| mattoliverau | notmyname: nice! | 00:01 |
|---|---|---|
| notmyname | the "starred by any core reviewer", in my mind, is "what are you thinking about and looking at". ie more for me :-) | 00:02 |
| *** ho has joined #openstack-swift | 00:03 | |
| *** cl__ has joined #openstack-swift | 00:06 | |
| *** annegentle has quit IRC | 00:14 | |
| *** chlong has quit IRC | 00:15 | |
| *** zaitcev has joined #openstack-swift | 00:20 | |
| *** ChanServ sets mode: +v zaitcev | 00:20 | |
| ho | good morning guys | 00:20 |
| mattoliverau | ho: morning | 00:22 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-swiftclient: Updated from global requirements https://review.openstack.org/89250 | 00:22 |
| ho | mattoliverau: morning! | 00:22 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 00:22 |
| *** tsg has joined #openstack-swift | 00:50 | |
| *** jrichli has joined #openstack-swift | 00:55 | |
| *** ujjain has quit IRC | 00:57 | |
| *** annegentle has joined #openstack-swift | 01:17 | |
| openstackgerrit | Tushar Gohad proposed openstack/swift: EC: Multi-phase commit conversation for PUTs https://review.openstack.org/153843 | 01:20 |
| tsg | peluse, torgomatic: ^^^^ (got rid of the diskfile changes and also removed the eventlet patch dependency) | 01:22 |
| *** nellysmitt has joined #openstack-swift | 01:25 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Fixed links in multi-server Swift documentation https://review.openstack.org/160196 | 01:26 |
| *** annegentle has quit IRC | 01:29 | |
| *** nellysmitt has quit IRC | 01:30 | |
| charz | clayg: That's similar one, thanks! I'll add unit tests to the patch. | 01:40 |
| *** annegentle has joined #openstack-swift | 01:45 | |
| *** dmorita has joined #openstack-swift | 01:59 | |
| *** dmsimard_away is now known as dmsimard | 02:01 | |
| *** zhill has quit IRC | 02:02 | |
| *** annegentle has quit IRC | 02:03 | |
| *** rmcall has joined #openstack-swift | 02:06 | |
| *** omame has quit IRC | 02:16 | |
| clayg | charz: awesome! | 02:20 |
| *** panbalag has quit IRC | 02:21 | |
| *** annegentle has joined #openstack-swift | 02:30 | |
| *** annegentle has quit IRC | 02:36 | |
| *** dmsimard is now known as dmsimard_away | 02:40 | |
| *** tsg has quit IRC | 02:47 | |
| *** MVenesio has joined #openstack-swift | 03:06 | |
| thurloat | awsum | 03:07 |
| *** amandap is now known as amandap_ | 03:24 | |
| *** nellysmitt has joined #openstack-swift | 03:26 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Support of the Linux socket option SO_REUSEPORT https://review.openstack.org/137659 | 03:28 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: wip: cleanup SO_REUSEPORT https://review.openstack.org/160621 | 03:28 |
| *** nellysmitt has quit IRC | 03:31 | |
| *** amandap_ is now known as amandap | 03:33 | |
| *** MVenesio has quit IRC | 03:47 | |
| *** MVenesio has joined #openstack-swift | 03:47 | |
| openstackgerrit | Merged openstack/swift-specs: Composite Tokens and Service Accounts are done https://review.openstack.org/159425 | 03:50 |
| *** jrichli has quit IRC | 03:54 | |
| *** bkopilov has quit IRC | 03:54 | |
| clayg | whooooo a done spec! | 04:12 |
| notmyname | :-) | 04:12 |
| clayg | mattoliverau: why you messing with docs - it's such a bikeshed fiesta | 04:13 |
| *** tanee has quit IRC | 04:14 | |
| mattoliverau | clayg: sigh.. i know.. I was just auditing bugs, found one that wasn't an issue anymore, but there was a problem with linking so thought I'll submit a simple patch so we can close it. #canofworms | 04:14 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Fixed links in multi-server Swift documentation https://review.openstack.org/160196 | 04:15 |
| clayg | oh no... i get it... I'm totally enthralled now | 04:16 |
| clayg | I'm just saying if you were to just try harder to ignore it - we wouldn't be in this mess! | 04:16 |
| mattoliverau | clayg: lol | 04:17 |
| clayg | honestly I think i'm just happy for any excuse to play with my autodoc script on vagrant-swift-all-in-one, plus I didn't like the way the one sentance page rendered with a big grey square? | 04:18 |
| * mattoliverau wonders if he should point out to clayg that debain install guide might be valid for old releases... but how can we tell :P | 04:18 | |
| mattoliverau | clayg: yeah I noticed that too.. however, there was still annoying grey on the old version, I'm going to have to build your patchset docs and look now :P | 04:18 |
| clayg | mattoliverau: yeah i'm down - if there's an old link that still works we could totally add it | 04:18 |
| clayg | might be useful if someone wanted to wing it with juno | 04:19 |
| clayg | another great reason to keep the "all the swift docs" links on one page we maintain - do it! | 04:19 |
| clayg | see | 04:19 |
| clayg | next time you'll remember it's better to put your head in the ground | 04:19 |
| mattoliverau | lol | 04:19 |
| clayg | portante and zaitcev got on kick fixing up the swift-all-in-one docs one time and we lost *weeks* i tell you! | 04:20 |
| mattoliverau | noted.. there is no such thing as, "oh it'll be simple change" when it comes to docs :P | 04:21 |
| clayg | that's the learning you only get at school of hard knocks brother | 04:22 |
| mattoliverau | teacher my teacher | 04:24 |
| *** tanee has joined #openstack-swift | 04:24 | |
| * clayg is just funning around | 04:25 | |
| clayg | doc changes are a whole different level of difficultly than your average "add new behavior to distributed storage system running production peta-byte scale clusters in a variety of deployemnts around the globe" change | 04:27 |
| notmyname | lol | 04:29 |
| clayg | notmyname knows what clayg is talking about | 04:29 |
| notmyname | +1000 | 04:29 |
| *** MVenesio has quit IRC | 04:30 | |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Fixed links in multi-server Swift documentation https://review.openstack.org/160196 | 04:36 |
| mattoliverau | clayg: debian link added to havana ^^ | 04:36 |
| clayg | what the crap is this ec multi-phase put: 'Object upload failed: remote and local object md5 did not match, 4f9e5b770657de3d48671d751dac2b7c != b4f01a1d0ba0b46b05c73b2ac909b1df\nremote object has not been removed.' | 04:38 |
| clayg | also what gives with the \n | 04:38 |
| clayg | what branch am I on | 04:38 |
| *** sandywalsh_ has quit IRC | 04:56 | |
| *** sandywalsh has joined #openstack-swift | 04:58 | |
| *** tsg has joined #openstack-swift | 05:02 | |
| *** david-lyle_afk has joined #openstack-swift | 05:06 | |
| *** ppai has joined #openstack-swift | 05:09 | |
| ho | mattoliverau: the link for Juno/Suse should be http://docs.openstack.org/juno/install-guide/install/zypper/content/ch_swift.html | 05:11 |
| ho | mattoliverau: other links work without no problem | 05:12 |
| ho | s/without no problem/with no problem/ | 05:14 |
| mattoliverau | ho: nice spot! new patchset coming | 05:18 |
| openstackgerrit | Matthew Oliver proposed openstack/swift: Fixed links in multi-server Swift documentation https://review.openstack.org/160196 | 05:19 |
| clayg | does wakefield hang out in here? | 05:20 |
| clayg | hmph - does Joel? | 05:20 |
| *** bkopilov has joined #openstack-swift | 05:20 | |
| notmyname | clayg: ya, I think I've seen joel's nick. but not seen him speak | 05:21 |
| notmyname | joeljwright | 05:22 |
| clayg | ah, yeah doesn't seem to be around atm - thanks | 05:22 |
| *** zaitcev has quit IRC | 05:22 | |
| mattoliverau | notmyname, clayg do you guys ever sleep? | 05:22 |
| notmyname | mattoliverau: I don't htink clayg does ;-) | 05:22 |
| clayg | i'm basically asleep now | 05:22 |
| mattoliverau | lol | 05:23 |
| notmyname | lol | 05:23 |
| clayg | idk, why I keep acting like it's a good idea to avoid setting up a seperate vm for ec... | 05:23 |
| notmyname | some idea that ec is supposed to work with existing swift? | 05:24 |
| clayg | it's not even that hard with a seperate clone of vagrant-swift-all-in-one - I think I just have to change the ip in localrc | 05:24 |
| clayg | but like for some reason I keep thinking i should be able to have one vm that does both? why do I keep thinking that? | 05:24 |
| notmyname | clayg: last time I looked at vSAIO, I thought thee would be a nice use for support of multiple config files | 05:25 |
| clayg | oh vSAIO i like that - sounds like something sony built | 05:25 |
| clayg | also I don't think i know what you mean :P | 05:26 |
| notmyname | well if it reveals anything about how recently I looked at it, I'm 18 commits behind your upstream | 05:27 |
| *** nellysmitt has joined #openstack-swift | 05:27 | |
| clayg | ho: how you feelin on mattoliverau's last change to multi-node doc links - i'm bout ready to put it to bed! | 05:27 |
| clayg | acpahpcahcapahcpahpcahph - you don't even have a opensource license | 05:28 |
| ho | clayg: I checked all links and they worked. | 05:29 |
| clayg | ho: ditto! | 05:29 |
| clayg | speaking of seperate vSAIO for ec - doug just added that thing to specify a branch - i wonder if I can set it to feature/ec | 05:30 |
| *** nellysmitt has quit IRC | 05:31 | |
| notmyname | so I just looked in your github account for the vsaio. then I realized it's in swiftstack's. then I realized I can just merger stuff there ;-) | 05:32 |
| * notmyname tries to remember what a github workflow looks like | 05:33 | |
| clayg | merger stuff - or murder stuff? | 05:34 |
| notmyname | both? ;-) | 05:34 |
| notmyname | https://github.com/swiftstack/vagrant-swift-all-in-one/pull/14 | 05:35 |
| clayg | notmyname: well but *mine* is acctually called localrc | 05:36 |
| notmyname | done | 05:37 |
| notmyname | (that was easy) | 05:37 |
| clayg | dork | 05:37 |
| notmyname | I'm just enjoying pushing code and having it auto-update a pull request ;-) | 05:38 |
| clayg | man, i sure screwed up dipping my toe in the abandon ML thread | 05:39 |
| notmyname | clayg: while I do agree with you, I think fifieldt has some good wisdom there | 05:39 |
| clayg | i like tom a lot | 05:39 |
| fifieldt | wisdom, me? | 05:39 |
| fifieldt | impossible | 05:39 |
| clayg | oh shit - didn't realize he was listening - i take it back | 05:40 |
| notmyname | lol | 05:40 |
| fifieldt | :D | 05:40 |
| fifieldt | are you feeling ok clayg? | 05:40 |
| fifieldt | I haven't read many replies after yours | 05:40 |
| fifieldt | I hope I didn't offend with mine :( | 05:40 |
| clayg | lol, i'm *really* hard to offend | 05:40 |
| mattoliverau | its true, I've been trigin to offend him for months now. | 05:41 |
| mattoliverau | *tryin | 05:41 |
| clayg | i was just realizing that the whole notion of "abandoning a change might have a negative connotative signal" is not the realm where I can contribute in a meaningful way | 05:41 |
| fifieldt | personally I still think it was worthwhile | 05:42 |
| fifieldt | you're not wrong in your post :) | 05:42 |
| clayg | totally useful - *i'm* not *helpful* with that problem | 05:42 |
| notmyname | fifieldt: I think it does come down to the fact that an action is taken. but to your point, the phrasing is important. like with the change in working a while back to the -1 text in gerrit | 05:43 |
| clayg | the only reason I jumped in was to say that if we don't abandon them they'll just pile up | 05:43 |
| notmyname | ie when abandoning something, it's not just words. a patch by someone is actually abandoned | 05:43 |
| fifieldt | indeed! it is fair to use the abandon button for abandoned things :) and I'm not entirely convinced by the "let's make a new dashboard to fix this" argument either | 05:43 |
| *** david-lyle_afk has quit IRC | 05:44 | |
| * fifieldt shrugs | 05:44 | |
| fifieldt | storm in a teacup? | 05:44 |
| fifieldt | hmm, sorry, I think that is supposed to be "Tempest in a teapot" | 05:45 |
| clayg | ho: where is everyone making these charts that don't render for me | 05:45 |
| notmyname | fifieldt: actually that's where I'd defer to you. from my american perspective I think it is as you say. from a non-western view? I don' t know. I defer to you | 05:46 |
| *** SkyRocknRoll has joined #openstack-swift | 05:46 | |
| clayg | ho: OH! no i get it, 1 & 3 don't exist - delete_objs is always None if you use rsync ;) | 05:47 |
| fifieldt | I can't say I have much more to offer :D | 05:47 |
| fifieldt | argh. you know that time when you open a bunch of reviews in gerrit only to realise you're not logged in? hate that. | 05:48 |
| mattoliverau | notmyname, clayg: if your still here, are you guys happy if I +A tsg's EC PUT patch, peluse has +2, and in my limitted PUT testing its working. I'm happy to +2 to get it moving and do more testing once EC is more complete. I ask cause its sounds like (from earlier channel posts) clayg might still be looking at it. | 06:07 |
| mattoliverau | if clayg is still iterating on it, I'll just +2 it and leave it to him to +A | 06:07 |
| clayg | well i started to look at it | 06:07 |
| mattoliverau | clayg: before getting sidetracked on vSAIO? :P | 06:08 |
| * notmyname isn't currently looking at it | 06:08 | |
| clayg | mattoliverau: I'd so go ahead and land it if you like it - paul also said something about dlo doesn't work? it's whatever - we'll fix it all eventually | 06:08 |
| mattoliverau | yeah, DLO will be fixed in a later patch, or something, in my testing I was getting wierd GET behaviour, specifically not using the .durable files, but that is nothing to do with this patch, and will obviously be something added once this lands. | 06:11 |
| mattoliverau | notmyname: noted :) | 06:11 |
| mattoliverau | K, I need to go make a start on dinner so it'll be ready by the time the wife gets home. | 06:17 |
| *** ppai has quit IRC | 06:22 | |
| clayg | mattoliverau: idk, when i test it i get md5 mismatch after a round trip so idk | 06:22 |
| *** gyee has quit IRC | 06:22 | |
| mattoliverau | Oh really.. Bugger.. I'll have to jump on my box and take another look.. Maybe that should be a test then | 06:32 |
| *** ppai has joined #openstack-swift | 06:33 | |
| mattoliverau | clayg: when I put a file and then pull it out again md5s match... So hmm. | 06:39 |
| *** nellysmitt has joined #openstack-swift | 06:42 | |
| *** david-lyle_afk has joined #openstack-swift | 06:44 | |
| *** jroll has quit IRC | 06:49 | |
| openstackgerrit | Merged openstack/swift: Fixed links in multi-server Swift documentation https://review.openstack.org/160196 | 06:52 |
| *** reed has quit IRC | 06:58 | |
| *** rmcall has quit IRC | 06:58 | |
| tsg | mattoliverau: are you using swift client to test ec put/gets? | 06:59 |
| *** jroll has joined #openstack-swift | 06:59 | |
| *** mrda is now known as mrda-away | 07:04 | |
| *** SkyRocknRoll has quit IRC | 07:04 | |
| *** mmcardle has joined #openstack-swift | 07:05 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:18 | |
| *** SkyRocknRoll has quit IRC | 07:18 | |
| *** SkyRocknRoll has joined #openstack-swift | 07:18 | |
| *** foexle has joined #openstack-swift | 07:20 | |
| *** openstackgerrit has quit IRC | 07:22 | |
| *** openstackgerrit has joined #openstack-swift | 07:22 | |
| mattoliverau | Yup.. And curl | 07:29 |
| *** AnxiousGarlic has joined #openstack-swift | 07:29 | |
| *** AnxiousGarlic has left #openstack-swift | 07:30 | |
| *** ppai has quit IRC | 07:34 | |
| openstackgerrit | Merged openstack/swift: EC: Multi-phase commit conversation for PUTs https://review.openstack.org/153843 | 07:41 |
| *** ppai has joined #openstack-swift | 07:48 | |
| *** nshaikh has joined #openstack-swift | 08:06 | |
| *** rledisez has joined #openstack-swift | 08:08 | |
| openstackgerrit | Charles Hsu proposed openstack/python-swiftclient: Add ability to download objects to particular folder. https://review.openstack.org/160283 | 08:11 |
| *** jistr has joined #openstack-swift | 08:46 | |
| *** acoles_away is now known as acoles | 08:47 | |
| *** geaaru has joined #openstack-swift | 08:48 | |
| *** omame has joined #openstack-swift | 09:01 | |
| *** JoshNang has quit IRC | 09:07 | |
| *** jordanP has joined #openstack-swift | 09:23 | |
| *** ppai has quit IRC | 09:49 | |
| *** foexle has quit IRC | 09:51 | |
| *** ppai has joined #openstack-swift | 10:03 | |
| *** foexle has joined #openstack-swift | 10:08 | |
| *** dmorita has quit IRC | 10:13 | |
| *** ujjain has joined #openstack-swift | 10:25 | |
| *** ujjain has quit IRC | 10:25 | |
| *** ujjain has joined #openstack-swift | 10:25 | |
| *** bkopilov has quit IRC | 10:54 | |
| *** ho has quit IRC | 10:56 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 11:03 |
| *** tellesnobrega has quit IRC | 11:27 | |
| *** km has quit IRC | 11:28 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 11:29 |
| acoles | peluse: ^^ | 11:29 |
| *** tellesnobrega has joined #openstack-swift | 11:33 | |
| *** EmilienM|afk is now known as EmilienM | 12:05 | |
| peluse | acoles, cool! I'll look a bit later this morning. Have a packed schdule here coming up | 12:25 |
| *** MVenesio has joined #openstack-swift | 12:26 | |
| acoles | peluse: np. i think there are some opportunities to simplify some code but that can wait. important thing now is to validate that I got the ondisk file search 'algorithm' correct. | 12:28 |
| acoles | peluse: i didn't look at any of the suffix_hash stuff in that review, maybe get to that later today | 12:28 |
| *** mmcardle has quit IRC | 12:30 | |
| *** aix has quit IRC | 12:44 | |
| *** tongli has joined #openstack-swift | 13:03 | |
| *** jroll has quit IRC | 13:07 | |
| *** jroll has joined #openstack-swift | 13:07 | |
| *** Bsony has joined #openstack-swift | 13:09 | |
| *** mmcardle has joined #openstack-swift | 13:14 | |
| *** lpabon has joined #openstack-swift | 13:17 | |
| openstackgerrit | Alistair Coles proposed openstack/swift-specs: Updates to encryption spec https://review.openstack.org/154318 | 13:21 |
| *** Bsony has quit IRC | 13:22 | |
| *** aix has joined #openstack-swift | 13:23 | |
| *** Bsony has joined #openstack-swift | 13:23 | |
| *** panbalag has joined #openstack-swift | 13:28 | |
| *** stockpirate has joined #openstack-swift | 13:31 | |
| peluse | acoles, OK, on the sufix_hash stuff note the big comment there that could result is big simplication of the class. Might not even want separate subclasses but I hesitate to mke that change until we get everything fully working on one branch... | 13:42 |
| acoles | peluse: ok, noted thx | 13:42 |
| peluse | acoles, what's not in use at all right now is the 2nd by_fi dict | 13:42 |
| peluse | acoles, and really the only thing left in the reconstructor (functinliaty) is local file cleanup of some corner cases so I doubt we'll need it. But I really don't want to rip it out and have to put it back in (lots of little changes to test code) so that's why I'm waiting :) | 13:43 |
| acoles | peluse: makes sense! | 13:43 |
| * peluse off to shuttle kids to school... | 13:44 | |
| *** aix has quit IRC | 13:44 | |
| *** aix has joined #openstack-swift | 13:44 | |
| *** jbonjean has quit IRC | 13:46 | |
| *** jbonjean has joined #openstack-swift | 13:46 | |
| *** jbonjean has quit IRC | 13:47 | |
| *** jbonjean has joined #openstack-swift | 13:48 | |
| *** jbonjean has quit IRC | 13:48 | |
| *** jbonjean has joined #openstack-swift | 13:48 | |
| *** nshaikh has quit IRC | 13:49 | |
| *** bill_az has joined #openstack-swift | 14:00 | |
| *** mahatic has joined #openstack-swift | 14:09 | |
| *** rdaly2 has joined #openstack-swift | 14:11 | |
| *** dmsimard_away is now known as dmsimard | 14:17 | |
| *** ppai has quit IRC | 14:20 | |
| *** chlong has joined #openstack-swift | 14:27 | |
| *** mahatic has quit IRC | 14:29 | |
| *** anayag has joined #openstack-swift | 14:33 | |
| *** tongli has quit IRC | 14:36 | |
| anayag | Hi I am doing some experiments with swift. My cluster has 5 storage nodes and 1 proxy. I performed 90% writes and 10% reads to the cluster using ssbench having Read Quorum 1 Write Quorum 5 and Read Quorum 5 and Write Quorum 1BUt I could not observe any significant change at all | 14:37 |
| anayag | could anybody explain me this behaviour? | 14:37 |
| openstackgerrit | Alistair Coles proposed openstack/swift-specs: Updates to encryption spec https://review.openstack.org/154318 | 14:40 |
| *** mahatic has joined #openstack-swift | 14:42 | |
| *** MVenesio_ has joined #openstack-swift | 14:44 | |
| *** bkopilov has joined #openstack-swift | 14:45 | |
| *** MVenesio has quit IRC | 14:47 | |
| *** MVenesio has joined #openstack-swift | 15:05 | |
| *** MVenesio_ has quit IRC | 15:08 | |
| openstackgerrit | Alistair Coles proposed openstack/swift: Fix tempauth acl checks when simplejson has no speedups https://review.openstack.org/159530 | 15:13 |
| *** jasondotstar has joined #openstack-swift | 15:16 | |
| *** tsg has quit IRC | 15:17 | |
| openstackgerrit | Charles Hsu proposed openstack/python-swiftclient: Add ability to download objects to particular folder. https://review.openstack.org/160283 | 15:19 |
| *** rdaly2 has quit IRC | 15:26 | |
| *** rdaly2 has joined #openstack-swift | 15:33 | |
| *** jrichli has joined #openstack-swift | 15:41 | |
| *** annegentle has joined #openstack-swift | 15:42 | |
| peluse | acoles, there? | 15:56 |
| acoles | peluse: here | 15:56 |
| peluse | hey great job the GOF/HCL stuff! | 15:56 |
| peluse | still reviewing the test code then need to rebase it again since multiphase landed w/some changes, then I'll add it as a dependecy to the reconstructor and test from there.... | 15:57 |
| acoles | peluse: np. i rebased it this morning so I think its based on the multiphase commit. in fact, i used diskfile.write_durable_timestamp! | 15:58 |
| *** tsg has joined #openstack-swift | 15:59 | |
| *** rdaly2 has quit IRC | 15:59 | |
| *** rdaly2 has joined #openstack-swift | 16:00 | |
| peluse | acoles, ok cool. I thought i saw something that was outdated but may have been mistaken. really like test GOF tests too, looking through those now | 16:01 |
| *** lpabon has quit IRC | 16:01 | |
| *** rmcall has joined #openstack-swift | 16:02 | |
| *** tsg_ has joined #openstack-swift | 16:02 | |
| *** lpabon has joined #openstack-swift | 16:03 | |
| *** ekarlso has quit IRC | 16:03 | |
| *** tsg has quit IRC | 16:05 | |
| *** annegentle has quit IRC | 16:06 | |
| *** annegentle has joined #openstack-swift | 16:06 | |
| *** ekarlso has joined #openstack-swift | 16:07 | |
| *** nexusz99 has joined #openstack-swift | 16:08 | |
| *** homegrown_ has joined #openstack-swift | 16:10 | |
| *** JoshNang has joined #openstack-swift | 16:14 | |
| peluse | clayg, mattoliverau FYI I didn't have any md5 mismstaches testing the multiphase either.... just the DLO thing | 16:21 |
| notmyname | good morning | 16:21 |
| peluse | speaking of which... | 16:21 |
| peluse | tdasilva, if you're around you're now free to knock out the DB udpate thing on feature/EC now that both PUT and GET have landed | 16:21 |
| peluse | morning notmyname | 16:21 |
| tdasilva | peluse: but did acoles change make it to feature/EC ? | 16:22 |
| notmyname | interesting http://www.hgst.com/press-room/press-releases/HGST-Announces-Definitive-Agreement-to-Acquire-Amplidata | 16:23 |
| tdasilva | peluse: this one: https://review.openstack.org/#/c/158821/ | 16:23 |
| peluse | tdasilva, one sec | 16:23 |
| peluse | so I didn't merge master on Fri because I wanted multiphase to land w/ more complications so no.... will do that now | 16:25 |
| peluse | w/o more compications that is :) | 16:26 |
| tdasilva | peluse: got it! thanks, will send the patch once that lands | 16:27 |
| *** bill_az has quit IRC | 16:32 | |
| peluse | tdasilva, ha, only one tiny doc conflict. will run a few tests and throw it in the hopper.... | 16:33 |
| tdasilva | peluse: cool...almost done with obj. versioning, so timing is good :) | 16:33 |
| *** otoolee has quit IRC | 16:36 | |
| *** Bsony has quit IRC | 16:36 | |
| *** david-lyle_afk has quit IRC | 16:37 | |
| *** reed has joined #openstack-swift | 16:38 | |
| openstackgerrit | Denis Cavalcante proposed openstack/swift: Add storage policy support for sorting method https://review.openstack.org/160877 | 16:39 |
| *** otoolee has joined #openstack-swift | 16:44 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: versioned writes middleware https://review.openstack.org/134347 | 16:44 |
| *** jistr has quit IRC | 16:50 | |
| tdasilva | peluse: stepping out during lunch and will be back later... | 16:51 |
| *** Sanchit has joined #openstack-swift | 16:52 | |
| peluse | tdasilva, cool. working through some failures that are probably in my env but may not have it submitted til after a meeting I have scoming up here in 45 min or so.... stay tuned | 16:52 |
| *** annegent_ has joined #openstack-swift | 16:54 | |
| *** mahatic has quit IRC | 16:54 | |
| Sanchit | Hi, I am using Swift+Ceilometer Icehouse version. I want to know if the publishing calls from swift to rabbitmq/ceilometer are asyncronus or not. | 16:55 |
| notmyname | Sanchit: better to ask the ceilometer devs | 16:55 |
| notmyname | Sanchit: I know it used to be. and now it might not be. but I don't know when that may have changed | 16:56 |
| Sanchit | Okay notmyname . I'll ask it in the other channel. anyways, any input would be better. | 16:57 |
| *** SkyRocknRoll has quit IRC | 16:57 | |
| *** annegentle has quit IRC | 16:58 | |
| *** gyee has joined #openstack-swift | 16:58 | |
| openstackgerrit | paul luse proposed openstack/swift: Merge master to feature/ec https://review.openstack.org/160890 | 16:59 |
| *** nellysmitt has quit IRC | 17:03 | |
| *** Bsony has joined #openstack-swift | 17:14 | |
| *** chlong has quit IRC | 17:17 | |
| *** rledisez has quit IRC | 17:20 | |
| Sanchit | Just in case if anybody using swift with ceilometer has any idea regarding this query of mine: | 17:26 |
| Sanchit | I am using Ceilometer Icehouse version. I want to know the difference between normal rpc calls and rpc cast. Also, are the publishing calls from swift to rabbitmq/ceilometer asynchronus call or else? | 17:27 |
| *** mmcardle has quit IRC | 17:27 | |
| *** jordanP has quit IRC | 17:39 | |
| *** anayag has quit IRC | 17:40 | |
| *** tellesnobrega has quit IRC | 17:43 | |
| *** JoshNang has left #openstack-swift | 17:43 | |
| *** Anticimex has quit IRC | 17:54 | |
| *** Anticimex has joined #openstack-swift | 17:56 | |
| *** ChanServ changes topic to "Review Dashboard: http://goo.gl/uRzLBX | Overview Dashboard: http://goo.gl/9EI0Sz | Priority Reviews: https://wiki.openstack.org/wiki/Swift/PriorityReviews | Ideas: https://wiki.openstack.org/wiki/Swift/ideas" | 17:56 | |
| *** geaaru has quit IRC | 18:02 | |
| *** tellesnobrega has joined #openstack-swift | 18:04 | |
| *** acoles is now known as acoles_away | 18:07 | |
| *** Bsony has quit IRC | 18:10 | |
| *** Bsony has joined #openstack-swift | 18:10 | |
| *** tsg_ has quit IRC | 18:14 | |
| *** erlon has joined #openstack-swift | 18:17 | |
| *** nexusz99 has quit IRC | 18:26 | |
| *** tsg has joined #openstack-swift | 18:28 | |
| *** tsg has quit IRC | 18:39 | |
| *** zhill has joined #openstack-swift | 18:45 | |
| *** tsg has joined #openstack-swift | 18:50 | |
| *** tsg_ has joined #openstack-swift | 18:52 | |
| *** nexusz99 has joined #openstack-swift | 18:53 | |
| *** tsg has quit IRC | 18:55 | |
| *** nexusz99 has quit IRC | 18:57 | |
| clayg | peluse: yes, it's very strange - when I get the object with my old swift-ec-get script the resulting md5 matches the uploaded object - but not the etag? | 18:58 |
| clayg | peluse: md5 matched on small objects | 18:59 |
| *** rdaly2 has quit IRC | 18:59 | |
| clayg | it seems like PUT is giving me back one of the FA's ETags - it's not consistent | 19:04 |
| clayg | peluse: can you make sure you're running the latest version of the swift client tool (with md5 checking on upload) and try uploading an object that's a few mb | 19:05 |
| *** zaitcev has joined #openstack-swift | 19:10 | |
| *** ChanServ sets mode: +v zaitcev | 19:10 | |
| *** aix has quit IRC | 19:11 | |
| redbo | why's sync_method a required config for handoff deletes to happen now? https://github.com/openstack/swift/blob/master/swift/obj/replicator.py#L280 | 19:12 |
| *** Nadeem has joined #openstack-swift | 19:18 | |
| *** Bsony has quit IRC | 19:21 | |
| *** Bsony has joined #openstack-swift | 19:24 | |
| *** Bsony has quit IRC | 19:25 | |
| *** Bsony has joined #openstack-swift | 19:25 | |
| *** Nadeem has quit IRC | 19:26 | |
| clayg | redbo: because no one has merged https://review.openstack.org/#/c/159683/ | 19:30 |
| clayg | redbo: sorry about that! | 19:31 |
| redbo | ha, okay. We just ran into it in testing by accident. | 19:35 |
| clayg | yeah it totally fucking sucks, thank god for probetests, i feel like a sucker for letting it slip | 19:35 |
| clayg | redbo: it's those hack-a-thons, people fly in from japan and shit and I get all antsy to merge stuff thinking we can just clean it up later | 19:36 |
| clayg | well turns out now is better than never, but never is often better than *right* now | 19:37 |
| *** bkopilov has quit IRC | 19:42 | |
| *** cebruns has quit IRC | 19:49 | |
| clayg | tdasilva: thanks for the new versioned writes middleware patchset - first glance looks pretty good - i'll play around with it! | 19:53 |
| clayg | redbo: I didn't use self.sync_method == self.rsync (or whatever) because sometimes tests want to set replicator.sync_method == _fake_sync_method | 19:54 |
| tdasilva | clayg: thanks! yeah, still a bit more work to do, but wanted to get something out for review sooner rather than later | 19:54 |
| clayg | tdasilva: good call | 19:54 |
| clayg | i'm trying to grok around why the signatures for self.contaienr_request and self.object_request are so different | 19:55 |
| peluse | clayg, just got back from, well, out goofing off. Yes I will give it another try for sure. I've been using a 3MB random binary and doing a cmp of the original with what I get back so checking integrity bu not explcitly the calc'd md5 so I guess I was talking apples to semi-apples | 19:55 |
| clayg | I was also hoping you'd pull in the doc change to keep the version write landing page - I keep thinking it will come in handy as an anchor poing when we get around to https://review.openstack.org/#/c/155421/ (the undelete shadow copy version manifest pointer object thing0 | 19:56 |
| clayg | peluse: so digging into the proxy i'm not seeing it's do anything to graph the X-Object-Sysmeta-Ec-Etag instead of just the Etag in best_responses when the not policy.stores_object_verbaitem | 19:58 |
| clayg | ... so i'm not surprised that PUT is giving me back one of the FA's etags - but I am surprised no one else noticed it | 19:58 |
| peluse | clayg, which ver of swift client are you using (just so I know I have the same exact thing) | 19:59 |
| clayg | master | 19:59 |
| peluse | clayg, and hmmmm on your comment. I thought I saw that code in one of a zillion recent reviews | 19:59 |
| clayg | 925c01ebfb or so | 19:59 |
| clayg | maybe *I'm* on the wrong patch!? | 20:00 |
| clayg | hrmm... looks right 619ba2f4035 | 20:00 |
| *** hemanthm has joined #openstack-swift | 20:03 | |
| notmyname | redbo: zaitcev: thanks for landing that patch | 20:03 |
| zaitcev | well if I paid attention at hackathon, it would not be needed in the first place | 20:04 |
| hemanthm | Hey, swift folks! Can I get a review on this glance MP please? https://review.openstack.org/#/c/159223/2 | 20:05 |
| * clayg and zaitcev take turns punching each other in the face | 20:05 | |
| *** tacticus has quit IRC | 20:06 | |
| *** tsg_ has quit IRC | 20:06 | |
| clayg | hemanthm: multipart-manifest=delete seems like the right thing to do for SLO - but I don't think there's anything like that for DLO | 20:07 |
| clayg | hemanthm: in the tests I'm not seeing anything that shows that the DLO segments are deleted, it'd take a container list and a bunch of delete calls | 20:08 |
| clayg | hemanthm: oh, code seems to do it though, maybe i missed something in the test... | 20:09 |
| hemanthm | clayg: Glance does that already for DLOs | 20:09 |
| hemanthm | clayg: the tests I added in that MP are just related to SLO deletion | 20:11 |
| clayg | hemanthm: oic, the second test isn't for a dlo it's just for a whole objects | 20:11 |
| clayg | hemanthm: looks good to me | 20:11 |
| tdasilva | clayg: re: version doc. The reason for removing that was because I copied the content to versioned_writes.py, so I didn't want to have duplicated content. | 20:12 |
| hemanthm | clayg: cool, thanks! | 20:13 |
| peluse | clayg, hells bells you're right. I'll track it down (and the unit test escape) and take care of it (unless you've already done so?) | 20:13 |
| clayg | tdasilva: yeah in my follow up patch i did something that just inlined it | 20:13 |
| *** tsg_ has joined #openstack-swift | 20:13 | |
| hemanthm | clayg: while we are talking about DLO deletion, do you there is a better way of deleting DLOs? | 20:13 |
| clayg | peluse: well sorta - i mean it's leaky because that method wouldn't otherwise know about an ec specific header like X-Object-Sysmeta-Ec-Etag | 20:13 |
| clayg | peluse: but torgomatic is going to extract all that shit to the ECObjectController so it won't matter | 20:14 |
| clayg | hemanthm: no | 20:14 |
| clayg | hemanthm: well... for this class of large object slo's are just better | 20:14 |
| hemanthm | clayg: would you recommend bulk delete? | 20:14 |
| peluse | clayg, OK, I'll leave it be then. w/o the md5 check (just using curl) the data is still in tact at least so we've got that going for us (get matchs what was put) | 20:14 |
| clayg | hemanthm: yeah, ok, fair enough - you still have to do the container listing, and make the manifest, and at the end of the day the proxy is making the same requests the client would | 20:15 |
| peluse | clayg, I'll throw somthing on trello though | 20:15 |
| clayg | hemanthm: also currently the bulk delete doesn't do anything concurrently - so if you're managing the connections on your end you can spin up a dozen works and make the requests a little faster than you'd get out of bulk delete | 20:16 |
| hemanthm | clayg: yeah, we are looking at using SLOs, but since SLOs have an upper limit on segments, we gotta create SLO of SLOs I think | 20:16 |
| clayg | also if you're w/i close proximity to the swift endpoint some of the connection overhead is minimized | 20:16 |
| clayg | hemanthm: honestly it's not worth nearly the effort of getting everything on slo's | 20:16 |
| tdasilva | clayg: ha! I had made that from your follow up patch and just noticed it didn't make it in the last patchset, sorry... | 20:17 |
| clayg | hemanthm: rly? you could make your segments bigger | 20:17 |
| tdasilva | *made that change | 20:17 |
| hemanthm | clayg: its configurable, if some operator uses a small segment size, its very easy to go beyond 1000 segments | 20:18 |
| *** silor has joined #openstack-swift | 20:18 | |
| *** tsg_ has quit IRC | 20:19 | |
| clayg | man... | 20:19 |
| *** zhill has quit IRC | 20:20 | |
| clayg | I think 100 MB is reasonable segment size, with default 1000 segment limit that's like what - 100 GB? what a good size for an image? | 20:20 |
| hemanthm | clayg: I have seen images over 1TB :) | 20:21 |
| clayg | minimum segment size 1GB - done | 20:21 |
| clayg | :) | 20:21 |
| hemanthm | hehe | 20:21 |
| clayg | oh crap... you said *over* 1TB :\ | 20:22 |
| clayg | stupid data | 20:22 |
| hemanthm | clayg: would you mind leaving a comment/+1 on the review, glance cores will be happy that it has swift's blessings | 20:24 |
| clayg | oh yeah shit - i meant to do that - blame peluse and tdasilva | 20:24 |
| tdasilva | lol | 20:24 |
| tdasilva | clayg is blaming everyone today, japan, peluse, tdasilva :P | 20:25 |
| tdasilva | hehe | 20:25 |
| hemanthm | haha | 20:25 |
| *** gyee has quit IRC | 20:28 | |
| *** silor has quit IRC | 20:28 | |
| *** zhill has joined #openstack-swift | 20:32 | |
| peluse | hell I'll take the blame | 20:35 |
| peluse | what was it for again? | 20:35 |
| hemanthm | clayg: thanks for the review :) | 20:36 |
| *** mrda-away is now known as mrda | 20:44 | |
| clayg | peluse: jesus man - the container listing also has some random FA's hash? | 20:48 |
| clayg | yeah all three replicas of the container have a different has for the object | 20:48 |
| *** aserw1 has joined #openstack-swift | 20:48 | |
| *** zhill has quit IRC | 20:49 | |
| aserw1 | . | 20:49 |
| *** foexle has quit IRC | 20:50 | |
| aserw1 | Hello to all. Can anyone point me to good delegetion explanation with code samples? I am just learning with Lynda.com and do not understand their explanation. Thank you | 20:52 |
| aserw1 | I did some google, but mostly i get c# samples. | 20:53 |
| notmyname | aserw1: are you asking about the Swift programming language? | 20:54 |
| *** nexusz99 has joined #openstack-swift | 20:54 | |
| notmyname | (that's not what we do here) | 20:54 |
| aserw1 | yes. sorry, i thougt from the channel name that this channel is for swift users | 20:55 |
| notmyname | aserw1: no, this is openstack swift, a distributed object storage system. http://docs.openstack.org/developer/swift/ | 20:56 |
| *** zhill has joined #openstack-swift | 20:57 | |
| *** zhill has quit IRC | 20:57 | |
| aserw1 | sorry again) it seems i got confused by the channel name | 20:57 |
| notmyname | aserw1: no worries. I think there might be a #swift-lang or #swiftlang or something around here somewhere | 20:58 |
| glange | aserw1: try #taylorswift | 20:59 |
| *** nexusz99 has quit IRC | 20:59 | |
| notmyname | aserw1: also, ignore glange ;-) | 20:59 |
| glange | :) | 20:59 |
| aserw1 | I will:) | 20:59 |
| *** aserw1 has left #openstack-swift | 21:01 | |
| clayg | glange: +1 | 21:02 |
| glange | she knows a lot about computer security, it wasn't that ridiculous a suggestion: https://twitter.com/swiftonsecurity | 21:03 |
| portante | glange: wow, #taylorswift exists! | 21:04 |
| glange | portante: you sound surprised :) | 21:04 |
| portante | and regarding security, she just says, "shake it off" | 21:04 |
| glange | she's big, really big | 21:04 |
| glange | haha | 21:04 |
| clayg | these jokes mean nothing to me | 21:06 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Sync local regions first, and always shuffle nodes https://review.openstack.org/159241 | 21:10 |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Cleanup object-replicator handoff/rebalance partition syncing https://review.openstack.org/158927 | 21:10 |
| *** Bsony has quit IRC | 21:12 | |
| mattoliverau | Morning | 21:13 |
| peluse | mornin' | 21:18 |
| openstackgerrit | paul luse proposed openstack/swift: Minor cleanup in proxy obj controller https://review.openstack.org/160985 | 21:24 |
| openstackgerrit | Merged openstack/swift: Merge master to feature/ec https://review.openstack.org/160890 | 21:25 |
| *** zhill has joined #openstack-swift | 21:29 | |
| openstackgerrit | Merged openstack/swift: Fix object replicator partition cleanup https://review.openstack.org/159683 | 21:31 |
| peluse | tdasilva, OK, feature/ec has been updated so has the container override stuff on it now | 21:33 |
| tdasilva | peluse: got it, rebasing my patch now | 21:34 |
| tdasilva | thx! | 21:34 |
| peluse | hey, my pleasure :) | 21:34 |
| notmyname | wow. there's a lot of hate for eventlet in the openstack world | 21:40 |
| zaitcev | and the alternative is... Apache? | 21:40 |
| clayg | zaitcev: probably tulip | 21:41 |
| zaitcev | I didn't think tulip included the webserver like the one that comes with eventlet | 21:41 |
| clayg | oh, is the dog specifically on eventlet.wsgi's server then or just implicity co-operative yeilding? | 21:42 |
| notmyname | clayg: mostly the latter | 21:43 |
| notmyname | and #openstack-meeting right now is debating the proposed spec to force everyone to use asyncio | 21:44 |
| *** EmilienM has quit IRC | 21:44 | |
| *** EmilienM has joined #openstack-swift | 21:44 | |
| *** gyee has joined #openstack-swift | 21:44 | |
| notmyname | and since that would "just" be an oslo change, it's pretty much trivial. /s | 21:45 |
| clayg | not to be confused with asyncore! | 21:45 |
| clayg | *I* was under the impression asyncio is a real chore without yeild from | 21:46 |
| notmyname | ah. monkeypatching is bad, therefore rewrite everything away from eventlet | 21:46 |
| clayg | it's true monkeypatching is bad | 21:47 |
| notmyname | therefore rewrite in golang! ;-) | 21:47 |
| clayg | well you might be onto something there | 21:47 |
| clayg | oic, Trollius has yield From(...) | 21:48 |
| clayg | yeah Trollius looks good, we could rewrite eventlet.wsgi's http server and add http2 support while we're at it... | 21:52 |
| clayg | notmyname: put it on the docket for Liberty | 21:53 |
| notmyname | clayg: why not kilo? we've still got time, right? ;-) | 21:53 |
| clayg | nope - ec | 21:53 |
| notmyname | I know. "just use a different concurrency library" sounds like a pretty big change | 21:54 |
| notmyname | (my only response to seeing openstack governance in action today is laughter) | 21:54 |
| *** erlon has quit IRC | 22:01 | |
| *** MVenesio has quit IRC | 22:08 | |
| *** lpabon has quit IRC | 22:12 | |
| *** gyee has quit IRC | 22:13 | |
| *** devlaps has joined #openstack-swift | 22:13 | |
| zaitcev | clayg: Did you look closely into HTTP/2? Can you confirm my reading that they provide an equivalent of 100-Continue in an officially blessed way (via frames and flow control)? | 22:13 |
| *** devlaps has quit IRC | 22:13 | |
| *** Bsony has joined #openstack-swift | 22:19 | |
| *** Bsony has quit IRC | 22:23 | |
| openstackgerrit | paul luse proposed openstack/swift: Erasure Code Reconstructor https://review.openstack.org/131872 | 22:23 |
| openstackgerrit | paul luse proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 22:23 |
| *** rdaly2 has joined #openstack-swift | 22:24 | |
| peluse | ^ unless I totally jacked that up these are now carved apart with the recon being needed by mutiple FI support | 22:25 |
| *** gyee has joined #openstack-swift | 22:33 | |
| openstackgerrit | paul luse proposed openstack/swift: Multiple fragment Archive Index Support https://review.openstack.org/159637 | 22:35 |
| *** rmcall has quit IRC | 22:36 | |
| *** os1 has joined #openstack-swift | 22:36 | |
| os1 | Hi | 22:36 |
| os1 | Is it possible to make it so that xprofile only collects data at specified times? | 22:37 |
| notmyname | os1: no | 22:37 |
| os1 | Because if it's always running, then I imagine that would generate significant overhead. | 22:37 |
| *** jrichli has quit IRC | 22:39 | |
| *** rdaly2 has quit IRC | 22:41 | |
| os1 | notmyname : Is there anything that can be done about the overhead? How significant is it? | 22:43 |
| *** nexusz99 has joined #openstack-swift | 22:44 | |
| notmyname | os1: it's designed to be used as-needed and more for relative profile tests (as opposed to always-on debug info) | 22:44 |
| notmyname | os1: I wouldn't use it much outside of controlled areas | 22:44 |
| *** nexusz99 has quit IRC | 22:48 | |
| *** annegent_ has quit IRC | 22:49 | |
| *** openstackgerrit has quit IRC | 22:51 | |
| *** openstackgerrit has joined #openstack-swift | 22:52 | |
| *** km has joined #openstack-swift | 23:07 | |
| *** chlong has joined #openstack-swift | 23:31 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift: WIP: Use Container override header to update etag and size https://review.openstack.org/161030 | 23:32 |
| *** hemanthm has left #openstack-swift | 23:32 | |
| *** os1 has left #openstack-swift | 23:37 | |
| *** dmorita has joined #openstack-swift | 23:45 | |
| *** zhill has quit IRC | 23:47 | |
| *** annegentle has joined #openstack-swift | 23:49 | |
| *** dmsimard is now known as dmsimard_away | 23:51 | |
| clayg | tdasilva: that change will make the PUT return the whole object md5's to the proxy on PUT? | 23:52 |
| tdasilva | clayg: yeah | 23:53 |
| clayg | tdasilva: I sorta think http would rather we return the etag of the object that was just uploaded... I mean didn't the proxy *send* us the computed etag (mainly for the container update) | 23:53 |
| tdasilva | clayg: oh, are you referring to that last line I added in obj/server.py: etag = update_headers['x-etag'] | 23:55 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!