift | 08:07 | |
| *** haomai___ has quit IRC | 08:16 | |
| *** haomaiwang has joined #openstack-swift | 08:17 | |
| *** ashish_ has joined #openstack-swift | 08:20 | |
| *** mmcardle has joined #openstack-swift | 08:21 | |
| *** foexle has joined #openstack-swift | 08:24 | |
| foexle | notmyname: thx for your hint yesterday. It was helpful :) | 08:29 |
| *** ashish_ has quit IRC | 08:32 | |
| *** RockKuo_Office has joined #openstack-swift | 08:39 | |
| *** baojg_ has quit IRC | 08:40 | |
| *** fbo_away is now known as fbo | 08:40 | |
| *** baojg has joined #openstack-swift | 08:40 | |
| *** ashish_ has joined #openstack-swift | 08:44 | |
| *** nacim has quit IRC | 08:48 | |
| *** mkollaro has joined #openstack-swift | 08:52 | |
| *** nacim has joined #openstack-swift | 09:06 | |
| *** chandan_kumar has quit IRC | 09:08 | |
| *** ashish_ has quit IRC | 09:09 | |
| *** Longgee__ has quit IRC | 09:20 | |
| *** ashish_ has joined #openstack-swift | 09:21 | |
| *** zhiyan is now known as zhiyan_ | 09:50 | |
| *** zhiyan_ is now known as zhiyan | 09:52 | |
| *** zhiyan is now known as zhiyan_ | 09:54 | |
| *** baojg_ has joined #openstack-swift | 10:12 | |
| *** madhuri has joined #openstack-swift | 10:12 | |
| *** baojg has quit IRC | 10:16 | |
| *** mlipchuk has joined #openstack-swift | 10:37 | |
| *** mkollaro has quit IRC | 10:37 | |
| *** lnxnut has joined #openstack-swift | 10:44 | |
| *** mkollaro has joined #openstack-swift | 10:54 | |
| *** baojg_ has quit IRC | 11:01 | |
| *** mkollaro has quit IRC | 11:04 | |
| *** RockKuo_Office has quit IRC | 11:07 | |
| *** mlipchuk has quit IRC | 11:35 | |
| *** mlipchuk has joined #openstack-swift | 11:35 | |
| *** praveenkumar has quit IRC | 11:40 | |
| *** mmcardle has quit IRC | 11:41 | |
| *** mmcardle has joined #openstack-swift | 11:41 | |
| *** mlipchuk has quit IRC | 11:42 | |
| *** praveenkumar has joined #openstack-swift | 11:44 | |
| *** ashish_ has quit IRC | 11:53 | |
| *** marcusvrn has joined #openstack-swift | 12:05 | |
| *** Trixboxer has joined #openstack-swift | 12:16 | |
| *** mkollaro has joined #openstack-swift | 12:33 | |
| *** yuanz has quit IRC | 12:37 | |
| *** _bluev has left #openstack-swift | 12:37 | |
| *** zhiyan_ is now known as zhiyan | 12:38 | |
| *** d89 has quit IRC | 12:39 | |
| openstackgerrit | Julien Danjou proposed a change to openstack/swift: obj/server: add support for If-Match https://review.openstack.org/88002 | 12:49 |
| *** nosnos has quit IRC | 12:56 | |
| openstackgerrit | paul luse proposed a change to openstack/swift: Quick fix to Storage Policy Support for List Endpoints https://review.openstack.org/88079 | 12:58 |
| *** matsuhashi has quit IRC | 13:06 | |
| *** minnear has quit IRC | 13:16 | |
| *** minnear has joined #openstack-swift | 13:16 | |
| *** matsuhashi has joined #openstack-swift | 13:17 | |
| *** bach has joined #openstack-swift | 13:20 | |
| *** baojg has joined #openstack-swift | 13:27 | |
| *** chandan_kumar has joined #openstack-swift | 13:36 | |
| *** bach has quit IRC | 13:46 | |
| *** bach has joined #openstack-swift | 14:05 | |
| *** changbl has quit IRC | 14:07 | |
| *** bach_ has joined #openstack-swift | 14:12 | |
| *** bach has quit IRC | 14:14 | |
| *** bach_ has quit IRC | 14:28 | |
| *** bach has joined #openstack-swift | 14:29 | |
| *** matsuhashi has quit IRC | 14:30 | |
| *** lpabon has joined #openstack-swift | 14:30 | |
| *** chandan_kumar has quit IRC | 14:31 | |
| *** baojg has quit IRC | 14:32 | |
| *** baojg has joined #openstack-swift | 14:33 | |
| *** bach has quit IRC | 14:37 | |
| *** tanee is now known as tanee-away | 14:42 | |
| *** tanee-away is now known as tanee | 14:42 | |
| *** zematynnad has joined #openstack-swift | 14:43 | |
| *** tanee is now known as tanee-away | 14:45 | |
| *** praveenkumar has quit IRC | 14:46 | |
| *** tanee-away is now known as tanee | 14:49 | |
| *** baojg has quit IRC | 14:50 | |
| *** baojg has joined #openstack-swift | 14:51 | |
| *** baojg has quit IRC | 14:55 | |
| *** chandan_kumar has joined #openstack-swift | 14:57 | |
| *** foexle has quit IRC | 14:59 | |
| notmyname | good morning | 15:06 |
| zematynnad | anybody know if keystone integration was working for swift in grizzly? | 15:08 |
| notmyname | zematynnad: are you seeing a particular error? | 15:09 |
| *** tanee is now known as tanee-away | 15:09 | |
| zematynnad | just not able to get it to work yet and not sure about the correct values that are supposed to go in proxy-server.conf | 15:09 |
| zematynnad | so was just wondering if others had it working in grizzly at some point | 15:10 |
| creiht | zematynnad: it is how devstack is configured and tested, so I would expect it to work | 15:14 |
| zematynnad | I see | 15:14 |
| *** kevinc_ has joined #openstack-swift | 15:16 | |
| notmyname | zematynnad: yes, keystone has always "worked" for swift, in all versions (with varying degrees of performance) | 15:16 |
| chmouel | zematynnad: yes should work on grizzly and should not be too much different for configuration than the current ones | 15:16 |
| zematynnad | ok, I will take a look at devstack for a more concrete example of integrating the two then | 15:17 |
| zematynnad | thamks notmyname. creiht and chmouel | 15:17 |
| notmyname | it's openstack release day! | 15:21 |
| notmyname | peluse_: around? | 15:22 |
| peluse_ | notmyname: yup | 15:22 |
| notmyname | peluse_: I'm merging master into ec and getting one unit test failure: https://gist.github.com/notmyname/5743cce010dac01a5e19 | 15:22 |
| peluse_ | Cool, I'll take a look | 15:23 |
| notmyname | creiht: portante: chmouel: all 3 of you volunteered to be the oslo liason. shall I pick one of you or do one of you want it more? | 15:23 |
| creiht | lol | 15:24 |
| *** Anju_ has joined #openstack-swift | 15:24 | |
| notmyname | creiht: actually, I think you'd do a great job at it. you've got a great feel for appropriate library design and dependencies | 15:26 |
| creiht | yeah I'm fine with that | 15:26 |
| peluse_ | notmyname: have to call into a meeting so will be jsut a few but on first look I agree with the assertion that 2 in fact does not equal 3 :) | 15:26 |
| creiht | not sure the oslo people will be happy about it :) | 15:26 |
| notmyname | peluse_: :-) | 15:26 |
| *** chandan_kumar has quit IRC | 15:27 | |
| chmouel | either way i don't mind :) | 15:27 |
| chmouel | import random;['portante', 'chmouel', 'creith'][random.randrange(0, 2)] ? | 15:27 |
| creiht | lol | 15:27 |
| creiht | random.choice(['portante', 'chmouel', 'creiht']) | 15:27 |
| creiht | :) | 15:27 |
| *** chandan_kumar has joined #openstack-swift | 15:28 | |
| notmyname | import random; print random.choice(['portante', 'chmouel', 'creith']) | 15:28 |
| notmyname | heh. creiht beat me to it :-) | 15:28 |
| notmyname | of course, I don't know who "creith" is | 15:28 |
| *** bach has joined #openstack-swift | 15:29 | |
| chmouel | heh | 15:29 |
| chmouel | we should decide who own the c namespace :) | 15:29 |
| peluse_ | notmyname: did you have like a zillion conflicts to manually resolve? | 15:34 |
| notmyname | creiht: how does this work? https://gist.github.com/notmyname/27f434fb9b94f438f7cf where are the passed end parameters coming from in that test case? | 15:34 |
| notmyname | peluse_: nope. simple ones at then end of account and object tests that check for log formats, and one about the constraints given in /info. that's it (except for this failure) | 15:35 |
| notmyname | creiht: this is in test_internal_client.py | 15:36 |
| notmyname | I don't see any setUp methods anywhere | 15:36 |
| peluse_ | hmm, can you shoot me a note detailing the steps you took, I must be doing something wrong but want to have the same base as obviously so I can fix it | 15:36 |
| notmyname | peluse_: ok | 15:37 |
| notmyname | FYI, all, if you want to run for the openstack TC, today is the last day to nominate yourself | 15:37 |
| creiht | notmyname: Most of that mock stuff is magic to me | 15:38 |
| notmyname | creiht: :-) | 15:38 |
| notmyname | peluse_: do you get 3 conflicts? | 15:38 |
| creiht | notmyname: heh, I'm tired of not getting voted in :/ | 15:39 |
| creiht | they don't want anyone that doesn't tote the party line anyways | 15:39 |
| creiht | notmyname: ok with your previous question, patching like that basically lets you call anything to it and it just returns | 15:41 |
| creiht | then look lower where it checks with request.assert_called_With | 15:41 |
| *** chandan_kumar has quit IRC | 15:42 | |
| creiht | to check to make sure it got called with the right params | 15:42 |
| peluse_ | notmyname: I got many dozens... made a local branch tracking feature/ec and did git rebase -i remotes/origin/master and picked all | 15:42 |
| notmyname | peluse_: ah. I got the latest origin/feature/ec and then did `git merge master`. there are only 3 conflicts there | 15:43 |
| peluse_ | OK, I'll try that way | 15:43 |
| creiht | yeah everything is set at the beginning of the method | 15:43 |
| creiht | the patched versions are sent as the parameters (request and urlopen) | 15:43 |
| notmyname | creiht: is the stack of decorators sent as params...ah ok | 15:44 |
| creiht | yeah I'm not really fond of how it works | 15:44 |
| creiht | but it works I guess | 15:44 |
| creiht | heh | 15:44 |
| notmyname | that makes sense to me then (agreed that it's weird) | 15:44 |
| creiht | I still have to go reread the mock docs every time that I use it | 15:44 |
| *** mwstorer has joined #openstack-swift | 15:46 | |
| *** _80386sx has joined #openstack-swift | 15:48 | |
| *** tanee-away is now known as tanee | 15:49 | |
| *** baojg has joined #openstack-swift | 15:50 | |
| *** tanee is now known as tanee-away | 15:53 | |
| openstackgerrit | Julien Danjou proposed a change to openstack/swift: obj: allow to use ETags with If-{None-,}Match https://review.openstack.org/88346 | 15:54 |
| chmouel | that patch is a bit like inception by jd https://review.openstack.org/#/c/86016/9 | 15:56 |
| creiht | lol | 15:56 |
| *** gyee has joined #openstack-swift | 15:56 | |
| creiht | chmouel: hey can you tell julien that we probably shouldn't support if-none-match with etags yet | 15:57 |
| creiht | as there are some hairy edge cases | 15:57 |
| *** zematynnad has left #openstack-swift | 15:57 | |
| chmouel | i'll ping him to come over | 15:57 |
| notmyname | creiht: well get out your yak shaving shears ;-) | 15:58 |
| creiht | heh | 15:58 |
| creiht | and I'm not sure the use for if_match on put? | 15:58 |
| *** zhiyan is now known as zhiyan_ | 15:59 | |
| portante | creiht: right behind in on the oslo work! :) | 15:59 |
| notmyname | creiht: it might make sense, in a general sense, if the etag weren't the md5 sum of the data. (eg etag == some internal id of the data that doesn't mean it's unique). maybe? that may violate teh spec in other ways... | 16:00 |
| creiht | notmyname: well I'm thinking specifically for swift | 16:00 |
| notmyname | creiht: of course :-) | 16:00 |
| *** jd__ has joined #openstack-swift | 16:01 | |
| jd__ | long time no see #openstack-swift | 16:01 |
| jd__ | hi everyone | 16:01 |
| notmyname | creiht: did I ever tell you about the guy I interviewed who, for the first 30-45 minutes, told me how swift was getting RFC2616 wrong, and he knew this because he'd read over the rfc twice? | 16:01 |
| notmyname | jd__: welcome, stranger :-) | 16:02 |
| creiht | notmyname: heh | 16:02 |
| notmyname | Techniques on How to Not Interview, 101 | 16:02 |
| creiht | I don't think anyone gets rfc2616 right :) | 16:02 |
| *** chandan_kumar has joined #openstack-swift | 16:02 | |
| notmyname | true :-) | 16:03 |
| *** tanee-away is now known as tanee | 16:03 | |
| *** tanee is now known as tanee-away | 16:03 | |
| *** bach_ has joined #openstack-swift | 16:03 | |
| notmyname | portante: so I'm not sure I understood what you said to creiht about oslo liason. how passionate about it are you? | 16:04 |
| notmyname | jd__: we have some concern about etag + if-non-match edge cases (re your patch) | 16:04 |
| jd__ | hehe, yeah you probably know better than me | 16:05 |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Only encode metadata for user customed headers https://review.openstack.org/88351 | 16:05 |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Use quote/unquote from six module for py3 https://review.openstack.org/87287 | 16:05 |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Remove validate_headers https://review.openstack.org/87772 | 16:05 |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Fix test_raw_upload test https://review.openstack.org/87784 | 16:05 |
| creiht | jd__: howdy :) | 16:05 |
| jd__ | that's mainly just a try to start discussing if that's possible, and if so to help moving this forward since I'd love to have that feature | 16:05 |
| notmyname | jd__: creiht had looked into it. IIRC the hard parts are dealing with the eventual consistency nature of things | 16:05 |
| * jd__ nods | 16:05 | |
| creiht | jd__: howdy :) | 16:06 |
| jd__ | I've used GETorHEAD thinking it would use the quorum thingy to handle that for me, but I don't know if that's good enough | 16:06 |
| openstackgerrit | Chmouel Boudjnah proposed a change to openstack/python-swiftclient: Only encode metadata for user customed headers https://review.openstack.org/88351 | 16:06 |
| jd__ | creiht: hey! | 16:06 |
| *** bach has quit IRC | 16:06 | |
| chmouel | oh damn rebase | 16:06 |
| creiht | yeah that will not work, because it is just going to choose one at random | 16:07 |
| creiht | and we would really like to have something like this work without having to do a lot of extra head requests | 16:07 |
| *** changbl has joined #openstack-swift | 16:07 | |
| creiht | I was really close to having it work in the current direction I was going | 16:08 |
| jd__ | ah :) | 16:08 |
| creiht | what is missing is a reasonable way to update an object that already exists | 16:08 |
| jd__ | did you send a patch already? I'd love to help if I can | 16:08 |
| creiht | jd__: I backed the work out when I submitted the if_none_match: * patch | 16:09 |
| portante | notmyname: I was confused by your email, as I thought you sent the email after the IRC meeting where creiht volunteered, so I answered thinking that perhaps something transpired after the meeting where creiht did not want to do it | 16:09 |
| creiht | jd__: will you be at the summit? | 16:09 |
| creiht | It is kinda difficult to discuss over irc :) | 16:09 |
| jd__ | yes I will | 16:09 |
| notmyname | portante: ah, ok. I actually didn't see creiht volunteered during the meeting. conference wifi cut out in the middle of the meeting and that's what I lost | 16:10 |
| creiht | ahh | 16:10 |
| portante | ah | 16:10 |
| portante | that makes sense now | 16:10 |
| jd__ | creiht: well if you can make this work _before_ the summit that'd be even better though :D | 16:10 |
| creiht | haha | 16:10 |
| creiht | yeah that isn't likely | 16:10 |
| notmyname | portante: creiht: so that makes a lot of sense then. if creiht volunteered at the meeting, then that's completely ok with me :-) | 16:10 |
| creiht | the biggest issue is handling the eventual consistency issues | 16:10 |
| creiht | like for example, if you PUT v1 and goes on nodes a,b,c | 16:11 |
| creiht | yout PUT v2 and there are failures so it goes on nodes c,d,e | 16:12 |
| peluse_ | notmyname: OK, off my call now. Yup, 3 conflicts, will take a look at that failuire now | 16:12 |
| notmyname | peluse_: I've been trying some printf debugging, but still haven't found anything | 16:13 |
| notmyname | yet | 16:13 |
| creiht | if someone does an if-none-match with a v1 etag | 16:14 |
| creiht | a and b have the right etag, so don't need the data | 16:14 |
| creiht | c gets overwritten with the uploaded file (with a different timestamp) | 16:15 |
| *** bach_ has quit IRC | 16:15 | |
| creiht | now if c fails, the v2 in d and e would eventually replicate over | 16:15 |
| creiht | so what it really needed was something like the linux touch command, that would just update the timestamp of a,b | 16:16 |
| creiht | then things would work reasonably well | 16:16 |
| jd__ | sounds too easy to be true :) | 16:16 |
| creiht | hehe | 16:16 |
| creiht | yeah so turns out that isn't that easy :) | 16:17 |
| creiht | right now that's handled by default via POST as COPY | 16:17 |
| creiht | so if you do that to a 5GB object, you don't want to wait for the object to copy just to update the ts | 16:17 |
| *** foexle has joined #openstack-swift | 16:17 | |
| creiht | there are some other oddities | 16:17 |
| jd__ | hm, can't we add something like a TOUCH method to the object server? | 16:18 |
| creiht | well the problem is how to make the change in an atomic way | 16:18 |
| jd__ | err right | 16:18 |
| creiht | and at this point, my memory is getting a little fuzzy as to all of the issues with that | 16:19 |
| creiht | I would have to poke at the code again | 16:19 |
| *** bach has joined #openstack-swift | 16:20 | |
| creiht | we started trying to figure out how to just make the .meta post work right | 16:21 |
| *** tanee-away is now known as tanee | 16:21 | |
| creiht | so we wouldn't have to do post a copy | 16:21 |
| creiht | but we ran into the same problems we had before | 16:21 |
| creiht | if only content-type wasn't in the container listing :/ | 16:22 |
| notmyname | creiht: can't send an update to the container servers for the metadata? seems like it would be the same container update semantics as an object PUT, just without the body | 16:23 |
| notmyname | creiht: I'm not doubting at all, just curious about the problems you ran in to | 16:23 |
| creiht | yeah we talked about that but I forgot what the issue ended up being with that | 16:23 |
| creiht | hehe | 16:23 |
| creiht | when you start digging in the weeds, it is hard to keep all the details | 16:24 |
| notmyname | peluse_: hmm...it seems that the test case is patching Request to return URLError, not raise it. ...still learning | 16:25 |
| notmyname | I'm not sure yet why this works in master | 16:25 |
| notmyname | (or does it?) | 16:25 |
| notmyname | hmm... cschwede is in git blame for it. make him fix it ;-) | 16:26 |
| *** bach_ has joined #openstack-swift | 16:26 | |
| cschwede | i broke something? let me know :) | 16:26 |
| creiht | jd__: so the easy(ish) answer is, if we can figure out how to get fast POST to work with content-type updates, that would likely lead to being able to support etag if-none-match | 16:26 |
| notmyname | cschwede: I'm working on merging master into the ec branch. but a unit test for internal_client is failing. I'm not sure why yet | 16:27 |
| cschwede | notmyname: which one? | 16:27 |
| peluse_ | notmyname: not failing for me.... | 16:28 |
| notmyname | cschwede: swift@saio:~/swift/test/unit/common$ nosetests -svv test_internal_client.py:TestSimpleClient.test_get_with_retries | 16:28 |
| notmyname | peluse_: ah, interesting | 16:28 |
| *** bach has quit IRC | 16:28 | |
| cschwede | notmyname: checking | 16:28 |
| peluse_ | notmyname: want to push your complete branch up and I can compare to mine? | 16:29 |
| notmyname | peluse_: cschwede: interesting. it's also failing on master for me. /whatsgoingon | 16:29 |
| *** gyee has quit IRC | 16:29 | |
| peluse_ | notmyname: maybe that test doesn't liky you :) | 16:30 |
| notmyname | maybe a dependency thing. old version of mock. wasn't there something about that recently? | 16:30 |
| notmyname | peluse_: what version of mock do you have? | 16:30 |
| cschwede | notmyname: mhh, it’s failing for me on master as well | 16:31 |
| creiht | notmyname: heh... tests on master run for me | 16:32 |
| notmyname | creiht: what version of mock do you havE? | 16:32 |
| creiht | 1.0.1 | 16:34 |
| cschwede | notmyname: if i run .unittests, it also works for me (and if the test would always fail, we should see it on the gate too?) | 16:34 |
| cschwede | maybe it’s related to nosetest? | 16:35 |
| cschwede | err, no makes no sense, .unittest also calls nosetest | 16:35 |
| notmyname | mock 0.8.0 (what's in swift's test-requirements) fails. mock 1.0.1 passes | 16:35 |
| cschwede | nosetests test/unit/common/test_internal_client.py:TestSimpleClient works | 16:35 |
| peluse_ | notmyname: mother mocker | 16:35 |
| cschwede | nosetests test/unit/common/test_internal_client.py:TestSimpleClient.test_get_with_retries doesn’T work | 16:35 |
| notmyname | cschwede: same | 16:36 |
| notmyname | cschwede: wait, with 0.8.0? | 16:36 |
| peluse_ | hey, how do I check my mock ver? | 16:36 |
| cschwede | notmyname: mock 1.0.1 for me | 16:36 |
| notmyname | python -c 'import mock; print mock.__version__' | 16:37 |
| notmyname | cschwede: confirmed that it still breaks with 0.8.0 | 16:37 |
| cschwede | peluse_: python -c "import mock; print mock.__version__" | 16:37 |
| cschwede | too slow | 16:37 |
| peluse_ | yup 1.0.1 for me too | 16:37 |
| notmyname | in 1.0.0 alph 2 of mock: Any exceptions in an iterable side_effect will be raised instead of returned | 16:37 |
| notmyname | which is the problem | 16:38 |
| notmyname | so 0.8.0 can't work | 16:38 |
| peluse_ | mockin' A | 16:38 |
| cschwede | yep, 0.8.0 fails for me too | 16:38 |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: merged master into ec https://review.openstack.org/88359 | 16:39 |
| notmyname | peluse_: ^^ | 16:39 |
| *** tanee is now known as tanee-away | 16:39 | |
| notmyname | the change in https://review.openstack.org/#/c/88359/1/test/unit/proxy/test_server.py makes me happy :-) | 16:39 |
| cschwede | notmyname: so how do we continue on this issue with mock 0.8? | 16:40 |
| notmyname | thanks acoles :-) | 16:40 |
| *** bach_ has quit IRC | 16:40 | |
| notmyname | cschwede: patch test-requirements | 16:40 |
| notmyname | while >-1.0.0 would probably work, >=1.0.1 is probably fine | 16:41 |
| notmyname | meh, the global-requirements have ">=1.0", so we should do that | 16:41 |
| peluse_ | notmyname: identical to mine, check! | 16:42 |
| *** tanee-away is now known as tanee | 16:42 | |
| cschwede | notmyname: ok, uploading | 16:42 |
| notmyname | cschwede: hang on | 16:42 |
| cschwede | notmyname: ok | 16:42 |
| *** tanee is now known as tanee-away | 16:42 | |
| *** tanee-away is now known as tanee | 16:42 | |
| notmyname | cschwede: please add a reference in your commit message to http://www.voidspace.org.uk/python/mock/changelog.html#version-1-0-0-alpha-2 | 16:43 |
| notmyname | cschwede: other than that, thanks! | 16:43 |
| *** tanee is now known as tanee-away | 16:43 | |
| notmyname | cschwede: or specifically to "Any exceptions in an iterable side_effect will be raised instead of returned" | 16:43 |
| *** tanee-away is now known as tanee | 16:43 | |
| cschwede | notmyname: ok | 16:43 |
| *** tanee is now known as tanee-away | 16:43 | |
| *** changbl has quit IRC | 16:44 | |
| *** tanee-away is now known as tanee | 16:44 | |
| notmyname | peluse_: ack. merging | 16:45 |
| *** tanee is now known as tanee-away | 16:45 | |
| *** tanee-away is now known as tanee | 16:46 | |
| notmyname | ubuntu release and openstack release all in the same day. what an exciting time to be alive! | 16:47 |
| creiht | lol | 16:47 |
| *** mmcardle has quit IRC | 16:47 | |
| *** tanee is now known as tanee-away | 16:48 | |
| *** tanee-away is now known as tanee | 16:48 | |
| creiht | I wonder how long before we have images for the new lts | 16:48 |
| creiht | time to rebuild my saio :) | 16:48 |
| notmyname | creiht: at rax or for a vm? (for a vm: http://releases.ubuntu.com/trusty/) | 16:48 |
| creiht | for rax | 16:49 |
| notmyname | hmm..only 22 seeds for the 64-bit server torrent | 16:50 |
| notmyname | oh, there it goes | 16:50 |
| notmyname | 329 peers | 16:50 |
| notmyname | very nice to download at about 4MB/s. already done | 16:52 |
| creiht | http://www.ubuntu.com/ | 16:54 |
| creiht | Ubuntu 14.04 LTS, with OpenStack. | 16:54 |
| creiht | that's kinda interesting | 16:54 |
| openstackgerrit | Christian Schwede proposed a change to openstack/swift: Fix test-requirements.txt https://review.openstack.org/88363 | 16:55 |
| notmyname | cschwede: thanks! | 16:56 |
| cschwede | you’re welcome! | 16:56 |
| cschwede | notmyname: and i learned another thing today: also test with the lowest version in (test-)requirements.txt to ensure nothing breaks | 16:57 |
| cschwede | notmyname: so happy me :) | 16:57 |
| notmyname | cschwede: chmouel: do you have a tl;dr for py3 in swiftclient? with the icehouse release, I was looking at when we should cut another release there | 16:58 |
| notmyname | looks like there are a few things other than py3 changes that have landed, but the majority are for py3 | 16:58 |
| notmyname | I'd prefer to have full py3 support rather than "here's a release with new dependencies for something that doesn't work yet" | 16:59 |
| openstackgerrit | paul luse proposed a change to openstack/swift: Bring Object Auditor Coverage To 99% https://review.openstack.org/87101 | 16:59 |
| Alex_Gaynor | notmyname: As far as I can tell, we're like three patches away from having passing tests. | 16:59 |
| cschwede | notmyname: yes, that would be nice. is there any specific date you have in mind? | 16:59 |
| notmyname | Alex_Gaynor: ah, cool | 16:59 |
| Alex_Gaynor | notmyname: https://review.openstack.org/#/c/87772/ https://review.openstack.org/#/c/87784/ https://review.openstack.org/#/c/87287/ are the most important ones, if you want to do a review :-) | 17:00 |
| notmyname | cschwede: no particular date for swiftclient. if there's stuff that's useful and releasable, then let's release. if not, no need to release | 17:00 |
| notmyname | Alex_Gaynor: thanks | 17:00 |
| *** shri has joined #openstack-swift | 17:04 | |
| notmyname | creiht: py34 in 14.04: https://news.ycombinator.com/item?id=7604596 | 17:04 |
| *** mmcardle has joined #openstack-swift | 17:05 | |
| *** nacim has quit IRC | 17:09 | |
| *** mmcardle has quit IRC | 17:09 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Enqueue misplaced objects in object-updater https://review.openstack.org/88145 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Keep status_changed_at up-to-date with status changes. https://review.openstack.org/86714 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Extend direct_client https://review.openstack.org/88144 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Enqueue misplaced objects in container-replicator https://review.openstack.org/88146 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: add write_fake_ring to test.unit https://review.openstack.org/88151 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Add object-reconciler daemon https://review.openstack.org/85908 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Container Confict on obj update with wrong Storage-Policy-Index https://review.openstack.org/79731 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Extend interface on InternalClient https://review.openstack.org/88152 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Merge container storage_policy_index values. https://review.openstack.org/86720 | 17:12 |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: Expose container info on deleted containers. https://review.openstack.org/86708 | 17:12 |
| clayg | oh sorry, just fixing up some missing imports from my rebase dance yesterday | 17:12 |
| peluse_ | phew | 17:19 |
| *** chandan_kumar has quit IRC | 17:19 | |
| peluse_ | clayg: saw the word when its a good time to start looking at these (and in what order) | 17:19 |
| notmyname | peluse_: google+? | 17:20 |
| notmyname | oh! you finally got it working | 17:20 |
| peluse_ | notmyname: with my ASU account - not sure what Intel has against google+ but my intel.com email won't work there | 17:25 |
| notmyname | heh. maybe google is using AMD or something ;-) | 17:26 |
| peluse_ | hah! | 17:26 |
| peluse_ | something tells my its the IT security "police" | 17:27 |
| *** CrackerJackMack has quit IRC | 17:27 | |
| *** foexle has quit IRC | 17:30 | |
| clayg | peluse_: the start of the change is some bullshit test fix for some times when I tried to pretend POLICIES is a list | 17:31 |
| clayg | peluse_: after that just follow the "neededby" links | 17:32 |
| clayg | peluse_: https://review.openstack.org/#/c/88150/ | 17:32 |
| clayg | I think everything up to Merge container storage_policy_index values should be done | 17:32 |
| *** mlipchuk has joined #openstack-swift | 17:35 | |
| peluse_ | clayg: so its ready to start looking at now? | 17:36 |
| clayg | yeah i'm done with those for awhile | 17:36 |
| peluse_ | claygy: cool, I'll get some lunch and a few beers and start looking :) | 17:37 |
| notmyname | prerequisites for reviewing all code ;-) | 17:38 |
| *** CrackerJackMack has joined #openstack-swift | 17:47 | |
| *** devananda has joined #openstack-swift | 17:47 | |
| devananda | hi ya'll - we're looking to add swift tempurl support in Ironic | 17:48 |
| notmyname | devananda: cool! | 17:48 |
| devananda | and i'm wondering if anyone would have a little time to review the proposal / code | 17:48 |
| notmyname | devananda: got a link? | 17:48 |
| devananda | notmyname: https://blueprints.launchpad.net/ironic/+spec/swift-temp-urls | 17:48 |
| *** fbo is now known as fbo_away | 17:48 | |
| notmyname | devananda: not a lot of detail there. do you have some specific questions about it? | 17:49 |
| notmyname | devananda: that is, yes, it describes how tempurls work :-) | 17:49 |
| devananda | heh | 17:50 |
| devananda | mostly whether it looks like we're using them sanely | 17:50 |
| devananda | duplicating code or using libs where approrpiate | 17:50 |
| devananda | etc | 17:50 |
| devananda | *that* we want to use tempurls is clear. it's more a question of: is this the right way to use them | 17:50 |
| notmyname | is the "conductor" a glance thing or something in ironic? | 17:51 |
| devananda | ironic-conductor is the service that performs mgmt tasks on bare metal nodes | 17:51 |
| notmyname | ok, so the conductor sends a signed URL referencing the swift cluster to the workers on the bare metal nodes? | 17:51 |
| devananda | yep | 17:52 |
| devananda | that worker (ironic-python-agent) then downloads the image and writes it to disk | 17:52 |
| notmyname | sounds like a great use case | 17:52 |
| devananda | whereas today, we're passing a keystone auth token | 17:52 |
| notmyname | there's a couple of things to think about | 17:52 |
| notmyname | (which is why you're here, I think) | 17:52 |
| notmyname | glance? hot content. per user keys.(making a list so I don't forget one) | 17:53 |
| *** ChanServ sets mode: +v dfg | 17:54 | |
| notmyname | devananda: I don't know enough about glance to know if ironic needs to go through glance to get images or not. it sounds like you don't want that in your use case, but I don't know if it's required in the architecture | 17:54 |
| notmyname | devananda: ie does glance provide the pass-through for the images, or does glance simply tell the client where to go get stuff? | 17:55 |
| *** changbl has joined #openstack-swift | 17:56 | |
| devananda | notmyname: so that's an aspect of this I do not fully undersatnd, but, there's a doc string that seems to cover it here | 17:56 |
| devananda | https://review.openstack.org/#/c/81391/13/ironic/common/glance_service/v2/image_service.py | 17:56 |
| devananda | notmyname: and i'm hoping you know more about that than I :) | 17:57 |
| notmyname | nope :-) | 17:57 |
| devananda | heh | 17:57 |
| notmyname | devananda: does glance hand bakc the image or a URL to the image? | 17:57 |
| notmyname | hmm...maybe that doesn't matter | 17:58 |
| devananda | in the existing PXE driver, we aer passed a glance image id from Nova, and then actually fetch it from glance | 17:58 |
| devananda | this is architecturally different | 17:58 |
| notmyname | devananda: either way, glance is the one generating the tempurl, not ironic | 17:58 |
| notmyname | right? | 17:58 |
| devananda | no | 17:58 |
| notmyname | ok | 17:58 |
| devananda | direct_url = image_info['properties']['direct_url'] | 17:59 |
| notmyname | ok | 17:59 |
| devananda | ... return ('http://{host}/{url}?temp_url_sig={sig}&temp_url_expires=' | 17:59 |
| notmyname | ok | 17:59 |
| devananda | so irnoic is taking the glance direct_url, if present, and creating a swift tempurl from it | 17:59 |
| notmyname | glance returns the template, not the actual url? | 17:59 |
| *** bach has joined #openstack-swift | 18:00 | |
| devananda | there's a bunch of url parsing in that file that seems to be related to the structure of a swift tempurl | 18:00 |
| devananda | so i think glance is returning a "direct url" but not a tempurl | 18:00 |
| notmyname | is glance filling in the value of {sig}? | 18:00 |
| devananda | lemme pull josh in | 18:01 |
| chmouel | notmyname: with the last patches i have sent tests pass under py3 | 18:01 |
| notmyname | chmouel: yay | 18:01 |
| *** piyush has joined #openstack-swift | 18:01 | |
| *** zaitcev has joined #openstack-swift | 18:01 | |
| *** ChanServ sets mode: +v zaitcev | 18:01 | |
| chmouel | notmyname: when we get those merged we will maybe talk about having the gates running the py3 unitttests | 18:01 |
| *** JayF has joined #openstack-swift | 18:01 | |
| notmyname | chmouel: IIRC it already does, just non-voting right? | 18:02 |
| *** bach has quit IRC | 18:02 | |
| chmouel | notmyname: that's correct, | 18:02 |
| *** jroll has joined #openstack-swift | 18:02 | |
| *** JoshNang has joined #openstack-swift | 18:02 | |
| notmyname | ok. agreed that flippingthe voting switch is good | 18:02 |
| chmouel | fyi: the patches is listed here https://wiki.openstack.org/wiki/Python3/SprintPycon2014#Synchronize_work_on_OpenStack_components | 18:02 |
| devananda | JoshNang: hi! so in reading through https://review.openstack.org/#/c/81391/13/ironic/common/glance_service/v2/image_service.py I dont actually see anything communicating with swift. should I ? | 18:03 |
| chmouel | and that's the last one of the series https://review.openstack.org/#/c/88351/ which has the py3 unittests passing | 18:03 |
| JoshNang | devananda: from my understanding of how swift temp urls work, you don't need to contact swift. you have an object that exists, a set of methods, and a shared secret. you use the shared secret to sign a constructed URL with a TTL, and when swift gets a call to that URL, it compares the given signed URL with its shared secret and either allows or 401s the request | 18:04 |
| notmyname | JoshNang: devananda and I are talking about tempurl support in ironic, and glance is mentioned in the ironic design proposal | 18:04 |
| notmyname | JoshNang: devananda: correct | 18:04 |
| devananda | JoshNang: ah, cool. | 18:05 |
| notmyname | JoshNang: does glance generate the tempurl signature or does ironic? | 18:05 |
| JoshNang | notmyname: awesome. i'd love feedback on it. | 18:05 |
| JoshNang | ironic generates the tempurl as it stands | 18:05 |
| notmyname | JoshNang: devananda: and then ironic requests glance to fetch that tempurl? | 18:06 |
| JoshNang | sort of. it's passed to an agent running elsewhere and then fetched directly from swift. | 18:06 |
| JoshNang | it gets the original image info from glance before hitting the temp url function | 18:07 |
| notmyname | JoshNang: devananda: here's how I would imagine things working: | 18:07 |
| JoshNang | and uses the direct_url in that image info to construct it | 18:07 |
| jroll | JoshNang: well, ironic fetches the tempurl itself, and passes that to the agent | 18:07 |
| notmyname | 1) Ironic gets a list of images from glance for the current user | 18:07 |
| notmyname | 2) ironic gets, from glance, the tempurl to directly access the image | 18:08 |
| notmyname | 3) ironic passes that tempurl to the workers | 18:08 |
| notmyname | 4) the workers each directly download the image from swift | 18:08 |
| notmyname | (steps one and two should be combined) | 18:08 |
| notmyname | JoshNang: devananda: does that make sense? | 18:08 |
| jroll | (I'm wrong about that, ignore me) :) | 18:08 |
| devananda | notmyname: i'd replace (1) with: ironic gets the image metadata from nova for the specific image the user is requesting | 18:09 |
| JoshNang | notmyname: totally. | 18:09 |
| *** joeljwright1 has quit IRC | 18:09 | |
| JoshNang | ^ +1 | 18:09 |
| devananda | notmyname: re: 2, it sounds like you are suggesting that ironic not generate the swift tempurls, but leverage glance for that? | 18:09 |
| JoshNang | and as a follow up re: 2, is that currently supported in glance? | 18:10 |
| notmyname | devananda: yes, but that's also based on my idea of what I think glance is. not necessarily what glance actually does | 18:10 |
| devananda | heh | 18:10 |
| notmyname | and I don't know what's in metadata that nova has about the image, or why nova does that (again, my understanding of what I think glance is/does) | 18:10 |
| devananda | sounds like we need to ask the glance folks if they provide an API to get swift tempurls | 18:11 |
| notmyname | my understanding of glance is that it tracks metadata about system images. an indexer of sorts | 18:11 |
| notmyname | devananda: ok | 18:11 |
| notmyname | devananda: as to the other things I was thinking of (glad I made a list), I think you--or glance, whoever is generating it--should support per-user keys, not per-system keys. | 18:12 |
| devananda | notmyname: as an aside, is there any reason why ironic should not generate swift tempurls? | 18:12 |
| devananda | notmyname: oh, good point, except ironic does not today support users -- one needs to be a cloud admin to talk to the API at all | 18:13 |
| notmyname | devananda: I have no idea about "should", but ironic certainly "could" generate tempurls (it's all done locally based on a shared secret) | 18:13 |
| devananda | notmyname: we dont currently track "user" or "project" | 18:13 |
| notmyname | devananda: ok. then don't build a schema today that locks you in :-). tempurls are per-account (swift account, not user identity), and therefore there are N of them in a given swift cluster | 18:14 |
| devananda | in part because of the security ramifications of allocating a physical machine, that was used by user A, to user B, without being able to guarantee a firmware level wipe | 18:15 |
| notmyname | devananda: JoshNang: also, FWIW, tempurls actually support 2 secrets to allow for key rotation with no interruption. while not necessary in your first implementation, it's something you can take advantage of | 18:15 |
| devananda | notmyname: good to know | 18:15 |
| JoshNang | notmyname: definitely something i've got in mind before going to production with anything :) | 18:15 |
| *** haomaiwang has quit IRC | 18:15 | |
| *** haomaiwang has joined #openstack-swift | 18:16 | |
| notmyname | devananda: so in this case, the source swift account is a "system-level" account? ie write access would only be granted to a cluster admin | 18:17 |
| devananda | notmyname: correct | 18:17 |
| notmyname | devananda: there are 2 reasons to use tempurls. 1) to give time-limited read access without needing to pass credentials around and 2) to avoid auth requests on every ironic worker when fetching the image | 18:18 |
| devananda | (1) is our main reason here, IMO | 18:18 |
| notmyname | ok | 18:18 |
| devananda | passing a full "system level" keystone token over DHCP or TFTP is really not ideal | 18:19 |
| JoshNang | heh | 18:19 |
| notmyname | :-) | 18:19 |
| notmyname | devananda: so it seems like you have some research to do on how the ironic/glance interaction will work | 18:19 |
| devananda | JoshNang: can you follow up with the glance team on ^ ? | 18:20 |
| devananda | JoshNang: aside from that, eerything else is sounding good to me | 18:20 |
| JoshNang | devananda: certainly. if i can just get it from glance without having to do it myself and mess with direct_url, i'd be quite happy :) | 18:20 |
| notmyname | devananda: cool. let me shatter those happy feelings with the last thing on my list: hot content :-) | 18:20 |
| devananda | yay! what's that :) | 18:20 |
| notmyname | devananda: if you are storing and accessing one image, it's going to be stored in swift on 3 spindles (assuming 3 replicas and a <5GB image). therefore, accessing it from 1000 ironic workers all at once could pretty easily saturate those spindles' IO | 18:22 |
| notmyname | not to mention the network | 18:22 |
| devananda | notmyname: swift does not use the FS cache, correct? | 18:22 |
| notmyname | devananda: swift can be configured to use it. but caching large objects probably isn't a good idea, nor too effective in practice | 18:23 |
| devananda | notmyname: as far as network IO, there is some work (or at least discussion of work) for doing multicast streaming from glance | 18:23 |
| notmyname | that's cool | 18:23 |
| notmyname | devananda: there are a couple of things that can be done in a production system to help mitigate this issue. in increasing order of difficulty, here are 4 ideas | 18:24 |
| notmyname | 1) don't ask for the image all at once. use some splay function | 18:24 |
| devananda | notmyname: and I would like the same mode of authentication to be applicable to both, that is, using a tempurl, rather than passing a whole keystone token | 18:24 |
| notmyname | 2) use a cache layer like varnish | 18:24 |
| devananda | JoshNang: so if glance can continue to act as the front-end for the tempurl, that would seem to be beneficial when multicast starts being implemented | 18:25 |
| notmyname | 3) deploy different copies of the image and request different ones | 18:25 |
| devananda | 3) yuck. hard for deplyers to keep track of that | 18:26 |
| *** mariusv has joined #openstack-swift | 18:26 | |
| notmyname | 3a) write a "rand-ifest" feature that works like the current manifests but instead chooses a random one in the manifest | 18:26 |
| devananda | 2) seems straight forward and easy to me | 18:26 |
| notmyname | 4) write a "seed a torrent" feature in swift | 18:26 |
| devananda | 4) yes! JoshNang you guys want to do that, right? ;) | 18:27 |
| JoshNang | devananda: notmyname 4) yes! | 18:27 |
| notmyname | devananda: JoshNang: awesome! when can I expect to see the swift patches from you? ;-) | 18:27 |
| notmyname | devananda: 3a solves the "deployer yuk" part of different copies, but not the issue of storing many copies of the same data and using the extra space (but FWIW, space storage normally isn't the concern with storing a library of common images) | 18:28 |
| JoshNang | :) i'm not sure how we were planning on implementing yet, but i'd love to upstream it when we do (rather than just implementing it as a feature of the agent, for example) | 18:28 |
| notmyname | JoshNang: there are some people here who have looked at it, but AFAIK haven't yet solved it | 18:29 |
| devananda | JoshNang: rather than the agent fetch-then-seed, if swift provides the seed, the agent ought to be somewhat simpler | 18:29 |
| devananda | *agent code for this ought ... | 18:29 |
| JoshNang | devananda: agreed. and i like that. | 18:30 |
| notmyname | JoshNang: devananda: but for the realistic solution, it may be adequate to add docs that say "if you are using more than XX nodes, put a cache layer in front of swift for the hot content" | 18:30 |
| JoshNang | notmyname: definitely | 18:30 |
| devananda | ++ | 18:30 |
| devananda | JoshNang: fwiw, that is a drawback of the agent approach. the PXE driver definitely uses the FS local cache on teh conductors to mitigate that | 18:31 |
| notmyname | devananda: JoshNang: I'd love to see torrent support in swift, but it's a tricky problem, from what I hear. or at least it adds a lot of complexity | 18:31 |
| notmyname | it's in my "someday" bucket of stuff for swift | 18:31 |
| JayF | devananda: sometimes it's easy to forget that not everyone gets a pre-scaled awesome swift cluster to use :) | 18:31 |
| zaitcev | torrent (4) is not going to work, because torrent only works when peers particiapte, but in our cases as soon as workers download the images they are going either to move on or to modify the images or discard the images. | 18:31 |
| JoshNang | notmyname: i'll do some digging into it either after lunch or next week | 18:32 |
| devananda | zaitcev: not true with the agent that is being added to ironic | 18:32 |
| zaitcev | The randifest (3) is intriguing in its cleverness, but not helping network choke at proxy | 18:32 |
| JayF | fwiw; also, the torrent based approach adds more credence to the desire to have long running agents. They can keep seeding while waiting to be provisioned :) | 18:32 |
| JoshNang | well, in our case, the workers won't in all cases go directly into booting. they may sit around seeding for some amount of time | 18:32 |
| devananda | zaitcev: it can be long-lived and continue seeding | 18:32 |
| notmyname | JayF: :-) I like your ideas and wish to subscribe to your newsletter | 18:32 |
| devananda | JayF: yep. this may be the best argument for long-lived agents taht i've seen yet :) | 18:33 |
| devananda | JayF: adding some logic to ConductorManager to guage how many agents to pre-warm // keep seeding is going to be interesting.... | 18:33 |
| notmyname | devananda: FYI, we (swiftstack) had a customer with this exact problem and the solution was to store multiple copies of the image in swift | 18:33 |
| notmyname | ie with different names | 18:34 |
| devananda | notmyname: sure. but i dont think that's a general solution eg. for folks using this to install their cloud | 18:34 |
| devananda | (ie tripleo) | 18:34 |
| devananda | (3a) would be fine | 18:34 |
| notmyname | devananda: right. just color commentary and background on a near-term solution | 18:34 |
| notmyname | devananda: 3a would actually be pretty easy to write | 18:35 |
| notmyname | devananda: JoshNang: questions answered? anything else? | 18:36 |
| JoshNang | that works for me! i'll dive into bittorrent when i get a chance and talk to the glance guys after lunch. thanks for your input!! | 18:37 |
| *** gyee has joined #openstack-swift | 18:37 | |
| *** piyush has left #openstack-swift | 18:37 | |
| notmyname | glad to help | 18:38 |
| * notmyname steps away from the computer for a bit | 18:38 | |
| *** chandan_kumar has joined #openstack-swift | 18:43 | |
| devananda | notmyname: thanks much! | 18:43 |
| * devananda steps away as well | 18:43 | |
| *** kevinc_ has quit IRC | 18:49 | |
| *** bach has joined #openstack-swift | 18:50 | |
| *** bach has quit IRC | 18:55 | |
| *** pberis has joined #openstack-swift | 19:00 | |
| *** bach has joined #openstack-swift | 19:08 | |
| *** baojg has quit IRC | 19:27 | |
| *** baojg has joined #openstack-swift | 19:27 | |
| *** baojg has quit IRC | 19:32 | |
| *** annegentle has quit IRC | 19:33 | |
| openstackgerrit | A change was merged to openstack/swift: Fix test-requirements.txt https://review.openstack.org/88363 | 19:38 |
| *** kevinc_ has joined #openstack-swift | 19:39 | |
| notmyname | hmm...so with that test-requirement change, I guess that's pretty strong evidence that the gate isn't actually using swift's specified dependencies | 19:42 |
| notmyname | well, no, never mind. the gate would have simply installed mock, got the latest one, and used it. | 19:42 |
| creiht | notmyname: well obviously the gate needs to start testing every combination of every version of library supported ;) | 19:43 |
| notmyname | and interesting set of tests, then, would be to run against the earliest specified version and the lated allowed version | 19:43 |
| *** bach_ has joined #openstack-swift | 19:45 | |
| *** zhiyan_ is now known as zhiyan | 19:46 | |
| *** bach_ has quit IRC | 19:47 | |
| *** bach_ has joined #openstack-swift | 19:48 | |
| *** bach has quit IRC | 19:50 | |
| *** bach_ has quit IRC | 19:54 | |
| *** bach has joined #openstack-swift | 19:55 | |
| *** ashish_ has joined #openstack-swift | 19:55 | |
| *** annegentle has joined #openstack-swift | 20:04 | |
| shri | hi... | 20:12 |
| shri | Is there a list of talks that'll be part of the design summit next month? | 20:12 |
| swills | so i finally got back to looking at swift, and running the tests on my zfs install... | 20:17 |
| swills | i'm using 1.13.0... tests are still running, cause my box is slow, but so far only a few errors. | 20:17 |
| *** chandan_kumar has quit IRC | 20:22 | |
| swills | Ran 233 tests in 1088.017s | 20:29 |
| swills | FAILED (SKIP=2, errors=16) | 20:29 |
| *** kevinc_ has quit IRC | 20:31 | |
| swills | box was too busy it seems, some of the failures were just timesouts, rerunning... | 20:34 |
| *** ashish_ has quit IRC | 20:42 | |
| swills | hmm, still some issues | 20:42 |
| *** occupant has joined #openstack-swift | 20:42 | |
| swills | http://meatwad.mouf.net/~swills/swift_tests.txt | 20:42 |
| swills | creiht: ideas on these test failures? | 20:42 |
| *** bach has quit IRC | 20:47 | |
| *** nacim has joined #openstack-swift | 20:50 | |
| *** kevinc_ has joined #openstack-swift | 20:51 | |
| *** Anju_ has quit IRC | 20:51 | |
| *** lpabon has quit IRC | 20:52 | |
| openstackgerrit | Pete Zaitcev proposed a change to openstack/swift: Plumb insecure flag to functests https://review.openstack.org/88438 | 21:05 |
| *** fifieldt has quit IRC | 21:09 | |
| zaitcev | swills: The only way I know to get to the bottom is to add print(resp.body) in places where these 503 happen. That ought to save the tracebacks. Assuming the failures are repeatable, of course. | 21:12 |
| swills | zaitcev: ok, thanks. i'm thinking they are simply timeouts because things are slow. i am running 6 jails on one box so... | 21:16 |
| swills | that is, all the swift instances are on the same box | 21:16 |
| swills | i did get some tracebacks in logs tho | 21:17 |
| swills | i can show them if you want | 21:17 |
| zaitcev | swills: Overloaded nodes aren't not going to work, because Swift uses 10s timeouts in backends. Going that low is necessary to switch to a working backend if some nodes go down before client times out, and some clients do 30s. | 21:18 |
| swills | oh, i'm getting some things timeing out >10 sec... | 21:18 |
| swills | http://paste.mouf.net/pastes/nuhhemerjto74v0l8fuirj8j | 21:18 |
| swills | so perhaps all my nodes are timing out | 21:19 |
| *** mlipchuk has quit IRC | 21:20 | |
| *** fifieldt has joined #openstack-swift | 21:21 | |
| *** changbl has quit IRC | 21:21 | |
| *** kevinc_ has quit IRC | 21:24 | |
| *** ashish_ has joined #openstack-swift | 21:29 | |
| *** rukawz has joined #openstack-swift | 21:31 | |
| rukawz | hi, I just used put_object method from python-swiftclient and I wanted to know what is the best way to retrieve the URI of the object I just created | 21:32 |
| *** nacim has quit IRC | 21:33 | |
| *** mkollaro1 has joined #openstack-swift | 21:33 | |
| *** mkollaro has quit IRC | 21:33 | |
| *** bach has joined #openstack-swift | 21:34 | |
| rukawz | put_object just returns a ETag, I was wondering if I could get the URI used to PUT in the headers | 21:34 |
| Alex_Gaynor | rukawz: it's {account name}/{container name}/{object name} I don't think there's a particular API to retrieve this though | 21:34 |
| *** ashish_ has quit IRC | 21:34 | |
| rukawz | ok | 21:35 |
| rukawz | thank you, so I'll have to build myself the URI, is the storage_url available in service endpoints ? | 21:36 |
| chmouel | has anyone plugged serf into swift for managing ring management push? (serf: http://www.serfdom.io/) | 21:39 |
| *** mkollaro has joined #openstack-swift | 21:39 | |
| *** erlon has quit IRC | 21:39 | |
| *** mkollaro1 has quit IRC | 21:39 | |
| creiht | swills: you would have to look at the logs to see why they returned a 503 | 21:43 |
| creiht | oh looks like you already got that sorted out | 21:44 |
| *** bach has quit IRC | 21:44 | |
| *** pberis has quit IRC | 21:45 | |
| *** kevinc_ has joined #openstack-swift | 21:46 | |
| *** bach has joined #openstack-swift | 21:48 | |
| *** bach_ has joined #openstack-swift | 21:49 | |
| *** marcusvrn has quit IRC | 21:50 | |
| *** bach_ has quit IRC | 21:51 | |
| *** bach has quit IRC | 21:53 | |
| *** bach has joined #openstack-swift | 21:55 | |
| *** foexle has joined #openstack-swift | 21:56 | |
| *** _80386sx has quit IRC | 21:58 | |
| *** mariusv has quit IRC | 21:58 | |
| *** sungju has joined #openstack-swift | 21:59 | |
| *** mariusv has joined #openstack-swift | 21:59 | |
| *** mariusv has quit IRC | 21:59 | |
| *** mariusv has joined #openstack-swift | 21:59 | |
| *** kevinc_ has quit IRC | 22:01 | |
| *** bach has quit IRC | 22:03 | |
| *** bach has joined #openstack-swift | 22:04 | |
| *** rukawz has left #openstack-swift | 22:04 | |
| *** kevinc_ has joined #openstack-swift | 22:04 | |
| zaitcev | portante: I was looking everywhere for that sample config | 22:09 |
| *** bach has quit IRC | 22:10 | |
| openstackgerrit | Pete Zaitcev proposed a change to openstack/swift: Plumb insecure flag to functests https://review.openstack.org/88438 | 22:14 |
| *** zhiyan is now known as zhiyan_ | 22:29 | |
| *** sungju has quit IRC | 22:48 | |
| swills | creiht: yeah, i guess the issue is just that my box is too slow, not that anything is not working? | 22:53 |
| notmyname | chmouel: that looks pretty cool. I haven't used it, but it looks useful | 22:58 |
| *** mkollaro has quit IRC | 22:59 | |
| *** Trixboxer has quit IRC | 23:07 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!