| *** dmorita has joined #openstack-swift | 00:33 | |
| *** matsuhashi has joined #openstack-swift | 00:35 | |
| *** matsuhashi has quit IRC | 00:35 | |
| *** matsuhashi has joined #openstack-swift | 00:35 | |
| *** mlipchuk has quit IRC | 00:41 | |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: When a filesystem does't support xattr return a 507 https://review.openstack.org/99883 | 00:47 |
|---|---|---|
| *** nsquare has joined #openstack-swift | 00:47 | |
| *** haomaiwang has quit IRC | 00:51 | |
| *** haomaiwang has joined #openstack-swift | 00:54 | |
| *** nosnos has joined #openstack-swift | 00:55 | |
| *** zhiyan_ is now known as zhiyan | 02:00 | |
| *** yuan has quit IRC | 02:06 | |
| *** zhiyan is now known as zhiyan_ | 02:18 | |
| openstackgerrit | Matthew Oliver proposed a change to openstack/swift: When a filesystem does't support xattr return a 507 https://review.openstack.org/99883 | 02:22 |
| *** zhiyan_ is now known as zhiyan | 02:27 | |
| *** fifieldt__ is now known as fifieldt | 02:30 | |
| *** haomai___ has joined #openstack-swift | 02:33 | |
| *** haomaiwang has quit IRC | 02:34 | |
| *** bkopilov has quit IRC | 02:47 | |
| *** nosnos has quit IRC | 03:36 | |
| *** nsquare has quit IRC | 03:50 | |
| *** nosnos has joined #openstack-swift | 04:15 | |
| *** bkopilov has joined #openstack-swift | 04:22 | |
| *** bkopilov_ has joined #openstack-swift | 04:28 | |
| *** bkopilov_ has joined #openstack-swift | 04:29 | |
| *** bkopilov has quit IRC | 04:31 | |
| *** kashyapk has joined #openstack-swift | 04:39 | |
| *** haomai___ has quit IRC | 04:44 | |
| *** haomaiwa_ has joined #openstack-swift | 04:45 | |
| *** kashyapk has quit IRC | 04:52 | |
| *** kashyapk has joined #openstack-swift | 04:52 | |
| *** kashyapk_ has joined #openstack-swift | 04:57 | |
| *** kashyapk has quit IRC | 04:58 | |
| openstackgerrit | A change was merged to openstack/swift: replaced e.message with a bare "e" object ref https://review.openstack.org/93570 | 04:58 |
| *** haomai___ has joined #openstack-swift | 05:00 | |
| *** haomaiwa_ has quit IRC | 05:03 | |
| *** ppai has joined #openstack-swift | 05:05 | |
| *** psharma has joined #openstack-swift | 05:16 | |
| *** nshaikh has joined #openstack-swift | 05:17 | |
| *** kashyapk_ has quit IRC | 05:21 | |
| *** kashyapk has joined #openstack-swift | 05:22 | |
| *** kashyapk has quit IRC | 05:30 | |
| *** kashyapk has joined #openstack-swift | 05:30 | |
| *** asdfsf has quit IRC | 05:34 | |
| *** nsquare has joined #openstack-swift | 05:42 | |
| *** zhiyan is now known as zhiyan_ | 05:55 | |
| *** zhiyan_ is now known as zhiyan | 05:56 | |
| *** matsuhashi has quit IRC | 06:00 | |
| *** matsuhashi has joined #openstack-swift | 06:01 | |
| *** matsuhas_ has joined #openstack-swift | 06:38 | |
| *** kashyapk has quit IRC | 06:38 | |
| *** matsuhashi has quit IRC | 06:38 | |
| *** kashyapk_ has joined #openstack-swift | 06:39 | |
| *** itarchitectkev has joined #openstack-swift | 06:48 | |
| *** haomai___ has quit IRC | 06:53 | |
| *** haomaiwang has joined #openstack-swift | 06:53 | |
| *** jamiehannaford has joined #openstack-swift | 06:56 | |
| *** matsuhas_ has quit IRC | 07:01 | |
| *** kashyapk_ has quit IRC | 07:03 | |
| *** matsuhashi has joined #openstack-swift | 07:04 | |
| *** chandan_kumar has quit IRC | 07:05 | |
| *** haomaiw__ has joined #openstack-swift | 07:08 | |
| *** haomaiwang has quit IRC | 07:10 | |
| *** haomaiw__ has quit IRC | 07:14 | |
| *** nsquare has quit IRC | 07:30 | |
| *** mmcardle has joined #openstack-swift | 07:30 | |
| *** kashyapk has joined #openstack-swift | 07:35 | |
| *** kashyapk has quit IRC | 07:38 | |
| *** kashyapk has joined #openstack-swift | 07:39 | |
| *** scohen has joined #openstack-swift | 07:42 | |
| mattoliverau | Well I'm calling it a day, night all! And for those of you on the other side of the world, enjoy your Monday, mine is over :) | 07:42 |
| *** kashyapk has quit IRC | 07:45 | |
| *** kashyapk has joined #openstack-swift | 07:45 | |
| *** kashyapk has quit IRC | 07:47 | |
| *** kashyapk has joined #openstack-swift | 07:47 | |
| *** rieglflo has joined #openstack-swift | 07:48 | |
| *** nacim has joined #openstack-swift | 07:54 | |
| rieglflo | hi everyone. is there any way to restrict container/object access to only one user/account while prohibiting that user from changing the acl to give everyone access? | 07:55 |
| rieglflo | a hint would be really useful, thanks | 07:56 |
| *** ppai has quit IRC | 07:57 | |
| *** foexle has joined #openstack-swift | 08:03 | |
| hugokuo | rieglflo: well.... Swift has only container and account ACL. | 08:07 |
| hugokuo | rieglflo: no Object level ACL tho. | 08:08 |
| hugokuo | rieglflo: Would you like information about container or account level ACLs ? | 08:08 |
| hugokuo | rieglflo: For container or account ACLs, only account Admin is able to set the ACL. | 08:09 |
| *** kashyapk_ has joined #openstack-swift | 08:11 | |
| *** Midnightmyth has joined #openstack-swift | 08:11 | |
| *** kashyapk has quit IRC | 08:11 | |
| rieglflo | hugokuo: so setting X-Container-Read: .r:*,.rlistings is only possible via an admin role? | 08:14 |
| hugokuo | rieglflo: yup... | 08:15 |
| rieglflo | hugokuo: great. is there any other way for an authenticated user to make any objects publicly available expect the temp url plugin? | 08:17 |
| rieglflo | hugokuo: except ... via the temp url plugin? | 08:18 |
| *** fbo_away is now known as fbo | 08:20 | |
| hugokuo | grapsus_: no object level ACL implementation. So tempurl prolly the best way to make publicity on "single object" | 08:20 |
| hugokuo | rieglflo: no object level ACL implementation. So tempurl prolly the best way to make publicity on "single object" | 08:20 |
| *** ppai has joined #openstack-swift | 08:27 | |
| rieglflo | thank you very much! | 08:28 |
| *** nshaikh has quit IRC | 08:37 | |
| *** circ-user-eYrGs has joined #openstack-swift | 08:45 | |
| *** help has joined #openstack-swift | 08:48 | |
| *** help is now known as Guest17639 | 08:48 | |
| *** circ-user-eYrGs has quit IRC | 08:50 | |
| *** mkollaro has joined #openstack-swift | 08:53 | |
| *** nacim has quit IRC | 08:54 | |
| *** zhiyan is now known as zhiyan_ | 08:56 | |
| *** Guest17639 has quit IRC | 08:58 | |
| openstackgerrit | Eamonn O'Toole proposed a change to openstack/swift: Fix swift-recon --auditor bug https://review.openstack.org/100192 | 09:01 |
| *** nacim has joined #openstack-swift | 09:06 | |
| *** charz has joined #openstack-swift | 09:09 | |
| *** nacim has quit IRC | 09:10 | |
| openstackgerrit | Zhang Hua proposed a change to openstack/swift: Add distributed tracing capablities in logging. https://review.openstack.org/93677 | 09:12 |
| openstackgerrit | Eamonn O'Toole proposed a change to openstack/swift: Fix swift-recon --auditor bug https://review.openstack.org/100192 | 09:12 |
| *** JelleB has quit IRC | 09:16 | |
| *** AbyssOne has quit IRC | 09:16 | |
| *** nacim has joined #openstack-swift | 09:23 | |
| *** kashyapk_ has quit IRC | 09:25 | |
| *** kashyapk has joined #openstack-swift | 09:26 | |
| hugokuo | charz: | 09:26 |
| *** itarchitectkev_ has joined #openstack-swift | 09:26 | |
| *** itarchitectkev has quit IRC | 09:29 | |
| *** kashyapk has quit IRC | 09:30 | |
| *** kashyapk has joined #openstack-swift | 09:30 | |
| *** davidhadas_ has quit IRC | 09:32 | |
| *** ppai has quit IRC | 09:34 | |
| *** Midnightmyth has quit IRC | 09:46 | |
| *** luisbg has quit IRC | 09:50 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/python-swiftclient: Updated from global requirements https://review.openstack.org/89250 | 09:52 |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 09:53 |
| *** a1|away has joined #openstack-swift | 09:54 | |
| *** a1|away is now known as AbyssOne | 09:54 | |
| *** itarchitectkev_ has quit IRC | 09:56 | |
| *** itarchitectkev_ has joined #openstack-swift | 09:58 | |
| *** madhuri has joined #openstack-swift | 09:58 | |
| madhuri | Hi, can anyone tell me where does the actual file content is read in case of object GET? | 09:59 |
| madhuri | Mainly, the code part | 09:59 |
| *** AbyssOne_ has joined #openstack-swift | 09:59 | |
| *** ondergetekende has quit IRC | 10:05 | |
| *** openstackgerrit has quit IRC | 10:06 | |
| *** ppai has joined #openstack-swift | 10:06 | |
| *** openstackgerrit has joined #openstack-swift | 10:06 | |
| *** davidhadas has joined #openstack-swift | 10:09 | |
| *** Midnightmyth has joined #openstack-swift | 10:09 | |
| *** haomaiwa_ has joined #openstack-swift | 10:13 | |
| *** kashyapk has quit IRC | 10:14 | |
| acoles | madhuri: DiskFileReader class - https://github.com/openstack/swift/blob/master/swift/obj/diskfile.py#L747 | 10:15 |
| madhuri | acoles: thanks.But where does this __iter__ gets called? | 10:17 |
| *** charz has quit IRC | 10:21 | |
| *** haomaiwa_ has quit IRC | 10:23 | |
| *** haomaiwang has joined #openstack-swift | 10:23 | |
| *** matsuhashi has quit IRC | 10:24 | |
| *** matsuhashi has joined #openstack-swift | 10:24 | |
| portante | madhuri: it is invoked by the wsgi protocol itself | 10:24 |
| acoles | madhuri: it gets set in the response here https://github.com/openstack/swift/blob/master/swift/obj/server.py#L498. the iter will then be called by the wsgi server. | 10:24 |
| *** ondergetekende has joined #openstack-swift | 10:25 | |
| portante | the iterator is invoked when the wsgi code (in eventlet) starts to construct the body | 10:25 |
| *** haomaiw__ has joined #openstack-swift | 10:25 | |
| *** kashyapk has joined #openstack-swift | 10:26 | |
| *** haomaiwang has quit IRC | 10:28 | |
| madhuri | portante: Thanks a lot | 10:30 |
| madhuri | acoles: Thanks a lot. Got it | 10:30 |
| *** scohen has quit IRC | 10:31 | |
| *** matsuhashi has quit IRC | 10:33 | |
| *** matsuhashi has joined #openstack-swift | 10:34 | |
| *** openstackgerrit has quit IRC | 10:35 | |
| *** openstackgerrit has joined #openstack-swift | 10:35 | |
| *** uksysadmin has joined #openstack-swift | 10:39 | |
| *** matsuhashi has quit IRC | 10:41 | |
| *** matsuhashi has joined #openstack-swift | 10:41 | |
| *** itarchitectkev_ has quit IRC | 10:42 | |
| *** tdasilva has joined #openstack-swift | 10:57 | |
| *** nshaikh has joined #openstack-swift | 10:58 | |
| *** mmcardle has quit IRC | 11:00 | |
| *** davidhadas___ has joined #openstack-swift | 11:03 | |
| *** ppai has quit IRC | 11:03 | |
| *** scohen has joined #openstack-swift | 11:03 | |
| *** mmcardle has joined #openstack-swift | 11:04 | |
| *** davidhadas has quit IRC | 11:06 | |
| *** kashyapk is now known as kashyap_ | 11:08 | |
| *** kashyap_ is now known as kashyapk | 11:08 | |
| *** nosnos has quit IRC | 11:15 | |
| *** ppai has joined #openstack-swift | 11:23 | |
| *** kashyapk is now known as kopparam | 11:25 | |
| *** kopparam is now known as kashyapk | 11:26 | |
| *** kashyapk is now known as kopparam | 11:27 | |
| *** matsuhashi has quit IRC | 11:29 | |
| *** matsuhashi has joined #openstack-swift | 11:29 | |
| *** tdasilva has left #openstack-swift | 11:29 | |
| *** matsuhashi has quit IRC | 11:33 | |
| *** mmcardle has quit IRC | 11:45 | |
| *** dmorita has quit IRC | 11:46 | |
| *** Trixboxer has joined #openstack-swift | 11:54 | |
| *** mmcardle has joined #openstack-swift | 12:10 | |
| *** tdasilva has joined #openstack-swift | 12:14 | |
| *** wer has joined #openstack-swift | 12:15 | |
| *** davidhadas___ has quit IRC | 12:24 | |
| *** ppai has quit IRC | 12:24 | |
| *** scohen has quit IRC | 12:25 | |
| *** miqui has joined #openstack-swift | 12:33 | |
| openstackgerrit | A change was merged to openstack/swift: Remove unused variable https://review.openstack.org/97287 | 12:42 |
| *** scohen has joined #openstack-swift | 12:42 | |
| *** zhiyan_ is now known as zhiyan | 12:44 | |
| *** kopparam has quit IRC | 12:44 | |
| *** mmcardle has quit IRC | 12:57 | |
| *** haomaiw__ has quit IRC | 12:58 | |
| *** nshaikh has quit IRC | 13:00 | |
| *** sandywalsh has quit IRC | 13:01 | |
| *** ujjain has joined #openstack-swift | 13:02 | |
| *** ujjain has quit IRC | 13:04 | |
| *** ujjain has joined #openstack-swift | 13:04 | |
| *** sandywalsh has joined #openstack-swift | 13:04 | |
| *** bkopilov_ has quit IRC | 13:06 | |
| *** dmsimard_away is now known as dmsimard | 13:27 | |
| acoles | chmouel: if you have a moment could you look at this one again https://review.openstack.org/#/c/91788/ , now swiftclient tests are passing again :) thx | 13:28 |
| chmouel | acoles: will do thanks, unfortunately prob not have time before wed tho | 13:30 |
| acoles | chmouel: ok. thanks. | 13:30 |
| openstackgerrit | Eamonn O'Toole proposed a change to openstack/swift: Parallel object auditor https://review.openstack.org/59778 | 13:30 |
| *** kopparam has joined #openstack-swift | 13:40 | |
| *** psharma has quit IRC | 13:48 | |
| *** kopparam has quit IRC | 13:49 | |
| *** kopparam has joined #openstack-swift | 13:50 | |
| *** uksysadmin has quit IRC | 13:55 | |
| *** uksysadmin has joined #openstack-swift | 13:56 | |
| *** nsquare has joined #openstack-swift | 14:18 | |
| *** asdfsf has joined #openstack-swift | 14:23 | |
| *** itarchitectkev_ has joined #openstack-swift | 14:23 | |
| *** uksysadmin has quit IRC | 14:26 | |
| *** asdfsf has quit IRC | 14:31 | |
| *** mmcardle has joined #openstack-swift | 14:33 | |
| *** sts has joined #openstack-swift | 14:33 | |
| sts | hello folks. Is it possible to prevent a user from creating new containers in SWIFT? | 14:34 |
| *** chandan_kumar has joined #openstack-swift | 14:36 | |
| *** Kr4zy has joined #openstack-swift | 14:37 | |
| Kr4zy | What is the recommended SSL terminator for performance and scalability? I heard Nginx, but have also heard of issues. Pound sounds like a good alternative, but heard that it doesn't scale. | 14:38 |
| *** tdasilva has quit IRC | 14:38 | |
| *** tdasilva has joined #openstack-swift | 14:44 | |
| *** mmcardle has quit IRC | 14:44 | |
| otoolee | @cschwede I've posted comments on your swift-recon fix, and I've also posted an alternative fix which retains the generator (I need the generator for the parallel object auditor patch) - see https://review.openstack.org/#/c/100192/2 | 14:44 |
| *** lpabon has joined #openstack-swift | 14:45 | |
| *** bkopilov_ has joined #openstack-swift | 14:45 | |
| portante | Kr4zy: haproxy is another, stud (https://github.com/bumptech/stud) is yet another | 14:46 |
| portante | I believe RAX uses haproxy | 14:46 |
| *** mmcardle has joined #openstack-swift | 14:46 | |
| notmyname | good morning world | 14:47 |
| portante | notmyname: sts asked about container creation prevention for a given user above | 14:49 |
| notmyname | sounds like a non swift owner user | 14:49 |
| *** lpabon has quit IRC | 14:51 | |
| *** foexle has quit IRC | 14:55 | |
| otoolee | cschwede: (repeat of above comment) I've posted comments on your swift-recon fix, and I've also posted an alternative fix which retains the generator (I need the generator for the parallel object auditor patch) - see https://review.openstack.org/#/c/100192/2 | 14:55 |
| *** judd7 has joined #openstack-swift | 14:57 | |
| portante | sts, I looked at the code, and I don't immediately see a way to prevent a user from creating a container, but I could be missing something obvious | 14:59 |
| Kr4zy | portante: I am using HAProxy for Load balancing only on a dedicated server. I need SSL termination on the proxy servers. Is HAProxy or Stud the better of the bunch for SSL termination in terms of performance and scalability? | 15:00 |
| creiht | Kr4zy: haproxy | 15:01 |
| ahale | yeah we found newer dev releases of haproxy with ssl were better than stable + stud | 15:01 |
| portante | creiht: do you run haproxy locally for each proxy server or front a few haproxy servers in front of the swift proxy servers? | 15:01 |
| ahale | we have a bunch infront | 15:02 |
| creiht | howdy ahale :) | 15:02 |
| portante | ahale, pardon my ignorance, are you RAX? | 15:02 |
| creiht | portante: yeah he will know a lot more about that stuff than I will :) | 15:02 |
| * portante guesses yes | 15:02 | |
| ahale | yeah | 15:02 |
| ahale | ops side rather than dev | 15:03 |
| portante | ahale: where the cool kids play | 15:03 |
| notmyname | FWIW we run pound locally on each proxy box behind the LB tier. but we have a task in the queue to move away from pound to haproxy | 15:03 |
| creiht | portante: if you come to the openstack paris summit, then you will likely get a chance to meet the amazing ahale :) | 15:03 |
| creiht | and his sidekick omame | 15:04 |
| ahale | hehe | 15:04 |
| creiht | :) | 15:04 |
| portante | creiht: cool, trying to swing that, hopefully | 15:04 |
| ahale | when is that summit? | 15:04 |
| portante | Nov beg | 15:04 |
| creiht | though, I'm not really sure which is the sidekick :) | 15:04 |
| notmyname | "The next OpenStack Summit will be held in Paris, France, November 3-7, 2014. " | 15:05 |
| creiht | ahale: I guess I assumed you were going to go, will you and omame be able to make it? | 15:06 |
| omame | creiht: he took off, be back in ~15'. as for myself, I'm leaning towards buying a train ticket and head there | 15:10 |
| Kr4zy | creiht: how does haproxy compare to nginx and pound in terms of performance and scalability? haproxy ssl was just introduced in the latest version (1.5); is it ready for production? | 15:11 |
| creiht | Kr4zy: we've been running the dev versions for quite a while, and like ahale said, they have been more stable then the previous release | 15:12 |
| creiht | omame: cool | 15:12 |
| creiht | Kr4zy: and haproxy continues to beat the others in performance | 15:13 |
| creiht | I also would generally not reccomend nginx for swift, as it spools all requests | 15:14 |
| Kr4zy | creiht: thanks for the info :) | 15:15 |
| *** Midnightmyth has quit IRC | 15:16 | |
| *** gholt has joined #openstack-swift | 15:19 | |
| *** ChanServ sets mode: +v gholt | 15:19 | |
| *** zul has quit IRC | 15:22 | |
| *** zul has joined #openstack-swift | 15:23 | |
| notmyname | ok, IRC buffers read. email handled. coffee in mug. time to dive in for the day. | 15:24 |
| notmyname | first up is figuring out where we are with the SP patch chain | 15:25 |
| notmyname | looks like the non-SP patches that were reviewed have landed | 15:26 |
| notmyname | which leaves just the SP patches | 15:26 |
| notmyname | how do we feel about them? | 15:26 |
| portante | pretty good, more reviews coming on my part, gonna add test environment where in-memory object server is setup as a another storage policy by default | 15:32 |
| * portante heads out for a ride, back after 2 PM EDT | 15:33 | |
| notmyname | I'd like to see the SP patches approved today so that we can get the mechanics working for merging it (new feature branch, force push the chain, gate the merge commit). | 15:36 |
| serverascode | Hi All, another question...we have a small two region swift cluster, when we upload larger objects they fail to replicate due to a rsync timeout of 30 seconds...anyone have any suggestions as to how to choose a timeout value? | 15:37 |
| serverascode | ie. the timeout for object replicator, I think | 15:38 |
| *** gholt has quit IRC | 15:38 | |
| *** kopparam has quit IRC | 15:41 | |
| *** zhiyan is now known as zhiyan_ | 15:52 | |
| creiht | heh | 15:56 |
| creiht | http://blog.zhaw.ch/icclab/evaluating-the-performance-of-ceph-and-swift-for-object-storage-on-small-clusters/ | 15:56 |
| creiht | while noting some of the things that I reported, he didn't even mention the biggest thing which was the fact that his performance numbers didn't add up for swift :/ | 15:57 |
| ahale | creiht: yeah i think we plan to go to paris - I've been kinda grumpy that I haven't been allowed since boston | 16:01 |
| *** nacim has quit IRC | 16:01 | |
| *** scohen has quit IRC | 16:04 | |
| *** scohen has joined #openstack-swift | 16:05 | |
| *** Midnightmyth has joined #openstack-swift | 16:08 | |
| *** mwstorer has joined #openstack-swift | 16:09 | |
| *** scohen has quit IRC | 16:09 | |
| *** jergerber has joined #openstack-swift | 16:12 | |
| *** kevinc_ has joined #openstack-swift | 16:14 | |
| *** gyee has joined #openstack-swift | 16:26 | |
| *** Kr4zy has quit IRC | 16:29 | |
| goodes | good morning all | 16:30 |
| *** Ju_ has quit IRC | 16:31 | |
| notmyname | goodes: good morning | 16:32 |
| *** nsquare has quit IRC | 16:33 | |
| *** itarchitectkev_ has quit IRC | 16:35 | |
| goodes | noob question - when you see > name: hello there - is that somene typing ‘name: hello’ or is that an IRC command that you use to reference the person? | 16:36 |
| notmyname | goodes: that's what I'm typing to specifically address you. many IRC clients have some sort of alert associated with your current user name. this means that eg I can pick out stuff addressed specifically to me in a busy channel | 16:37 |
| notmyname | goodes: all IRC commends start with a "/" and aren't generally displayed here in the normal chat | 16:38 |
| notmyname | except for /me. for example: | 16:38 |
| * notmyname uses /me to show goodes | 16:38 | |
| goodes | but do you just type my name then semicolon or is that being done my an irc command ie /cmd handle hello there? | 16:38 |
| notmyname | goodes: well, my IRC client has a "tab-to-autocomplete" your nick, so I type "g" and hit tab and it expands to "goodes: " | 16:39 |
| goodes | notmyname: thanks, obvious now that you said it | 16:39 |
| *** kopparam has joined #openstack-swift | 16:43 | |
| notmyname | goodes: I think you were in here this weekend asking some questions. this channel is more busy during the week (most of the active people are in north america, although we have a few in the EU and Asia/Australia) | 16:43 |
| notmyname | goodes: feel free to ask questions, and someone will probably answer or at least point you to some docs | 16:43 |
| *** itarchitectkev_ has joined #openstack-swift | 16:44 | |
| *** scohen has joined #openstack-swift | 16:44 | |
| notmyname | acoles: chmouel: cschwede: may already be too late for you guys, but I'd love to have a go/no-go from you on the SP patches before your end-of-day today, please | 16:47 |
| *** kopparam has quit IRC | 16:50 | |
| *** mmcardle has quit IRC | 16:55 | |
| *** scohen has quit IRC | 16:58 | |
| *** scohen has joined #openstack-swift | 16:59 | |
| *** lpabon has joined #openstack-swift | 17:13 | |
| acoles | clayg: where are at with the timestamp offset format - current patch is still 8 digits, are you going to bump this up to 12? | 17:14 |
| acoles | notmyname: working on it... | 17:14 |
| *** nsquare has joined #openstack-swift | 17:23 | |
| *** jamiehannaford has quit IRC | 17:26 | |
| *** mkollaro has quit IRC | 17:29 | |
| *** mkollaro has joined #openstack-swift | 17:31 | |
| *** nsquare has quit IRC | 17:32 | |
| *** nsquare has joined #openstack-swift | 17:36 | |
| *** ChanServ changes topic to "Storage Policies have been proposed to master. Soft freeze in effect until these patches land | First SP patch: https://review.openstack.org/#/c/96026/ | Last SP patch: https://review.openstack.org/#/c/96049/" | 17:40 | |
| *** zaitcev has joined #openstack-swift | 17:43 | |
| *** ChanServ sets mode: +v zaitcev | 17:43 | |
| openstackgerrit | Samuel Merritt proposed a change to openstack/swift: Fix SLO test with old simplejson https://review.openstack.org/99805 | 17:44 |
| *** annegentle has joined #openstack-swift | 17:46 | |
| *** Kr4zy has joined #openstack-swift | 17:48 | |
| Kr4zy | anyone know why proxy-buffering on nginx is not recommended for Swift? | 17:50 |
| notmyname | Kr4zy: you don't want 10 concurrent users uploading a 5GB all at the same time | 17:51 |
| notmyname | or 1000 | 17:51 |
| creiht | Kr4zy: it works fairly well for small objects, but as your objects get larger, it starts to fall apart | 17:51 |
| creiht | and what notmyname said :) | 17:51 |
| Kr4zy | Okay | 17:51 |
| Kr4zy | thanks | 17:51 |
| Kr4zy | Since HAProxy is still in dev, is it ready for production? Anyone using it in production? | 17:52 |
| Kr4zy | The latest 1.5 release | 17:52 |
| creiht | Kr4zy: yeah we (rackspace) have been using it in production for quite a while | 17:52 |
| creiht | almost a year I think | 17:52 |
| creiht | Kr4zy: of course, we are just some random dudes on the internet :) | 17:54 |
| Kr4zy | lol ;) | 17:54 |
| torgomatic | Kr4zy: also horrible timeouts; your client sends the last byte of a 5GB object, then nginx starts to send stuff to the proxy... this takes some time, during which your client decides the server is dead and times out the connection | 17:55 |
| creiht | the above was more in reference to the nginx comments | 17:55 |
| torgomatic | also also instead of spreading the IO load around all your big, fat, disk-stuffed object servers, it concentrates things on your poor /tmp fs on your proxies | 17:55 |
| creiht | yup | 17:55 |
| torgomatic | basically nothing good happens with nginx in front of swift | 17:56 |
| *** lpabon_ has joined #openstack-swift | 17:56 | |
| *** lpabon_ has quit IRC | 17:56 | |
| *** lpabon has quit IRC | 17:57 | |
| creiht | otherwise, nginx is really awesome :) | 17:57 |
| zaitcev | HAProxy does not terminate SSL, however, so it's only useful for large installations where you are already in hw termination zone. For small deployments I think Pound is still appropriate. | 17:58 |
| creiht | zaitcev: it does now | 17:59 |
| *** lpabon has joined #openstack-swift | 17:59 | |
| *** lpabon has quit IRC | 17:59 | |
| notmyname | zaitcev: and pound prevents the use of expect: 100-continue | 17:59 |
| *** haomaiwang has joined #openstack-swift | 18:02 | |
| *** lpabon has joined #openstack-swift | 18:02 | |
| openstackgerrit | Joel Wright proposed a change to openstack/python-swiftclient: Add importable SwiftService incorporating shell.py logic https://review.openstack.org/85453 | 18:09 |
| *** joeljwright has joined #openstack-swift | 18:16 | |
| *** Kbee has joined #openstack-swift | 18:38 | |
| *** zaitcev is now known as zaitcev_tooth | 18:38 | |
| Kbee | Hello, while doing a PUT, i'm getting 503 error - with the following logged in syslog in storage node | 18:38 |
| Kbee | ERROR __call__ error with PUT /sdb1/15/AUTH_2a5b67092df94b19a9e33d6dcd2a3ad7/cont : #012Traceback (most recent call last):#012 File "/usr/lib/python2.7/dist-packages/swift/account/server.py", line 284, in __call__#012 res = method(req)#012 File "/usr/lib/python2.7/dist-packages/swift/common/utils.py", line 2217, in wrapped#012 return func(*a, **kw)#012 File "/usr/lib/python2.7/dist-packages/swift/common/utils.py", line 837, in _timing_stat | 18:38 |
| Kbee | s#012 resp = func(ctrl, *args, **kwargs)#012 File "/usr/lib/python2.7/dist-packages/swift/account/server.py", line 125, in PUT#012 broker.put_container(container, req.headers['x-put-timestamp'],#012 File "/usr/lib/python2.7/dist-packages/swift/common/swob.py", line 219, in __getitem__#012 return self.environ[self._normalize(key)]#012KeyError: 'HTTP_X_PUT_TIMESTAMP' (txn: tx3975efcea54346fb9669b-00539f38c9) | 18:38 |
| Kbee | can someone help in rootcausing this one ? | 18:39 |
| Kr4zy | creiht: what about pound? What's wrong with it? | 18:39 |
| notmyname | Kr4zy: pound is ok, but it limits some of the HTTP functionality. there's really no reason to use pound now that HAProxy supports ssl termination and pipelined requests | 18:40 |
| creiht | Kr4zy: it is reasonbly capable for a lot of things, but you will have a hard time saturating 10Gb ethernet with ssl | 18:40 |
| *** kopparam has joined #openstack-swift | 18:40 | |
| *** nsquare has quit IRC | 18:41 | |
| notmyname | Kr4zy: and both pound and haproxy use openssl, so the engine is the same and you could expect similar performance from both as to ssl-termination | 18:41 |
| *** scohen has quit IRC | 18:41 | |
| notmyname | creiht: right? ^ | 18:41 |
| creiht | notmyname: that isn't clear to me | 18:41 |
| notmyname | ah | 18:41 |
| Kbee | sorry pasted content instead of gist- here is error message - http://paste.openstack.org/show/84200/ | 18:42 |
| creiht | I know our guys had a hard time getting pound to perform reasonably a while back | 18:42 |
| Kbee | While doing a PUT, i'm getting 503 error - with the following logged in syslog in storage node - error -> http://paste.openstack.org/show/84200/ | 18:42 |
| notmyname | Kbee: grep your whole cluster's logs for tx3975efcea54346fb9669b-00539f38c9 (the transaction id for that request) and pastbin that | 18:42 |
| *** scohen has joined #openstack-swift | 18:43 | |
| creiht | Kr4zy: but realistically the best thing to do would be to set up a cluster with each of the load balancers you would like to evaluate and benchmark them | 18:44 |
| *** scohen has quit IRC | 18:44 | |
| *** scohen has joined #openstack-swift | 18:44 | |
| Kbee | notmyname, pasted.. | 18:44 |
| Kr4zy | notmyname: and creiht, if I want to use HAProxy as dedicated LBs and it as SSL terminators on the proxy servers, would I do a SSL passthrough on the LBs? | 18:44 |
| creiht | Kr4zy: in theory that sounds right, though we do the ssl termination at the LB | 18:46 |
| *** joeljwright has quit IRC | 18:46 | |
| Kr4zy | creiht: I would too, but security requirement. | 18:46 |
| creiht | ahh | 18:46 |
| *** kopparam has quit IRC | 18:47 | |
| Kr4zy | creiht: Since there's a layer in front of the proxy service, how would I handle health checks on the HAProxy? | 18:48 |
| notmyname | Kr4zy: note that swift assumes a secure, private network within the cluster (ie proxy->storage communication or replication traffic isn't encrypted) | 18:48 |
| notmyname | Kr4zy: you could do simple IP:port connect checks or you could still use the /healthcheck endpoint for each proxy | 18:48 |
| creiht | Kr4zy: also, I'm just a dev that is parroting most of what I've heard our ops guys say over the years :) | 18:50 |
| Kbee | notmyname: http://paste.openstack.org/show/84201/ | 18:50 |
| notmyname | creiht: lol. ditto :-) | 18:51 |
| Kr4zy | notmyname: right, our setup does put the Swift cluster in a private network. We encrypt only the communication between the proxies and the Swift clients. proxy -> storage is not encrypted. Security wants user credentials to be encrypted. For us, it is the Keystone token. We also configured the proxy services to listen on localhost and have only the SSL terminators publicly accessible. So we can't really run a health checks to t | 18:52 |
| Kr4zy | notmyname: at least I don't know how. | 18:52 |
| Kr4zy | notmyname: I am new to HAProxy btw. | 18:52 |
| notmyname | Kr4zy: I too am new to HAProxy configs. but looking at the online docs it seems pretty straightforward. honestly all I can do is point you to docs someone else has written | 18:53 |
| *** lpabon has quit IRC | 18:54 | |
| creiht | Kr4zy: also I think in a setup with ssl passthrough you lose the client source ip | 18:55 |
| *** nsquare has joined #openstack-swift | 18:55 | |
| creiht | Kbee: sorry, we still need more context to try to help | 18:56 |
| notmyname | creiht: but can't you tell haproxy to add the x-forwarded-for header with the client ip? | 18:56 |
| Kr4zy | notmyname: I currently have health checks to the proxy ssl terminators which works if I turn off the ssl terminator. The client will failover to the other HAProxy LB. But if I don't turn off the SSL terminator, but turns off the swift proxy service, HAProxy will still thinks the service is online. Clients connecting to it will hang and eventually fail. | 18:56 |
| creiht | notmyname: ahh true | 18:57 |
| Kbee | creiht: what other information is needed ? | 18:58 |
| creiht | Kbee: it looks like that is just the logs for the one server | 18:58 |
| creiht | can you grep the logs on all of your servers (proxy, account, container, object) | 18:59 |
| *** mlipchuk has joined #openstack-swift | 18:59 | |
| creiht | for that txid, that helps us to understand the request that was going through | 18:59 |
| notmyname | Kr4zy: ya, I think the setup you describe loses the backpressure feature you get with healthcheck endpoints. that is, you've got the ssl terminator in the middle of the data path, so if you can't check the proxy service directly then you can't make intelligent decisions at the LB layer | 18:59 |
| * notmyname grabs the popcorn and watches the mailing list thread about the openstack-ci gates throwing too many errors and running too many tests | 19:01 | |
| portante | clayg, acoles, notmyname, torgomatic: so now the end of the chain for SP is 96026, and the beginning is 96027? | 19:01 |
| torgomatic | portante: sounds about right | 19:01 |
| notmyname | portante: current end point is in the topic | 19:01 |
| Kr4zy | notmyname: yeah, any idea on going around this? btw, what are you guys using for identify service | 19:01 |
| notmyname | portante: wait. that doesn't sound right | 19:01 |
| portante | notmyname: does not look right | 19:01 |
| notmyname | looks like I fixed the topic with an old version of it | 19:02 |
| * notmyname goes hunting | 19:02 | |
| acoles | portante: notmyname: it is that way | 19:02 |
| portante | acoles, torgomatic: thanks for confirming | 19:02 |
| *** nsquare has quit IRC | 19:02 | |
| notmyname | ah yes, https://review.openstack.org/#/c/96026/ is now the very end (the docs) | 19:03 |
| *** ChanServ changes topic to "Storage Policies have been proposed to master. Soft freeze in effect until these patches land | First SP patch: https://review.openstack.org/#/c/96027/ | Last SP patch: https://review.openstack.org/#/c/96026/" | 19:03 | |
| notmyname | there | 19:03 |
| Kbee | creiht, notmyname - updated with all logs - http://paste.openstack.org/show/84203/ | 19:04 |
| creiht | notmyname: but I thought nothing was broken, it is the way things *must* be done :) | 19:04 |
| Kr4zy | notmyname: also, there aren't any concern about having the user credentials being pass in plain text between the LBs and the proxy servers in a private network? Does this comply with the Openstack security guide? | 19:04 |
| * notmyname invites creiht to share his popcorn | 19:04 | |
| creiht | notmyname: heh | 19:05 |
| creiht | I haven't read that thread since last week | 19:05 |
| acoles | portante: also, 96031 has morphed to 99979, but the content is pretty similar | 19:06 |
| creiht | Kbee: hrm, do you know what version of swift you are running? | 19:06 |
| *** nsquare has joined #openstack-swift | 19:06 | |
| portante | acoles: thanks | 19:07 |
| *** mmcardle has joined #openstack-swift | 19:07 | |
| notmyname | Kbee: creiht: that is a really weird error. why would the timestamp not be in the environ? | 19:07 |
| torgomatic | notmyname: well, if race conditions are slipping in due to random chance, then clearly what we need is a deterministic race-condition detector | 19:07 |
| creiht | notmyname: yeah that's what I'm wondering | 19:07 |
| notmyname | torgomatic: does it have to also be correct? I can deterministically say "yes, there is a race" for every question ;-) | 19:08 |
| torgomatic | notmyname: might help if it were correct ;) | 19:08 |
| Kbee | creiht, notmyname swift 2.0.3, icehouse bits ubuntu | 19:08 |
| notmyname | Kbee: that's the client version. not the server version | 19:09 |
| notmyname | Kbee: I guess "ubuntu icehouse" is the server-side | 19:09 |
| notmyname | I guess they packaged 1.13.1 correctly? | 19:09 |
| Kbee | notmyname, correct.. | 19:09 |
| notmyname | Kbee: can you do `curl -i <swift domain>/info`? that should have the swift version number in it | 19:10 |
| Kbee | notmyname, whats <swift domain> ?? | 19:12 |
| notmyname | Kbee: whatever hostname/IP you're using to talk to your swift cluster. looks like you are using "192.168.33.20:8080". I didn't put that in earlier because I was getting a temporary error from the pastbin | 19:13 |
| * notmyname goes to eat lunch | 19:14 | |
| *** scohen has quit IRC | 19:14 | |
| *** scohen has joined #openstack-swift | 19:14 | |
| Kbee | notmyname, "version": "1.13.1" | 19:15 |
| *** mmcardle has quit IRC | 19:18 | |
| *** mmcardle has joined #openstack-swift | 19:18 | |
| *** joeljwright has joined #openstack-swift | 19:23 | |
| *** kevinc_ has quit IRC | 19:24 | |
| *** leews has quit IRC | 19:25 | |
| *** miqui has quit IRC | 19:25 | |
| *** miqui has joined #openstack-swift | 19:26 | |
| *** joeljwright has quit IRC | 19:26 | |
| *** joeljwright has joined #openstack-swift | 19:29 | |
| *** Kbee has quit IRC | 19:36 | |
| *** Kbee has joined #openstack-swift | 19:36 | |
| Kbee | notmyname: "version": "1.13.1" | 19:45 |
| *** kopparam has joined #openstack-swift | 19:45 | |
| *** itarchitectkev_ has quit IRC | 19:46 | |
| *** annegentle has quit IRC | 19:48 | |
| *** kopparam has quit IRC | 19:50 | |
| *** haomaiwang has quit IRC | 19:52 | |
| notmyname | /mack | 19:56 |
| notmyname | or "back", as the case may be | 19:57 |
| notmyname | my 4 year-old just invented the word "frizzed" as in the past-tense of throwing something like a frisbee. eg "yesterday, I frizzed the plate across the room" | 19:59 |
| *** kevinc_ has joined #openstack-swift | 20:01 | |
| *** mkollaro has quit IRC | 20:01 | |
| portante | cool | 20:02 |
| Kbee | notmyname: "version": "1.13.1" | 20:03 |
| *** jamiehannaford has joined #openstack-swift | 20:13 | |
| portante | clayg, peluse_, others: a gist to bring coverage of storage policies module to 100%: https://gist.github.com/portante/6a7aecc635731f1a905b | 20:13 |
| portante | I can submit that after this lands, if you want. | 20:13 |
| notmyname | cool | 20:14 |
| *** scohen_ has joined #openstack-swift | 20:14 | |
| *** scohen has quit IRC | 20:14 | |
| *** Midnightmyth has quit IRC | 20:14 | |
| Kbee | notmyname, any leads ? | 20:14 |
| *** annegentle_ has joined #openstack-swift | 20:16 | |
| *** mmcardle has quit IRC | 20:17 | |
| *** mlipchuk has quit IRC | 20:20 | |
| peluse_ | portante: awesome, thanks! | 20:23 |
| *** zaitcev_tooth is now known as zaitcev | 20:23 | |
| *** Kbee has quit IRC | 20:26 | |
| *** CaioBrentano has joined #openstack-swift | 20:27 | |
| *** mmcardle has joined #openstack-swift | 20:34 | |
| notmyname | ttx: I'd like to move our 1:1 earlier this week. we're normally scheduled at 8:30 (pacific). I'd like to do it sometime between 6am and 8am | 20:37 |
| *** scohen_ has quit IRC | 20:37 | |
| *** foexle has joined #openstack-swift | 20:43 | |
| *** annegentle_ has quit IRC | 20:47 | |
| *** Midnightmyth has joined #openstack-swift | 20:49 | |
| notmyname | and I just realized that was supposed to go into the IRC window just above this one. | 20:52 |
| portante | ;) | 20:53 |
| peluse_ | i figured it was the drugs still | 20:54 |
| notmyname | I wonder how long I'll be able to blame stuff on that ;-) | 20:55 |
| peluse_ | milk it for as long as you can :) | 20:58 |
| *** mmcardle1 has joined #openstack-swift | 21:08 | |
| *** joeljwright has quit IRC | 21:08 | |
| *** joeljwright has joined #openstack-swift | 21:08 | |
| *** mmcardle has quit IRC | 21:10 | |
| *** foexle has quit IRC | 21:10 | |
| *** Trixboxer has quit IRC | 21:13 | |
| *** jamiehannaford has quit IRC | 21:16 | |
| notmyname | looks like there's been lots of reviews on the current SP patches. but none yet really from RAX. creiht, dfg, glange, redbo, [gholt] | 21:26 |
| notmyname | aside from a few small things, the parts that have been reviewed are trending towards "this looks good", from what I can tell | 21:26 |
| *** Midnightmyth has quit IRC | 21:35 | |
| *** kevinc_ has quit IRC | 21:40 | |
| *** kevinc_ has joined #openstack-swift | 21:43 | |
| *** dmsimard is now known as dmsimard_away | 21:48 | |
| *** kopparam has joined #openstack-swift | 21:49 | |
| scotticus | notmyname: really? | 21:49 |
| scotticus | didn' | 21:49 |
| scotticus | didn't redbo and creiht go to the hackathon? | 21:49 |
| scotticus | and gholt has been here thoroughly going over the patches | 21:49 |
| notmyname | scotticus: yes. I only mean on the reviews in gerrit. they were both there and provided valuable feedback and insight | 21:49 |
| scotticus | dfg is on vacation... | 21:50 |
| scotticus | they aren't done yet. | 21:50 |
| notmyname | scotticus: I'm not saying y'all aren't doing anything. just that the current gerrit patches don't have any rackspace comments on them. | 21:50 |
| scotticus | gholt is still reviewing, afaik everyone else is too. we're doing it :) | 21:51 |
| notmyname | awesome! :-) | 21:51 |
| *** kopparam has quit IRC | 21:54 | |
| *** joeljwright has quit IRC | 21:59 | |
| *** foexle has joined #openstack-swift | 22:01 | |
| creiht | notmyname: yeah I'm mostly done... I know gholt was still poking at it | 22:28 |
| notmyname | creiht: cool, thanks | 22:29 |
| creiht | though I'm mostly done due to attrition | 22:35 |
| *** jergerber has quit IRC | 22:35 | |
| notmyname | creiht: you're not the first person I've heard that from ;-) | 22:36 |
| creiht | hehe | 22:36 |
| *** dmsimard_away is now known as dmsimard | 22:49 | |
| *** Kr4zy has quit IRC | 22:53 | |
| mattoliverau | Morning all! | 22:56 |
| notmyname | mattoliverau: hi! | 22:56 |
| mattoliverau | notmyname: how you going? recovering well? | 22:57 |
| notmyname | mattoliverau: I'm alive. recovering pretty well I think :-) | 22:57 |
| *** mkollaro has joined #openstack-swift | 23:10 | |
| *** dmsimard is now known as dmsimard_away | 23:11 | |
| goodes | How critical is it to use one email address of all of openstack work? Does it screw things up if you start changing emails (i.e. jobs)? Would it be best just to use your personal standard github email? | 23:12 |
| torgomatic | goodes: that's what .mailmap is for | 23:12 |
| goodes | torgomatic: thanks | 23:14 |
| mattoliverau | notmyname: being alive is a good start :P Glad to hear recovery is going well, if there anything you need then don't hesitate to ask.. my distance might be a hindrence, tho. But if it's Australian Tim Tams in the mail then consider it done ;) You might not know this, but in Atlanta when you got a cold/flu pack to Mikal, they were passed on to me later in the week when I was hit with the flu.. so | 23:18 |
| mattoliverau | thank you!! | 23:18 |
| notmyname | mattoliverau: nice! | 23:18 |
| *** tdasilva has left #openstack-swift | 23:19 | |
| *** acoles has quit IRC | 23:20 | |
| *** mkollaro has quit IRC | 23:24 | |
| *** acoles has joined #openstack-swift | 23:44 | |
| *** ChanServ sets mode: +v acoles | 23:44 | |
| *** mwstorer has quit IRC | 23:48 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!