| *** chlong has joined #openstack-swift | 00:19 | |
| *** hrou has joined #openstack-swift | 00:21 | |
| openstackgerrit | Janie Richling proposed openstack/swift: Encrypting/Decrypting account and container metadata https://review.openstack.org/203454 | 00:29 |
|---|---|---|
| openstackgerrit | Janie Richling proposed openstack/swift: Encrypting/Decrypting account and container metadata https://review.openstack.org/203454 | 00:32 |
| *** jrichli has quit IRC | 00:55 | |
| *** jrichli has joined #openstack-swift | 00:56 | |
| *** Kennan2 has joined #openstack-swift | 02:05 | |
| *** Kennan has quit IRC | 02:05 | |
| *** Kennan2 is now known as Kennan | 02:07 | |
| openstackgerrit | Janie Richling proposed openstack/swift: Encrypting/Decrypting account and container metadata https://review.openstack.org/203454 | 02:39 |
| *** jrichli has quit IRC | 02:43 | |
| *** wbhuber has quit IRC | 03:32 | |
| *** sanchitmalhotra has joined #openstack-swift | 03:44 | |
| *** pbourke has quit IRC | 03:56 | |
| *** pbourke has joined #openstack-swift | 03:57 | |
| openstackgerrit | paul luse proposed openstack/swift: Add ability for GET path to see/select alternate frag archs https://review.openstack.org/201283 | 04:06 |
| *** cdelatte has joined #openstack-swift | 04:11 | |
| *** delattec has joined #openstack-swift | 04:11 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 04:18 | |
| *** darrenc is now known as darrenc_afk | 04:20 | |
| *** sanchitmalhotra has quit IRC | 04:21 | |
| *** sanchitmalhotra has joined #openstack-swift | 04:29 | |
| *** sanchitmalhotra1 has quit IRC | 04:31 | |
| *** ppai has joined #openstack-swift | 04:34 | |
| *** darrenc_afk is now known as darrenc | 04:35 | |
| *** briancurtin has quit IRC | 04:58 | |
| *** briancurtin has joined #openstack-swift | 04:59 | |
| *** nottrobin has quit IRC | 05:02 | |
| *** nottrobin has joined #openstack-swift | 05:05 | |
| *** chlong has quit IRC | 05:27 | |
| *** Protux has quit IRC | 05:35 | |
| *** serverascode has quit IRC | 05:35 | |
| *** h00327910_ has quit IRC | 05:36 | |
| *** briancurtin has quit IRC | 05:36 | |
| *** chlong has joined #openstack-swift | 05:40 | |
| *** serverascode has joined #openstack-swift | 05:41 | |
| *** briancurtin has joined #openstack-swift | 05:44 | |
| *** h00327910_ has joined #openstack-swift | 05:48 | |
| *** Protux has joined #openstack-swift | 05:48 | |
| *** sanchitmalhotra1 has joined #openstack-swift | 06:00 | |
| *** sanchitmalhotra has quit IRC | 06:02 | |
| *** hrou has quit IRC | 06:03 | |
| *** chlong has quit IRC | 06:20 | |
| *** chlong has joined #openstack-swift | 06:36 | |
| *** sanchitmalhotra has joined #openstack-swift | 06:46 | |
| *** sanchitmalhotra1 has quit IRC | 06:48 | |
| *** jith_ has joined #openstack-swift | 06:59 | |
| *** ig0r_ has joined #openstack-swift | 07:01 | |
| *** rledisez has joined #openstack-swift | 07:05 | |
| *** hseipp has joined #openstack-swift | 07:20 | |
| *** hseipp has quit IRC | 07:20 | |
| *** hseipp has joined #openstack-swift | 07:20 | |
| *** rledisez has quit IRC | 07:30 | |
| *** nottrobin has quit IRC | 07:31 | |
| *** Protux has quit IRC | 07:31 | |
| *** zhiyan has quit IRC | 07:31 | |
| *** cppforlife_ has quit IRC | 07:31 | |
| *** serverascode has quit IRC | 07:31 | |
| *** chlong has quit IRC | 07:34 | |
| *** ntt has joined #openstack-swift | 07:41 | |
| *** pcaruana has quit IRC | 07:42 | |
| *** nottrobin has joined #openstack-swift | 07:44 | |
| *** zhiyan has joined #openstack-swift | 07:48 | |
| *** rledisez has joined #openstack-swift | 07:49 | |
| *** cppforlife_ has joined #openstack-swift | 07:53 | |
| *** jistr has joined #openstack-swift | 07:54 | |
| *** serverascode has joined #openstack-swift | 07:59 | |
| *** Protux has joined #openstack-swift | 08:02 | |
| *** silor has joined #openstack-swift | 08:16 | |
| *** jordanP has joined #openstack-swift | 08:26 | |
| *** silor has quit IRC | 08:29 | |
| openstackgerrit | Ondrej Novy proposed openstack/swift: Rsync port in ring file. https://review.openstack.org/192198 | 08:29 |
| openstackgerrit | Ondrej Novy proposed openstack/swift: Time synchronization check in recon. https://review.openstack.org/192562 | 08:31 |
| *** jordanP has quit IRC | 08:31 | |
| *** zhill has joined #openstack-swift | 08:38 | |
| *** pcaruana has joined #openstack-swift | 08:43 | |
| *** jordanP has joined #openstack-swift | 08:45 | |
| *** zhill has quit IRC | 08:54 | |
| *** ppai is now known as ppai|afk | 08:56 | |
| *** rledisez has quit IRC | 09:25 | |
| openstackgerrit | Ondrej Novy proposed openstack/swift: Rsync port in ring file. https://review.openstack.org/192198 | 09:41 |
| *** aix has quit IRC | 09:41 | |
| *** SkyRocknRoll has joined #openstack-swift | 09:42 | |
| jith_ | hi all.. what is swift_store_key in configuring swift for glance backend... is it password for swift user?? | 09:44 |
| *** SkyRocknRoll has quit IRC | 09:47 | |
| *** SkyRocknRoll has joined #openstack-swift | 10:01 | |
| *** lcurtis has joined #openstack-swift | 10:15 | |
| *** ppai|afk is now known as ppai | 10:19 | |
| *** ig0r_ has quit IRC | 10:24 | |
| *** lcurtis has quit IRC | 10:35 | |
| *** rvasilets__ has joined #openstack-swift | 10:36 | |
| *** lcurtis has joined #openstack-swift | 10:48 | |
| *** jasondotstar has joined #openstack-swift | 10:50 | |
| *** lcurtis has quit IRC | 10:54 | |
| *** ig0r_ has joined #openstack-swift | 11:08 | |
| *** ntt has quit IRC | 11:29 | |
| *** sanchitmalhotra has quit IRC | 11:37 | |
| *** aix has joined #openstack-swift | 12:00 | |
| *** ppai has quit IRC | 12:03 | |
| *** ig0r_ has quit IRC | 12:05 | |
| *** alexcampbell has joined #openstack-swift | 12:06 | |
| *** delatte has joined #openstack-swift | 12:14 | |
| *** ppai has joined #openstack-swift | 12:15 | |
| *** cdelatte has quit IRC | 12:15 | |
| *** delattec has quit IRC | 12:17 | |
| *** chlong has joined #openstack-swift | 12:34 | |
| *** hrou has joined #openstack-swift | 12:36 | |
| *** rvasilets__ has quit IRC | 12:44 | |
| *** aagrawal has joined #openstack-swift | 13:10 | |
| *** ppai has quit IRC | 13:16 | |
| *** aagrawal has quit IRC | 13:17 | |
| tdasilva- | good morning | 13:25 |
| *** annegentl_ has joined #openstack-swift | 13:30 | |
| *** alexcampbell has quit IRC | 13:33 | |
| *** janonymous has joined #openstack-swift | 13:34 | |
| *** killa_kyle has joined #openstack-swift | 13:41 | |
| *** killa_kyle has left #openstack-swift | 13:42 | |
| *** annegentl_ has quit IRC | 13:48 | |
| *** breitz has joined #openstack-swift | 14:05 | |
| *** annegentl_ has joined #openstack-swift | 14:11 | |
| *** wbhuber has joined #openstack-swift | 14:16 | |
| *** alexcampbell has joined #openstack-swift | 14:19 | |
| *** haomaiwang has quit IRC | 14:21 | |
| *** haomaiwang has joined #openstack-swift | 14:21 | |
| *** pgbridge has quit IRC | 14:25 | |
| *** jrichli has joined #openstack-swift | 14:31 | |
| *** wbhuber_ has joined #openstack-swift | 14:35 | |
| *** _hrou_ has joined #openstack-swift | 14:35 | |
| *** sc68cal_ has joined #openstack-swift | 14:36 | |
| *** mgriffin_ has joined #openstack-swift | 14:36 | |
| *** occup4nt has joined #openstack-swift | 14:36 | |
| *** sc68cal has quit IRC | 14:37 | |
| *** sc68cal_ is now known as sc68cal | 14:37 | |
| *** matt____ has joined #openstack-swift | 14:38 | |
| *** darrenc_ has joined #openstack-swift | 14:38 | |
| *** jordan_ has joined #openstack-swift | 14:38 | |
| openstackgerrit | Hiroshi Miura proposed openstack/python-swiftclient: fix flake8 config https://review.openstack.org/202909 | 14:39 |
| *** xnox_ has joined #openstack-swift | 14:41 | |
| *** dmsimard has quit IRC | 14:42 | |
| *** wasmum- has quit IRC | 14:42 | |
| *** sudorandom has quit IRC | 14:42 | |
| *** briancline has quit IRC | 14:42 | |
| *** jordanP has quit IRC | 14:42 | |
| *** xnox has quit IRC | 14:42 | |
| *** ujjain has quit IRC | 14:42 | |
| *** jamielennox has quit IRC | 14:42 | |
| *** mattoliverau has quit IRC | 14:42 | |
| *** CrackerJackMack has quit IRC | 14:42 | |
| *** klrmn has quit IRC | 14:42 | |
| *** wbhuber has quit IRC | 14:42 | |
| *** hrou has quit IRC | 14:42 | |
| *** occupant has quit IRC | 14:42 | |
| *** mgriffin has quit IRC | 14:42 | |
| *** darrenc has quit IRC | 14:42 | |
| *** wer has quit IRC | 14:42 | |
| *** sudorandom has joined #openstack-swift | 14:42 | |
| *** briancline has joined #openstack-swift | 14:42 | |
| *** CrackerJackMack has joined #openstack-swift | 14:42 | |
| *** wasmum has joined #openstack-swift | 14:42 | |
| *** xnox_ is now known as xnox | 14:42 | |
| *** jamielennox has joined #openstack-swift | 14:42 | |
| *** ujjain has joined #openstack-swift | 14:43 | |
| *** ujjain has joined #openstack-swift | 14:43 | |
| *** dmsimard has joined #openstack-swift | 14:43 | |
| *** wer has joined #openstack-swift | 14:43 | |
| *** klrmn has joined #openstack-swift | 14:45 | |
| *** jistr is now known as jistr|mtg | 15:13 | |
| *** minwoob has joined #openstack-swift | 15:14 | |
| *** janonymous_ has joined #openstack-swift | 15:16 | |
| *** gyee has joined #openstack-swift | 15:19 | |
| *** thebloggu has joined #openstack-swift | 15:22 | |
| *** jistr|mtg is now known as jistr | 15:24 | |
| *** proteusguy has quit IRC | 15:26 | |
| *** nadeem has joined #openstack-swift | 15:27 | |
| *** marzif has joined #openstack-swift | 15:35 | |
| *** tongli has joined #openstack-swift | 15:35 | |
| *** proteusguy has joined #openstack-swift | 15:39 | |
| thebloggu | do I need special configurations to support server-side copy of objects with X-Copy-From in a PUT request or a COPY request? I'm getting an unauthorized 401 error everytime I try to do it (even though I can GET/POST other objects) | 15:42 |
| *** mgarza_ has joined #openstack-swift | 15:51 | |
| thebloggu | :q | 16:01 |
| thebloggu | ups, wrong window, sorry :) | 16:01 |
| clayg | nothing in lastlog to clayg - so I think i'm just going to skip the scroll back and get the overview from notmyname | 16:02 |
| *** gyee has quit IRC | 16:11 | |
| *** alexcampbell has quit IRC | 16:13 | |
| openstackgerrit | janonymous proposed openstack/swift: Metaclass Python 3.x Compatibility https://review.openstack.org/203714 | 16:13 |
| *** gyee has joined #openstack-swift | 16:15 | |
| *** jistr has quit IRC | 16:24 | |
| *** blmartin has joined #openstack-swift | 16:25 | |
| *** jordan_ has quit IRC | 16:32 | |
| *** SkyRocknRoll has quit IRC | 16:34 | |
| notmyname | good morning | 16:35 |
| peluse | morning | 16:35 |
| notmyname | I had a good weekend at the community leadership summit. took lots of notes and now I need to parse them to see if they make any sense. but I got some inspiration, so that's good I think | 16:36 |
| peluse | clayg, https://review.openstack.org/#/c/201283/ is the GET alt frag index thing and the fix for the bug that was about EC overwrite (1457691). Not done w/test code and may refactor a bit now that it works but ahve a gander when you get a chance. | 16:37 |
| notmyname | clayg is back! | 16:37 |
| notmyname | IIRC, acoles_away should be back this week too | 16:37 |
| peluse | also there's some code in the ResumeGetter that I don't understand that I might have duplciated - need to talk to torgomatic | 16:37 |
| peluse | oh yeah, welcome back :) | 16:38 |
| * peluse left glasses at home today. this should be interesting... | 16:39 | |
| notmyname | peluse: that's a really big deal for you. | 16:39 |
| notmyname | peluse: I think you need to ask your boss for an even bigger monitor, then :-) | 16:39 |
| peluse | they don't make a bigger one :) | 16:40 |
| *** annegentl_ has quit IRC | 16:40 | |
| *** lyrrad has joined #openstack-swift | 16:41 | |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Minor change in development saio setup doc https://review.openstack.org/203726 | 16:42 |
| clayg | does anyone know how to gerrit? I just want to star patch 198429 so I can make sure to answer torgomatic's question today | 16:42 |
| patchbot | clayg: https://review.openstack.org/#/c/198429/ | 16:42 |
| clayg | peluse: i saw that update to the bug report | 16:43 |
| clayg | lol @ "they don't make a bigger one" | 16:43 |
| peluse | clayg, where did ya go anyway? | 16:43 |
| clayg | texas | 16:43 |
| peluse | aren't we going to TX in a few weeks? :) | 16:43 |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Minor change in development saio setup doc https://review.openstack.org/203726 | 16:48 |
| clayg | whoa - there's a lot of starred patches | 16:52 |
| clayg | peluse: yeah I should have just waited! | 16:52 |
| notmyname | clayg: I starred all the patches that are associated with an EC bug report | 16:54 |
| *** occupant has joined #openstack-swift | 17:00 | |
| *** occupwnt has joined #openstack-swift | 17:01 | |
| peluse | notmyname, we're still "on track" (as much as we ever claim to be) for encryption middleware in Liberty right? Enough to claim so in a public preso? | 17:01 |
| *** occup4nt has quit IRC | 17:03 | |
| notmyname | peluse: that's a good question | 17:03 |
| notmyname | I briefly chatted with jrichli about that late last week | 17:03 |
| notmyname | there's still a bunch to do | 17:03 |
| notmyname | I've got a list (from jrichli) of some of the big outstanding things, and I'll be going through that early this week to try to highlight some of the areas that need help | 17:04 |
| * peluse bites his tongue (fingers) after almost typing the word " | 17:04 | |
| peluse | beta" | 17:04 |
| notmyname | ie where people can work | 17:04 |
| notmyname | peluse: my gut reaction is that I don't want to do that :-) | 17:05 |
| *** occupant has quit IRC | 17:05 | |
| peluse | indeed! | 17:05 |
| peluse | one beta 'something' per year should be enough :) | 17:07 |
| notmyname | my top priority is getting the EC stuff finished up. I want to see the outstanding issues closed and do a swift release with them (with enough time for another release for Liberty combined release | 17:07 |
| peluse | agreed | 17:08 |
| peluse | the patch I just posted took a really unacceptable amount of time to get working... working on slides today for all these danged upcoming conferences but will continue on that one after I'm done and then start hitting bugs... | 17:10 |
| thebloggu | I'm getting 401 unauthorized errors when I try to copy a file server side with COPY/PUT. can someone help me? | 17:10 |
| *** jordanP has joined #openstack-swift | 17:12 | |
| *** mahatic has joined #openstack-swift | 17:12 | |
| *** openstackgerrit has quit IRC | 17:16 | |
| *** openstackgerrit has joined #openstack-swift | 17:17 | |
| *** zhill has joined #openstack-swift | 17:21 | |
| *** zhill_desktop has joined #openstack-swift | 17:23 | |
| minwoob | thebloggu: Are you trying to do this in the context of single-ranged read? | 17:30 |
| thebloggu | minwoob, do you mean if the object is copied with a single range? if so, yes. I'm copying the whole object, I'm not even using range headers | 17:31 |
| notmyname | thebloggu: first thing to look at is if your token is still valid. if so, then you need to look at the ACLs on the source and destination containers | 17:32 |
| minwoob | thebloggu: Hmm then it should be fine. I was referring to the case with a single range header. | 17:32 |
| thebloggu | notmyname, the token is valid since I can make other GET and PUT requests. i'm copying inside the same container and i've got the default ACL's since I didn't change them (swift stat shows them empty so I believe I've got read and write permission right?) | 17:35 |
| thebloggu | notmyname, actually, sorry but I was wrong, I can seem to make a GET request to the object | 17:37 |
| openstackgerrit | Azhagu Selvan SP proposed openstack/swift: Minor change in development saio setup doc https://review.openstack.org/203726 | 17:38 |
| *** occupwnt is now known as occupant | 17:43 | |
| notmyname | thebloggu: ok, so you can make a GET request to the source object. that's good. can you PUT a new object to the destination container? | 17:43 |
| *** hseipp has quit IRC | 17:44 | |
| *** annegentl_ has joined #openstack-swift | 17:44 | |
| *** aix has quit IRC | 17:46 | |
| thebloggu | notmyname, I can. just for context I'm using python-swiftclient so I believe if the token's not valid it requests another one, right? | 17:48 |
| *** ccavanna has joined #openstack-swift | 17:55 | |
| *** annegentl_ has quit IRC | 17:56 | |
| *** annegentl_ has joined #openstack-swift | 17:56 | |
| *** pcaruana has quit IRC | 18:00 | |
| notmyname | it should ;-) | 18:04 |
| notmyname | but a new token doesn't solve ACLs issues. definitely try putting an object directly into the target container | 18:05 |
| notmyname | if that doesn't work, then that's likely your problem | 18:06 |
| notmyname | if it does work, then it's likely something more subtle. maybe related to your config | 18:06 |
| ccavanna | Hi. I ran into a problem starting swift after executing tox. I hope anyone here knows what may be going on. | 18:08 |
| ccavanna | I executed tox earlier today and after that I am no longe able to start Swift. | 18:09 |
| ccavanna | I get "pkg_resources.DistributionNotFound: swift==2.3.1.dev33" | 18:09 |
| ccavanna | Interestingly, under swift/swift.egg-info/PKG-INFO I see: | 18:09 |
| ccavanna | Version: 2.3.1.dev34 | 18:10 |
| ccavanna | And I also noticed that /usr/local/bin/swift-init shows | 18:10 |
| *** jasondotstar has quit IRC | 18:10 | |
| ccavanna | 'swift==2.3.1.dev33' as a requirement. | 18:10 |
| ccavanna | I noticed swift/swift.egg-info/PKG-INFO has a timestamp from today, even though I installed this a while ago. | 18:11 |
| ccavanna | So something was updated, it seems. | 18:11 |
| notmyname | IIRC that's an issue with not having the latest python-setuptools (pip) | 18:12 |
| thebloggu | notmyname, it was indeed a more subtle error. I was accidentally printing a \n to a metadata header. thank you very much for your help :) | 18:15 |
| notmyname | thebloggu: ah. yeah, that's tricky. glad you found it! | 18:16 |
| openstackgerrit | Greg Lange proposed openstack/swift: go: Adds memcache ring https://review.openstack.org/203758 | 18:28 |
| *** thebloggu has quit IRC | 18:31 | |
| *** mahatic has quit IRC | 18:35 | |
| clayg | torgomatic: so no, patch 185679 addressed the empty-dir-post-reclaim a little differently than patch 185679 | 18:37 |
| patchbot | clayg: https://review.openstack.org/#/c/185679/ | 18:37 |
| torgomatic | clayg: ah, darn... I was hoping acoles_away would just tell me "yep, it's all taken care of" | 18:37 |
| clayg | torgomatic: honestly I had forgotten about patch 185679 | 18:38 |
| patchbot | clayg: https://review.openstack.org/#/c/185679/ | 18:38 |
| clayg | I mean I *think* i looked at it once :\ | 18:38 |
| notmyname | jrichli: around? | 18:40 |
| jrichli | notmyname: hello | 18:40 |
| jrichli | reading now | 18:40 |
| notmyname | jrichli: I've reformatted the encryption todos email you sent me in an etherpad so we can all look at the stuff outstanding (from a higher level than trello) | 18:41 |
| *** jlhinson has joined #openstack-swift | 18:42 | |
| jrichli | notmyname: great! link? | 18:43 |
| notmyname | https://etherpad.openstack.org/p/swift_encryption_issues | 18:43 |
| notmyname | for everyone, these are the big problems still outstanding for encryption ^^ | 18:43 |
| jrichli | looks great, thanks! should I put a link to this on the encryption section of priority reviews page? | 18:45 |
| tdasilva- | clayg: hi! can you comment on the effect of versionining middleware to encryption? | 18:46 |
| clayg | notmyname: the look roughly ordered by priority to me - was that on purpose? | 18:46 |
| notmyname | clayg: nope | 18:46 |
| notmyname | I think they should be moved to a priority order, but it's roughly in the order jrichli gave them to me | 18:46 |
| clayg | tdasilva-: versioning and copy middleware would help encryption *greatly* | 18:47 |
| jrichli | and I roughly gave them priority order - but not completely | 18:47 |
| jrichli | i put the 2 highest at the top | 18:47 |
| clayg | tdasilva-: like as in merging version middlware moves encryption from "not going to be done by tokyo" to "probably not going to be *done, done* by tokyo" | 18:47 |
| clayg | notmyname: so you're saying you just do whatever jrichli tells you | 18:48 |
| clayg | ... smart | 18:48 |
| notmyname | lol | 18:48 |
| jrichli | lol | 18:48 |
| tdasilva- | clayg: ok, I was reading this comment: "clay seemed to think that the versioning patch for moving to middleware would affect encryption. I am not sure of the connection." and thought there was some issue with it... | 18:48 |
| jrichli | notmyname: maybe some filtering was required ;-) | 18:49 |
| jrichli | but I really do need to be educated on that | 18:50 |
| clayg | tdasilva-: ok yeah... alright let me see if I can remember how versioned middlware + encryption works - i'm *entirely* sure that encyrption + in-proxy-versions is doomed | 18:50 |
| clayg | yeah so in middleaware it shoudl be fine - but there's some questions with key management and how encryption is scoped - if the cluster manages keys and public objects work then encryption would be transparent to the versioned-objects-*middleware* | 18:51 |
| notmyname | I've put some of these topics down for this week's meeting https://wiki.openstack.org/wiki/Meetings/Swift | 18:51 |
| jrichli | thx! | 18:52 |
| clayg | notmyname: meeting looks good | 18:54 |
| notmyname | don't let that stop discussing in here now :-) | 18:54 |
| jrichli | I am still absorbing what clayg just said. On a different note: the copy as middleware would be really good to have - so we dont have to write a copy hook | 18:56 |
| sweeper | any idea why swift nodes would be reporting unmounted drives that no longer exist in the ringfile? | 18:57 |
| redbo | Reporting how? If you mean from recon, it doesn't even look at the ring. | 18:58 |
| sweeper | yea recon | 18:58 |
| sweeper | so just commented lines in fstab will make that happen? | 18:59 |
| redbo | It just looks for directories under the root that aren't mounted. | 18:59 |
| ccavanna | notmyname: I updated python-pip with apt-get. I am assuming I will have to update my swift files as well (git checkout?) | 19:00 |
| *** DericHorn-HP has joined #openstack-swift | 19:00 | |
| notmyname | ccavanna: no, you shouldn't | 19:01 |
| ccavanna | MMm. But there will still be this explicit dependency in swift-init (I'm still getting the same error). | 19:02 |
| openstackgerrit | Samuel Merritt proposed openstack/swift: Rename WsgiStringIO -> WsgiBytesIO. https://review.openstack.org/203775 | 19:03 |
| sweeper | oic | 19:03 |
| sweeper | interesting | 19:03 |
| *** losingle has joined #openstack-swift | 19:05 | |
| *** robefran has joined #openstack-swift | 19:05 | |
| clayg | depnding on how new your debian/ubuntu is you probably don't want the system python-pip | 19:06 |
| *** janonymous_ has quit IRC | 19:10 | |
| ccavanna | It's part of the installation steps for SAIO, for Ubuntu 14. | 19:15 |
| *** DericHorn-HP has quit IRC | 19:15 | |
| clayg | well - then that's a doc bug - but only because pip releases a new version every other day | 19:17 |
| wasmum | swift-bench v2 compatable? if so, anyone know the format for bench config file? | 19:18 |
| clayg | wasmum: I would have swore that chmouel_ added v2 support to swift-bench | 19:20 |
| wasmum | +clayg: saw the issue in the tracker | 19:21 |
| wasmum | not any docs on how or if the auth_version setting will allow anything other that "1" | 19:21 |
| wasmum | I'll keep digging | 19:21 |
| clayg | i'm looking at ef3e55 - in the swift code base | 19:22 |
| clayg | it ws moved into it's own repo some some of the history is in the swift tree | 19:22 |
| clayg | yeah I see -V option? | 19:23 |
| clayg | but it's just a wrapper around swiftclient get_auth? | 19:23 |
| wasmum | ah | 19:24 |
| wasmum | ok, I'll wrap that up. Just looking into a sall job for jenkins to run | 19:24 |
| wasmum | thanks, sir | 19:24 |
| wasmum | **small | 19:25 |
| *** zaitcev has joined #openstack-swift | 19:25 | |
| *** ChanServ sets mode: +v zaitcev | 19:25 | |
| *** zhill has quit IRC | 19:25 | |
| clayg | ah so client.get_auth does the ':'.split trick to tenant:username for auth v2/v3 | 19:26 |
| *** nadeem has quit IRC | 19:26 | |
| clayg | yeah idk, that might work | 19:27 |
| openstackgerrit | Thiago da Silva proposed openstack/swift: Add container reverse listings https://review.openstack.org/120709 | 19:28 |
| openstackgerrit | Greg Lange proposed openstack/swift: go: add ability to lock device/drive with file on file system https://review.openstack.org/203789 | 19:29 |
| *** hemanthm has left #openstack-swift | 19:31 | |
| openstackgerrit | Greg Lange proposed openstack/swift: go: add ability to lock device/drive with file on file system https://review.openstack.org/203789 | 19:31 |
| *** miand has joined #openstack-swift | 19:35 | |
| *** miand has quit IRC | 19:35 | |
| *** DericHorn-HP has joined #openstack-swift | 19:37 | |
| *** zhill has joined #openstack-swift | 19:38 | |
| openstackgerrit | Joel Wright proposed openstack/python-swiftclient: Reduce memory usage for download/delete and add --no-shuffle option to st_download https://review.openstack.org/130339 | 19:45 |
| *** tdasilva- is now known as tdasilva | 19:53 | |
| *** DericHorn-HP has quit IRC | 19:59 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift: Handle removed suffix dirs the same as empty suffix dirs https://review.openstack.org/185679 | 20:11 |
| clayg | torgomatic: oh - it kept you as the owner - i should have put myself as second billing :P | 20:12 |
| torgomatic | clayg: authored, co-authored, written, produced, directed by, and starring... ME :) | 20:12 |
| clayg | glange: I'm +1 just on the *title* of patch 203789 | 20:12 |
| patchbot | clayg: https://review.openstack.org/#/c/203789/ | 20:12 |
| notmyname | torgomatic: https://pbs.twimg.com/media/BhV6LA0CIAI120o.jpg | 20:13 |
| clayg | glange: esspecially if you plan on using it to lock out PUTS on full drives ;) | 20:13 |
| *** ChanServ sets mode: +v tdasilva | 20:14 | |
| glange | well, our ops_ guys asked for that -- sometimes they want to umount and do something with a drive but they have to catch it win it's not busy | 20:15 |
| clayg | glange: yeah fer-real | 20:15 |
| notmyname | glange: clayg: did I miss the first part of that conversation? | 20:15 |
| clayg | notmyname: the title of patch 203789? | 20:16 |
| patchbot | clayg: https://review.openstack.org/#/c/203789/ | 20:16 |
| notmyname | at the summit we had requests from ops people to be able to mark a drive as disabled | 20:16 |
| notmyname | ah ok | 20:16 |
| clayg | notmyname: and it is now possible - by the power of go | 20:16 |
| glange | haha | 20:16 |
| glange | somebody should port that to python at the hackathon | 20:16 |
| notmyname | well, all the cool stuff I've seen so far in hummingbird doesn't really have anything to do with go | 20:17 |
| ccavanna | No luck yet, but I traced back what I did since I could successfully run swift for the last time. I did execute this git command, as part of my check in: "git rebase -i HEAD~2". Will that change 2.3.1.dev33 to 2.3.1.dev34 ? | 20:17 |
| clayg | glange: nah - let's just hack on the the hummingbird branch of vagrant-swift-all-in-one | 20:17 |
| clayg | and have a 'how the fuck do you guys develop on this shit' getting started session | 20:17 |
| clayg | ccavanna: if the bin scripts are throwing version errors it normally means you need to reinstall the source again - but that normally squares it - i'm sure i'm missing some interesting context | 20:19 |
| clayg | ccavanna: but just about any time you change the commits if you installed with 'pip install .' instead of 'pip install -e .' then you probably need to re-run 'pip install .' | 20:19 |
| *** zhill has quit IRC | 20:20 | |
| Guest48719 | notmyname: ? | 20:20 |
| jrichli | lol, welcome back clayg. I hope you had a good time in TX. i have to admit, the channel was a little boring without you | 20:20 |
| *** Guest48719 is now known as scotticus | 20:21 | |
| notmyname | scotticus: yes? | 20:21 |
| scotticus | what cool things have you seen in hummingbird? | 20:21 |
| notmyname | scotticus: the per-drive locking we were just talking about and limiting the drives to a certain number of connections are the 2 that quickly spring to mind | 20:22 |
| notmyname | I feel like there was amother/more | 20:23 |
| scotticus | i think that was sort of made easier with Go... | 20:23 |
| clayg | jrichli: L-O-L | 20:23 |
| notmyname | ah, the async container update on an object put (for which there is a patch for the python object server) | 20:23 |
| clayg | notmyname: oh - yeah - is that patch ready? | 20:24 |
| notmyname | clayg: yup. been sitting there a while for ya :-) | 20:24 |
| clayg | awww you were *saving* it for me! | 20:24 |
| notmyname | scotticus: could be. I think there's some really interesting stuff that y'all have done on the golang side, but most of what I've seen is better algorithms than a language-specific thing | 20:25 |
| notmyname | clayg: https://review.openstack.org/#/c/189080/ | 20:25 |
| clayg | notmyname: found it on priority reviews - thanks! | 20:25 |
| notmyname | scotticus: what plans do you have for making and sharing performance comparisons? | 20:26 |
| notmyname | scotticus: of python vs golang | 20:26 |
| notmyname | I'd love to see something in Tokyo like that | 20:27 |
| *** zhill has joined #openstack-swift | 20:29 | |
| ccavanna | clayg: I'm sure I did not use pip install -e, since I was following the SAIO instructions. | 20:30 |
| ccavanna | And that's installing from a requirements file. | 20:31 |
| ccavanna | are you suggesting to just run "pip install" with no other options? What will that do? | 20:31 |
| clayg | ccavanna: I'm really sorry those instructions are out of date :\ | 20:34 |
| clayg | no - "pip install ." where "." means current directory because bash | 20:34 |
| ccavanna | Oh, we all used those instructions here :/ | 20:35 |
| ccavanna | I confirmed it shows dev34, pip list shows: | 20:36 |
| ccavanna | swift (2.3.1.dev34, /home/swift/swift) | 20:36 |
| ccavanna | I'm sorry for the 101 questions, I'm really new to all these possible problems. | 20:36 |
| clayg | ccavanna: so you used "python setup.py develop" | 20:41 |
| clayg | ccavanna: what's your `pip --version` | 20:42 |
| ccavanna | yes, I used that command (again, from SAIO :) ) | 20:42 |
| ccavanna | pip 1.5.4 from /usr/lib/python2.7/dist-packages (python 2.7) | 20:42 |
| clayg | ok - so just run that command again | 20:43 |
| ccavanna | I executed that (from python-swiftclient, as I originally did). I noticed in the output that it says: | 20:48 |
| ccavanna | Installing swift script to /usr/local/bin | 20:48 |
| ccavanna | which is probably what I'm looking for. But the files there were not overwritten. | 20:48 |
| clayg | `sudo python setup.py develop` do it in all the git dirs - it's fine | 20:50 |
| *** blmartin has quit IRC | 20:52 | |
| *** nadeem has joined #openstack-swift | 20:57 | |
| ccavanna | Great! | 20:58 |
| ccavanna | it updated when I exectued from inside the swift directory. | 20:58 |
| ccavanna | 10^6 thanks clay. | 20:59 |
| *** nadeem has quit IRC | 21:01 | |
| *** nadeem has joined #openstack-swift | 21:01 | |
| openstackgerrit | Robert Francis proposed openstack/swift: add test_copy_to_same https://review.openstack.org/203820 | 21:02 |
| notmyname | robefran: thanks for your first patch! | 21:09 |
| robefran | notmyname: just a func test for copy_to_same, mostly to go through the dev/review process, let's see if it passes the auto-testing | 21:12 |
| clayg | ccavanna: maybe you could recommend (via a bug report, or gerrit change) something that should be updated on those docs that would have helped you out? | 21:12 |
| notmyname | robefran: did it pass locally? if so, gate checks should be no problem | 21:12 |
| robefran | notmyname: yes, it passed tox (py27 pep8) and .functests locally | 21:13 |
| notmyname | great | 21:13 |
| clayg | robefran: your first attempt to contribute to the project was a new functional test of existing behavior? | 21:13 |
| * clayg hugs robefran | 21:13 | |
| notmyname | clayg: sounds like a pretty cool guy, that robefran ;-) | 21:13 |
| clayg | notmyname: wait - do I know him? | 21:13 |
| clayg | robefran: do I know you? | 21:13 |
| notmyname | I don't | 21:13 |
| clayg | nice | 21:13 |
| robefran | clayg: yes, I know there is already a unit test for copy_to_same, but I thought a functest would be a small/safe addition | 21:13 |
| clayg | robefran: good call | 21:14 |
| notmyname | robefran: I see from your contrib email that you're with IBM. are you with jrichli or wbhuber_ in Austin? | 21:14 |
| clayg | timburke: joel keeps trying to get patch 130339 merged - i apparently tried to +2 about six revisions ago - is it worth another look? | 21:15 |
| patchbot | clayg: https://review.openstack.org/#/c/130339/ | 21:15 |
| robefran | notmyname: yes, I work with Janie, but I'm in Hamdi's (hrou) and Carlos (ccavanna) team in Toronot | 21:15 |
| robefran | arrg, Toronto | 21:15 |
| peluse | torgomatic, ping me when you're available. Wanna chat a little about used_source_etag in ResummingGetter as it related to patch 201283 | 21:15 |
| patchbot | peluse: https://review.openstack.org/#/c/201283/ | 21:15 |
| notmyname | robefran: ah ok. (did we talk on the phone about SMR drives last week?) | 21:16 |
| clayg | robefran: cool! does that mean you'll be in austin? | 21:16 |
| wbhuber_ | notmyname: no, robefran is with IBM but in a different location. | 21:16 |
| clayg | fwiw google auto corrects Toronot to Toronto | 21:16 |
| ccavanna | clayg: yes, I will do that. I need to think what would be the best change. | 21:16 |
| robefran | notmyname: I think you were talking to Hamdi (hrou) about SMR | 21:16 |
| notmyname | robefran: yup | 21:16 |
| ccavanna | notmyname: and me. | 21:17 |
| notmyname | ah, ok! | 21:17 |
| * peluse wonders what a toronot is | 21:17 | |
| clayg | ccavanna: that would be so very helpful - people that have worked on the project a long time loose all perspective when it comes to encountering this stuff for the first time :'( | 21:17 |
| timburke | clayg: yes. seems good as of last patchset; haven't looked at latest, but sounds like it's just resolving a merge conflict | 21:18 |
| clayg | peluse: maybe it's like a torus - but not? | 21:18 |
| ccavanna | clayg: it happens to everone after a while. | 21:18 |
| clayg | peluse: maybe we could design a new placement algorithm that's based on a toronot? | 21:18 |
| jrichli | peluse: sounds like something from my topology class | 21:18 |
| peluse | heh | 21:19 |
| dfg_ | are there some instructions somewhere about how to get you SAIO unit tests running? | 21:19 |
| clayg | dfg_: do yours not work? | 21:20 |
| dfg_ | no | 21:21 |
| clayg | dfg_: maybe related to pyeclib if you haven't been following along? you can try to pip a few things? | 21:21 |
| clayg | `sudo pip install -r requirements.txt ; sudo pip install -r test-requirements.txt` maybe? | 21:21 |
| dfg_ | i think i tried that | 21:23 |
| notmyname | or if your SAIO is out of sync enough, might just be easier to set it up again | 21:23 |
| dfg_ | notmyname: tried that too :) | 21:23 |
| *** losingle has quit IRC | 21:23 | |
| clayg | dfg_: any particuarl group of tests failing for any particularlly coherent reason? | 21:23 |
| dfg_ | like 30 failures 40 something errors or somehting? | 21:24 |
| dfg_ | as far as people know if you follow the SAIO instructions on the docs they should pass? | 21:24 |
| notmyname | yup. running mine now | 21:25 |
| clayg | dfg_: that might not be true for the latest versions of some stuff that pip drags in - esspecially on an old distro using system pip | 21:25 |
| dfg_ | i created a brand new VM, precise, followed the instructions and got a ton of failures | 21:25 |
| clayg | dfg_: but ccavanna went through this recently apparenlty - so he mgiht now? | 21:25 |
| clayg | *precise* | 21:25 |
| clayg | dfg_: I'm guessing there's a few battles in there for you | 21:26 |
| dfg_ | i don't want to fight these battles | 21:26 |
| clayg | dfg_: I'd drop system setuptools and pip asap | 21:26 |
| clayg | dfg_: then use vagrant-swift-all-in-one | 21:26 |
| clayg | no one has time to dick with that crap | 21:26 |
| notmyname | Ran 4051 tests in 121.474s (all OK, on precise) | 21:26 |
| dfg_ | :/ | 21:27 |
| clayg | notmyname: you have a working precise install! | 21:27 |
| clayg | notmyname: i'm floored | 21:27 |
| dfg_ | so the instructions on how to get this working are to use the swiftstack vagrant saio image? | 21:28 |
| notmyname | I normally try to have the min version we say is required installed | 21:28 |
| dfg_ | or use trusty? | 21:28 |
| dfg_ | or what? | 21:28 |
| clayg | dfg_: NO ONE IN THIS FUCKING CHANNEL HAS ANY IDEA WHAT ERRORS YOU ARE SEEING | 21:28 |
| *** robefran has quit IRC | 21:28 | |
| clayg | dfg_: you want help tell me what the problem is | 21:29 |
| notmyname | dfg_: no, the instructions are the saio docs at swift.openstack.org like always. but if you don't want to mess with all the nasty details and just get it working, you can use the vagrant swift all in one | 21:29 |
| clayg | you don't wnat to say what's wrong try the vagrant image or trusty or something else | 21:29 |
| clayg | geez | 21:29 |
| dfg_ | clayg: do you need a nap or something? | 21:29 |
| notmyname | lol | 21:29 |
| clayg | dfg_: probably moar coffee ;) | 21:29 |
| dfg_ | i'll try it again. thx | 21:29 |
| *** pgbridge has joined #openstack-swift | 21:29 | |
| openstackgerrit | Janie Richling proposed openstack/swift: Encrypting/Decrypting account and container metadata https://review.openstack.org/203454 | 21:34 |
| *** jordanP has quit IRC | 21:39 | |
| *** ccavanna has quit IRC | 21:57 | |
| *** thurloat_isgone is now known as thurloat | 21:58 | |
| *** peterbusque has joined #openstack-swift | 22:00 | |
| *** tongli has quit IRC | 22:00 | |
| openstackgerrit | Joel Wright proposed openstack/python-swiftclient: Log and report trace on service operation fails https://review.openstack.org/171692 | 22:13 |
| *** robefran has joined #openstack-swift | 22:14 | |
| *** matt____ is now known as mattoliverau | 22:15 | |
| *** ChanServ sets mode: +v mattoliverau | 22:15 | |
| mattoliverau | morning everyone | 22:16 |
| notmyname | wbhuber_: does the comment from peluse make sense on https://review.openstack.org/#/c/201791/? | 22:16 |
| notmyname | mattoliverau: hullo | 22:16 |
| mattoliverau | notmyname: have fun at the weekend conference? | 22:16 |
| peluse | do my comments ever make sense? | 22:16 |
| notmyname | mattoliverau: yeah, it was good. I met https://twitter.com/dfflanders from your neighborhood | 22:18 |
| *** mgriffin_ has quit IRC | 22:18 | |
| *** mgriffin_ has joined #openstack-swift | 22:18 | |
| *** mgriffin_ is now known as mgriffin | 22:18 | |
| notmyname | mattoliverau: I got some good ideas to think about and some stuff to work on. so that was good | 22:19 |
| *** zhill has quit IRC | 22:19 | |
| wbhuber_ | notmyname: yep. his comments made sense. i've to go close and abandon it. | 22:20 |
| mattoliverau | notmyname: wow, he follows me, he looks familar so he must be one of the guys from the openstack meetup group here in melb. | 22:20 |
| notmyname | mattoliverau: yeah. he's at uni melb | 22:21 |
| notmyname | runs some of the nectar stuff there, I believe | 22:21 |
| mattoliverau | ahh, yes I do know him then :) | 22:21 |
| robefran | my patch https://review.openstack.org/#/c/203820/ failed in gate-swift-dsvm-functional, what is gate-swift-dsvm-functional? and how do I run it locally? | 22:21 |
| notmyname | robefran: it's the `tox -e func` tests | 22:22 |
| clayg | looks like a build error tho | 22:22 |
| notmyname | robefran: which are the `./.functests` | 22:22 |
| *** jrichli has quit IRC | 22:22 | |
| notmyname | right, not your code | 22:22 |
| clayg | distutils.errors.DistutilsError: Could not find suitable distribution for Requirement.parse('pbr') | 22:22 |
| mattoliverau | notmyname: so I have a beta version of shrinking in container sharding written, so had a productive Friday (and weekend) :) | 22:22 |
| robefran | I ran ./.functests directly, should I run tox -e func instead? | 22:22 |
| notmyname | mattoliverau: nice!! | 22:22 |
| notmyname | robefran: no | 22:23 |
| mattoliverau | now just need to test | 22:23 |
| notmyname | robefran: note the last comment on the review now | 22:23 |
| notmyname | robefran: the elastic recheck bot noticed the failure and corollated it with other failures that have been seen. there's already a bug for it | 22:23 |
| robefran | notmyname: how to I trigger a recheck? | 22:24 |
| notmyname | so I added the recheck comment | 22:24 |
| notmyname | robefran: I just did :-) | 22:24 |
| notmyname | syntax is "recheck [bug number]" | 22:24 |
| notmyname | wbhuber_: thanks | 22:26 |
| *** jlhinson has quit IRC | 22:27 | |
| wbhuber_ | peluse: the charts you sent is based for 3-way replica? | 22:28 |
| wbhuber_ | Not sure about "tuned" vs "untuned" | 22:29 |
| *** nadeem has quit IRC | 22:32 | |
| peluse | wbhuber_, yes 3x only for now and 'tuned' just means where Bill was able to tweak settings for incremental performance | 22:34 |
| *** zhill has joined #openstack-swift | 22:37 | |
| *** _hrou_ has quit IRC | 22:42 | |
| *** km has joined #openstack-swift | 22:51 | |
| *** chlong has quit IRC | 22:54 | |
| *** zhill has quit IRC | 22:59 | |
| clayg | notmyname: crap - there's local trick that you have to do to keep the transaction id in the spanwned greenthread for the container update :'( | 23:00 |
| notmyname | bah! | 23:01 |
| notmyname | got a pointer to it? | 23:01 |
| clayg | notmyname: looking | 23:01 |
| clayg | just noticed when reviewing and trying to figure out what's going on in the logs :'( | 23:02 |
| torgomatic | 0x107940120 | 23:02 |
| clayg | then I remember the thing | 23:02 |
| clayg | torgomatic has a pointer | 23:02 |
| clayg | kudos | 23:02 |
| torgomatic | someone had to say it | 23:02 |
| clayg | notmyname: some proxy code is going to do it - there's probably a comment | 23:02 |
| clayg | swifterdarrell: ran into it during the per-disk-object-server testing too I think - not sure why - his thing may have just been an existing bug? | 23:03 |
| *** peterbusque has left #openstack-swift | 23:06 | |
| clayg | notmyname: so this was swifterdarrell's bug -> https://bugs.launchpad.net/swift/+bug/1409302 | 23:10 |
| openstack | Launchpad bug 1409302 in OpenStack Object Storage (swift) "Use of GreenAsyncPile can lose txn_id logging" [Undecided,New] | 23:10 |
| *** hrou has joined #openstack-swift | 23:11 | |
| notmyname | clayg: looks straightforward https://gist.github.com/notmyname/4121933a86aa3477b2f6 | 23:13 |
| notmyname | I'll write a test and resubmit it | 23:14 |
| clayg | ummm gists are different | 23:14 |
| clayg | notmyname: we've got to quit doing this -> https://gist.github.com/clayg/b36f6196156999f1671c | 23:15 |
| notmyname | lol | 23:15 |
| notmyname | ah, yeah. there's one other call. so your way is better with the named param | 23:17 |
| clayg | notmyname: yeah i wasn't sure what tests would have to say about the new arg if I didn't make it optional :\ | 23:17 |
| clayg | notmyname: I was sorta liking your way of making it required and forcing it | 23:18 |
| notmyname | delete_at_update() is the only other caller | 23:18 |
| clayg | notmyname: oh it defaults to node_timeout :'( | 23:18 |
| clayg | yeah the default should be "better" | 23:18 |
| notmyname | yup agreed. we had talked about that. I'll do it with this patchset | 23:19 |
| notmyname | node timeout is 3. new default = 1? my tests in the commit message used .75 | 23:19 |
| *** bitblt has joined #openstack-swift | 23:20 | |
| clayg | notmyname: no no - saying there *should* be a better default is the easy part - acctually *picking* the better default gets into the realm of biksheadding - glwt | 23:20 |
| notmyname | ok, it's 1.0 | 23:21 |
| *** kei_yama has joined #openstack-swift | 23:22 | |
| clayg | bah - i tried putting the option in the container-server config :'( | 23:27 |
| *** minwoob has quit IRC | 23:27 | |
| notmyname | and? | 23:28 |
| clayg | heh - oddly it did not effect the value of the object-server's container_update_timeout value!? | 23:29 |
| *** janonymous has quit IRC | 23:31 | |
| *** mgarza_ has quit IRC | 23:47 | |
| peluse | torgomatic, got a sec to talk EC GET fun? | 23:47 |
| torgomatic | peluse: sure | 23:47 |
| clayg | peluse: that *does* sound fun | 23:48 |
| peluse | Cool... will try IRC first and if I make no sense resort to telephone.. patch is 201283 | 23:48 |
| torgomatic | patchbot: patch 201283 | 23:48 |
| patchbot | torgomatic: https://review.openstack.org/#/c/201283/ | 23:48 |
| patchbot | torgomatic: https://review.openstack.org/#/c/201283/ | 23:48 |
| peluse | so... we fire off ndata GETs and if we get back <n where one is old (overwrite during node down) we'd basically puke | 23:49 |
| peluse | what I did was start paying attention to ETAGs of the original object and sort the incoming FAs into ETAG buckets, then find the largest one and use that | 23:49 |
| torgomatic | seems reasonable | 23:49 |
| peluse | and if it wasn't big enough, jump back to building anohter set of getters to go fetch from the parity nodes | 23:49 |
| peluse | works pretty good. | 23:50 |
| peluse | quesiton is on ResumingGetter and an element in there that looks like it might have been trying to do something similar | 23:50 |
| peluse | one sec | 23:50 |
| peluse | used_source_etag - I think you added that one? I tried to sort out what its goal in life is but can't seem to get it | 23:50 |
| peluse | (BTW this is just a part of the patch, will cover the rest here in a sec after we talk about this used_source_etag thing) | 23:51 |
| peluse | while you're reading code, my best guess was that it was trying to track etags on GET and if it found one that didn't match a previous find would 404 forcing the node_iter to pick up the next node. However... | 23:52 |
| torgomatic | that doesn't look familiar, but it appears to be enforcing that we don't ever mix two FAs with different source etags | 23:52 |
| peluse | either I misunderstand otally (probably) or since there's one ResumeGetterper FA GET, it'd never find a 2nd | 23:53 |
| openstackgerrit | Merged openstack/swift: Minor change in development saio setup doc https://review.openstack.org/203726 | 23:53 |
| torgomatic | the only relevant bit is "if self.used_source_etag != src.headers.get('x-object-sysmeta-ec-etag', ...)" | 23:53 |
| peluse | OK, so I wasn't far off... but how, when there's one instance of the ResumeGetter per GET and it gets just one FA would there ever be a 2nd one to not match? | 23:53 |
| torgomatic | uhh | 23:54 |
| peluse | this line confuses my small brain: "if not self.newest: # one good source is enough" | 23:54 |
| torgomatic | good question; let me throw out that whole if branch and see if anything fails | 23:55 |
| peluse | in _get_source_and_node() | 23:55 |
| clayg | peluse: torgomatic: I'm not sure x-newest makes sense at all in EC land | 23:55 |
| peluse | I think it might be dead code | 23:55 |
| notmyname | clayg: or in non-ec land, really ;-) | 23:55 |
| clayg | it'd be one weird schema failure if you can rebuild multiple versions of an ec object | 23:55 |
| torgomatic | peluse: ECObjectController always calls ResumingGetter(..., newest=False) | 23:56 |
| torgomatic | so you're right, x-newest + EC is nonsense | 23:56 |
| clayg | notmyname: why doesn't x-newest make sense in the replicated land - if you have multiple versions - why *wouldn't* you want the newest! | 23:56 |
| peluse | torgomatic, ahh, I missed that false parm... | 23:56 |
| *** nadeem has joined #openstack-swift | 23:56 | |
| notmyname | clayg: yeah, but "newest" is defined by "latest timestamp found on primary nodes", not latest in the cluster. x-newest is ok, but it certainly isn't a guarantee | 23:57 |
| clayg | notmyname: x-newest always digs handoffs | 23:57 |
| notmyname | really!? | 23:57 |
| clayg | it doesn't check the *whole* cluster - but it's not terrible | 23:57 |
| clayg | yeah post-as-copy hates your cluster | 23:57 |
| notmyname | I thought it was just the timestamps on primaries | 23:57 |
| clayg | :'( | 23:57 |
| clayg | sorry bro | 23:57 |
| notmyname | ok, so better for users, worse for the cluster, and apps still need to know about eventual consistency ;-) | 23:58 |
| peluse | OK, so seems like self.used_source_etag was clearly added for EC though (it checks the ec etag header) but still AFAICT its dead code. Or at least with the changes I made on that patch its not needed I dont think | 23:58 |
| *** darrenc_ is now known as darrenc | 23:58 | |
| peluse | maybe :) | 23:58 |
| torgomatic | peluse: well, when I comment out that code block, I get a couple failing tests, so something's using that functionality | 23:59 |
| torgomatic | OH | 23:59 |
| torgomatic | right | 23:59 |
| torgomatic | so it's a ResumingGetter, right? | 23:59 |
| peluse | heh, test code | 23:59 |
| peluse | yeah | 23:59 |
| torgomatic | if shit breaks and it resumes, it's gotta resume from something with the same etag | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
