| tdasilva | mattoliverau: great, I need to go take a look at your latest spec | 00:04 |
|---|---|---|
| *** carrigan has joined #openstack-swift | 00:14 | |
| *** devlaps has quit IRC | 00:18 | |
| *** mfalatic_ has joined #openstack-swift | 00:18 | |
| *** lyrrad_ has joined #openstack-swift | 00:19 | |
| *** jrichli has joined #openstack-swift | 00:20 | |
| *** bapalm_ has joined #openstack-swift | 00:20 | |
| *** mfalatic has quit IRC | 00:21 | |
| *** bapalm has quit IRC | 00:21 | |
| *** lyrrad has quit IRC | 00:21 | |
| *** lyrrad_ is now known as lyrrad | 00:21 | |
| *** portante has quit IRC | 00:21 | |
| *** portante has joined #openstack-swift | 00:22 | |
| *** esker has quit IRC | 00:24 | |
| jrichli | mattoliverau: good to hear you are feeling better. hope you are 100% soon | 00:24 |
| *** esker has joined #openstack-swift | 00:25 | |
| *** minwoob has quit IRC | 00:29 | |
| *** matt____ has joined #openstack-swift | 00:29 | |
| *** m_kazuhiro has joined #openstack-swift | 00:30 | |
| *** dmorita has joined #openstack-swift | 00:31 | |
| *** wasmum- has joined #openstack-swift | 00:31 | |
| *** StevenK_ has joined #openstack-swift | 00:31 | |
| *** mlanner_ has joined #openstack-swift | 00:31 | |
| *** StevenK has quit IRC | 00:32 | |
| *** StevenK_ is now known as StevenK | 00:33 | |
| *** dipe has quit IRC | 00:35 | |
| *** mlanner has quit IRC | 00:35 | |
| *** mattoliverau has quit IRC | 00:36 | |
| *** wasmum has quit IRC | 00:36 | |
| *** mlanner_ is now known as mlanner | 00:36 | |
| *** zhill has quit IRC | 00:39 | |
| *** carrigan has quit IRC | 00:41 | |
| *** hrou has joined #openstack-swift | 00:41 | |
| *** dipe has joined #openstack-swift | 00:44 | |
| *** matt____ is now known as mattoliverau | 00:46 | |
| *** ChanServ sets mode: +v mattoliverau | 00:46 | |
| *** Vinsh has quit IRC | 00:56 | |
| clayg | q | 00:56 |
| *** DericHorn-HP has joined #openstack-swift | 01:00 | |
| *** vinsh has joined #openstack-swift | 01:04 | |
| mattoliverau | clayg: don't quit! | 01:11 |
| *** wbhuber has joined #openstack-swift | 01:12 | |
| *** baojg has joined #openstack-swift | 01:15 | |
| *** cdelatte has joined #openstack-swift | 01:18 | |
| *** garthb has quit IRC | 01:19 | |
| *** baojg has quit IRC | 01:20 | |
| openstackgerrit | Zack M. Davis proposed openstack/swift: replace our use of deprecated rfc822.Message with a helper function https://review.openstack.org/203304 | 01:21 |
| openstackgerrit | Alexandra Settle proposed openstack/swift: Moving DLO functionality doc to the middleware code https://review.openstack.org/219991 | 01:25 |
| *** hezhiqiang has joined #openstack-swift | 01:34 | |
| *** hezhiqiang has quit IRC | 01:34 | |
| *** baojg has joined #openstack-swift | 01:38 | |
| *** zhill has joined #openstack-swift | 01:38 | |
| *** zhill has quit IRC | 01:43 | |
| *** m_kazuhiro has quit IRC | 01:47 | |
| *** hezhiqiang has joined #openstack-swift | 01:48 | |
| *** wshao has quit IRC | 01:51 | |
| clayg | well I be... I didn't know the DiskFile.create would optomistically clean up tmppath on a failed PUT - how helpful! | 01:54 |
| *** cdelatte has quit IRC | 01:56 | |
| clayg | fuh - i was the second to last person to touch that code - but it was during the first portante based diskfile refactor - so I wasn't paying that close of attention | 01:56 |
| clayg | but after me it looks like Prashanth Pai optomized it! nice | 01:57 |
| clayg | 42c790d0 -> Avoid unnecessary unlink() on every successful PUT | 01:57 |
| openstackgerrit | Janie Richling proposed openstack/swift: Decrypting Container Listing https://review.openstack.org/214438 | 01:57 |
| jrichli | mahatic: ^^ rebase on this, and then you can apply the comments I made on patch 216974 | 01:58 |
| patchbot | jrichli: https://review.openstack.org/#/c/216974/ | 01:58 |
| *** jkugel has joined #openstack-swift | 02:05 | |
| *** asettle has joined #openstack-swift | 02:06 | |
| * portante sings "what am I doin' hangin' round, I should be on that train and gone, I should be ridin' on that train to San Anton..." | 02:07 | |
| clayg | portante: the Monkeys? | 02:09 |
| portante | ;) | 02:09 |
| *** zhill has joined #openstack-swift | 02:10 | |
| *** zhill has quit IRC | 02:14 | |
| *** haomaiwang has joined #openstack-swift | 02:14 | |
| *** m_kazuhiro has joined #openstack-swift | 02:17 | |
| *** DericHorn-HP has quit IRC | 02:17 | |
| *** gyee has quit IRC | 02:24 | |
| *** baojg has quit IRC | 02:28 | |
| *** wbhuber has quit IRC | 02:28 | |
| *** baojg has joined #openstack-swift | 02:29 | |
| *** zhill has joined #openstack-swift | 02:29 | |
| *** zhill has quit IRC | 02:34 | |
| *** kota_ has joined #openstack-swift | 02:34 | |
| *** ChanServ sets mode: +v kota_ | 02:34 | |
| *** haomaiwang has quit IRC | 02:37 | |
| *** bkopilov has quit IRC | 02:38 | |
| *** haomaiwang has joined #openstack-swift | 02:43 | |
| *** haomaiwang has quit IRC | 02:50 | |
| *** haomaiwang has joined #openstack-swift | 02:52 | |
| *** haomaiwang has quit IRC | 03:01 | |
| *** DericHorn-HP has joined #openstack-swift | 03:01 | |
| *** haomaiwang has joined #openstack-swift | 03:01 | |
| notmyname | hello, world | 03:02 |
| mattoliverau | notmyname: o/ | 03:03 |
| notmyname | now I'm in new york and catching up what I missed online today | 03:03 |
| mattoliverau | notmyname: how was the drive? | 03:04 |
| notmyname | long ;-) | 03:04 |
| notmyname | not bad though | 03:04 |
| notmyname | I have now been to delaware. yay, I guess | 03:04 |
| mattoliverau | Everytime I head delaware I think of: https://www.youtube.com/watch?v=TK_FiRm6ZK8 | 03:06 |
| notmyname | ;-) | 03:06 |
| notmyname | I saw some really cool stuff today. sciency things where they use robots to do experiments and they want to use swift to store the results of the hundreds of thousands of tests they are running | 03:07 |
| *** janonymous has joined #openstack-swift | 03:07 | |
| *** baojg has quit IRC | 03:07 | |
| mattoliverau | wow, cool! | 03:07 |
| notmyname | I have some pictures, but it seems I need to be on a better connection than my hotel wifi to sync and upload them | 03:08 |
| mattoliverau | bugger, pics would be awesome | 03:08 |
| notmyname | so they have a tray filled with thousands of little holes. and they put chemicals in each hole, only varying the thing they are testing | 03:08 |
| jrichli | notmyname: sounds neat! btw, if I didnt mention before, DE was the state I used to live in. What city did you see? | 03:09 |
| notmyname | so they want to see if it makes a reaction more or less active | 03:09 |
| *** mahatic has joined #openstack-swift | 03:09 | |
| notmyname | jrichli: some highway and a rest stop ;-) | 03:09 |
| notmyname | so then they use robots to run these tests of all these chemical variations and take a picture of the tray of holes | 03:09 |
| jrichli | notmyname: ah, not the interesting parts, I guess. :-) | 03:09 |
| notmyname | then they analyze the pics and see what's interesting | 03:10 |
| notmyname | oh, and they're taking pictures at as high a resolution as 0.1 microns. they use that to build a 3d model of the chemical | 03:10 |
| notmyname | so this is "chem-informatics" (like bio-informatics) | 03:10 |
| notmyname | this is basically the very first stage of developing new drugs | 03:11 |
| *** kota_ has quit IRC | 03:11 | |
| clayg | why do we have to send 0\r\n\r\n in the middle of our mime document? | 03:11 |
| notmyname | biologists and chemists set up the tests and look at the results, and then the results can be shared with pharma companies and they develop drugs | 03:11 |
| clayg | this is some fucked up wizardry | 03:12 |
| notmyname | clayg: looks like chunked transfer encoding end marker | 03:12 |
| *** DericHorn-HP has quit IRC | 03:12 | |
| clayg | yeah - specifically the end of a chunked transfer request - in the middle of our chunked transfer request | 03:12 |
| *** zhill has joined #openstack-swift | 03:13 | |
| mahatic | notmyname: what is this all about? insights on drug making stuff? :D | 03:13 |
| clayg | i know we sort made all this crap up with the second expect 100 continue stuff - but ... well we probably should have just done http2 | 03:13 |
| notmyname | mahatic: someone I visited today that is looking at using swift to store a bunch of data | 03:14 |
| mattoliverau | clayg: are we ending a chunked transfer request between each document in the mime? Says me who isn't looking at the code | 03:14 |
| notmyname | mattoliverau: I got my phone synced. now trying to upload stuff ;-) | 03:14 |
| mattoliverau | notmyname: \o/ | 03:14 |
| mahatic | notmyname: oh cool! | 03:14 |
| clayg | mattoliverau: no - just after the 2nd before the 3rd? | 03:14 |
| mahatic | jrichli: ack on rebase | 03:15 |
| mahatic | will look at that | 03:15 |
| clayg | mattoliverau: but even that would be bs - a mimedocument has it's *own* boundaries 0\r\n\r\n means the next thing on the wire better be the start of the next request - how did we trick eventlet.wsgi into doing this? | 03:15 |
| jrichli | mahatic: great. thx. let me know if it does the trick. | 03:15 |
| *** DericHorn-HP has joined #openstack-swift | 03:15 | |
| mahatic | jrichli: yup, will do | 03:15 |
| notmyname | here's an empty tray that they load the reagents into http://d.not.mn/empty_tray.jpg | 03:17 |
| mattoliverau | clayg: or for a _second_ expect 100, is it to tell the server the data has stopped and so an 100-conintue can happen so we can pass something like footers. | 03:17 |
| * mattoliverau needs to open up some code now... so much for lunch :P | 03:17 | |
| *** zhill has quit IRC | 03:18 | |
| *** fifieldt has joined #openstack-swift | 03:18 | |
| notmyname | http://d.not.mn/trays_loaded.jpg is the big machines where they put a bunch of these trays | 03:18 |
| notmyname | then the robot does science! http://d.not.mn/robot.m4v | 03:18 |
| clayg | mattoliverau: yeah i mean it has to be something like that - but it's bs | 03:18 |
| notmyname | then this is one example of a results image that is produced http://d.not.mn/results.jpg | 03:19 |
| clayg | mattoliverau: i'm looking for the bug in eventlet.wsgi right now | 03:19 |
| jrichli | notmyname: looks kind of like a test board | 03:19 |
| notmyname | jrichli: yeah, really similar. | 03:19 |
| mattoliverau | clayg: lol, yeah.. well good luck with that :P | 03:20 |
| mattoliverau | notmyname: \o/ robots doing science! | 03:20 |
| *** baojg has joined #openstack-swift | 03:22 | |
| jrichli | notmyname: looks like its software just for those trays. dont suppose you know who does the image processing software? | 03:22 |
| notmyname | nope. no idea | 03:22 |
| notmyname | but when you have 600000 experiments with sets of images taken at 0.1 micron vertical slices and you want to track it, I do know a good storage system that can handle that ;-) | 03:24 |
| *** mragupat has joined #openstack-swift | 03:24 | |
| clayg | ugh - that's bs - f'ing eventlet wsgi won't return a chunked transfer body from the client that's let than the network size | 03:25 |
| *** mragupat_ has joined #openstack-swift | 03:25 | |
| clayg | so we FUCKING just write 0\r\n\r\n on the wire to shut down the read method - then - like bastards - just because we can - we start reading off wsgi.input again (even tho '' is supposed to mean the end of the request) - and for some reason the chunked read let's it fly | 03:26 |
| notmyname | clayg: I know I'm out of town, but let me preempt where this is going. no, everything isn't terrible. yes there's a lot to do. we'll get to it. ;-) | 03:26 |
| clayg | LOL | 03:26 |
| clayg | gah - yeah the content-length based read can cap what you're able to pull off wsgi.input because it knows the size - if you've already eaten your request you start getting '' | 03:27 |
| notmyname | doesn't eventlet also completely hide the fact it's even chunked? | 03:28 |
| *** mragupat has quit IRC | 03:29 | |
| clayg | well on the body yeah - but you still have to that an empty chunk means break | 03:31 |
| notmyname | next swift release notes will contain: "acoles got something he wanted" (https://review.openstack.org/#/c/223854/) ;-) | 03:33 |
| mattoliverau | lol | 03:34 |
| mattoliverau | that would be the best note ever | 03:34 |
| *** nakagawamsa has joined #openstack-swift | 03:36 | |
| mattoliverau | ok, now I really need to go to lunch (I need to time to drive to the other side of Melbourne). bbl from a cafe. | 03:36 |
| mattoliverau | *need the time | 03:37 |
| * mattoliverau feels his typing just keeps getting worse | 03:37 | |
| *** david-lyle has joined #openstack-swift | 03:41 | |
| openstackgerrit | Merged openstack/swift: Update EC Support on how to build an EC ring with replicas count https://review.openstack.org/223304 | 03:46 |
| clayg | how do I not have eventlet source code checkedout? | 03:50 |
| clayg | nm, found it | 03:51 |
| *** trifon has quit IRC | 03:51 | |
| clayg | + # Reinitialize chunk_length (expect more data) | 03:52 |
| clayg | + self.chunk_length = -1 | 03:52 |
| clayg | Tushar ... | 03:53 |
| clayg | ok. so there ends that rabbit hole. how did i get here | 03:54 |
| *** haomaiwang has quit IRC | 03:58 | |
| *** haomaiwang has joined #openstack-swift | 03:58 | |
| *** DericHorn-HP has quit IRC | 03:59 | |
| *** haomaiwang has quit IRC | 04:01 | |
| *** haomaiwa_ has joined #openstack-swift | 04:01 | |
| *** changbl has joined #openstack-swift | 04:04 | |
| *** bkopilov has joined #openstack-swift | 04:04 | |
| *** DericHorn-HP has joined #openstack-swift | 04:26 | |
| notmyname | time for bed | 04:30 |
| jrichli | me too. good night! | 04:32 |
| clayg | ok, so I got my review that started this mess finished (surprise surprise it was a -1) | 04:32 |
| clayg | ... but now I really don't know what to do with this eventlet thing - like I guess it works - but it seems really fragile | 04:32 |
| *** jrichli has quit IRC | 04:32 | |
| clayg | maybe I can chat with sam about it tomorrow - the problem is a fix would have to co-ordinate the up-rev of eventlet :\ we're probably stuck with | 04:33 |
| *** baojg has quit IRC | 04:33 | |
| clayg | maybe best we could do is try and get some more tests into eventlet upstream like "yeah sorry this is gross, but it works this way now, please don't break us trying to fix this!" | 04:33 |
| notmyname | why are we stuck with older eventlet? backwards support? | 04:34 |
| clayg | it's not that we can't have new eventlet - but yeah it'd be like during upgrades | 04:34 |
| notmyname | yeah, got it | 04:35 |
| clayg | did all of kota's fixes for the content-length on invalid ranges and shit get merged? | 04:35 |
| notmyname | I think so | 04:36 |
| clayg | patch 198632 from forever! sorry kota! | 04:36 |
| patchbot | clayg: https://review.openstack.org/#/c/198632/ | 04:36 |
| notmyname | ok, time for me to sleep on it. early start tomorrow, and it's already nearly 1am here | 04:37 |
| *** wshao has joined #openstack-swift | 04:38 | |
| clayg | notmyname: yeah looks like they're all merged | 04:39 |
| *** baojg has joined #openstack-swift | 04:40 | |
| *** DericHorn-HP has quit IRC | 04:40 | |
| *** david-lyle has quit IRC | 04:42 | |
| *** david-lyle has joined #openstack-swift | 04:45 | |
| *** trifon has joined #openstack-swift | 04:56 | |
| *** amit213 has quit IRC | 04:59 | |
| *** amit213 has joined #openstack-swift | 04:59 | |
| *** haomaiwa_ has quit IRC | 05:01 | |
| *** haomaiwang has joined #openstack-swift | 05:01 | |
| *** garthb has joined #openstack-swift | 05:14 | |
| *** mragupat_ has quit IRC | 05:24 | |
| *** baojg has quit IRC | 05:28 | |
| *** hrou has quit IRC | 05:33 | |
| *** ppai has joined #openstack-swift | 05:35 | |
| *** zhill has joined #openstack-swift | 05:40 | |
| *** zhill has quit IRC | 05:45 | |
| *** haomaiwang has quit IRC | 06:01 | |
| *** haomaiwa_ has joined #openstack-swift | 06:02 | |
| *** mahatic has quit IRC | 06:09 | |
| *** mahatic has joined #openstack-swift | 06:13 | |
| *** asettle has quit IRC | 06:29 | |
| *** wshao has quit IRC | 06:35 | |
| *** haomaiwa_ has quit IRC | 07:01 | |
| *** zhill has joined #openstack-swift | 07:01 | |
| *** haomaiwang has joined #openstack-swift | 07:02 | |
| *** zhill has quit IRC | 07:05 | |
| *** wer_ has quit IRC | 07:07 | |
| *** m_kazuhiro has quit IRC | 07:11 | |
| *** garthb has quit IRC | 07:12 | |
| *** zhill has joined #openstack-swift | 07:13 | |
| *** wer_ has joined #openstack-swift | 07:16 | |
| *** zhill has quit IRC | 07:17 | |
| *** rledisez has joined #openstack-swift | 07:24 | |
| *** m_kazuhiro has joined #openstack-swift | 07:26 | |
| *** wer_ has quit IRC | 07:39 | |
| *** wer_ has joined #openstack-swift | 07:47 | |
| *** geaaru has joined #openstack-swift | 07:52 | |
| *** jordanP has joined #openstack-swift | 07:53 | |
| *** haomaiwang has quit IRC | 08:01 | |
| *** acoles_ is now known as acoles | 08:02 | |
| *** haomaiwa_ has joined #openstack-swift | 08:04 | |
| acoles | notmyname: or maybe the release notes will say "clayg fixed acoles" ;) | 08:05 |
| *** zhill has joined #openstack-swift | 08:13 | |
| *** zhill has quit IRC | 08:17 | |
| *** jistr has joined #openstack-swift | 08:19 | |
| *** kota_ has joined #openstack-swift | 08:31 | |
| *** ChanServ sets mode: +v kota_ | 08:31 | |
| *** haomaiwa_ has quit IRC | 08:39 | |
| *** nakagawamsa has quit IRC | 08:40 | |
| *** haomaiwang has joined #openstack-swift | 08:40 | |
| *** SkyRocknRoll has joined #openstack-swift | 08:47 | |
| kota_ | ack, no meeting tommorow. | 08:52 |
| *** km has quit IRC | 08:57 | |
| *** tab_ has joined #openstack-swift | 09:00 | |
| *** haomaiwang has quit IRC | 09:01 | |
| *** kei_yama has quit IRC | 09:01 | |
| *** haomaiwang has joined #openstack-swift | 09:02 | |
| *** zhill has joined #openstack-swift | 09:14 | |
| *** aix has joined #openstack-swift | 09:18 | |
| *** zhill has quit IRC | 09:19 | |
| openstackgerrit | Daisuke Morita proposed openstack/swift: Show each policy's information on audited results in recon https://review.openstack.org/138697 | 09:21 |
| *** dmorita has quit IRC | 09:26 | |
| *** janonymous has quit IRC | 09:32 | |
| *** tab_ has quit IRC | 09:35 | |
| *** m_kazuhiro has quit IRC | 09:44 | |
| *** baojg has joined #openstack-swift | 09:48 | |
| *** zhill has joined #openstack-swift | 09:50 | |
| *** mahatic has quit IRC | 09:52 | |
| *** mahatic has joined #openstack-swift | 09:53 | |
| *** zhill has quit IRC | 09:54 | |
| acoles | kota_: oh yeah, no meeting today :) | 09:59 |
| *** haomaiwang has quit IRC | 10:01 | |
| *** haomaiwang has joined #openstack-swift | 10:01 | |
| kota_ | acoles: morning :) | 10:03 |
| *** hezhiqiang has quit IRC | 10:06 | |
| *** zhill has joined #openstack-swift | 10:12 | |
| mahatic | acoles: Morning! A question: did tox -e func run fine with my changes yesterday? The headers patch, did it give errors? | 10:13 |
| mahatic | this patch 223489 | 10:15 |
| patchbot | mahatic: https://review.openstack.org/#/c/223489/ | 10:15 |
| *** zhill has quit IRC | 10:17 | |
| mattoliverau | What no meeting tomorrow? | 10:22 |
| kota_ | mattoliverau: yes, https://wiki.openstack.org/wiki/Meetings/Swift | 10:23 |
| mattoliverau | \o/ sleep in! | 10:23 |
| mattoliverau | kota_: thanks | 10:24 |
| kota_ | mattoliverau: np :P | 10:24 |
| *** baojg has quit IRC | 10:30 | |
| *** baojg has joined #openstack-swift | 10:31 | |
| *** baojg has quit IRC | 10:35 | |
| acoles | mahatic: no. tox -e func failed on my SAIO (with keystone auth) same as on jenkins. But when I made the change I suggested in review on my local copy it passed. | 10:35 |
| mahatic | acoles: oh, also was encryption enabled when you ran the tests? | 10:36 |
| acoles | mahatic: no. it has nothing to do with encryption as far as i can see. | 10:37 |
| mahatic | acoles: interesting and confusing. It somehow has an effect. when I remove encryption from pipeline and run swiftclient tests, all of the pass without your suggested change. Otherwise, there is etag mismatch | 10:38 |
| *** eranrom has joined #openstack-swift | 10:38 | |
| mahatic | am I missing something very obvious? | 10:38 |
| acoles | mahatic: are you using keystone auth? swiftclient tox -e func passed with tempauth IIRC. | 10:39 |
| mahatic | acoles: I use tempauth | 10:39 |
| mahatic | acoles: oh did it pass without your review changes on tempauth? | 10:40 |
| mahatic | http://paste.openstack.org/show/464439/ | 10:41 |
| *** janonymous has joined #openstack-swift | 10:41 | |
| acoles | mahatic: let me just confirm | 10:42 |
| mahatic | acoles: sure, thanks | 10:42 |
| acoles | mahatic: yes your original patch passes with tempauth on my saio | 10:46 |
| acoles | the test failure in your paste bin is different | 10:46 |
| *** chenhuayi has quit IRC | 10:47 | |
| mahatic | acoles: oh, thanks for confirming! that leaves with a couple of questions, how would I test if it passes on tempauth? | 10:48 |
| mahatic | and why do I have a diff error | 10:49 |
| *** aix has quit IRC | 10:50 | |
| acoles | mahatic: to your first question - you said you are using tempauth. take all encryption middleware out of proxy pipeline, restart and run the swiftclient tests | 10:52 |
| acoles | mahatic: to your second question - i guess you are maybe seeing an error because you have encryption in your proxy pipeline? | 10:52 |
| mahatic | acoles: yes, I do. when I remove encryption, tests pass | 10:53 |
| mahatic | the thing is, with encryption out of proxy pipeline, tests pass *without* your suggested changes | 10:55 |
| mahatic | or maybe I should just setup keystone (!) | 10:59 |
| *** haomaiwang has quit IRC | 11:01 | |
| *** haomaiwang has joined #openstack-swift | 11:01 | |
| *** ho has quit IRC | 11:02 | |
| *** aix has joined #openstack-swift | 11:03 | |
| *** cdelatte has joined #openstack-swift | 11:15 | |
| *** miurahr has joined #openstack-swift | 11:24 | |
| acoles | yes the test passes with tempauth but not keystone. the jenkins job is using keystone. | 11:26 |
| *** bkopilov has quit IRC | 11:26 | |
| *** DericHorn-HP has joined #openstack-swift | 11:34 | |
| *** haomaiwang has quit IRC | 11:35 | |
| *** haomaiwang has joined #openstack-swift | 11:38 | |
| mahatic | acoles: thanks, will setup keystone | 11:44 |
| acoles | mahatic: having a keystone setup can be a good thing, but to make progress with your patch you could rely on the jenkins job. | 11:46 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 11:47 |
| mahatic | acoles: I was going back and forth on that thought, thanks! | 11:51 |
| *** haomaiwang has quit IRC | 12:01 | |
| *** haomaiwa_ has joined #openstack-swift | 12:01 | |
| *** [1]eranrom has joined #openstack-swift | 12:03 | |
| *** eranrom has quit IRC | 12:04 | |
| *** [1]eranrom is now known as eranrom | 12:04 | |
| openstackgerrit | Merged openstack/swift: Allows to configure the rsync modules where the replicators will send data https://review.openstack.org/192186 | 12:06 |
| *** ppai has quit IRC | 12:08 | |
| *** jkugel has quit IRC | 12:11 | |
| *** haomaiwa_ has quit IRC | 12:11 | |
| *** ppai has joined #openstack-swift | 12:21 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Fix object-expirer to prevent orphaned objects https://review.openstack.org/224055 | 12:26 |
| kota_ | eranrom: I didn't notice your message, sorry. | 12:27 |
| *** MVenesio has joined #openstack-swift | 12:28 | |
| *** ppai has quit IRC | 12:33 | |
| *** tongli has joined #openstack-swift | 12:38 | |
| *** bkopilov has joined #openstack-swift | 12:45 | |
| *** dipe has quit IRC | 12:46 | |
| *** dipe has joined #openstack-swift | 12:47 | |
| *** DericHorn-HP has quit IRC | 12:50 | |
| *** bkopilov has quit IRC | 12:51 | |
| *** bkopilov has joined #openstack-swift | 12:51 | |
| *** kota_ has quit IRC | 12:53 | |
| *** annegentle has joined #openstack-swift | 12:54 | |
| *** bill_az has joined #openstack-swift | 12:55 | |
| *** aix has quit IRC | 12:56 | |
| *** SkyRocknRoll has quit IRC | 13:05 | |
| *** DericHorn-HP has joined #openstack-swift | 13:05 | |
| *** esker has quit IRC | 13:05 | |
| *** jkugel has joined #openstack-swift | 13:08 | |
| *** aix has joined #openstack-swift | 13:10 | |
| *** MVenesio has quit IRC | 13:14 | |
| *** MVenesio has joined #openstack-swift | 13:15 | |
| *** dustins has joined #openstack-swift | 13:21 | |
| *** hrou has joined #openstack-swift | 13:27 | |
| *** janonymous has quit IRC | 13:28 | |
| *** annegentle has quit IRC | 13:30 | |
| *** annegentle has joined #openstack-swift | 13:30 | |
| *** dustins has quit IRC | 13:32 | |
| *** dustins has joined #openstack-swift | 13:39 | |
| *** hezhiqiang has joined #openstack-swift | 13:41 | |
| *** DericHorn-HP has quit IRC | 13:45 | |
| *** ganv has joined #openstack-swift | 14:06 | |
| ganv | Hi. I am trying to deploy openstack swift as a backend storage to backing up cinder volumes. Is it possible to do this without keystone in the picture? | 14:06 |
| *** mragupat has joined #openstack-swift | 14:09 | |
| *** mragupat has quit IRC | 14:10 | |
| *** aix has quit IRC | 14:11 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: diskfile: make get_ondisk_files return a dict https://review.openstack.org/222706 | 14:11 |
| *** jrichli has joined #openstack-swift | 14:14 | |
| *** trifon has quit IRC | 14:18 | |
| *** DericHorn-HP has joined #openstack-swift | 14:22 | |
| *** aix has joined #openstack-swift | 14:22 | |
| *** acoles is now known as acoles_ | 14:24 | |
| *** pgbridge has joined #openstack-swift | 14:25 | |
| *** DericHorn-HP has quit IRC | 14:26 | |
| *** jlhinson has joined #openstack-swift | 14:30 | |
| *** DericHorn-HP has joined #openstack-swift | 14:30 | |
| *** garthb has joined #openstack-swift | 14:31 | |
| *** openstackgerrit has quit IRC | 14:31 | |
| *** openstackgerrit has joined #openstack-swift | 14:31 | |
| *** DericHorn-HP has quit IRC | 14:32 | |
| *** eranrom has quit IRC | 14:34 | |
| *** wbhuber has joined #openstack-swift | 14:34 | |
| *** tongli_ has joined #openstack-swift | 14:35 | |
| *** tongli has quit IRC | 14:37 | |
| *** david-lyle has quit IRC | 14:37 | |
| tdasilva | 14:39 | |
| *** haomaiwang has joined #openstack-swift | 14:39 | |
| *** lcurtis has joined #openstack-swift | 14:40 | |
| *** acoles_ is now known as acoles | 14:44 | |
| *** ganv___ has joined #openstack-swift | 14:49 | |
| ganv___ | Hi. I am trying to deploy openstack swift as a backend storage to backing up cinder volumes. Is it possible to do this without keystone in the picture? | 14:49 |
| *** ganv has quit IRC | 14:49 | |
| *** hezhiqiang has quit IRC | 14:51 | |
| *** annegent_ has joined #openstack-swift | 14:54 | |
| *** annegentle has quit IRC | 14:54 | |
| *** haomaiwang has quit IRC | 14:56 | |
| *** jistr is now known as jistr|call | 14:58 | |
| *** minwoob has joined #openstack-swift | 14:59 | |
| *** openstackgerrit has quit IRC | 15:01 | |
| *** openstackgerrit has joined #openstack-swift | 15:01 | |
| *** mahatic has quit IRC | 15:02 | |
| lcurtis | hello all...we are running statsd, and unfortunately starting to run into performance issues as noted here: https://ask.openstack.org/en/question/1772/swift-replicator-timeouts/ | 15:04 |
| jrichli | mahatic: have you made all the name changes needed for the latest encryption code? the filename changed, and the name of the middleware in the pipeline changed | 15:04 |
| lcurtis | wondering if anyone else has found some tricks to running statsd | 15:04 |
| hrou | jrichli, mahatic: just an FYI I updated the etherpad with the results I've seen with your probe test patch, I'll add swiftclient test results too (but we know about the open issue there without crypto) | 15:09 |
| *** mahatic has joined #openstack-swift | 15:09 | |
| *** ganv___ has quit IRC | 15:24 | |
| jrichli | ganv___ : What auth service are you looking to use? | 15:28 |
| jrichli | oops, too late | 15:28 |
| *** jistr|call is now known as jistr | 15:30 | |
| *** slo___ has joined #openstack-swift | 15:30 | |
| *** annegent_ has quit IRC | 15:32 | |
| *** annegentle has joined #openstack-swift | 15:33 | |
| *** haomaiwa_ has joined #openstack-swift | 15:33 | |
| *** miurahr has quit IRC | 15:33 | |
| *** david-ly_ has joined #openstack-swift | 15:35 | |
| mahatic | jrichli: hi, morning. name changes? while I was doing rebase? | 15:40 |
| mahatic | also jrichli thanks for updating me with the messages :) | 15:41 |
| mahatic | hrou: what open issue about crypto? | 15:41 |
| *** david-ly_ is now known as david-lyle | 15:41 | |
| hrou | mahatic, without crypto in the pipeline but with your changes client changes (using keystone auth). | 15:41 |
| hrou | mahatic, what you were chatting with acoles about earlier. | 15:42 |
| mahatic | hrou: oh, I'm pushing those changes in a while. The only issue is I don't use keystone :D | 15:43 |
| hrou | mahatic, ah : ) Ha no worries yea I'd just push anyway, I can validate them, I have keystone - can help you set it up if you like as well, its not too bad. | 15:44 |
| *** trifon has joined #openstack-swift | 15:44 | |
| *** jordanP has quit IRC | 15:45 | |
| jrichli | mahatic: you will need to change the middleware name in the proxy-server.conf file to be trivial_keymaster. Did you do that change? | 15:45 |
| mahatic | hrou: sure thanks. about setting up keystone, will let you know | 15:45 |
| mahatic | jrichli: oh yes I did. the rebase wouldn't have worked otherwise | 15:46 |
| mahatic | jrichli: you should have noticed them when you pulled those changes I believe? | 15:46 |
| jrichli | mahatic: did you change your setup.cfg file to say: trivial_keymaster = swift.common.middleware.trivial_keymaster:filter_factory | 15:46 |
| jrichli | the config file instance used is not technically part of the repo. only the example file is. | 15:47 |
| mahatic | jrichli: correct, but acoles mentioned it as one of his review comments (thanks acoles !) | 15:48 |
| mahatic | jrichli: you ask this because of the headers thing? | 15:49 |
| acoles | don't blame me i only put the _ in because hrou asked for it :P | 15:52 |
| hrou | acoles, yes its always hrou's fault ; - ) | 15:52 |
| * acoles happened to agree with hrou but will still pass the blame :P | 15:53 | |
| mahatic | haha, yes that's what you have put in the comments :P | 15:53 |
| hrou | jrichli, acoles, mahatic you sure setup.cfg isn't part of the repo though ? I thought it was. | 15:53 |
| mahatic | hrou: setup.cfg is | 15:53 |
| hrou | I don't recall having to manually patch that file. | 15:53 |
| mahatic | hrou: proxy conf is not | 15:53 |
| hrou | Oh, yea of course : ) Ahhh sorry that's what you were talking about, gotcha. | 15:53 |
| hrou | Cool | 15:53 |
| mahatic | proxy-conf sample is | 15:53 |
| hrou | yep | 15:53 |
| *** jistr has quit IRC | 15:54 | |
| *** devlaps has joined #openstack-swift | 15:54 | |
| *** garthb has quit IRC | 15:55 | |
| jrichli | hrou: yes, i only mentioned setup.cfg because i sometimes have to think about it: like when i update my testing with COPY middleware integrating different changes by hand | 15:59 |
| *** haomaiwa_ has quit IRC | 16:01 | |
| *** 16WAAAFWE has joined #openstack-swift | 16:01 | |
| *** DericHorn-HP has joined #openstack-swift | 16:01 | |
| minwoob | clayg: https://review.openstack.org/#/c/215276/ | 16:03 |
| minwoob | I can go ahead and implement what you were thinking in the last comment. | 16:03 |
| *** pberis has joined #openstack-swift | 16:04 | |
| *** chsc has joined #openstack-swift | 16:04 | |
| hrou | jrichli, ah ! Yea I could see that in that case, makes sense, you're doing all the tough work with copy middleware ; ) | 16:05 |
| jrichli | hrou: not so tough now that we know what our issue was | 16:06 |
| *** gyee has joined #openstack-swift | 16:06 | |
| hrou | jrichli, getting to that point : ) | 16:07 |
| hrou | mahatic, just an fyi that for me swift client func tests are clean as well with temp auth (but not keystone) i.e. I'm seeing the same as acoles. | 16:09 |
| *** blmartin has joined #openstack-swift | 16:11 | |
| *** saltsa has quit IRC | 16:12 | |
| mahatic | hrou: okay, thanks | 16:19 |
| *** mahatic has quit IRC | 16:19 | |
| *** rledisez has quit IRC | 16:20 | |
| *** annegentle has quit IRC | 16:20 | |
| openstackgerrit | David Goetz proposed openstack/swift: go: make BeginReplication concurrent https://review.openstack.org/224181 | 16:21 |
| clayg | minwoob: sure - give me more time to review all your other patches ;) | 16:21 |
| *** annegentle has joined #openstack-swift | 16:21 | |
| clayg | dfg: nice concurrent all the things! | 16:21 |
| clayg | although... seems like making FinishReplication concurrent would have been even better? | 16:21 |
| *** mahatic has joined #openstack-swift | 16:22 | |
| ahale | i'd settle for making replication finish | 16:22 |
| clayg | ahale: :D | 16:22 |
| clayg | briancline: what's alan's handle? | 16:22 |
| minwoob | clayg: ajiang | 16:23 |
| openstackgerrit | Mahati Chamarthy proposed openstack/python-swiftclient: Add headers parameter https://review.openstack.org/223489 | 16:26 |
| *** garthb has joined #openstack-swift | 16:28 | |
| *** saltsa has joined #openstack-swift | 16:28 | |
| acoles | clayg: question for you... | 16:29 |
| clayg | shoot! | 16:29 |
| * clayg get's ready for "why are you suck a jerk all the time?" | 16:30 | |
| acoles | clayg: how do you read my mind from so far away? :P | 16:30 |
| clayg | heh | 16:30 |
| acoles | clayg: EC GET - if proxy gets some frags but not ec_ndata, should it return 404 or 503? | 16:31 |
| clayg | acoles: we talked about it in Austin and were leaning towards 404 - I think it's definately a scale | 16:31 |
| clayg | did it find *one* fragment and a bunch of tombstones - 404 - did it find (ec_ndata - 1) and a bunch of empty handoffs - less clear | 16:32 |
| acoles | clayg: we have one test that says return 404 if other backend responses were 404s, and another test that says return 503 if other backend responses were 507s | 16:32 |
| clayg | ho gosh, 507 - normally the proxy won't talk to a 507'd device very long | 16:32 |
| *** mc_nair has joined #openstack-swift | 16:33 | |
| *** breitz has joined #openstack-swift | 16:33 | |
| acoles | clayg: test.unit.proxy.test_server.TestECMismatchedFA#test_mixing_different_objects_fragment_archives | 16:34 |
| acoles | clayg: i should be clearer - 2 different etags plus a 507 -> 503 | 16:34 |
| clayg | acoles: that test is pretty gross | 16:36 |
| acoles | clayg: so i am trying to test a scenario where ALL primaries and handoffs return a frag but there is no set that reaches ec_ndata | 16:36 |
| acoles | so i have no "bad" response | 16:36 |
| acoles | and on master i get a 503 | 16:36 |
| acoles | but i feel it should be a 404? IDK | 16:36 |
| clayg | acoles: i'm not sure how much it matters - you have to make a response - just do whatever feels most consistent - if we feel we have to break consistenty later to meet client expectations (think DELETE quorum on 404 => 204) we can special case it | 16:38 |
| clayg | hopefully it won't come up that often? | 16:38 |
| acoles | i made it a 404 in optimistic gets patch but then that other test blew up. | 16:39 |
| acoles | waddya think i change the existing test to expect a 404 too ? | 16:39 |
| acoles | i.e. test.unit.proxy.test_server.TestECMismatchedFA#test_mixing_different_objects_fragment_archives | 16:39 |
| clayg | acoles: well that test seems sorta suspect to me - it was testig the mixed frags 503's case - and it didn't even pop when we did the etag_buckets - so what's going on with that? | 16:39 |
| clayg | acoles: yeah I think that test should change - i'm not sure how exactly - mabybe just delete it? I think etag_buckets/frag-mixing is pretty well covered in the test.unit.proxy.controller.test_obj tests these days | 16:40 |
| clayg | acoles: i guess that test is sorta interesting in that it hits real servers - and I think it writes objects to disks? | 16:41 |
| acoles | yeah it creates objects for realsy | 16:41 |
| acoles | on master if you get no bucket with ndata then you use the bad_gets, which has the 507, and i guess that gets morphed to a 503 somewhere?? | 16:42 |
| acoles | but in my scenario you get no bucket with ndata and no bad_gets, but we go and use bad_gets and get a 503 for having no responses | 16:43 |
| clayg | 2+1 EC with exactly three devices :\ | 16:43 |
| acoles | i'll change the expect 507 test to expect 404 and we can chew on it on gerrit | 16:43 |
| acoles | yeah - i spent >>1 minute on those tests before realising it was a 2+1 | 16:44 |
| *** DericHorn-HP has quit IRC | 16:45 | |
| clayg | acoles: i guess [200, 503, 503] => 404 does *seem* sorta counter intuitive | 16:45 |
| clayg | oh - but the 503's get skipped in the resuming getter - you never see them up top | 16:46 |
| clayg | so by the time your node iter runs out you've got [200] and you're like - k, now what? | 16:47 |
| acoles | they get skipped if there are other nodes to go to but returned if the iter is done, no? | 16:48 |
| clayg | orly!? | 16:48 |
| clayg | I thought it would just return None | 16:48 |
| clayg | mahatic: is the swiftclient headers patch awesome now!? | 16:49 |
| acoles | hmmm, you can get a None if the getter never gets a node, but I'm sure I have seen 404's from getters when they exhaust the iter | 16:49 |
| mahatic | clayg: we'll wait for jenkins :P and then you to decide | 16:50 |
| acoles | clayg: checked, i do see the 507 from the getter | 16:51 |
| acoles | clayg: maybe is should make it configurable, response_upon_inconclusive_data = [404|503] :D | 16:53 |
| clayg | acoles: I don't see how ResumingGetter._get_source_and_node could be doing that | 16:53 |
| clayg | acoles: no | 16:53 |
| acoles | jk | 16:54 |
| *** bill_az_ has joined #openstack-swift | 16:54 | |
| *** jrichli has quit IRC | 16:55 | |
| acoles | clayg: even when possible source is not a good source it saves its status to self.statuses, so last_status can be a 404 or 507 or whatever | 16:56 |
| clayg | yeah last statuses | 16:58 |
| acoles | this stuff fries my brain :( | 16:58 |
| clayg | and we iter over (getter, parts_iter) - so a series of 404's will come out (getter, None) | 16:59 |
| *** annegentle has quit IRC | 16:59 | |
| *** annegentle has joined #openstack-swift | 17:00 | |
| *** 16WAAAFWE has quit IRC | 17:01 | |
| clayg | acoles: so yeah I need to see it gerrit - it seems like best_response over the bad_gets should kind of just take all the statuses and make up something the best it can | 17:01 |
| *** haomaiwang has joined #openstack-swift | 17:01 | |
| clayg | i'm not sure why the statues = [507, 507] would come out of best_response as a 404 :) | 17:01 |
| openstackgerrit | Merged openstack/swift: Fixes recon bug with initially missing rings https://review.openstack.org/219617 | 17:02 |
| acoles | clayg: ok i'll put a TODO inline with my workaround and we can figure uot the right thing to do. | 17:03 |
| acoles | clayg: thanks! | 17:04 |
| *** wer_ has quit IRC | 17:05 | |
| clayg | acoles: if you wanna test the case of a bunch of mixed up successful responses you could special case "if not bad_gets" and stuff in some 404's or 503's - either | 17:05 |
| clayg | like _get_put_responses for replicated objects down at the bottom fills in 503's while len(resp) < len(nodes) | 17:05 |
| clayg | acoles: np! good look! | 17:06 |
| clayg | *luck | 17:06 |
| *** eranrom has joined #openstack-swift | 17:12 | |
| *** haomaiwang has quit IRC | 17:12 | |
| *** zhill has joined #openstack-swift | 17:19 | |
| *** geaaru has quit IRC | 17:24 | |
| *** blmartin has quit IRC | 17:26 | |
| *** wer has joined #openstack-swift | 17:37 | |
| *** mahatic has quit IRC | 17:49 | |
| redbo | I keep getting "fatal error: erasurecode.h: No such file or directory" when it tries to install pyeclib. Anyone have a hint? | 17:51 |
| redbo | I've tried not installing liberasurecode and installing liberasurecode with different prefixes.. | 17:53 |
| *** MVenesio has quit IRC | 17:59 | |
| *** DericHorn-HP has joined #openstack-swift | 17:59 | |
| *** silor has joined #openstack-swift | 18:02 | |
| redbo | OH, I think I got it to work. | 18:03 |
| *** jrichli has joined #openstack-swift | 18:10 | |
| *** thurloat_isgone is now known as thurloat | 18:12 | |
| *** annegentle has quit IRC | 18:12 | |
| *** annegentle has joined #openstack-swift | 18:12 | |
| *** openstackgerrit has quit IRC | 18:16 | |
| *** openstackgerrit has joined #openstack-swift | 18:16 | |
| *** changbl has quit IRC | 18:23 | |
| *** david-lyle has quit IRC | 18:30 | |
| clayg | redbo: good work! | 18:33 |
| clayg | what was the trick? | 18:33 |
| tdasilva | clayg: he re-wrote it in go ;) | 18:35 |
| clayg | yay rsync module per disk! | 18:43 |
| *** aix has quit IRC | 18:46 | |
| redbo | I wasn't doing the pip install requirements.txt before setup.py. I only had test-requirements.txt. | 18:48 |
| *** slo___ has quit IRC | 18:49 | |
| clayg | redbo: don't use 'setup.py develop' anymore - it uses setuptools and setuptools is broken - use `pip install -e .` - you get the same thing except the dependency installation of the install_requires/setup.cfg works out of the box - all hail our new python^Wopenstack packaging overlords! | 18:50 |
| jrichli | clayg: the setup.py develop step is still in the SAIO instructions. I guess that is ok if setuptools being broken is only a temporary thing. Otherwise, do you think that we should change SAIO? | 18:52 |
| clayg | redbo: i'm constantly amused when some python ecosystem tool I've used for years suddenly stops working in totally insane ways - so I go dig into the project to figure "WTF is going on over here!?" and it's the same openstack people everywhere! openstack is taking over python - soon everything you loved about python will be replaced with everything you love about openstack! Net win right! | 18:52 |
| clayg | jrichli: yeah we should change everything - it's the openstack way! If you have the option of doing it the way that's worked in the past or the new way - always prefer the new way. | 18:52 |
| clayg | SHINY! | 18:53 |
| jrichli | clayg: imagine notmyname is there to console you | 18:53 |
| clayg | well w/e - swift does things the swift way - openstack and python and however the else can GTFO-of-the-way | 18:54 |
| jrichli | SWIFTY! | 18:54 |
| clayg | that's right | 18:55 |
| clayg | damn stright | 18:55 |
| redbo | I understand packaging is sort of complex, but lots of things manage to work with "python setup.py install". I've said that too many times, I'm going to quit. Just know I still believe it. | 18:56 |
| clayg | redbo: noted! | 18:58 |
| *** rohit_ has joined #openstack-swift | 19:01 | |
| *** esker has joined #openstack-swift | 19:04 | |
| *** nadeem has joined #openstack-swift | 19:06 | |
| *** nadeem has quit IRC | 19:06 | |
| *** nadeem has joined #openstack-swift | 19:07 | |
| *** wasmum- has quit IRC | 19:08 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Enable object server to return non-durable data https://review.openstack.org/215276 | 19:09 |
| acoles | clayg: peluse ^^ now with proxy optimistic gets. its a long way from perfect but pushing what i have so far. | 19:10 |
| hrou | yea +1 to jrichli question as the new http://docs.openstack.org/developer/swift/first_contribution_swift.html#post-rebase-instructions docs also state to use setup.py after rebasing : ) | 19:16 |
| *** DericHorn-HP has quit IRC | 19:19 | |
| *** acoles is now known as acoles_ | 19:22 | |
| *** aix has joined #openstack-swift | 19:25 | |
| *** bill_az_ has quit IRC | 19:31 | |
| *** bill_az has quit IRC | 19:35 | |
| *** wasmum has joined #openstack-swift | 19:38 | |
| clayg | acoles_: thanks for pushing early! ahve a good night! | 19:38 |
| clayg | FWIW today I freak out about lp bug #1496205 | 19:39 |
| openstack | Launchpad bug 1496205 in OpenStack Object Storage (swift) "EC: Client Disconnect leaves inaccessible data on disk" [Undecided,New] https://launchpad.net/bugs/1496205 | 19:39 |
| clayg | I also want to find torgomatic and whine at him about sending 0\r\n\r\n in the *middle* of a chunked transfer only to reset wsgi input reading with a "set_hunred_continue_response_headers" call | 19:41 |
| *** Fin1te has joined #openstack-swift | 19:41 | |
| *** bill_az has joined #openstack-swift | 19:51 | |
| clayg | notmyname: jesus! in the new gerrit view I can't even see/edit the messages I've typed during the review? | 19:53 |
| clayg | notmyname: it's like "here's a bunch of random things no one asked for - oh an btw we broke a bunch of useful stuff too - you're welcome" | 19:53 |
| clayg | no wonder -infra has made it optional off by default | 19:54 |
| briancline | I hate that I can no longer hover over a reviewer's name to see their email address | 19:59 |
| * briancline is often nosey and likes to know where people are from | 20:00 | |
| tdasilva | clayg, briancline: weird, i haven't seen this new view yet... | 20:00 |
| *** openstackstatus has quit IRC | 20:02 | |
| *** silor has quit IRC | 20:04 | |
| *** openstackstatus has joined #openstack-swift | 20:04 | |
| *** ChanServ sets mode: +v openstackstatus | 20:04 | |
| clayg | tdasilva: if you click on your name in the upper right go to Settings => Preferences => Change View => "New Screen" - then you can be just as excited and then quickly dissappointed as the rest of us | 20:12 |
| clayg | you *do* want a fleating moment of hope that things might get better - DON'T YOU!? | 20:12 |
| tdasilva | clayg: tried it and changed it right back | 20:13 |
| tdasilva | :D | 20:13 |
| *** Fin1te has quit IRC | 20:13 | |
| tdasilva | gluster team had made the same change, i'm not a big fan | 20:14 |
| clayg | ah | 20:14 |
| *** Fin1te has joined #openstack-swift | 20:15 | |
| *** proteusguy_ has quit IRC | 20:15 | |
| hrou | tdasilva, clayg yea I've been playing around with it for a bunch of reviews I've been doing in the storlet project, switched back too : ) But not sure what you mean by "can't see / edit messages", you mean inline code comments ? That seems to work fine. | 20:15 |
| clayg | and so I finally get a chance to use the "change base" option with the new view - and surprise suprise - it's exactly the same problem as the old view | 20:20 |
| clayg | you get all the changes from itermediate merges before the rebase showing up in the diff making it impossible to see what's changed from one revision of the patch to the next | 20:22 |
| clayg | hrou: normally a the end of of a review when I'm putting in my final thoughts I'll look back at all the comments I made along the way to decide if any of them should be blockers - but in the new view the "reply" button is just a pop-up at the top and I can't see all the comments I made (I also like to re-read them and try fix typos and stuff before I submit) | 20:23 |
| clayg | hrou: maybe there's another way to leave a review besides the reploy drop-down at the top? | 20:24 |
| hrou | clayg, Ah gotcha ! Yea that would suck, I must have just skimmed over that aspect actually, leaving review comments inline and going back and forth between files is Ok. Let me check the final summary now. | 20:24 |
| openstackgerrit | Clay Gerrard proposed openstack/python-swiftclient: Prompt for missing password https://review.openstack.org/189387 | 20:25 |
| clayg | oh my god - there's a *rebase* button? | 20:25 |
| clayg | timburke: sorry I just rebased patch 189387 - it didn't ask for confirmation or anything!? | 20:26 |
| patchbot | clayg: https://review.openstack.org/#/c/189387/ | 20:26 |
| hrou | clayg, yep you're right : ) you can see the drafts you left (no check that, you can see that you left a comment in a given file, but not the comment it self). Agreed that's annoying. | 20:26 |
| clayg | but I do sorta think that if you rebase before you git review you might get a cleaner "what's changed from the last revision" - i'll try that next time I'm pushing up fixups over a change that's been up for awhile. | 20:26 |
| openstackgerrit | Merged openstack/swift: go: make BeginReplication concurrent https://review.openstack.org/224181 | 20:27 |
| clayg | it looks like Gerrit 2.11.3 is available - maybe *that's* better? | 20:28 |
| *** proteusguy_ has joined #openstack-swift | 20:28 | |
| *** dustins has quit IRC | 20:35 | |
| *** Fin1te has quit IRC | 20:43 | |
| *** Fin1te has joined #openstack-swift | 20:50 | |
| *** vkmc has joined #openstack-swift | 20:52 | |
| *** barker has joined #openstack-swift | 20:57 | |
| *** barker has quit IRC | 20:58 | |
| timburke | clayg: haha i remember zackmdavis doing the same thing not so long ago! | 21:06 |
| *** jrichli has quit IRC | 21:06 | |
| *** barker has joined #openstack-swift | 21:08 | |
| *** vkmc has left #openstack-swift | 21:19 | |
| *** barker has quit IRC | 21:21 | |
| *** dustins has joined #openstack-swift | 21:24 | |
| *** Fin1te has quit IRC | 21:24 | |
| *** hrou has quit IRC | 21:24 | |
| *** cdelatte has quit IRC | 21:27 | |
| *** Fin1te has joined #openstack-swift | 21:27 | |
| *** changbl has joined #openstack-swift | 21:30 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: More cleanup https://review.openstack.org/224319 | 21:36 |
| *** tongli_ has quit IRC | 21:46 | |
| *** jkugel has quit IRC | 21:46 | |
| dfg | clayg: turns out more time is spent generating hashes than sending the data over the wire :) and thats true even when you are populting a lot of new nodes in the cluster. | 21:48 |
| dfg | well- i guess that won't be true after that fix goes in. | 21:49 |
| clayg | fascinating | 21:49 |
| dfg | anyway- even without that change replication passes are x3 faster :) | 21:49 |
| clayg | ... wait - which fix? the BeginReplication? | 21:50 |
| clayg | dfg: whooot! Do you get a free soda for winning replication!? | 21:50 |
| dfg | clayg: ya the concurrent BeginReplication fix should lower timeGetHashesRemote so that it won't be the highest number of the three | 21:50 |
| dfg | which it is a lot of the time now | 21:51 |
| dfg | so next up index.db :) maybe... | 21:51 |
| clayg | ... right | 21:51 |
| clayg | dfg: but so - while I'm sure that could be true in aggregate - it can't (?) be true in general. | 21:52 |
| dfg | which? | 21:52 |
| *** barker has joined #openstack-swift | 21:52 | |
| dfg | that gethashes remote is super slow? | 21:52 |
| clayg | if you're adding a node, get remote hashes should be *very* quick - and fork-lifting the part will take time (even if golang is 3x on writes) | 21:52 |
| dfg | you'd think that | 21:53 |
| dfg | but i think it turns out that generating the hashes the way we do is just that crappy | 21:53 |
| clayg | I mean if the part already exists on the remote sure walking the suffixes and making hashes is teh suck - worse the more files you get - could easily dominate the transfer time - esspecially if the get hashes warms up the file system cache | 21:54 |
| *** Fin1te has quit IRC | 21:54 | |
| clayg | maybe - i'm glad you guys are making these discoveries - it's really awesome - congrats - tell shaggy to buy you a steak | 21:54 |
| *** changbl has quit IRC | 21:55 | |
| *** david-lyle has joined #openstack-swift | 21:56 | |
| dfg | ya- i'll have some better numbers before we do that next talk. but so far its pretty cool. | 21:56 |
| *** dustins has quit IRC | 22:07 | |
| *** pberis has quit IRC | 22:08 | |
| *** barker has quit IRC | 22:09 | |
| mattoliverau | Morning | 22:10 |
| *** wbhuber has quit IRC | 22:11 | |
| *** barker has joined #openstack-swift | 22:12 | |
| *** annegentle has quit IRC | 22:16 | |
| *** barker has quit IRC | 22:21 | |
| openstackgerrit | Zack M. Davis proposed openstack/swift: replace use of deprecated rfc822.Message with a helper utility https://review.openstack.org/203304 | 22:28 |
| *** david-lyle has quit IRC | 22:29 | |
| *** hezhiqiang has joined #openstack-swift | 22:29 | |
| *** esker has quit IRC | 22:29 | |
| *** chsc has quit IRC | 22:29 | |
| clayg | torgomatic: oh - so should I file a bug on the 0\r\n\r\n in set_hundred_continue/fuck-wsgi-input thing? Where? eventlet or swift? | 22:39 |
| torgomatic | clayg: probably swift; we shouldn't ever be sending that stuff | 22:40 |
| *** jlhinson has quit IRC | 22:40 | |
| briancline | clayg: acoles_: jrichli: wbhuber: thanks for the reviews yesterday :) | 22:41 |
| *** cdelatte has joined #openstack-swift | 22:42 | |
| *** lcurtis has quit IRC | 22:42 | |
| *** thurloat is now known as thurloat_isgone | 22:46 | |
| *** nadeem has quit IRC | 22:46 | |
| clayg | briancline: thanks for asking! | 22:48 |
| *** bill_az_ has joined #openstack-swift | 22:55 | |
| *** thurloat_isgone is now known as thurloat | 22:55 | |
| *** bill_az has quit IRC | 22:56 | |
| *** bill_az has joined #openstack-swift | 22:57 | |
| *** bill_az_ has quit IRC | 22:57 | |
| *** kota_ has joined #openstack-swift | 22:59 | |
| *** ChanServ sets mode: +v kota_ | 22:59 | |
| kota_ | good morning | 22:59 |
| clayg | kota_: good morning! | 22:59 |
| clayg | I'm sort of scared to look at your expiring objects thing - isn't aerwin3 going to fix all that for us? | 23:00 |
| kota_ | clayg: patch 224055? | 23:00 |
| patchbot | kota_: https://review.openstack.org/#/c/224055/ | 23:00 |
| *** minwoob has quit IRC | 23:04 | |
| kota_ | clayg: sorry, I'm missing your mean for aerwin3. what's that? | 23:04 |
| *** chsc has joined #openstack-swift | 23:05 | |
| *** km has joined #openstack-swift | 23:07 | |
| *** mc_nair has quit IRC | 23:09 | |
| *** wer has quit IRC | 23:21 | |
| redbo | kota_: https://review.openstack.org/#/c/221914/ | 23:24 |
| *** hezhiqiang has quit IRC | 23:24 | |
| *** hrou has joined #openstack-swift | 23:24 | |
| kota_ | redbo: thanks, is that about improvements of object-expirer brought at Austin? | 23:26 |
| *** sneti has joined #openstack-swift | 23:27 | |
| redbo | I know we talked about it a bit in Austin, but I don't remember how much. | 23:28 |
| kota_ | redbo: ok, let me look at. | 23:29 |
| kota_ | ah, aerwin3 means a nick in IRC! make sense! | 23:30 |
| *** kei_yama has joined #openstack-swift | 23:30 | |
| kota_ | clayg, redbo: thanks for poking me to head up to the spec :) | 23:34 |
| *** wer has joined #openstack-swift | 23:34 | |
| *** gyee has quit IRC | 23:35 | |
| *** jkugel has joined #openstack-swift | 23:37 | |
| *** chlong has quit IRC | 23:41 | |
| clayg | torgomatic: what do you think on lp bug #1496636 | 23:46 |
| openstack | Launchpad bug 1496636 in OpenStack Object Storage (swift) "EC: Chunked transfer/commit protocol is *not* HTTP" [Undecided,New] https://launchpad.net/bugs/1496636 | 23:46 |
| clayg | torgomatic: honestly it's probably related to lp bug #1496205 | 23:49 |
| openstack | Launchpad bug 1496205 in OpenStack Object Storage (swift) "EC: Client Disconnect leaves inaccessible data on disk" [Undecided,New] https://launchpad.net/bugs/1496205 | 23:49 |
| torgomatic | clayg: looks about right to me | 23:49 |
| clayg | torgomatic: which is the root cause of lp bug #1469094 | 23:49 |
| openstack | Launchpad bug 1469094 in OpenStack Object Storage (swift) "Timeout writing .durable can cause error on GET (under failure)" [Critical,Confirmed] https://launchpad.net/bugs/1469094 - Assigned to paul luse (paul-e-luse) | 23:49 |
| clayg | *probalby | 23:49 |
| *** wer_ has joined #openstack-swift | 23:52 | |
| *** ho has joined #openstack-swift | 23:53 | |
| *** wer has quit IRC | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
