Thursday, 2014-08-14

openstackgerritSamuel Merritt proposed a change to openstack/swift: Refactor object PUT to make EC easier to add
*** nosnos has joined #openstack-swift00:11
openstackgerritA change was merged to openstack/swift: Add POST and DELETE to tempurl default methods
*** addnull has joined #openstack-swift00:40
*** tdasilva has quit IRC00:46
*** nosnos has quit IRC00:46
*** shri has quit IRC00:51
*** Krast has quit IRC01:17
*** Krast has joined #openstack-swift01:17
*** addnull has quit IRC01:37
*** nosnos has joined #openstack-swift01:38
*** tsg has joined #openstack-swift01:48
*** tgohad has quit IRC01:48
*** addnull has joined #openstack-swift01:53
*** haomaiw__ has quit IRC01:53
*** haomaiwang has joined #openstack-swift01:53
*** nexusz99 has joined #openstack-swift02:09
mattoliveraunotmyname: are you still around?02:31
*** haomaiw__ has joined #openstack-swift02:40
*** haomaiwang has quit IRC02:44
*** tongli has quit IRC03:06
*** nosnos has quit IRC03:32
*** jyoti-ranjan has joined #openstack-swift03:34
*** mkerrin has quit IRC03:43
*** nexusz99_ has joined #openstack-swift03:48
*** addnull_ has joined #openstack-swift03:48
*** mkerrin has joined #openstack-swift03:50
*** dmsimard_away has quit IRC03:52
*** joearnold has quit IRC03:52
*** bgmccollum has quit IRC03:52
*** charz has quit IRC03:52
*** nexusz99 has quit IRC03:52
*** mordred has quit IRC03:52
*** CybergeekDK has quit IRC03:52
*** addnull has quit IRC03:52
*** jroll has quit IRC03:52
*** infotection has quit IRC03:52
*** CybergeekDK has joined #openstack-swift03:52
*** bgmccollum has joined #openstack-swift03:52
*** joearnold has joined #openstack-swift03:53
*** charz has joined #openstack-swift03:53
*** dmsimard_away has joined #openstack-swift03:53
*** infotect- has joined #openstack-swift03:53
*** mordred has joined #openstack-swift03:53
*** dmsimard_away is now known as dmsimard03:54
*** gyee has quit IRC03:54
*** jroll has joined #openstack-swift03:54
*** mkerrin has quit IRC03:55
*** nosnos has joined #openstack-swift04:01
*** bgmccollum_ has joined #openstack-swift04:06
*** joearnold_ has joined #openstack-swift04:07
*** bgmccollum has quit IRC04:07
*** joearnold has quit IRC04:07
*** CybergeekDK has quit IRC04:07
*** joearnold_ is now known as joearnold04:07
*** CybergeekDK has joined #openstack-swift04:07
*** zaitcev has quit IRC04:22
*** jyoti-ranjan has quit IRC04:24
*** nexusz99_ has quit IRC04:28
*** Midnightmyth has joined #openstack-swift04:40
*** kota_ has joined #openstack-swift04:43
*** ppai has joined #openstack-swift04:45
*** chandankumar has joined #openstack-swift04:51
*** ZBhatti has joined #openstack-swift04:55
openstackgerritMatthew Oliver proposed a change to openstack/swift: Treat 404s as 200 on object delete in proxy
*** jyoti-ranjan has joined #openstack-swift05:00
*** Midnightmyth has quit IRC05:03
*** mkerrin has joined #openstack-swift05:08
*** kopparam has joined #openstack-swift05:12
*** kopparam has quit IRC05:13
*** mkerrin has quit IRC05:13
*** kopparam has joined #openstack-swift05:13
*** akp has quit IRC05:13
*** akp has joined #openstack-swift05:14
*** mahati has joined #openstack-swift05:19
*** haomaiw__ has quit IRC05:31
*** haomaiwang has joined #openstack-swift05:31
*** chandankumar has quit IRC05:40
*** haomai___ has joined #openstack-swift05:42
*** haomaiwang has quit IRC05:46
*** ppai has quit IRC05:48
*** mkerrin has joined #openstack-swift05:50
*** chandankumar has joined #openstack-swift06:02
*** ttrumm has joined #openstack-swift06:02
*** mkerrin has quit IRC06:02
*** ppai has joined #openstack-swift06:03
*** ttrumm_ has joined #openstack-swift06:03
*** ttrumm has quit IRC06:07
*** mkerrin has joined #openstack-swift06:11
*** tsg has quit IRC06:13
openstackgerritMatthew Oliver proposed a change to openstack/swift: Treat 404s as 200 on object delete in proxy
openstackgerritMatthew Oliver proposed a change to openstack/swift: Treat 404s as 204 on object delete in proxy
*** k4n0 has joined #openstack-swift06:30
*** ppai has quit IRC06:36
*** bvandenh has joined #openstack-swift06:49
*** ppai has joined #openstack-swift06:49
*** madhuri has joined #openstack-swift06:57
madhuriHi can any one tell me what is the significance of "used_source_etag"?06:58
*** addnull_ is now known as addnull07:00
*** kopparam has quit IRC07:00
*** haomai___ has quit IRC07:12
*** haomaiwang has joined #openstack-swift07:12
*** haomaiw__ has joined #openstack-swift07:15
*** haomaiwang has quit IRC07:18
*** mkerrin has quit IRC07:22
*** occup4nt has joined #openstack-swift07:24
*** ppai has quit IRC07:27
*** occupant has quit IRC07:27
*** occupwnt has joined #openstack-swift07:28
*** mkerrin has joined #openstack-swift07:30
*** occup4nt has quit IRC07:31
*** occupant has joined #openstack-swift07:32
*** occupwnt has quit IRC07:34
*** ppai has joined #openstack-swift07:44
*** mahati has quit IRC07:44
*** occup4nt has joined #openstack-swift07:46
*** occupant has quit IRC07:49
*** occupwnt has joined #openstack-swift07:49
*** ppai has quit IRC07:49
*** mrsnivvel has joined #openstack-swift07:51
*** occup4nt has quit IRC07:52
*** ppai has joined #openstack-swift07:53
*** mrsnivvel has left #openstack-swift07:53
*** kopparam has joined #openstack-swift07:54
*** occupant has joined #openstack-swift07:56
*** homegrown has quit IRC07:56
*** occupwnt has quit IRC07:56
*** kopparam has quit IRC07:58
*** occup4nt has joined #openstack-swift08:00
*** occupwnt has joined #openstack-swift08:01
*** occupant has quit IRC08:02
*** occup4nt has quit IRC08:04
*** haomaiw__ has quit IRC08:04
*** haomaiwang has joined #openstack-swift08:05
*** ppai has quit IRC08:07
*** mkollaro has joined #openstack-swift08:10
*** haomai___ has joined #openstack-swift08:11
*** haomaiwang has quit IRC08:14
*** kopparam has joined #openstack-swift08:15
*** jyoti-ranjan has quit IRC08:19
*** ppai has joined #openstack-swift08:20
*** foexle has joined #openstack-swift08:22
*** jyoti-ranjan2 has joined #openstack-swift08:23
*** joeljwright has joined #openstack-swift08:25
*** ppai has quit IRC08:26
*** sts_ is now known as sts08:30
*** sts has left #openstack-swift08:31
*** ppai has joined #openstack-swift08:39
*** leopoldj has joined #openstack-swift08:48
*** kopparam has quit IRC09:08
*** kopparam has joined #openstack-swift09:09
*** haomai___ has quit IRC09:11
*** haomaiwa_ has joined #openstack-swift09:11
*** jyoti-ranjan2 has quit IRC09:11
*** kopparam has quit IRC09:13
acolesportante: so that line has changed since i did setup! there was no 'tr'. sounds like you have it sorted but FWIW my lib/keystone:
*** ppai has quit IRC09:29
*** jyoti-ranjan has joined #openstack-swift09:30
*** kota_ has quit IRC09:36
*** kopparam has joined #openstack-swift09:49
*** nosnos has quit IRC10:16
*** chandankumar has quit IRC10:19
goodesgood morning10:22
*** haomaiw__ has joined #openstack-swift10:24
*** haomaiwa_ has quit IRC10:27
*** chandankumar has joined #openstack-swift10:31
*** stevie99 has joined #openstack-swift10:38
*** haomaiw__ has quit IRC10:50
*** haomaiwang has joined #openstack-swift10:51
*** bvandenh has quit IRC11:00
portanteacoles: yes, at least my full setup works.11:05
portanteso can you share how to get my devstack to use my other swift instances in my SAIO's?11:06
*** bvandenh has joined #openstack-swift11:06
acolesportante: yes. sec11:07
acolesportante: that'll be this link
*** jyoti-ranjan has quit IRC11:08
acolesportante: do you have openstackclient working?11:08
portanteacoles: have not taken that step yet, but will be today. :)11:13
acolesportante: you have the notion of a swift service in the devstack kesytone by default (assuming you configured swift to run in your devstack as per that wiki page)11:15
acolesportante: you need to add your SAIO as an endpoint for that service11:16
acolesso its not too much work - if you can authenticate and talk to the devstack swift, then you'll use same auth credentials to talk to your SAIO swift once its set as an endpoint11:17
acolesendpoint can be added to keystone using the keystoneclient cli or openstack client, but to set up user/project in a non-default domain you will need openstackclient11:19
acoles...then follow the instructions on that wiki - first to add domain/project/user to exercise the v 3 features, then to add your own SAIO as endpoint and point your SAIO to the devstack keystone.11:21
acolesportante: ping me if you need more11:22
*** leopoldj has quit IRC11:29
*** kopparam has quit IRC11:30
openstackgerritA change was merged to openstack/swift: Catch permissions errors when writing StatsD packets
*** Krast has quit IRC11:31
*** tongli has joined #openstack-swift11:36
portanteacoles: thanks!11:47
*** kopparam has joined #openstack-swift11:51
*** addnull has quit IRC12:03
*** infotect- has quit IRC12:04
*** kopparam has quit IRC12:04
*** infotection has joined #openstack-swift12:10
*** stevie99 has quit IRC12:22
*** tdasilva has joined #openstack-swift12:30
*** bvandenh has quit IRC12:31
*** jyoti-ranjan has joined #openstack-swift12:37
openstackgerritOpenStack Proposal Bot proposed a change to openstack/swift: Updated from global requirements
*** bvandenh has joined #openstack-swift12:41
openstackgerritSalvatore Pinto proposed a change to openstack/swift: Add Keystone URI to 401 responces
*** jyoti-ranjan has quit IRC13:09
*** nshaikh has quit IRC13:20
*** annegentle_ has joined #openstack-swift13:35
*** annegentle has quit IRC13:35
*** bill_az_ has joined #openstack-swift13:39
*** tsg has joined #openstack-swift13:43
portanteacoles: still there?13:59
acolesportante: yup13:59
portanteI was able to list the endpoints with, openstack --os-url http://devstack-0-perf44:5000/v3 --os-identity-api-version 3 --os-token=azertytoken endpoint list13:59
portantebut I get the following error when I try to create a new region14:00
*** bvandenh has quit IRC14:00
acolesnot good14:00
portante--publicurl is required14:00
acolesok rings a bell, hang on14:00
acolesportante: i think the command line syntax may have changed (mea culpa for not checking) see
acoleslooks like publicurl is now flagged rather than positional14:02
portanteyes, and when I changed that, then I get ...14:02
portante"openstack No service with a type, name or ID of 'object-store' exists"14:03
portanteso I am guessing that should be "swift"14:03
portantethat does not work either14:04
acolesif you do openstack service list what do you see - swift, object-store ?14:04
portanteservice name: swift, service type, object-store14:05
portantethere are three listed, admin, public, internal14:05
portanteacoles: maybe I have to specify all three urls?14:06
*** jyoti-ranjan has joined #openstack-swift14:07
*** astellwag has quit IRC14:07
peluseportante:  have to run my daughter to school but when I get back we can compare notes if you want, mine still works14:08
acolesmaybe, but that doc shows only public being required. could you try passing the service ID rather than type (listed by openstack service list)?14:08
acolesportante: i'll update my openstackclient to reproduce14:10
*** xianghuihui has joined #openstack-swift14:10
portantek, thanks14:10
*** astellwag has joined #openstack-swift14:13
*** ttrumm_ has quit IRC14:24
*** dmsimard is now known as dmsimard_away14:30
*** dmsimard_away is now known as dmsimard14:30
*** tsg has quit IRC14:38
*** k4n0 has quit IRC14:45
*** nshaikh has joined #openstack-swift14:45
*** foexle has quit IRC14:45
*** kenhui has joined #openstack-swift14:47
*** aerwin has joined #openstack-swift14:50
peluseportante:  I'm back now, here's some sample output from my working v3 setup if it helps:
* portante looks ...14:57
*** nshaikh has quit IRC14:59
portantepeluse: so what environment variables do you have set for the keystone service-list command to work?15:01
acolesportante: peluse: been in dependency hell! but i think my openstackclient is up to date - openstack --version -> openstack
portanteI have 0.4.015:02
portanteacoles: ^^^15:02
peluseportante: export OS_TENANT_NAME=service15:04
peluseexport OS_USERNAME=swift15:04
peluseexport OS_PASSWORD=intel12315:04
peluseexport OS_AUTH_URL=http://controller:35357/v2.015:04
acolesportante: i get usage: openstack endpoint create [-h] [-f {shell,table,value}] [-c COLUMN]15:04
acoles                                 [--max-width <integer>] [--variable VARIABLE]15:04
acoles                                 [--prefix PREFIX] [--region <region>]15:04
acoles                                 [--enable | --disable]15:04
acoles                                 <service> <interface> <url>15:04
portantek thx, peluse15:04
peluseportante:  and my ver is openstack
portantegotta update15:04
acolesi.e. it does NOT like --publicurl ? but you see different15:04
pelusend I don't use --publicurl on any of my cmd lines for testing this...15:06
acolesi can generate portante's error but only by passing bogus service type: "No service with a type, name or ID of 'banana' exists"15:07
*** kevinc_ has joined #openstack-swift15:08
acoles'object-store' works fine, ther service ID works fine, 'swift' doesn't work15:08
*** zaitcev has joined #openstack-swift15:09
*** ChanServ sets mode: +v zaitcev15:09
*** tsg has joined #openstack-swift15:09
openstackgerritA change was merged to openstack/python-swiftclient: Clean up raw policy stats in account stat
*** joeljwright has quit IRC15:15
*** __lgw4__ has joined #openstack-swift15:16
portanteacoles, peluse: so I am using the openstack command on my devstack15:20
portantecould that be the problem?15:20
peluseportante:  I went the non-devstack route so not sure but I wouldn't think so15:21
*** IRTermite has quit IRC15:24
*** IRTermite has joined #openstack-swift15:24
*** peterbusque has joined #openstack-swift15:24
acolesportante: on my devstack i have openstackclient
acoles and i see same cli requirements15:25
*** peterbusque has left #openstack-swift15:25
portantehmm, openstackclient is the same as python-openstackclient repo, and provides the openstack command, right?15:29
portanteacoles: ^^^15:30
peluseportante:  yup15:30
acolespretty sure. to be more precise, on my devstack openstack --version returns
peluseat least that's what I thought :)15:30
*** tong_ has joined #openstack-swift15:31
portanteseems funny to use a different openstackclient version than the one that comes with a devstack15:33
acolesportante: yeah. so back to your original error, did you try using the service id with the --public-url fomr of the command?15:39
portantewell, no, I am trying to work with the 0.4.0 openstack client, as I cannot seem to install the latest openstackclient on my other Fedora VM ...15:39
*** tongli has quit IRC15:40
*** omame has quit IRC15:40
* acoles is probably getting confused15:42
*** omame has joined #openstack-swift15:42
acolescan you use 'endpoint create --public-url http://saio:8080/v1 <service id>15:43
portanteokay, I think I have worked this out, was able to use the 0.4.0 client to create public, internal and admin interface15:43
portanteI did it by first using all environment variables for --os-url, identity version, token, until I had a plain command line like:15:44
portanteopentack endpoint list15:44
portanteopenstack endpoint create --region SAIO_0_Region object-store internal 'http://saio-0-perf44:8080/v1/AUTH_$(tenant_id)s'15:44
portantethat works with this15:45
portanteI don't know why this is that different from what you have here:
acolesyeah, i am perplexed!15:47
peluseinteresting, I just used keystone to create my endpoints...15:47
acolesand does 'openstack endpoint list' show your SAIO_)_Region endpoints?15:48
portanteyes, I now have the default RegionOne end ponts for the swift instance in the devstack environment, and the two regions I created for my two SAIOs15:49
portanteacoles: peluse: heading out for a ride, will continue with all this when I get back.15:51
acolesok, so hopefully you can continue with the steps on the wiki, to config your saio-0-perf44 to use devstack keystone15:51
portantethanks for helping me out15:51
portanteyes, hopefully15:51
acolesportante: no problem.15:52
*** Midnightmyth has joined #openstack-swift15:55
notmynamegood morning16:27
goodesnotmyname: good morning16:32
notmynamewhat's the news from (my) overnight?16:32
*** mkollaro1 has joined #openstack-swift16:35
*** mkollaro has quit IRC16:35
acolesnotmyname: morning. news is that portante's openstackclient is different to mine :( but we're getting there i hope.16:39
*** chandankumar has quit IRC16:46
*** pconstantine has quit IRC16:54
*** chandankumar has joined #openstack-swift17:02
*** pberis has joined #openstack-swift17:04
*** pconstantine has joined #openstack-swift17:09
*** pberis has quit IRC17:10
*** pberis has joined #openstack-swift17:12
peluseand different still from the one I used :)17:19
*** Midnightmyth has quit IRC17:20
*** shri has joined #openstack-swift17:21
*** pconstantine has quit IRC17:43
notmynamehmmm..what is this policy constraints middleware?17:52
zaitcevwe discussed it in the meeting yesterday and you ask it now?17:55
zaitcevI haven't looked it though17:55
notmynameI've slept since then!17:57
notmynameI'm looking at it now17:57
zaitcevOkay. I don't understand why it should be a middleware, but we've got a bunch of core features implemented so, like DLO.17:58
notmynametdasilva: ping17:58
notmynametdasilva: we're talking about your code :-)17:59
*** pconstantine has joined #openstack-swift17:59
notmynameI can understand per-policy constraints17:59
*** Midnightmyth has joined #openstack-swift18:00
notmynametdasilva: my initial impressing is that a "mem_constraints" middleware seems unnecessary.18:03
notmynamesetting the env callback (swift.constraints) seems an interesting way to solve it, too18:03
notmynamebut I'd like to see not "here's the default one in and then I had an override in this other middleware" but rather just one way. that is, do it all in a middleware or do it all in the module (eg pulled from the config)18:05
notmynametdasilva: did you consider allowing the storage policies sections in swift.conf to set constraints? eg set the default max object size to X, but in storage policy 36 set it to Y18:06
zaitcevmy thought too - at least from sysadmin's perspective18:08
notmynametdasilva doesn't seem to be active online right now. I'll copy/paste these comments into gerrit18:08
*** erlon has quit IRC18:10
*** jyoti-ranjan has quit IRC18:11
*** annegentle has joined #openstack-swift18:13
*** acoles is now known as acoles_away18:17
*** chandankumar has quit IRC18:22
*** annegentle has quit IRC18:26
portantezaitcev, notmyname: one advantage to the middleware side is that if you pull the middleware, you pull the constraints18:26
*** kevinc_ has quit IRC18:27
notmynameportante: or if you deprecate the policy, same. but this time all of the policy config is in one place, not two18:27
portanteby adding them to swift.conf, you have the potential to cause a screw up of swift.conf for other things18:27
notmynameusing middleware and they can screw up the proxy server config file ;-)18:28
portantehmmm, I can see that, too.18:29
notmynameportante: I'm mostly just trying to push and giving you a hard time. you're certainly not wrong. and middleware is certainly one way to solve this problem. personally, I think doing it all in swift.conf is better. I just haven't been convinced on the middleware side yet :-)18:30
portanteso doing it in swift.conf means more mechanism to implement which is not there, were with middleware the mechanism is already in place18:31
portantebut the down side seems to be the two locations for policy information18:31
notmynamenow that is true. it would probably take a little more code refactoring to ensure that the constraints are checked per policy. probably would need something like passing the req to the constraints or even setting up a callback in the proxy server18:32
notmynamebut it seems that the current patch does do some of that anyway (set up the callback). granted, it's probably less code this way.18:32
portantethe second problem is that for ALL requests, we have to add this check for which set of constraints to use, but with middleware we only pay for that on the ones that need it18:33
notmynamenot really. the middleware is still evalauted for every request. it just wouldn't set the callback for some of them18:33
portanteyes, if you enable that policy18:33
portanteif you don't use that policy, you don't pay18:33
portanteso somebody who does not use swift-on-file does not need to pay for the check overhead18:34
portantethe concern in a slow creep of drag on all requests18:34
notmynameI'm imagining something very early in the proxy server where it might set a swift.constraints callback (or something) based on the info in the request18:34
portantebut maybe that should be solved later18:34
portantepossibly dove tailed with other checks at the time, so perhaps18:35
notmynamewell, my thought is that there would be one mechanism for every constraints check. that is, don't use the "normal" module and maybe a middleware18:35
notmynamejust use one or the other (or some third way)18:35
notmynameless options, less code paths, less complexity (but maybe more refactoring initially)18:35
notmynameanyway, I'm just commenting on the code. tdasilva is actually writing it :-)18:36
portantemight be worth the unfront effort for the reduced complexity18:36
notmynameand my calendar just told me I have a lunch meeting in 10 minutes18:36
notmynameso I gotta run18:36
notmynamebut thanks very much for the discussion :-)18:36
notmynamehelps me figure out what I want too :-)18:36
zaitcevuseful to articulate18:36
* portante plows ahead with keystone v3 ...18:40
*** occupwnt is now known as occupant18:43
* tdasilva needs to learn to set tdasilva away :)18:50
tdasilvacatching up now18:51
*** NM has joined #openstack-swift18:51
*** jyoti-ranjan has joined #openstack-swift18:55
*** acoles_away is now known as acoles18:58
*** jyoti-ranjan has quit IRC19:00
acolesportante: i'm packing cases but will keep an eye on channel so ping if you have question19:01
portanteacoles: thanks, almost there...19:04
portantejust verified my SAIO is properly functional with master, now to change the config to use V3 auth19:05
portanteacoles: no module named keystonemiddleware.auth_token19:15
portanteshouldn't that be keystoneclient.middleware.auth_token?19:16
openstackgerritDavid Goetz proposed a change to openstack/swift: Sleep for longer at a time in lock_path.
acolesportante: recent change, auth_token got moved to its own project keystonemiddleware. keystoneclient.middleware.auth_token should still work19:17
portanteoh, already I am passe19:17
portantejust put me out to pasture19:17
acoleswhichever, you will need it installed on your SAIO19:18
acolesno. its crazy. i think keystonemiddleware imports from keystoneclient!!19:19
portanteokay, that worked, at least the proxy server is now started19:20
peluseportante:  yeah, I went though that too19:20
*** kenhui has quit IRC19:21
openstackgerritOpenStack Proposal Bot proposed a change to openstack/python-swiftclient: Updated from global requirements
portanteacoles: okay, so I have your patches installed on my client node19:26
portantenow to see if I can't get the client to work ...19:28
acolesportante: remember that if using v3 credentials to put -V 3 on swift command line and the auth-url ends /v319:30
*** miqui has quit IRC19:33
portanteacoles: thanks!19:38
*** kevinc_ has joined #openstack-swift19:38
*** miqui has joined #openstack-swift19:40
*** ryao is now known as gentoofan19:42
*** gentoofan is now known as ryao19:43
tdasilvanotmyname, zaitcev, portante, I posted some comments there on the patch. Thanks for looking at this...19:54
zaitcevtdasilva: any time, but I haven't looked properly yet, sorry. Fixing up Kodiak's crash right now.19:54
tdasilvazaitcev, no problem :)19:57
*** aerwin has quit IRC19:59
openstackgerritPete Zaitcev proposed a change to openstack/python-swiftclient: Fix crash when downloading a pseudo-directory
portanteacoles: using your script I get a 409 from keystone20:17
*** peterbusque has joined #openstack-swift20:17
acolesportante: do you get message with it? could just be conflict trying to create admin role because it already exists i.e. not a problem20:19
acolesportante: openstack [user|project|domain] list will show if user u1, project p1 were created in domain d120:21
*** peterbusque has quit IRC20:21
portanteacoles: checking ...20:28
portanteacoles: all three exist, yeah!20:29
portantedo I win a gum ball?20:30
acolesportante: one gum ball each20:31
* portante goes to get some20:31
portanteokay now, can I configure the swiftclient functional tests to use this setup?20:31
pelusetdasilva:  Wrt the middleware patch, why did you go that route? I asked myself that while reviewing and the answer I came up with was becuase it allowed per policy contraint enforcement via new code w/o affecting the core code base.20:32
pelusetdsilva: Given something like a different back-end that can be confiured w/o code changes I figured that kind of extensibility would be nice.  I dunno, off base?20:32
acolesportante: they pull in from test.conf but see tests/sample.conf20:33
acolesportante: i tend to run them on my saio so have an /etc/swift/test.conf sat there already20:34
peluseacoles:  wtf,I didn't get any gumballs??  :)20:34
tdasilvapeluse: the main driver is that for swiftonfile it was not sufficient to just have different values for the existing constraints. For example, have a different size of max_object_name_length20:35
acolespeluse: what can i say? i'll bring the to boston20:35
notmynametdasilva: I have thoughts on that. want me to respond here or in gerrit (..just got back from lunch)20:36
tdasilvahumm...would gerrit be ok? i need to step away again...driving to boston now to see mark seger's talk at the openstack-boston meetup20:37
notmynametdasilva: sure :-)20:37
portanteacoles, I can't get the example to list the container I think I created20:38
acolesportante: which example are you trying?20:39
portantebottom of this page20:39
portantewell, it was the bottom, just before the SAIO setup20:42
acolesok, with you - the example using openstackclient to list the container or using swift?20:44
acolesswift worked?20:44
acolesie example before that last one20:45
portanteah, I see, it is not using the regions I setup20:45
portanteso I have to specify the region on the swift client command line20:45
portanteI think I left that off20:45
acolesyes. --os-region <wotever>20:46
portanteokay, sec20:46
portantehow do admins use this?20:46
portanteI cannot get over how complicated and involved all this is20:46
portantewhat have we engineered?20:46
acoleslots of gumballs20:46
*** tsg has quit IRC20:47
notmynamey'all are working on keystone integration? is that the "complicated and involved" thing?20:47
acolesthose examples were for swift running in devstack, so only one region. right at the bottom of that wiki is example with --os-region to select the endpoint for your saio20:47
acolesnotmyname: keystone disintegration20:48
notmynameacoles: :-)20:48
notmynameacoles: thanks for working late at night on it20:49
*** LarsN has left #openstack-swift20:49
acolesnotmyname: k. i have 7 hours before i leave the internet behind for a while :)20:49
portanteacoles, I'll suffer through the rest of this to see if I can verify20:50
notmynameacoles: wow. that's a rough vacation. leaving work I get. leaving the internet. that's a serious commitment in today's world ;-)20:51
acolesportante: so what you have when swift authenticates with keystone it is now returning a token plus catalog with two swift endpoints - one in the devstack Region and one in the region you setup for your saio20:51
* notmyname will stop distracting now20:52
acolesportante: swift will choose the first one it parses unless you pass in --os-region (thats not new behavior)20:52
acolesportante: we#re just using the region here to integrate in your saio, nothing to do with v 320:54
acolesnotmyname: i exaggerate ;)20:55
portanteokay, I am going to try that now ...20:56
portanteokay, acoles, got the python swift client to use the right region and have the newly created container show up21:02
portantebut how do I get the openstack client to do the same, or do we not care and declare victory here?21:02
* acoles secretly doesn't care21:03
acolesportante: thats a great question !) let me take a look...21:03
* portante won't tell anybody21:03
*** Midnightmyth has quit IRC21:04
acolesbtw i'd say 'more gumballs' but peluse will get jealous21:04
portantehe better keep up!21:04
portanteokay, can't run the python-swiftclient functional tests, says "ImportError: No module named testrepository", end of the day, though another problem to solve tomorrow.21:09
portanteacoles: thanks for your help and guidance21:09
acolesportante: hmm, not seen that one. ok, thanks for sticking with it. If the patch needs any straightforward changes please feel free to push over in my absence and i'm sure peluse & cschwede_ would review again21:12
*** __lgw4__ has quit IRC21:13
*** tsg has joined #openstack-swift21:15
acolesportante: might just need a 'pip install -r test-requirements.txt'21:17
acolescschwede_ might be able to help re swiftclient functional tests^^ ?21:17
*** mkollaro1 has quit IRC21:19
peluseacoles:  so stupid question (maybe), I don't need to worry about the fact that I only used the default region, regionOne, in my testing right?21:23
*** tongli has joined #openstack-swift21:23
*** mwstorer has joined #openstack-swift21:24
*** tong_ has quit IRC21:26
acolespeluse: nope. portante wanted to test against his own saio, thats why we did the region thing.21:26
pelusephew :)21:30
acolespeluse: but you only get gumballs for using regions ;)21:31
acolesno, seriously, the region stuff has nothing to do with the v3 stuff.21:31
peluseyeah, that's what I thought so wanted to make sure I wasn't totally missing the forrest for the tress21:32
pelusehey, I read that 'reverify no bug' isn't supported anymore and that w/reverify we have to supply a bug # - is that the case? Anyone know?21:34
*** annegentle has joined #openstack-swift21:35
acolespeluse: i have a vague memory i read that21:35
peluseacoles:  yeah, its on and I ask because tsg has a patch that just keeps getting kicked back and we're sorta at the point of trying a comment like "yeah, recheck or reverify - whatever you need to do just do it"21:37
notmynamepeluse: yes, that's the case. however, since all things go in to the check queue first, it's possible to "recheck no bug" and it will actually still make it all the way through the gate queue21:39
*** annegentle has quit IRC21:39
pelusetorgomatic:  might be that I've had too much cold medicine today but man your PUT patch is making my head spin :)  Will try to step through a few of the parts that aren't adding up for me but may have to pick your brain a bit...21:40
pelusenotmyname:  cool, thanks.  that's what I've been using when I know for sure its not something in the patch (well, as for sure as can be)21:40
tsgnotmyname: any ideas why the TestProfileMiddleware keep failing though?21:40
torgomaticpeluse: don't blame the medicine, it's entirely possible I goofed it up hard :)21:40
tsgnotmyname: I haven't been able to reproduce it locally at all21:40
pelusetsg:  ditto, runs fine for me locally too and its has passed a few times on Jenkins, seems like the pass rate is like 1 in 10 though21:41
notmynametsg: which patch?21:41
tsgthe EC policy patch21:41
tsgnotmyname: it has passed the check a few times .. but then it fails in verify21:42
peluseis there an echo in here?21:42
notmynameah, interesting21:42
tsgpeluse, "echo"? :)21:43
notmynameok, so somethign on the swift unit tests is somethign that needs to be investigated on our end and fixed. there's a very low false negative rate there (granted, this still could be a false negative)21:43
peluseyeah, looks like it just passed the check21:44
tsgnotmyname: I ran the profile test about 100 times in a row today, passes every single time21:44
tsgnotmyname: looking for ideas to try21:44
pelusetsg:  if it fails verify and you don't have time to dig in (instrument, fully unserstand the failing test) let me know and I'll look at it too21:44
notmynametsg: hmm...jsut for fun, run the xprofile middleware on your saio and then run `swift-bench --saio` against it. a while back I saw that fail (but didn't have time to look in to it). but maybe there is a system load issue or something else21:45
notmynamewhere "I saw that fail" == swift started spewing tracebacks in the logs for every request21:46
tsgnotmyname: system load is what I was thinking as well21:46
*** openstackgerrit has quit IRC21:46
tsgnotmyname: does the py27 gate job run in ubuntu 12.04 env?21:47
notmynametsg: I think they upgraded them to trusty21:47
tsgok .. let me set up a Trusty SAIO just to be sure21:47
*** openstackgerrit has joined #openstack-swift21:48
pelusetsg:  I've got one ready to go, I'll fire it up too and let you know21:49
*** acoles is now known as acoles_away21:49
tsgcool, thanks peluse21:49
pelusetsg:  oops, I lied its not ready to go.  Not as "trusty" as I thought :)  Will still give a whirl too though so we have 2 data points21:51
tsg:) ok21:51
*** annegentle has joined #openstack-swift21:54
*** jergerber has joined #openstack-swift21:57
notmynameFYI DefCore is recommending to the BoD that Swift isn't a designated section of OpenStack. that is, swift's code does not matter when determining trademark protection for the openstack brand22:01
notmynameIOW, the proposal means that you can use any codebase or commercial product you want, as long as it claims "swift API support" and that's perfectly ok for OpenStack trademark usage22:02
*** annegentle has quit IRC22:10
*** NM has quit IRC22:17
*** kevinc_ has quit IRC22:18
*** kevinc_ has joined #openstack-swift22:20
*** kevinc_ has quit IRC22:21
*** kevinc_ has joined #openstack-swift22:21
notmynamemattoliverau: hi22:21
*** dmsimard is now known as dmsimard_away22:29
pelusetsg:  ran the unit tests for 100x on trusty SAIO, no issues here22:40
tsgpeluse: same experience here, can you also try to load the machine with swift-bench22:41
tsgpeluse: trying now myself also .. just had a long chat with yuan on the trailer patch and eventlet changes22:42
*** annegentle has joined #openstack-swift23:05
tsgpeluse: even with swift-bench, no failures over 100 runs23:06
tsgpeluse, notmyname: the ec policy patch passed both check and gate today!23:09
tsgjust when we were to get it to fail .. :)23:09
*** annegentle has quit IRC23:10
peluseyeah, I almost had it root caused over here :)23:22
pelusetsg:  you mean the common failing test passed right?  looks like its still in the gate with that one success but several others still running (that could fail for non-swift reason).  Hope I didn't just jinx it23:23
*** marcusvrn has quit IRC23:41
tsgpeluse: yeah I realized :( so far so good23:42
*** kevinc_ has quit IRC23:47

Generated by 2.14.0 by Marius Gedminas - find it at!