*** lyrrad has quit IRC | 00:35 | |
*** acoles_ is now known as acoles | 07:14 | |
*** openstack has joined #swift3 | 11:33 | |
*** openstackgerrit has quit IRC | 11:35 | |
*** openstackgerrit has joined #swift3 | 11:51 | |
*** kota_ has quit IRC | 13:43 | |
*** bill_az has joined #swift3 | 13:46 | |
*** ajiang has quit IRC | 14:05 | |
*** ajiang has joined #swift3 | 14:05 | |
*** bill_az has quit IRC | 14:38 | |
*** openstack has joined #swift3 | 16:10 | |
*** lyrrad has quit IRC | 16:32 | |
*** lyrrad has joined #swift3 | 16:58 | |
*** cschwede has quit IRC | 18:39 | |
*** openstackgerrit has quit IRC | 18:39 | |
*** cschwede has joined #swift3 | 18:45 | |
*** openstackgerrit has joined #swift3 | 18:55 | |
*** acoles is now known as acoles_ | 19:17 | |
*** acoles_ is now known as acoles | 19:17 | |
*** acoles is now known as acoles_ | 19:23 | |
*** timburke_ has joined #swift3 | 20:40 | |
*** hrou_ has joined #swift3 | 20:44 | |
*** timburke has quit IRC | 20:47 | |
*** hrou has quit IRC | 20:47 | |
*** acoles_ is now known as acoles | 20:58 | |
*** bill_az_ has joined #swift3 | 20:59 | |
*** kota_ has joined #swift3 | 21:46 | |
*** timburke_ is now known as timburke | 21:48 | |
*** acoles is now known as acoles_ | 21:51 | |
*** kota_ has quit IRC | 22:00 | |
*** kota_ has joined #swift3 | 22:15 | |
kota_ | hello | 23:00 |
---|---|---|
kota_ | someone? | 23:00 |
kota_ | timburke? | 23:01 |
timburke | hi! sorry, forgot to watch the time | 23:01 |
kota_ | no worries, and thanks for your ack :-) | 23:02 |
kota_ | anyone else, here? | 23:02 |
kota_ | ok, let get started... | 23:03 |
bill_az_ | Hi Kota - Bill here | 23:03 |
kota_ | bill_az!!! | 23:03 |
bill_az_ | Sorry I've not been joining - but want to check in today - hope all is well! | 23:03 |
kota_ | bill_az_: welcome and thanks for joining :) | 23:03 |
kota_ | the agenda is at https://wiki.openstack.org/wiki/Meetings/swift3 | 23:04 |
kota_ | i think today we have some important topics. | 23:04 |
kota_ | fist one, signature v4. | 23:05 |
timburke | that's actually the reason i wasn't watching the time :) looking at your patch now | 23:05 |
kota_ | i think we've spent a lot of time to get progress for the patches. | 23:06 |
kota_ | timburke: awesome, thanks. | 23:06 |
bill_az_ | I took a look earlier - great work! | 23:06 |
kota_ | and, imo it is getting good shape so maybe closely we can land the patch. | 23:06 |
kota_ | I put 3 patches into agenda. | 23:06 |
kota_ | that is because I'd like to hear opinions for some my concerns. | 23:07 |
kota_ | first, | 23:07 |
kota_ | As you can see the first patch in the wiki, we need to bump the keystone version | 23:08 |
kota_ | to enable sigv4 stuff. | 23:08 |
timburke | oh right! there's been a release that has support, so we don't have to grab tip of development! love it | 23:08 |
kota_ | timburke: good to hear | 23:09 |
kota_ | so the minimum version will be 9.0.0 | 23:09 |
kota_ | for keystone. | 23:09 |
kota_ | however, we cannot make the requirement because swift can allow the development model w/o keystone. | 23:09 |
kota_ | i.e. we cannot write "keystone>=9.0.0" in requirements.txt | 23:10 |
kota_ | just make a note into somewhere (conf file or commit log) | 23:10 |
bill_az_ | what is the dependency in mitaka keystone? | 23:11 |
kota_ | keystone-mitaka is 9.0.0 | 23:11 |
kota_ | released around a few weeks ago. | 23:11 |
bill_az_ | but why does sigv4 need version? | 23:12 |
kota_ | bill_az_: good question | 23:12 |
timburke | previous versions of keystone didn't have v4 signature support. it was added in https://github.com/openstack/keystone/commit/f11d396 | 23:13 |
kota_ | sig v4 requires new algorithms for header parsing and signature calculation. | 23:13 |
kota_ | yeah, that patch adds the new algorithms. | 23:13 |
bill_az_ | ahh - ok - got it | 23:14 |
kota_ | for now, the sig v4 is working only with keystone and we need continous work to improve (maybe with tempauth?) somewhat. | 23:15 |
bill_az_ | In the ibm os distribution we had been using, the s3 code was pulled out of keystone - we packaged separately - but I got it now :-) | 23:15 |
kota_ | bill_az_: ah, it might be affected. | 23:16 |
kota_ | circle back to my first question, can we make it to land with the keystone dependency? | 23:17 |
kota_ | just run functional test in gerrit gate with keystone 9.0.0 | 23:17 |
kota_ | i.e. we are changing default support keystone into 9.0.0 | 23:19 |
timburke | for our tests? absolutely. for deployers, i think documentation should be sufficient; this isn't worth adding to the extras section in setup.cfg | 23:19 |
kota_ | ah, good catch. need to change setup.cfg too. | 23:19 |
kota_ | ok, thanks | 23:19 |
kota_ | moving on to second concern. | 23:20 |
kota_ | I put 3rd patch in the wiki to make sure what's changed with keystone conversation and amazingly, keystone is no care about the region information v4 stuff using for authentication. | 23:21 |
timburke | yep. for better or worse, keystone doesn't really care about the region in the request | 23:22 |
kota_ | in the original patch for v4 andrey wrote, he said "we need to set region name with same value of keystone setting" | 23:22 |
kota_ | i'm not sure for now, but I'm douting that it is really reasonable for authentication (or not) | 23:23 |
timburke | this seems consistent with what region-less auth systems will do, though; swauth and tempauth don't have that concept (as far as i know) | 23:23 |
kota_ | yeah, an user should use the region name which is same value of *swift3* setting. | 23:25 |
kota_ | timburke: exactly, if we go this way, it will be easy to apply v4 calculation to any other auth system which don't have the concept | 23:26 |
kota_ | hmm | 23:26 |
kota_ | but some auth info bleed to swift side. | 23:27 |
kota_ | s/some/an/ | 23:27 |
kota_ | because I'm not security guy, I'm not sure it's ok swift3 keeps authenticaion info itself. | 23:28 |
timburke | i think i'm fine with it; i could see an argument for region being part of authorization (and so swift3's responsibility, at least with s3_acl enabled) rather than authentication (keystone's responsibility) | 23:30 |
timburke | i'd be open to reconsidering that if we ever had some sort of multi-region support, but i'm not sure how we'd do that currently | 23:30 |
kota_ | timburke: thanks, and ok keep it in my mind. For now, we can keep the current way. | 23:31 |
kota_ | that's all for sigv4 from me. | 23:32 |
kota_ | anything else for this topic? | 23:32 |
timburke | don't think so. i think i'll have comments ready for you later this week | 23:33 |
kota_ | ok, let's moving to the next. | 23:33 |
kota_ | Release timing | 23:33 |
kota_ | I think, we don't have to make a version bump *before* sigv4 stuff landing | 23:33 |
kota_ | because we have really a few things to land since last updated (sorry slow reviews) | 23:34 |
bill_az_ | That seems reasonable to me - v1.11 has sigv4 + small stuff | 23:34 |
timburke | sounds good | 23:34 |
kota_ | k, we will make the release after sigv4 stuff done. | 23:35 |
kota_ | and the last topic today from me is Austin summit. | 23:35 |
timburke | i'll be there! | 23:36 |
kota_ | Summit will happen in the week since 25 Apr. | 23:36 |
kota_ | timburke: great! | 23:36 |
kota_ | bill_az_: how about you? | 23:36 |
bill_az_ | Yes - I will be there! | 23:37 |
kota_ | great! | 23:37 |
kota_ | so my question is that do we need to make an extra time (e.g. mini design summit) for swift3 specific? | 23:38 |
kota_ | I'll be there too | 23:38 |
kota_ | *BUT* | 23:38 |
bill_az_ | I would like to meet up there - | 23:38 |
kota_ | i'll stay really short time for this time (flight in Monday and flyout Friday morning) | 23:38 |
kota_ | bill_az_: good to hear, me too. | 23:39 |
bill_az_ | could we meet on Wed sometime? or thurs? (I will leave on Thursday late afternoon) | 23:39 |
timburke | i'll be there all week; i think any time that works for you guys will be fine for me | 23:40 |
bill_az_ | timburke: looking forward to meeting you! | 23:40 |
kota_ | yup, maybe we have a hole slot of Swift design summit in Wed? | 23:41 |
bill_az_ | do you know if lyrrad will be attending? | 23:41 |
kota_ | it might be swap with other project, not sure. | 23:41 |
timburke | unfortunately, lyrrad won't be there. there'll be some other swiftstackers about, but not as involved in swift3 | 23:42 |
kota_ | that's too bad:/ | 23:42 |
timburke | wednesday sounds good to me | 23:42 |
kota_ | I'm missing design summit schedule slot | 23:42 |
kota_ | :/ | 23:42 |
bill_az_ | I wanted to check on the posting of results of ceph-s3 tests - he had done some work on that, but I don't know where it ended up | 23:43 |
kota_ | i think, timburke also is working on the stuff | 23:43 |
timburke | i can talk to notmyname about working us in; worst case, we can just go off in a corner sometime | 23:43 |
timburke | i have been! see patch 304367 | 23:44 |
kota_ | https://review.openstack.org/#/c/304367/ | 23:44 |
patchbot | timburke: https://review.openstack.org/#/c/304367/ - swift3 - WIP: add ceph-s3-tests tox env | 23:44 |
patchbot | kota_: patch 304367 - swift3 - WIP: add ceph-s3-tests tox env | 23:44 |
timburke | yeah :-) that one | 23:44 |
kota_ | same one :) | 23:44 |
kota_ | i think it's great stuff waiting for dropping WIP tag with coffee :) | 23:45 |
kota_ | ok, maybe swift3 meeting will happen at Wed, I'll check the free slot later and ping you all. | 23:46 |
bill_az_ | great - Thanks! | 23:46 |
timburke | i still need to look into getting that as a (non-voting) gate job, though. and i need to think about how to (1) use a known-failures file and (2) update the known-failures file on merges to master | 23:46 |
kota_ | bill_az_: do you want to send the ping via E-mail? or Is IRC enough for you? | 23:46 |
timburke | then we could turn it into a voting gate if we wanted | 23:46 |
bill_az_ | do you want to post it on the swift3 wiki? or if you email me I can add to wiki - billowen@us.ibm.com | 23:47 |
bill_az_ | I will be on IRC more these next weeks too | 23:48 |
kota_ | bill_az_: ok, I'll update the wiki and send the notification for you. | 23:48 |
kota_ | timburke: great | 23:48 |
bill_az_ | Thanks kota_: | 23:48 |
kota_ | timburke: sounds like really useful (I don't have the idea yet) | 23:50 |
kota_ | timburke: thanks for working on that :) | 23:51 |
kota_ | anything else for today? | 23:51 |
timburke | happy to; it's one of those useful tools i keep forgetting to use :P | 23:51 |
timburke | you probably noticed a flurry of patches from me recently | 23:51 |
timburke | i figure it might be nice to have a summary of what they were | 23:51 |
timburke | i've got a couple testing-infrastructure niceties (patch 294782, patch 294849) | 23:51 |
patchbot | timburke: https://review.openstack.org/#/c/294782/ - swift3 - Make Swift/Keystone logs available in gate results | 23:51 |
patchbot | timburke: https://review.openstack.org/#/c/294849/ - swift3 - Fix up some warnings | 23:51 |
timburke | a couple regressions on newer swift (patch 301976, patch 301937) | 23:52 |
patchbot | timburke: https://review.openstack.org/#/c/301976/ - swift3 - Catch segment-too-small errors from 2.6.0+ swift | 23:52 |
patchbot | timburke: https://review.openstack.org/#/c/301937/ - swift3 - Allow multipart uploads to have an empty final part | 23:52 |
timburke | and a little patch to prevent an exception in some clients (patch 302475) | 23:52 |
patchbot | timburke: https://review.openstack.org/#/c/302475/ - swift3 - Include '-' in multipart ETags | 23:52 |
timburke | i also noticed a behavioral difference between AWS and swift when deleting a key that doesn't exist; AWS returns 204 while swift (and therefore swift3) returns 404 | 23:52 |
timburke | don't have a patch for that one yet, though | 23:53 |
kota_ | ok, I will look at. | 23:54 |
kota_ | thanks for bringing up. | 23:54 |
kota_ | looking at the titles. all of them are reasonable. | 23:55 |
kota_ | anything? | 23:56 |
timburke | that's it for me | 23:57 |
kota_ | ok | 23:58 |
kota_ | bill_az_ ? | 23:58 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!