| portante | achhabra: https://wiki.openstack.org/wiki/Meetings#Swift_team_meeting | 00:01 |
|---|---|---|
| achhabra | portante: Thank you. | 00:05 |
| *** dmorita has joined #openstack-swift | 00:51 | |
| *** achhabra has quit IRC | 00:55 | |
| *** IgnacioCorderi has joined #openstack-swift | 01:10 | |
| *** IgnacioCorderi has quit IRC | 01:12 | |
| *** erlon has joined #openstack-swift | 01:47 | |
| *** nosnos has joined #openstack-swift | 01:51 | |
| *** HenryG_ is now known as HenryG | 02:01 | |
| *** echevemaster has joined #openstack-swift | 02:03 | |
| *** judd7 has joined #openstack-swift | 02:11 | |
| *** bkopilov has quit IRC | 02:30 | |
| *** judd7 has quit IRC | 02:56 | |
| *** echevemaster has quit IRC | 03:19 | |
| *** tkay has joined #openstack-swift | 03:19 | |
| *** mrsnivvel has joined #openstack-swift | 03:23 | |
| *** tkay has quit IRC | 03:23 | |
| *** dvorkbjel has quit IRC | 03:23 | |
| *** IgnacioCorderi has joined #openstack-swift | 03:40 | |
| *** IgnacioCorderi has quit IRC | 03:41 | |
| *** erlon has quit IRC | 04:00 | |
| *** bkopilov has joined #openstack-swift | 04:00 | |
| *** IgnacioCorderi has joined #openstack-swift | 05:07 | |
| *** kopparam has joined #openstack-swift | 05:10 | |
| *** kopparam has quit IRC | 05:26 | |
| *** tsg has quit IRC | 05:27 | |
| *** kopparam has joined #openstack-swift | 05:27 | |
| *** kopparam has quit IRC | 05:31 | |
| *** ppai has joined #openstack-swift | 05:37 | |
| *** geaaru has quit IRC | 05:44 | |
| *** nosnos has quit IRC | 05:57 | |
| *** nosnos_ has joined #openstack-swift | 06:00 | |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: Treat 404s as 204 on object delete in proxy https://review.openstack.org/114120 | 06:02 |
| *** kopparam has joined #openstack-swift | 06:06 | |
| *** nshaikh has joined #openstack-swift | 06:09 | |
| *** IgnacioCorderi has quit IRC | 06:38 | |
| *** kopparam_ has joined #openstack-swift | 07:14 | |
| *** kopparam has quit IRC | 07:16 | |
| *** ujjain has quit IRC | 07:17 | |
| *** ujjain has joined #openstack-swift | 07:17 | |
| *** zul has quit IRC | 07:24 | |
| *** geaaru has joined #openstack-swift | 07:31 | |
| *** tattabbum has joined #openstack-swift | 07:39 | |
| tattabbum | Hi there! | 07:39 |
| tattabbum | I would to upload a zip file (file.zip) into a swift container by means of REST API. The container's path is: http://SWIFT_IP:8080/v1/AUTH_043c13d5c76747f3befed94215b0b0fd/ApplicationRepositoryPubblico | 07:41 |
| tattabbum | How can I do it? | 07:41 |
| tattabbum | The file.zip has a dimension of 30MB | 07:42 |
| tattabbum | I am using this guide: http://docs.openstack.org/api/openstack-object-storage/1.0/content/PUT_createOrReplaceObject__v1__account___container___object__storage_object_services.html | 08:08 |
| ahale | tattabbum: something like this: curl -XPUT -T file.zip http://swift/v1/account/container/file and headers for whatever auth you are using and content type I guess | 08:13 |
| *** Dafna has joined #openstack-swift | 08:24 | |
| tattabbum | ahale: Thank you. It works! I have used: curl -X PUT -T kunagi.war http://SWIFT_IP:8080/v1/AUTH_043c13d5c76747f3befed94215b0b0fd/ApplicationRepositoryPubblico/file.war -H "X-Auth-Token: $token" -H "Content-Type: application" | 08:28 |
| ahale | awesome | 08:28 |
| ahale | hm, is that content type right? should be application/zip or something like that ? a two part thing anyway | 08:29 |
| tattabbum | ahale: I have uploaded a war file. So, inserting -H "Content-Type: application" it works. | 08:30 |
| ahale | ah cool :) just looked weird is all - glad its working | 08:30 |
| ahale | remembered the X-Detect-Content-Type:true header too that makes swift guess, if you're on a relatively recent swift | 08:31 |
| tattabbum | ahale: I am using swift 1.6.0 | 08:32 |
| ahale | ah thats probably not got that feature then | 08:32 |
| *** aix has joined #openstack-swift | 08:39 | |
| *** vr2 has joined #openstack-swift | 08:41 | |
| *** infotection has quit IRC | 08:46 | |
| *** infotection has joined #openstack-swift | 08:50 | |
| *** nshaikh has quit IRC | 08:53 | |
| *** homegrown has joined #openstack-swift | 08:57 | |
| *** infotection has quit IRC | 09:06 | |
| *** infotection has joined #openstack-swift | 09:07 | |
| openstackgerrit | Alistair Coles proposed a change to openstack/swift: moving object validation checks to top of PUT method https://review.openstack.org/115995 | 09:16 |
| acoles | tdasilva: fix for the py26 failure ^^ | 09:16 |
| *** mkollaro has joined #openstack-swift | 09:22 | |
| *** zul has joined #openstack-swift | 09:24 | |
| *** bvandenh has joined #openstack-swift | 09:32 | |
| *** mkollaro has quit IRC | 09:36 | |
| tattabbum | Hi there! How many custom metadata can I create using the X-Object-Meta-{name} header? In other words, I need to create 2 custom metadata for an object: "X-Object-Meta-Name_1=Value_1" and "X-Object-Meta-Name_2=Value_2" | 09:37 |
| tattabbum | Is it possible? | 09:37 |
| *** mkollaro has joined #openstack-swift | 09:44 | |
| *** nshaikh has joined #openstack-swift | 09:44 | |
| openstackgerrit | Prashanth Pai proposed a change to openstack/swift: Mention storage backends in Associated Projects https://review.openstack.org/119724 | 10:37 |
| mattoliverau | tattabbum: so long as the names are unique you can make as many as you want. | 10:57 |
| *** ppai has quit IRC | 11:07 | |
| *** Midnightmyth has joined #openstack-swift | 11:17 | |
| *** dmsimard_away is now known as dmsimard | 11:17 | |
| *** ppai has joined #openstack-swift | 11:23 | |
| *** ppai has quit IRC | 11:47 | |
| tdasilva | acoles: good morning! thanks for taking care of that | 11:47 |
| *** kopparam_ has quit IRC | 12:06 | |
| *** tdasilva has quit IRC | 12:06 | |
| *** dmorita has quit IRC | 12:06 | |
| *** kopparam has joined #openstack-swift | 12:07 | |
| *** kopparam has quit IRC | 12:11 | |
| *** nosnos_ has quit IRC | 12:13 | |
| *** nosnos has joined #openstack-swift | 12:14 | |
| *** nosnos has quit IRC | 12:18 | |
| *** Trixboxer has joined #openstack-swift | 12:32 | |
| *** marcusvrn has joined #openstack-swift | 12:46 | |
| *** bkopilov has quit IRC | 12:48 | |
| *** sandywalsh_ has joined #openstack-swift | 13:21 | |
| *** sandywalsh has quit IRC | 13:24 | |
| *** annegent_ has joined #openstack-swift | 13:24 | |
| openstackgerrit | Alistair Coles proposed a change to openstack/swift: Extra unit tests for check_delete_headers https://review.openstack.org/119756 | 13:37 |
| *** miqui has joined #openstack-swift | 13:49 | |
| *** nshaikh has quit IRC | 13:57 | |
| *** tattabbum has quit IRC | 14:08 | |
| *** kopparam has joined #openstack-swift | 14:11 | |
| *** kopparam has quit IRC | 14:16 | |
| *** tsg has joined #openstack-swift | 14:22 | |
| *** tdasilva has joined #openstack-swift | 14:27 | |
| openstackgerrit | John Dickinson proposed a change to openstack/swift: make the bind_port config setting required https://review.openstack.org/118200 | 14:29 |
| *** zaitcev has joined #openstack-swift | 14:30 | |
| *** ChanServ sets mode: +v zaitcev | 14:30 | |
| *** mahatic has joined #openstack-swift | 14:50 | |
| *** judd7 has joined #openstack-swift | 14:52 | |
| openstackgerrit | Yummy Bian proposed a change to openstack/swift: Too many if clauses in the _clean_incoming_headers and the _clean_outgoing_headers routines of the tempurl module https://review.openstack.org/119782 | 14:53 |
| *** jergerber has joined #openstack-swift | 14:55 | |
| openstackgerrit | David Goetz proposed a change to openstack/swift: Delete expired objects in slightly smarter way. https://review.openstack.org/118434 | 15:00 |
| acoles | notmyname: morning, process trivia: does this https://review.openstack.org/#/c/119724/ need DocImpact? | 15:00 |
| *** marcusvrn has quit IRC | 15:01 | |
| *** Midnightmyth has quit IRC | 15:07 | |
| *** mwstorer has joined #openstack-swift | 15:09 | |
| *** aix has quit IRC | 15:13 | |
| *** aix has joined #openstack-swift | 15:15 | |
| *** vr2 has quit IRC | 15:37 | |
| *** vr3 has joined #openstack-swift | 15:37 | |
| *** tattabbum has joined #openstack-swift | 15:38 | |
| *** vr3 has quit IRC | 15:38 | |
| *** vr2 has joined #openstack-swift | 15:40 | |
| tattabbum | Dear all, I would like that two users (Alice and Bob) of the same tenant (TENANT) see own objects they have uploaded. In other words, Alice haven't see the Bob's objects and viceversa. Is this scenario realizable? Thank you | 15:41 |
| *** bvandenh has quit IRC | 15:42 | |
| openstackgerrit | Alistair Coles proposed a change to openstack/swift: Make probe tests tolerate deprecated policies https://review.openstack.org/119798 | 15:56 |
| *** kyles_ne has joined #openstack-swift | 15:59 | |
| *** gyee has joined #openstack-swift | 15:59 | |
| acoles | tattabbum: does you scenario use keystone auth? | 16:03 |
| *** annegent_ has quit IRC | 16:03 | |
| *** homegrown has left #openstack-swift | 16:03 | |
| tattabbum | acoles: yes, it does. | 16:05 |
| acoles | tattabbum: If Alice and Bob both have admin roles on the same tenant then they can both read all containers/objects in that tenant, so they would be able to read each other's objects. | 16:07 |
| acoles | tattabbum: If Alice and Bob do _not_ have admin roles on the tenant then a third user (with admin role) could grant Alice read/write access to one container and Bob read/write access to another container | 16:08 |
| acoles | using X-Container-Read and X-Container-Write headers to set container access control. | 16:09 |
| acoles | there is no way to apply different access controls to objects in the _same_ container. | 16:10 |
| *** mahatic has quit IRC | 16:10 | |
| tattabbum | acoles: Thanks for your reply. Alice and Bob do _not_ have admin roles on the tenant | 16:12 |
| *** geaaru has quit IRC | 16:13 | |
| *** kevinc_ has joined #openstack-swift | 16:16 | |
| tattabbum | Dear all, I would like that the user Bill with admin role on the tenant TENANT can access the SWIFT's REST API using a "static token", so that he no need to request a new token every time he have to either upload or download an object. | 16:20 |
| tattabbum | How can do it? Thank you. | 16:22 |
| ekarlso- | how do I get a tempurl ? | 16:24 |
| *** geaaru has joined #openstack-swift | 16:24 | |
| acoles | tattabbum: this may be useful http://blog.fsquat.net/?p=40 | 16:26 |
| notmyname | good morning | 16:28 |
| notmyname | acoles: no. that doesn't need DocImpact AFAIK | 16:28 |
| *** mahatic has joined #openstack-swift | 16:28 | |
| notmyname | tattabbum: the token you use (and its validity) is up to the auth system. so if you're using keystone, then you'll have to ask keystone for a non-expiring token. | 16:30 |
| acoles | notmyname: thx | 16:30 |
| acoles | tattabbum: i _think_ that by default a keystone token is valid for 1 hour | 16:31 |
| tattabbum | notmyname: how can I ask keystone a a non-expiring token? | 16:31 |
| notmyname | acoles: I think DocImpact is for things that would need to change in operator docs (eg configs) and API docs. I don't think the docs team maintains any ecosystem stuff | 16:31 |
| notmyname | tattabbum: I don't know (or even if that's possible). maybe acoles knows. or you'll need to ask keystone devs in #openstack-dev | 16:32 |
| tattabbum | notmyname: thank you | 16:32 |
| acoles | notmyname: yeah thats what i figured but just wanted to doublecheck | 16:32 |
| notmyname | tattabbum: however, while I'm not questioning your use case, "forever" sounds kinda scary when thinking about auth tokens. doesn't seem (to me, not knowing your use case) too hard to account for a 401 response to do a reauth | 16:33 |
| *** mahatic has quit IRC | 16:33 | |
| notmyname | especially since you'll need to have most of the "go get a valid token" code anyway. and you'll already need to handle other error conditions for retries | 16:33 |
| acoles | tattabbum: sorry i don't know about non-expiring token. try #openstack-keystone | 16:34 |
| notmyname | tattabbum: eg there's a pretty robust retry method in the python-swiftclient codebase that you could use as a pattern. it handles not only reauth but also ratelimiting and other errors. https://github.com/openstack/python-swiftclient/blob/master/swiftclient/client.py#L1219 | 16:35 |
| *** mahatic has joined #openstack-swift | 16:36 | |
| *** kenhui has joined #openstack-swift | 16:36 | |
| tattabbum | notmyname: acoles: thank you all! | 16:37 |
| *** tattabbum has quit IRC | 16:38 | |
| acoles | tattabbum: no problem | 16:38 |
| *** tkay has joined #openstack-swift | 16:50 | |
| *** vr2 has quit IRC | 16:50 | |
| *** vr1 has joined #openstack-swift | 16:50 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 16:54 |
| *** judd7 has quit IRC | 16:59 | |
| *** vr1 has quit IRC | 17:01 | |
| *** judd7 has joined #openstack-swift | 17:02 | |
| *** annegent_ has joined #openstack-swift | 17:04 | |
| *** kyles_ne has quit IRC | 17:08 | |
| *** kyles_ne has joined #openstack-swift | 17:09 | |
| *** kopparam has joined #openstack-swift | 17:10 | |
| *** gyee has quit IRC | 17:12 | |
| *** kopparam has quit IRC | 17:15 | |
| *** kevinc_ has quit IRC | 17:15 | |
| *** tsg has quit IRC | 17:16 | |
| *** kevinc_ has joined #openstack-swift | 17:18 | |
| notmyname | acoles: FWIW, just asked a wikimedia guy if they explicitly set the bind_port for swift and it turns out they already do | 17:20 |
| acoles | notmyname: same with our public service | 17:22 |
| notmyname | great! | 17:22 |
| acoles | tripleo is the issue | 17:23 |
| acoles | but gerry's on it | 17:23 |
| notmyname | also, thinking about lifeless's idea of binding to multiple ports, that seems to me that it would be trickier for operators since it's another port to configure/secure and also at least one other config to enable/disable | 17:23 |
| notmyname | so I prefer the current plan of making it explicit | 17:23 |
| acoles | hmmm so that would mean we'd need to explicitly disable a second port or get a surprise | 17:25 |
| notmyname | right. or run with both open | 17:25 |
| acoles | or, could be that if a port is config'd, use only that, otherwise run with bot | 17:26 |
| acoles | h | 17:26 |
| notmyname | I can't see that surprising anyone, ever /s | 17:26 |
| notmyname | that's the kind of thing pentest guys love :-) | 17:26 |
| acoles | well you gotta keep them in a job ;) | 17:27 |
| *** gyee has joined #openstack-swift | 17:29 | |
| *** jon___ has joined #openstack-swift | 17:32 | |
| *** annegent_ has quit IRC | 17:34 | |
| *** kenhui has quit IRC | 17:35 | |
| *** kenhui has joined #openstack-swift | 17:35 | |
| notmyname | chmouel: will making the bind_port explicit cause any problems with devstack? | 17:40 |
| jon___ | I tried before using 127.0.0.1 because of DHCP issues. It worked for me | 17:44 |
| *** gyee has quit IRC | 17:46 | |
| *** geaaru has quit IRC | 17:47 | |
| *** annegent_ has joined #openstack-swift | 17:50 | |
| *** homegrown has joined #openstack-swift | 17:50 | |
| *** IgnacioCorderi has joined #openstack-swift | 17:51 | |
| homegrown | I'm looking at some pointers to speed up the listing of objects. I have a bout 500k objects in a container with some pseudo folders. | 17:54 |
| homegrown | Currently using static web, but would be happy with plain text | 17:54 |
| clayg | morning... | 17:55 |
| *** tab_ has joined #openstack-swift | 17:55 | |
| homegrown | evening... ;-) | 17:55 |
| notmyname | homegrown: what are you looking to do? get a full list of all 500k objects | 17:59 |
| notmyname | ? | 17:59 |
| homegrown | yes, as quick as i can | 18:00 |
| homegrown | not sure where to do the tuning / put the resources to speed it up | 18:00 |
| notmyname | homegrown: infrastructure wise, I always recommend using flash storage for container dbs. but otherwise, you'll have to page through the results | 18:01 |
| notmyname | homegrown: now with your UI, there are some tricks you could do, I'd think | 18:02 |
| homegrown | ok, it's currently on fairly old hardware, so sure ssd would speed it up | 18:02 |
| notmyname | homegrown: for example, you have some sort of hierarchy in the object naming your said, so you could eg do a listing with a delimiter and display that immediately then do other concurrent listings in the background | 18:03 |
| notmyname | homegrown: so if your delimiter is '/', then do that listing first, then you can do subsequent (and concurrent!) listings with the prefixes you get back from the first request. effectively walking the tree of listings with breadth-first search. | 18:04 |
| homegrown | possibly, its also important to "find" the latest objects in each pseudo folder first | 18:04 |
| notmyname | homegrown: that would only work if you have structure in the name | 18:04 |
| homegrown | although the name of the object wouldn't necessarily be known when doing the query | 18:05 |
| notmyname | homegrown: ah, swift doesn't index on anything but the name of the object, so you can't get "newest first" or anything like that (interestingly I've heard that request recently elsewhere though) | 18:05 |
| torgomatic | even if you don't have structure you could do prefix=%01, prefix=%02, ..., prefix=%FF and get a 255-way fanout (no %00 because I don't think we allow null bytes) | 18:05 |
| torgomatic | probably some stuff you can remove there too since obj names are UTF-8, so not every byte can start a UTF-8 string | 18:06 |
| notmyname | homegrown: if you want to do listings, sorting, or searching on anything other than the name you'll have to keep your own index of what's in the container | 18:06 |
| homegrown | some more background - it's to use as an artifact store | 18:07 |
| torgomatic | like warehouse 13? neat. | 18:07 |
| notmyname | lol | 18:07 |
| jon___ | I am looking at running my proxy nodes on KVM. I have a 8 VCPU VM, I set it to 16 workers. I was testing with just going from 1,2,3,4,5... current client. I noticed that only 1 of the 16 processes were taking requests. I was not up to closer to 8 or 10 current client that a second process we being used. But on physical hardware I saw the process have a MECH more even usage. Has anyone seen this? | 18:08 |
| homegrown | like apache ivy | 18:08 |
| homegrown | jokers ;-0 | 18:09 |
| jon___ | any ideas? This seems kind of odd to me. | 18:11 |
| torgomatic | jon___: nope, never seen anything like it | 18:11 |
| homegrown | notmyname: torgomatic: maintaining a separate index is not ideal, but i might be able to do something by maintaining a more predictable structure. | 18:11 |
| torgomatic | homegrown: sounds like a job for a static index file and a cronjob, tbh | 18:12 |
| torgomatic | every 5 minutes, walk the whole container (blah blah concurrently blah blah BFS, whatever) and generate an index, then store that index in Swift | 18:13 |
| torgomatic | then when clients go looking for jars or wars or whatev-ars, they'll get the index | 18:14 |
| homegrown | torgomatic: possibly ... will post back here if i manage something better. for now it's beer o'clock in sunny London. | 18:14 |
| torgomatic | I mean, that does assume that uploaders are okay with it taking a few minutes for their stuff to become available, but meh | 18:14 |
| torgomatic | homegrown: hah! well, enjoy. :) | 18:14 |
| jon___ | +torgomatic: thanks. I will create a dummy eventlet wsgi app to see if that is it/ | 18:15 |
| notmyname | homegrown: feel free to keep asking questions :-) | 18:15 |
| homegrown | torgomatic: that's the problem, nobody is willing to wait | 18:15 |
| *** homegrown has quit IRC | 18:15 | |
| *** IgnacioCorderi has quit IRC | 18:20 | |
| *** aix has quit IRC | 18:21 | |
| *** homegrown has joined #openstack-swift | 18:21 | |
| *** tsg has joined #openstack-swift | 18:37 | |
| *** annegent_ has quit IRC | 18:38 | |
| *** homegrown has quit IRC | 18:49 | |
| *** openstackgerrit has quit IRC | 18:51 | |
| *** mkollaro has quit IRC | 18:51 | |
| chmouel | notmyname: humm, let me check | 18:53 |
| *** kenhui has quit IRC | 18:54 | |
| *** swat30 has quit IRC | 18:54 | |
| chmouel | notmyname: yeah i think it will, https://github.com/openstack-dev/devstack/blob/b6f183ad09c51f0d116fecc0c526d6752424a35b/lib/swift#L259 | 18:55 |
| chmouel | notmyname: send a patch and we will see the devstack results if it does if not i will make the iniuncomment function more robust | 18:55 |
| *** thurloat has quit IRC | 18:56 | |
| *** mkollaro has joined #openstack-swift | 19:03 | |
| *** tdasilva has quit IRC | 19:03 | |
| *** kevinc_ has quit IRC | 19:06 | |
| *** tdasilva has joined #openstack-swift | 19:06 | |
| *** openstackgerrit has joined #openstack-swift | 19:08 | |
| *** thurloat has joined #openstack-swift | 19:11 | |
| *** swat30 has joined #openstack-swift | 19:11 | |
| *** kopparam has joined #openstack-swift | 19:15 | |
| *** thurloat has quit IRC | 19:16 | |
| *** swat30 has quit IRC | 19:16 | |
| peluse | torgomatic, any chance you can complete at least a first pass of https://review.openstack.org/#/c/106910/ here soon? | 19:18 |
| torgomatic | peluse: sure; I'm slogging through reviews as we speak | 19:18 |
| peluse | torgomatic, cool. BTW I'll slip over to the PUT side of your slice set here in the next few days as well... | 19:19 |
| torgomatic | peluse: yeah, whenever you get a chance... the socket stuff in there is kind of fun though ;) | 19:19 |
| * peluse might have to save a special beer for that one then... | 19:20 | |
| *** kopparam has quit IRC | 19:20 | |
| *** thurloat has joined #openstack-swift | 19:25 | |
| openstackgerrit | Clay Gerrard proposed a change to openstack/swift: test tempurl header sanitization priority https://review.openstack.org/119885 | 19:26 |
| *** tkay has quit IRC | 19:42 | |
| *** annegent_ has joined #openstack-swift | 19:45 | |
| *** echevemaster has joined #openstack-swift | 19:51 | |
| mahatic | can someone please help me how to add a new patchset for my first commit? I made the changes and did a "git commit -a". It created a new commit. I then did a "git revert HEAD" and then a "git commit -a --amend" | 20:01 |
| mahatic | here is what "git log" looks like: http://paste.openstack.org/show/108446/ | 20:02 |
| *** gyee has joined #openstack-swift | 20:03 | |
| torgomatic | mahatic: so, you're going to want to run "git rebase -i 80b03304^" so that you rebase on the parent of your first commit | 20:05 |
| torgomatic | that'll give you an $EDITOR window with three lines, one per commit | 20:05 |
| torgomatic | keep 80b03304, squash 79e883b into it, and throw away 4457a61 (the revert of 79e883b) | 20:06 |
| torgomatic | that should result in one commit that has all the changes in it | 20:06 |
| *** anderstj has quit IRC | 20:12 | |
| *** anderstj has joined #openstack-swift | 20:13 | |
| *** tdasilva has quit IRC | 20:15 | |
| *** anticw has quit IRC | 20:15 | |
| mahatic | torgomatic, i did a revert of 79e883b because i did not want those changes | 20:15 |
| mahatic | torgomatic, maybe i can just rebase it to original like you said and remove the rest of the two. Then modify my document, then run a "git commit -a --amend". Will that do? | 20:15 |
| clayg | so i'm a little bitter at Ceilometer... | 20:16 |
| *** anticw has joined #openstack-swift | 20:16 | |
| clayg | I could stop there I 'spose; but I have a specific in this case | 20:16 |
| notmyname | clayg: what's up clayg? | 20:16 |
| clayg | I didn't care for Najc claiming Greg's aweomse writeup on the rings was somehow "lacking" | 20:16 |
| torgomatic | mahatic: yeah, if you don't like those other two commits, that'll work | 20:17 |
| clayg | I wanted to post back - but I can hear the disdain coming through -> https://gist.github.com/ | 20:17 |
| clayg | maybe someone should tell me to bite my tongue | 20:17 |
| clayg | here i mean https://gist.github.com/clayg/d46146719c4412fedb02 | 20:17 |
| clayg | oh well, maybe it was just therepudic to get it out and I can go back to ignoring it | 20:19 |
| *** thurloat has quit IRC | 20:19 | |
| notmyname | clayg: I think it's a good response, especially calling out the mistake of just building two rings instead of actually rebalancing | 20:19 |
| notmyname | clayg: there's a few things that could be reworded | 20:20 |
| *** echevemaster has quit IRC | 20:20 | |
| notmyname | clayg: probably taking out the first sentence of the 2nd paragraph would be enough to remove the disdain coming through :-) | 20:21 |
| *** thurloat has joined #openstack-swift | 20:23 | |
| clayg | was that the one about thinking for ourselves or copying and pasting your ring off some blog post on the internet? | 20:25 |
| *** mahatic has quit IRC | 20:25 | |
| notmyname | clayg: thinking for yourself :-) | 20:25 |
| *** erlon has joined #openstack-swift | 20:26 | |
| mattoliverau | Morning | 20:26 |
| notmyname | mattoliverau: hi | 20:26 |
| mattoliverau | You all have a good weekend? | 20:27 |
| *** mahatic has joined #openstack-swift | 20:27 | |
| *** jergerber has quit IRC | 20:29 | |
| openstackgerrit | Mahati proposed a change to openstack/swift: Made changes according to review comments. https://review.openstack.org/119193 | 20:29 |
| smart_developer | Has anyone run into this error when trying a "swift-init all restart"? | 20:29 |
| smart_developer | "No handlers could be found for logger "swift" | 20:29 |
| smart_developer | This happened after changing the proxy-server.conf file to use authtoken/keystoneauth instead of tempauth. | 20:31 |
| notmyname | smart_developer: what version of swift and keystone are you using? | 20:32 |
| chmouel | clayg: i was thinking about the same the guy didn't read the full blog post, i think john's reply was excellent tho | 20:32 |
| smart_developer | swift: 1.13.1 | 20:32 |
| smart_developer | keystone: 1.0, I believe | 20:32 |
| chmouel | smart_developer: i think this has to be with the keystoneclient version instead | 20:32 |
| chmouel | which has the auth_token | 20:32 |
| notmyname | chmouel: yeah, that :-) | 20:33 |
| smart_developer | swift: 1.13 | 20:34 |
| notmyname | smart_developer: I think chmouel has the right idea. it's an issue with whatever version of keystoneclient you're using | 20:34 |
| torgomatic | mahatic: you should put the commit message back to what it was before; right now the whole commit message is just "Made changes according to review comments", and you're gonna get dinged for that | 20:35 |
| mahatic | torgomatic, oh yeah, my bad! | 20:35 |
| smart_developer | I did "keystone --version" my Keystone server and it returned 0.7.1. | 20:35 |
| notmyname | mahatic: (which you can fix with a `git commit --amend` without having to worry about rebasing) | 20:35 |
| notmyname | and if you do `git commit -amend` you get a new commit with the commit message of "end" | 20:36 |
| openstackgerrit | A change was merged to openstack/swift: Pay attention to all punctual nodes https://review.openstack.org/119484 | 20:36 |
| chmouel | smart_developer: to be honest i think we fixed that but i can't rem in which part of the stack this was fixed, keystoneauth (i.e: swift), swift core or keystoneclient | 20:36 |
| notmyname | which is why I have a git alias of "git amend" -> "git commit --amend" cause I'm kinda bad at typing sometimes | 20:37 |
| chmouel | (there is even a keystone auth token middleware repo now but i doubt you have that unless you run git head) | 20:37 |
| *** jon___ has quit IRC | 20:37 | |
| openstackgerrit | Mahati proposed a change to openstack/swift: Added instructions to create a label or UUID to the XFS volume and mount using it. https://review.openstack.org/119193 | 20:38 |
| mahatic | notmyname, great. torgomatic, done. Thank you both! | 20:40 |
| notmyname | mahatic: one tiny thing there. in the UUID example, you're using tabs for indention, not spaces. can you change it? | 20:40 |
| *** robemcallister has joined #openstack-swift | 20:41 | |
| mahatic | notmyname, oh okay. sure, i will. | 20:41 |
| notmyname | thanks | 20:42 |
| *** robemcallister has quit IRC | 20:42 | |
| *** robemcallister has joined #openstack-swift | 20:43 | |
| *** miqui has quit IRC | 20:43 | |
| *** robemcallister is now known as rmcall | 20:44 | |
| *** annegent_ has quit IRC | 20:44 | |
| mahatic | notmyname, is it advised that i make a quick change and push it? or maybe just wait for the comments, put them together and then make a commit? | 20:45 |
| notmyname | mahatic: go ahead and make this short one. I only asked since I guessed you still have it open and haven't moved on to something else. otherwise I could just push over your patch whenever I get around to it | 20:46 |
| mahatic | notmyname, yup. it is still open. i can make it now | 20:46 |
| notmyname | mahatic: thanks | 20:46 |
| mattoliverau | Boarding a plane, be back in an hour or so. Unless they have WiFi (which I doubt) | 20:47 |
| notmyname | mattoliverau: you could use gertty! | 20:47 |
| notmyname | mattoliverau: where are you going? | 20:47 |
| mattoliverau | I have it installed so reviews will happen :) | 20:47 |
| notmyname | lol. awesome :-) | 20:47 |
| *** HenryG has quit IRC | 20:48 | |
| mattoliverau | I'm off to visit and work from the folks place (at the beach) :) being a good son :) | 20:49 |
| notmyname | nice | 20:50 |
| *** rmcall has quit IRC | 20:50 | |
| *** occupant has joined #openstack-swift | 20:51 | |
| openstackgerrit | Mahati proposed a change to openstack/swift: Added instructions to create a label or UUID to the XFS volume and mount using it. https://review.openstack.org/119193 | 20:53 |
| *** rmcall has joined #openstack-swift | 20:53 | |
| *** annegent_ has joined #openstack-swift | 21:00 | |
| *** HenryG has joined #openstack-swift | 21:00 | |
| *** tgohad has joined #openstack-swift | 21:01 | |
| *** tsg has quit IRC | 21:02 | |
| *** mahatic has quit IRC | 21:06 | |
| *** tsg has joined #openstack-swift | 21:08 | |
| *** tgohad has quit IRC | 21:10 | |
| *** kevinc_ has joined #openstack-swift | 21:25 | |
| *** annegent_ has quit IRC | 21:36 | |
| *** shri has joined #openstack-swift | 21:36 | |
| *** tgohad has joined #openstack-swift | 21:43 | |
| *** tsg has quit IRC | 21:46 | |
| hurricanerix_ | torgomatic: you around? | 21:52 |
| *** tgohad has quit IRC | 22:00 | |
| *** jasondotstar has quit IRC | 22:03 | |
| torgomatic | hurricanerix_: yep | 22:05 |
| *** miqui has joined #openstack-swift | 22:13 | |
| *** kopparam has joined #openstack-swift | 22:25 | |
| hurricanerix_ | torgomatic: sorry, i stepped away for a bit. I just wanted to talk about that formpost test some to see if there is a better solution. | 22:26 |
| torgomatic | hurricanerix_: if you take off the exception handler, then a unit test with a too-large file should fail with an error, I think | 22:27 |
| hurricanerix_ | torgomatic: Yeah, my main issue was that the unit test would pass (since the formpost middleware would catch the exception). But when you ran your saio, it would never see the exception. So having the test didn't really help any. | 22:28 |
| mattoliverau | Landed, it was the bumpiest flight I've ever taken (small regional airline = small plane + bad weather). Interesting trying to review via gertty while bumping around.. I'll call it extreme reviewing :p | 22:28 |
| hurricanerix_ | torgomatic: *never see it since the real app wraps everything in a try/except Exception/Timeout. | 22:29 |
| hurricanerix_ | torgomatic: So I was just trying to modify the test so without the change, the test would fail similar to how it works in a saio/prod. | 22:30 |
| *** kopparam has quit IRC | 22:30 | |
| torgomatic | hurricanerix_: that makes sense, it just makes it harder to write new tests because your no-such-method error gets swallowed | 22:33 |
| hurricanerix_ | torgomatic: Yeah. So you think I should just revert it, or do you have a better approach that I can try? I don't want to make writing new tests harder, but at the same time, if they don't catch problems, why bother writing them? =) | 22:35 |
| torgomatic | hurricanerix_: I'd write a test with a too-large file that looks for a 400 status code; right now it looks like it'll get a 499, yes? | 22:35 |
| torgomatic | so that'll be a nice failure, and it'll help document the intent of the code | 22:36 |
| hurricanerix_ | No, it will get the 400 | 22:36 |
| torgomatic | so clients right now see 400 on too-large file? | 22:36 |
| hurricanerix_ | we get the 499 because in a real saio, the app catches the IOError and thinks the client disconnected. | 22:36 |
| hurricanerix_ | Since in the test, the fake app does not wrap everything in a try/except, the formpost is able to catch the IOError | 22:37 |
| torgomatic | so formpost replies to the client with a 400, and some other request looks disconnected and gets a 499? | 22:37 |
| *** tsg has joined #openstack-swift | 22:37 | |
| torgomatic | also, is "too large" too large for Swift, or too large for the max_file_size param of the form? | 22:38 |
| hurricanerix_ | *EOFError | 22:38 |
| hurricanerix_ | So formpost raises the EOFError when the file size is exceeded. | 22:38 |
| hurricanerix_ | The formpost middleware attempts to catch it | 22:38 |
| *** IgnacioCorderi has joined #openstack-swift | 22:38 | |
| *** IgnacioCorderi has quit IRC | 22:38 | |
| hurricanerix_ | But, because the app wraps everything in a try/except, it thinks it is the client disconnecting. | 22:38 |
| *** tgohad has joined #openstack-swift | 22:39 | |
| hurricanerix_ | So if you run the unit test, the test passes checking for a 400, because the fake app does not have that try/except. | 22:39 |
| *** tgohad__ has joined #openstack-swift | 22:39 | |
| hurricanerix_ | Bug if you spin up a saio, and post a file larger than the max file size, you will get a 499 back. | 22:39 |
| hurricanerix_ | *But | 22:39 |
| torgomatic | Oh... so formpost replaces the request's wsgi.input with Folger's crystals^W^Wa _CappedFileLikeObject, and that raises an exception when it's done to blow up | 22:40 |
| hurricanerix_ | yeah | 22:40 |
| torgomatic | blow up in the proxy code, that is | 22:40 |
| hurricanerix_ | Yeah, and the app catches it and it never gets relayed back to the middleware. | 22:41 |
| *** tsg has quit IRC | 22:41 | |
| *** tgohad has quit IRC | 22:42 | |
| hurricanerix_ | So I added that flag, so that the middleware could check if the file size was exceeded. But the unit test will always pass, so I thought I would change it also, so in the future it could catch the error. | 22:42 |
| torgomatic | and so, against a real Swift proxy, the client receives a 499 status code? | 22:42 |
| hurricanerix_ | Yeah | 22:42 |
| hurricanerix_ | Sorry if I am doing a terrible job at explaining this. =) | 22:42 |
| torgomatic | Yikes... okay, I'd change the test's app so that it catches EOFError and emits a 499 just like the proxy does, then go ahead and fix formpost | 22:42 |
| torgomatic | no, I got it :) | 22:42 |
| hurricanerix_ | Ok, thanks. | 22:44 |
| torgomatic | sure :) | 22:44 |
| *** dmsimard is now known as dmsimard_away | 22:49 | |
| *** mkollaro has quit IRC | 23:08 | |
| *** dmsimard_away is now known as dmsimard | 23:09 | |
| *** mwstorer has quit IRC | 23:15 | |
| *** tab_ has quit IRC | 23:16 | |
| *** kyles_ne has quit IRC | 23:20 | |
| *** Trixboxer has quit IRC | 23:23 | |
| *** tsg has joined #openstack-swift | 23:30 | |
| *** tgohad__ has quit IRC | 23:30 | |
| *** tgohad has joined #openstack-swift | 23:32 | |
| *** tsg has quit IRC | 23:35 | |
| *** rmcall has quit IRC | 23:56 | |
| *** rmcall has joined #openstack-swift | 23:57 | |
| *** kyles_ne has joined #openstack-swift | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!