| ormandj | coreycb: thanks so much and sorry for the slow response! much appreciated | 00:12 |
|---|---|---|
| *** diablo_rojo has joined #openstack-swift | 00:32 | |
| *** manuvakery has joined #openstack-swift | 01:08 | |
| timburke | tdasilva, i'm starting to wonder if bulk needs to learn more about versions :-/ | 01:11 |
| timburke | mainly because SLO needs to learn more about versions | 01:12 |
| tdasilva | mm..was hoping that it would be transparent to bulk | 01:15 |
| tdasilva | but i'm assuming you looking somewhere around here: https://github.com/openstack/swift/blob/master/swift/common/middleware/slo.py#L1518 ? | 01:16 |
| timburke | yeah | 01:16 |
| timburke | new set of patches coming in a bit | 01:19 |
| tdasilva | k | 01:21 |
| *** diablo_rojo has quit IRC | 02:03 | |
| *** openstackgerrit has joined #openstack-swift | 03:02 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: New Object Versioning mode https://review.opendev.org/682382 | 03:02 |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Implement object versioning API https://review.opendev.org/673682 | 03:02 |
| timburke | unit tests are pretty busted, but i htink func tests should demonstrate where i want to go | 03:02 |
| *** manuvakery has quit IRC | 03:18 | |
| *** psachin has joined #openstack-swift | 03:38 | |
| *** mkrai has joined #openstack-swift | 04:19 | |
| *** evrardjp has quit IRC | 05:34 | |
| *** evrardjp has joined #openstack-swift | 05:34 | |
| *** gyee has quit IRC | 05:47 | |
| openstackgerrit | Tim Burke proposed openstack/swift stable/train: Allow bulk to fwd some headers at tar extraction https://review.opendev.org/704108 | 06:45 |
| openstackgerrit | Tim Burke proposed openstack/swift master: container-sync: Sync static links similar to how we sync SLOs https://review.opendev.org/698139 | 07:29 |
| openstackgerrit | Tim Burke proposed openstack/swift master: New Object Versioning mode https://review.opendev.org/682382 | 07:29 |
| openstackgerrit | Tim Burke proposed openstack/swift master: s3api: Implement object versioning API https://review.opendev.org/673682 | 07:29 |
| *** mkrai has quit IRC | 07:44 | |
| *** rdejoux has joined #openstack-swift | 08:08 | |
| *** mkrai has joined #openstack-swift | 08:09 | |
| *** tkajinam has quit IRC | 08:20 | |
| *** ccamacho has joined #openstack-swift | 08:20 | |
| *** pawan-gupta has quit IRC | 08:32 | |
| *** tesseract has joined #openstack-swift | 08:34 | |
| *** rpittau|afk is now known as ironic | 08:51 | |
| *** ironic is now known as rpittau | 08:51 | |
| *** evrardjp has quit IRC | 08:54 | |
| *** pcaruana has joined #openstack-swift | 09:14 | |
| *** mkrai has quit IRC | 11:14 | |
| *** rpittau is now known as rpittau|bbl | 11:28 | |
| *** mkrai has joined #openstack-swift | 11:33 | |
| *** mkrai has quit IRC | 11:53 | |
| *** pawan-gupta has joined #openstack-swift | 11:54 | |
| *** pawan-gupta_ has joined #openstack-swift | 11:54 | |
| *** pawan-gupta_ has quit IRC | 12:25 | |
| *** pawan-gupta has quit IRC | 12:26 | |
| *** pawan-gupta has joined #openstack-swift | 12:26 | |
| *** mkrai has joined #openstack-swift | 12:27 | |
| mkrai | HI, I am trying to find an openstack command to create an account in Swift, what is it? | 12:27 |
| seongsoocho | mkrai: Hi, Are you using keystone for swift auth? | 12:58 |
| mkrai | seongsoocho, Yes | 12:59 |
| DHE | then you usually don't need to. there's a proxy-server.conf setting to auto-create an account. | 13:00 |
| seongsoocho | mkrai: OK, Then, If account_autocreate=true in your proxy-server.conf, the account will be create automatically. | 13:00 |
| mkrai | seongsoocho, Ok and what if I want to create a new account manually? | 13:00 |
| mkrai | seongsoocho, Is there any command for it? | 13:00 |
| seongsoocho | mkrai: Maybe You can use REST API for account. The PUT http method. | 13:01 |
| DHE | is there a keystone "god admin" user rule? | 13:02 |
| DHE | because with keystone, you wouldn't have a means of accessing the account once created anyway unless you had such a user | 13:02 |
| DHE | I guess you could add some ACLs to some containers... | 13:02 |
| *** rpittau|bbl is now known as rpittau | 13:40 | |
| *** mkrai has quit IRC | 13:56 | |
| *** jvisser has joined #openstack-swift | 13:57 | |
| *** seongsoocho has quit IRC | 14:49 | |
| *** seongsoocho has joined #openstack-swift | 14:49 | |
| *** psachin has quit IRC | 14:52 | |
| *** coreycb has quit IRC | 14:57 | |
| *** coreycb has joined #openstack-swift | 14:57 | |
| *** mkrai has joined #openstack-swift | 15:04 | |
| *** gregwork has quit IRC | 15:55 | |
| *** gyee has joined #openstack-swift | 16:14 | |
| *** diablo_rojo has joined #openstack-swift | 17:03 | |
| *** rdejoux has quit IRC | 17:10 | |
| clayg | timburke: so did you already get into that unavailable versions container change? | 17:24 |
| *** tesseract has quit IRC | 17:27 | |
| *** rpittau is now known as rpittau|afk | 17:35 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: New Object Versioning mode https://review.opendev.org/682382 | 17:35 |
| timburke | clayg, yup. thanks for fixing up the pep8 fail! | 17:50 |
| clayg | i have the container-sync change checked out - i was just reading your notes | 17:50 |
| timburke | on https://review.opendev.org/#/c/698139/ -- that used to be the "Skip container-sync when versioning enabled" patch, but i pulled the versioning-specific stuff out to the versioning patch | 17:51 |
| patchbot | patch 698139 - swift - container-sync: Sync static links similar to how w... - 16 patch sets | 17:51 |
| clayg | I guess fixing syncing of the static links is helpful somewhat orthogonally to versioning | 17:51 |
| clayg | and the "don't container-sync versions" requires the symlink header shuffling to work the way we want? | 17:52 |
| timburke | in large part i wanted to get https://review.opendev.org/#/c/682382/78/swift/common/middleware/container_sync.py to be really crisp -- there was a bunch of other middleware re-org going on :-/ | 17:53 |
| patchbot | patch 682382 - swift - New Object Versioning mode - 78 patch sets | 17:53 |
| clayg | it was interesting that rledisez is saying he needs container-sync to work before he can make a move on versioning, since we kinda added the "don't be undefined with container-sync" knowing that it was an issue | 17:53 |
| clayg | ok, yeah that part looks great! | 17:53 |
| timburke | i think we also have been pretty clear that we *want* the features to play well together, we just don't want to block forward progress on it | 17:54 |
| clayg | well... I mean... to the extent we "want" container-sync to be a thing - my thinking on that rolls into bucket policies and pulling 1space out of crawling containers | 17:56 |
| clayg | as we get more and more aws compatible having versioning work with "the stuff formally known as container-sync" starts to be interesting again | 17:57 |
| clayg | maybe there's a middle step that's just "teach container-sync how to versions" | 17:57 |
| timburke | clayg, how do you feel about the 412 around https://review.opendev.org/#/c/682382/78/test/probe/test_object_versioning.py@204 ? i don't know what status i *want*, but i don't think that's it... | 18:05 |
| patchbot | patch 682382 - swift - New Object Versioning mode - 78 patch sets | 18:05 |
| clayg | any idea where the 412 is coming from? | 18:06 |
| timburke | i mean, the *hope* is that it's a transient server error and the versions container was just temporarily unavailable... | 18:06 |
| timburke | https://review.opendev.org/#/c/682382/78/swift/common/middleware/versioned_writes/object_versioning.py@389 | 18:07 |
| patchbot | patch 682382 - swift - New Object Versioning mode - 78 patch sets | 18:07 |
| clayg | oh, so it's very specific we can return whatever we want | 18:07 |
| timburke | yup. 412 was just carry-over from old-style versioning | 18:08 |
| clayg | yeah any 400 series is fine, 404 is probably about as good as 412 - if we can throw a body/message in there that's unique eventually clients would grow robust to re-enabling versioning as needed | 18:08 |
| clayg | well, i guess _check_response_error is somewhat generic, maybe a 404 there isn't always a missing versions container | 18:10 |
| timburke | yup -- see also https://review.opendev.org/#/c/682382/74/test/unit/common/middleware/test_object_versioning.py@1812 | 18:10 |
| patchbot | patch 682382 - swift - New Object Versioning mode - 78 patch sets | 18:10 |
| clayg | oic, yeah I think it'd be reasoable to change that - did it *used* to be a 404 before we special cased the 404 => 412 translation then? | 18:11 |
| timburke | i think it would've 500ed or 503ed, looking at the rest of _check_response_error | 18:13 |
| clayg | yeah so that code must have come in before or along with the repair-version-link api - cause it wouldn't make sense to server error | 18:27 |
| clayg | mayb we could pull the generic handling out and try to catch the 404's and return more context specific 400's | 18:28 |
| clayg | i'm still on p 698139 right now tho - I think it needs more unitests and I'm not sure I understand all the header forwarding that's going on (like static links to slos) | 18:30 |
| patchbot | https://review.opendev.org/#/c/698139/ - swift - container-sync: Sync static links similar to how w... - 16 patch sets | 18:30 |
| *** mkrai has quit IRC | 18:30 | |
| *** psachin has joined #openstack-swift | 18:47 | |
| clayg | timburke: i 100% agree with your take on http://lists.openstack.org/pipermail/openstack-discuss/2020-January/012158.html are you gunna respond then? | 18:49 |
| timburke | clayg, yeah, i can do that | 18:56 |
| clayg | timburke: awesome thanks for catching that one - i totally missed it! | 18:58 |
| *** mvkr has quit IRC | 19:00 | |
| *** psachin has quit IRC | 19:23 | |
| *** theintern_ has joined #openstack-swift | 19:27 | |
| *** theintern_ has quit IRC | 19:38 | |
| clayg | so I would have swore `echo '' | swift upload test -H 'x-symlink-target: test/test' --object-name link -` was working 🤔 | 19:41 |
| clayg | but I guess that's why @charz did p 694211 | 19:42 |
| patchbot | https://review.opendev.org/#/c/694211/ - python-swiftclient - Support uploading Swift symlinks without content. - 8 patch sets | 19:42 |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: container-sync: Sync static links similar to how we sync SLOs https://review.opendev.org/698139 | 19:47 |
| timburke | clayg, the echo adds a newline -- `< /dev/null swift upload test -H 'x-symlink-target: test/test' --object-name link -` should work, but it's definitely cumbersome | 20:09 |
| timburke | clayg, you gonna restack the versioning patches, or should i? | 20:17 |
| clayg | oh the newline!!! I've probably done echo -n or something | 20:17 |
| timburke | yeah, that'd work, too | 20:18 |
| clayg | I wasn't immediately working on a restack - it doesn't seem like these things are going to land imminantly | 20:18 |
| timburke | but really, we should probably just land the client patch -- i should look at that next | 20:18 |
| clayg | i'm not sure about that interface on the client patch 😬 | 20:18 |
| timburke | yeah, maybe. gotta remember what it's like | 20:19 |
| clayg | it's not very condusive to doing "batch" sorts of things like some of the other commands with their --prefix and the like | 20:19 |
| clayg | I think @tdasilva may have suggested it, or maybe @charz came up with it on his own | 20:19 |
| clayg | i can imagine migrating a replicated container <foo> to <foo>_ec and then wanting to symlink the lot of 'em | 20:20 |
| clayg | I guess as long as the service api is reasonable it's less important that we support that use-case from the command line | 20:20 |
| timburke | idk that there's a good way to make it more batch-y though -- reminds me that i once played around with writing a swiftclient with a mode that basically adds a DSL for scripting... | 20:22 |
| clayg | anyway - on the versioning patches, it seems like things are setup to want us to land container-sync pre-req first | 20:22 |
| clayg | so maybe wait for p 698139 to land and then rebase? | 20:22 |
| patchbot | https://review.opendev.org/#/c/698139/ - swift - container-sync: Sync static links similar to how w... - 17 patch sets | 20:22 |
| clayg | timburke: ultimately I think `swift copy` and `swift link` should have "about" the same api 🤷 | 20:22 |
| timburke | fair enough | 20:23 |
| timburke | anyway, if we can agree on what we want the missing-archive status(es) to be, i think the swift side at least is ready to go | 20:23 |
| clayg | ok, well then before we respin maybe I can pull that down and play with some different responses that might work see what looks good | 20:25 |
| timburke | i think there's a good chance we'll want to distinguish between 404-archive-is-missing and 404-version-id-not-found fwiw | 20:26 |
| clayg | timburke: yes i agree | 20:32 |
| clayg | I think some of the reason we ended up with 412 was wanting to distinguish between PUT-repair-versioning-symlink "container not found" vs "target version-id not found" | 20:33 |
| clayg | I think "status" might be too course grained and we just need to say 400 "read the body for more info" | 20:33 |
| clayg | but, idk | 20:33 |
| timburke | idk that what we have *makes* that distinction, tho :-/ | 20:36 |
| clayg | I think the 412 on "target version-id not found" is unambiguously NOT a 404 "container not found" 🤷 | 20:51 |
| clayg | but I think ultimately there's to many things going through the _check_resp_error | 20:51 |
| *** rdejoux has joined #openstack-swift | 21:28 | |
| timburke | clayg, so do we want sync_to in p 698139 or no? 'cause we've got some unit test failures to do with that... | 21:33 |
| patchbot | https://review.opendev.org/#/c/698139/ - swift - container-sync: Sync static links similar to how w... - 17 patch sets | 21:33 |
| *** diablo_rojo has quit IRC | 21:48 | |
| *** rdejoux has quit IRC | 21:51 | |
| *** rcernin has quit IRC | 22:34 | |
| openstackgerrit | Merged openstack/swift master: Allow Timestamp comparisons against out-of-range values https://review.opendev.org/702950 | 23:37 |
| *** mvkr has joined #openstack-swift | 23:38 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!