| seongsoocho | timburke: The swift Meeting logs still link to 2019. https://wiki.openstack.org/wiki/Meetings/Swift | 00:14 |
|---|---|---|
| *** diablo_rojo has quit IRC | 00:33 | |
| *** gyee has quit IRC | 00:48 | |
| *** xiaolin has joined #openstack-swift | 00:59 | |
| openstackgerrit | Merged openstack/swift master: Revert "Make rolling-upgrade job non-voting until virtualenv problem is solved" https://review.opendev.org/708719 | 01:02 |
| timburke | seongsoocho, thank! updated | 02:00 |
| *** openstackstatus has joined #openstack-swift | 02:56 | |
| *** ChanServ sets mode: +v openstackstatus | 02:56 | |
| *** psachin has joined #openstack-swift | 03:32 | |
| *** evrardjp has quit IRC | 03:42 | |
| *** evrardjp has joined #openstack-swift | 03:45 | |
| *** xiaolin has quit IRC | 03:54 | |
| kota_ | timburke: if you have any pointer for Vancouver Dev Event survey you asked in the meeting today, could you let me know? | 03:56 |
| kota_ | I checked my e-mail directory but I couldn't find any info like attendees, whether the project will have the dev meeting there. | 03:58 |
| zaitcev | Does anyone have a library that parses our log files? I cannot just line.split() because of Referrer having spaces. It's surrounded by quotes. | 04:33 |
| clayg | zaitcev: oh no! we should probably quote that before logging it. | 04:38 |
| zaitcev | too late for me though, I'm analyzing some logs from customer site | 04:39 |
| zaitcev | Date has a space too, and it's in square brackets. Consistency is not something we respect. | 04:39 |
| clayg | zaitcev: do you know anything about keystone? https://bugs.launchpad.net/swift/+bug/1863680 | 04:55 |
| openstack | Launchpad bug 1863680 in OpenStack Object Storage (swift) "python3 swift-dispersion-report issue" [High,In progress] - Assigned to Sam Morrison (sorrison) | 04:55 |
| zaitcev | clayg: sorry, I don't know anything about it. | 04:57 |
| zaitcev | clayg: smorrison mentioned in the bug that Keystone must be on HTTPS: for this to reproduce. | 05:03 |
| clayg | oh, good to know! I bet the gate could reproduce it 🤔 | 05:04 |
| zaitcev | Also, I seriously doubt that the order of imports is what gets us. I suspect that the change from calling patcher.monkey_patch() with no arguments to the version in swift.common.utils is what does that. Heck, there's even a comment about it in the code. | 05:05 |
| clayg | oh, the new comment - I thought I missed something | 05:05 |
| clayg | is he online? sorrison ohai! | 05:05 |
| zaitcev | dunno... I typoed his nick with m anyway | 05:06 |
| zaitcev | If you can reproduce, than you can try leaving patching in main(). | 05:07 |
| clayg | zaitcev: I think I could just move the swiftclient import under main (after existing patch) and it should work | 05:07 |
| clayg | or maybe the "problem" is an import side-effect in swiftclient | 05:08 |
| *** ccamacho has quit IRC | 05:17 | |
| *** evrardjp has quit IRC | 05:34 | |
| *** evrardjp has joined #openstack-swift | 05:34 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: Preserve query params in tempurl https://review.opendev.org/333331 | 06:41 |
| clayg | so if we're interested in that change the next step would be a functest so we can enumerate the auth behaviors we care about | 06:43 |
| *** tesseract has joined #openstack-swift | 08:05 | |
| openstackgerrit | Charles Hsu proposed openstack/python-swiftclient master: Allow users to list containers/objects by size. https://review.opendev.org/708074 | 08:24 |
| *** rpittau|afk is now known as rpittau | 08:26 | |
| *** ccamacho has joined #openstack-swift | 08:32 | |
| *** tesseract has quit IRC | 08:38 | |
| *** tesseract has joined #openstack-swift | 08:39 | |
| *** mikecmpbll has joined #openstack-swift | 08:55 | |
| *** tkajinam has quit IRC | 09:19 | |
| *** mvkr has joined #openstack-swift | 09:24 | |
| *** StevenK has quit IRC | 09:52 | |
| *** StevenK has joined #openstack-swift | 09:52 | |
| *** mikecmpbll has quit IRC | 09:54 | |
| *** mikecmpbll has joined #openstack-swift | 09:57 | |
| *** tesseract has quit IRC | 09:58 | |
| *** tesseract has joined #openstack-swift | 10:11 | |
| *** rpittau is now known as rpittau|bbl | 11:20 | |
| *** ccamacho has quit IRC | 12:14 | |
| *** jvisser has joined #openstack-swift | 12:17 | |
| *** theintern_ has joined #openstack-swift | 12:20 | |
| *** jvisser has quit IRC | 12:26 | |
| *** jvisser has joined #openstack-swift | 12:27 | |
| *** jvisser has quit IRC | 12:47 | |
| *** jvisser has joined #openstack-swift | 12:49 | |
| *** psachin has quit IRC | 12:49 | |
| *** jvisser has quit IRC | 13:00 | |
| *** jvisser has joined #openstack-swift | 13:00 | |
| *** jvisser has quit IRC | 13:01 | |
| *** jvisser has joined #openstack-swift | 13:01 | |
| *** rpittau|bbl is now known as rpittau | 13:07 | |
| *** ccamacho has joined #openstack-swift | 13:17 | |
| viks___ | Hi,I'm checking the logs in SAIO and my local cluster and the swift proxy logs shows `HTTP/1.0` instead of `HTTP/1.1`. i.e. i get some message like the below: | 13:17 |
| viks___ | ``` | 13:17 |
| viks___ | Feb 20 14:26:28 saio proxy-server[2878]: 127.0.0.1 127.0.0.1 20/Feb/2020/14/26/28 GET /info HTTP/1.0 200 - curl/7.58.0 - - 1666 - txd554055608e247ebb1d30-005e4e9714 - 0.0009 - - 1582208788.357263088 1582208788.358170986 - | 13:17 |
| viks___ | Feb 20 14:26:28 saio proxy-server[2878]: STDERR: 127.0.0.1 - - [20/Feb/2020 14:26:28] "GET /info HTTP/1.1" 200 1899 0.009609 (txn: txd554055608e247ebb1d30-005e4e9714) | 13:17 |
| viks___ | Feb 20 14:27:00 saio proxy-server[2878]: STDERR: (2878) accepted ('127.0.0.1', 60684) | 13:17 |
| viks___ | ``` | 13:17 |
| viks___ | Why this happens? Is it normal? | 13:17 |
| *** jvisser has quit IRC | 13:18 | |
| *** jvisser has joined #openstack-swift | 13:18 | |
| *** jvisser has quit IRC | 13:23 | |
| *** psachin has joined #openstack-swift | 13:37 | |
| *** jvisser has joined #openstack-swift | 14:00 | |
| *** diablo_rojo has joined #openstack-swift | 14:56 | |
| *** theintern_ has quit IRC | 15:06 | |
| *** psachin has quit IRC | 15:22 | |
| DHE | well the user agent is curl, so that was curl's decision. for a single request http 1.0 is fine | 16:11 |
| *** gyee has joined #openstack-swift | 16:14 | |
| *** ccamacho has quit IRC | 16:32 | |
| *** rpittau is now known as rpittau|afk | 16:53 | |
| *** tesseract has quit IRC | 17:00 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Preserve query params in tempurl https://review.opendev.org/333331 | 17:16 |
| *** diablo_rojo has quit IRC | 17:22 | |
| *** evrardjp has quit IRC | 17:34 | |
| *** mikecmpbll has quit IRC | 17:34 | |
| *** evrardjp has joined #openstack-swift | 17:35 | |
| *** mikecmpbll has joined #openstack-swift | 18:04 | |
| clayg | timburke: wow, I my brain was stuck on "I have a key to generate PUT tempurls" vs "I have been *given* a PUT tempurl" | 18:15 |
| timburke | clayg, i wonder if we could bake a set of query params into the signature, so the signer would have to know what the tempurl could be used for... i'd be much more on-board with that | 18:16 |
| clayg | the prefix approach sounds interesting, but probably not when you want to create an SLO at "c/big.object" to segments like "c+segments/fooXX"? | 18:17 |
| clayg | timburke: it's still asking a lot of the application developer understand the security implication of "PUT tempurls for SLOs leak data" | 18:17 |
| clayg | but yeah - better than making a hole for ALL PUT tempurls!!! | 18:18 |
| timburke | ...maybe? i need to refresh my memory about how much of a prefix you need, and where it starts | 18:18 |
| clayg | I almost think it'd be better to put the tempurl sig IN the manifest - just intermingle it all up | 18:18 |
| clayg | so generally speaking a PUT tempurl let's you create an object - but unless you have a valid PUT/HEAD sig for the segments you're not going to be able to create your SLO | 18:19 |
| clayg | it's alot more coupled that way tho 😬 | 18:19 |
| timburke | oh jeeze 🤮 | 18:19 |
| clayg | yeah if you can make a prefix of 'foo' that enables PUT/HEAD into all both 'foobar' and 'foobaz' containers that might reasonable for some SLO/MPU workloads! | 18:20 |
| clayg | timburke: also kudos for the pointer to p 659499 | 18:20 |
| patchbot | https://review.opendev.org/#/c/659499/ - swift - WIP: Functests that stress tempurl xlo PUTs - 1 patch set | 18:20 |
| clayg | if we can successfully pick a direction I think functests are going to be instrumental to understanding/validating the authorization behaviors we want. | 18:21 |
| timburke | ☝ | 18:25 |
| *** mvkr has quit IRC | 19:01 | |
| clayg | timburke: tdasilva: do either of you have the vsaio zipkin support enabled in your vsaio to test -> https://github.com/swiftstack/vagrant-swift-all-in-one/pull/93 | 19:51 |
| clayg | ^ or kota_ rledisez mattoliverau if you want to see what the distributed tracing with zipkin hubub is all about | 19:52 |
| *** mikecmpbll has quit IRC | 20:08 | |
| rledisez | clayg: can't wait to play with it, it looks interesting (I still need to read that PDF you link yesterday) | 20:10 |
| timburke | clayg, my vsaio's in a bad state right now, won't come up. stupid https://github.com/aws/aws-cli/issues/4828 | 20:12 |
| clayg | timburke: oh yeah I hit that when I switched to py3 i think | 20:12 |
| timburke | i'm trying to get to a state that i can repro https://bugs.launchpad.net/swift/+bug/1863680 -- i'm like 99% sure it's not keystone specific and you'd be able to repro w/ https://github.com/swiftstack/vagrant-swift-all-in-one/pull/90/ | 20:14 |
| openstack | Launchpad bug 1863680 in OpenStack Object Storage (swift) "python3 swift-dispersion-report issue" [High,In progress] - Assigned to Sam Morrison (sorrison) | 20:14 |
| timburke | i know i hit similar issues in https://review.opendev.org/#/c/675227/ and https://review.opendev.org/#/c/705579/ | 20:17 |
| patchbot | patch 675227 - swift - py3: Finish porting func tests (MERGED) - 4 patch sets | 20:17 |
| patchbot | patch 705579 - swift - probe tests: Work when fronted by a TLS terminator - 2 patch sets | 20:17 |
| timburke | hmm... maybe i'm misattributing my error; i guess the real trouble is that (1) something got python3-yaml installed via system packages, providing pyyaml 3.12, and (2) something *else* wants to upgrade that, but pip can't uninstall the system package | 20:32 |
| *** jvisser has quit IRC | 20:53 | |
| *** jvisser has joined #openstack-swift | 21:25 | |
| clayg | timburke: oh it'd be great if the difference was SSL and not keystone! | 21:28 |
| clayg | it looks like I'm due for another look @ https://github.com/swiftstack/vagrant-swift-all-in-one/pull/90/ - do you think it's ready to go, or do you have more local changes? | 21:28 |
| timburke | i think it's gtg -- you might be happier if you got https://review.opendev.org/#/c/705579/ (i know *i* am) | 21:29 |
| patchbot | patch 705579 - swift - probe tests: Work when fronted by a TLS terminator - 2 patch sets | 21:29 |
| clayg | awesome! that looks like a fun rabbit hole 🤣 | 21:30 |
| timburke | it wasn't near as bad as i'd feared -- tho that monkey-patching should maybe only apply to socket | 21:31 |
| timburke | clayg, sure enough, can repro w/ vsaio+tempauth+TLS | 21:48 |
| timburke | and i apparently need to template out /etc/swift/dispersion.conf :-/ | 21:48 |
| clayg | awesome! are you groking what I was getting at on moving the imports around so we don't monkey patch as a side-effect of importing? Can you try and validate a different change that fixes the issue? | 21:49 |
| clayg | timburke: ugh yeah, i hate it when i have to move chef file to a template 😡 | 21:50 |
| clayg | I'm not sure if it's annoying because of something inherent in chef's design or I did something wrong when I setup the project layout 🤔 | 21:51 |
| timburke | i almost got good at it as part of that PR ;-) | 21:51 |
| timburke | ok, clayg: which do you dislike more? imports with side-effects, or delayed imports? | 21:59 |
| clayg | Port with side effect categorically worse. | 22:05 |
| mattoliverau | morning | 22:07 |
| mattoliverau | clayg: ooohh nice! Me do wanna play with that hubub :) | 22:07 |
| *** jvisser has quit IRC | 22:22 | |
| *** tkajinam has joined #openstack-swift | 22:58 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Delay importing swiftclient until after monkey-patching https://review.opendev.org/708274 | 23:02 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!