| *** sandywalsh_ has joined #openstack-swift | 03:34 | |
| *** sandywalsh has quit IRC | 03:34 | |
| *** Gues_____ has quit IRC | 03:57 | |
| openstackgerrit | Monty Taylor proposed openstack/python-swiftclient: Add socket-level timeout parameter https://review.openstack.org/168673 | 05:42 |
|---|---|---|
| *** bkopilov has joined #openstack-swift | 07:48 | |
| *** chlong has joined #openstack-swift | 08:38 | |
| *** bkopilov has quit IRC | 12:41 | |
| *** EmilienM|PTO is now known as EmilienM | 12:42 | |
| *** silor has joined #openstack-swift | 13:24 | |
| *** silor1 has joined #openstack-swift | 13:45 | |
| *** silor has quit IRC | 13:48 | |
| *** silor has joined #openstack-swift | 14:21 | |
| *** silor1 has quit IRC | 14:25 | |
| *** dmsimard has quit IRC | 14:43 | |
| *** dmsimard has joined #openstack-swift | 14:45 | |
| *** gsilvis has quit IRC | 15:00 | |
| *** gsilvis has joined #openstack-swift | 15:01 | |
| openstackgerrit | Monty Taylor proposed openstack/python-swiftclient: Add socket-level timeout parameter https://review.openstack.org/168673 | 15:04 |
| *** bkopilov has joined #openstack-swift | 16:26 | |
| *** khivin has joined #openstack-swift | 16:54 | |
| *** sandywalsh_ has quit IRC | 17:03 | |
| *** panbalag has joined #openstack-swift | 17:04 | |
| *** sandywalsh has joined #openstack-swift | 17:05 | |
| *** panbalag has quit IRC | 17:38 | |
| *** zaitcev has joined #openstack-swift | 18:02 | |
| *** ChanServ sets mode: +v zaitcev | 18:02 | |
| *** chlong has quit IRC | 18:15 | |
| *** panbalag has joined #openstack-swift | 18:37 | |
| *** panbalag has joined #openstack-swift | 18:39 | |
| *** mmcardle has joined #openstack-swift | 18:39 | |
| *** panbalag has quit IRC | 18:40 | |
| *** barra204 has quit IRC | 19:04 | |
| *** shakamunyi has quit IRC | 19:04 | |
| *** silor has quit IRC | 19:12 | |
| *** david-ly_ has joined #openstack-swift | 19:21 | |
| *** david-lyle has quit IRC | 19:23 | |
| clayg | hey guys | 20:04 |
| clayg | torgomatic: i thought we stopped getting those weird error limit things when we took the counts out of the rhings | 20:05 |
| clayg | maybe... we have some of the real socket tests that re-use the application between tests doing some failure injection now? | 20:05 |
| clayg | birthday! | 20:06 |
| clayg | noone ever rebases *my* patches :'( | 20:06 |
| zaitcev | maybe they respect your authorship too much | 20:09 |
| clayg | zaitcev: ! hi! | 20:11 |
| *** Gu_______ has joined #openstack-swift | 20:11 | |
| clayg | ok, so some stuff got merged - that's good - i'll try to audit what we have outstanding and make sure it's all mergable | 20:12 |
| zaitcev | what backend do we use under liberasurecode, realistically speaking? jerasure, I presume? | 20:13 |
| *** Gu_______ has quit IRC | 20:17 | |
| *** Gu_______ has joined #openstack-swift | 20:19 | |
| *** jamielennox is now known as jamielennox|away | 20:58 | |
| mattoliverau | Morning | 21:01 |
| notmyname | clayg: mattoliverau: I'm going to eat something, then I'll be looking at what's outstanding, too | 21:37 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fix AttributeError and logging in ECAppIter https://review.openstack.org/168543 | 21:43 |
| clayg | what happend in bug #1437870 | 21:46 |
| openstack | bug 1437870 in OpenStack Object Storage (swift) "python-swiftclient 2.4.0 broke WAL-E" [Undecided,New] https://launchpad.net/bugs/1437870 | 21:46 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Add Fragment Index filter support to ssync https://review.openstack.org/165188 | 21:52 |
| clayg | ^ missed it by *this* much! | 21:52 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: wip: ec reconstructor probe test https://review.openstack.org/164291 | 21:53 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Erasure Code Reconstructor https://review.openstack.org/131872 | 21:53 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Guard against invalid Policy https://review.openstack.org/168548 | 22:00 |
| clayg | mattoliverau: good catch on the storage_policy key leaking between tests - but I think it's a more general problem with the setup just attaching the mutable class attribute to each test methods's self.app directly instead of a fresh copy for each test | 22:01 |
| clayg | mattoliverau: I think fixing the setup to not do something so likely to cause problems is better than any amount of cleanup we could hope to achive in the indvidual test cases | 22:01 |
| clayg | good eye though - i was totally thinking each test got a fresh app with a fres container_info dict :\ | 22:02 |
| notmyname | here | 22:07 |
| clayg | notmyname: i rebased everyhting and now nothing is verfiedy for merge :P | 22:08 |
| clayg | notmyname: I suppose i'm mostly thinking that I'm going to start trying to pull apart the first few changes for the rebased merge chain | 22:08 |
| clayg | either that or review the reconstructor and see what I can do help peluse with a head start on Monday | 22:09 |
| clayg | ... probably both | 22:09 |
| clayg | ok take a break and go play with kids for awhile :P | 22:09 |
| notmyname | I'm more worried about the reconstructor than the patch chain at this point. maybe just because the reconstructor has gotta be in there | 22:10 |
| clayg | ok that's good direction - THANKS! | 22:11 |
| notmyname | clayg: you need an ec_review feature branch, right? is that what we decided last week? | 22:11 |
| clayg | yeah that's what we came up with in the end | 22:11 |
| notmyname | I'm getting my head around the patches outstanding right now, then I'm going to start looking at any chain leading to the reconstructor | 22:11 |
| clayg | but remember you shoudln't ask me - as cschwede - you and he were the only ones that seemed to remmeber correctly how it worked last time | 22:11 |
| notmyname | lol | 22:11 |
| notmyname | ok, I'll get that in the morning (the feature branch) from -infra or ttx | 22:12 |
| clayg | notmyname: yeah np, the first one in the chain up to the reconstructor is pretty great obvious cleanup - should be easy | 22:12 |
| clayg | the ssync frag index support may need more tests - i need a critical review on it | 22:12 |
| clayg | the reconstructor change is great - but probably isn't in it's final form TBH | 22:12 |
| notmyname | -2! (there. that's critical) | 22:12 |
| clayg | well - you have to have constructive critisim too :P | 22:13 |
| clayg | it may be fine - i haven't really given it a "review" perspective - and I don't think i'm hte only one who's been working on it | 22:13 |
| *** jamielennox|away is now known as jamielennox | 22:20 | |
| notmyname | torgomatic: I don't know if this uses multi-range requests or just single ranges, but it seems possible that it uses multi-ranges. http://www.flussonic.com/forums/content.php?155-Streaming-files-from-HTTP-storage | 22:21 |
| notmyname | torgomatic: ...in answer to your question about "who cares about multi range anyway" | 22:21 |
| mattoliverau | clayg: yeah fair enough (re: setup refreshing object rather then copying mutable). Sorry Monday morning email, meeting boring stuff.. Back to proper work now. | 22:22 |
| mattoliverau | (Well until an 11 am meeting) | 22:22 |
| notmyname | ok, starred patches cleaned up | 22:31 |
| notmyname | ultimately, I'd love to see https://review.openstack.org/#/c/164291/ land in the beta (ec probe test) | 22:31 |
| clayg | pepAAAAAAAATE! | 22:31 |
| clayg | i *litterally* exercise that code in tests - how is that a namerror? | 22:32 |
| notmyname | I'm unsure about https://review.openstack.org/#/c/168254/. the commit message sounds like a pretty important piece of functionality | 22:33 |
| clayg | oh heh - nope not *that* line :P | 22:33 |
| clayg | notmyname: no, i don't think so | 22:34 |
| notmyname | ack | 22:34 |
| clayg | I think yaun may not realize how much error handling torgomatic managed to get into the proxy (he should look at test.unit.proxy.controller.test_obj) - or I'm drastically misunderstanding what he's trying to accomplish with that patch | 22:35 |
| notmyname | ok. that was my confusion. I thought it had been handled already by torgomatic | 22:35 |
| notmyname | I'll ignore it until torgomatic is around to comment on it | 22:36 |
| clayg | idk, we might have to let tsg or kevin what are the tradeoffs/advantages of feeding decode k or k + m fragments | 22:38 |
| notmyname | that sounds like a wonderful thing to suss out in a beta ;-) | 22:38 |
| clayg | I think that if you can connect to the first k nodes there's absoluately zero value in waiting for the other m nodes to hand you the parity bits - because decode is going to throw them away | 22:38 |
| clayg | otoh if you have a say 8 + 6 schema or something and only 7 data nodes come in - there may be some significant advantage to having all of the parity bits available instead of just one? | 22:39 |
| notmyname | it might be a more generic thing with other EC stuff. eg I know kota has their own NTT lib that they'll be using. not sure if it's same algo with faster math like isa-l or if it's whole new algos, though | 22:39 |
| notmyname | s/stuff/encodings/ | 22:39 |
| clayg | notmyname: if it's going to be that specific (per lib) we may just make the the ec_ndata configurable and idk - validate it's great than k or something? | 22:40 |
| *** Gu_______ has quit IRC | 23:03 | |
| mattoliverau | clayg: I wouldn't have thought that wrapping the container_info dict in a dict() would fix what was happening here.. nice one. | 23:23 |
| *** kei_yama has joined #openstack-swift | 23:24 | |
| *** annegentle has joined #openstack-swift | 23:26 | |
| *** jrichli has joined #openstack-swift | 23:32 | |
| clayg | mattoliverau: well it creates a new dict with the same keys and values - which is mostly good enough to prevent anything a test does to self.app.container_info from manipulating the state in the next test run - *technially* I probably should deepcopy or whatever it's called - i can never remember | 23:33 |
| clayg | right now I'm being all bitter about functools.partialmethod having only been added in python3.4 | 23:33 |
| mattoliverau | clayg: meh, it worked! jenkins likes it, and it passed on my local machine, so it does enough :) | 23:33 |
| clayg | bha - what a, it's a freaking property descriptor (of course) - i don't want to backport all tha tnoise | 23:34 |
| notmyname | https://review.openstack.org/#/c/168076/ landing | 23:35 |
| mattoliverau | clayg: it's a weekend and getting later in the day.. that means it's not frowned apon to start drinking :P | 23:35 |
| notmyname | ec is there a list of stuff for probe tests? | 23:43 |
| notmyname | ie a list of things in ec that need probe tests? | 23:43 |
| notmyname | s/ec/clayg | 23:43 |
| notmyname | wait wait. let me start over | 23:44 |
| notmyname | clayg: is there a list of stuff in ec that still needs probetests? | 23:44 |
| clayg | pshshshhs not | 23:53 |
| clayg | the cally me ec | 23:54 |
| clayg | bah fixing tests is *so* borning | 23:56 |
| openstackgerrit | John Dickinson proposed openstack/swift: Guard against invalid Policy https://review.openstack.org/168548 | 23:57 |
| mattoliverau | off to a meeting, bbs | 23:57 |
| notmyname | ^^ fixed the docstring nit and approved | 23:59 |
| clayg | yeah looks good! | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!