| *** kei_yama has quit IRC | 00:01 | |
| *** kei_yama_ has joined #openstack-swift | 00:01 | |
| clayg | torgomatic: oh sorry - didn't notice you already had domain sockets wip'd | 00:01 |
|---|---|---|
| stewie925 | hey clayg - after applying the patch the failure is now down to just the test_delete_policy_override_params | 00:07 |
| stewie925 | and I see you mentioned youre not sure how the 'test_delete_policy_override_params' is related | 00:11 |
| clayg | stewie925: no, i see it now - that was my rant on the ValueError handling in obj/replicatory.py being poorly organized | 00:13 |
| clayg | but basically that test is causing the replicator to call get_part_nodes with 99 and it's blowing up "handling it" and leaving the part around | 00:13 |
| clayg | there might be other tests in there that hit the same code path but aren't triggered to assert on the behavior | 00:14 |
| clayg | I think you need to change your code to raise a different exception that's not a ValueError - you'll get a bunch more errors - you should be able to mostly track them down to that code path in the replicator (or something similar in the other consistency daemons) | 00:14 |
| clayg | in most cases they're probably not code paths that are mostly just tickled by tests throwing garbage at the fake rings - but each one is an oppertunity to think about how the real code should really handle this condition when it sees it in the wild | 00:15 |
| clayg | you could even just grep around for everywhere that we call get_part_nodes and make sure they're prepared to handle and log the ValueError - then decide where the code goes from there - return an error to a client? skip the part and go onto the next one? depends on the context | 00:16 |
| stewie925 | hi clayg, yeah I am looking for a list of errors that may be used for assertRaises | 00:16 |
| clayg | then any tests that are hitting that path can either be updated to expect the new behavior (garbage in garbage out) or updated to provide parts that are reasonable/valid for the test rings they're setting up | 00:16 |
| clayg | assertRaises doesn't care what exception you give it? | 00:17 |
| *** diogogmt has joined #openstack-swift | 00:17 | |
| stewie925 | am hoping to get a list to see which type of error that I can replace the valueerror with | 00:18 |
| stewie925 | also, I assume we need to change this in both ring.py as well as test_ring.py | 00:19 |
| clayg | just do a plane Exception for now | 00:23 |
| clayg | plain? | 00:24 |
| stewie925 | oh ok, will do - thanks | 00:24 |
| stewie925 | ! | 00:24 |
| *** NM has left #openstack-swift | 00:35 | |
| stewie925 | hi clayg - I ve changed the assertRaises(ValueError...) to assertRaises(Exception...) - here are the test results http://paste.openstack.org/show/507059/ | 00:37 |
| *** suyash has quit IRC | 00:38 | |
| stewie925 | and thank you very much for your help so far, clayg - will try to get in touch with you tomorrow | 00:38 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Suppress InsecureRequestWarning when using --insecure https://review.openstack.org/324148 | 00:38 |
| clayg | sorta sucks to unittest behavior of external dependencies don't it :\ | 00:42 |
| stewie925 | :\ | 00:42 |
| clayg | code change - call library like this - unit test - validate called libaary like this - #WOMM | 00:42 |
| stewie925 | i apologize can you elaborate the above? | 00:44 |
| clayg | stewie925: i was talking to tb | 00:44 |
| stewie925 | oops sorry | 00:44 |
| clayg | stewie925: np | 00:45 |
| stewie925 | well, thank you so much again and good night | 00:45 |
| clayg | don't know unless you ask - it's np | 00:45 |
| clayg | stewie925: yeah - thanks for working on that one - sorry to setup the bomb on you - i'm sure you'll get it all squared if you stick with it! keep asking questions! | 00:45 |
| * stewie925 waves | 00:46 | |
| stewie925 | will do, and appreciate your helpfulness | 00:46 |
| *** stewie925 has quit IRC | 00:47 | |
| timburke | clayg: yeah, it's not great. supposedly it's the way to do it, though | 00:52 |
| clayg | lol | 00:52 |
| clayg | oh you ment supposedly the way to disable the warning :P | 00:52 |
| clayg | i ment supposedly the way to unittest interactions with external dependencies - which is also true as I've come to understand it! | 00:53 |
| clayg | yeah I mean the only way to know if that code works is to get a self signed cert going on your dev saio and test it out | 00:53 |
| timburke | hehe also true | 00:53 |
| clayg | on the plus side if I review it there's a fair chance vsaio will grow a ssl=true option :P | 00:53 |
| clayg | which i might want to do to test torgomatic's domain socket thing anyway - so maybe it's wins all around | 00:54 |
| timburke | clayg: if it does, it might be preferable to set OS_CACERT | 00:54 |
| clayg | unless jrichli gets wind of any of this - then I'll be in trouble for not reviewing crypto :\ | 00:54 |
| timburke | clayg: you don't want to make pandas cry, do you?? | 00:55 |
| clayg | timburke: is cacert plumbed into the swift endpoint/storage-url or is it just for auth? | 00:57 |
| *** elambert_ has quit IRC | 00:57 | |
| timburke | if it's just auth, we need to fix it | 00:57 |
| timburke | clayg: looks like it's good? https://github.com/openstack/python-swiftclient/blob/master/swiftclient/client.py#L385 | 00:58 |
| clayg | yeah I think i went to the original introduction and didn't realize it got improved domewhere along the line -> https://review.openstack.org/#/c/18393/2/swiftclient/client.py | 00:59 |
| patchbot | clayg: patch 18393 - python-swiftclient - Add --os-cacert (MERGED) | 00:59 |
| clayg | timburke: any idea what merged that blew up the copy command change? :'( | 01:03 |
| clayg | timburke: swiftclient of course | 01:03 |
| timburke | clayg: ima guess https://review.openstack.org/#/c/323975/ | 01:06 |
| patchbot | timburke: patch 323975 - python-swiftclient - Fixed a typo in SwiftService API doc (MERGED) | 01:06 |
| clayg | joel - i used to think docs were good - now i hate them | 01:06 |
| * clayg thinks that didn't take much :\ | 01:07 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: boolean logic cleanup in service.Swift[Copy|Post]Object https://review.openstack.org/321808 | 01:08 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Add copy object method https://review.openstack.org/280200 | 01:08 |
| timburke | clayg: better? | 01:08 |
| clayg | timburke: did you use the rebase buttom!? | 01:08 |
| timburke | no, just did it on my machine | 01:09 |
| clayg | timburke: whoa - and the dependent patches too | 01:09 |
| clayg | joel ok, docs can stay - apparently we have tburke-bot for this kind of thing | 01:09 |
| timburke | well, one, anyway. was there another? | 01:09 |
| clayg | timburke: oh yeah, but everyone hates patch 321815 anyway | 01:09 |
| patchbot | clayg: https://review.openstack.org/#/c/321815/ - python-swiftclient - Use requests CaseInsensitiveDict in swiftclient.co... | 01:09 |
| timburke | i dunno that we *hate* it... | 01:10 |
| clayg | oic, hate *me* | 01:11 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Use requests CaseInsensitiveDict in swiftclient.copy_object https://review.openstack.org/321815 | 01:14 |
| *** klrmn has quit IRC | 01:24 | |
| *** sheel has joined #openstack-swift | 01:28 | |
| *** amit213 has joined #openstack-swift | 01:39 | |
| *** amit213 has quit IRC | 01:41 | |
| *** amit213 has joined #openstack-swift | 01:44 | |
| *** kei_yama_ has quit IRC | 01:45 | |
| *** kei_yama has joined #openstack-swift | 01:45 | |
| *** bill_az has quit IRC | 01:47 | |
| *** nadeem has joined #openstack-swift | 01:55 | |
| *** kei_yama has quit IRC | 01:56 | |
| *** klamath has quit IRC | 01:57 | |
| *** klamath has joined #openstack-swift | 01:57 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Suppress InsecureRequestWarning when using --insecure https://review.openstack.org/324148 | 01:57 |
| *** nadeem has quit IRC | 01:58 | |
| *** tanee is now known as tanee_away | 01:59 | |
| *** tanee_away is now known as tanee | 02:00 | |
| *** kei_yama has joined #openstack-swift | 02:03 | |
| kota_ | back to online again at office | 02:11 |
| *** nadeem has joined #openstack-swift | 02:19 | |
| *** klrmn has joined #openstack-swift | 02:21 | |
| *** nadeem has quit IRC | 02:26 | |
| *** dmorita has quit IRC | 02:26 | |
| *** StraubTW has joined #openstack-swift | 02:49 | |
| *** diogogmt has quit IRC | 03:10 | |
| *** StraubTW has quit IRC | 03:20 | |
| *** links has joined #openstack-swift | 03:35 | |
| *** zaitcev has quit IRC | 03:40 | |
| *** elambert_ has joined #openstack-swift | 03:43 | |
| *** nadeem has joined #openstack-swift | 03:47 | |
| *** natarej_ has quit IRC | 03:48 | |
| *** natarej_ has joined #openstack-swift | 03:49 | |
| *** nadeem has quit IRC | 03:54 | |
| *** arch-nemesis has joined #openstack-swift | 03:59 | |
| openstackgerrit | Merged openstack/swift: Adding sorting_method to admin_guide https://review.openstack.org/324097 | 04:01 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-swiftclient: Updated from global requirements https://review.openstack.org/89250 | 04:09 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 04:10 |
| *** sams-gleb has joined #openstack-swift | 04:13 | |
| *** sams-gleb has quit IRC | 04:17 | |
| *** patchbot has quit IRC | 04:22 | |
| *** patchbot has joined #openstack-swift | 04:22 | |
| *** mlanner has quit IRC | 04:23 | |
| *** ctennis has quit IRC | 04:23 | |
| *** Lickitysplitted has quit IRC | 04:23 | |
| *** Lickitysplitted has joined #openstack-swift | 04:23 | |
| *** swifterdarrell has quit IRC | 04:24 | |
| *** dmellado has quit IRC | 04:24 | |
| *** cschwede has quit IRC | 04:24 | |
| *** mmotiani_ has quit IRC | 04:24 | |
| *** cschwede has joined #openstack-swift | 04:25 | |
| *** ctennis has joined #openstack-swift | 04:25 | |
| *** swifterdarrell has joined #openstack-swift | 04:25 | |
| *** ChanServ sets mode: +v swifterdarrell | 04:25 | |
| *** mlanner has joined #openstack-swift | 04:26 | |
| *** dmorita has joined #openstack-swift | 04:26 | |
| *** mmotiani_ has joined #openstack-swift | 04:26 | |
| *** dmellado has joined #openstack-swift | 04:27 | |
| *** dmorita has quit IRC | 04:31 | |
| *** nadeem has joined #openstack-swift | 04:36 | |
| *** psachin has joined #openstack-swift | 04:47 | |
| *** jamielennox is now known as jamielennox|away | 04:48 | |
| *** nadeem has quit IRC | 04:57 | |
| openstackgerrit | Merged openstack/swift: Fix swift-dispersion in multi-region setups https://review.openstack.org/322820 | 04:57 |
| *** nadeem has joined #openstack-swift | 04:58 | |
| *** furlongm_ is now known as furlongm | 04:59 | |
| *** jamielennox|away is now known as jamielennox | 05:04 | |
| *** ChubYann has quit IRC | 05:19 | |
| *** hosanai has joined #openstack-swift | 05:23 | |
| *** ChanServ sets mode: +v hosanai | 05:23 | |
| *** janonymous has quit IRC | 05:26 | |
| *** ppai has joined #openstack-swift | 05:26 | |
| *** cguadall_ has joined #openstack-swift | 05:27 | |
| *** cguadall has quit IRC | 05:30 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:31 | |
| *** delatte has quit IRC | 05:44 | |
| *** rcernin has joined #openstack-swift | 05:48 | |
| *** mmcardle has joined #openstack-swift | 05:51 | |
| *** wer has joined #openstack-swift | 05:51 | |
| *** greghaynes has quit IRC | 05:58 | |
| *** jamielennox is now known as jamielennox|away | 06:01 | |
| *** cdelatte has joined #openstack-swift | 06:03 | |
| *** greghaynes has joined #openstack-swift | 06:10 | |
| *** sams-gleb has joined #openstack-swift | 06:14 | |
| *** jamielennox|away is now known as jamielennox | 06:16 | |
| *** sams-gleb has quit IRC | 06:19 | |
| *** arch-nemesis has quit IRC | 06:20 | |
| *** zul has quit IRC | 06:21 | |
| *** mmcardle has quit IRC | 06:22 | |
| *** arch-nemesis has joined #openstack-swift | 06:35 | |
| *** zul has joined #openstack-swift | 06:36 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Imported Translations from Zanata https://review.openstack.org/324234 | 06:41 |
| *** ChanServ sets mode: +v cschwede | 06:51 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Add Pros/Cons table for global clusetr consideration https://review.openstack.org/304382 | 06:51 |
| *** mvk_ has quit IRC | 06:52 | |
| *** tesseract- has joined #openstack-swift | 06:53 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: pids in probe is no longer used https://review.openstack.org/324246 | 06:55 |
| *** sams-gleb has joined #openstack-swift | 06:55 | |
| *** SkyRocknRoll has quit IRC | 07:05 | |
| *** nadeem has quit IRC | 07:05 | |
| *** nadeem has joined #openstack-swift | 07:12 | |
| *** rledisez has joined #openstack-swift | 07:17 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:22 | |
| *** fbo has joined #openstack-swift | 07:27 | |
| kota_ | I noticed, launchpad says "mitaka series is the current focus of development", interesting. | 07:33 |
| *** hoonetorg has quit IRC | 07:38 | |
| *** janonymous has joined #openstack-swift | 07:40 | |
| *** elambert_ has quit IRC | 07:43 | |
| *** cguadall_ has quit IRC | 07:46 | |
| *** klrmn has quit IRC | 07:48 | |
| *** hoonetorg has joined #openstack-swift | 07:50 | |
| *** anticw_ has joined #openstack-swift | 07:55 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto combine crypto.py with crypto_utils.py https://review.openstack.org/323978 | 07:56 |
| *** acoles_ is now known as acoles | 07:56 | |
| acoles | clayg: cheer up, we *love* you and your patches :D | 07:58 |
| *** SkyRocknRoll has quit IRC | 08:01 | |
| acoles | clayg: notmyname huh, I see i just strayed out of my quadrant with my use of the word "love" there, it did feel kindof unnatural | 08:02 |
| *** SkyRocknRoll has joined #openstack-swift | 08:08 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/swift: [WIP] Testing latest u-c https://review.openstack.org/318441 | 08:10 |
| *** nadeem has quit IRC | 08:21 | |
| *** jmccarthy has quit IRC | 08:31 | |
| *** janonymous has quit IRC | 08:32 | |
| *** jmccarthy has joined #openstack-swift | 08:32 | |
| *** arch-nemesis has quit IRC | 08:45 | |
| *** psachin has quit IRC | 08:45 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto - add tests for override headers https://review.openstack.org/324299 | 08:47 |
| *** arch-nemesis has joined #openstack-swift | 08:48 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto - add tests for override headers https://review.openstack.org/324299 | 08:51 |
| *** mmcardle has joined #openstack-swift | 08:54 | |
| *** psachin has joined #openstack-swift | 09:01 | |
| *** psachin has quit IRC | 09:01 | |
| *** psachin has joined #openstack-swift | 09:01 | |
| *** mmcardle has quit IRC | 09:02 | |
| *** mmcardle has joined #openstack-swift | 09:09 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto - refactor footers callback https://review.openstack.org/324313 | 09:10 |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto - catch errors when decrypting update override sysmeta (WIP) https://review.openstack.org/321638 | 09:19 |
| *** arch-nemesis has quit IRC | 09:22 | |
| *** dmorita has joined #openstack-swift | 09:27 | |
| *** dmorita has quit IRC | 09:32 | |
| *** SkyRocknRoll has quit IRC | 09:33 | |
| *** SkyRocknRoll has joined #openstack-swift | 09:49 | |
| *** mmcardle has quit IRC | 10:01 | |
| *** mmcardle has joined #openstack-swift | 10:02 | |
| *** mvk_ has joined #openstack-swift | 10:41 | |
| *** hosanai has quit IRC | 10:45 | |
| openstackgerrit | Merged openstack/swift: Imported Translations from Zanata https://review.openstack.org/324234 | 10:47 |
| openstackgerrit | Merged openstack/swift: crypto combine crypto.py with crypto_utils.py https://review.openstack.org/323978 | 10:49 |
| openstackgerrit | Merged openstack/swift: crypto - add tests for override headers https://review.openstack.org/324299 | 10:49 |
| *** mmuffinman has quit IRC | 11:00 | |
| *** mmuffinman has joined #openstack-swift | 11:00 | |
| *** psachin has quit IRC | 11:06 | |
| *** mmcardle has quit IRC | 11:19 | |
| *** mmcardle has joined #openstack-swift | 11:20 | |
| *** raildo-afk is now known as raildo | 11:53 | |
| *** furlongm has quit IRC | 11:56 | |
| *** furlongm has joined #openstack-swift | 11:56 | |
| acoles | notmyname: please consider patch 324392 backport to liberty | 12:04 |
| patchbot | acoles: https://review.openstack.org/#/c/324392/ - swift (stable/liberty) - Put correct Etag and Accept-Ranges in EC 304 and 4... | 12:04 |
| *** psachin has joined #openstack-swift | 12:11 | |
| *** hoonetorg has quit IRC | 12:12 | |
| *** sams-gleb has quit IRC | 12:18 | |
| tdasilva | acoles: haha, too funny...keep sharing the love | 12:23 |
| *** hoonetorg has joined #openstack-swift | 12:27 | |
| mahatic | acoles: " so I am going to be bold and go ahead with approving." on 323978. good decision! :) | 12:32 |
| *** mmcardle has joined #openstack-swift | 12:34 | |
| *** sams-gleb has joined #openstack-swift | 12:35 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto - remove default root secret https://review.openstack.org/318208 | 12:52 |
| *** sams-gleb has quit IRC | 12:54 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto - stop using key id to derive GET path keys https://review.openstack.org/318148 | 12:55 |
| *** NM has joined #openstack-swift | 12:56 | |
| *** NM has quit IRC | 12:56 | |
| *** NM has joined #openstack-swift | 12:57 | |
| *** kei_yama has quit IRC | 12:57 | |
| *** _JZ_ has joined #openstack-swift | 13:02 | |
| *** ppai has quit IRC | 13:02 | |
| *** sams-gleb has joined #openstack-swift | 13:08 | |
| *** dfg_ has quit IRC | 13:13 | |
| *** klamath has quit IRC | 13:15 | |
| *** klamath has joined #openstack-swift | 13:16 | |
| *** pauloewerton has joined #openstack-swift | 13:22 | |
| *** diogogmt has joined #openstack-swift | 13:22 | |
| *** SkyRocknRoll has quit IRC | 13:33 | |
| *** bill_az has joined #openstack-swift | 13:37 | |
| *** SkyRocknRoll has joined #openstack-swift | 13:45 | |
| *** mmuffinman has quit IRC | 13:49 | |
| *** ametts has joined #openstack-swift | 13:53 | |
| *** bill_az has quit IRC | 13:57 | |
| *** bill_az has joined #openstack-swift | 13:57 | |
| *** sams-gleb has quit IRC | 14:02 | |
| *** sams-gleb has joined #openstack-swift | 14:03 | |
| cschwede | notmyname: I submitted a patch for stable/libery, that we already backported to kilo but somehow missed for Liberty: https://review.openstack.org/#/c/324551/1 | 14:09 |
| patchbot | cschwede: patch 324551 - swift (stable/liberty) - Fix replicator intersection exception when sync da... | 14:09 |
| cschwede | s/libery/liberty/ | 14:09 |
| *** links has quit IRC | 14:12 | |
| *** diogogmt has quit IRC | 14:19 | |
| *** diogogmt has joined #openstack-swift | 14:20 | |
| *** SkyRocknRoll has quit IRC | 14:21 | |
| *** MikeG451 has joined #openstack-swift | 14:24 | |
| *** dmorita has joined #openstack-swift | 14:27 | |
| *** vint_bra has joined #openstack-swift | 14:28 | |
| *** mmcardle has quit IRC | 14:29 | |
| *** dmorita has quit IRC | 14:32 | |
| *** elambert_ has joined #openstack-swift | 14:33 | |
| *** mmcardle has joined #openstack-swift | 14:33 | |
| *** elambert_ has joined #openstack-swift | 14:35 | |
| *** elambert_ has quit IRC | 14:39 | |
| *** david-lyle has joined #openstack-swift | 14:44 | |
| *** diogogmt has quit IRC | 14:44 | |
| *** cdelatte has quit IRC | 14:47 | |
| *** thumpba has joined #openstack-swift | 14:49 | |
| *** sams-gleb has quit IRC | 14:49 | |
| *** adu has joined #openstack-swift | 14:50 | |
| openstackgerrit | Cheng Li proposed openstack/python-swiftclient: Add an option: disable etag check on downloads https://review.openstack.org/323150 | 14:54 |
| *** sams-gleb has joined #openstack-swift | 14:57 | |
| *** thumpba has quit IRC | 15:00 | |
| *** thumpba has joined #openstack-swift | 15:01 | |
| *** diogogmt has joined #openstack-swift | 15:04 | |
| *** thumpba has quit IRC | 15:05 | |
| *** arch-nemesis has joined #openstack-swift | 15:08 | |
| *** klrmn has joined #openstack-swift | 15:12 | |
| acoles | tdasilva: fyi patch 324607 (the bot failed to report it here) | 15:15 |
| patchbot | acoles: https://review.openstack.org/#/c/324607/ - swift - Use source etag when copying unless using Range he... | 15:15 |
| notmyname | good morning | 15:15 |
| notmyname | acoles: cschwede: ack on the liberty backports | 15:16 |
| acoles | notmyname: goo dmorning | 15:16 |
| *** dmorita has joined #openstack-swift | 15:16 | |
| notmyname | acoles: well this is going to be interesting for liberty backports http://logs.openstack.org/92/324392/1/check/gate-swift-tox-func-in-process-fast-post/71bba98/console.html#_2016-06-02_12_58_29_177 | 15:17 |
| *** thumpba has joined #openstack-swift | 15:17 | |
| acoles | notmyname: :( | 15:19 |
| notmyname | acoles: maybe an easy way to fix it is to land a change to tox.ini for liberty that defines func-in-process-fast-post exactly the same way as func (ie doesn't set env and just calls .functests) | 15:19 |
| tdasilva | acoles: ack, looking now | 15:19 |
| acoles | notmyname: yes, or maybe the same job will pass, i have run func tests with fast post omm for a long time now | 15:20 |
| acoles | notmyname: or i figure out the correct yaml for it to not be run on the stable branches | 15:20 |
| notmyname | patch 324611 | 15:23 |
| patchbot | notmyname: https://review.openstack.org/#/c/324611/ - swift (stable/liberty) - added func-in-process-fast-post to mirror func | 15:23 |
| notmyname | acoles: ^ | 15:23 |
| notmyname | patch set 2 fixes the spelling error :-) | 15:24 |
| notmyname | acoles: so let's see if the gate works for that, then land than one first? | 15:24 |
| acoles | notmyname: gets my vote. thanks | 15:25 |
| notmyname | tdasilva: email on the ML putting swiftonfile in a will-be-EOL'd list | 15:26 |
| notmyname | I think? the email is hard to read | 15:27 |
| notmyname | oh, maybe that's just to mark the kilo branch as EOL | 15:28 |
| acoles | timburke: you were +1 on version 6 of this https://review.openstack.org/#/c/318208 - comments addressed in v7 and just rebases since then, are you still +ve? | 15:29 |
| timburke | acoles: looking...probably... | 15:29 |
| timburke | acoles: still kinda weirds me out that we're raising the *type* ValueError instead of an *instance of* ValueError | 15:32 |
| acoles | timburke: its valid syntax - the raise creates an instance of the type | 15:34 |
| timburke | k | 15:35 |
| notmyname | https://review.openstack.org/#/c/323501/ landed! new client docs for everyone, now | 15:37 |
| patchbot | notmyname: patch 323501 - openstack-infra/project-config - Publish python-swiftclient docs on every patch (MERGED) | 15:37 |
| notmyname | I'm guessing the next client patch that lands should trigger that | 15:37 |
| tdasilva | notmyname: looking for email | 15:39 |
| notmyname | tdasilva: the -dev list. '[stable][all] Takking kilo-eol for "the world"' | 15:40 |
| acoles | timburke: thanks | 15:41 |
| *** psachin has quit IRC | 15:41 | |
| notmyname | tdasilva: *tagging | 15:42 |
| acoles | tdasilva: timburke when i cherry pick patch 324607 to feature/crypto the new func test for copy with a Range fails with incorrect etag in container listing. I strongly suspect that is because we copy across the etag override header from the source. | 15:43 |
| patchbot | acoles: https://review.openstack.org/#/c/324607/ - swift - Use source etag when copying unless using Range he... | 15:43 |
| tdasilva | notmyname: that email is confusing, swiftonfile shows up in list 2, which is described as: "A project is in list 2 that would like to opt *IN* to tagging/EOLing", so I'm assuming it won't be in the list to EOL | 15:43 |
| notmyname | tdasilva: yeah, definitely a confusing email | 15:43 |
| timburke | acoles: makes sense. i was just starting to look at that patch and make sense of what was going on | 15:44 |
| acoles | tdasilva: timburke since I don't expect us to stop copying ranges in the next 24 hours I am going to purge the override headers during the copy and hopefully fix things | 15:44 |
| tdasilva | notmyname: and openstack-ansible is waiting for swift??? | 15:44 |
| timburke | acoles: so should copy always drop the override, or only when doing a ranged GET? | 15:45 |
| acoles | timburke: took me a while to realise that sink_req.etag = source_resp.etag achieves absolutely nothing ;) | 15:45 |
| tdasilva | I wonder where that code came from :P | 15:46 |
| notmyname | tdasilva: that makes sense. since swift isn't EOL'd yet, it doesn't make sense to EOL the ansible-swift stuff either | 15:46 |
| timburke | yeah, thanks for calling out what was going on in the commit message. sure enough, there's only a _resp_etag_protperty | 15:46 |
| notmyname | tdasilva: AFAIK, ansible-swift comes from andymccr | 15:46 |
| acoles | tdasilva: I'm not being critical of anyone - it's a language feature that allows it to be written :) | 15:46 |
| tdasilva | notmyname: no no, sorry, that was a reply to acoles | 15:47 |
| tdasilva | acoles: haha | 15:47 |
| timburke | acoles: that's right, get back in your quadrant :P | 15:47 |
| tdasilva | lol | 15:48 |
| * tdasilva is going to grab some lunch and will be right back | 15:48 | |
| acoles | timburke: idk. I am inclined to keep it simple. well, to start with and get things passing the test. | 15:48 |
| acoles | timburke: tdasilva i think we should spend every Friday in one of our 3 alien quadrants | 15:49 |
| notmyname | acoles: :-) | 15:51 |
| notmyname | acoles: "last week I was expressive/non-confrontational. this week, I'll be unexpressive/confrontational" | 15:53 |
| acoles | notmyname: that's such a bad idea it makes me rage | 15:56 |
| acoles | oops, it's not Friday yet ;) | 15:56 |
| notmyname | lol | 15:57 |
| *** tesseract- has quit IRC | 15:59 | |
| *** rcernin has quit IRC | 16:01 | |
| *** mvk_ has quit IRC | 16:03 | |
| acoles | jrichli: could I beg you to look at patch 318148 again? | 16:06 |
| patchbot | acoles: https://review.openstack.org/#/c/318148/ - swift (feature/crypto) - crypto - stop using key id to derive GET path keys | 16:06 |
| *** gyee has joined #openstack-swift | 16:08 | |
| *** bill_az has quit IRC | 16:11 | |
| *** bill_az has joined #openstack-swift | 16:11 | |
| *** elambert_ has joined #openstack-swift | 16:14 | |
| tdasilva | acoles: problem is having lived in a different country for more than half of my life, i'm really not sure which quadrant I should be anymore... just something different everyday | 16:21 |
| *** david-lyle has quit IRC | 16:24 | |
| *** rledisez has quit IRC | 16:25 | |
| *** tesseract has joined #openstack-swift | 16:44 | |
| jrichli | acoles: lol. you only have to ask! | 16:49 |
| *** tesseract has quit IRC | 16:51 | |
| *** sams-gleb has quit IRC | 16:52 | |
| *** bill_az has quit IRC | 16:55 | |
| *** suyash has joined #openstack-swift | 16:56 | |
| *** mmuffinman has joined #openstack-swift | 17:03 | |
| tdasilva | acoles: re patch 324607, func tests failed for me on my first run. resetting everything and running again... | 17:11 |
| patchbot | tdasilva: https://review.openstack.org/#/c/324607/ - swift - Use source etag when copying unless using Range he... | 17:11 |
| *** mvk_ has joined #openstack-swift | 17:14 | |
| *** dmorita_ has joined #openstack-swift | 17:14 | |
| notmyname | cschwede: what do you think about backporting https://review.openstack.org/#/c/303994/ ? | 17:16 |
| patchbot | notmyname: patch 303994 - swift - Don't delete misplaced dbs if not replicated (MERGED) | 17:16 |
| *** dmorita has quit IRC | 17:17 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: crypto - purge crypto sysmeta from responses https://review.openstack.org/320579 | 17:18 |
| jrichli | acoles: I took an initial look. I left a comment explaining the part I need to think more on before giving a score. But if you needed to move this forward today, then I can still catch up with you on it later. | 17:18 |
| acoles | timburke: tdasilva patch 320579 now selectively removes the override headers during ranged copies. the func test in patch 324607 will help verify but obviously that's not on feature/crypto. | 17:20 |
| patchbot | acoles: https://review.openstack.org/#/c/320579/ - swift (feature/crypto) - crypto - purge crypto sysmeta from responses | 17:20 |
| patchbot | acoles: https://review.openstack.org/#/c/324607/ - swift - Use source etag when copying unless using Range he... | 17:20 |
| tdasilva | acoles: looks like there's some issue with *lo in patch 324607 | 17:21 |
| patchbot | tdasilva: https://review.openstack.org/#/c/324607/ - swift - Use source etag when copying unless using Range he... | 17:21 |
| tdasilva | http://logs.openstack.org/07/324607/1/check/gate-swift-tox-func/9ce038c/console.html | 17:21 |
| *** mmcardle has quit IRC | 17:22 | |
| notmyname | working on the changelog. in this release, we have patches that fix the times when swift might delete data when it's not supposed to and patches that fix the times when swift wasn't deleting data when it's supposed to | 17:28 |
| tdasilva | acoles: so timburke was correct, the issue with actually assigning the etag to the correct header is that you now get a 422 from the object server when copying *LO objects | 17:31 |
| acoles | tdasilva: ok, well we learnt something then :) | 17:32 |
| timburke | notmyname: i've got a concern that in fixing bug 1582723, we may have created another one (specifically, that containers will report the wrong size for an object that was copied from an EC policy to a replicated one) | 17:32 |
| openstack | bug 1582723 in OpenStack Object Storage (swift) "EC object may have wrong size in container listing after POST" [High,Fix released] https://launchpad.net/bugs/1582723 - Assigned to Alistair Coles (alistair-coles) | 17:32 |
| timburke | will investigate further, but thought you might want to know for release-related reasons | 17:33 |
| notmyname | timburke: ok, let me know what you find | 17:33 |
| acoles | jrichli: ack. left reply on gerrit. | 17:33 |
| *** gyee has quit IRC | 17:40 | |
| acoles | timburke: are you seeing wrong size after the copy, or after a post to the copied object? | 17:41 |
| timburke | acoles: i haven't actually seen it yet; it was a thought experiment. i'm becoming more convinced that it's not a real problem, though. the override should be the client-facing size, so it should be the same for both replicated and EC | 17:42 |
| timburke | worst case, we get an override for an object that didn't technically need it | 17:43 |
| timburke | notmyname: ^^ i don't think it's actually a problem | 17:43 |
| clayg | looks like we're porting some good fixes back to liberty? anything particular prompting that? | 17:43 |
| acoles | timburke: ok, phew, that was how i saw it work out, but I do suspect that the x-object-sysmeta-ec-etag/size leaks from the source to the copy. they just happen to have the right values (I think). | 17:44 |
| notmyname | clayg: AFAICT, noticing the good fixes that have landed since we're kinda looking back at them now as part of a new release | 17:44 |
| *** nadeem has joined #openstack-swift | 17:44 | |
| timburke | also, couldn't hurt in getting that stable-branch checkbox on http://www.openstack.org/software/releases/mitaka/components/swift | 17:45 |
| acoles | timburke: but I'd prefer them not to leak across, if they do, cos it will be very confusing when looking at swift-get-info | 17:45 |
| notmyname | timburke: ack for the copy | 17:45 |
| *** nadeem has quit IRC | 17:46 | |
| *** nadeem has joined #openstack-swift | 17:47 | |
| acoles | clayg: QA people come to me and tell me tempest tests fail with EC | 17:47 |
| acoles | and I have been dragging my feet | 17:47 |
| *** thumpba has quit IRC | 17:48 | |
| timburke | acoles: yup. maybe something that ECObjectController._fix_response should clean up, like you're doing with the crypto meta? | 17:48 |
| * acoles afk | 17:48 | |
| acoles | timburke: exactly my thought | 17:48 |
| * acoles afk really | 17:49 | |
| *** dtoast has joined #openstack-swift | 17:51 | |
| *** sams-gleb has joined #openstack-swift | 17:53 | |
| *** thumpba has joined #openstack-swift | 17:54 | |
| *** links has joined #openstack-swift | 17:56 | |
| *** sams-gleb has quit IRC | 17:58 | |
| *** diogogmt has quit IRC | 18:03 | |
| *** diogogmt has joined #openstack-swift | 18:04 | |
| openstackgerrit | Mohit Motiani proposed openstack/python-swiftclient: Inaccurate info about read and write ACL syntax https://review.openstack.org/324739 | 18:04 |
| openstackgerrit | Shashirekha Gundur proposed openstack/swift: Add content-disposition header to tempurl request https://review.openstack.org/293907 | 18:05 |
| *** ChubYann has joined #openstack-swift | 18:05 | |
| openstackgerrit | John Dickinson proposed openstack/swift: AUTHORS/CHANGELOG updates for 2.8.0 release https://review.openstack.org/324742 | 18:06 |
| notmyname | for your consideration ^ | 18:06 |
| *** david-lyle has joined #openstack-swift | 18:09 | |
| *** diogogmt has quit IRC | 18:10 | |
| *** mvk has joined #openstack-swift | 18:10 | |
| *** mvk_ has quit IRC | 18:13 | |
| *** diogogmt has joined #openstack-swift | 18:13 | |
| *** dmorita has joined #openstack-swift | 18:14 | |
| *** dmorita_ has quit IRC | 18:14 | |
| notmyname | acoles: cschwede: ok, I'm landing https://review.openstack.org/#/c/324611/ now that the gate passed on it. please rebase your liberty backports on top of it | 18:21 |
| patchbot | notmyname: patch 324611 - swift (stable/liberty) - added func-in-process-fast-post to mirror func | 18:21 |
| acoles | jrichli: I get you now - I thought you meant passing key_id as an arg of the callback, rather than in response. It shoudl be in the response dict, I have done something bad on that patch while rebasing :/ | 18:21 |
| *** bill_az has joined #openstack-swift | 18:21 | |
| acoles | notmyname: ack | 18:21 |
| acoles | jrichli: I'll have to fix it tomorrow | 18:22 |
| *** zaitcev has joined #openstack-swift | 18:22 | |
| *** ChanServ sets mode: +v zaitcev | 18:22 | |
| acoles | jrichli: looks like I based v7 on v3 :/ | 18:24 |
| *** acoles is now known as acoles_ | 18:27 | |
| *** garthb has joined #openstack-swift | 18:34 | |
| *** nadeem has quit IRC | 18:34 | |
| *** david-lyle has quit IRC | 18:35 | |
| openstackgerrit | Merged openstack/swift: crypto - remove default root secret https://review.openstack.org/318208 | 18:39 |
| *** links has quit IRC | 18:39 | |
| jrichli | acoles: I thought the diff looked strange! I was thinking i got confused between all the recent patches | 18:41 |
| *** ChubYann has quit IRC | 18:53 | |
| *** MikeG451 has quit IRC | 18:53 | |
| *** sams-gleb has joined #openstack-swift | 18:54 | |
| *** sams-gleb has quit IRC | 19:00 | |
| *** ChubYann has joined #openstack-swift | 19:06 | |
| *** MikeG451 has joined #openstack-swift | 19:08 | |
| *** thumpba has quit IRC | 19:09 | |
| *** thumpba has joined #openstack-swift | 19:11 | |
| openstackgerrit | Merged openstack/swift: pids in probe is no longer used https://review.openstack.org/324246 | 19:22 |
| *** david-lyle has joined #openstack-swift | 19:24 | |
| *** Jeffrey4l_ has quit IRC | 19:30 | |
| *** Jeffrey4l_ has joined #openstack-swift | 19:31 | |
| *** dtoast has quit IRC | 19:36 | |
| *** openstackstatus has quit IRC | 19:39 | |
| *** openstack has joined #openstack-swift | 19:41 | |
| openstackgerrit | John Dickinson proposed openstack/swift: AUTHORS/CHANGELOG updates for 2.8.0 release https://review.openstack.org/324742 | 19:44 |
| *** david-lyle has quit IRC | 19:57 | |
| *** thumpba has quit IRC | 20:12 | |
| *** tqtran has joined #openstack-swift | 20:14 | |
| NM | Hello guys! | 20:17 |
| NM | Can someone point out docs or explain how proxies handle write failures? Our proxy returned a 503 for a PUT resquest. However it only tried to write on two object-servers. After those tries, it gave up. Should it have tried more servers? (We use 4 replicas and 12 object-servers) | 20:18 |
| notmyname | NM: what version of swift? | 20:19 |
| *** rcernin has joined #openstack-swift | 20:20 | |
| notmyname | NM: and do you have a pastebin for the logs with that trans id ? | 20:21 |
| *** gyee has joined #openstack-swift | 20:31 | |
| *** sheel has quit IRC | 20:35 | |
| openstackgerrit | Michael Barton proposed openstack/swift: go: portability work https://review.openstack.org/324807 | 20:40 |
| *** raildo is now known as raildo-afk | 20:56 | |
| *** sams-gleb has joined #openstack-swift | 20:56 | |
| *** nadeem has joined #openstack-swift | 21:00 | |
| *** sams-gleb has quit IRC | 21:01 | |
| Vinsh | anyone know if/when ubuntu will be adding initd scripts for swift-object-replicator? or maybe that just isn't in the plans? | 21:01 |
| Vinsh | Maybe they are only going to support systemd ? | 21:02 |
| Vinsh | I mean reconstructor not replicator. | 21:02 |
| Vinsh | Most folks using EC probably do so from git or some non package method... but it probably makes the feature less accessible to the others not doing so. | 21:03 |
| Vinsh | I see a bug from 6 months ago with no activity https://bugs.launchpad.net/ubuntu/+source/swift/+bug/1522448 | 21:03 |
| openstack | Launchpad bug 1522448 in swift (Ubuntu) "initd script missing for swift-object-reconstructor process" [Undecided,New] | 21:03 |
| notmyname | Vinsh: is canonical/ubuntu doing anything specific for swift or are they just picking up whatever debian does? if the latter, then onovy is your guy. | 21:05 |
| Vinsh | They seem to be doing bare-minimum, so likely copy debian | 21:06 |
| clarkb | I thought that zigo said ubuntu does lal of their own packaging for openstack things | 21:06 |
| Vinsh | Though they do offer the latest versions of swift in UCA.. they don't appear to have upstart support for swift-object-reconstructor yet. | 21:07 |
| Vinsh | I'm looking to see if they have it for systemd. same for debian/centos. looking | 21:07 |
| Vinsh | ah, debian upstream does support it: https://packages.debian.org/sid/all/swift-object/filelist | 21:09 |
| Vinsh | so it's just ubuntu slacking | 21:09 |
| Vinsh | I understand their package team has been understaffed for some time. I'll see what I can do to correct this/help them. | 21:10 |
| clayg | Vinsh: you the man! | 21:10 |
| NM | notmyname: 2.5.0 upgrading to 2.7.0 | 21:13 |
| clayg | NM, if it was in the middle of writing the object when they went offline (with like restart instead of reload) there's nothing it can do to resend the already sent chunks to a different node. | 21:14 |
| *** adu has quit IRC | 21:16 | |
| *** bill_az has quit IRC | 21:20 | |
| *** pauloewerton has quit IRC | 21:22 | |
| NM | clayg: Actually the object storage logged a 201 but they took more then 10sec, the timeout limit | 21:23 |
| clayg | NM, ah - yeah that's annoying | 21:23 |
| NM | http://pastebin.com/KuPWDBTt | 21:24 |
| clayg | did you have some exception IO going on - like in *addition* to the upgrade? | 21:24 |
| NM | I just wonder why the proxy just gave up after two tries | 21:24 |
| clayg | what do you mean "gave up" - it was *done* - what else was there for it to do? wait on the backends forever? | 21:25 |
| NM | Since we have 12 object storage servers, I thought it would try to write on other servers. | 21:26 |
| clayg | it had already written it? | 21:26 |
| notmyname | you don't have any write affinity turned on, do you? | 21:27 |
| NM | Write no. Just read. | 21:27 |
| NM | I was wondering if there is any rule like 'If the proxy fail to write replicas / 2 without success it wont retry.' | 21:28 |
| clayg | what are those nodes doing where they're taking >10s to eat a 100K object tho? even the "fast" nodes where 6 seconds? is this common in this cluster? how much ram what's your part power? do you have containers on these disks? | 21:28 |
| clayg | NM the proxy does not buffer the object - once the connections for the backend nodes are made - the bytes stream through - end the end it will gather status from the reaining connections ot determine the success - if the majority has failed/timedout it will return an error - since the bytes have already left the proxy - there is no oppertunity to retry | 21:29 |
| clayg | FWIW in this case the fact that all nodes ultimately returned success means the object was fully durable - presumably in the primary locations - so the 5XX response in this case turned out to be conservative | 21:30 |
| clayg | Also, we moved the needle on quorum for even replicas in this last release (2.8) hopefully @notmyname is going to do a great job selling that in the release nodes ;) | 21:31 |
| clayg | *notes | 21:31 |
| notmyname | clayg: patch 324742 line 36. definitely open to different words | 21:33 |
| patchbot | notmyname: https://review.openstack.org/#/c/324742/ - swift - AUTHORS/CHANGELOG updates for 2.8.0 release | 21:33 |
| notmyname | I didn't do anything to "sell" it, just stated that it changed | 21:33 |
| NM | One of our clients put millions of object in a single container, so our load is quite high. | 21:34 |
| *** bill_az has joined #openstack-swift | 21:34 | |
| notmyname | containers are on the same drive? | 21:35 |
| NM | clayg: your explanation about the proxy streaming answer my question :) | 21:35 |
| NM | notmyname: yes :( | 21:36 |
| notmyname | NM: I only say this because you already know it: don't do that | 21:36 |
| clayg | I EAT IOPS FOR BREAKFAST | 21:37 |
| briancline | funny you guys mention this 503 frontend vs 201 backend issue... we have just seen this very recently | 21:37 |
| NM | I'll print this chat and show our manager "The guys who created swift are saying to not do that. Can you please give me those ssds?" | 21:37 |
| notmyname | lol | 21:38 |
| clayg | nice ;) | 21:38 |
| *** NM has quit IRC | 21:38 | |
| clayg | NM it's acctually a pretty good win if you can just put the containers on different devices form the the object devices | 21:38 |
| clayg | you said like 12 nodes? | 21:38 |
| *** rcernin_ has joined #openstack-swift | 21:38 | |
| briancline | in our case it took a long time for at least one of the data nodes to issue a 100 Continue (longer than node_timeout) | 21:38 |
| notmyname | looks like he just left? | 21:38 |
| clayg | so 2 disks out of 4 of those are probably pleanty for your container durability availablity at that size, or 1 disk out of each node - something like that | 21:39 |
| clayg | it should just be ring manipulation and a few rebalances | 21:39 |
| CaioBrentano | notmyname clayg NM had to leave to pick up his son in school | 21:39 |
| notmyname | briancline: that's the kind of thing that really intrigues me about the OVH setup. I'd be really interested to know if they've seen that too | 21:39 |
| notmyname | CaioBrentano: ah ok | 21:39 |
| clayg | briancline: in *that* case it should try to grab the next node off the node iter | 21:39 |
| CaioBrentano | NM typing: I'm sorry to leave that way. I'm really late :( | 21:40 |
| notmyname | no worries. I completely understand | 21:40 |
| *** CaioBrentano has quit IRC | 21:41 | |
| *** adu has joined #openstack-swift | 21:41 | |
| briancline | clayg: yeah, I think we only had a wireshark trace from one node showing that. though all three still logged 201s (this was on a PUT) | 21:41 |
| briancline | notmyname: yeah, me too | 21:42 |
| briancline | things like power saving settings and the like can heavily affect that sequence | 21:43 |
| clayg | briancline: i don't see why the node that was slow to give back the 100 continue would have ever seen anymore bytes from the proxy - can't imagine how that turns into a 201 | 21:43 |
| clayg | but having nodes timeout in the get final status stanza is pretty common - it's where the fsync is | 21:43 |
| clayg | and fucking disks are slow as shit | 21:43 |
| clayg | unless you write to them with golang | 21:43 |
| zaitcev | kernel can make it worse with various memory balancing issues | 21:44 |
| notmyname | briancline: IIRC we had to disable the power saving functions in the drive firmware. that stuff is terrible for swift | 21:44 |
| zaitcev | Like for example if someone plugs a USB key into a storage node | 21:44 |
| briancline | clayg: I... er.. I think I'm confusing two issues that behaved in the same manner | 21:44 |
| zaitcev | Suddenly every I/O has to wait until something gets drained to that key. | 21:45 |
| clayg | zaitcev: use golang | 21:45 |
| zaitcev | clayg: Now you're kidding. | 21:45 |
| clayg | it's like pixy dust - did you hear the tweet from PHP CEO? | 21:45 |
| briancline | notmyname: yeah. bios too. even the kernel will complain about it | 21:45 |
| clayg | briancline: write the bios in golang | 21:45 |
| clayg | sorry i can't stop | 21:46 |
| clayg | ... i'll stop | 21:46 |
| zaitcev | Yes. It was a tweet urging to rewrite everything in APPLE Swift, right? | 21:46 |
| notmyname | zaitcev: https://twitter.com/php_ceo/status/738088338024857602 | 21:46 |
| zaitcev | notmyname: that sounds like a parody account | 21:46 |
| briancline | so, discounting the 100-Continue bit I metnioned.. what was strange in the 5xx/201 case is that each object node logged that they completed the write in <1 sec. but the proxy logged an entry node_timeout seconds later than all of the entries with its 5xx status | 21:47 |
| clayg | briancline: that *is* interesting! | 21:47 |
| notmyname | sounds like eventlet hub starvation | 21:47 |
| clayg | briancline: could point to network congestion, or hub starvation on the object server nodes - sometimes they lie about how long they take - were you looking at the log line timestamps or just the time recording on the log line? | 21:47 |
| briancline | I'd have to check the ticket but I think we're still trying to figure that one out | 21:47 |
| zaitcev | I'm wondering if a switch could cause that. | 21:48 |
| zaitcev | I'll shut up now | 21:48 |
| briancline | notmyname: yeah. that was one of my initial hunches | 21:48 |
| clayg | yeah it could totally just be packets in the wind | 21:48 |
| clayg | but hub starvation is a real deal - if your clocks are syncronized pretty good sometimes you can get hints of it from timestamps - esp if your node timeout is big enough where you're gunna spot that kind of thing | 21:48 |
| clayg | I had a patch to make eventlet tell the truth a little more in the storage nodes - but i didn't follow through | 21:49 |
| notmyname | clayg: we'll just rewrite that part in golang, right? | 21:49 |
| briancline | lol | 21:50 |
| zaitcev | *groan* | 21:50 |
| clayg | notmyname: srly, hub starvation shouldn't be a thing we have to talk about - fucking python | 21:50 |
| zaitcev | Why not. It's a basic technology and modern switches unfortunately do very deep packet inspection. It's a reality with which we have to deal. | 21:51 |
| briancline | clayg: on the timestamps I believe I was looking at syslog's timestamp. on the proxy I can see a line where container-quotas checked the container when the request came in, then the 5xx status line for the actual request a bit later | 21:51 |
| clayg | briancline: found it! https://review.openstack.org/#/c/173622/ | 21:52 |
| patchbot | clayg: patch 173622 - swift - Better storage server eventlet logging (ABANDONED) | 21:52 |
| briancline | guess I'd better double check the actual request start timestamp in the proxy log | 21:52 |
| briancline | oh yeah, I remember this now | 21:53 |
| briancline | man, that feels way more recent than 9 months ago :| | 21:53 |
| clayg | oh yeah i was going to statsd that shit! | 21:55 |
| clayg | :'( | 21:55 |
| clayg | sigh | 21:55 |
| *** ametts has quit IRC | 22:03 | |
| *** nadeem has quit IRC | 22:04 | |
| *** nadeem has joined #openstack-swift | 22:06 | |
| *** rcernin_ has quit IRC | 22:06 | |
| *** nadeem has quit IRC | 22:07 | |
| mattoliverau | Morning | 22:16 |
| *** klrmn has quit IRC | 22:17 | |
| *** klrmn has joined #openstack-swift | 22:18 | |
| *** klamath has quit IRC | 22:19 | |
| *** openstackgerrit has quit IRC | 22:19 | |
| *** openstackgerrit has joined #openstack-swift | 22:20 | |
| torgomatic | morning | 22:34 |
| *** _JZ_ has quit IRC | 22:34 | |
| *** klrmn has quit IRC | 22:39 | |
| *** klrmn has joined #openstack-swift | 22:39 | |
| *** _JZ_ has joined #openstack-swift | 22:41 | |
| *** NM has joined #openstack-swift | 22:46 | |
| *** NM has quit IRC | 22:48 | |
| *** adu has quit IRC | 22:52 | |
| *** sams-gleb has joined #openstack-swift | 22:58 | |
| *** sams-gleb has quit IRC | 23:03 | |
| *** rcernin has quit IRC | 23:10 | |
| *** dtoast has joined #openstack-swift | 23:11 | |
| *** MikeG451 has quit IRC | 23:12 | |
| *** arch-nemesis has quit IRC | 23:16 | |
| *** david-lyle has joined #openstack-swift | 23:20 | |
| *** adu has joined #openstack-swift | 23:23 | |
| zaitcev | I'm going to ask something terrible | 23:30 |
| zaitcev | Do we even have a document like old install_multinode.rst that would be like "do this, this, this to add EC policy to your cluster" | 23:30 |
| *** CaioBrentano has joined #openstack-swift | 23:32 | |
| *** kei_yama has joined #openstack-swift | 23:33 | |
| notmyname | zaitcev: http://docs.openstack.org/developer/swift/overview_erasure_code.html#using-an-erasure-code-policy is probably the closest | 23:34 |
| notmyname | tdasilva: looks like we need to update links on http://docs.openstack.org/developer/swift/overview_erasure_code.html#pyeclib-external-erasure-code-library | 23:34 |
| zaitcev | notmyname: I already have that document, but thanks a lot, I'll run through it again | 23:35 |
| notmyname | zaitcev: ok. I was thinking "you just gotta add the stanza to swift.conf" and that's pretty much it | 23:36 |
| zaitcev | yay | 23:36 |
| zaitcev | the example starts with 2, so I presume discontiguous policy indexes are permissible. | 23:37 |
| mattoliverau | zaitcev: you can use any index you want so long as you continue using it for that policy. | 23:40 |
| notmyname | and it's an integer >= 1 | 23:41 |
| *** hosanai has joined #openstack-swift | 23:43 | |
| *** ChanServ sets mode: +v hosanai | 23:43 | |
| *** _JZ_ has quit IRC | 23:45 | |
| *** bill_az has quit IRC | 23:49 | |
| *** hosanai has quit IRC | 23:49 | |
| *** bill_az has joined #openstack-swift | 23:51 | |
| zaitcev | I'm enjoying devops if only of a short while. | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!