| openstackgerrit | Merged openstack/swift: Make ECDiskFile require a fragment index https://review.openstack.org/168076 | 00:12 |
|---|---|---|
| *** ho has joined #openstack-swift | 00:12 | |
| ho | good morning guys! | 00:20 |
| notmyname | hello | 00:22 |
| *** dmorita has joined #openstack-swift | 00:24 | |
| *** annegentle has quit IRC | 00:26 | |
| mattoliverau | ho: Morning | 00:28 |
| *** annegentle has joined #openstack-swift | 00:29 | |
| ho | notmyname: mattoliverau: hello, thanks! | 00:34 |
| ho | Do we have priority review in this week? | 00:35 |
| mattoliverau | clayg: are you doing something on patch 168543, otherwise I'm about to push a patchset to fix the pep8 issue | 00:36 |
| patchbot | mattoliverau: https://review.openstack.org/#/c/168543/ | 00:36 |
| *** haomaiwang has quit IRC | 00:36 | |
| mattoliverau | ho: there are always priority reviews :P.. but in regards to EC, yeah still some more to be merged, so still more need to be reviewed. Getting closer tho (famous last words) | 00:37 |
| ho | mattoliverau: thanks! yeah, we always have prioirty reviews :) Can I have patch numbers of priority reviews? | 00:40 |
| mattoliverau | ho: they are always on the dash board (in the channel title) and on our priority reviews wiki page. | 00:41 |
| notmyname | ho: check out the starred things on http://goo.gl/uRzLBX | 00:42 |
| *** Gu_______ has joined #openstack-swift | 00:42 | |
| notmyname | ho: basically for the next ~24 hours (or less), we need to get the final things for the ec beta into the feature/ec branch so that we can begin the merge to master | 00:42 |
| ho | mattoliverau: notmyname: Do we have any blocking patch to proceed it? | 00:43 |
| notmyname | ho: what's "it"? | 00:44 |
| *** kota_ has joined #openstack-swift | 00:44 | |
| notmyname | ho: so the 2 starred patches there don't have any blockers | 00:44 |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Fix AttributeError and logging in ECAppIter https://review.openstack.org/168543 | 00:45 |
| ho | notmyname: from this week we will start merging right? in order to do it, I would like to know | 00:45 |
| notmyname | ho: right | 00:45 |
| ho | notmyname: whether there is *really* priority reviews | 00:45 |
| notmyname | ho: there's nothing blocking the merge except having the right code *to* merge. ie the feature/ec branch | 00:45 |
| mattoliverau | clayg: ^^ pep8 issue fixed | 00:45 |
| notmyname | ho: you mean outside of feature/ec? | 00:45 |
| ho | notmyname: I understand our current focus is ec. | 00:46 |
| notmyname | ho: so for other stuff outside of feature/ec, there's isn't anything on gerrit that must land before feature/ec. of course, there's some good stuff there, but the focus is ec | 00:47 |
| ho | notmyname: i see. On dashboard, there are ec related patches. I would like to spend my time for high priority patches is what I wanted to say/do today. | 00:49 |
| notmyname | ok :-) | 00:49 |
| ho | notmyname: which is high priority "Starred" or "Erasure Code work" on dashboard? | 00:52 |
| notmyname | ho: "starred" | 00:52 |
| ho | notmyname: thanks! I will try. | 00:52 |
| notmyname | ho: the most important thing on ec right now is getting https://review.openstack.org/#/c/165188/ (the reconstructor patch chain) beat into shape and landed. that's the last crucial piece of missing functionality on ec | 00:53 |
| notmyname | well, that's the first of 3 patches in a dependency chain | 00:53 |
| ho | notmyname: thanks. I understand. | 00:55 |
| kota_ | notmyname: for now, someone else is working on chainging those? | 00:55 |
| kota_ | notmyname: I am worried I could review stale patches on the chain. | 00:55 |
| kota_ | nomyname: anways, good morning : ) | 00:56 |
| mattoliverau | kota_: morning | 00:56 |
| notmyname | kota_: good question, and good morning | 00:56 |
| kota_ | mattoliverau: morning : ) | 00:56 |
| mattoliverau | kota_: even stale reviews are good as you may see something that others miss :) | 00:56 |
| notmyname | I'd defer to clayg on that, but AFAIK the major structure of the patches are good. ie if there are additional changes, it's to fix stuff that's found in review | 00:57 |
| kota_ | mattoliverau, notmyname: OK, I am going to start to review, today ;-) | 00:59 |
| mattoliverau | kota_: cheers, more eyes on the code the better :) | 00:59 |
| mattoliverau | ho: thanks to you as well :) | 01:00 |
| *** shakamunyi has joined #openstack-swift | 01:07 | |
| *** shakamunyi_ has joined #openstack-swift | 01:07 | |
| *** Gu_______ has quit IRC | 01:09 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix AttributeError and logging in ECAppIter https://review.openstack.org/168543 | 01:13 |
| clayg | mattoliverau: it wasn't even pep8 - it was a whole big mess | 01:14 |
| clayg | oh - you had already done the import - yeah that was step 0 i guess :P | 01:14 |
| mattoliverau | lol | 01:14 |
| clayg | mattoliverau: you eve seen this one -> https://www.youtube.com/watch?v=SY99U3cpsIw | 01:16 |
| *** annegentle has quit IRC | 01:19 | |
| clayg | ... and before you know it you have the engine pulled out of the car | 01:19 |
| mattoliverau | clayg: haha, awesome.. ahhh, so step 0 :P | 01:20 |
| clayg | oh shit, the stupid doc comment code :\ | 01:20 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix AttributeError and logging in ECAppIter https://review.openstack.org/168543 | 01:21 |
| clayg | ^ yeah that's better! | 01:24 |
| clayg | ok, dinner time | 01:24 |
| openstackgerrit | Merged openstack/swift: Guard against invalid Policy https://review.openstack.org/168548 | 01:24 |
| clayg | bbiab | 01:24 |
| clayg | oh yeah heh i remember when jd clicked approve on that one... HOURS AGO :P | 01:25 |
| mattoliverau | ha, well I'm going to take a lunch break. bbs | 01:28 |
| *** Guest____ has joined #openstack-swift | 01:55 | |
| openstackgerrit | Daisuke Morita proposed openstack/swift-specs: Changing Policies spec https://review.openstack.org/168761 | 02:09 |
| *** BAKfr has quit IRC | 02:11 | |
| *** haomaiwang has joined #openstack-swift | 02:13 | |
| *** BAKfr has joined #openstack-swift | 02:13 | |
| *** annegentle has joined #openstack-swift | 02:45 | |
| *** annegentle has quit IRC | 02:51 | |
| *** shakamunyi__ has joined #openstack-swift | 02:51 | |
| *** david-ly_ is now known as david-lyle | 02:52 | |
| *** shakamunyi has quit IRC | 02:54 | |
| *** shakamunyi_ has quit IRC | 02:54 | |
| *** shakamunyi has joined #openstack-swift | 02:54 | |
| *** branne has joined #openstack-swift | 03:00 | |
| *** haigang has joined #openstack-swift | 03:04 | |
| *** jrichli has quit IRC | 03:07 | |
| *** branne has quit IRC | 03:17 | |
| clayg | i thought patch 168543 must have not been failing on my machine because i have recon configured and the test doesn't mock it out | 03:24 |
| patchbot | clayg: https://review.openstack.org/#/c/168543/ | 03:24 |
| clayg | ... but it looks like it is passsing in a mkdtemp as the recon_cache_path - so I don't know what exception is being caught in jenkins | 03:25 |
| clayg | hrmm... acctually I had an idea that might help - man I'm really reworking the fake logger here :\ | 03:25 |
| *** sandywalsh has quit IRC | 03:28 | |
| notmyname | I'm happy about s/RUGGEDIZE/SSYNC | 03:28 |
| *** ppai has joined #openstack-swift | 03:30 | |
| clayg | notmyname: on master it'll just be REPLICATION -> SSYNC | 03:30 |
| clayg | got a great time for any other breaking protocol changes | 03:30 |
| mattoliverau | clayg: on test_failed_delete_continue_on the lists it asserts are not the same, the error_lines has an additional line (exception): 'Exception dumping recon cache' | 03:35 |
| *** sandywalsh has joined #openstack-swift | 03:37 | |
| mattoliverau | now to debug to see where that is coming from | 03:37 |
| clayg | mattoliverau: yeah i was thinking it was just working for me because I have recon setup - but acctually the test seems to be setting up a mkdtemp rcache path - so I don't know what's going on. | 03:40 |
| mattoliverau | I yeah, now I see you previous msg.. /me fails to read scroll back :P | 03:40 |
| clayg | bah, i was just thinking that it'd be nice if the FakeLogger handler did something to include soe of the exception in the log line - there's a few tests that still dig into log_dict because it's the easiest way to get at the exception from exc_info | 03:44 |
| clayg | i was playing around with adding like .... '\n%s: %s' % (exc_info[0], exc_info[1]) ... or someting minimal like that to the log lines that include exc_info | 03:44 |
| clayg | but... i'm not really sure I like how it's turning out | 03:45 |
| clayg | i really just wish I knew what that gate check failed to dump the recon cache | 03:45 |
| clayg | i was about to just mock.patch('.dump_recon_cache') before calling run_once - but i feel like tht's sorta dodging the issue | 03:45 |
| mattoliverau | clayg: it seems to be dying when trying to lock_file on '/var/cache/swift/object.recon' I assume it doesn't exist | 03:50 |
| mattoliverau | at least on my machine | 03:50 |
| mattoliverau | a mock may be in order | 03:50 |
| clayg | why is it trying to ... oh | 03:52 |
| clayg | i | 03:52 |
| clayg | no... | 03:52 |
| clayg | isn't it creating the expirer with the recon_cache_path of the mkdtemp? | 03:52 |
| clayg | yeah lockfile seems to think it can create the file | 03:57 |
| mattoliverau | lol, if I: x = expirer.ObjectExpirer({'recon_cache_path': self.rcache}, | 03:58 |
| mattoliverau | logger=self.logger) | 03:58 |
| mattoliverau | then the test passes.. turns out we need to pass in the tmp recon cache | 03:58 |
| mattoliverau | clayg: ^ | 03:58 |
| notmyname | mattoliverau: what's different about your saio that makes it fail? works on my machine too | 03:59 |
| notmyname | you don't have recon set up? | 03:59 |
| mattoliverau | not, I'm just running the unit tests on the code (as jenkins does) | 04:00 |
| mattoliverau | wow, my typing is awesome today | 04:00 |
| notmyname | well, me too | 04:00 |
| notmyname | but I have 'sudo mkdir -p /var/cache/swift \n sudo chown -R swift:swift /var/cache/swift' in my resetswift script | 04:01 |
| mattoliverau | so I'm not running it on a saio, so the cache doesn't exist.. unless I pass the tmpdir created | 04:01 |
| clayg | mattoliverau: no i thought the test already did that? | 04:01 |
| clayg | like which line are you doing that in? | 04:01 |
| mattoliverau | it doens't, I had to add it | 04:01 |
| notmyname | line 568, rihgt? | 04:01 |
| mattoliverau | 568 | 04:01 |
| mattoliverau | yup | 04:01 |
| mattoliverau | the test passes in an empty config, so I can define the 'recon_cache_path' and use the tmp file the test has created (but isn't passing) | 04:02 |
| notmyname | mattoliverau: and 424 and 373 | 04:02 |
| notmyname | (the other tests that are failing | 04:03 |
| mattoliverau | I'll try, I was just debugging one at a time.. give me a sec | 04:03 |
| clayg | oh wow - test_container_timestamp_break setts it the self.rcache one time it creates it and the next time it does not | 04:04 |
| clayg | mattoliverau: no it's cool - you found it - nice work - i'll audito all the instantiation and make sure they're all using the tempdir | 04:04 |
| *** thumpba has joined #openstack-swift | 04:05 | |
| notmyname | (there are 26 matches for "ObjectExpirer({}" in that file) | 04:05 |
| clayg | notmyname: yeah I think some of them are just testing the process sharding config option stuff - they're easier to read left alone I think | 04:07 |
| clayg | notmyname: I think it's only the ones that call run_once | 04:07 |
| notmyname | clayg: see the Hal youtube video you left in here earlier.... ;-) | 04:07 |
| mattoliverau | ok in test_object_timestamp_break, it still fails. but it's becuase the output lines don't match. | 04:08 |
| mattoliverau | particularly: 'Pass completed in 10s; 0 objects expired' is generated, but is expecting 'Pass completed in 0s; 0 objects expired' | 04:08 |
| mattoliverau | is that just a typo? 10 secs rather then 0 secs | 04:09 |
| mattoliverau | I can debug to find out.. but in case you know | 04:09 |
| notmyname | I need to go put kids to bed | 04:09 |
| notmyname | looks like there are 4 patches outstanding: 3 for the reconstructor chain and the one you're just looking at | 04:10 |
| notmyname | which seems doable, especially as EU comes online | 04:11 |
| notmyname | I'll get the feature/ec_review branch set up in the morning | 04:11 |
| notmyname | (my morning) | 04:12 |
| notmyname | all of those are starred | 04:13 |
| notmyname | ok, gotta run. good night | 04:14 |
| mattoliverau | So test_failed_delete_keeps_entry works with the tmpdir fix if you set the option when the expirer is created _both_ times. | 04:14 |
| mattoliverau | notmyname: night man | 04:14 |
| mattoliverau | So I'll go back to timestamp_break and debug some more | 04:15 |
| mattoliverau | opps, yup timestamp_breaks is fixed by the config too.. just make sure you make the change in _both_ places in the test too :P | 04:16 |
| clayg | hehehe | 04:17 |
| clayg | yeah i got my macro going now - wasted a few mintutes trying to figure out why it wasn't working because I didn't save the buffer - but I think i'm getting close now | 04:18 |
| clayg | mattoliverau: thanks for the help! | 04:18 |
| mattoliverau | clayg: nps, anytime :) | 04:18 |
| clayg | acctually that may be it | 04:23 |
| clayg | ok I deleted /var/cache/swift - i'm running all the unittests now | 04:23 |
| mattoliverau | nice | 04:23 |
| mattoliverau | I'm happy to run it for you too.. just running it with those 3 tests changed atm | 04:24 |
| clayg | a'ight gerrit try this! | 04:25 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix AttributeError and logging in ECAppIter https://review.openstack.org/168543 | 04:25 |
| mattoliverau | yay passed on mine now | 04:28 |
| *** zaitcev has quit IRC | 04:32 | |
| *** zaitcev has joined #openstack-swift | 04:32 | |
| *** ChanServ sets mode: +v zaitcev | 04:32 | |
| zaitcev | Weird, Xchat lost its marbles and started hogging CPU for no reason. Could be an attack of some kind. | 04:33 |
| kota_ | mattoliverau: congrats | 04:35 |
| mattoliverau | zaitcev: xchat probably just wanted some attention :P | 04:37 |
| mattoliverau | or the internet thinks your too productive that you must be a botnet | 04:38 |
| *** shakamunyi_ has joined #openstack-swift | 04:39 | |
| zaitcev | I'm trying to package Tushar's liberasurecode, and it fails the build of a test binary with rpmbuild. Works fine with plain "make". | 04:40 |
| mattoliverau | clayg: and the new patchet passes on my machine.. plus the test_verioned_manifest tests that were taking up to 20+ seconds each and now < 1/2 as long.. what did you do? | 04:40 |
| zaitcev | For some reason we already added PyECLib to master's requirements.txt, so I cannot build Swift unless I package the whole enchilada, stat. | 04:40 |
| mattoliverau | zaitcev: bugger :( | 04:41 |
| *** shakamunyi__ has quit IRC | 04:41 | |
| *** shakamunyi has quit IRC | 04:41 | |
| clayg | mattoliverau: wat? no those still take forever on me | 04:41 |
| *** shakamunyi has joined #openstack-swift | 04:41 | |
| clayg | zaitcev: yeah i don't know why that got accepted into master already unless infra needed it there to add to global requirements or something? | 04:42 |
| mattoliverau | clayg: maybe my machine just evolved (or found some more CPU) :P | 04:42 |
| clayg | mattoliverau: must be! | 04:42 |
| ho | clayg: mattoliverau: You fixed the following three FAILs in test_expirer (patch 168543)? http://paste.openstack.org/show/197386/ | 04:42 |
| patchbot | ho: https://review.openstack.org/#/c/168543/ | 04:42 |
| mattoliverau | ho: yup :) I and passes | 04:43 |
| mattoliverau | gerrit has passed py27 check, now just waiting on all the tempest stuff | 04:43 |
| ho | mattoliverau: thanks! I will rebase it for my review. | 04:44 |
| mattoliverau | ho: awesome :) | 04:45 |
| *** welldannit has joined #openstack-swift | 04:46 | |
| clayg | ho: mattoliverau says their fixed | 04:47 |
| *** ppai_ has joined #openstack-swift | 04:47 | |
| clayg | ho: i'm looking at your comment son patch 165188 - not sure I prefer the alrger try block to catch the assertion error all in one go | 04:47 |
| patchbot | clayg: https://review.openstack.org/#/c/165188/ | 04:47 |
| mattoliverau | well my tox and gerrit thinks so :) | 04:47 |
| clayg | ho: using the asssertion error in the debug message is ok maybe - but if it's going to be that wide might have to add exc_info and nab the traceback so you can tell who exactly raised the assertion error - which seems like a bunch of noise for a debug message | 04:48 |
| *** ppai has quit IRC | 04:48 | |
| *** zaitcev has quit IRC | 04:49 | |
| clayg | ho: I think we sort of assume skipping these invalid hash paths is just part of normal operation - a better place for the traceback would be in the auditor something - maybe quarantine them? I'm not quite sure | 04:49 |
| ho | clayg: I'm checking my comment and code. Please give a minute | 04:51 |
| clayg | ho: as far as moving the test methods - yield_hashes is a Manager method - so I think for an organization stand point it's quite reasonable to move them from TestDiskFile to TestDiskFileManager | 04:51 |
| ho | clayg: thanks for the explanation of the moving. I understand. | 04:52 |
| clayg | ho: well i just applied your diff - and I'm looking at the big try block catching something fairly generic like an AssertionError - it's probably fine i guess -> https://gist.github.com/clayg/b6ae381139621d135d5b | 04:53 |
| clayg | ho: it *is* a little less code - and I acctually screwed up the exception Assertion: log; conitinue the first time I wrote it (i forgot the continue) | 04:54 |
| clayg | so maybe it's better to put everything in one - trying to think if there's anything better we could do - maybe log.debug(exc_info=1) or something? | 04:54 |
| clayg | ho: hrm... yeah that assertionError message is acctually pretty good | 04:58 |
| ho | clayg: sorry to respond you late. it's touch for me to follow your speed :) | 04:59 |
| clayg | ho: not it's fine - i'm just thinking out loud - I think you're onto something | 05:02 |
| clayg | yeah ok, i'm definately +2 on including the assertion error output | 05:03 |
| ho | clayg: I can not find the info of log.debug(exe_info=1). could you point it out? | 05:03 |
| *** shakamunyi_ has quit IRC | 05:04 | |
| clayg | you can make any log message level include the traceback like logger.exception by passing in exc_info=True | 05:05 |
| ho | clayg: sorry, I found it. type for error can be gotten. | 05:05 |
| clayg | you can think of logger.exception as an alias for log.error(exc_info=True) | 05:05 |
| clayg | ho well it does the whole traceback - which would be useful if you needed to suss out which of the two or three lines that could have raised that assertion error acctually did | 05:06 |
| clayg | but i'm not sure it's important which one raised the assertion - maybe i would just rather it raise something more specific - like "DiskFileInvalidHashdirContents" or something :\ | 05:07 |
| ho | clayg: the messages are quite useful I think and distinguish the line of code. | 05:08 |
| clayg | acoles_away: it bugs me that hash_listdir_cleanup calls gather_ondisk_files, then in yeild_hashes after hash_listdir_cleanup we have to call gather_ondisk_files again | 05:08 |
| clayg | ho: how do they distinguish - i honestly can't tell when hash_listdir_cleanup is raising the assertion or when it's gather_ondisk_files | 05:09 |
| clayg | in a few months I won't know if it was one of those or parse_on_disk_filename | 05:09 |
| clayg | acoles_away: oh crap - i probably should have named that parse_ondisk_filename (drop the _ in on_disk) | 05:09 |
| clayg | bah _gather_ondisk_files and gather_ondisk_files :\ | 05:12 |
| clayg | on and get_ondisk_files too! | 05:12 |
| ho | clayg: I wanted to say the message is unique string (L2081-2082) http://paste.openstack.org/show/197387/ when i investigate a trouble, I can use the string for grep. | 05:13 |
| clayg | _get_ondisk_file -> get_ondisk_files -> gather_ondisk_files -> _gather_ondisk_files | 05:14 |
| clayg | that's way simple | 05:14 |
| mattoliverau | clayg: lol, so logical :P | 05:15 |
| clayg | ho: yeah I'm not sure how much it matters to yield_hashes - the point is some hashdir had some garbage in it and was skipped :P | 05:16 |
| clayg | ho: I *really* don't understand if maybe the second call to gather_ondisk_files in yield_hashes maybe never happens if the path is invalid? | 05:16 |
| clayg | ho: probably on acoles_away will be able to refactor it | 05:16 |
| clayg | s/path is invalid/path contains an invalid file set/ | 05:17 |
| ho | clayg: i think so too :-) | 05:18 |
| *** shakamunyi has quit IRC | 05:18 | |
| clayg | ho: hehe | 05:20 |
| clayg | maybe if ECDiskFileManger just returns the gather_ondisk_files results dict instead of just the files list? | 05:21 |
| clayg | I wonder if it'd be wise try and bury the tombstone expiration in gather_ondisk_files - and hash_cleanup_listdir get's a little simpler | 05:22 |
| *** shakamunyi has joined #openstack-swift | 05:22 | |
| clayg | _gather_ondisk_files is sorta trixy with it's little lambda's over the filename | 05:24 |
| clayg | and it's updating of the passed in context | 05:24 |
| clayg | that acoles_away :\ | 05:24 |
| mattoliverau | clayg: yeah, that did my head in while reading it for the first time :) | 05:25 |
| clayg | i think i sorta glassed over trying to grok it and went stright to the tests :D | 05:25 |
| *** thumpba has quit IRC | 05:28 | |
| clayg | yay verified! | 05:40 |
| mattoliverau | nice! I for one will +2 it :) | 05:44 |
| clayg | bah - it's too much hash_cleanup_listdir can't both find all the files that can be reaped and also select the .data matching the frag_index - unless the result dict can return a list for .data | 05:53 |
| kota_ | clayg: are you working on 165188? | 05:55 |
| clayg | kota_: yeah thinking about ho's comments and just trying to make the most of acoles work on the gather_ondisk_files methods | 05:55 |
| kota_ | clayg: your log seems similer the thing I'm just now writing comments on 165188 | 05:56 |
| clayg | kota_: oh good - maybe you can figure out the right thing to do | 05:56 |
| clayg | i think i'm going to give up and move on down the chain | 05:56 |
| kota_ | clayg: ok, I am going to push it right now even though I don't finish my walking though. | 05:57 |
| kota_ | clayg: done, right now :D | 05:58 |
| kota_ | clayg: I hope it could help your work to figure out the right thing. | 05:58 |
| *** ktsuyuzaki has joined #openstack-swift | 06:03 | |
| ktsuyuzaki | clayg: anyways, I am going to continue to think about ho's comment. | 06:03 |
| ktsuyuzaki | clayg: ah, I am kota_ even the name was changed. | 06:04 |
| *** kota_ has quit IRC | 06:05 | |
| *** ktsuyuzaki has quit IRC | 06:05 | |
| *** kota_ has joined #openstack-swift | 06:05 | |
| kota_ | retreiving my name expired w/ timeout. | 06:06 |
| *** ktsuyuzaki has joined #openstack-swift | 06:08 | |
| ktsuyuzaki | wow... my wifi condition seems to have problems... | 06:09 |
| *** kota_ has quit IRC | 06:11 | |
| clayg | whoa | 06:11 |
| *** kota_ has joined #openstack-swift | 06:14 | |
| *** ktsuyuzaki has quit IRC | 06:17 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:17 | |
| mattoliverau | ktsuyuzaki or kota_ or whoever you are, pull yourself together man! you seem to becoming two :P | 06:17 |
| *** kota_ has quit IRC | 06:18 | |
| *** jamielennox is now known as jamielennox|away | 06:20 | |
| ktsuyuzaki | mattoliverau: wow... ktsuyuzaki and kota_ are same person. ktsuyuzaki is a sub nick on my irc client when kota_ didn't be expired for some reason. | 06:20 |
| *** silor has joined #openstack-swift | 06:21 | |
| ktsuyuzaki | hmm...I'm stopping to retry to login until my the network fixed. So sorry. | 06:22 |
| *** ktsuyuzaki has quit IRC | 06:22 | |
| clayg | mattoliverau: lol | 06:23 |
| *** shakamunyi has quit IRC | 06:25 | |
| *** Akshat has joined #openstack-swift | 06:42 | |
| *** nshaikh has joined #openstack-swift | 06:53 | |
| *** SkyRocknRoll has joined #openstack-swift | 06:54 | |
| *** SkyRocknRoll has joined #openstack-swift | 06:54 | |
| *** kota_ has joined #openstack-swift | 07:00 | |
| kota_ | I'm comming back. | 07:00 |
| ho | kota_: welcome back! | 07:01 |
| mattoliverau | Yay! Your you again! | 07:02 |
| *** Akshat has quit IRC | 07:03 | |
| kota_ | ho, mattoliverau: thanks ; ) | 07:04 |
| *** joeljwright has joined #openstack-swift | 07:20 | |
| *** km has joined #openstack-swift | 07:31 | |
| *** jordanP has joined #openstack-swift | 07:38 | |
| *** ppai_ has quit IRC | 07:44 | |
| openstackgerrit | Merged openstack/swift: Fix AttributeError and logging in ECAppIter https://review.openstack.org/168543 | 07:52 |
| *** ppai_ has joined #openstack-swift | 07:57 | |
| clayg | oh - that's good | 08:04 |
| clayg | oh goodness, that was a big change :P | 08:05 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Update contianer sync to use internal client https://review.openstack.org/143791 | 08:07 |
| *** jistr has joined #openstack-swift | 08:09 | |
| clayg | torgomatic: patch 166576 is all conflict-y I don't even know | 08:10 |
| patchbot | clayg: https://review.openstack.org/#/c/166576/ | 08:10 |
| openstackgerrit | Merged openstack/swift: Add Swift Design Principles to CONTRIBUTING.md https://review.openstack.org/168221 | 08:10 |
| clayg | cschwede: good morning! | 08:13 |
| cschwede | clayg: Good Morning - still awake? | 08:13 |
| clayg | yeah - i have a thing in the mornig so i'll get started kinda late tomorrow | 08:16 |
| clayg | wanna get a jump start | 08:16 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Add Fragment Index filter support to ssync https://review.openstack.org/165188 | 08:17 |
| *** haigang has quit IRC | 08:18 | |
| *** haigang has joined #openstack-swift | 08:18 | |
| *** haigang has quit IRC | 08:19 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Removed an unused variable in decode_segments_from_fragments https://review.openstack.org/168810 | 08:22 |
| *** admin6_ has joined #openstack-swift | 08:23 | |
| kota_ | cschwede: still here? | 08:52 |
| cschwede | kota_: Hey Kota! Yes | 08:53 |
| kota_ | cshwede: morning | 08:53 |
| kota_ | cshwede: your comment on 165188 | 08:53 |
| *** ppai_ has quit IRC | 08:53 | |
| cschwede | kota_: i just revoked the comment | 08:54 |
| cschwede | a minute ago :) | 08:54 |
| kota_ | cshwede: the "break" at replicated is to stop the for-loop hash_cleanup_listdir | 08:54 |
| kota_ | cshwede: oh, really | 08:54 |
| cschwede | yes, i just realized that as well | 08:54 |
| *** haigang has joined #openstack-swift | 08:55 | |
| kota_ | cschwede: oh, sorry for my nosy | 08:55 |
| cschwede | kota_: no worries, good to get some confirmation! | 08:55 |
| kota_ | cschwede: : ) In addition, I have a thing to wonder on the patch. | 08:56 |
| kota_ | Could you see my comments on patch set 13? | 08:56 |
| kota_ | cshwede: espicially about test/unit/obj/test_diskfile.py | 08:57 |
| cschwede | kota_: checking | 08:57 |
| kota_ | cshwede: thanks | 08:58 |
| kota_ | cschwede: On the EC, diskfile won't have both .meta and .ts because gather_ondisk_files ignores .meta newer than .ts. | 08:59 |
| kota_ | cschwede: However the diskfile changes in the patch seems prior to .meta than .ts... | 09:00 |
| kota_ | cschwede: I'm not sure it is the right thing but still wondering... | 09:00 |
| cschwede | kota_: regarding the comment in test_diskfile - i think it is called in reverse order, thus newer timestamps are checked first | 09:05 |
| clayg | oh yeah I ment to point to acoles_away that there was some comments on the older change sets that he should look at | 09:05 |
| cschwede | kota_: your comment is right if the older timestamps are checked first, but i think this is not the case | 09:06 |
| *** ppai_ has joined #openstack-swift | 09:07 | |
| kota_ | cshwede: I still think the reverse order case... | 09:07 |
| kota_ | cshwede: in gather_ondisk_file | 09:08 |
| kota_ | cshwede: set_valid_fileset() is called only when .durable or .ts found | 09:09 |
| kota_ | cshwede: right? | 09:09 |
| kota_ | cshwede: and then, if .ts found before .durable found, context['meta'] is assgined None becase have_valid_fileset() is False. | 09:10 |
| kota_ | cshwede: in any timestamp cases...i think. | 09:11 |
| cschwede | kota_ yes, but have a look at gather_ondisk_files, that calls _gather_on_disk_file - it will start with the newest file first | 09:11 |
| kota_ | cschwede: yes, ts2 is newer than ts1. so the first ts2.meta will appeared in _gather_on_diskf_file. | 09:13 |
| kota_ | cschwede: it will be set context['meta'] but set_valid__fileset is not called. | 09:13 |
| kota_ | cschwede: and then, ts1 will appear in _gather_ondisk_file it'll make a valid_file set like as context = {'.ts': ts1} | 09:15 |
| kota_ | cschwede: oh, sorry I'm looking at _gather_on_disk_file, right now. | 09:16 |
| kota_ | cschwede: i.e. ts2.meta is set but dropped when ts1.ts is set and it results in {'.ts': ts1}, i think. | 09:18 |
| *** bkopilov has quit IRC | 09:18 | |
| kota_ | because .meta doesn't trigger have_valide_fileset() to be True... | 09:21 |
| *** bkopilov has joined #openstack-swift | 09:23 | |
| cschwede | hmm | 09:27 |
| cschwede | is there ever a .meta without a .durable context? | 09:27 |
| kota_ | cschwede: wait a minutes, I might find the test code on feature/ec branch unit test. | 09:27 |
| kota_ | cschwede: In my memory .durable allows to be newer .meta. | 09:29 |
| kota_ | cschwede: similar test is here, https://github.com/openstack/swift/blob/feature/ec/test/unit/obj/test_diskfile.py#L731-L735 | 09:30 |
| cschwede | kota_: yes, but i’m wondering if there is ever a .meta without a .durable? i don’t think so? and then set_valid_fileset is set in line 2007? | 09:31 |
| cschwede | kota_: your linked test is not the same - the tombstone has the same timestamp, not an older one | 09:31 |
| kota_ | cschwede: yes doesn't same, ah, I might miss something...thinkingg... | 09:32 |
| kota_ | cschwede: yes, if .durable exists _gather_on_diskf_file call set_valid_fileset. | 09:36 |
| *** bkopilov has quit IRC | 09:37 | |
| *** bkopilov has joined #openstack-swift | 09:39 | |
| *** acoles_away is now known as acoles | 09:39 | |
| kota_ | cschwede: I jsut realized I might have misunderstanding, so sorry for that. Maybe it need to take atime to look at more deeply. | 09:40 |
| acoles | morning! | 09:41 |
| cschwede | kota_: no worries, it’s good to talk about the code! | 09:41 |
| cschwede | Hello Alistair! | 09:41 |
| kota_ | acoles: morning! | 09:41 |
| acoles | cschwede: hi, just scrolling back, | 09:41 |
| acoles | kota_: hi | 09:42 |
| kota_ | acoles: hi | 09:42 |
| openstackgerrit | Merged openstack/swift: Removed an unused variable in decode_segments_from_fragments https://review.openstack.org/168810 | 09:46 |
| acoles | cschwede: kota_ : i see you were discussing gather_ondisk_files - anything i can help explain? | 09:47 |
| acoles | clayg: yeah the call to hash_cleanup_listdir followed by gather_ondisk_files in yield_hashes has been annoying me all weekend, i just needed a little more time to refactor HCL so i could make a single call to get qhat yield_hashes wants | 09:49 |
| acoles | s/qhat//what/ | 09:49 |
| cschwede | acoles: thanks, i think i got it. i’m currently thinking if it might make sense to add a few comment lines into _gather_on_disk_file | 09:51 |
| acoles | cschwede: i was going to write a book about it ;) | 09:52 |
| cschwede | haha, very nice! | 09:52 |
| acoles | cschwede: seriously i'll add comments | 09:53 |
| acoles | cschwede: kota_ : the gather_ondisk_files -> _gather_ondisk_file makes more sense when you see how simple it is to refactor and share code with the replication get_ondisk_files, but clayg wouldn't let me go that far (yet) ;P | 10:09 |
| *** dmorita has quit IRC | 10:17 | |
| clayg | acoles: oh i see how you're putting it on me now | 10:29 |
| *** ppai_ has quit IRC | 10:29 | |
| acoles | clayg: ;) | 10:30 |
| acoles | clayg: thanks for cleaning up the unbound 'files' reference in yield_hashes. | 10:31 |
| * acoles slaps himself | 10:31 | |
| *** km has quit IRC | 10:32 | |
| *** km has joined #openstack-swift | 10:32 | |
| *** geaaru has joined #openstack-swift | 10:37 | |
| *** bkopilov has quit IRC | 10:37 | |
| *** haigang has quit IRC | 10:40 | |
| *** ppai_ has joined #openstack-swift | 10:41 | |
| *** haomaiwang has quit IRC | 10:47 | |
| *** bkopilov has joined #openstack-swift | 10:47 | |
| clayg | acoles: well then ho made me change it again | 10:53 |
| acoles | clayg: you mean relocating the excepts? | 10:53 |
| clayg | so is everyone happy with 165188 | 10:53 |
| clayg | acoles: yeah putting everything in one big try | 10:54 |
| clayg | i liked getting the assertion error strs in there it turned out | 10:54 |
| acoles | yeah i'm cool with that | 10:54 |
| acoles | cschwede: according to this https://docs.python.org/2/library/unittest.html#unittest.TestCase.assertRaises self.assertRaises as a context manager is new in 2.7, but I'm sure I have seen it in swift recently | 10:55 |
| acoles | clayg: i'm happy with 165188 but I am author (or am I ? I have lost track !) | 10:59 |
| acoles | gerrit says i am | 10:59 |
| acoles | i'm annoyed with myself by the duplication in calling hcl then gather_ondisk_files | 11:00 |
| acoles | which i could fix now or later. | 11:00 |
| cschwede | acoles: ah yes, you’re right! i think i saw something similar in swiftclient recently | 11:05 |
| acoles | cschwede: but swiftclient still tests py26 i think? | 11:05 |
| acoles | cschwede: anyway, since we have just moved those tests i'd prefer not to have to change them in 165188, | 11:06 |
| cschwede | acoles: fine with me - as i said in the comments, it’s not a blocker for me | 11:07 |
| cschwede | acoles: just checked - this is where i saw it: https://github.com/openstack/python-swiftclient/blob/master/tests/unit/test_swiftclient.py#L1184-L1185 but that’s not a context manager. | 11:09 |
| cschwede | heh, the amount of available reviewers to approve patch 165188 is somewhat limited - at least three people submitted patchsets ;) | 11:12 |
| patchbot | cschwede: https://review.openstack.org/#/c/165188/ | 11:12 |
| *** kei_yama has quit IRC | 11:13 | |
| *** ho has quit IRC | 11:15 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: WIP: older tombstone is valid rather than newer meta https://review.openstack.org/168853 | 11:18 |
| kota_ | cschwede: I wrote a unit for my idea at 168853 | 11:19 |
| kota_ | cschwede: it might be wrong but i wonder if you could review again whether the assumption is right or not. | 11:20 |
| kota_ | patch 168853 is passed in my env so _get_ondisk_files() absolutely return older tombstone rather than newer .meta if no .durable or newer .ts exists. | 11:23 |
| patchbot | kota_: https://review.openstack.org/#/c/168853/ | 11:23 |
| acoles | kota_: that test *should* fail because of line 2020 in diskfile.py https://github.com/openstack/swift/blob/feature/ec/swift/obj/diskfile.py#L2020 | 11:24 |
| kota_ | acoles: good test? | 11:25 |
| kota_ | acoles: L2020 drops .meta, yes, the newest .meta should be dropped so... | 11:26 |
| acoles | kota_: sorry, ignore my last remark, it should pass | 11:26 |
| acoles | kota_: the .ts will be returned, the .meta will be left on disk | 11:26 |
| kota_ | acoles: ok, it fits on my idea. | 11:27 |
| *** km has quit IRC | 11:27 | |
| cschwede | kota_: i think that behavior is as expected (because of the missing .durable) | 11:27 |
| kota_ | cschwede: ok, returning back the test_yield_hashes_yields_meta_timestamp on patch 165188 | 11:29 |
| *** SkyRocknRoll has quit IRC | 11:29 | |
| patchbot | kota_: https://review.openstack.org/#/c/165188/ | 11:29 |
| kota_ | I guess, this test doesn't have .durable, right? | 11:29 |
| kota_ | if it is true, L1010 on test_diskfile of 165188 patch set14 should be ts1.ts because no .durable file and .ts comes up on the gathering job. | 11:32 |
| cschwede | kota_: i justo observed that you need a .durable and .data to ignore the .ts. I added the following lines to your patch: http://paste.openstack.org/show/197521/ | 11:33 |
| cschwede | that only works if there is a .data, and not if there is only a .durable (which makes sense to me) | 11:33 |
| kota_ | cschwede: thanks! | 11:35 |
| kota_ | cschwede: yeah, the paste fits my idea. the .data works with the same timestamp .durable it makes me sense. Hoever.. | 11:36 |
| kota_ | cschwede: test_yield_hashes_yields_meta_timestamp seems not to have .durable, that is the last thing I still wonder... | 11:37 |
| acoles | kota_: line 1010 in tests_diskfile.py is testing replication policy not EC | 11:38 |
| kota_ | Oh my god... | 11:39 |
| kota_ | acoles: ya, absolutely | 11:39 |
| kota_ | acoles, cschwede: so sorry to work for that, it is definitly my mistake... | 11:40 |
| cschwede | kota_: no worries, it’s better we think one more time over all of this! | 11:41 |
| acoles | kota_: see line 1419. I put those tests in because some of the behaviour around .meta files is possibly confusing, so the tests illustrate the behavior | 11:46 |
| mattoliverau | evening all, did you guys need to check out patch 165188, I'm like the only one who hasn't submitted a patchset :P | 11:46 |
| patchbot | mattoliverau: https://review.openstack.org/#/c/165188/ | 11:46 |
| acoles | mattoliverau: if you can the i'm sure it would add to your tally of beers in vancouver | 11:47 |
| kota_ | acoles: I see, it makes me sense, much appreciated. | 11:47 |
| acoles | kota_: tbh i think that yield_hashes on master is broken when there is a t2.meta and t1.ts - ssync will attempt to replicate the file at t2, but error when trying to open the file since there is no .data | 11:49 |
| acoles | kota_: but I guess no one has used ssync with fast-post meta files much | 11:51 |
| kota_ | acoles: lol | 11:51 |
| acoles | kota_: but *thank you* for checking these tests thoroughly | 11:52 |
| kota_ | acoles: I didn't notice that too, let's fix that after EC relased on Kilo :-) | 11:52 |
| kota_ | acoles: sure, I'm very interested those patches on EC, it's really fun to review awsome patches :-) | 11:54 |
| *** jistr has quit IRC | 12:07 | |
| *** jistr has joined #openstack-swift | 12:08 | |
| *** Gues_____ has joined #openstack-swift | 12:15 | |
| mattoliverau | acoles: you mention in the review "the call to hash_cleanup_listdir followed by gather_ondisk_files can be simplified".. are you planning on doing that now or post beta (or a follow up patch)? Just needing to know if we want to get the change merged now | 12:15 |
| mattoliverau | s/review/patchset comment/ | 12:16 |
| kota_ | Thanks guys, I'm going to leave my office to go back home : ) | 12:23 |
| acoles | kota_: thanks have a good night | 12:23 |
| *** kota_ has quit IRC | 12:24 | |
| *** panbalag has joined #openstack-swift | 12:24 | |
| acoles | mattoliverau: i'm working on it but can be a follow up patch since we are accumulating +'s on 165188 | 12:25 |
| *** Gues_____ has quit IRC | 12:25 | |
| acoles | mattoliverau: i'll also add some more commentary to gather_ondisk_files... | 12:26 |
| acoles | mattoliverau: so yeah, maybe follow up is best, keep the train rolling forwards | 12:26 |
| cschwede | yeah, more merges! | 12:28 |
| mattoliverau | Kk, sounds like a plan, thanks acoles, I'll +A and then head to bed :) | 12:28 |
| acoles | mattoliverau: thanks for that! | 12:29 |
| cschwede | acoles: which patch in https://review.openstack.org/#/q/status:open+project:openstack/swift+branch:feature/ec,n,z is the next most important from your point of view? | 12:29 |
| acoles | cschwede: idk! i guess the reconstructor 131872 or container sync path 143791 | 12:33 |
| acoles | i assume clayg is not still awake ? ^^ | 12:33 |
| cschwede | acoles: thx, i was thinking the same | 12:33 |
| mattoliverau | Done, night guys! Nice work btw | 12:33 |
| acoles | cschwede: i will probably head to 131872 | 12:33 |
| acoles | mattoliverau: g'nite | 12:33 |
| mattoliverau | I hope clayg is asleep.. His was online my entire day.. Well pretty much.. He's a machine! | 12:34 |
| acoles | i think "clayg" is actually a globally distributed team | 12:34 |
| cschwede | haha, that would explain a lot :D | 12:35 |
| *** Trixboxer has joined #openstack-swift | 12:36 | |
| clayg | ye of little faith | 12:36 |
| acoles | ! | 12:37 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: wip: ec reconstructor probe test https://review.openstack.org/164291 | 12:37 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Erasure Code Reconstructor https://review.openstack.org/131872 | 12:37 |
| clayg | i've just been busy fixing shit | 12:37 |
| *** sandywalsh has quit IRC | 12:37 | |
| acoles | clayg: ye of little sleep | 12:37 |
| cschwede | jeez! | 12:37 |
| clayg | well how can anyone sleep when there's shit like this -> https://gist.github.com/clayg/959a5d77353a91a4b10a | 12:38 |
| clayg | i sent an email to tsg he normally patches shit I have no idea - in the meantime Im just working around it | 12:39 |
| clayg | anyway i'm trying to get something going for the reconstrcutor - it needs a bunch more tests | 12:39 |
| clayg | but all the probetests are passing for me relibably I think? | 12:39 |
| clayg | ya'll figure out what you wanna do with the ssync-fi thing | 12:39 |
| *** sandywalsh has joined #openstack-swift | 12:40 | |
| clayg | i'ma take a nap for a few - i won't be on till like idk - 1pm or something PST - so i'll see most of you when you come online in the morning | 12:40 |
| acoles | clayg: its en route to land | 12:40 |
| acoles | clayg: mattoliverau +"A'd | 12:40 |
| clayg | peluse: welcome back (when you get here) | 12:40 |
| clayg | sweet - i hope I rebased on the right revision | 12:41 |
| acoles | clayg: rest well | 12:41 |
| clayg | why *did* we merge the liberasure dependency into master? | 12:42 |
| clayg | anyway - we've probably got to update something to mention that whatever liberasure devel headers are required to install swift now? | 12:42 |
| clayg | idk, seems strange | 12:43 |
| clayg | k - later | 12:43 |
| *** ppai_ has quit IRC | 12:45 | |
| cschwede | clayg: i submitted it to feature/ec, don’t know why it needed to be on master | 12:48 |
| cschwede | https://review.openstack.org/#/c/164172/ | 12:48 |
| *** shakamunyi has joined #openstack-swift | 12:57 | |
| *** annegentle has joined #openstack-swift | 13:08 | |
| *** haigang has joined #openstack-swift | 13:12 | |
| *** shakamunyi has quit IRC | 13:13 | |
| openstackgerrit | Merged openstack/swift: Add Fragment Index filter support to ssync https://review.openstack.org/165188 | 13:17 |
| openstackgerrit | Yuan Zhou proposed openstack/swift: EC: Allow proxy to read from all SN in EC policy https://review.openstack.org/168254 | 13:24 |
| *** chlong has joined #openstack-swift | 13:36 | |
| *** nshaikh has quit IRC | 13:51 | |
| *** dencaval has joined #openstack-swift | 13:52 | |
| *** jrichli has joined #openstack-swift | 13:56 | |
| *** haomaiwang has joined #openstack-swift | 14:01 | |
| openstackgerrit | Christian Schwede proposed openstack/swift: Check if REST API version is valid https://review.openstack.org/168925 | 14:17 |
| *** zaitcev has joined #openstack-swift | 14:21 | |
| *** ChanServ sets mode: +v zaitcev | 14:21 | |
| *** haomaiwang has quit IRC | 14:24 | |
| *** haomaiw__ has joined #openstack-swift | 14:25 | |
| *** G________ has joined #openstack-swift | 14:30 | |
| *** lcurtis has joined #openstack-swift | 14:36 | |
| *** egon has quit IRC | 14:39 | |
| *** me has joined #openstack-swift | 14:42 | |
| *** me is now known as Guest54459 | 14:43 | |
| *** G________ has quit IRC | 14:44 | |
| *** egon has joined #openstack-swift | 14:45 | |
| *** haomaiwa_ has joined #openstack-swift | 14:47 | |
| *** zul has quit IRC | 14:47 | |
| *** Akshat has joined #openstack-swift | 14:48 | |
| Akshat | Hi Clayg | 14:49 |
| Akshat | Hi Ctennis | 14:49 |
| Akshat | clayg:I still could not overcome that error | 14:49 |
| *** haomaiw__ has quit IRC | 14:49 | |
| *** haigang has quit IRC | 14:53 | |
| *** zul has joined #openstack-swift | 14:53 | |
| *** G________ has joined #openstack-swift | 14:53 | |
| notmyname | good morning | 14:58 |
| notmyname | looks like I just barely missed clayg | 14:59 |
| notmyname | in answer to the question posed last night... | 15:00 |
| *** mahatic has joined #openstack-swift | 15:00 | |
| notmyname | yes, pyeclib landed on master because of two things | 15:00 |
| *** reed has joined #openstack-swift | 15:01 | |
| Akshat | Hi notmyname | 15:01 |
| *** zul has quit IRC | 15:02 | |
| *** zul has joined #openstack-swift | 15:03 | |
| notmyname | first, the global-requirements issues with the feature branch (since resolved, but was an issue). also, we had the openstack requirement for getting dependencies in by march 19 | 15:03 |
| notmyname | since there was no release between then and the one with ec, we landed the requirement on master | 15:04 |
| Akshat | Hey I am facing issue with my swift cluster | 15:04 |
| Akshat | when I am loading the cluster...I am getting errors to update container,account server | 15:04 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Minor cleanup to EC yield_hashes and gather_ondisk_files https://review.openstack.org/168941 | 15:08 |
| *** zul has quit IRC | 15:08 | |
| *** zul has joined #openstack-swift | 15:08 | |
| *** Akshat has quit IRC | 15:08 | |
| acoles | mattoliverau: cschwede: kota: ^^ cleanups from patch 165188 | 15:09 |
| patchbot | acoles: https://review.openstack.org/#/c/165188/ | 15:09 |
| *** Akshat has joined #openstack-swift | 15:10 | |
| *** haomaiwa_ has quit IRC | 15:18 | |
| *** haomaiwa_ has joined #openstack-swift | 15:18 | |
| zaitcev | notmyname: I knew there would be a reason to add PyECLib, just forgot what it was. | 15:18 |
| *** annegentle has quit IRC | 15:19 | |
| *** annegentle has joined #openstack-swift | 15:19 | |
| *** shakamunyi has joined #openstack-swift | 15:25 | |
| *** chlong has quit IRC | 15:26 | |
| clayg | just missed me by like 3 hours | 15:46 |
| *** haomaiwa_ has quit IRC | 15:46 | |
| clayg | cschwede: you can't put that on me! | 15:48 |
| cschwede | clayg: what? why not? it’s your patch, thus you get the kudos | 15:49 |
| clayg | cschwede: I think patch 168925 was acctually closer on than patch 168509 | 15:49 |
| patchbot | clayg: https://review.openstack.org/#/c/168925/ | 15:49 |
| clayg | you mean I get the *blame* when it breaks someone's client - i see what you're doing. | 15:50 |
| cschwede | clayg: heh, well, i didn’t look at it from that side | 15:51 |
| clayg | cschwede: here we'll try that | 15:52 |
| clayg | ;) | 15:52 |
| *** shakamunyi has quit IRC | 15:52 | |
| clayg | i'm honestly just not working on it - really do think what you had done with the method extraction (maybe in constraints) would be better. | 15:53 |
| cschwede | clayg: shall i push a new patchset on top of yours? or restore my change? | 15:54 |
| clayg | yeah we should keep everything on he same patch - just push over that change id we already have going | 15:55 |
| *** Akshat has quit IRC | 15:58 | |
| cschwede | clayg: hmm, just trying to cherry-pick my patch on top of yours. actually i’m not sure what to keep, for example commit msg | 15:59 |
| clayg | cschwede: i'm pretty sure my commit msg was crap - i exected it to change - was yours any good? | 16:00 |
| cschwede | clayg: i’m trying to make the best out of both msgs | 16:01 |
| clayg | nice - you rock | 16:01 |
| acoles | clayg: peluse: i'm looking at patch 131872 and see self.job['policy'].ts_to_fname in ssync_sender - didn't ts_to_fname go away or did i miss something? | 16:02 |
| patchbot | acoles: https://review.openstack.org/#/c/131872/ | 16:02 |
| acoles | line 352 | 16:02 |
| cschwede | clayg: i will submit a new patchset later, leaving for dinner now | 16:04 |
| *** annegentle has quit IRC | 16:07 | |
| clayg | cschwede: np, thanks! | 16:08 |
| *** vinsh has joined #openstack-swift | 16:09 | |
| *** shakamunyi has joined #openstack-swift | 16:10 | |
| *** Akshat has joined #openstack-swift | 16:13 | |
| jordanP | hi guys. If you could take a look at https://review.openstack.org/#/c/168990/ (swift-devstack integration issue, w.r.t PyEClib), that would be great :) | 16:18 |
| jordanP | ^^ chmouel, | 16:18 |
| ctennis | sorry Akshat I don't have any more ideas, it sounds to me like something is misconfigured. | 16:19 |
| chmouel | joearnold: wouldn't need the full build-essential thing ? | 16:20 |
| *** Akshat has quit IRC | 16:22 | |
| jordanP | chmouel, hummm :) | 16:23 |
| chmouel | cause make is nice but i guess gcc/glibc-devel etc.. would be needed? | 16:23 |
| jordanP | I think make is enough... I can double check that | 16:24 |
| jordanP | chmouel, gcc is already in files/debs/general | 16:25 |
| jordanP | but I will double check anyway | 16:25 |
| chmouel | jordanP: cool, would be good you can test on a pristine vm | 16:25 |
| jordanP | sure | 16:26 |
| *** jistr has quit IRC | 16:44 | |
| peluse | acoles, I have some catch up reading to do and my head is *killing* me :) Besides wine still running through my veins, didn't get to sleep til about 3am. I'm mostly worthless today I think, just FYI | 16:57 |
| acoles | peluse: sounds like you have a had a great weekend! | 16:57 |
| acoles | peluse: i'm writing some unit tests for the ssync_sender changes in patch 131872 fyi | 16:58 |
| patchbot | acoles: https://review.openstack.org/#/c/131872/ | 16:58 |
| *** erlon has joined #openstack-swift | 16:59 | |
| jordanP | chmouel, 'make' is enough. I've left a comment on the code review | 17:05 |
| clayg | acoles: there's about 10 different tests that will look like test.unit.obj.test_reconstructor.test_reconstruct_fa_(no_errors) - but I won't be writing them till this afternoon | 17:08 |
| notmyname | good morning (again) | 17:08 |
| notmyname | peluse: glad you made it back | 17:08 |
| clayg | peluse: when you say you weren't seeing issues I think that just means you weren't testing it enough - that or you're using the isa-l that doesn't seem to exibit the issue - which is also cool! | 17:08 |
| acoles | clayg: ok so if i concentrate on test_ssync_sender.py we don't overlap effort? | 17:09 |
| clayg | peluse: because basically if you want to rebuilt 5 and you have 0, 1, 2, 4 you get back garbage - so it really all just depends on what nodes you turn off | 17:10 |
| clayg | acoles: k, makes sense - thanks! | 17:10 |
| clayg | acoles: unless you were going to try to get that tombstone cleanup out of that otherwise beautiful bit of work in the _gather_ondisk_file cleanup! | 17:10 |
| clayg | k, i'm out of pocket for a bit then i think | 17:11 |
| peluse | clayg, yeah I switched to ISAL a while back... | 17:12 |
| clayg | peluse: good call apparently! | 17:13 |
| *** G________ has quit IRC | 17:13 | |
| peluse | clayg, acoles so what do you guys want me looking at right now? | 17:13 |
| acoles | clayg: heh. i'd love to do that but i figure ssync_sender tests might be a priority - did you see my comment above re "self.job['policy'].ts_to_fname in ssync_sender " | 17:13 |
| acoles | clayg: like, that can't be tested | 17:14 |
| *** tsg_ has joined #openstack-swift | 17:16 | |
| acoles | peluse: i was out of loop myself over weekend, clayg probably has best view on the state of things. everything below 131872 (recon) has landed in that chain | 17:17 |
| peluse | OK cool. Thinking I'd start wiht patch 168941 | 17:17 |
| patchbot | peluse: https://review.openstack.org/#/c/168941/ | 17:17 |
| notmyname | peluse: the starred patches on the review dashboard are where we are. acoles has one patch. and then there's the reconstructor | 17:17 |
| notmyname | I think we really need to get those landed today | 17:18 |
| *** Gues_____ has joined #openstack-swift | 17:19 | |
| acoles | notmyname: today which time zone? anyone in hawaii? :P | 17:22 |
| notmyname | :-) | 17:22 |
| notmyname | clayg's time zone is what matters, since he'll be doing the merge to master. but, to be fair, I'm not actually sure clayg follows "time zones" (or anything resembling a normal schedule) ;-) | 17:23 |
| clayg | notmyname: torgomatic need to rebase the multi-rage get last I checkede | 17:24 |
| swifterdarrell | clayg: best accidentally-appropriate typo ever | 17:25 |
| acoles | "multi-rage" " | 17:25 |
| notmyname | clayg: right, but there's still an issue in it | 17:25 |
| clayg | the recon stuff needs more tests i think - but acoles is working on some of them - maybe peluse or I can do some more | 17:25 |
| notmyname | as of right now, I don't think the beta will have multi-range or container sync | 17:25 |
| peluse | he, clayg said 'recon' and not 'reconstructor' | 17:26 |
| clayg | notmyname: I *think* i'm ready to maybe start pulling out some piecies of feature/ec and see if we can start to play around with the patch chain on ec-review - are you going to get that setup today (is it already?) | 17:26 |
| clayg | peluse: DAMNIT - reconstrcu;lajksdfj - that fucking word sucks - we can't call it recon either | 17:26 |
| acoles | peluse: he's been doing that a lot | 17:26 |
| clayg | this is the most critical issue facing the future of ec right now - we have to rename it to the recoder | 17:27 |
| chmouel | jordanP: cool, will review later on :) | 17:27 |
| peluse | #vote | 17:27 |
| acoles | no, no - object-fixer !! :P | 17:27 |
| clayg | object-do-the-right-thing-button | 17:27 |
| peluse | we could always being back "the ruggedizer" | 17:27 |
| acoles | NO | 17:27 |
| peluse | :) | 17:28 |
| notmyname | the hard work, high stress, and lack of sleep have made you all crazy | 17:28 |
| peluse | acoles, nice stuff on the refactor of yield_hashes and gather | 17:28 |
| acoles | lets make the name of the daemon configurable | 17:29 |
| acoles | that always avoids making a decision | 17:29 |
| * peluse googles visitor pattern | 17:29 | |
| clayg | peluse: *right* - acoles is on his a game | 17:30 |
| peluse | no shit! | 17:30 |
| acoles | peluse: thx but clayg diss'd me leaving the "ugly" tombstone handling so i 'll have to go fix :P | 17:30 |
| * clayg mostly uses the make-the-crap-i-wrote-last-year-less-annoying-to-me-now pattern | 17:30 | |
| clayg | acoles: seriously it's like a fart in the middle of bethoven | 17:31 |
| peluse | acoles, you mean the first part of cleanup_ondisk_files()? | 17:31 |
| acoles | peluse: google 'gang of four patterns' | 17:31 |
| clayg | notmyname: awww no container-sync!? what was wrong with container-sync | 17:31 |
| peluse | that sounds like it should be rated X | 17:31 |
| acoles | peluse: yeah | 17:32 |
| clayg | notmyname: shit i forgot about the stupid bug with acccount-reaper | 17:32 |
| clayg | notmyname: anyway - what about the ec-review branch? | 17:32 |
| notmyname | clayg: next on my todo for today. on it | 17:32 |
| acoles | peluse: :) when we're done with EC go read their book and your life will change | 17:32 |
| clayg | acoles: meh | 17:32 |
| peluse | that's what I need... I life changing OO book :) | 17:33 |
| acoles | clayg: actually i'm just gonna delete that first piece of cleanup?ondisk and see if any tests fail | 17:33 |
| clayg | lol | 17:34 |
| clayg | good start! | 17:34 |
| peluse | acoles, I dopn't really understand the history behind that first part anyway, assumed it was an opmitization for a special case of some kinds | 17:35 |
| clayg | peluse: someone needed to do it - that way seemed least likely to do the wrong thing? | 17:35 |
| clayg | ok i'm off - see you gusy later | 17:36 |
| *** annegentle has joined #openstack-swift | 17:37 | |
| *** zhill has joined #openstack-swift | 17:38 | |
| *** silor has quit IRC | 17:41 | |
| *** annegentle has quit IRC | 17:43 | |
| *** annegentle has joined #openstack-swift | 17:44 | |
| *** admin6_ has quit IRC | 17:44 | |
| *** jordanP has quit IRC | 17:54 | |
| *** geaaru has quit IRC | 18:00 | |
| *** Gues_____ has quit IRC | 18:01 | |
| openstackgerrit | Christian Schwede proposed openstack/swift: Check if REST API version is valid https://review.openstack.org/168509 | 18:02 |
| cschwede | clayg: ^^ i refactored the patch just a tiny bit | 18:03 |
| *** tsg_ has quit IRC | 18:05 | |
| *** tsg has joined #openstack-swift | 18:05 | |
| *** greghaynes has quit IRC | 18:07 | |
| *** jrichli has quit IRC | 18:10 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Even more cleanup to EC on-disk file cleanup https://review.openstack.org/169035 | 18:11 |
| acoles | clayg: so peluse approved before i could cleanup the cleanup ^^ it wasn't as straightforward as i hoped | 18:12 |
| peluse | is it ever? | 18:15 |
| *** Gues_____ has joined #openstack-swift | 18:16 | |
| acoles | peluse: nope | 18:19 |
| *** jrichli has joined #openstack-swift | 18:21 | |
| openstackgerrit | Merged openstack/swift: Minor cleanup to EC yield_hashes and gather_ondisk_files https://review.openstack.org/168941 | 18:46 |
| openstackgerrit | Christian Schwede proposed openstack/swift: Check if REST API version is valid https://review.openstack.org/168509 | 19:17 |
| openstackgerrit | Alistair Coles proposed openstack/swift: WIP: EC specific ssync_sender fixes and tests https://review.openstack.org/169052 | 19:18 |
| acoles | peluse clayg ^^ WIP back later | 19:19 |
| peluse | k | 19:19 |
| *** mahatic has quit IRC | 19:20 | |
| *** acoles is now known as acoles_away | 19:20 | |
| tdasilva | cschwede, clayg: re 168509, shouldn't the response be a 501? 404 seems strange to me... | 19:21 |
| *** nikhil|afk has quit IRC | 19:21 | |
| cschwede | tdasilva: i think 5xx errors are server errors, but the error is on the client side in this case? | 19:22 |
| cschwede | actually i’m thinking if 422 is makes sense | 19:23 |
| notmyname | cschwede: ya, 400 or 422 seem reasonable (over 404) | 19:26 |
| tdasilva | cschwede: yeah, either 412 or 422 would be better in my mind, 404 seems deceiving | 19:26 |
| *** Gues_____ has quit IRC | 19:26 | |
| tdasilva | or 400 like notmyname said | 19:26 |
| notmyname | cschwede: tdasilva: so without listing an account, ... , then swift will currently return 412 | 19:31 |
| notmyname | (which I think could be better, but perhaps the consistency is beneficial? | 19:32 |
| cschwede | notmyname: i was using 412 in my first patchset as well. i don’t know - there are good reasons for 400,412 and 422 imo | 19:33 |
| cschwede | at least we can agree that 404 is not the best idea ;) | 19:33 |
| notmyname | if I did it all over again, I'd vote 422. but 412 seems more consistent with existing behavior | 19:34 |
| tdasilva | notmyname: yeah, in that case 412 is returned regardless of version thou, right? | 19:34 |
| notmyname | tdasilva: yes | 19:34 |
| notmyname | tdasilva: it's 412 if it can't figure out what the request was fo | 19:34 |
| tdasilva | right | 19:34 |
| notmyname | eg saio/xyzzy/ | 19:34 |
| *** Gues_____ has joined #openstack-swift | 19:36 | |
| tdasilva | I'd vote for 400 | 19:42 |
| *** dencaval has quit IRC | 19:42 | |
| *** teploxoh has joined #openstack-swift | 19:42 | |
| *** annegentle has quit IRC | 20:04 | |
| *** Gues_____ has quit IRC | 20:11 | |
| *** Gues_____ has joined #openstack-swift | 20:15 | |
| *** annegentle has joined #openstack-swift | 20:25 | |
| *** tsg has quit IRC | 20:31 | |
| *** tsg has joined #openstack-swift | 20:35 | |
| *** Gues_____ has quit IRC | 20:35 | |
| *** Gues_____ has joined #openstack-swift | 20:38 | |
| *** annegentle has quit IRC | 20:41 | |
| *** annegentle has joined #openstack-swift | 20:52 | |
| briancline | reading the 1.1 spec, 400 seems to make the most sense.. mostly since it stipulates the request should not be resent without some change/correction and makes no assumption about whether the rest of the request was valid or not | 20:53 |
| briancline | iirc, 412 preconditions are assumed to be header-based (like when being provided an expected etag for a PUT) | 20:54 |
| clayg | are we freaking out!? | 20:55 |
| briancline | that said, who actually follows the rfc on this stuff | 20:55 |
| clayg | i'm checking out patch 131872 again | 20:57 |
| patchbot | clayg: https://review.openstack.org/#/c/131872/ | 20:57 |
| *** bsdkurt has quit IRC | 21:00 | |
| *** bsdkurt has joined #openstack-swift | 21:01 | |
| openstackgerrit | Monty Taylor proposed openstack/python-swiftclient: Add socket-level read timeout parameter https://review.openstack.org/168673 | 21:02 |
| *** tsg has quit IRC | 21:04 | |
| *** acoles_away is now known as acoles | 21:06 | |
| *** rdaly2 has joined #openstack-swift | 21:07 | |
| *** Gues_____ has quit IRC | 21:09 | |
| clayg | acoles: what are you doing here!? | 21:11 |
| sweeper | lurking imo | 21:12 |
| sweeper | like everyone else | 21:12 |
| acoles | clayg: writing tests... | 21:12 |
| notmyname | acoles: can you help me grok https://review.openstack.org/#/c/169035/1 ? | 21:12 |
| notmyname | acoles: I' trying to be helpful but feel that I came into that discussion partway through | 21:13 |
| *** tsg has joined #openstack-swift | 21:13 | |
| notmyname | acoles: what's the "single junk in dir that may violate ondisk contract" | 21:14 |
| acoles | notmyname: ok, so i nearly called that one 'remove the fart from beethoven' cos thats how clayg described the previous cleanup | 21:14 |
| clayg | awwwwww we should have waited to merge patch 168941 is acoles was going to keep working on it - my fault for +2ing it I suppose | 21:14 |
| patchbot | clayg: https://review.openstack.org/#/c/168941/ | 21:14 |
| acoles | ;) | 21:15 |
| notmyname | acoles: ya, I liked that ;-) | 21:15 |
| acoles | looking | 21:15 |
| clayg | acoles: but you didn't remove it - you just added comments | 21:15 |
| clayg | :P | 21:15 |
| clayg | acoles: which tests broke when you just pulled it out? | 21:15 |
| notmyname | acoles: I was just trying to figure out the scenario. 'cause I think it needs a test. and I can write it, but I don't understand the "why" yet | 21:15 |
| vinsh | acoles: Head back to Colorado and drink some more beer :) | 21:15 |
| *** Gues_____ has joined #openstack-swift | 21:16 | |
| acoles | vinsh: that a helluva way for me to go for beer, but i have a glass of wine by my side ;) | 21:16 |
| vinsh | Just fine ;) | 21:16 |
| clayg | acoles: I'm pretty sure there's at least one test that specifically validates reclaimed tombstones - but I don't remember anything being to specific about that wierd "there has to be one file in the directory in order to reap too old tombstones" behavior of the current crap | 21:16 |
| *** rdaly2 has quit IRC | 21:16 | |
| acoles | clayg: notmyname there was a test that had a single .meta on the dir iirc. that breaks get_ondisk_files but legacy hash_cleanup_listdir allows it because of the 'if only one file don't even try get_ondisk_files' | 21:17 |
| acoles | clayg: notmyname so i was very careful not to change behavior wrt legacy even though i don't get the legacy behavior, hence the comment, but cleaning up the filename parsing to use our super new parse_ondisk_filename | 21:19 |
| acoles | like, i think if you had an obj dir with only junk.foo in it, that would just be ignored on master | 21:20 |
| *** annegentle has quit IRC | 21:21 | |
| acoles | notmyname: clayg: what i *wanted to do* was just call gather_ondisk_files, test the results for a .ts that is older than reclaim_age and if so move it to the obsolete files list, it was sweet, but some tests broke and tbh it doesnt seem a priority to make the code more beautiful :) | 21:22 |
| acoles | try removing the 'if .ts' condition and i expect the same tests will blow up | 21:23 |
| clayg | acoles: yeah do what you wanted to do - if there is a stupid test that only works on legacy - then move it to the replicated case of the mixin and rewrite a more reasonable test for the ec case | 21:26 |
| clayg | you don't have to do any of this of course - you're right about it not being the #1 priority | 21:26 |
| clayg | acoles: where did you end up on some ssync tests you said? | 21:26 |
| acoles | clayg: totally agree it should be done but i'm focussed on recosntructor and adding thise ssync tests in patch 169052 | 21:27 |
| patchbot | acoles: https://review.openstack.org/#/c/169052/ | 21:27 |
| clayg | redbo: YES! http://bouk.co/blog/monkey-patching-in-go/ | 21:28 |
| clayg | acoles: right on | 21:28 |
| acoles | clayg: peluse : so i have a question about the sync_revert case with ssync ... | 21:29 |
| acoles | sender starts with a durable and frag archives index 1 and 2, sends FA 2 to the receiver, then wants to cleanup FA2 from local node right? | 21:30 |
| peluse | yes | 21:31 |
| acoles | clayg: peluse : the 'cleanup' should not remove durable, right? cos FA 1 is still on local node, correct? | 21:31 |
| peluse | currect, jsut the FA | 21:31 |
| peluse | I'm working on the cleanup stuff right now, its got a few small issues | 21:32 |
| acoles | but if there is no FA 1 on local node then we *should* cleanup the durable? or can that not happen - no of course it can happen yeas? | 21:32 |
| peluse | what? | 21:32 |
| acoles | peluse: is the starting state is that local node has only FA 2 and durable, sends FA 2 to receiver, then cleanup needs to remove the durable?? | 21:33 |
| acoles | s/is/if/ | 21:33 |
| clayg | i'm realizing I'm not entirely following the high level flow - I get why a job would have a partition/path and and frag_index, but why does it have a "suffix" key - our jobs should be segregated mainly on the node(s) they're connecting to - which has everything to do with which fi's are in the part - which in turn tells us what kind of job it is and therefore who and how many nodes to connect to | 21:33 |
| peluse | yes | 21:34 |
| clayg | what does the suffix have to do with it - it should always be suffixES | 21:34 |
| clayg | as in - all the suffixes in this part that will meet the critiera for this job (i.e. it has some of these fragment indexes in it) | 21:34 |
| acoles | peluse: was that 'yes' to me or clayg ? :) | 21:34 |
| peluse | acoles to you | 21:35 |
| peluse | clayg, not even sure that's used anywhere, might be left over junk | 21:35 |
| clayg | peluse: oic | 21:36 |
| *** jamielennox|away is now known as jamielennox | 21:36 | |
| peluse | yeah, I'm not seeing it used anywhere.... | 21:36 |
| *** annegentle has joined #openstack-swift | 21:36 | |
| peluse | recall this job processor changed just a few times :) | 21:36 |
| clayg | peluse: yeah i don't understand the cardinality of the jobs - these tests that run over the canned and shared crazy set of all kind of objects all up in the setUp is going to be impossible to maintain | 21:37 |
| clayg | like anytime we want to test a new sceanario the most ovbious place to add it is to the shared setup - but then *every* test in the case has to change | 21:38 |
| clayg | I'm going to start ripping that testcase apart | 21:38 |
| peluse | clayg, yeah, it turned turn out to be very maintainable. sorry 'bout that | 21:39 |
| clayg | peluse: s/very/not very/ ? | 21:40 |
| acoles | peluse: ok. cool. i got part way to fixing that and testing in patch 169052, was going to carry on with it now ie. cleaning up where ssync_sender updates removes delete_list items so that durable delete depends on whether there is another frag archive | 21:40 |
| patchbot | acoles: https://review.openstack.org/#/c/169052/ | 21:40 |
| peluse | yeah, not... my brain is pretty fried today | 21:40 |
| clayg | peluse: no problem | 21:40 |
| *** jamielennox is now known as jamielennox|away | 21:46 | |
| *** jamielennox|away is now known as jamielennox | 21:53 | |
| mattoliverau | Morning | 21:54 |
| acoles | mattoliverau: hi | 21:55 |
| mattoliverau | acoles your still here! Yay overlap :) | 21:55 |
| *** teploxoh has quit IRC | 22:05 | |
| *** Gues_____ has quit IRC | 22:13 | |
| *** Gues_____ has joined #openstack-swift | 22:14 | |
| *** annegentle has quit IRC | 22:19 | |
| jrichli | mattoliverau acoles: hi! Yes, acoles is working late!! Ok, I am out ... | 22:21 |
| *** jrichli has quit IRC | 22:21 | |
| *** tsg has quit IRC | 22:24 | |
| *** tsg has joined #openstack-swift | 22:31 | |
| *** panbalag has quit IRC | 22:39 | |
| *** annegentle has joined #openstack-swift | 22:41 | |
| *** panbalag has joined #openstack-swift | 22:44 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: support multiple ranges for GET requests https://review.openstack.org/166576 | 22:46 |
| briancline | does anyone remember (maybe in the last 11-12 months) any bugs/changesets related to json unicode warnings emitted from DatabaseBroker.update_metadata? | 22:46 |
| notmyname | briancline: ya, something about unicode rings a bell | 22:47 |
| briancline | if memory serves the ones I saw were ultimately raised from within simplejson | 22:48 |
| briancline | came up during some upgrade testing on account/container servers | 22:48 |
| briancline | I might've been using an older tag.. I'll try against the latest to double-check | 22:49 |
| notmyname | briancline: I72b322e7513f7da32e8dc75c6bf0e7e016948c88 and Id2ac129828dbdf55b609d839ce4d9d42437ee0a3 and I82359c87a0919d71693e49ecf08e0f1eedc9d18e and I82359c87a0919d71693e49ecf08e0f1eedc9d18e | 22:49 |
| briancline | and to capture output.. it got lost somewhere in my 5 billion open terminals | 22:49 |
| briancline | ah | 22:50 |
| notmyname | briancline: those are the gerrit change ids. search for them on review.openstack.org | 22:50 |
| notmyname | briancline: that's from the last year that has "unicode" in the subject | 22:50 |
| briancline | notmyname: aha, this sounds precisely like what I saw | 22:51 |
| briancline | notmyname: thanks for grabbing those | 22:51 |
| briancline | based on those dates I was definitely on an older tag | 22:54 |
| *** annegentle has quit IRC | 22:55 | |
| *** panbalag has quit IRC | 22:55 | |
| clayg | wasn't there something with ipv6 that made us grow an is_local_device function somewhere? | 23:12 |
| *** tsg has quit IRC | 23:15 | |
| clayg | nm, i found it - from swift.common.ring.utils import is_local_device | 23:16 |
| zaitcev | yes, ho did it | 23:20 |
| zaitcev | right | 23:20 |
| *** petertr7 has quit IRC | 23:21 | |
| *** tsg has joined #openstack-swift | 23:22 | |
| *** logan2 has quit IRC | 23:25 | |
| *** mattoliverau has quit IRC | 23:29 | |
| *** mattoliverau has joined #openstack-swift | 23:29 | |
| *** ChanServ sets mode: +v mattoliverau | 23:30 | |
| notmyname | FYI https://wiki.openstack.org/wiki/Swift/version_map | 23:31 |
| *** ho has joined #openstack-swift | 23:31 | |
| *** logan2 has joined #openstack-swift | 23:33 | |
| mattoliverau | ok, I'm dangerously under caffinated.. think it's time for a coffee run, bbs | 23:38 |
| acoles | mattoliverau: me too! | 23:38 |
| mattoliverau | acoles: your still here, are you attempting to beat the clayg record of lack of sleep? | 23:39 |
| acoles | mattoliverau: no chance :P | 23:40 |
| *** Nadeem_ has joined #openstack-swift | 23:43 | |
| ho | good morning guys! | 23:46 |
| *** theanalyst has quit IRC | 23:46 | |
| mattoliverau | ho: Morning | 23:46 |
| * mattoliverau seems to failing at this coffee thing | 23:46 | |
| ho | mattoliverau: morning! | 23:47 |
| *** km has joined #openstack-swift | 23:50 | |
| *** theanalyst has joined #openstack-swift | 23:52 | |
| torgomatic | so is the K release going to be the last one where we support python 2.6? | 23:52 |
| *** Tahmina has joined #openstack-swift | 23:52 | |
| notmyname | torgomatic: IOW you can stop writing py26 code in about 3 weeks? | 23:53 |
| torgomatic | notmyname: can I start the long, tedious process of extracting simplejson's tentacles from our codebase | 23:53 |
| torgomatic | we'll see how many Unicode-related bugs I can uncover as I go | 23:54 |
| *** jrichli has joined #openstack-swift | 23:54 | |
| notmyname | lucid support dies in a few weeks, right? what about cent/rhel? what's reasonable to expect there? | 23:54 |
| notmyname | ie LTS distro releases that have/had py26 | 23:54 |
| *** shakamunyi has quit IRC | 23:55 | |
| torgomatic | yeah, Lucid has at most until the end of April for support | 23:56 |
| *** annegentle has joined #openstack-swift | 23:56 | |
| ho | What is the highest priority review today? (may be the patch 169035?) | 23:59 |
| patchbot | ho: https://review.openstack.org/#/c/169035/ | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!