| *** baojg has joined #openstack-swift | 00:16 | |
| *** takashi has quit IRC | 00:28 | |
| *** baojg has quit IRC | 00:42 | |
| *** m_kazuhiro has joined #openstack-swift | 00:54 | |
| *** mingyu has quit IRC | 01:01 | |
| *** chlong_ has joined #openstack-swift | 01:12 | |
| *** chlong has joined #openstack-swift | 01:12 | |
| zhengyin | good morning | 01:30 |
|---|---|---|
| zhengyin | mattoliverau,kota_: morning | 01:30 |
| mattoliverau | zhengyin: o/ | 01:31 |
| kota_ | zhengyin: o/ | 01:34 |
| kota_ | zhengyin: btw, where timezone in you are? it looks Asian Pacific? | 01:35 |
| kota_ | s/in you are/are you in/ | 01:35 |
| kota_ | maybe/ | 01:35 |
| kota_ | my bad english .. :/ | 01:35 |
| zhengyin | kota_: I am in China | 01:39 |
| kota_ | zhengyin: gotcha, almostly same timezone with me (maybe 1 hour difference?) | 01:40 |
| zhengyin | kota_:yes, you are an hour faster than me | 01:41 |
| zhengyin | now, it is 9:41 in China | 01:41 |
| kota_ | zhengyin: I'm happy looking contributors in similar timezone in active ;-) | 01:42 |
| zhengyin | kota_: thank you, I just graduated from school， I 阿吗 | 01:44 |
| zhengyin | I am a newer in here, I will learn from you and community :) | 01:46 |
| kota_ | zhengyin: nice, anytime, feel free to ask everything ;-) | 01:48 |
| zhengyin | kota_: thanks a lot. | 01:50 |
| zhengyin | :) | 01:51 |
| *** jamielennox is now known as jamielennox|away | 02:00 | |
| *** jamielennox|away is now known as jamielennox | 02:04 | |
| *** vint_bra has joined #openstack-swift | 02:17 | |
| mattoliverau | kota_, zhengyin: you clocks are both too slow :P Looks like I'm 2 hours ahead of zhengyin. Although it's Spring here so day light savings will change things soon a suspect :) | 02:17 |
| *** lifeless_ has quit IRC | 02:19 | |
| zhengyin | mattoliverau: you are in our East and in Australia？ ：） | 02:23 |
| mattoliverau | zhengyin: yup East coast of Australia :) | 02:23 |
| kota_ | beach office!!! | 02:24 |
| jrichli | zhengyin: will you be going to the summit in Barcelona? | 02:25 |
| mattoliverau | kota_: that's right! | 02:25 |
| mattoliverau | jrichli: you? | 02:25 |
| zhengyin | mattoliverau： I know Sydney is a seaside city and it is beautiful | 02:26 |
| jrichli | mattoliverau: yes. you? I remember you saying that you were waiting on approval | 02:26 |
| mattoliverau | jrichli: yup \o/. Finally got the approval :) | 02:26 |
| jrichli | Yay! | 02:26 |
| jrichli | hope you didnt have to go with a shady hotel miles away! | 02:27 |
| mattoliverau | zhengyin: I'm about 3 hours south (driving) from Sydney, in a costal town on the beach :) | 02:27 |
| jrichli | the hotel search was a bit challenging to stay in policy this time | 02:28 |
| mattoliverau | jrichli: not this time, Rackspace has a block and actually had room for us Aussies thus time :) | 02:28 |
| jrichli | nice! | 02:28 |
| *** vint_bra has quit IRC | 02:29 | |
| zhengyin | jrichli: sorry, Leadership is not approved，I will not go to the summit in Barcelona, | 02:29 |
| mattoliverau | zhengyin: always next time :( | 02:30 |
| jrichli | zhengyin: I am disappointed to here that we won't get to meet you in person yet :-( maybe the ... what is it called ... PTG? | 02:30 |
| jrichli | the dev meetup organized by openstack | 02:30 |
| jrichli | in Feb | 02:30 |
| *** tqtran has joined #openstack-swift | 02:31 | |
| zhengyin | I'm sorry that I can't see you this time | 02:31 |
| jrichli | s/here/hear/ ... yes, jrichli does know how to spell. most words. | 02:31 |
| mattoliverau | jrichli: good for you, cause I don't :P | 02:32 |
| jrichli | mattoliverau: well, I don't even know many words - but my interactions with acoles has increased my number! | 02:33 |
| mattoliverau | zhengyin: PTG in Atlanta, then Forum (summit) in Boston.. the the next forum would be better cause it'll be in Sydney (but I might be a bit biased) :P | 02:33 |
| *** lifeless has joined #openstack-swift | 02:34 | |
| mattoliverau | jrichli: lol | 02:34 |
| mattoliverau | of course | 02:34 |
| jrichli | Can't wait for Sydney! | 02:34 |
| *** tqtran has quit IRC | 02:35 | |
| * mattoliverau might go enjoy this Spring weather with a walk on the beach ;) | 02:35 | |
| mattoliverau | bbs | 02:35 |
| jrichli | nice, have fun! | 02:36 |
| zhengyin | mattoliverau: have fun :) | 02:37 |
| zhengyin | mattoliverau: do you have time to help me review patch 366689? :) | 02:45 |
| patchbot | https://review.openstack.org/#/c/366689/ - swift - Add test cases in the common/test_db | 02:45 |
| *** StraubTW has quit IRC | 02:49 | |
| *** Jeffrey4l has quit IRC | 02:52 | |
| *** Jeffrey4l has joined #openstack-swift | 02:58 | |
| openstackgerrit | Merged openstack/swift: Add troubleshooting tips to keystoneauth docs https://review.openstack.org/366868 | 03:15 |
| notmyname | hello, again | 03:17 |
| notmyname | zhengyin: I'm sorry I won't see you in barcelona | 03:17 |
| zhengyin | notmyname: hi, notmyname, I may see you and friends of Community next time :) | 03:21 |
| notmyname | zhengyin: I hope so | 03:21 |
| zhengyin | notmyname: when is this version released? | 03:23 |
| *** okdas has quit IRC | 03:24 | |
| notmyname | the next version of swift or the next combined openstack release? well, actually right now it's the same answer | 03:24 |
| notmyname | the openstack newton release will be on the week of october 3, and it will include swift 2.10.0 | 03:25 |
| notmyname | https://releases.openstack.org/newton/schedule.html | 03:26 |
| *** okdas has joined #openstack-swift | 03:26 | |
| *** okdas has joined #openstack-swift | 03:26 | |
| notmyname | in fact, tonight I'm finishing up a presentation and then looking at the starred patches to see what can land for 2.10.0 | 03:26 |
| notmyname | specifically, we need to have the 2.10.0 release by the end of the month. so, basically, any point between now and end of the month we can tag it and call it 2.10.0 | 03:27 |
| zhengyin | well, I know and I will commit some patchs before this release :) | 03:28 |
| notmyname | :-) | 03:29 |
| zhengyin | and review some patch | 03:29 |
| notmyname | yes, definitely do reviews. that's very helpful | 03:29 |
| notmyname | I'll be updating https://wiki.openstack.org/wiki/Swift/PriorityReviews soon | 03:30 |
| notmyname | but in the meantime, looking at the starred reviews is good | 03:30 |
| notmyname | oh. that link isn't in the topic any more | 03:30 |
| notmyname | hang on | 03:30 |
| zhengyin | ok, I start to learn to review patch now, and it may have mistake :) | 03:31 |
| *** ChanServ changes topic to "Let's talk, we're nice. | Reviews: http://goo.gl/mtEv1C | Ideas: https://wiki.openstack.org/wiki/Swift/ideas | Logs: http://eavesdrop.openstack.org/irclogs/%23openstack-swift/ | Meetings: https://wiki.openstack.org/wiki/Meetings/Swift" | 03:32 | |
| notmyname | http://goo.gl/mtEv1C | 03:32 |
| notmyname | zhengyin: don't worry about making review "mistakes". it's not even possible to be a mistake to write down your thoughts on a patch | 03:33 |
| zhengyin | notmyname: ok, I will try my best to do it :) | 03:35 |
| zhengyin | notmyname: thanks notmyname, I will have lunch now :) | 03:37 |
| *** baojg has joined #openstack-swift | 03:43 | |
| *** baojg has quit IRC | 03:47 | |
| openstackgerrit | Merged openstack/python-swiftclient: Fix order of arguments in assertIs https://review.openstack.org/367798 | 04:06 |
| jrichli | new versions of the docs will be released with this, right? so it'd be good to have patch 354767 | 04:09 |
| patchbot | https://review.openstack.org/#/c/354767/ - swift - Corrections for the API specifications in api-ref | 04:09 |
| notmyname | jrichli: I'd assume those are built at every commit like the ones in doc/ | 04:17 |
| notmyname | that would be a good thing to check, actually | 04:17 |
| notmyname | anyone know the py3 version available in the gate? py34? py35? | 04:21 |
| *** ppai has joined #openstack-swift | 04:22 | |
| mattoliverau | notmyname: depends on if you get a trusty or xenial cloud server | 04:26 |
| mattoliverau | I think | 04:26 |
| mattoliverau | or maybe it's all 3.5 and it either 3.5.1 or 3.5.2, I can't remember | 04:27 |
| notmyname | mattoliverau: in the context of the tox envs available, do you know? | 04:27 |
| mattoliverau | not of the top of my head, would need to ask someone in infra, or call out to the infra guys who dont sleep :P .. clarkb? | 04:29 |
| notmyname | yeah, I asked in -infra too ;-) | 04:29 |
| mattoliverau | sigh debugging my first attempt at read only container while sharding. Meaning while in sharding state have 2 sqlite containers, and have one node as the pivot point scanner as to get the pivot points as fast as possible to redirect ingress traffic faster and get to the ideal state faster.... | 04:32 |
| mattoliverau | and boy, what was I on when I wrote some of this :P | 04:33 |
| * mattoliverau reworks the sharder yet again | 04:33 | |
| notmyname | mattoliverau: did you see the thing clayg posted earlier (IIRC) that was about remote sqlite DBs? | 04:34 |
| notmyname | mattoliverau: "2016-09-07_freenode.txt:11:08 clayg: mattoliverau: holy crap! https://www.sqlite.org/lang_attach.html <- from sqlite ML - you can tie two databases together so you can inserting into kind of statements? pretty wild" | 04:37 |
| mattoliverau | yup, it looks pretty good, book marked it and will give a more detailed look soon, but is based of table names, so you'd need to use different table names in the two.. not that that's nessarily bad. | 04:38 |
| mattoliverau | the idea is to only have the tables you need in the new database, but at the moment I've making an entire new container db (with pivot tables). Cause it's easier and I need to track things like replication etc as I don't know how long sharding will take. | 04:39 |
| notmyname | so I haven't talked to him about it, but I would *love* to see kota_ working with you on the container sharding problem. if that happened, I think we could get something soon(-sh) in prod | 04:40 |
| mattoliverau | notmyname: cool! that would be awesome, in any case, I'm just pushing ahead, will try and keep the trello board as up to date as possible. I want to have something that works (with the big database that isn't written to) and see how much better it performs. | 04:42 |
| mattoliverau | ^ by/before barcelona | 04:42 |
| notmyname | mattoliverau: I'd love to see sharding in the Ocata release. :-) | 04:44 |
| openstackgerrit | Merged openstack/swift-bench: Remove support for py33/py26 https://review.openstack.org/264486 | 04:45 |
| notmyname | wow. swift-bench patches merge so quickly | 04:45 |
| notmyname | dozens of seconds, tops. | 04:45 |
| *** psachin has joined #openstack-swift | 04:47 | |
| mattoliverau | lol, I'm sure we have a bunch of tests there :) | 04:47 |
| mattoliverau | notmyname: I'd love to see it merged.. then I can pick some new feature to work on :) | 04:48 |
| notmyname | honestly, the think I hate most about swift-bench is that is shines a light on how painfully slow it is to get patches to swift merged. | 04:48 |
| notmyname | mattoliverau: I've got a list. ;-) | 04:49 |
| mattoliverau | it would be good to get more/another eye on the sharding POC, that I hope will be turned in to code... it's gone through so many POCs there is probably cruft from old things that don't need to be done any more.. I try and start from a clean slate each time, but I do get lazy and still finding stiff :P | 04:50 |
| notmyname | patch 162243 would be nice to land this week | 04:50 |
| patchbot | https://review.openstack.org/#/c/162243/ - swift - Make object creation more atomic in Linux | 04:50 |
| mattoliverau | +1 | 04:52 |
| notmyname | mattoliverau: oh was that you echoing what you left in gerrit? ;-) | 04:52 |
| mattoliverau | lol | 04:52 |
| mattoliverau | notmyname: I keep getting distracted, I'll try and play with it again this week :) | 04:53 |
| notmyname | acoles_: I'd love to see https://review.openstack.org/#/c/215276/ land this week to be in the release, but it's got a -1 from kota_. what do you think about it? | 04:53 |
| patchbot | patch 215276 - swift - Enable object server to return non-durable data | 04:53 |
| notmyname | mattoliverau: getting close to release time. no time to be distracted ;-) | 04:53 |
| notmyname | mattoliverau: actually, yeah. reviewing that would be more important than container sharding this week | 04:53 |
| mattoliverau | kk :) | 04:54 |
| notmyname | is reedip banerjee online? | 05:00 |
| *** Jeffrey4l has quit IRC | 05:00 | |
| *** Jeffrey4l has joined #openstack-swift | 05:00 | |
| jrichli | prob doesn't have to be in next release, but a change that was talked about with ops way back in Vancouver is in patch 218490 - which is ready to review again | 05:02 |
| patchbot | https://review.openstack.org/#/c/218490/ - swift - Automatic refresh of memcache config settings | 05:02 |
| notmyname | oh that hasn't landed yet? :-( | 05:03 |
| notmyname | oh. I'm the one who last gave it a -1. carry on | 05:03 |
| *** chlong_ has quit IRC | 05:03 | |
| *** chlong has quit IRC | 05:03 | |
| jrichli | :-) but i think i addressed all that now | 05:03 |
| notmyname | acoles_ has marked https://review.openstack.org/#/c/348495/ as a critical bug (https://bugs.launchpad.net/swift/+bug/1605597). it affects prod, so it's an important one for the release | 05:06 |
| openstack | Launchpad bug 1605597 in OpenStack Object Storage (swift) "container sync cannot copy SLOs" [Critical,In progress] - Assigned to Alistair Coles (alistair-coles) | 05:06 |
| patchbot | patch 348495 - swift - Make container sync copy SLO manifests | 05:06 |
| notmyname | I'm turning in for the night. talk to you tomorrow | 05:15 |
| notmyname | tomorrow my plan is to have the priority reviews wiki page updated | 05:15 |
| *** chlong_ has joined #openstack-swift | 05:16 | |
| *** chlong has joined #openstack-swift | 05:20 | |
| *** chlong has quit IRC | 05:25 | |
| *** ppai has quit IRC | 05:26 | |
| *** tqtran has joined #openstack-swift | 05:33 | |
| *** rcernin has joined #openstack-swift | 05:33 | |
| *** tqtran has quit IRC | 05:37 | |
| zhengyin | jrichli: hi, jrichli， do you have time to help me review a patch 355858? thanks :) | 05:43 |
| patchbot | https://review.openstack.org/#/c/355858/ - swift - simplify code | 05:43 |
| mattoliverau | notmyname: o/ | 05:44 |
| * mattoliverau sorry me went to make a new coffee and got sidetracked with a cute baby :) | 05:45 | |
| *** ppai has joined #openstack-swift | 05:52 | |
| *** ChubYann has quit IRC | 05:55 | |
| *** chsc has joined #openstack-swift | 05:55 | |
| *** chsc has quit IRC | 06:01 | |
| *** chlong_ has quit IRC | 06:05 | |
| mahatic | clayg: just wanted to remind you on patch 346865, while it's still on your mind (somewhat ;) | 06:05 |
| patchbot | https://review.openstack.org/#/c/346865/ - swift - Delete old tombstones | 06:05 |
| mahatic | of patch* | 06:06 |
| *** chsc has joined #openstack-swift | 06:25 | |
| *** chsc has quit IRC | 06:30 | |
| *** pcaruana has joined #openstack-swift | 06:49 | |
| *** tesseract- has joined #openstack-swift | 07:00 | |
| *** rledisez has joined #openstack-swift | 07:06 | |
| *** hseipp has joined #openstack-swift | 07:25 | |
| *** tqtran has joined #openstack-swift | 07:34 | |
| *** tqtran has quit IRC | 07:38 | |
| *** oshritf has joined #openstack-swift | 07:42 | |
| *** sams-gleb has joined #openstack-swift | 07:53 | |
| *** asettle has joined #openstack-swift | 08:02 | |
| openstackgerrit | Tuan Luong-Anh proposed openstack/swift: Use ConfigParser instead of SafeConfigParser https://review.openstack.org/368617 | 08:02 |
| openstackgerrit | zheng yin proposed openstack/swift: Add test_long_names in the test_account and test_container https://review.openstack.org/368627 | 08:20 |
| *** joeljwright has joined #openstack-swift | 08:34 | |
| *** ChanServ sets mode: +v joeljwright | 08:34 | |
| *** bikmak has joined #openstack-swift | 08:55 | |
| *** openstackgerrit has quit IRC | 09:04 | |
| *** openstackgerrit has joined #openstack-swift | 09:05 | |
| *** hoonetorg has joined #openstack-swift | 09:05 | |
| *** mingyu has joined #openstack-swift | 09:07 | |
| *** oshritf has quit IRC | 09:09 | |
| *** zhugaoxiao has quit IRC | 09:16 | |
| *** zhugaoxiao has joined #openstack-swift | 09:17 | |
| *** geaaru has joined #openstack-swift | 09:21 | |
| *** trananhkma has joined #openstack-swift | 09:21 | |
| *** tqtran has joined #openstack-swift | 09:35 | |
| *** tqtran has quit IRC | 09:40 | |
| *** mingyu_ has joined #openstack-swift | 09:44 | |
| *** mingyu has quit IRC | 09:46 | |
| *** mingyu has joined #openstack-swift | 09:59 | |
| zhengyin | clayg: hi clayg, are you here? | 10:02 |
| *** mingyu_ has quit IRC | 10:02 | |
| *** mingyu has quit IRC | 10:02 | |
| *** mingyu has joined #openstack-swift | 10:02 | |
| *** mingyu has quit IRC | 10:07 | |
| *** oshritf has joined #openstack-swift | 10:21 | |
| *** kei_yama has quit IRC | 10:22 | |
| *** psachin has quit IRC | 10:33 | |
| *** nikivi has joined #openstack-swift | 10:37 | |
| *** mingyu has joined #openstack-swift | 10:46 | |
| *** diogogmt_ has joined #openstack-swift | 10:52 | |
| *** diogogmt has quit IRC | 10:53 | |
| *** diogogmt_ is now known as diogogmt | 10:53 | |
| *** oshritf has quit IRC | 10:58 | |
| *** acoles_ is now known as acoles | 11:00 | |
| *** trananhkma has quit IRC | 11:01 | |
| *** hoonetorg has quit IRC | 11:03 | |
| acoles | notmyname: as it happens I planned to work on patch 215276 today - Kota has been rescuing it from merge conflict, IIRC it's needed for his Global EC work | 11:03 |
| patchbot | https://review.openstack.org/#/c/215276/ - swift - Enable object server to return non-durable data | 11:03 |
| *** silor has joined #openstack-swift | 11:08 | |
| *** mingyu has quit IRC | 11:09 | |
| *** cdelatte has joined #openstack-swift | 11:10 | |
| *** daemontool has joined #openstack-swift | 11:19 | |
| *** silor has quit IRC | 11:35 | |
| *** silor has joined #openstack-swift | 11:35 | |
| *** tqtran has joined #openstack-swift | 11:37 | |
| *** ppai has quit IRC | 11:38 | |
| *** silor has quit IRC | 11:39 | |
| *** silor has joined #openstack-swift | 11:40 | |
| *** tqtran has quit IRC | 11:41 | |
| *** silor1 has joined #openstack-swift | 11:43 | |
| *** silor has quit IRC | 11:44 | |
| *** silor1 is now known as silor | 11:44 | |
| *** bikmak has quit IRC | 11:45 | |
| *** ppai has joined #openstack-swift | 11:52 | |
| *** daemontool has quit IRC | 11:56 | |
| *** daemontool has joined #openstack-swift | 12:05 | |
| zhengyin | clayg: hi, clayg. I have a patch 367825 to remove "TODO" message that you left. I want to know you will add more tests or perhaps you forget to remove "TODO". thanks | 12:07 |
| patchbot | https://review.openstack.org/#/c/367825/ - swift - delete TODO in test_manager | 12:07 |
| *** psachin has joined #openstack-swift | 12:09 | |
| *** mingyu has joined #openstack-swift | 12:10 | |
| *** mingyu has quit IRC | 12:14 | |
| *** daemontool has quit IRC | 12:27 | |
| *** daemontool has joined #openstack-swift | 12:28 | |
| *** oshritf has joined #openstack-swift | 12:35 | |
| *** m_kazuhiro has quit IRC | 12:53 | |
| *** mingyu has joined #openstack-swift | 12:58 | |
| *** nikivi has quit IRC | 12:58 | |
| *** Jeffrey4l has quit IRC | 12:59 | |
| *** Jeffrey4l has joined #openstack-swift | 12:59 | |
| *** nikivi has joined #openstack-swift | 13:08 | |
| *** mingyu has quit IRC | 13:08 | |
| *** nikivi has quit IRC | 13:12 | |
| *** mingyu has joined #openstack-swift | 13:14 | |
| *** openstackgerrit has quit IRC | 13:19 | |
| *** openstackgerrit has joined #openstack-swift | 13:19 | |
| *** bikmak has joined #openstack-swift | 13:22 | |
| *** vint_bra has joined #openstack-swift | 13:32 | |
| *** gabor_antal_ has joined #openstack-swift | 13:34 | |
| *** gabor_antal has quit IRC | 13:34 | |
| *** klamath_ has joined #openstack-swift | 13:35 | |
| *** tqtran has joined #openstack-swift | 13:38 | |
| *** ppai has quit IRC | 13:41 | |
| *** tqtran has quit IRC | 13:43 | |
| *** psachin has quit IRC | 13:43 | |
| *** mingyu has quit IRC | 13:49 | |
| *** ametts has joined #openstack-swift | 13:54 | |
| *** lcurtis has quit IRC | 14:02 | |
| *** tongli has joined #openstack-swift | 14:03 | |
| *** bikmak has quit IRC | 14:13 | |
| *** bikmak has joined #openstack-swift | 14:14 | |
| *** kodoku has joined #openstack-swift | 14:16 | |
| kodoku | HI, anyone use the module puppet swift for deploy ? | 14:20 |
| *** bikmak has quit IRC | 14:24 | |
| *** dfg_ has joined #openstack-swift | 14:25 | |
| *** mingyu has joined #openstack-swift | 14:26 | |
| *** baojg has joined #openstack-swift | 14:27 | |
| *** bikmak has joined #openstack-swift | 14:28 | |
| *** ChanServ sets mode: +v tdasilva | 14:32 | |
| *** mvk has quit IRC | 14:37 | |
| *** hoonetorg has joined #openstack-swift | 14:38 | |
| *** bikmak has quit IRC | 14:43 | |
| *** _JZ_ has joined #openstack-swift | 14:43 | |
| *** bikmak has joined #openstack-swift | 14:45 | |
| *** Fin1te has joined #openstack-swift | 14:46 | |
| *** _JZ_ has quit IRC | 14:47 | |
| *** _JZ_ has joined #openstack-swift | 14:48 | |
| *** chsc has joined #openstack-swift | 14:53 | |
| *** mingyu has quit IRC | 15:05 | |
| *** mingyu has joined #openstack-swift | 15:06 | |
| *** daemontool_ has joined #openstack-swift | 15:07 | |
| *** daemontool has quit IRC | 15:10 | |
| *** zul has joined #openstack-swift | 15:12 | |
| *** chsc has quit IRC | 15:17 | |
| *** baojg has quit IRC | 15:20 | |
| *** dfg_ has quit IRC | 15:24 | |
| ntata | Good Morning! | 15:26 |
| *** oshritf has quit IRC | 15:28 | |
| acoles | ntata: good morning | 15:28 |
| acoles | kota_: I started work on your fanatastic review of optimistic GETs, but only part way through today. There are some mistakes form rebase i think, so I will fix and push a new version once I am finsihed. | 15:29 |
| acoles | finished* | 15:29 |
| kota_ | Thanks acoles! And sorry my bad rebase... | 15:30 |
| *** sams-gleb has quit IRC | 15:34 | |
| *** Jeffrey4l has quit IRC | 15:35 | |
| acoles | kota_: not your fault, just rebase collateral damage :) | 15:36 |
| *** Jeffrey4l has joined #openstack-swift | 15:36 | |
| *** acoles is now known as acoles_ | 15:37 | |
| *** tqtran has joined #openstack-swift | 15:40 | |
| *** thebloggu has joined #openstack-swift | 15:41 | |
| *** thebloggu has quit IRC | 15:43 | |
| *** nikivi has joined #openstack-swift | 15:43 | |
| *** tsg has joined #openstack-swift | 15:44 | |
| *** tqtran has quit IRC | 15:45 | |
| *** corvus is now known as jeblair | 15:50 | |
| *** rcernin has quit IRC | 15:53 | |
| *** pcaruana has quit IRC | 15:56 | |
| *** diogogmt has quit IRC | 15:57 | |
| notmyname | good morning | 16:00 |
| *** tongli has quit IRC | 16:01 | |
| *** zul has quit IRC | 16:04 | |
| clayg | zhengyin: checking | 16:04 |
| *** thumpba has joined #openstack-swift | 16:06 | |
| *** ChubYann has joined #openstack-swift | 16:08 | |
| *** daemontool_ has quit IRC | 16:09 | |
| *** daemontool has joined #openstack-swift | 16:10 | |
| *** dfg_ has joined #openstack-swift | 16:15 | |
| *** takashi has joined #openstack-swift | 16:15 | |
| *** diogogmt has joined #openstack-swift | 16:19 | |
| *** hseipp has quit IRC | 16:26 | |
| *** rledisez has quit IRC | 16:40 | |
| *** Fin1te has quit IRC | 16:42 | |
| *** dmorita has joined #openstack-swift | 16:45 | |
| *** ametts has quit IRC | 16:45 | |
| *** diogogmt has quit IRC | 16:47 | |
| notmyname | tdasilva: can you look at mattoliverau's comment on https://review.openstack.org/#/c/162243/ ? | 16:51 |
| patchbot | patch 162243 - swift - Make object creation more atomic in Linux | 16:51 |
| tdasilva | notmyname: I think ppai replied to mattoliverau's comment | 16:53 |
| notmyname | oh, ok ;-) | 16:53 |
| *** hurricanerix has quit IRC | 16:53 | |
| notmyname | ah yes. I'd just seen the emails | 16:53 |
| tdasilva | :) | 16:54 |
| *** hurricanerix has joined #openstack-swift | 16:54 | |
| notmyname | ok, so it looks likely that mattoliverau will change his vote when he wakes up. should see the patch land, then, for the release (yay!) | 16:55 |
| *** tesseract- has quit IRC | 16:56 | |
| *** Suyi_ has joined #openstack-swift | 17:02 | |
| openstackgerrit | Merged openstack/swift: delete TODO in test_manager https://review.openstack.org/367825 | 17:14 |
| *** joeljwright has quit IRC | 17:17 | |
| clayg | ^ phew, glad we got *that* taken care of | 17:18 |
| *** diogogmt has joined #openstack-swift | 17:21 | |
| *** klrmn has joined #openstack-swift | 17:21 | |
| *** tqtran has joined #openstack-swift | 17:27 | |
| *** ukaynar has joined #openstack-swift | 17:27 | |
| notmyname | clayg: now that we got the critical TODO deletion landed... | 17:29 |
| notmyname | clayg: can you take a look at https://review.openstack.org/#/c/348495/? acoles_ has it as a critical bug and it's affecting their prod | 17:30 |
| patchbot | patch 348495 - swift - Make container sync copy SLO manifests | 17:30 |
| clayg | YESSIR! | 17:31 |
| notmyname | thanks :-) | 17:31 |
| pdardeau | clayg: nice posting to ML! | 17:31 |
| clayg | pdardeau: no no no - i'm pretty sure this is me -> http://66.media.tumblr.com/tumblr_m9yx3gTr9d1rzupqxo1_500.png | 17:33 |
| *** asettle has quit IRC | 17:33 | |
| clayg | you're comments on the review where *much* more helpful | 17:33 |
| *** takashi has quit IRC | 17:34 | |
| clayg | I'm sure there's other contributors like me that feel compleatly illequipped to get involved in these soft mushy OpenStack debates - but it's not because we don't have an opinon on what it is we're supposed to be doing here - so while I maybe can't help review - i sure can vote! | 17:34 |
| notmyname | jrichli: you left a +1 on https://review.openstack.org/#/c/357559/. can you look at it again with an eye to giving it a +2, or do you need timburke to answer your question first? | 17:35 |
| patchbot | patch 357559 - swift - Include metadata in PUT/POST responses | 17:35 |
| tdasilva | yeah, would be great to see that merged | 17:35 |
| notmyname | tdasilva: yeah, why'd timburke add you as co-author? ;-) | 17:36 |
| jrichli | I did want to know Tim's thoughts on my last question | 17:36 |
| timburke | ah, right! yes, my thinking was that there would (eventually) be similar patches for accounts/containers | 17:37 |
| timburke | notmyname: i could always drop tdasilva... :P | 17:37 |
| jrichli | ok. I jsut wanted to be sure we had thought about whether or not a consistent approach would be ok for acct/cont | 17:37 |
| jrichli | timburke: no need for that! I will +2 in a bit | 17:38 |
| timburke | i haven't thought very *deeply* about it, but it seems like it should work fine (provided we document some caveats about eventual consistency and all that) | 17:38 |
| tdasilva | timburke: i'm actually looking at patch 347538 atm :) | 17:39 |
| patchbot | https://review.openstack.org/#/c/347538/ - swift - Store SLO Etag and size in sysmeta | 17:39 |
| timburke | yay! | 17:40 |
| tdasilva | timburke: at first I thought you were also changing the way the etag is calculated, as in, being the real etag, but no way around that yet | 17:43 |
| clayg | acoles_: reading the bug report and commit message for patch 348495 | 17:44 |
| patchbot | https://review.openstack.org/#/c/348495/ - swift - Make container sync copy SLO manifests | 17:44 |
| timburke | tdasilva: no there is not :-( although i had some crazy idea a few weeks ago that might let me store an honest-to-god Content-MD5 if you trust your upload connection enough | 17:45 |
| clayg | not sure I agree with the assumption that the destination cluster need not validate the sending clusters manifest - while I think container sync realms supposes some co-ordination between the two clusters they may have different operators on different release/upgrade schedules - i'm not sure exactly how much trust and conformity is really expected and required (lack of experience on my part) | 17:45 |
| timburke | although it would require a single connection and, as oxinabox noticed, that'll kill throughput | 17:46 |
| timburke | trade-offs. all about trade-offs | 17:46 |
| clayg | acoles_: but secondly, unlike DLO - i'm not sure it's sufficient to sync the manifest before the targeted segments have been sent - obviously in the ideal the slo manifest would exist in the destination iff it's complete and validated in the destination cluster | 17:47 |
| notmyname | timburke: your comment ("...if you trust your upload connection") reminds me of number 1: https://en.wikipedia.org/wiki/Fallacies_of_distributed_computing | 17:47 |
| clayg | ain't no body wants to give up all that throughput just for checksums | 17:48 |
| clayg | acoles_: I think the minimum thing we might be able to get away with is to not have cotainer sync progress halt when it hits a SLO (or any other object that is rejected by the remote cluster's validation for whatever reason?) | 17:49 |
| *** nikivi has quit IRC | 17:49 | |
| clayg | notmyname: did we ever get anything like per-policy constraints? | 17:50 |
| notmyname | clayg: not yet | 17:50 |
| *** zul has joined #openstack-swift | 17:50 | |
| timburke | notmyname: hey, sometimes you're willing to accept that the object stopped abruptly. the particular use-case that made me think of it was streaming surveillance video -- i have an indefinite amount of data i want to store and i want to minimize my buffering. if the connection drops, w/e, i'll start another one | 17:50 |
| timburke | clayg: having a useful checksum was just a bonus | 17:51 |
| *** dfg_ has quit IRC | 18:04 | |
| *** chsc has joined #openstack-swift | 18:05 | |
| clayg | acoles_: ok, the 409 seems reasonable.... | 18:06 |
| *** Fin1te has joined #openstack-swift | 18:13 | |
| *** nikivi has joined #openstack-swift | 18:14 | |
| *** dmorita has quit IRC | 18:15 | |
| *** dmorita has joined #openstack-swift | 18:16 | |
| *** vinsh has joined #openstack-swift | 18:17 | |
| *** nikivi has quit IRC | 18:24 | |
| *** ukaynar has quit IRC | 18:26 | |
| *** dmorita has quit IRC | 18:30 | |
| notmyname | https://wiki.openstack.org/wiki/Swift/PriorityReviews has been updated with review priorities for 2.10.0 | 18:32 |
| tdasilva | notmyname: are you maintaining that list or is it ok to add there? | 18:33 |
| notmyname | well, it's a big list already, and based on what has been brought up before. I'd prefer that you coordinate with me before updating it | 18:34 |
| *** dmorita has joined #openstack-swift | 18:36 | |
| tdasilva | notmyname: ok | 18:36 |
| tdasilva | notmyname: was hoping patch 361439 would make the cut | 18:37 |
| patchbot | https://review.openstack.org/#/c/361439/ - swift - Add option to set default version mode | 18:37 |
| *** dmorita has quit IRC | 18:39 | |
| *** dmorita_ has joined #openstack-swift | 18:39 | |
| jrichli | tdasilva: I have held off on that one because clayg gave a convincing argument that the default should remain with past behavior. At least, that was my take form the discussion. | 18:40 |
| tdasilva | jrichli: oh, I'm not changing the default at al | 18:40 |
| tdasilva | jrichli: the patch only provides the option for operators to change the default if they choose too | 18:41 |
| clayg | yeah but you're giving cluster operators the option of changing the default when the mode is underspecified | 18:41 |
| tdasilva | clayg: correct | 18:41 |
| clayg | yeah, i'm pretty sure that's actively hostile to apps that want to make use of swift's versioning | 18:42 |
| clayg | and they've already suffered enough | 18:42 |
| tdasilva | clayg: for current apps, nothing changes (as long as operators don't change anything). This change is geared towards new installations where ideally new apps would not even know about stack mode | 18:43 |
| clayg | x-versions-location should die in a fire, x-history-container and x-stack-container FTW! | 18:43 |
| jrichli | yes, that is what I had taken from clayg said before. that we shouldnt even provide the option. His reasoning showed me how this is a different case than other configuragbles we add in later | 18:43 |
| clayg | "as long as operators don't change anything" is the worst pre-req for adding an operator tunable in the world! | 18:43 |
| clayg | here's a knob - DO NOT TOUCH IT EVAR | 18:44 |
| tdasilva | it's not DO NOT touch, it's you should know what you are doing and what your users want | 18:44 |
| tdasilva | "openstack for operators and operators for users" ;-) | 18:44 |
| clayg | tdasilva: i don't out-right reject that it's *possible* for swift clusters to exist in isolation - but I would like it not to be the expectation that "only apps written after the deployment of this cluster will be used with this cluster" | 18:44 |
| jrichli | i think its more about the fact that the change is visible to the client's actions | 18:45 |
| clayg | tdasilva: I think it's a bandaid for us writing a weird/bad bi-modal api | 18:45 |
| clayg | yes, we don't want users to have to say x-versions-location AND x-version-mode - that's not great | 18:45 |
| clayg | no, we don't want users to not know what they get if they DON'T set x-version-mode | 18:45 |
| clayg | we would prefer they set x-version-history-container and x-version-stack-container and always know what they get | 18:46 |
| clayg | (and hope they never choose x-version-stack-container because they are not mad men) | 18:46 |
| tdasilva | so you propose yet a new api? | 18:46 |
| tdasilva | this reminds of a xkcd i saw somewhere | 18:47 |
| clayg | i don't think the current one is released - but yes - I think the x-version-mode is weird/bad - it might make sense at the data model layer - and i'm not against leaving the internals exposed - but the one we should document and advertise should look like you're picking between two different things by setting two different things (not by flipping a mode on an old bad crufty thing) | 18:48 |
| timburke | tdasilva: or for that matter, one of my comments on https://review.openstack.org/#/c/347538/ -- "I was getting annoyed at needing to send the same value twice, so I found a way to send it three times instead >_<" | 18:48 |
| patchbot | patch 347538 - swift - Store SLO Etag and size in sysmeta | 18:48 |
| clayg | I think the data model is acctually nice - if we had a layer that *just* turned x-version-stack-container and x-version-history-container into the always setting of the two backend flags if you try to change between one or the two it might always overwrite it? | 18:48 |
| clayg | timburke: tdasilva: ok, perhaps I'm in the minority with this mode flag being cumbersome - but I think that fact that tdasilva is suggesting we have a operator tunable to change it smells like I'm not the only one? | 18:50 |
| clayg | I would rather it stay as is than expose the tunable tho - i'd be happy to say as much as on the review - but I didn't think that's as helpful as suggesting an alternative path forward - but if no one likes it... maybe i'm better to keep my mouth shut :D | 18:51 |
| timburke | clayg: i'm actually rather OK with your suggestion -- this thing behaves in two rather different ways; send one of two rather different headers, and error out if you send both | 18:52 |
| tdasilva | clayg: i don't disagree with it being cumbersome, i just wish we had had this conversation before, so that we wouldn't have to fix it now | 18:52 |
| clayg | tdasilva: sorry - i kept meaning to look at it - but it seemed stalled - then by the time it was unblocked it was merged | 18:53 |
| clayg | tdasilva: totally unfair for me to complain about it now - but here we are! | 18:53 |
| tdasilva | clayg: no worries and it is not your fault, i think this was one of those features that actually got very little eyes on it | 18:54 |
| clayg | either way - the more I think the data model the more I think this is just syntactic sugar - i'm sure it'll be easy to gloss up later | 18:54 |
| timburke | tdasilva: sounds like versioned_writes all right ;-) | 18:54 |
| tdasilva | lol | 18:55 |
| *** Fin1te has quit IRC | 18:56 | |
| clayg | timburke: tdasilva: so is there something that needs to happen in this release - or is there additional debate - or is the operator default config option useful in addition to any future syntax/ux work? | 18:56 |
| timburke | clayg: if we get the sugar in this release, we never have to support a user-toggle-able x-versions-mode | 18:57 |
| jrichli | timburke: so, ... you already have it working and just need to write the tests, am I right? | 18:58 |
| timburke | lol i'm gonna need more coffee... | 18:58 |
| clayg | timburke: well based on the data model i'm not sure that x-versions-location and x-versions-mode wouldn't always be exposed - with the other two headers as an alias to explictly set them both with one flag? | 19:00 |
| *** zul has quit IRC | 19:00 | |
| clayg | well, i guess we always need to expose x-versions-location - so the question is just about if we continue to persist x-version-history-container as x-versions-location + x-versions-mode | 19:01 |
| clayg | ... I kinda still think that's a reasonable model - I don't want to have x-versions-location and x-version-history-container both written down because of some eventual consistency bullshit | 19:01 |
| *** Fin1te has joined #openstack-swift | 19:03 | |
| *** pcaruana has joined #openstack-swift | 19:04 | |
| tdasilva | clayg: not sure about this release, but i'd almost prefer we take a step back and think the design a bit more. For example, another possibility would be to start from scratch with this a new middleware that uses symlinks | 19:06 |
| *** tongli has joined #openstack-swift | 19:06 | |
| *** silor has quit IRC | 19:11 | |
| *** asettle has joined #openstack-swift | 19:14 | |
| clayg | acoles_: so this slo change is surprinsgly small for all the test doc changes you did to make it look big and scary | 19:19 |
| *** asettle has quit IRC | 19:26 | |
| *** asettle has joined #openstack-swift | 19:28 | |
| *** asettle has quit IRC | 19:28 | |
| openstackgerrit | Merged openstack/swift: Include correct version in install-guide https://review.openstack.org/363938 | 19:28 |
| *** asettle has joined #openstack-swift | 19:28 | |
| *** CaioBrentano has joined #openstack-swift | 19:29 | |
| *** hseipp has joined #openstack-swift | 19:30 | |
| CaioBrentano | Hi all!! I set weight zero to a device, but it still has some partitions. How is that possible? | 19:32 |
| glange | partitions on disk? | 19:32 |
| clayg | on the disk or assigned in the ring? | 19:32 |
| *** asettle has quit IRC | 19:33 | |
| clayg | i guess I could imagine duplicate part assignment on a disk leading to only picking up one part-replica off the disk in the first pass | 19:33 |
| *** asettle has joined #openstack-swift | 19:33 | |
| jrichli | asettle: howdy. we are preparing for our Newton release. patch 354767 makes some big improvements on the api-ref. Are those docs built at every commit like the ones in doc/ ? | 19:33 |
| patchbot | https://review.openstack.org/#/c/354767/ - swift - Corrections for the API specifications in api-ref | 19:33 |
| clayg | failing the disk would be an option - but if you drain it - probably just doing a second cycle rebalance would be great | 19:33 |
| clayg | CaioBrentano: ^^ | 19:34 |
| CaioBrentano | clayg: both! the partitions are assigned in the ring... and the disk still has objects | 19:36 |
| *** Fin1te has quit IRC | 19:36 | |
| *** StraubTW has joined #openstack-swift | 19:38 | |
| CaioBrentano | clayg: I set weight zero because I want to remove these disks... | 19:38 |
| *** asettle has quit IRC | 19:38 | |
| clayg | notmyname: I sorta think that lp bug #1580678 was fixed by cschwede somewhere along the way? | 19:38 |
| openstack | Launchpad bug 1580678 in OpenStack Object Storage (swift) "UnicodeDecodeError when rebalancing a ring" [High,In progress] https://launchpad.net/bugs/1580678 - Assigned to Christian Schwede (cschwede) | 19:38 |
| clayg | CaioBrentano: so pushing out the rings with the parts unassigned will let the devices drain what they can | 19:39 |
| *** Guest22444 has joined #openstack-swift | 19:39 | |
| CaioBrentano | clayg: nice! thanks! | 19:39 |
| clayg | CaioBrentano: after a replication cycle the devices on the disk should match what's assigned in the ring (which I could imagine might be non-zero iff those parts were already reassigned this min_part_hours) | 19:39 |
| clayg | then for the few remaining parts - you can either wait min_part_hours and rebalance and push again with 0 parts assigned | 19:40 |
| clayg | or you could fail the devices - the parts will immediately be reassigned and back filled from the other replicas - but the deleted device will not drain it's parts from the disk - the replicator on that node will simply ignore the deleted device and the parts that are on it | 19:41 |
| CaioBrentano | clayg: how could this "duplicate part assignment" happen? | 19:42 |
| Guest22444 | exit | 19:44 |
| *** Guest22444 has quit IRC | 19:44 | |
| clayg | CaioBrentano: old builer codef used to be stupid -> lp bug #1452431 | 19:45 |
| openstack | Launchpad bug 1452431 in OpenStack Object Storage (swift) "some parts replicas assigned to duplicate devices in the ring" [High,Fix released] https://launchpad.net/bugs/1452431 - Assigned to Samuel Merritt (torgomatic) | 19:45 |
| clayg | CaioBrentano: although, it may not have been that bug specifically depending on your cluster topology - I think the same issue would come up if you had two part-replicas assigned to the same server/zone and you set both of those devices to zero weight in the same pass | 19:45 |
| clayg | which really - is the desired behavior | 19:46 |
| notmyname | clayg: maybe cschwede fixed that bug at some point. would be good to get confirmation. but you could be thinking of https://bugs.launchpad.net/swift/+bug/1597210 (linked from the bottom of the first one) | 19:46 |
| openstack | Launchpad bug 1597210 in OpenStack Object Storage (swift) "Downed object-server during a PUT with unicode characters causes UnicodeDecodeError and 500 status code" [Critical,Fix released] - Assigned to Brian Cline (briancline) | 19:46 |
| CaioBrentano | clayg: thanks! | 19:49 |
| hogepodge | notmyname: I'm going to ping you on this again, just because I got pinged on it again. https://review.openstack.org/#/c/324739/5 | 19:49 |
| patchbot | patch 324739 - python-swiftclient - Insufficent info about read and write ACL syntax | 19:49 |
| *** hseipp has quit IRC | 19:50 | |
| hogepodge | notmyname: A decision on how to fix the inline docs would be nice. I'm not too invested in how the decision lands, just that there is one. | 19:50 |
| notmyname | hogepodge: I'll need to refamiliarize myself with the question :-) | 19:50 |
| notmyname | hogepodge: ah yes. you have a test or something that is based on docs, and the docs aren't actually how it works | 19:51 |
| hogepodge | notmyname: tempauth and keystone auth have different ACL syntax. python-swiftclient assumes tempauth | 19:51 |
| hogepodge | notmyname: yeah, if there's documentation somewhere online that points to how keystone auth does it, that would work for me too. We could link to it and call the job done | 19:52 |
| notmyname | hogepodge: rereading the comments, acoles_ seems to have the right idea--as normal ;-) | 19:53 |
| notmyname | hogepodge: how would you feel about the references to v1 auth api vs non-v1 api? | 19:53 |
| clayg | notmyname: I can't place it - I tought Al fixed something too - there's this one -> https://review.openstack.org/#/c/339360/1 | 19:53 |
| patchbot | patch 339360 - swift - Make verbose unicode test output working | 19:53 |
| notmyname | mmotiani: ping. we're talking about your patch 324739 | 19:53 |
| patchbot | https://review.openstack.org/#/c/324739/ - python-swiftclient - Insufficent info about read and write ACL syntax | 19:53 |
| notmyname | clayg: the common theme here is that you keep linking stuff you've put a -1 on ;-) | 19:54 |
| clayg | hrm... that doesn't sound so helpful :\ | 19:54 |
| *** zul has joined #openstack-swift | 19:54 | |
| *** Guess456787654 has joined #openstack-swift | 19:55 | |
| clayg | notmyname: no look! i'm helping -> https://review.openstack.org/#/c/338950/ | 19:55 |
| patchbot | patch 338950 - swift - Fix unicode errors in object controller logging (MERGED) | 19:55 |
| notmyname | lol | 19:55 |
| hogepodge | notmyname: I'm not sure it would clear things up for our target audience. the first thing I'd ask myself was "am I running v1 or non v1?" and then shrug my shoulders | 19:55 |
| notmyname | hogepodge: yeah, but you've got that issue anyway with v2 vs v3 | 19:56 |
| MooingLemur | if a DELETE on an object only succeeds on one replica, does the client get a 204? | 19:56 |
| notmyname | hogepodge: and I strongly suspect most people shrug and walk away when confronted with that | 19:56 |
| clayg | only *one* - might depend on the result of the other backend requests | 19:56 |
| notmyname | MooingLemur: no. same quorum rules apply to DELETEs as PUTs | 19:56 |
| hogepodge | notmyname: if I hear | 19:56 |
| clayg | notmyname: MooingLemur: wheeeeelll - except that a backend 404 on DELETE can count towards the q of 204's because the result is the same | 19:57 |
| clayg | or something | 19:57 |
| clayg | mattoliverau: wrote it | 19:57 |
| hogepodge | notmyname: keystone vs tempauth, I have a much better chance of saying "It's a keystone system! I know this!" | 19:57 |
| clayg | I also think mattoliverau was working on the unicode business | 19:57 |
| MooingLemur | notmyname: so if it's 404, 204, 404, (404s on handoffs) client'll get what? | 19:57 |
| notmyname | clayg: ah yes. 404 counts as a 2xx for DELETEs | 19:57 |
| * hogepodge makes oblique jurrasic park reference... | 19:57 | |
| MooingLemur | aha, alrighty | 19:57 |
| MooingLemur | but if all backends return 404, client gets 404 | 19:58 |
| notmyname | hogepodge: oooo. can we get 3D models of the cloud to fly over? | 19:58 |
| clayg | yeah, maybe a single 204 can make the client see success... that probably makes sense | 19:58 |
| hogepodge | notmyname: horizon doesn't provide that? | 19:59 |
| notmyname | hogepodge: definitely a missing feature. you can add that to defcore, right? ;-) | 19:59 |
| *** daemontool has quit IRC | 19:59 | |
| *** tongli has quit IRC | 20:01 | |
| *** MVenesio has joined #openstack-swift | 20:04 | |
| *** tongli has joined #openstack-swift | 20:07 | |
| *** tongli has quit IRC | 20:12 | |
| *** tongli has joined #openstack-swift | 20:13 | |
| torgomatic | clayg: can you take a look at patch 344898 when you get a second? I want to make sure I'm saying reasonable things in the review comments. | 20:15 |
| patchbot | https://review.openstack.org/#/c/344898/ - swift - add dev_losing_part['parts_wanted'] += 1 | 20:15 |
| 17SAA00LA | notmyname: I am still trying to figure that out. Currently, I am looking at the integration of swift with keystoneauth to see how it works | 20:16 |
| clayg | torgomatic: if you're -1 I probably agree ;) | 20:16 |
| notmyname | 17SAA00LA: mmotiani? | 20:16 |
| 17SAA00LA | Yup. | 20:16 |
| 17SAA00LA | Oh wait | 20:16 |
| 17SAA00LA | whats that | 20:17 |
| notmyname | 17SAA00LA: /nick might help :-) | 20:17 |
| 17SAA00LA | I don't understand why is it some random nick. Sorry about that | 20:19 |
| notmyname | 17SAA00LA: what part are you still trying to figure out wrt the ACL syntax docs? | 20:19 |
| clayg | torgomatic: so what you're saying there makes sense to me - and I don't understand what exactly Cheng Li is after with his change (beyond what it claims - which is "this changed the code to do the thing I said the code should say" | 20:20 |
| clayg | torgomatic: but looking at the linked lp bug #1605841 - it *seems* like he's trying to address some issue? | 20:20 |
| openstack | Launchpad bug 1605841 in OpenStack Object Storage (swift) "add dev_losing_part['parts_wanted'] += 1" [Undecided,In progress] https://launchpad.net/bugs/1605841 - Assigned to Cheng Li (shcli) | 20:20 |
| clayg | Reassigned 128 (50.00%) partitions vs Reassigned 9 (3.52%) partitions after 3.5 -> 3.0 | 20:21 |
| torgomatic | clayg: yeah, there's something going on in there, but I'm not entirely sure what it is | 20:21 |
| 17SAA00LA | notmyname: to check acl syntax doc, I was trying to setup swift with keystoneauth instead of tempauth. I am currently stuck at the client side authentication with keystoneauth | 20:21 |
| clayg | he also claims that balance gets better | 20:21 |
| torgomatic | clayg: yeah, so maybe the trick is to call set_parts_wanted *after* fiddling with the replica counts? | 20:23 |
| *** thumpba has quit IRC | 20:27 | |
| jrichli | timburke: thanks for the review on memcache reload! | 20:27 |
| *** d0ugal has quit IRC | 20:28 | |
| timburke | jrichli: sure! the exception handling is the big thing for me; i don't see any reason to fail the request when trying to reload a bad config | 20:28 |
| *** thumpba has joined #openstack-swift | 20:29 | |
| jrichli | timburke: ok, that sounds like something i can be on board with. i hope to get to this soon, but it seems my list has just grown over the last few days. | 20:30 |
| clayg | torgomatic: lol - i just said as much on the review | 20:31 |
| clayg | torgomatic: I think we also need to true up our goals and problem statement on the bug report | 20:31 |
| clayg | torgomatic: thanks for reminding me about that | 20:31 |
| clayg | ... that patch | 20:31 |
| torgomatic | clayg: phew! that means that it's not completely nuts ;) | 20:31 |
| clayg | torgomatic: no sir you are not! | 20:31 |
| clayg | torgomatic: do you have any good jokes about working on the ring and riding a bike? | 20:32 |
| torgomatic | clayg: that's a very specific thing to joke about; nothing comes to mind. I'll let you know. :) | 20:36 |
| clayg | torgomatic: I don't get it | 20:37 |
| clayg | torgomatic: how about working on the ring is like riding a bike - it's hard to balance - and hurts if you're not careful? | 20:38 |
| * torgomatic will take it | 20:38 | |
| *** tongli has quit IRC | 20:39 | |
| timburke | something something velodrome? | 20:39 |
| clayg | timburke: I was trying to get at despire torgomatic | 20:40 |
| clayg | ... torgomatic's recent efforts he has *not* forgotten how the ring works ;) | 20:41 |
| clayg | oh, *despite even | 20:41 |
| *** MVenesio has quit IRC | 20:43 | |
| *** d0ugal has joined #openstack-swift | 20:44 | |
| *** pcaruana has quit IRC | 20:46 | |
| openstackgerrit | Merged openstack/swift: Apply go fmt to tempauth https://review.openstack.org/367685 | 20:46 |
| clayg | timburke: do all the filesystem api's in py3 return unicode? e.g. os.listdir? | 20:47 |
| timburke | iirc, yes | 20:47 |
| clayg | do they have any opinion about byte strings from the disk that it can't seem to convert to unicode based on the default locale? I think there's like a "substitute" mechanic? | 20:48 |
| timburke | yup; look for "surrogateescape" | 20:48 |
| timburke | https://www.python.org/dev/peps/pep-0383/ | 20:48 |
| *** mvk has joined #openstack-swift | 21:07 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Store SLO Etag and size in sysmeta https://review.openstack.org/347538 | 21:24 |
| *** Guess456787654 has quit IRC | 21:34 | |
| *** vint_bra has quit IRC | 21:34 | |
| clayg | i love computers *so* much -> "Emit a malformed UTF-16 sequence for every byte in a malformed UTF-8 sequence" -- the clear RightThingToDo[TM] | 21:37 |
| *** thumpba has quit IRC | 21:49 | |
| *** CaioBrentano has quit IRC | 21:56 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Bring "egregious range request" comment in line with reality https://review.openstack.org/369060 | 22:06 |
| *** zigo has quit IRC | 22:08 | |
| *** zigo has joined #openstack-swift | 22:11 | |
| *** zigo is now known as Guest41388 | 22:11 | |
| pdardeau | clayg: that's where your image is appropriate - http://66.media.tumblr.com/tumblr_m9yx3gTr9d1rzupqxo1_500.png | 22:13 |
| *** cebruns__ has quit IRC | 22:13 | |
| *** cebruns has joined #openstack-swift | 22:15 | |
| *** Guest41388 is now known as zigo_ | 22:17 | |
| *** cebruns has quit IRC | 22:19 | |
| *** thumpba has joined #openstack-swift | 22:19 | |
| *** cebruns has joined #openstack-swift | 22:21 | |
| *** StraubTW has quit IRC | 22:21 | |
| clayg | pdardeau: not at all - I think you made a genuine contribution to that discussion on that patch and elicited meaningful discourse - nice work! | 22:22 |
| *** cebruns has quit IRC | 22:22 | |
| pdardeau | clayg: thx, but i was referring to your comment about the malformed utf sequence | 22:23 |
| clayg | oh - rofl | 22:24 |
| *** thumpba has quit IRC | 22:24 | |
| clayg | yeah, idk, there was a somewhat convincing argument in that spec for the byte smuggling surrogateescape strategy! | 22:24 |
| clayg | gah "spec" :'( | 22:25 |
| clayg | pep | 22:25 |
| *** cebruns has joined #openstack-swift | 22:29 | |
| *** vint_bra has joined #openstack-swift | 22:34 | |
| timburke | huh. anyone else see this? POST response under fast-post includes an HTML body, but with post-as-copy, there's no body | 22:40 |
| *** cebruns has quit IRC | 22:41 | |
| *** cebruns has joined #openstack-swift | 22:42 | |
| mattoliverau | morning | 22:43 |
| clayg | timburke: who runs past-as-copy!? | 22:45 |
| timburke | clayg: my old, janky saio? customers that haven't been watching the release notes with bated breath? | 22:47 |
| clayg | timburke: anyway - confirmed | 22:47 |
| clayg | do you think it's a regression or just an old difference that no one had previously thought to care about? | 22:48 |
| clayg | also it's weird | 22:48 |
| timburke | no idea. do we have a preference for which way it should behave? | 22:48 |
| clayg | i get 202 accepted in both cases | 22:48 |
| timburke | yup, me too | 22:48 |
| *** openstackgerrit has quit IRC | 22:48 | |
| clayg | i'm guessing best_response cooks up the 202 from a collection of the backends on fast-post and probably post-as-copy does something to translate the 201 into a 202 and has always done the wrong thing? | 22:48 |
| *** openstackgerrit has joined #openstack-swift | 22:49 | |
| timburke | i think it might be vaguely related to the swob changes torgomatic has in https://review.openstack.org/#/c/352657/ ? | 22:50 |
| patchbot | patch 352657 - swift - Support multi-range GETs for static large objects. | 22:50 |
| clayg | timburke: but i'm seeing the difference in behavior on an unrelated change - and that isn't merged? does that change *fix* it!? | 22:51 |
| clayg | i'm not sure where the copy middleware is translating the 201 from the put response into a 202 on post-as-copy? | 22:52 |
| timburke | it fixes a similar problem that was happening during 416 responses for SLOs | 22:52 |
| clayg | oh, _adjust_put_response | 22:52 |
| timburke | yeah, was just about to send a link to https://github.com/openstack/swift/blob/2.9.0/swift/common/middleware/copy.py#L249 | 22:53 |
| torgomatic | that patch shouldn't affect this; the swob changes there are guarded by "if ranges == []" in range handling; POST shouldn't get anywhere near that code | 22:53 |
| clayg | so it might even be *weirder* if 201 returned a body? | 22:53 |
| timburke | torgomatic: i said *vaguely* related :P | 22:54 |
| timburke | clayg: almost certainly | 22:54 |
| clayg | timburke: ok, so when you file the bug - it'd be great if you could see if the difference existed before the copy middleware extraction | 22:55 |
| timburke | i wonder if we maybe *used to* include a body on PUT? looking at http://developer.openstack.org/api-ref/object-storage/index.html?expanded=create-or-replace-object-detail the example response has `Content-Length: 116` (but no actual body) | 22:56 |
| timburke | also, we somehow managed to include an example that 408s :-/ | 22:57 |
| clayg | timburke: idk :\ | 22:57 |
| *** vint_bra has quit IRC | 22:59 | |
| timburke | ugh, and we claim that Date and Last-Modified will look like CCYY-MM-DDThh:mm:ss±hh:mm :-( | 23:06 |
| timburke | hurray for donagh! https://review.openstack.org/#/c/354767/12/api-ref/source/parameters.yaml | 23:07 |
| patchbot | patch 354767 - swift - Corrections for the API specifications in api-ref | 23:07 |
| clayg | hooray for donagh! | 23:08 |
| *** _JZ_ has quit IRC | 23:09 | |
| *** klamath_ has quit IRC | 23:13 | |
| openstackgerrit | Tim Burke proposed openstack/swift: Include metadata in PUT/POST responses https://review.openstack.org/357559 | 23:14 |
| *** cebruns has quit IRC | 23:16 | |
| *** cebruns has joined #openstack-swift | 23:19 | |
| *** Guess456787654 has joined #openstack-swift | 23:22 | |
| *** cebruns has quit IRC | 23:27 | |
| *** zul has quit IRC | 23:28 | |
| *** cebruns has joined #openstack-swift | 23:30 | |
| *** kei_yama has joined #openstack-swift | 23:30 | |
| *** tsg has quit IRC | 23:38 | |
| *** Jeffrey4l has quit IRC | 23:43 | |
| *** Jeffrey4l has joined #openstack-swift | 23:43 | |
| *** Guess456787654 has quit IRC | 23:46 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
