Wednesday, 2014-03-19

*** jergerber has quit IRC00:17
*** matsuhashi has joined #openstack-swift00:20
*** shri has left #openstack-swift00:41
*** tusharg has left #openstack-swift00:46
*** tusharg has joined #openstack-swift00:46
*** nosnos has joined #openstack-swift00:51
*** nosnos_ has joined #openstack-swift00:53
*** nosnos has quit IRC00:57
*** d89 has joined #openstack-swift01:06
*** mkollaro has quit IRC01:14
*** csd has quit IRC01:18
*** bada has quit IRC01:25
*** bada has joined #openstack-swift01:26
*** tusharg is now known as tgohad02:11
*** shakamunyi has joined #openstack-swift02:28
*** chandan_kumar has joined #openstack-swift02:37
*** gyee has quit IRC02:59
*** matsuhashi has quit IRC03:13
*** zanc_ has quit IRC03:16
*** changbl has joined #openstack-swift03:17
*** matsuhashi has joined #openstack-swift03:20
*** matsuhashi has quit IRC03:25
*** nosnos_ has quit IRC03:41
*** chandan_kumar has quit IRC03:41
openstackgerritClay Gerrard proposed a change to openstack/swift: Add containeralias middleware
claygwhy do i have to re-brew install everything after that stupid mac ssl upgrade thing04:18
swifterdarrellclayg: gah! wat wat?04:18
swifterdarrellclayg: i think i'm all ssl'ed up but didn't have to re-brew-install anything?04:19
h6wSounds interesting...04:20
*** matsuhashi has joined #openstack-swift04:23
*** h6w has quit IRC04:33
*** nosnos has joined #openstack-swift04:35
*** h6w has joined #openstack-swift04:38
madhuri_clayg: there?05:03
*** nshaikh has joined #openstack-swift05:24
*** zaitcev has quit IRC05:30
*** psharma has joined #openstack-swift05:58
*** saju_m has joined #openstack-swift05:58
*** matsuhashi has quit IRC05:59
*** matsuhashi has joined #openstack-swift06:01
*** h6w has left #openstack-swift06:05
*** matsuhashi has quit IRC06:47
*** matsuhashi has joined #openstack-swift06:50
*** bada has quit IRC07:25
*** bada has joined #openstack-swift07:26
*** matsuhashi has quit IRC07:30
*** matsuhashi has joined #openstack-swift07:31
*** shakamunyi has quit IRC08:07
openstackgerritChristian Schwede proposed a change to openstack/swift: Add containeralias middleware
*** jokke_ has joined #openstack-swift08:33
*** shakamunyi has joined #openstack-swift08:33
*** mlipchuk has joined #openstack-swift08:37
*** chandankumar_ has quit IRC08:37
*** shakamunyi has quit IRC08:38
openstackgerritChmouel Boudjnah proposed a change to openstack/python-swiftclient: Fix temporary pypy gate issue with setuptools
*** chandan_kumar has joined #openstack-swift08:39
*** nshaikh has quit IRC08:41
*** nacim has joined #openstack-swift08:50
*** nacim has quit IRC08:54
*** nacim has joined #openstack-swift08:55
*** tanee-away has quit IRC09:17
*** saschpe_ has joined #openstack-swift09:19
*** mmcardle has joined #openstack-swift09:22
*** sileht has quit IRC09:33
*** shakamunyi has joined #openstack-swift09:34
*** shakamunyi has quit IRC09:38
*** tanee has joined #openstack-swift09:41
*** akscram has quit IRC09:43
*** tanee is now known as tanee-away09:46
*** d89 has quit IRC09:50
*** sileht has joined #openstack-swift09:50
*** mmcardle has quit IRC09:59
*** mmcardle has joined #openstack-swift10:02
*** tanee-away has quit IRC10:03
*** jamieh has joined #openstack-swift10:05
*** jamieh is now known as Guest7278610:05
*** d89 has joined #openstack-swift10:06
*** mkollaro has joined #openstack-swift10:19
*** tanee has joined #openstack-swift10:20
openstackgerritChristian Schwede proposed a change to openstack/python-swiftclient: Make bin/swift testable part 1
openstackgerritChristian Schwede proposed a change to openstack/python-swiftclient: Fix temporary pypy gate issue with setuptools
*** shakamunyi has joined #openstack-swift10:35
*** shakamunyi has quit IRC10:39
openstackgerritA change was merged to openstack/swift: remove debugging print
*** d89 has quit IRC10:52
*** bada_ has joined #openstack-swift10:58
*** bada has quit IRC11:00
*** dvas has joined #openstack-swift11:03
*** tdasilva has left #openstack-swift11:04
*** otoolee has joined #openstack-swift11:07
*** d89 has joined #openstack-swift11:09
*** dvas has quit IRC11:10
*** dvas has joined #openstack-swift11:11
*** mkollaro has quit IRC11:16
*** matsuhashi has quit IRC11:22
*** matsuhashi has joined #openstack-swift11:24
*** mmcardle has quit IRC11:26
*** shakamunyi has joined #openstack-swift11:36
*** matsuhashi has quit IRC11:37
*** akscram has joined #openstack-swift11:39
*** shakamunyi has quit IRC11:40
*** Trixboxer has joined #openstack-swift11:56
*** mmcardle has joined #openstack-swift12:00
*** mmcardle has joined #openstack-swift12:03
*** chandan_kumar has quit IRC12:23
*** piyush1 has joined #openstack-swift12:28
*** d89 has quit IRC12:32
*** shakamunyi has joined #openstack-swift12:36
*** shakamunyi has quit IRC12:41
*** mkollaro has joined #openstack-swift12:48
*** JuanManuelOlle has joined #openstack-swift12:53
*** chandan_kumar has joined #openstack-swift12:56
*** psharma has quit IRC13:03
*** mmcardle has quit IRC13:11
*** tdasilva has joined #openstack-swift13:19
*** nosnos has quit IRC13:24
*** piyush1 has quit IRC13:32
*** shakamunyi has joined #openstack-swift13:37
*** shakamunyi has quit IRC13:41
*** Guest72786 is now known as jamie_h13:44
*** saschpe__ has joined #openstack-swift13:49
*** bsdkurt has quit IRC13:52
*** bsdkurt has joined #openstack-swift13:54
openstackgerritA change was merged to openstack/swift: Remove some debugging prints from tests
*** byeager has joined #openstack-swift14:03
*** mmcardle has joined #openstack-swift14:04
*** saschpe__ has quit IRC14:06
*** saschpe__ has joined #openstack-swift14:06
*** saschpe__ has quit IRC14:07
*** saschpe_ has quit IRC14:07
*** dmsimard has joined #openstack-swift14:12
*** chandan_kumar has quit IRC14:13
*** judd7 has joined #openstack-swift14:15
*** nshaikh has joined #openstack-swift14:16
*** chandan_kumar has joined #openstack-swift14:17
*** piyush1 has joined #openstack-swift14:17
*** shakamunyi has joined #openstack-swift14:18
*** saschpe- has joined #openstack-swift14:36
*** pconstantine_ has quit IRC14:39
*** pconstantine_ has joined #openstack-swift14:40
*** saschpe- has quit IRC14:42
*** saschpe- has joined #openstack-swift14:42
*** saschpe- is now known as saschpe_14:43
*** saschpe_ is now known as saschpe__14:43
*** shakamunyi has quit IRC14:52
*** shakamunyi has joined #openstack-swift14:52
*** byeager has quit IRC15:01
*** byeager has joined #openstack-swift15:01
*** piyush1 has quit IRC15:06
*** dvas has quit IRC15:14
*** nshaikh has quit IRC15:18
*** saschpe__ is now known as saschpe15:24
*** dvas has joined #openstack-swift15:32
*** lpabon has joined #openstack-swift15:37
portantenotmyname: have you heard of these guys:
openstackgerritChristian Schwede proposed a change to openstack/swift: Add containeralias middleware
*** zaitcev has joined #openstack-swift15:42
*** ChanServ sets mode: +v zaitcev15:42
zaitcevportante: Not sure if you realize that I merged my tip that has PBE with your in-process functional tests, and ran the result. It looked fine, bring 66108 out of WIP, I'll retest and +2.15:44
*** dvas has quit IRC15:46
cschwedeportante: i was testing that a while ago - it's blazingly fast. it works a little bit different than other object storages, for example you can't set the object name, but get an uuid for each uploaded object. and no object listing15:46
*** dvas has joined #openstack-swift15:47
portantecschwede: you are referring to the container alias?15:47
cschwedeportante: sorry, no, to the DDN WOS15:48
portanteah, thanks15:48
portantecschwede: yes, I worked on a system like that for tabblo.com15:48
creihtcschwede: heh, swift would be super fast if we didn't have to do object listings either :)15:49
cschwedeportante: page 9 on the whitepaper lists some functionalities:
cschwedecreiht: haha, true :)15:49
creihtfacebook's haystack was kinda like that15:51
cschwedecreiht: but they are also extremely fast for getting / storing data, for example if you put a 5GB blob onto it there should be no significant  delay due to DB updates15:51
*** dvas has quit IRC15:51
otooleecreiht, Sorry to pester, but could you please have a quick look at and let me know if I'm on the right track?15:51
creihtotoolee: sorry also been sick15:51
*** dvas has joined #openstack-swift15:52
cschwedecreiht: they don't use a filesystem on their disks, and it's more like a black box15:52
otooleecreiht, Oh, sorry to hear that.15:52
creihtI'm trying to finish up a patch now, then will get back to reviews15:52
creihtno worries15:52
creihtcschwede: yeah, I've been thinking about a non-file system backend for swift15:52
creihtthe problem is that a filesystem is terribly convenient15:52
*** chandankumar_ has joined #openstack-swift15:54
*** chandan_kumar has quit IRC15:56
notmynameportante: ya, I've come across them from time to time16:01
*** dvas has quit IRC16:01
creihtalso just a thought16:03
creihtif people are interested in talking about stuff at the summit, it would probably be a good idea to get things in sooner rather than later16:04
creihtto make sure we get a decent number of slots16:04
notmynamecreiht: yes, exactly!16:05
creihtI don't want to be the only one suggesting topics for the summit16:05
creihtcorrection: you probably don't want me being the only one suggestion topics for the summit ;)16:06
*** judd7 has quit IRC16:09
*** jamie_h has quit IRC16:11
*** saschpe has quit IRC16:11
*** saschpe has joined #openstack-swift16:12
*** jamieh has joined #openstack-swift16:13
*** jamieh is now known as jamie_h16:14
*** saschpe has quit IRC16:14
*** saschpe has joined #openstack-swift16:15
*** dvas has joined #openstack-swift16:15
*** bada_ has quit IRC16:21
*** bada_ has joined #openstack-swift16:21
notmynamegood morning everyone! (just got into the office and marked all the emails as read)16:25
notmynamereminder that we have a swift team meeting in #openstack-meeting in 2.5 hours16:26
*** mkollaro has quit IRC16:26
torgomaticcat /dev/null > /var/spool/mail/$USER.mbox  # Inbox Zero!16:27
torgomaticnotmyname: something like that? ;)16:27
* hugokuo laughing 16:29
*** byeager has quit IRC16:32
*** byeager has joined #openstack-swift16:32
*** byeager has quit IRC16:36
*** byeager has joined #openstack-swift16:38
notmynamecreiht: portante: if you want swift to be faster on writes and not do DB updates, may be something to check out :-)16:39
*** shakamunyi has quit IRC16:40
*** gyee has joined #openstack-swift16:43
portantenotmyname: hmm, two unanswered questions ... ;)16:49
notmynameportante: well, I've looked at it about as much as anyone else recently (ie not at all ;-))16:49
notmynameIIRC it was some random idea combined with a few minutes of downtime ;-)16:50
*** mmcardle has quit IRC16:52
*** gvernik has joined #openstack-swift16:52
*** d89 has joined #openstack-swift16:54
portantecreiht, notmyname, zaitcev: add two, PBEs and DiskFile16:56
notmynameportante: ?16:56
portantedesign summit topics?16:57
notmynameah, cool16:57
portanteotoolee: perhaps it would be worth proposing a summit design session on object auditors16:57
notmynameportante: I'd like to see some discussion around making FS traversal better. we've got several processes that do it (at different speeds), and os.walk() isn't too efficient17:01
*** chandan_kumar has joined #openstack-swift17:05
*** d89 has quit IRC17:09
creihtnotmyname: hrm, not sure I like that idea17:11
*** shri has joined #openstack-swift17:11
notmynamecreiht: you don't like making FS traversal better? ;-)17:14
creihtthe patch you pointed me at17:15
notmynameor you don't want just one walker (I'm not sure I want that either)17:15
notmynamecreiht: ah ok :-)17:15
creihtyou move to a definate consistency window17:15
notmynamecreiht: ya, I'm not sure that's a good idea either ("here's a loaded gun pointed at your foot. have fun")17:15
notmynamecreiht: but in some cases, it's interesting to consider17:15
creihtyou also move what was an error case to a normal case17:15
creihtwhere as before, it was easy to track, and if your pendings start shooting up, you know you have some sort of issue17:16
notmynameoh, I completely get that :-)17:16
creihtI don't think it would be right for our use case17:16
notmynameeg I've seen several people interested in swift but completely don't need listings or container metadata or rolled up stats.17:16
notmynamecreiht: I agree it would be horrible for cloud files17:17
*** chandan_kumar has quit IRC17:17
creihtbut I can see in the case where you don't really care about listings, it might be alright17:17
creihtbut in the case if you really don't care about listings, you could drop container services all-together17:18
*** mlipchuk has quit IRC17:18
notmynamenot without a lot more changes, since that would affect the proxies and auth a lot17:19
creihtoh sure17:19
creihtI struggle with, how many use cases should swift cover?17:20
creihtor what should be consider out-of-bounds17:21
* torgomatic is working on RFC2324 support for Swift17:28
*** JuanManuelOlle has quit IRC17:30
*** JuanManuelOlle1 has joined #openstack-swift17:30
luisbgtorgomatic, "Coffee brewing is an art, but the distributed intelligence of the web-connected world transcends art.  Thus, there is a strong, dark, rich requirement for a protocol designed espressoly for the brewing of coffee."17:43
torgomaticall I'm saying is I want a coffee replicator17:47
luisbg+1, or even +217:48
*** bada_ has quit IRC17:51
*** gvernik has quit IRC17:51
*** [1]gvernik has joined #openstack-swift17:51
swifterdarrelltorgomatic: luisbg: new OpenStack project, STAT!17:52
*** bada_ has joined #openstack-swift17:52
*** mmcardle has joined #openstack-swift17:52
luisbgswifterdarrell, url?17:52
swifterdarrellluisbg: i'm just sayin' we need one for the coffee thing17:53
luisbgoooh START :)17:53
*** tanee is now known as tanee-away17:53
werwhat's all this coffee talk?  I'm in!17:53
portantenotmyname: yes, I have a patch I am in the process of testing which I think is going to significantly help the directory traversal17:54
portantetrying to get some numbers generated using a big workload to make sure it is good17:55
portantewe used the same technique at another customer to reduce head seeks on the disk17:55
portantebasically, if you sort the inodes for the lowest level hash directories, and access those directories in ascending inode number order, you'll access the files in the order they are written on disk, avoiding the head seeks.17:59
*** nacim has quit IRC18:01
torgomaticportante: is that an XFS-ism?18:01
torgomaticI mean, I guess it doesn't matter too much, as it'll be no worse on other filesystems18:01
portanteI don't think so18:02
portantebut maybe18:02
portanteit worked well for xfs18:02
portanteI'll post the patch as work-in-progress18:03
portanteuntil we finish our numbers gathering18:03
torgomaticsounds good; I'm always up for perf gains, especially when they don't involve huge code restructurings :)18:04
*** mmcardle has quit IRC18:06
creihtoh man18:08
openstackgerritPeter Portante proposed a change to openstack/swift: WIP: Speed up obj aud. sorting hsh dirs by inode #
creihtwe have very little unit testing in our proxy code :/18:09
portantetorgomatic: ^^^18:09
portantecreiht: yes18:09
* creiht rolls up his sleeves18:11
portantego, I'm right behind ya18:11
creihtwell I'm not guaranteeing 100% coverage :)18:12
creihtalso I'm just looking at the object controller right now for this patch18:12
portantebut we get coverage numbers by controller, right?18:12
creihtright now it is 88% for object controller18:13
creihtand most of that is incidental18:13
creihtmean it gets coverage just because it happens to run the code, not because it is actully tested :)18:13
portanteyes, it seems that if you only run the tests that target a given module we have very low coverage18:13
*** byeager has quit IRC18:17
*** byeager has joined #openstack-swift18:17
*** byeager_ has joined #openstack-swift18:20
*** mmcardle has joined #openstack-swift18:21
*** bada_ has quit IRC18:21
*** bada_ has joined #openstack-swift18:21
*** byeager has quit IRC18:22
*** mmcardle has quit IRC18:25
*** dvas has quit IRC18:31
portantenotmyname: are you going to suggest a session on disk walking?18:34
notmynameportante: hmm..should I? I don't have anything to say on it right now other than "let's make it better"18:35
portantemaybe we can come up with a program that shakes an entire rack off its moarings using a GET pattern on swift ...18:35
portantenotmyname: sounds like a good basis for a design summit session18:35
notmynameportante: done18:38
notmynamemeeting in 15 minutes18:45
*** jamie_h has quit IRC18:49
*** gvernik has joined #openstack-swift18:52
*** [1]gvernik has quit IRC18:52
*** jamieh has joined #openstack-swift18:56
*** jamieh is now known as Guest71918:57
*** bada_ has quit IRC18:57
*** bada_ has joined #openstack-swift18:58
*** Guest719 has quit IRC18:58
notmynamemeeting time in #openstack-meeting18:59
*** saju_m has quit IRC19:21
*** saju_m has joined #openstack-swift19:22
*** bada_ has quit IRC19:25
*** bada_ has joined #openstack-swift19:26
*** tgohad has left #openstack-swift19:37
*** tgohad has joined #openstack-swift19:37
*** tgohad is now known as tusharsg19:37
*** keving1 has joined #openstack-swift19:44
*** dvas has joined #openstack-swift19:46
*** jairo has joined #openstack-swift19:46
*** tusharsg has left #openstack-swift19:46
*** tusharsg has joined #openstack-swift19:46
*** tusharsg has left #openstack-swift19:46
*** tusharsg has joined #openstack-swift19:47
jairohey guys I was wondering if you guys could guide me on how to run this script    I just try to run it and it just stays there doing nothing, thx19:48
gholtjairo: Hi, it could be that it hasn't found anything to do yet.19:49
gholtI assume you're running that with the DELETE line commented out still?19:50
*** gvernik has quit IRC19:51
gholtIn case I'm not here when you respond: I would add some extra print statements here and there so you can check on just what the script is doing.19:51
*** gvernik has joined #openstack-swift19:51
jairoyeah still commented out19:52
*** tusharsg_ has joined #openstack-swift19:52
notmynamethanks for coming to the meeting. /me is going to get lunch now19:59
creihtif someone would like to take a look :)20:00
glangethat has html in it20:01
portantecreiht: pulling ...20:02
portantegonna see how it fares on Fedora 19 SAIO20:02
creihtportante: thanks20:02
zaitcevportante: what did you want to discuss about PBE? you mentioned doing something after meeting20:02
peluseportante:  just looking over the beginning of the meeting minutes (I missed the start)...20:03
*** tusharsg_ has quit IRC20:03
peluseportante:  unit test coverage on SP is solid20:03
peluseportante:  however functional and probe testing is lacking20:03
portantepeluse: getting coverage there will help20:03
zaitcevportante: how about I'll prepare the old and new flows like you did in wide text files and then we'll look at those? In 25 mintes I'll need to start cooking here20:03
*** gvernik has quit IRC20:03
portantebut as clayg said, the code paths might be easily covered by a config change to the default policy20:03
portantezaitcev: sure20:04
portantesounds like a plan20:04
cschwedecreiht: will have a look at that patch tomorrow morning20:04
portanteI have a hard stop at 5:30 PM EDT20:04
peluseportante:  yup, totally agree.  brought that up a while back and we have a TODO item on trellow that still, well, TODO :)20:04
portantecreiht: would like to see the 62% coverage of xprofile a bit higher. :)20:05
creihtit isn't my code, I would just like to see the functionality :)20:06
portanteI thought since you were jumping in on the proxy server unit tests you could just get the remaining 38% for that patch as well20:06
portantea unit test is a unit test, right?20:07
*** saju_m has quit IRC20:08
portantefolks, get object versioning not enabled in my functional tests runs20:08
portanteis that normal?20:08
portantedo I have a bad config?20:08
portantein my SAIO?20:08
*** saju_m has joined #openstack-swift20:08
*** tongli has joined #openstack-swift20:09
torgomaticportante: you mean the functional tests pass with object versioning turned off?20:10
portantetest_overwriting is SKIPPED20:10
portante and UTF8 version20:10
torgomaticseems reasonable; if you haven't turned on object versioning, the tests are skipped20:11
*** Trixboxer has quit IRC20:11
portantehmm, thought I had that ...20:12
torgomaticthat guy's weird; it's in the container-server config20:12
portanteshould we have that on by default in a SAIO?20:12
torgomaticprobably, IMO; if you're doing development, all the stuff should be on so your functional tests test everything20:13
torgomatic(at least by default)20:13
portantetorgomatic: I would think so as well20:13
zaitcevmakes sense20:13
zaitcevexcept perhaps it's okay to skip when packages are missing20:13
bsdkurthi. checking in to see if there's any comment on that direct io patch. :-)20:14
creihtbsdkurt: I want to look at it, but haven't had a chance yet20:14
portantebsdkurt: same here20:14
bsdkurtok. thanks.20:15
bsdkurtI know it isn't commit worthy as it stands, but it should serve as a good starting point for discussion20:15
portantebsdkurt: did you mark it work-in-progress?20:16
portanteand do you have some numbers comparing some swift API benchmarks?20:16
creihtbsdkurt: absolutely20:16
bsdkurthow do you mark it WIP?20:16
portantetorgomatic: nice rendition of sinatra20:16
torgomaticportante: oh, the branch name in my Swift fork? :)20:17
jairogholt: even if it doesn't find anything, I should see activity on the syslog, going through it, or shouldn't I?20:17
notmynamebsdkurt: there's a button in gerrit
bsdkurtnotmyname: ahh. how did I miss that? thanks!20:19
* portante has missed many a tricky button press20:19
openstackgerritPeter Portante proposed a change to openstack/swift: Enable object servers for SAIO by default
creihtgerrit doesn't exactly help you in that respect20:20
openstackgerritPeter Portante proposed a change to openstack/swift: Enable object versions for SAIO by default
bsdkurtportante: the testing I did was all somewhat custom and focused on retrieving large objects. is there a standard way to do perf tests for changes like this?20:23
portantessbench, swiftbench, which others are more versed in20:24
notmynameportante: for anticw would be a good person to get comments from20:24
portanteswifterdarrell might be able to offer you some guidance20:24
notmynameportante: but, he's also a very busy guy :-)20:24
portantek thanks20:25
bsdkurtportante: okay. I'll take a look at ssbench and swiftbench20:25
jeblair"swift --debug list" against rackspace dfw region is hanging for me20:31
jeblairit's stuck at "INFO:requests.packages.urllib3.connectionpool:Starting new HTTPS connection (1):"20:31
jeblairord works20:31
jeblairpandemicsyn, ahale, notmyname: ^20:32
*** dvas has quit IRC20:32
jeblairi'm using git master, with latest python-keystoneclient release and the OS_ env variables20:33
jeblairoh, interesting...20:33
jeblairi am unable to upload files in the web ui to dfw, but ord works there20:34
jeblairgholt, creiht: ^20:35
*** shri has quit IRC20:37
ahalethat sounds messed up jeblair , have you got your swift account handy i can look and see whats going on, or if i need to get someone in our identity to look ?20:38
jeblairahale: the rs cloud account username is 'openstackci'; if there's a separate swift account i'm not familiar with it20:39
ahaleah cool, thats enough20:39
ahalei'll dig around, let you know when i get somewhere20:41
jeblairahale: cool.  i'm not blocked since ord is working; let me know if you need anything else20:41
ahaleok cool20:41
*** shri has joined #openstack-swift20:44
*** lpabon has quit IRC20:52
ahalejeblair: thats weird, it all looks good and im showing 2xx codes, both direct and through the control panel.. i'm going to check with the identity people see if they have anything going on20:54
ahaleare you doing this from ord? i see there its hitting us on ipv6, i wonder if theres something weird with ip6 to dfw.. but then not if control panel is doing the same I guess20:55
claygwhat channel am I in?20:57
*** dvas has joined #openstack-swift20:58
ahaleif i wasn't so terrified of reviewers, and better at writing code theres an idea I had… but alas, this is the only way I can help openstack at the moment sorry21:00
*** JuanManuelOlle1 has quit IRC21:00
claygahale: sorry - my default mode is "not helpful" - I think most people just ignore me and then try throwing things21:02
*** gvernik has joined #openstack-swift21:05
werhrm, I am getting things truncated when curl PUT'ing things into swift :/  Even though I supply the Content-Type the Created response tells me it is Content-Type: text/html; charset=UTF-8... then when I go head the request again it has my supplied content type  Content-Type: image/jpeg  but the file got truncated.21:05
werI don't know what I am going wrong as I don't remember this happening in the past.21:06
claygwer: can you duplicate the issue with the swift cli?  Otherwise it's probably time to paste your curl command.21:06
wercurl -i -H 'Content-Type: image/jpeg' -X PUT --data "@./dd1b9400-548c-44f2-ab79-3819cc52eb83.jpg" http://swifthost:8080/v1/AUTH_user/asset/00000000-0000-0000-0000-000000000010.jpg21:07
werno the swift client behaves correctly.21:07
clayg--data-binary ?21:08
claygif you run that with -v can you see the transmitted headers and verb?21:08
werclayg: I'm an idiot. --data-binary21:09
claygi feel like when I'm doing PUT's with curl I mostly just use -T21:09
werSo it escaped a null or something.21:09
claygcurl's crazy21:09
weragreed :)21:09
werty so much21:09
* clayg tries to gain back some karma for being a nob to poor ahale 21:10
ahalei feel bad for making you feel bad now clay :(21:10
*** gvernik has quit IRC21:10
claygahale: you'd be better off trying to throw something than spending any time worrying about me - i get what I deserve21:11
* clayg ducks21:11
*** mmcardle has joined #openstack-swift21:19
portantezaitcev: let's meet up tomorrow then, i'll post something on your calendar21:25
jogoBug 126290621:26
jogonotmyname: ^ is the new #2 gate failure
zaitcevportante: okay, but please not too early (I had some errands in the morning -- nothing too important, honestly, but already set..)21:26
zaitcevportante: 2 p.m. E?21:26
portante2pm it is21:26
*** bvandenh has quit IRC21:27
portantejogo: does that recheck page offer a link to that bug somewhere?21:27
jogoportante: it does click  launchpad at the bottom21:28
jogoportante: we are discussing it in #openstack-qa21:28
*** saju_m has quit IRC21:28
portantejogo: I have a 5:30 PM EDT hard stop, so I gotta run, but will be on later if you need anything21:29
*** mmcardle has quit IRC21:32
*** mmcardle has joined #openstack-swift21:33
notmynamejogo: I just got back from lunch and read the scrollback21:38
jogonotmyname: it looks like a devstack issue, we are only seeing this in hp and not rax21:38
notmynamejogo: ok, thanks for taking a look21:38
notmynamejogo: also "jogo" not "jog0" anymore?21:39
notmynameor was it j0go?21:39
jogonotmyname: haha it was jog021:39
jogobut someone else had that name for a while, but it just expireed21:39
*** krtaylor has quit IRC21:46
*** dvas has quit IRC21:49
*** tdasilva has left #openstack-swift21:51
*** mmcardle has quit IRC21:51
*** krtaylor has joined #openstack-swift21:52
*** dvas_ has joined #openstack-swift21:53
*** wer is now known as wer_mocha21:54
zaitcevI also happen to review patches in other projects and apparently "recheck no bug" is more common elsewhere. We are more dilligent about helping infrastructure people in Swift.21:58
zaitcevChange subject: Add http(s) protocol support to fetch_review and list_reviews21:58
zaitcevPatch Set 7:  recheck no bug21:58
openstackgerritChuck Thier proposed a change to openstack/swift: Add "If-None-Match: *" support to PUT
*** keving1 has left #openstack-swift21:59
creihtportante: ^- there's some extra tests22:00
creihtI just added some of the basic requests, but it is a start22:00
*** fifieldt has joined #openstack-swift22:02
openstackgerritChuck Thier proposed a change to openstack/swift: Add "If-None-Match: *" support to PUT
creihthehe.. silly print statement :)22:02
claygcreiht: is the goal there mainly to save wasted effort if the file is already there or acctually *prevent* uploading if the file already exists?  I worry that it looks like stronger consistency than it really is in the face of failures + rebalance22:03
anticwportante: does that auditor speedup work?22:03
anticwi ask because the write patterns for PUT cause badly skewed block allocation and it might not be that effective22:04
creihtclayg: well the goal is to have similar semantics to a HEAD before22:04
notmynameanticw: I think he had to step out for a bit, but he should be able to read buffer playback later22:04
creihtI don't think it is possible to absolutely *prevent* if the file is *supposed* to already exist22:05
claygwhere similar doesn't include handoffs ;)22:05
creihtit would if the cannonical locations were not available22:05
claygtrue, first three servers that are up22:06
claygit's probably fine - i'll try to poke at it - thanks!22:06
creihtyeah feel free to shoot holes in it22:06
creihtI don't want to open up some weird use cases22:06
creihtthat try to abuse something like that22:06
claygi think the best I'll be able to get is upload with 2 primaries up, take out a device and rebalance, then fail one of the primaries, and watch it still upload 50% of the time until replication kicks in22:07
creihtone of the reasons I didn't include etag support initially is that there were so many weird edge cases22:07
creihtyeah this is for a pretty specific use case22:07
claygcreiht: yeah I like that bit22:07
creihtand will help out one of our customers greatly22:08
claygoh, doh!  debug print in object controller line 59022:09
creihtclayg: reload22:09
creihtI should have gotten rid of that22:09
openstackgerritJenkins proposed a change to openstack/swift: Updated from global requirements
*** mlipchuk has joined #openstack-swift22:16
*** dmsimard has quit IRC22:29
*** byeager_ has quit IRC22:49
*** byeager has joined #openstack-swift22:49
*** byeager has quit IRC22:53
*** tusharsg has quit IRC22:58
*** shri1 has joined #openstack-swift23:11
*** shri has quit IRC23:13
*** tongli has quit IRC23:13
*** mlipchuk has quit IRC23:19
*** shri1 has quit IRC23:29
claygzaitcev: does the PBE branch fix/address the stupid/useless/broken status/status_changed_at fields on the account and container databases?  (I call it stupid/useless - it's possible the account-reaper acctually cares about those fields, but it gets on my nerves that recreated containers still say they're DELETED)23:47
zaitcevclayg: No, sorry. I didn't dare mixing it in.23:47
zaitcevclayg: However, my experiments show that it's possible to have Swift that sits on top of something that is deleted when it's deleted. E.g. container is a directory and GlusterFS just does rmdir.23:49
zaitcevclayg: This may interest you because I think it demonstrates that if you can adjust internal users such as account-reaper, you should be safe about nothing leaking up to user API.23:50
claygzaitcev: not entirely sure I follow - acctually pretty sure I don't - but that's ok ;)23:56
zaitcevclayg: Okay! So in short, no, I didn't do any cleanups of the deleted state. Only made sure it's not in the API of backends and contained fully inside legacy backend.23:57
claygah, I think I see then - yeah that's great!23:58

Generated by 2.14.0 by Marius Gedminas - find it at!