| *** echevemaster has joined #openstack-swift | 00:02 | |
| *** ho has joined #openstack-swift | 00:13 | |
| *** bkopilov has quit IRC | 00:24 | |
| ho | good morining! | 00:24 |
|---|---|---|
| mattoliverau | ho morning | 00:28 |
| *** bkopilov has joined #openstack-swift | 00:29 | |
| ho | mattoliverau: morning | 00:29 |
| *** bkopilov has quit IRC | 00:48 | |
| *** bkopilov has joined #openstack-swift | 00:54 | |
| *** fbo has quit IRC | 01:16 | |
| *** fbo has joined #openstack-swift | 01:17 | |
| *** panbalag has joined #openstack-swift | 01:27 | |
| *** bkopilov has quit IRC | 01:40 | |
| *** Guest____ has joined #openstack-swift | 02:02 | |
| *** kota_ has joined #openstack-swift | 02:05 | |
| *** Guest____ has quit IRC | 02:06 | |
| *** Guest____ has joined #openstack-swift | 02:08 | |
| *** Guest____ has quit IRC | 02:13 | |
| *** panbalag has quit IRC | 02:14 | |
| *** jrichli has quit IRC | 02:16 | |
| *** Gu_______ has joined #openstack-swift | 02:19 | |
| peluse | mattoliverau, awesome thanks. Just made the suggested change, yeah very cool. Plus I found another typo in a comment, good lord I'm falling apart | 02:25 |
| mattoliverau | lol | 02:25 |
| openstackgerrit | paul luse proposed openstack/swift: Make get_dev_path() treat mount_check literally... https://review.openstack.org/166307 | 02:26 |
| mattoliverau | peluse: cool, once you've uploaded the new patchset I'll go check it out and +2 it (hopefully) :) | 02:26 |
| mattoliverau | and there it is, lol | 02:26 |
| peluse | good timing :) | 02:26 |
| *** haomaiwang has joined #openstack-swift | 02:46 | |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Clarify the description of backward compatibility in Keystoneauth's docstring https://review.openstack.org/166681 | 03:15 |
| *** bkopilov has joined #openstack-swift | 03:46 | |
| *** kbee has joined #openstack-swift | 03:56 | |
| openstackgerrit | John Dickinson proposed openstack/swift: EC: support single ranges for GET requests https://review.openstack.org/163620 | 04:20 |
| notmyname | torgomatic: my change is to update the functional tests to include the values I was seeing on Friday that initially revealed the issue. https://review.openstack.org/#/c/163620/5/test/functional/tests.py | 04:21 |
| notmyname | also, I can't see anything that's not awesome about a one-liner that's written as a 25-line-nested-tertiary-statements-with-comments-too statement ;-) | 04:24 |
| *** bkopilov has quit IRC | 04:25 | |
| *** bkopilov has joined #openstack-swift | 04:28 | |
| *** echevemaster has quit IRC | 04:47 | |
| notmyname | torgomatic: I'm still seeing a functional test fail in test_slo_ranged_submanifest() with a 416. but that's the only range functional test that fails | 04:48 |
| openstackgerrit | Hisashi Osanai proposed openstack/swift: Refactor the getting roles logic in _keystone_identity https://review.openstack.org/166696 | 05:01 |
| notmyname | ah, there's another one. test_object.py:test_manifest | 05:04 |
| *** pcaruana has quit IRC | 05:07 | |
| *** ppai has joined #openstack-swift | 05:30 | |
| *** bkopilov has quit IRC | 05:33 | |
| *** bkopilov has joined #openstack-swift | 05:34 | |
| *** Sandeep has joined #openstack-swift | 05:36 | |
| Sandeep | hi | 05:36 |
| *** Sandeep is now known as Guest21005 | 05:36 | |
| Guest21005 | hi | 05:37 |
| *** Guest21005 has left #openstack-swift | 05:37 | |
| notmyname | hi. and bye | 05:37 |
| *** bishtsanddev has joined #openstack-swift | 05:38 | |
| bishtsanddev | hi | 05:39 |
| notmyname | hi again | 05:39 |
| bishtsanddev | hi notmyname | 05:41 |
| bishtsanddev | is this possible in to migrate objects from one devstack to another | 05:42 |
| bishtsanddev | ? | 05:42 |
| notmyname | isn't devstack for a dev environment? ie why do you need to keep the data? | 05:43 |
| notmyname | but to answer the question...yes, probably, with a read from one and a write to another | 05:43 |
| * notmyname doesn't use devstack | 05:43 | |
| bishtsanddev | i am thinking of geo redundancy modal Object migration from one cloud to another (what you prefer for Openstack development) | 05:44 |
| bishtsanddev | ? | 05:44 |
| notmyname | Swift supports geographically disperse clusters. see https://swiftstack.com/blog/2013/07/02/swift-1-9-0-release/ from2013 | 05:46 |
| bishtsanddev | i am interested in contributing to openstack-Could you please guide me on this | 05:47 |
| notmyname | bishtsanddev: openstack is a huge and diverse collection of different source code repos and teams of people. do you have a particular area of interest? | 05:48 |
| bishtsanddev | neutron / Swift | 05:49 |
| bishtsanddev | SDN-NFV | 05:49 |
| *** bkopilov has quit IRC | 05:50 | |
| notmyname | sdn/nfv, of course, would be in the realm of neutron (instead of swift). for that, you should check out #openstack-neutron (I think) | 05:51 |
| notmyname | but if you're interested in object storage, swift is great, and this is the right place :-) | 05:51 |
| bishtsanddev | Yes i am interested in it | 05:52 |
| bishtsanddev | I am focusing in these two area | 05:53 |
| bishtsanddev | Please guide me for swift development | 05:54 |
| notmyname | bishtsanddev: right now in swift we're busy with a final push to a major feature. we're working on getting erasure codes supported (in addition to replicated storage). this week is a big push to get the last stuff done (then we've got some integration work). you can see the channel topic message for some info | 05:54 |
| notmyname | bishtsanddev: but, for general stuff, definitely check out https://swiftstack.com/blog/2013/02/12/swift-for-new-contributors/ and https://wiki.openstack.org/wiki/Swift/ideas and https://bugs.launchpad.net/swift/+bugs | 05:55 |
| notmyname | bishtsanddev: and we have weekly meeings in #openstack-meeting on wednesdays at 1900UTC | 05:56 |
| notmyname | meetings agendas are tracked at https://wiki.openstack.org/wiki/Meetings/Swift | 05:58 |
| bishtsanddev | Great i will join that, first few week i need to learn Openstack-swift | 05:58 |
| notmyname | bishtsanddev: http://swift.openstack.org and https://swiftstack.com/openstack-swift/ are two places for high-level understanding of how swift works and what it does | 05:59 |
| bishtsanddev | currently in home i have MAC with 4gb ram, its enough for development purpose | 06:00 |
| bishtsanddev | ? | 06:00 |
| bishtsanddev | or i have to purchase some high configuration PC (BOTH Development of Swift and Testing) | 06:01 |
| notmyname | http://docs.openstack.org/developer/swift/development_saio.html | 06:02 |
| *** bkopilov has joined #openstack-swift | 06:03 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:04 | |
| *** Gu_______ has quit IRC | 06:04 | |
| bishtsanddev | Is there any way to contact you offline (mean) mail? | 06:05 |
| *** kota_ has quit IRC | 06:05 | |
| *** kota_ has joined #openstack-swift | 06:08 | |
| notmyname | bishtsanddev: the best way, if you're not on IRC, is to use the openstack-dev mailing list | 06:09 |
| notmyname | lists.openstack.org | 06:10 |
| *** ktsuyuzaki has quit IRC | 06:10 | |
| bishtsanddev | i am in maling list...Thanks | 06:12 |
| bishtsanddev | one more question- Requirement to join meeting | 06:13 |
| bishtsanddev | ? | 06:13 |
| bishtsanddev | its Chat based meeting or Confrence ? | 06:13 |
| notmyname | it's in irc | 06:13 |
| bishtsanddev | THanks | 06:13 |
| bishtsanddev | Thanks what is your exactly name.....My Mom said ....say thanks with name :) | 06:14 |
| notmyname | heh. in here I'm notmyname. otherwise, http://not.mn | 06:15 |
| bishtsanddev | Thanks John Dickinson. Nice Talking too you... | 06:16 |
| *** ktsuyuzaki has joined #openstack-swift | 06:19 | |
| *** bishtsanddev has quit IRC | 06:20 | |
| *** kota_ has quit IRC | 06:21 | |
| *** kota_ has joined #openstack-swift | 06:24 | |
| *** ktsuyuzaki has quit IRC | 06:25 | |
| *** ktsuyuzaki has joined #openstack-swift | 06:26 | |
| *** jamielennox is now known as jamielennox|away | 06:28 | |
| *** kota_ has quit IRC | 06:28 | |
| *** kota_ has joined #openstack-swift | 06:29 | |
| *** ktsuyuzaki has quit IRC | 06:31 | |
| notmyname | torgomatic: I found an issue with DLOs and suffix ranges with the current version of the patch. leaving a comment now | 06:31 |
| notmyname | (and then to bed) | 06:31 |
| *** bkopilov is now known as bkopilov_wfh | 06:37 | |
| notmyname | torgomatic: ok, comment left. | 06:38 |
| notmyname | for any wanting to take a look at something, EC + zero-byte with a suffix range causes problems (traceback in the logs). https://review.openstack.org/#/c/163620/ | 06:39 |
| notmyname | and now it's midnight, so I'll see you tomorrow | 06:39 |
| ho | I'm looking for doc to build a saio with ec. Is there any doc? | 06:42 |
| *** ppai has quit IRC | 06:51 | |
| openstackgerrit | Merged openstack/swift: Fix typo in swift/test/unit/account/test_backend.py https://review.openstack.org/166421 | 06:53 |
| *** ppai has joined #openstack-swift | 06:57 | |
| *** pcaruana has joined #openstack-swift | 06:58 | |
| *** nshaikh has joined #openstack-swift | 06:59 | |
| *** Bsony has joined #openstack-swift | 06:59 | |
| *** silor has joined #openstack-swift | 07:00 | |
| *** wer has quit IRC | 07:01 | |
| *** wer has joined #openstack-swift | 07:03 | |
| *** ttrumm has joined #openstack-swift | 07:34 | |
| *** ppai has quit IRC | 07:35 | |
| *** bkopilov_wfh has quit IRC | 07:37 | |
| *** bkopilov_wfh has joined #openstack-swift | 07:44 | |
| *** ppai has joined #openstack-swift | 07:47 | |
| *** bkopilov has joined #openstack-swift | 07:51 | |
| *** bkopilov_wfh has quit IRC | 07:52 | |
| *** chlong has quit IRC | 07:58 | |
| *** bkopilov has quit IRC | 08:19 | |
| *** jordanP has joined #openstack-swift | 08:21 | |
| *** mmcardle has quit IRC | 08:28 | |
| *** geaaru has joined #openstack-swift | 08:32 | |
| *** mmcardle has joined #openstack-swift | 08:34 | |
| *** SkyRocknRoll has joined #openstack-swift | 08:40 | |
| *** SkyRocknRoll has joined #openstack-swift | 08:40 | |
| *** ppai has quit IRC | 08:53 | |
| *** joeljwright has joined #openstack-swift | 08:57 | |
| *** jistr has joined #openstack-swift | 09:02 | |
| *** ondergetekende_ has quit IRC | 09:03 | |
| *** ppai has joined #openstack-swift | 09:07 | |
| *** jordanP has quit IRC | 09:27 | |
| *** kota_ has quit IRC | 09:27 | |
| *** tab___ has joined #openstack-swift | 09:28 | |
| *** jordanP has joined #openstack-swift | 09:28 | |
| *** ho has quit IRC | 09:44 | |
| openstackgerrit | Kota Tsuyuzaki proposed openstack/swift: Small ec diskfile refactor https://review.openstack.org/166754 | 09:46 |
| *** kota_ has joined #openstack-swift | 09:47 | |
| *** ktsuyuzaki has joined #openstack-swift | 09:52 | |
| *** kota_ has quit IRC | 09:54 | |
| *** kota_ has joined #openstack-swift | 09:54 | |
| *** ktsuyuzaki has quit IRC | 09:57 | |
| *** ktsuyuzaki has joined #openstack-swift | 09:57 | |
| *** kota_ has quit IRC | 09:59 | |
| *** kota_ has joined #openstack-swift | 09:59 | |
| *** ktsuyuzaki has quit IRC | 10:01 | |
| *** ktsuyuzaki has joined #openstack-swift | 10:02 | |
| *** kota_ has quit IRC | 10:04 | |
| *** kota_ has joined #openstack-swift | 10:05 | |
| *** ktsuyuzaki has quit IRC | 10:06 | |
| *** ktsuyuzaki has joined #openstack-swift | 10:07 | |
| *** kota_ has quit IRC | 10:09 | |
| *** kota_ has joined #openstack-swift | 10:10 | |
| *** ktsuyuzaki has quit IRC | 10:12 | |
| *** ktsuyuza_ has joined #openstack-swift | 10:12 | |
| *** ppai has quit IRC | 10:13 | |
| *** ondergetekende has joined #openstack-swift | 10:13 | |
| *** kota_ has quit IRC | 10:14 | |
| *** kota_ has joined #openstack-swift | 10:17 | |
| *** ktsuyuza_ has quit IRC | 10:18 | |
| *** ktsuyuzaki has joined #openstack-swift | 10:21 | |
| *** kota_ has quit IRC | 10:23 | |
| *** ktsuyuzaki has quit IRC | 10:26 | |
| *** ppai has joined #openstack-swift | 10:27 | |
| *** nottrobin is now known as nottrobi | 10:29 | |
| *** nottrobi is now known as nottrobin | 10:29 | |
| *** haomaiwang has quit IRC | 10:38 | |
| *** SkyRocknRoll has quit IRC | 10:41 | |
| *** bkopilov has joined #openstack-swift | 10:44 | |
| *** panbalag has joined #openstack-swift | 10:45 | |
| *** panbalag has quit IRC | 10:45 | |
| *** panbalag has joined #openstack-swift | 10:46 | |
| *** ttrumm has quit IRC | 10:53 | |
| *** kbee has quit IRC | 10:55 | |
| *** Bsony_ has joined #openstack-swift | 11:00 | |
| *** kota_ has joined #openstack-swift | 11:02 | |
| *** ppai has quit IRC | 11:02 | |
| *** Bsony has quit IRC | 11:04 | |
| *** silor has quit IRC | 11:13 | |
| *** ppai has joined #openstack-swift | 11:16 | |
| *** zul has joined #openstack-swift | 11:32 | |
| *** bkopilov has quit IRC | 11:34 | |
| *** bkopilov has joined #openstack-swift | 11:34 | |
| *** nshaikh has quit IRC | 11:39 | |
| openstackgerrit | Lorcan Browne proposed openstack/swift: Add swift-recon feature to track swift-drive-audit error count https://review.openstack.org/163889 | 11:46 |
| *** ppai has quit IRC | 11:49 | |
| *** ppai has joined #openstack-swift | 12:03 | |
| chmouel | clayg: yeah it's the thing i wrote some time ago (cloudfs) and guys from memset took over and made it basically working | 12:18 |
| *** bkopilov has quit IRC | 12:19 | |
| *** kota_ has quit IRC | 12:21 | |
| *** jrichli has joined #openstack-swift | 12:41 | |
| *** bkopilov has joined #openstack-swift | 12:45 | |
| openstackgerrit | Merged openstack/swift: Test swift-object-info opens meta and ts files https://review.openstack.org/165395 | 12:50 |
| *** silor has joined #openstack-swift | 12:51 | |
| openstackgerrit | Merged openstack/swift: Per-policy DiskFile classes https://review.openstack.org/165125 | 12:51 |
| *** Bsony has joined #openstack-swift | 13:01 | |
| *** ppai has quit IRC | 13:03 | |
| *** Bsony_ has quit IRC | 13:04 | |
| tab___ | one quick question. Having 12 disk node. how much CPU cores would you recommend within swift set-up to have on such system for expecting for example write throughput of just about 3-4 MB/s, i gess no big demands. Typical HP DL380 has for example 2xCPU, each for example with 6 cores. Since there will also be some other high demand appliaction on this system, would 2 cores be enough for Swift to process such a demand?, | 13:04 |
| *** annegentle has joined #openstack-swift | 13:16 | |
| *** bkopilov has quit IRC | 13:32 | |
| *** cdelatte has quit IRC | 13:32 | |
| *** delattec has quit IRC | 13:32 | |
| *** bkopilov has joined #openstack-swift | 13:33 | |
| *** bkopilov has quit IRC | 13:39 | |
| *** bkopilov has joined #openstack-swift | 13:42 | |
| *** acoles_away is now known as acoles | 13:52 | |
| tdasilva | dencaval: hi!, I was just reading your comment on patch 156825, but i'm not sure I understand what you mean | 13:58 |
| patchbot | tdasilva: https://review.openstack.org/#/c/156825/ | 13:58 |
| openstackgerrit | Martin Kletzander proposed openstack/swift: Fix common misspellings https://review.openstack.org/166843 | 14:00 |
| dencaval | tdasilva hm, It's just an idea to have the policy object created early instead at _store_object | 14:01 |
| tdasilva | dencaval: oh I see...I think the idea is that everything in the PUT method itself is more like req validation and some API handling, and in _store_object is where data gets transferred to the backend storage nodes, so I think that's where policy would come into play, what do you think? | 14:04 |
| *** bkopilov has quit IRC | 14:04 | |
| *** mragupat has joined #openstack-swift | 14:08 | |
| dencaval | dasilva hm, I do not know yet if some req validation could take some different decision according to the policy, but It could be easy to do if policy is available | 14:09 |
| *** Gue______ has joined #openstack-swift | 14:11 | |
| *** acoles is now known as acoles_away | 14:17 | |
| *** PeterTr7 has joined #openstack-swift | 14:17 | |
| PeterTr7 | Hi, is anyone familiar with Python-swiftclient? I have a feature I was wondering if I should build out myself or potentially contribute | 14:20 |
| *** kallebe has joined #openstack-swift | 14:20 | |
| tdasilva | dencaval: yeah, at least in the ec code so far, it was not needed, but if that comes up, we could change it later...my 2 cents :-) | 14:22 |
| *** bkopilov has joined #openstack-swift | 14:22 | |
| *** lpabon has joined #openstack-swift | 14:26 | |
| *** petertr7_ has joined #openstack-swift | 14:27 | |
| *** annegentle has quit IRC | 14:29 | |
| *** annegentle has joined #openstack-swift | 14:34 | |
| *** annegentle has quit IRC | 14:34 | |
| joeljwright | PeterTr7: If you give more details it might start a discussion | 14:34 |
| PeterTr7 | Thank you joeljwright! | 14:35 |
| *** Fin1te has joined #openstack-swift | 14:35 | |
| joeljwright | PeterTr7: If you're interested in contributing you should take a look here too http://docs.openstack.org/infra/manual/developers.html | 14:38 |
| PeterTr7 | Thank you! I'm reading through this and some more getting started documentation | 14:39 |
| *** mahatic has joined #openstack-swift | 14:47 | |
| PeterTr7 | I'm interested in creating an "upload continuation" feature for users who upload several TB of data to a Swift cluster via python-swiftclient. In particular, uploading files exceeding the 5gb limit and need to be segmented. The primary use case for this is when a user is tasked with uploading a lot of data and maybe interrupted (disconnected, lunch break, etc). I would like the user to be able to continue the upload of some larg | 14:48 |
| *** annegentle has joined #openstack-swift | 14:51 | |
| notmyname | https://metacpan.org/release/MASAKYST/Net-OpenStack-Swift-0.02 <-- anyone know this person? | 14:51 |
| *** acoles_away is now known as acoles | 14:56 | |
| *** mragupat has quit IRC | 15:03 | |
| *** Fin1te has quit IRC | 15:06 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: refactor PUT method https://review.openstack.org/164561 | 15:07 |
| openstackgerrit | Thiago da Silva proposed openstack/swift: Validate the PUT method extraction for EC https://review.openstack.org/164950 | 15:07 |
| *** kallebe has quit IRC | 15:29 | |
| *** Fin1te has joined #openstack-swift | 15:31 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: Validate the PUT method extraction for EC https://review.openstack.org/164950 | 15:36 |
| tdasilva | clayg: ^^^ squashed the two patches to make it easier to review | 15:38 |
| *** bkopilov has quit IRC | 15:51 | |
| *** Gue______ has quit IRC | 15:51 | |
| *** kallebe has joined #openstack-swift | 15:53 | |
| *** Gue______ has joined #openstack-swift | 15:54 | |
| *** kallebe has left #openstack-swift | 15:54 | |
| *** reed has joined #openstack-swift | 15:55 | |
| *** PeterTr7 has quit IRC | 15:58 | |
| *** petertr7_ is now known as petertr7 | 15:58 | |
| notmyname | good morning | 16:04 |
| acoles | notmyname: good morning | 16:05 |
| notmyname | big week this week | 16:07 |
| peluse | mornin' | 16:07 |
| *** Bsony has quit IRC | 16:08 | |
| peluse | I went though acoles patch chain and +2'd each one. They all look and work cominbed with the recon which was just tons of fun to switch over :) | 16:08 |
| acoles | peluse: cool, i bet it wasn't 'fun' | 16:08 |
| acoles | peluse: fyi i am mid-review of clayg's patch 159637 | 16:09 |
| patchbot | acoles: https://review.openstack.org/#/c/159637/ | 16:09 |
| *** gyee has joined #openstack-swift | 16:09 | |
| peluse | If we can land those today and I can make some progress on ecrecon we should be able to get the PUT refactor in next, any of torgomatics GET range stuff, more ssync tests and then the recon possibly most of those this week | 16:09 |
| peluse | acoles, cool - that one surprised me. I figured "no way will this work" :) | 16:10 |
| notmyname | and by "this week" you mean by wednesday, right? ;-) | 16:10 |
| peluse | yes, wed | 16:10 |
| acoles | peluse: oh, and re ssync tests, i spent some time over weekend reworking those end-to-end tests i had (to make them way more compact/re-usable i hope) - not quite there yet though | 16:10 |
| peluse | I have spotted time today but almost all day Tue/Wed both | 16:11 |
| acoles | notmyname: weds? aren't we targetting 27th? | 16:11 |
| notmyname | acoles: ya :-) | 16:11 |
| acoles | notmyname: you just piling on the pressure there> :) | 16:11 |
| notmyname | peluse is out for something like "I've been married multiple decades and am going somewhere with my wife" on thursday and friday | 16:11 |
| notmyname | acoles: heh | 16:12 |
| notmyname | acoles: you've got until friday. peluse has to be done by wednesday :-) | 16:12 |
| peluse | yeah, San Fran and Napa! | 16:12 |
| peluse | will text you Thu eve to see if you want to pop down to thr RickHouse fo a drink with us... | 16:12 |
| acoles | peluse: right! 'hey darling, all the guys from work are coming along to our anniversary dinner...' | 16:13 |
| notmyname | lol | 16:13 |
| acoles | peluse: congrats by the way. | 16:14 |
| peluse | heh - the celebration starts in Napa on Fri. Thu my sister and her husband are also in town so its more about just having fun and less about the anniverary then. | 16:14 |
| peluse | acoles, thanks :) | 16:14 |
| acoles | notmyname: peluse : i am in Galway this week, where the team have just moved into a brand new building. I am not sure yet whether my pass will allow me to leave overnight ;) | 16:16 |
| * notmyname goes to figure out gerrit patch chains | 16:17 | |
| notmyname | looks like just one more +2 needed for https://review.openstack.org/#/c/156825/ | 16:19 |
| peluse | notmyname, yeah I was thinking we should land scoles and claygs MFI patches first since they're a bit more complex then if there are conflicts it'd be easier to deal with them on the PUT refator landing | 16:20 |
| notmyname | ok. on master or feature/ec | 16:21 |
| notmyname | tdasilva: after that one lands ^ what's the state of the version for feature/ec? pick it up from master? or still work on https://review.openstack.org/#/c/164561/ | 16:21 |
| *** gyee has quit IRC | 16:21 | |
| peluse | I'm only talking feature/ec. shit I was reading the wrong patch | 16:22 |
| notmyname | peluse: acoles: looks like https://review.openstack.org/#/c/165188/3 is the current start of the reconstructor chain | 16:22 |
| notmyname | is that correct? | 16:22 |
| peluse | I have to take my daughter to a Dr appt but when I get back, if it hasn't already landed, I'll review the master version | 16:22 |
| notmyname | thanks | 16:22 |
| peluse | notmyname, that one is still outside of the reconstructor | 16:23 |
| notmyname | peluse: looks like there's a dependency chain there | 16:23 |
| peluse | yeah, I neeed it for sure | 16:23 |
| peluse | but it can land first w/o screwing anything else up | 16:23 |
| *** gyee has joined #openstack-swift | 16:23 | |
| notmyname | ok. that's all I'm looking at. the root of the various patch chains | 16:24 |
| acoles | notmyname: yes looks like thats now the start of the chain | 16:24 |
| peluse | the chain all the way up to https://review.openstack.org/#/c/159637/ can land as "pre reconstructor" work | 16:24 |
| acoles | why does git review -d for *any* reviewon a chain create the *same* local branch name? :( its so confusing | 16:24 |
| peluse | BTW, acoles my hat is off to you for your work there. Really, really nice work! | 16:25 |
| peluse | and clayg of course as well | 16:25 |
| acoles | its all clayg | 16:26 |
| peluse | OK, I'm outta here for a few hrs.... back later | 16:26 |
| notmyname | acoles: what's the status of https://review.openstack.org/#/c/162249/ | 16:26 |
| notmyname | marked WIP with a merge conflict and a referenced patch that has landed | 16:26 |
| acoles | notmyname: abandoned as of 10 seconds ago | 16:27 |
| notmyname | thanks | 16:27 |
| *** Fin1te has quit IRC | 16:29 | |
| *** tsg_ has joined #openstack-swift | 16:35 | |
| *** zaitcev has joined #openstack-swift | 16:44 | |
| *** ChanServ sets mode: +v zaitcev | 16:44 | |
| *** reed has quit IRC | 16:45 | |
| *** AbyssOne is now known as a1|away | 16:46 | |
| *** reed has joined #openstack-swift | 16:46 | |
| *** Gue______ has quit IRC | 16:49 | |
| *** Gue______ has joined #openstack-swift | 16:49 | |
| *** jordanP has quit IRC | 16:56 | |
| *** annegentle has quit IRC | 16:59 | |
| *** annegentle has joined #openstack-swift | 17:10 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 17:14 |
| *** zhill has joined #openstack-swift | 17:21 | |
| tdasilva | notmyname: I tried to squash 164561 and 164950 and I think I got it right (checking now), so I should be able to abandon 164561, so that there's only one refactoring to review in feature/ec. | 17:22 |
| notmyname | thanks | 17:22 |
| tdasilva | notmyname: I think if patch 156825 lands on master and we merge master to feature/ec, then it will be even easier to review the EC refactor | 17:24 |
| patchbot | tdasilva: https://review.openstack.org/#/c/156825/ | 17:24 |
| notmyname | right. that's the plan, as I understand it | 17:24 |
| tdasilva | ok, thanks | 17:24 |
| *** kallebe has joined #openstack-swift | 17:27 | |
| *** annegentle has quit IRC | 17:37 | |
| *** cdelatte has joined #openstack-swift | 17:39 | |
| *** delattec has joined #openstack-swift | 17:39 | |
| *** khivin has quit IRC | 17:40 | |
| *** jest__ has joined #openstack-swift | 17:44 | |
| *** zhill_ has joined #openstack-swift | 17:51 | |
| *** jistr has quit IRC | 17:54 | |
| *** bkopilov has joined #openstack-swift | 17:55 | |
| kallebe | Hello. Since assertEquals is deprecated for quite some time ( http://bugs.python.org/issue9424 and https://docs.python.org/2/library/unittest.html#deprecated-aliases ), I think it would be good if the tests were refactored to use assertEqual instead of assertEquals. I grep'ed the code and there were 4635 occurences of assertEquals. I am thinking ab | 17:58 |
| kallebe | out doing the replace (simple sed and check the indentation for some cases). However, all this change, if merged, would cause much merge conflict in current change requests. What do you think? | 17:58 |
| *** bkopilov has quit IRC | 17:59 | |
| *** bkopilov has joined #openstack-swift | 17:59 | |
| *** Fin1te has joined #openstack-swift | 18:00 | |
| kallebe | Forgot to add that there is already a hacking check about this, but apparently it it nos enabled for swift: https://github.com/openstack-dev/hacking/blob/master/hacking/checks/python23.py#L106 | 18:01 |
| *** sandywalsh_ has joined #openstack-swift | 18:03 | |
| *** sandywalsh has quit IRC | 18:06 | |
| zaitcev | I think "Who in the world cares" | 18:07 |
| kallebe | Yes, it is not really important, but one day this will have to be done. We don't know how long assertEquals will be compatible | 18:08 |
| *** bkopilov has quit IRC | 18:12 | |
| zaitcev | We're all hands on deck for EC in Kilo right now. Once that ships in April, then file away, someone will hopefuly waddle through and verify, +2. | 18:18 |
| tdasilva | kallebe: I think there should also be a discussion about enabling the hacking check after you change, otherwise there would be no point in having you make all those changes, just to allow people to go and add assertEquals in new tets... | 18:27 |
| *** silor has quit IRC | 18:27 | |
| kallebe | tdasilva yes, I was thinking about this too. I think it will not be worth to do this huge change for now because it will only delay all current patch sets because almost all test files have assertEquals | 18:29 |
| zaitcev | Enabling that goes without saying | 18:29 |
| tdasilva | kallebe: but like zaitcev said, everyone is very focused on the EC work right now, so if you submit a patch it probably won't be looked at for a while. My other suggestion would be that if it makes sense, try to break up into smaller patches (e.,g maybe one for functional tests and another for unit)...just an idea... | 18:31 |
| zaitcev | no, don't mix it | 18:31 |
| zaitcev | although a good idea by itself | 18:32 |
| *** mmcardle has quit IRC | 18:32 | |
| tdasilva | don't mix? or don't break up? | 18:32 |
| zaitcev | don't do breakup and assertEqual(s) in one patch | 18:32 |
| zaitcev | I'd hate to review that hairball. | 18:32 |
| lpabon | kallebe: i think that is a great idea, i look forward to your patch or patches | 18:41 |
| clayg | ohai | 18:46 |
| clayg | i have a bunch of outstanding code that I don't know how to move forward with - i think i'm spending too much time thinking about the right way to order the diffs for the next week that I'm not focusing on fixing the things that are broken quickly enough | 18:47 |
| clayg | e.g. the handoff get node index thing | 18:48 |
| *** jest__ has quit IRC | 18:56 | |
| peluse | clayg, I had some ideas there - what did you have in mind? | 18:58 |
| peluse | clayg, oh, just saw email | 18:59 |
| *** raginbajin has quit IRC | 19:04 | |
| *** swat30 has quit IRC | 19:04 | |
| *** raginbajin has joined #openstack-swift | 19:04 | |
| *** Dieterbe has quit IRC | 19:04 | |
| *** omame has quit IRC | 19:04 | |
| *** bus-104 has quit IRC | 19:04 | |
| *** Dieterbe has joined #openstack-swift | 19:04 | |
| *** swat30 has joined #openstack-swift | 19:04 | |
| *** bus-104 has joined #openstack-swift | 19:05 | |
| *** geaaru has quit IRC | 19:05 | |
| peluse | tsg_, any chance you were able to look into the requiements.txt issue on feature/ec? | 19:07 |
| *** Gue______ has quit IRC | 19:08 | |
| notmyname | peluse: I can follow up with that one (perhaps after lunch) | 19:10 |
| notmyname | peluse: for the time being, can you repropose that merge from master without the requirements change? that way we won't be blocked by it | 19:10 |
| *** Gues_____ has joined #openstack-swift | 19:10 | |
| *** erlon_away has joined #openstack-swift | 19:10 | |
| peluse | notmyname, I can't figure out how to get it removed from the patch | 19:14 |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Remove all DLO segments on upload of replacement https://review.openstack.org/161972 | 19:15 |
| openstackgerrit | John Dickinson proposed openstack/swift: Merge master to feature/ec https://review.openstack.org/166558 | 19:16 |
| notmyname | peluse: there | 19:16 |
| peluse | well well | 19:17 |
| notmyname | peluse: well, all I did was use the requeirements.txt from master | 19:17 |
| notmyname | `git checkout master -- requirements.txt` | 19:17 |
| notmyname | peluse: please validate it. if it works, land it. I'm going to grab some lunch | 19:18 |
| peluse | notmyname, so yeah I tried that and Jenkins still puked. I was trying to totally remove it from the patch - that's what I couldn't figure out how to do. | 19:18 |
| notmyname | oh. hmm | 19:18 |
| peluse | hell, maybe I did it wrong :) we | 19:18 |
| notmyname | ya, I'm not sure why it's still showing up | 19:18 |
| peluse | will see | 19:18 |
| acoles | clayg: i'm reviewing patch 159637, got a question for you | 19:19 |
| patchbot | acoles: https://review.openstack.org/#/c/159637/ | 19:19 |
| clayg | acoles: sup! | 19:20 |
| clayg | thanks for fixing that test bug thing last week or whatever | 19:20 |
| acoles | clayg: i'm trying to convince myself that all the test coverage has been retained :) so here https://review.openstack.org/#/c/159637/20/test/unit/obj/test_diskfile.py line 901-916 in the *base* version | 19:21 |
| acoles | i don't see those two tests translated down to the new test class, but i think the scenarios are covered, correct? | 19:21 |
| acoles | clayg: like, i broke the code and *some* tests still failed | 19:21 |
| *** dencaval has quit IRC | 19:21 | |
| acoles | clayg: so just checking thats part of your plan :) | 19:22 |
| clayg | i would have swore the raise unavailable test was in there | 19:22 |
| clayg | test_get_hashes_bad_dev | 19:22 |
| acoles | clayg: oh crap, sorry, wrong lines!! | 19:22 |
| acoles | hang on | 19:22 |
| *** erlon_away is now known as erlon | 19:23 | |
| clayg | then test_get_hashes_create... fine | 19:23 |
| acoles | clayg: line 187 - 196 | 19:23 |
| clayg | yeah those are probably there too - let's find them | 19:23 |
| acoles | clayg: i think you have just covered in tests on get_hashes? | 19:23 |
| clayg | yeah I never tested hash_suffix directly because it's not part of the public interface in any fashion - it should just be an implemenation detail | 19:24 |
| clayg | the interface of when it raises and whatever is a total coupling fuck | 19:25 |
| acoles | clayg: thats what i figured | 19:25 |
| clayg | so the assertraises pathnotdir - turned into test_get_hashes_hash_suffix_enotdir probably | 19:25 |
| clayg | yeah, then test_get_hashes_hash_suffix_other_oserror | 19:26 |
| clayg | i think you're cover | 19:26 |
| clayg | d | 19:26 |
| clayg | thanks for checking! | 19:26 |
| acoles | clayg: yup makes sense those are tests that failed when i broke the code. | 19:27 |
| clayg | lol ;) | 19:27 |
| clayg | i guess they were there for that! | 19:27 |
| acoles | clayg: ok i just wanted to be sure it wasn't a cut'n'paste blooper | 19:27 |
| clayg | acoles: totally could have happened - i really appreciate you double checking me | 19:27 |
| acoles | clayg: well, my eyes are dropping on those tests, i've been up since 5am (flew over to ireland today), so i'm not sure i'm doing a great job | 19:30 |
| clayg | i'll take a sleepy acoles over almost any other review any time! | 19:30 |
| acoles | clayg: lol. so....ECDiskFileManager.invalidate_hash - we don't need it do we? | 19:31 |
| acoles | clayg: like i just deleted it and tests passed ;D | 19:31 |
| clayg | hrmmm..... | 19:32 |
| * acoles feels invalidate_hash is his nemesis | 19:32 | |
| clayg | it's an exact dupe of the module function right? | 19:32 |
| acoles | yup | 19:32 |
| clayg | ok sweet - don't need it! | 19:32 |
| peluse | clayg, try deleting some more code and see what happens - you could be on a roll | 19:32 |
| tdasilva | lol, are we also replicating code now? | 19:33 |
| peluse | data *and* code :) | 19:33 |
| tdasilva | :-) | 19:33 |
| acoles | peluse: clayg: so this is *erasure* coded code yeah, like we can delete pieces and it keeps working? | 19:33 |
| peluse | oh man, don't go there! | 19:34 |
| tsg_ | peluse: on the requirements issue on feature/ec, looks like Jenkins still is complaining .. I am looking | 19:34 |
| acoles | clayg: i'm pretty sure i diff'd the two functions and they were the same , i'll double check myself | 19:34 |
| peluse | clayg, if you're going to work on the GET FI thing I'll keep on the ECrecon however if not I'll see if I can nail it here pretty quick - your call chief | 19:35 |
| peluse | pretty sure? heh | 19:36 |
| *** jest_____ has joined #openstack-swift | 19:36 | |
| clayg | acoles: I think there were needlessly different - i vaugly remember seeing something in a self review that reminded me I'm stupid | 19:36 |
| acoles | clayg: ah, there is a diff -you took out the os.path.exists(hashes_file) check before lock_partition | 19:37 |
| clayg | yeah i don't recall at all why that was a good idea | 19:37 |
| peluse | or why it would be different EC vs repl | 19:38 |
| clayg | I think it was a hold over of refctoring out the class things - the whole thing can be dropped most likely | 19:38 |
| clayg | acoles: ding me on the review and I'll get it cleaned up | 19:38 |
| acoles | clayg: my guess is that the check is done for early exit before taking the lock, so imho probably best left there | 19:38 |
| clayg | I think i'm going to rebase the world by EOB | 19:38 |
| clayg | acoles: should probably have a behavior test - like I said, I think i was just copying pauls code into the per policy diskfiles - and the lock was probably ahppening inside of a class call that I replaced | 19:39 |
| acoles | clayg: there were a few totally nit typos i found and fixed locally, you want me to just push over with invalidate_hash gone too? or you prefer i leave it to you? | 19:40 |
| acoles | yeah a test would be good | 19:40 |
| clayg | oh, yeah if you've already got a diff going just push - you can make me write the test | 19:40 |
| acoles | k will do | 19:40 |
| clayg | well otoh - just push the typos - the broken code may be useful for the test | 19:40 |
| clayg | it's either way | 19:40 |
| acoles | i'll write a test it np | 19:41 |
| clayg | oh ptptphthtfhfhhfh - even better :P | 19:41 |
| acoles | but if it proves tricky i'll give up and go get a beer ;D | 19:41 |
| openstackgerrit | paul luse proposed openstack/swift: Make get_dev_path() treat mount_check literally... https://review.openstack.org/166307 | 19:43 |
| *** omame has joined #openstack-swift | 19:46 | |
| *** kallebe has left #openstack-swift | 19:55 | |
| *** os1 has joined #openstack-swift | 19:58 | |
| os1 | Hi | 19:58 |
| os1 | Is there any way I could see an updates/specs page in regards to EC? | 19:58 |
| os1 | Just out of curiosity. | 19:59 |
| peluse | usre, one sec | 19:59 |
| *** Gues_____ has quit IRC | 20:00 | |
| peluse | well, maybe like 5 min on the spec page but wrt current WIP see https://trello.com/b/LlvIFIQs/swift-erasure-codes | 20:00 |
| os1 | peluse: Thanks. | 20:03 |
| os1 | Just out of curiosity, once EC is done, what are the "next big features" that the community plans to work on? | 20:03 |
| os1 | It seems like EC was big, in terms of effort and resources involved. | 20:04 |
| peluse | all sorts of good stuff :) notmyname may chime in, I'm trying to get a few changes to the spec done here before I have to run off to a meeting (yes EC=big) | 20:04 |
| os1 | wondering it that's normally the case, or if there is something particular to EC that made it a big accomplishment. | 20:05 |
| acoles | clayg: i can fix that stupid invalidate_hash forwarding method that i put into DiskFileWriter as well | 20:06 |
| *** tsg_ has quit IRC | 20:07 | |
| *** tsg has joined #openstack-swift | 20:08 | |
| clayg | acoles: waht do you mean now? | 20:11 |
| *** annegentle has joined #openstack-swift | 20:11 | |
| clayg | acoles: I think it's probably worth it in the long run to forward the majority of module methods to the manager - i stopped just sort of read/write_metadata | 20:11 |
| clayg | i even forwarded quarantine renamer in one of my dependent patches - but I didn't acctually implement anything better for the ECDiskFileManager | 20:12 |
| *** Fin1te has quit IRC | 20:12 | |
| clayg | oh god - i didn't realize we merged the per policy diskfile change already - i need to rebase badly :\ | 20:13 |
| peluse | os1, patch coming here in just another minute or two.... | 20:14 |
| acoles | clayg: i only mean this http://paste.openstack.org/show/195551/ , still routing to manager, just without the hop through the writer method | 20:15 |
| clayg | acoles: yeah that was a good change - i have that in my chain somewhere too - i bit up too much rebasing in this cleanup - cost me too much time and folks (rightfully) moved on without me. | 20:17 |
| openstackgerrit | Alistair Coles proposed openstack/swift: Multiple Fragment Archive support for suffix hashes https://review.openstack.org/159637 | 20:22 |
| acoles | clayg: peluse: ^^hope i didn't break it! gotta go find my hotel and food now | 20:23 |
| openstackgerrit | paul luse proposed openstack/swift-specs: Updates to the reconstructor section, quick scrub of everything https://review.openstack.org/142146 | 20:23 |
| peluse | acoles, that sounds enoucraging :) | 20:24 |
| peluse | acoles, off to a meeting. Will check it out later | 20:24 |
| peluse | os1, once that builds ^^ click on the docs link in there and you'll see the latest | 20:25 |
| peluse | os1, here's the direct link http://docs-draft.openstack.org/46/142146/4/check/gate-swift-specs-docs/6910dcb//doc/build/html/specs/in_progress/erasure_coding.html | 20:30 |
| *** bkopilov has joined #openstack-swift | 20:31 | |
| *** zhill_ has quit IRC | 20:32 | |
| *** acoles is now known as acoles_away | 20:35 | |
| tdasilva | tsg: hi, when was ECPyECLibException removed? | 20:37 |
| *** tsg has quit IRC | 20:38 | |
| *** Fin1te has joined #openstack-swift | 20:42 | |
| *** lpabon has quit IRC | 20:49 | |
| openstackgerrit | John Dickinson proposed openstack/swift: Merge master to feature/ec https://review.openstack.org/167004 | 20:50 |
| openstackgerrit | Janie Richling proposed openstack/swift: Enable middleware to set metadata on object POST https://review.openstack.org/158401 | 20:53 |
| peluse | tdasilva, yeah I think that changed with a recent version of pyeclib | 20:55 |
| *** mahatic has quit IRC | 20:56 | |
| *** annegentle has quit IRC | 20:59 | |
| *** jamielennox|away is now known as jamielennox | 21:02 | |
| openstackgerrit | John Dickinson proposed openstack/swift: Merge master to feature/ec https://review.openstack.org/166558 | 21:03 |
| notmyname | bah!! | 21:03 |
| peluse | crazy huh? | 21:04 |
| notmyname | well, this one might wrok | 21:04 |
| notmyname | maybe | 21:04 |
| notmyname | so my goal is not to get the proper requirements. just to mark it so that requirements.txt isnt' changed so it doesn't trigger | 21:05 |
| notmyname | that maybe it's just a diff against the previous commit that matters (ie not against master) | 21:06 |
| peluse | yeah, that's what I gave up on a while ago :) | 21:06 |
| peluse | might be easier to get the requiements at the right version and have the jenkins issues addressed | 21:07 |
| torgomatic | mutable state is basically the root of all evil | 21:08 |
| notmyname | peluse: also, I have no idea why pep8 is failing. | 21:08 |
| mattoliverau | Morning | 21:11 |
| notmyname | hi mattoliverau | 21:11 |
| notmyname | hmm... `$ tox -epep8` works on saio, so I'm guessing it might be a side-effect of the requirements shenanigans | 21:12 |
| torgomatic | if I have to fix one more stupid bug caused by this big steaming pile of mutable WSGI environment getting passed back and forth and back and forth all over the place, I'm rewriting the whole thing in Clojure. | 21:12 |
| mattoliverau | Busy week! Let's go! | 21:12 |
| peluse | notmyname, yeah WFM too | 21:12 |
| notmyname | mattoliverau: typey typey! | 21:13 |
| mattoliverau | Lol, perfect use | 21:13 |
| *** annegentle has joined #openstack-swift | 21:14 | |
| notmyname | peluse: ya, I was hoping to sidestep requirements so it's not a blocker for the merge from master. I don't think it's hard to find the right values for requirements.txt, it's just that the job fails because it doesn't work on the feature branch | 21:14 |
| peluse | notmyname, we could just do a new patch to manually update feature/ec (w/o using merge) | 21:17 |
| notmyname | peluse: that sounds terrible ;-) | 21:17 |
| peluse | yeah, I typed before I thought :) | 21:17 |
| *** Fin1te has quit IRC | 21:17 | |
| os1 | peluse: Thank you! :) | 21:18 |
| peluse | no problem! | 21:20 |
| notmyname | peluse: nope. hasn't finished all the checks yet, but the requirements one is still busted | 21:23 |
| * notmyname takes a journey up the mountain to -infra to see if it can be fixed | 21:23 | |
| *** jrichli has quit IRC | 21:25 | |
| *** joeljwright has quit IRC | 21:29 | |
| notmyname | peluse: clarkb is helping out in -infra | 21:31 |
| peluse | ahh cool | 21:31 |
| notmyname | saying "feature/* should be treated like master" and looking in to what needs to change | 21:31 |
| *** tsg has joined #openstack-swift | 21:40 | |
| os1 | What are some "big item" work items, once EC work has been released? | 21:43 |
| os1 | coming up | 21:44 |
| os1 | Just out of curiosity :) | 21:45 |
| os1 | I'm wondering what other items could hold such a high importance as to have almost everybody focus on it, like the case of erasure codes. | 21:46 |
| os1 | In the near/far future. | 21:47 |
| zaitcev | I'm going to circle around with PBEs and Thiago is coming with the single-process mode. | 21:47 |
| zaitcev | But you may be sure that EC will take a year to settled down with its reconstructors, trailers, fragment argives and all other machinery. | 21:48 |
| zaitcev | Kilo is merely a beta. | 21:48 |
| *** joeljwright has joined #openstack-swift | 21:49 | |
| peluse | "merely a beta" ouch! | 21:49 |
| zaitcev | You know best of all how big a change it is. | 21:50 |
| peluse | but for sure we all agree its needs exposure/feedback/testing before we can stand behind it as production :) | 21:50 |
| mattoliverau | os1: you can take a look at the swift-specs repo to see some proposed/potential larger changes/additions to swift. | 21:53 |
| peluse | os1, just keep in mind that those are not all approved (so not to be confused with a roadmap or anything) | 21:54 |
| notmyname | os1: that's the fun problem of project management in an open source project. next priority is whatever people in the community work on :-) | 21:56 |
| peluse | that's the best part!! | 21:57 |
| notmyname | os1: of course, that' a little naive too. in reality, look at the specs repo. look at the ideas wiki page. those are some of the things that have been talked about that people are working on or have talked about | 21:57 |
| *** Nadeem has joined #openstack-swift | 21:58 | |
| * peluse wonders if notmyname just called him naive... | 21:58 | |
| notmyname | nah, it was a response to myself. you just type faster :-) | 21:58 |
| peluse | hey but if the shoe fits.... | 21:59 |
| notmyname | IMO it's naive to think there is zero prioritization in the community and it's all "whatever you want to do" | 21:59 |
| notmyname | peluse: is there anything on the master2ec merge that can't wait until tomorrow? -infra can hack it today if needed, or tomorrow it can be done right | 22:01 |
| notmyname | looks like tdasilva's patch hasn't landed yet, and that's all I think is on master that's required for ec | 22:01 |
| peluse | notmyname, yeah it can wait | 22:02 |
| notmyname | ok, thanks for confirming | 22:02 |
| peluse | yup, I like the sound of "done right" :) | 22:02 |
| peluse | FYI I'll be looking at the PUT refactor here shorlty as well.... | 22:02 |
| zaitcev | I'd like believe that we have certain accountability and/or responsibility to operators. In particular it makes sense for Box to desire for EC, so they help with manpower, and then this makes progress. | 22:04 |
| jest_____ | Quick question: let's say that i would want to choose saving data with EC, insted of 3-way replica for saving data with certain amount of robustness. How much more CPU usage I could expect on the system when reading/writing such data in comparison with current Swift mode using replicas? | 22:19 |
| torgomatic | those two statements are incompatible | 22:19 |
| torgomatic | "quick question" and the rest | 22:19 |
| notmyname | quick answer: "more" | 22:20 |
| clayg | surely you jest | 22:20 |
| torgomatic | no, I'm serious, and don't call me Shirley | 22:20 |
| torgomatic | that joke doesn't work nearly so well in text, sadly | 22:20 |
| notmyname | "surely you jest_____" FTFY | 22:21 |
| notmyname | jest_____: there isn't an answer to that question. it depends on what EC scheme, what EC library, what workload, what size of data, what hardware, etc | 22:22 |
| notmyname | jest_____: but at this point, all we really know is "EC takes more CPU than replicas". you definitely need to benchmark it (and otherwise test) before using it | 22:22 |
| notmyname | also, it's not even finished yet, so there's that :-) | 22:23 |
| jest_____ | torgomaic, notmyname: ok yes too general question. thank your comments. Maybe than this question, would you rather use EC just for strictly archival solution, Where also read are rare or at least not web-scaled like? | 22:26 |
| peluse | jest_____, so some of that answer will depend on where we wind up w/performance. EC is pretty popular for archival but it doesn't have to be that way if perf is good enough (or other cahing solutions are added for example to make up for latency) | 22:27 |
| peluse | good enough for the usage model in question that is.... | 22:28 |
| zaitcev | Consider that Microsoft Azure is purely EC based and never had a replication implementation to begin with. If Microsoft's claims are to be trusted, it has more objects than Amazon S3. What's not web-scale about it? | 22:31 |
| zaitcev | It's manifastly obvious that you can build a RAX-sized cluster with EC, the only small problem is to implement it right. | 22:31 |
| notmyname | zaitcev: any idea if they do a haystack-style scheme with large data slabs? | 22:31 |
| notmyname | "small problem" lol | 22:32 |
| zaitcev | hehe | 22:32 |
| notmyname | it's trivially easy to build AI. the only small problem is typing in the right things | 22:32 |
| peluse | well, the Azure implementation is a bit more complex than ours... | 22:33 |
| peluse | we're starting with basic RS, they do all sorts of cool stuff for optimizing perf, minimziing network overhead | 22:33 |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: support single ranges for GET requests https://review.openstack.org/163620 | 22:47 |
| *** erlon has quit IRC | 22:51 | |
| *** annegentle has quit IRC | 22:53 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: Support If-Match and If-None-Match on GET+HEAD https://review.openstack.org/166579 | 22:55 |
| *** joeljwright has quit IRC | 23:13 | |
| *** os1 has left #openstack-swift | 23:17 | |
| *** Nadeem has quit IRC | 23:22 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: multi-range GET support https://review.openstack.org/166576 | 23:23 |
| *** jest_____ has quit IRC | 23:25 | |
| *** chlong has joined #openstack-swift | 23:30 | |
| *** dmorita has joined #openstack-swift | 23:38 | |
| *** annegentle has joined #openstack-swift | 23:41 | |
| *** km has joined #openstack-swift | 23:46 | |
| openstackgerrit | Samuel Merritt proposed openstack/swift: EC: multi-range GET support https://review.openstack.org/166576 | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!