| *** tusharsg has quit IRC | 00:05 | |
| *** Midnightmyth has quit IRC | 00:09 | |
| *** dmorita has joined #openstack-swift | 00:25 | |
| openstackgerrit | A change was merged to openstack/swift: Fix "Handoff requested (N)" log spam https://review.openstack.org/109791 | 00:32 |
|---|---|---|
| *** bill_az has quit IRC | 00:40 | |
| *** cpen has quit IRC | 00:41 | |
| *** mitz has quit IRC | 00:43 | |
| *** tsg has joined #openstack-swift | 00:44 | |
| *** shri has left #openstack-swift | 00:47 | |
| *** mitz has joined #openstack-swift | 00:48 | |
| *** addnull has joined #openstack-swift | 00:56 | |
| *** addnull has quit IRC | 00:58 | |
| *** addnull has joined #openstack-swift | 00:59 | |
| *** mitz has quit IRC | 01:00 | |
| *** mitz has joined #openstack-swift | 01:03 | |
| *** mitz has quit IRC | 01:04 | |
| *** mitz has joined #openstack-swift | 01:06 | |
| *** mitz has quit IRC | 01:17 | |
| *** addnull has quit IRC | 01:21 | |
| *** addnull has joined #openstack-swift | 01:21 | |
| *** tsg has quit IRC | 01:23 | |
| *** mrmoje has quit IRC | 01:23 | |
| *** addnull has quit IRC | 01:26 | |
| *** judd7 has quit IRC | 01:28 | |
| *** nosnos has joined #openstack-swift | 01:38 | |
| *** addnull has joined #openstack-swift | 01:43 | |
| *** haomaiwang has joined #openstack-swift | 01:43 | |
| *** haomaiwang has quit IRC | 01:44 | |
| *** haomaiwang has joined #openstack-swift | 01:45 | |
| *** haomaiw__ has joined #openstack-swift | 01:58 | |
| *** haomaiwang has quit IRC | 02:01 | |
| *** tsg has joined #openstack-swift | 02:11 | |
| *** jyoti-ranjan has joined #openstack-swift | 03:15 | |
| *** abhirc has joined #openstack-swift | 03:17 | |
| *** morganfainberg is now known as morganfainberg_Z | 03:20 | |
| *** morganfainberg_Z is now known as morganfainberg | 03:22 | |
| *** gyee has quit IRC | 03:23 | |
| *** nosnos has quit IRC | 03:28 | |
| *** addnull has quit IRC | 03:46 | |
| *** addnull has joined #openstack-swift | 03:47 | |
| *** addnull has quit IRC | 03:51 | |
| *** nosnos has joined #openstack-swift | 04:03 | |
| *** ppai has joined #openstack-swift | 04:04 | |
| *** jyoti-ranjan has quit IRC | 04:05 | |
| *** haomaiw__ has quit IRC | 04:09 | |
| *** haomaiwang has joined #openstack-swift | 04:10 | |
| *** Shivani has quit IRC | 04:17 | |
| *** addnull has joined #openstack-swift | 04:17 | |
| *** addnull has quit IRC | 04:22 | |
| *** haomaiwang has quit IRC | 04:29 | |
| *** haomaiwa_ has joined #openstack-swift | 04:29 | |
| allyunion | Anyone use Puppet with Swift OpenStack? | 04:41 |
| allyunion | Er, OpenStack Swift. | 04:41 |
| openstackgerrit | Tushar Gohad proposed a change to openstack/swift: EC: Add support for policy types, 'erasure_coding' policy https://review.openstack.org/103644 | 04:42 |
| *** addnull has joined #openstack-swift | 04:51 | |
| *** tusharsg has joined #openstack-swift | 04:58 | |
| *** morganfainberg is now known as morganfainberg_Z | 04:59 | |
| *** tsg has quit IRC | 05:01 | |
| *** haomaiwa_ has quit IRC | 05:04 | |
| *** haomaiwang has joined #openstack-swift | 05:04 | |
| *** kopparam has joined #openstack-swift | 05:14 | |
| *** haomaiw__ has joined #openstack-swift | 05:14 | |
| *** haomaiwang has quit IRC | 05:17 | |
| *** jyoti-ranjan has joined #openstack-swift | 05:40 | |
| *** kopparam_ has joined #openstack-swift | 05:42 | |
| *** chandankumar has joined #openstack-swift | 05:44 | |
| *** kopparam has quit IRC | 05:44 | |
| *** tusharsg has quit IRC | 05:45 | |
| mattoliverau | notmyname, acoles_away, torgomatic, portante, clayg_: (and anyone else I've missed or want to voice an opinion) can you all take a look at the subject and email template I've put together and can I get your feedback or corrections on this etherpad: https://etherpad.openstack.org/p/abandoner_email | 06:09 |
| mattoliverau | when your all up and at work of course :) | 06:10 |
| fifieldt_ | mattoliverau, some new contributors might not know what rebase is | 06:11 |
| mattoliverau | fifieldt_: good call :) | 06:11 |
| fifieldt_ | I'd also add a stanza about how friendly the swift team is | 06:12 |
| fifieldt_ | and where they can go for help if they need it | 06:12 |
| fifieldt_ | I normally link to https://wiki.openstack.org/IRC and https://wiki.openstack.org/wiki/MailingLists#Future_Development | 06:13 |
| fifieldt_ | but, otherwise | 06:13 |
| fifieldt_ | love your work :) | 06:13 |
| mattoliverau | fifieldt_: thanks I appreciate you taking a look!! | 06:14 |
| *** fifieldt_ is now known as fifieldt | 06:15 | |
| *** haomaiw__ has quit IRC | 06:19 | |
| *** fifieldt has quit IRC | 06:19 | |
| *** haomaiwang has joined #openstack-swift | 06:19 | |
| *** ho has joined #openstack-swift | 06:30 | |
| *** haomaiwang has quit IRC | 06:35 | |
| *** haomaiwang has joined #openstack-swift | 06:36 | |
| *** bvandenh has joined #openstack-swift | 06:38 | |
| *** haomaiw__ has joined #openstack-swift | 06:42 | |
| *** haomaiwang has quit IRC | 06:42 | |
| *** haomai___ has joined #openstack-swift | 06:58 | |
| *** haomaiw__ has quit IRC | 07:02 | |
| *** psharma has joined #openstack-swift | 07:19 | |
| *** astellwag|afk is now known as astellwag | 07:19 | |
| *** fifieldt has joined #openstack-swift | 07:23 | |
| *** nexusz99 has joined #openstack-swift | 07:35 | |
| nexusz99 | I have question. Is there any request for deleting dynamic large object both manifest file and segments? if i wan't to delete dynamic large object, i have to know all segments and request delete one by one?? | 07:41 |
| *** leopoldj has joined #openstack-swift | 08:01 | |
| *** haomai___ has quit IRC | 08:08 | |
| *** haomaiwang has joined #openstack-swift | 08:09 | |
| *** mrmoje has joined #openstack-swift | 08:14 | |
| *** mkollaro has joined #openstack-swift | 08:16 | |
| *** mmcardle has joined #openstack-swift | 08:18 | |
| *** haomaiw__ has joined #openstack-swift | 08:24 | |
| *** haomaiwang has quit IRC | 08:26 | |
| *** jyoti-ranjan has quit IRC | 08:47 | |
| *** ho has quit IRC | 08:49 | |
| *** nosnos has quit IRC | 08:49 | |
| *** marzif has joined #openstack-swift | 09:01 | |
| *** openstackgerrit has quit IRC | 09:16 | |
| *** openstackgerrit has joined #openstack-swift | 09:17 | |
| *** jyoti-ranjan has joined #openstack-swift | 09:25 | |
| *** nosnos has joined #openstack-swift | 09:30 | |
| *** infotection has quit IRC | 09:38 | |
| *** sungju has joined #openstack-swift | 09:42 | |
| *** sungju has quit IRC | 09:43 | |
| *** sts_ has joined #openstack-swift | 09:44 | |
| *** j_king_ has joined #openstack-swift | 09:44 | |
| *** occupant has quit IRC | 09:44 | |
| *** wer has quit IRC | 09:44 | |
| *** amandap has quit IRC | 09:44 | |
| *** MooingLemur has quit IRC | 09:44 | |
| *** thurloat has quit IRC | 09:44 | |
| *** bsdkurt has quit IRC | 09:44 | |
| *** creiht has quit IRC | 09:44 | |
| *** jd__ has quit IRC | 09:44 | |
| *** sts has quit IRC | 09:44 | |
| *** anderstj has quit IRC | 09:44 | |
| *** alpha_ori has quit IRC | 09:44 | |
| *** j_king has quit IRC | 09:44 | |
| *** torgomatic has quit IRC | 09:44 | |
| *** hurricanerix has quit IRC | 09:44 | |
| *** briancline has quit IRC | 09:44 | |
| *** torgomatic has joined #openstack-swift | 09:44 | |
| *** MooingLe1ur has joined #openstack-swift | 09:44 | |
| *** MooingLe1ur has quit IRC | 09:44 | |
| *** MooingLe1ur has joined #openstack-swift | 09:44 | |
| *** ChanServ sets mode: +v torgomatic | 09:44 | |
| *** occupant has joined #openstack-swift | 09:44 | |
| *** wer has joined #openstack-swift | 09:44 | |
| *** jd__ has joined #openstack-swift | 09:44 | |
| *** bsdkurt has joined #openstack-swift | 09:44 | |
| *** briancline has joined #openstack-swift | 09:44 | |
| *** amandap has joined #openstack-swift | 09:45 | |
| *** hurricanerix has joined #openstack-swift | 09:45 | |
| *** thurloat has joined #openstack-swift | 09:45 | |
| *** nexusz99 has quit IRC | 09:45 | |
| *** infotection has joined #openstack-swift | 09:45 | |
| *** creiht_ has joined #openstack-swift | 09:45 | |
| *** anderstj has joined #openstack-swift | 09:45 | |
| *** nexusz99 has joined #openstack-swift | 09:45 | |
| *** alpha_ori has joined #openstack-swift | 09:46 | |
| *** nexusz99 has quit IRC | 09:50 | |
| *** infotection has quit IRC | 09:52 | |
| *** infotection has joined #openstack-swift | 09:53 | |
| *** nexusz99 has joined #openstack-swift | 09:59 | |
| *** pconstantine has quit IRC | 09:59 | |
| *** Midnightmyth has joined #openstack-swift | 09:59 | |
| *** pconstantine has joined #openstack-swift | 10:00 | |
| *** infotection has quit IRC | 10:05 | |
| *** pconstantine has quit IRC | 10:05 | |
| *** infotection has joined #openstack-swift | 10:10 | |
| *** mmcardle has quit IRC | 10:18 | |
| *** pconstantine has joined #openstack-swift | 10:19 | |
| *** infotection has quit IRC | 10:19 | |
| *** ppai has quit IRC | 10:20 | |
| *** nosnos has quit IRC | 10:22 | |
| *** infotection has joined #openstack-swift | 10:23 | |
| *** mmcardle has joined #openstack-swift | 10:25 | |
| *** ppai has joined #openstack-swift | 10:33 | |
| *** jyoti-ranjan has quit IRC | 10:43 | |
| *** jyoti-ranjan has joined #openstack-swift | 10:43 | |
| *** mkollaro has quit IRC | 10:48 | |
| *** devpadak has joined #openstack-swift | 10:49 | |
| *** nexusz99 has quit IRC | 10:49 | |
| *** mkollaro has joined #openstack-swift | 11:07 | |
| *** mmcardle has quit IRC | 11:14 | |
| *** aswadr has joined #openstack-swift | 11:22 | |
| *** mitz has joined #openstack-swift | 11:27 | |
| *** mitz has quit IRC | 11:28 | |
| *** mitz has joined #openstack-swift | 11:30 | |
| *** mitz has quit IRC | 11:31 | |
| *** mitz has joined #openstack-swift | 11:33 | |
| *** kopparam has joined #openstack-swift | 11:34 | |
| *** kopparam_ has quit IRC | 11:36 | |
| *** ondergetekende has quit IRC | 11:46 | |
| *** ondergetekende has joined #openstack-swift | 11:48 | |
| *** ondergetekende has quit IRC | 11:53 | |
| *** ondergetekende has joined #openstack-swift | 11:54 | |
| *** mmcardle has joined #openstack-swift | 12:14 | |
| *** mrsnivvel has quit IRC | 12:16 | |
| *** tdasilva has quit IRC | 12:22 | |
| *** zslot has joined #openstack-swift | 12:25 | |
| *** mmcardle has quit IRC | 12:26 | |
| *** mmcardle has joined #openstack-swift | 12:28 | |
| *** mrsnivvel has joined #openstack-swift | 12:29 | |
| *** dmorita has quit IRC | 12:32 | |
| *** mrsnivvel has quit IRC | 12:37 | |
| *** addnull has quit IRC | 12:39 | |
| *** ppai has quit IRC | 12:47 | |
| *** mrsnivvel has joined #openstack-swift | 12:56 | |
| *** kopparam has quit IRC | 12:57 | |
| *** devpadak has quit IRC | 12:58 | |
| *** bvandenh has quit IRC | 13:02 | |
| *** bill_az has joined #openstack-swift | 13:12 | |
| *** bvandenh has joined #openstack-swift | 13:17 | |
| *** mmcardle has quit IRC | 13:17 | |
| *** psharma has quit IRC | 13:25 | |
| *** mmcardle has joined #openstack-swift | 13:26 | |
| *** devpadak has joined #openstack-swift | 13:30 | |
| *** nexusz99 has joined #openstack-swift | 13:30 | |
| *** mmcardle has quit IRC | 13:39 | |
| *** mmcardle has joined #openstack-swift | 13:40 | |
| *** tdasilva has joined #openstack-swift | 13:51 | |
| *** judd7 has joined #openstack-swift | 13:53 | |
| *** kopparam has joined #openstack-swift | 13:53 | |
| *** dmsimard_away is now known as dmsimard | 13:57 | |
| *** kopparam has quit IRC | 13:58 | |
| *** kopparam has joined #openstack-swift | 13:58 | |
| *** tsg has joined #openstack-swift | 14:01 | |
| *** kopparam has quit IRC | 14:03 | |
| *** Wimpers has joined #openstack-swift | 14:04 | |
| *** mmcardle has quit IRC | 14:11 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 14:12 |
| *** tsg has quit IRC | 14:20 | |
| *** vr1 has joined #openstack-swift | 14:21 | |
| *** zul has joined #openstack-swift | 14:23 | |
| *** ujjain has quit IRC | 14:25 | |
| *** ujjain has joined #openstack-swift | 14:27 | |
| *** SHiker has joined #openstack-swift | 14:29 | |
| SHiker | If I configure swift to only maintain 2 copies of an object, will it still only return a successful PUT if both copies are written or will it return success if one copy is written to object server? | 14:30 |
| *** tsg has joined #openstack-swift | 14:40 | |
| *** abhirc has quit IRC | 14:41 | |
| *** bvandenh has quit IRC | 14:43 | |
| *** zul has quit IRC | 14:47 | |
| *** mmcardle has joined #openstack-swift | 14:49 | |
| *** chandankumar_ has joined #openstack-swift | 14:50 | |
| pconstantine | SHiker: return (n // 2) + 1 | 14:51 |
| *** marcusvrn has quit IRC | 14:51 | |
| *** chandankumar has quit IRC | 14:53 | |
| SHiker | PConstantine: Thanks. | 15:04 |
| *** SHiker has quit IRC | 15:05 | |
| *** tusharsg has joined #openstack-swift | 15:06 | |
| *** tsg has quit IRC | 15:09 | |
| *** tsg has joined #openstack-swift | 15:10 | |
| *** tusharsg has quit IRC | 15:11 | |
| *** mwstorer has joined #openstack-swift | 15:13 | |
| *** zul has joined #openstack-swift | 15:17 | |
| *** kevinc_ has joined #openstack-swift | 15:18 | |
| *** astellwag is now known as astellwag|afk | 15:19 | |
| nexusz99 | I have question. Is there any request for deleting dynamic large object both manifest file and segments? if i wan't to delete dynamic large object, i have to know all segments and request delete one by one?? | 15:22 |
| *** gyee has joined #openstack-swift | 15:28 | |
| *** mmcardle has quit IRC | 15:28 | |
| pconstantine | nexusz99: http://docs.openstack.org/developer/swift/middleware.html#module-swift.common.middleware.bulk | 15:31 |
| *** mmcardle has joined #openstack-swift | 15:31 | |
| dfg_ | nexusz99: yes. static large objects has a delete-all function but not dynamic | 15:33 |
| vr1 | hi all, we diskfile backend developers, how are we supposed to manage underlying I/O exceptions to the upper layers ? | 15:34 |
| pconstantine | vr1: you can raise one of the HTTPException classes, obj-server will catch it and create a proper response | 15:38 |
| vr1 | pconstantine: thx | 15:39 |
| pconstantine | vr1: at least that's what I do, works quite well | 15:39 |
| *** Wimpers has quit IRC | 15:39 | |
| *** tdasilva has quit IRC | 15:45 | |
| *** chandankumar_ has quit IRC | 15:45 | |
| *** mrmoje has quit IRC | 15:47 | |
| *** kevinc_ has quit IRC | 15:47 | |
| *** MooingLe1ur is now known as MooingLemur | 15:50 | |
| *** gyee has quit IRC | 15:51 | |
| nexusz99 | dfg_: Thanks. I use s3multi with swift3. The s3multi implements multipart upload by Dynamic large object.. So. i have no idea how to delete multipart object.. | 15:54 |
| *** kevinc_ has joined #openstack-swift | 15:56 | |
| *** tdasilva has joined #openstack-swift | 15:58 | |
| nexusz99 | pconstantine: Thanks. I will refer to your opinion. | 16:03 |
| pconstantine | nexusz99: you can list parts of multipart object using s3multi and then delete all of them with bulk delete or one by one | 16:05 |
| *** gyee has joined #openstack-swift | 16:05 | |
| dfg_ | nexusz99: ya, what pconstantine said. you can get the listing by doing a HEAD on the DLO object, getting the X-Object-Manifest header, and doing a container listing with prefix using that header | 16:07 |
| *** devpadak has quit IRC | 16:08 | |
| *** gyee has quit IRC | 16:09 | |
| nexusz99 | pconstantine: dfg_: thank you for your answer | 16:11 |
| *** cutforth has joined #openstack-swift | 16:12 | |
| *** pberis has quit IRC | 16:21 | |
| *** pberis has joined #openstack-swift | 16:21 | |
| *** nexusz99 has quit IRC | 16:29 | |
| *** pberis has quit IRC | 16:30 | |
| *** vr1 has left #openstack-swift | 16:31 | |
| openstackgerrit | Tushar Gohad proposed a change to openstack/swift: EC: Add support for policy types, 'erasure_coding' policy https://review.openstack.org/103644 | 16:32 |
| *** pberis has joined #openstack-swift | 16:33 | |
| *** jyoti-ranjan has quit IRC | 16:34 | |
| *** shri has joined #openstack-swift | 16:40 | |
| *** pberis has quit IRC | 16:44 | |
| *** pberis1 has joined #openstack-swift | 16:44 | |
| notmyname | good morning | 16:48 |
| *** pberis has joined #openstack-swift | 16:52 | |
| *** cpen has joined #openstack-swift | 16:53 | |
| *** pberis1 has quit IRC | 16:55 | |
| *** peluse_ has quit IRC | 16:55 | |
| *** peluse_ has joined #openstack-swift | 16:57 | |
| peluse_ | good morning! | 16:57 |
| notmyname | great openstack reactions post: "trying to use devstack" http://openstackreactions.enovance.com/2014/07/trying-to-use-devstack/ | 16:59 |
| *** leopoldj has quit IRC | 17:06 | |
| cpen | I’ve got a question about setting ACLs on the object level | 17:07 |
| peluse_ | notmyname: heh, I like that little clip | 17:08 |
| *** morganfainberg_Z is now known as morganfainberg | 17:09 | |
| chmouel | notmyname: lol | 17:14 |
| *** bvandenh has joined #openstack-swift | 17:14 | |
| chmouel | /this is posted from a devstack core devs :p/ | 17:15 |
| chmouel | but that's not as bad :) | 17:15 |
| *** astellwag|afk is now known as astellwag | 17:19 | |
| cpen | it seems like Swift doesn’t support object-level ACLs. I first wanted to make sure that I wasn’t missing something. but presuming that’s true, I was wondering if there are any thoughts on why that’s the case | 17:20 |
| *** bill_az has quit IRC | 17:20 | |
| *** mkollaro has quit IRC | 17:22 | |
| *** pberis has quit IRC | 17:25 | |
| *** pberis has joined #openstack-swift | 17:25 | |
| *** abhirc has joined #openstack-swift | 17:30 | |
| *** cutforth has quit IRC | 17:31 | |
| *** pberis1 has joined #openstack-swift | 17:32 | |
| *** pberis has quit IRC | 17:32 | |
| notmyname | cpen: that is correct. swift doesn't support object ACLs | 17:34 |
| cpen | bummer :[ | 17:34 |
| cpen | are there any conference notes or blog posts, etc. that discuss this design decision? | 17:35 |
| cpen | I tried looking around before coming here, but couldn’t find anything | 17:35 |
| portante | notmyname: swift would support object acls if you put one object per container. :) | 17:36 |
| cpen | :] | 17:36 |
| notmyname | cpen: the real reason (as with all things programming) is because nobody has written it. the "design decision" is because almost everything in swift is grouped on containers (eg ACLs today, static web, versioning, storage policies) | 17:37 |
| *** mmcardle1 has joined #openstack-swift | 17:37 | |
| cpen | that makes sense | 17:37 |
| *** chandankumar_ has joined #openstack-swift | 17:38 | |
| cpen | I’m a bit surprised, actually. S3 has supported object-level ACLs since its inception, if I recall correctly. they’ve been around for at least a long time. | 17:38 |
| notmyname | mattoliverau: I made a couple of edits to the etherpad email tempate | 17:39 |
| *** mmcardle has quit IRC | 17:39 | |
| cpen | in any case, I may implement a piece of middleware implementing some of this at some point in the future | 17:39 |
| cpen | I’ll make sure to mention it here if/when I do! | 17:39 |
| notmyname | cpen: why in middleware? | 17:41 |
| cpen | easiest place to implement for my organization | 17:41 |
| cpen | not really ready to start hacking on swift itself | 17:41 |
| notmyname | cpen: if you're considering writing the functionality, maybe you could add it to the proxy server itself | 17:41 |
| cpen | yeah, I can see how that might be a superior implementation | 17:42 |
| cpen | I’ll keep it in mind | 17:42 |
| notmyname | cpen: are you using swift as an internal s3 replacement? | 17:43 |
| cpen | nah. s3 was never used where I’m working now. but I’ve used it in past gigs. | 17:43 |
| *** tsg has quit IRC | 17:43 | |
| notmyname | ah ok | 17:43 |
| *** anderstj has quit IRC | 17:44 | |
| *** amandap has quit IRC | 17:44 | |
| *** wer has quit IRC | 17:44 | |
| *** occupant has quit IRC | 17:44 | |
| *** anderstj has joined #openstack-swift | 17:45 | |
| *** occupant has joined #openstack-swift | 17:45 | |
| *** amandap has joined #openstack-swift | 17:45 | |
| notmyname | oh yeah. swift team meeting in 1.25 hours (at 1900utc) in #openstack-meeting | 17:48 |
| *** abhirc has quit IRC | 17:51 | |
| *** morganfainberg is now known as morganfainberg_Z | 17:51 | |
| pconstantine | cpen: container acls are cached in memcache, memcaching object acls will put a lot of pressure on memcache, and not caching them may lead to unessesary reads and object-server disk cache eviction | 17:53 |
| pconstantine | now when I'm thinking about it, using memcache on object servers actually can lead to a better performance, as disk cache is highly ineffective without proper hints, and not too many hints are at the developer disposal | 17:56 |
| cpen | yeah, I suspected that’s why container acl management is done in the proxy server in the first place | 17:56 |
| *** tsg has joined #openstack-swift | 17:56 | |
| pconstantine | cpen: but, extending container db to hold object metadata, can solve the problem quite easily | 17:56 |
| cpen | if I was going to do it with middleware, I first thought of doing it on the object server anyway, but was then thinking that I should do it on the proxy server to properly integrate with the auth pipeline there | 17:56 |
| cpen | yeah, that’s another solution | 17:57 |
| cpen | more complex, of course | 17:57 |
| pconstantine | cpen: not complex at all, like 10 lines of code :) | 17:57 |
| notmyname | I don't think you want to add more per-object info to the container DB though. they can already get quite large. also, the migration story would be harder | 17:57 |
| notmyname | and as cpen mentioned, doing it on a storage node makes it harder because of the auth integration | 17:58 |
| cpen | complex in the sense of having to consider performance & resource implications, etc. | 17:58 |
| pconstantine | notmyname: dunno, it's quite a lot of info anyway, and sqlite uses full pages if I'm not mistaken | 17:58 |
| pconstantine | notmyname: using object-level metadata is easy here, just short read on GET and properly kill it on object server | 18:00 |
| notmyname | it's not that a particular object will store a lot of info for ACLs, it's that a container has millions (or more) of them | 18:00 |
| notmyname | pconstantine: you could do it in the proxy server and not cache it. as soon as the headers are read, do the ACL validation and close the internal connection before streaming the data. | 18:02 |
| notmyname | pconstantine: but as you mentioned, if there are a lot of requests that are denied, then this becomes quite taxing on the cluster as a whole | 18:02 |
| pconstantine | pconstantine: that's the "short read" I'm referring to, you will get errors in logs :) | 18:02 |
| pconstantine | notmyname: AFAIK, sqlite will use at least 1k per row in object table anyway | 18:03 |
| pconstantine | so, if you have less than 1k data - you're not changing anything | 18:03 |
| cpen | I was initially consdiering just using object-level metadata for an initial implementation, and doing it on the object server for better distribution of work. but for a complete implementation, you want full auth integration, so that would necessitate doing it on the proxy server, as best I understand | 18:04 |
| *** gyee has joined #openstack-swift | 18:04 | |
| notmyname | cpen: ya, that sounds correct | 18:04 |
| *** gyee has quit IRC | 18:05 | |
| *** gyee has joined #openstack-swift | 18:05 | |
| pconstantine | notmyname: yep, most certainly Swift is using 1024 byte pages and never vacuumint, which means that each row in the object table is 1024 bytes | 18:12 |
| *** bvandenh has quit IRC | 18:13 | |
| *** astellwag is now known as astellwag|afk | 18:14 | |
| torgomatic | notmyname: heh, thanks for writing some positive words after my mailing list message :) | 18:16 |
| notmyname | torgomatic: :-) | 18:17 |
| *** gyee has quit IRC | 18:18 | |
| *** mmcardle1 has quit IRC | 18:19 | |
| *** gyee has joined #openstack-swift | 18:19 | |
| *** cutforth has joined #openstack-swift | 18:21 | |
| *** zslot has quit IRC | 18:23 | |
| *** bvandenh has joined #openstack-swift | 18:28 | |
| *** chandankumar_ has quit IRC | 18:33 | |
| *** tdasilva has quit IRC | 18:34 | |
| *** tdasilva has joined #openstack-swift | 18:37 | |
| *** kevinc_ has quit IRC | 18:43 | |
| *** gvernik has joined #openstack-swift | 18:47 | |
| *** tdasilva_ has joined #openstack-swift | 18:59 | |
| mattoliverau | notmyname: thanks I'll take a look post meeting. | 19:00 |
| notmyname | yup, it's meeting time | 19:00 |
| *** ChanServ sets mode: +v dfg_ | 19:02 | |
| *** tdasilva has quit IRC | 19:03 | |
| *** zul has quit IRC | 19:04 | |
| *** mrsnivvel has quit IRC | 19:05 | |
| *** mrsnivvel has joined #openstack-swift | 19:07 | |
| *** ophuk has joined #openstack-swift | 19:09 | |
| *** creiht_ is now known as creiht | 19:09 | |
| ophuk | when using swift to upload an object you have an option for object-threads to upload said objects with multiple threads. Is there an equivalent for swiftclient? | 19:12 |
| torgomatic | ophuk: the command-line tool "swift" is swiftclient | 19:12 |
| ophuk | torgomatic: that's what I thought but I couldn't figure out how to tell it to use multiple threads in my script | 19:13 |
| *** kevinc_ has joined #openstack-swift | 19:14 | |
| ophuk | torgomatic: I import swiftclient.client as swiftClient, and when I upload it I use swiftClient.put_object(SWIFT_ENDPOINT, TOKEN, contName, fileName, file_object, None, None, None, None, None, None, None, None, None) - I couldn't find an option for that one | 19:15 |
| ophuk | am I just looking in the wrong spot? | 19:16 |
| torgomatic | ophuk: yes... there's some work going on to make that stuff reusable, but at the moment, swiftclient only has lower-level single-threaded operations | 19:16 |
| torgomatic | if you need concurrency, it's roll-your-own time | 19:17 |
| ophuk | torgomatic: gotcha, so I'm not going crazy:) | 19:17 |
| torgomatic | ophuk: https://review.openstack.org/#/c/85453/ | 19:17 |
| ophuk | thought I was just missing something | 19:17 |
| ophuk | torgomatic: so it is being worked on, excellent. Thanks for your help! | 19:21 |
| torgomatic | ophuk: no problem; if you can, you should try taking that patch for a test drive and see how you like the API and if it works for you, then leave feedback | 19:22 |
| torgomatic | that'll help it land faster :) | 19:22 |
| *** aswadr has quit IRC | 19:22 | |
| *** elambert has joined #openstack-swift | 19:30 | |
| *** bvandenh has quit IRC | 19:32 | |
| *** mmcardle has joined #openstack-swift | 19:37 | |
| *** tsg has quit IRC | 20:00 | |
| mattoliverau | Well I'm going back to bed for the last hour or so of the night, bbl | 20:00 |
| *** tdasilva_ has quit IRC | 20:00 | |
| *** tsg has joined #openstack-swift | 20:00 | |
| *** gvernik has quit IRC | 20:02 | |
| *** tusharsg has joined #openstack-swift | 20:05 | |
| *** gvernik has joined #openstack-swift | 20:07 | |
| *** tsg has quit IRC | 20:08 | |
| *** gvernik has left #openstack-swift | 20:15 | |
| *** elambert has quit IRC | 20:55 | |
| *** cpen has quit IRC | 20:58 | |
| *** tusharsg has quit IRC | 21:06 | |
| *** tsg has joined #openstack-swift | 21:06 | |
| *** mmcardle has quit IRC | 21:27 | |
| *** zaitcev has joined #openstack-swift | 21:36 | |
| *** ChanServ sets mode: +v zaitcev | 21:36 | |
| *** abhirc has joined #openstack-swift | 21:58 | |
| *** kevinc_ has quit IRC | 22:02 | |
| mattoliverau | Morning all | 22:07 |
| *** dmsimard is now known as dmsimard_away | 22:20 | |
| *** ophuk has quit IRC | 22:21 | |
| *** cpen has joined #openstack-swift | 22:25 | |
| *** Midnightmyth has quit IRC | 22:33 | |
| *** elambert has joined #openstack-swift | 22:38 | |
| notmyname | blaaaaaaaaaaargh | 22:45 |
| peluse_ | bless you | 22:46 |
| notmyname | want to hear a funny story? | 22:46 |
| peluse_ | sure | 22:46 |
| notmyname | so my probe tests on my SAIo have been borked | 22:47 |
| peluse_ | still?? | 22:47 |
| notmyname | and I've been trying to figure out why | 22:47 |
| notmyname | so through a lot of printf debugging and strace and netstat and etc etc | 22:47 |
| notmyname | I found that the error is that when trying to start up a server it was getting EADDRINUSE when trying to bind | 22:48 |
| notmyname | but, here's the funny part, **there was nothing listening on that port** | 22:48 |
| notmyname | spooky | 22:48 |
| peluse_ | still a mystery? | 22:48 |
| notmyname | no. here's what happened | 22:49 |
| notmyname | the check_server call for the proxy does a head request to the account (actually this seems unecessary) | 22:49 |
| notmyname | but anyway it was doing it | 22:49 |
| notmyname | well, that left a half-open connection to the account server, so there was a socket in TIME WAIT on that port. so the new server couldn't bind | 22:50 |
| notmyname | here's the fun part | 22:50 |
| notmyname | it turns out that we use the eventlet.listen convenience function to bind to the port | 22:50 |
| notmyname | and in eventlet 0.9.16 and on, it automatically sets SO_REUSEADDR | 22:51 |
| notmyname | but, our requirements have 0.9.15 | 22:51 |
| notmyname | and because of other stuff like this in the past, I've taken to running the earliest versions of our dependencies | 22:51 |
| notmyname | so, eventlet 0.9.15 doesn't do the sockopt | 22:51 |
| notmyname | but you can't pass that in to eventlet.listen | 22:52 |
| notmyname | so, you have to set the sysctl net.ipv4.tcp_tw_recycle = 1 | 22:52 |
| peluse_ | sounds like you've had fun | 22:52 |
| notmyname | so, the summary is: if you are using the oldest supported version of eventlet, and you haven't set the sysctl stuff to reuse timewait sockets, then probetests don't work | 22:52 |
| notmyname | and you'd never see it if you are running newer eventlet or had set the sysctl | 22:53 |
| notmyname | fun! | 22:53 |
| peluse_ | I figured out my keystone swift issue too... nowhere near the excisting story though | 22:53 |
| notmyname | and thank you torgomatic for helping me finish debugging it | 22:53 |
| notmyname | what'd you find on the keystone stuff? | 22:54 |
| peluse_ | I've been chasing 401 issues for a few days as I could sworn I had it working til a reboot... my keystone server had time drifted from my swift server (VMs) and auth_token.py was calling it expired... | 22:54 |
| notmyname | oh wow | 22:54 |
| notmyname | that's another hard one to debug | 22:55 |
| peluse_ | like no helpful log messages whatsoever, the messges I was getting had me chasing 6 differnt leads that had nothing to do with it :) | 22:55 |
| peluse_ | but it was fun (sort of) | 22:55 |
| peluse_ | now maybe tomorrow I can look at alcoles v3 patches - finally | 22:57 |
| notmyname | ugh. just ran ./.probetests after resetting my codebase back to master. looks like some errors still. wonder what now | 22:57 |
| *** cutforth has quit IRC | 22:58 | |
| peluse_ | check your date/time :) | 22:59 |
| notmyname | heh | 22:59 |
| peluse_ | BTW, I did notice one failing probe test that I hit sometimes if I forget to unset http_proxy... | 23:00 |
| mattoliverau | notmyname: is it time to up the requirements to 0.9.16 then :) | 23:07 |
| torgomatic | precise is 0.9.16... maybe it's RHEL 6 that's a little older? | 23:08 |
| torgomatic | I think there's a distro with 0.9.15 and that's why it's what we use | 23:08 |
| notmyname | torgomatic: lucid IIRC | 23:09 |
| torgomatic | notmyname: I thought lucid's was even way older than that... could be wrong though | 23:09 |
| notmyname | torgomatic: hmm, maybe so | 23:11 |
| notmyname | https://gist.github.com/notmyname/45e6a9e32822c8f98ddd | 23:12 |
| notmyname | wait. why do we have 0.9.15 as the requirement? | 23:18 |
| notmyname | precise has 0.9.16 https://gist.github.com/notmyname/f48719565fbdc1ee1364 | 23:19 |
| notmyname | and lucid actually doesn't have it at all | 23:19 |
| notmyname | maybe RHEL | 23:20 |
| notmyname | portante: zaitcev: ^^? | 23:20 |
| zaitcev | gimme a moment | 23:21 |
| zaitcev | [root@rhev-a24c-01 ~]# rpm -q python-eventlet | 23:22 |
| zaitcev | python-eventlet-0.9.17-2.el6.noarch | 23:22 |
| notmyname | hmm | 23:22 |
| *** dmsimard_away is now known as dmsimard | 23:22 | |
| zaitcev | RHEL 6 GA might be older, but 6.4 is that | 23:22 |
| torgomatic | well, time to bump it to 0.9.16 | 23:34 |
| *** dmsimard is now known as dmsimard_away | 23:37 | |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: bumped the eventlet minimum version to 0.9.16 https://review.openstack.org/110815 | 23:37 |
| notmyname | torgomatic: ^^ | 23:37 |
| notmyname | now let's see if we're allowed to do that (the global requirements has >=0.13 | 23:40 |
| notmyname | last issue with the probe tests is that I got unlucky with the randomness, and one of my policies with 3.25 replicas messed it up | 23:50 |
| torgomatic | oh, fractional replicas... such fun | 23:53 |
| *** gyee has quit IRC | 23:55 | |
| *** morganfainberg_Z is now known as morganfainberg | 23:55 | |
| notmyname | woohoo! probetests actually passed | 23:56 |
| notmyname | torgomatic: ya, one of those "other things" I've tried to keep around to break things | 23:57 |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: Multithread optimization for object updater https://review.openstack.org/75941 | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!