| *** andrewbogott has quit IRC | 00:01 | |
| *** andrewbogott has joined #openstack-dev | 00:16 | |
| *** anotherjesse is now known as anotherjesse_zz | 00:17 | |
| *** anotherjesse_zz is now known as anotherjesse | 00:20 | |
| *** anotherjesse is now known as anotherjesse_zz | 00:30 | |
| *** ncode has joined #openstack-dev | 00:31 | |
| *** ncode has joined #openstack-dev | 00:31 | |
| *** shevek_ has quit IRC | 00:33 | |
| *** jakedahn_zz is now known as jakedahn | 00:38 | |
| *** andrewbogott has quit IRC | 00:46 | |
| *** pvo-away is now known as pvo | 00:51 | |
| *** ncode has joined #openstack-dev | 01:01 | |
| *** dtroyer is now known as dtroyer_zzz | 01:02 | |
| *** ncode has quit IRC | 01:10 | |
| *** dolphm has joined #openstack-dev | 01:41 | |
| *** novas0x2a|laptop has quit IRC | 01:53 | |
| *** martine has joined #openstack-dev | 02:27 | |
| *** danwent has quit IRC | 02:39 | |
| *** ywu has quit IRC | 02:52 | |
| *** novas0x2a|laptop has joined #openstack-dev | 03:05 | |
| *** sdake has quit IRC | 03:06 | |
| *** dolphm has quit IRC | 03:07 | |
| *** flaviamissi_ has joined #openstack-dev | 03:15 | |
| *** flaviamissi has quit IRC | 03:18 | |
| *** flaviamissi_ has quit IRC | 03:19 | |
| *** Mandell has joined #openstack-dev | 03:24 | |
| *** devcamcar has quit IRC | 04:02 | |
| *** kpepple_ has quit IRC | 04:02 | |
| *** openstackgerrit has quit IRC | 04:02 | |
| *** devcamcar has joined #openstack-dev | 04:06 | |
| *** kpepple_ has joined #openstack-dev | 04:06 | |
| *** openstackgerrit has joined #openstack-dev | 04:06 | |
| *** danwent has joined #openstack-dev | 04:14 | |
| *** asalkeld has joined #openstack-dev | 04:23 | |
| *** asalkeld has quit IRC | 04:24 | |
| *** asalkeld has joined #openstack-dev | 04:24 | |
| *** martine has quit IRC | 04:36 | |
| *** ncode has joined #openstack-dev | 04:37 | |
| *** asalkeld has quit IRC | 04:37 | |
| *** anotherjesse_zz is now known as anotherjesse | 04:38 | |
| *** andrewsmedina has quit IRC | 04:48 | |
| *** andrewsmedina has joined #openstack-dev | 04:52 | |
| *** andrewsmedina has quit IRC | 04:59 | |
| *** mjfork has quit IRC | 05:07 | |
| *** anotherjesse is now known as anotherjesse_zz | 05:08 | |
| *** zigo has joined #openstack-dev | 05:13 | |
| *** ncode has quit IRC | 05:13 | |
| *** dolphm has joined #openstack-dev | 05:30 | |
| *** Mandell has quit IRC | 05:31 | |
| *** hattwick has quit IRC | 05:32 | |
| *** wwkeyboard has quit IRC | 05:32 | |
| *** thingee has joined #openstack-dev | 05:34 | |
| bcwaldon | jeblair: when you get a chance, glance unit tests are failing due to too many links in /tmp | 05:35 |
|---|---|---|
| *** Mandell has joined #openstack-dev | 05:46 | |
| *** zaitcev has quit IRC | 05:46 | |
| *** dolphm has quit IRC | 05:50 | |
| *** dolphm has joined #openstack-dev | 05:52 | |
| *** dtroyer_zzz is now known as dtroyer | 06:02 | |
| *** dtroyer is now known as dtroyer_zzz | 06:09 | |
| *** Shrews has joined #openstack-dev | 06:14 | |
| *** dolphm has quit IRC | 06:16 | |
| *** dolphm has joined #openstack-dev | 06:24 | |
| *** danwent has quit IRC | 06:28 | |
| *** littleidea has quit IRC | 06:52 | |
| *** Shrews has quit IRC | 06:55 | |
| *** Mandell has quit IRC | 07:08 | |
| *** dolphm has quit IRC | 07:29 | |
| *** dolphm has joined #openstack-dev | 07:52 | |
| *** hattwick has joined #openstack-dev | 07:59 | |
| *** dachary has joined #openstack-dev | 08:03 | |
| *** thingee is now known as thingee_zzz | 08:08 | |
| *** roge has joined #openstack-dev | 08:15 | |
| *** dolphm has quit IRC | 08:17 | |
| *** roge has quit IRC | 09:16 | |
| *** shevek_ has joined #openstack-dev | 10:05 | |
| *** martine has joined #openstack-dev | 10:33 | |
| *** martine has quit IRC | 10:38 | |
| *** berendt has joined #openstack-dev | 11:06 | |
| *** Ryan_Lane has quit IRC | 11:38 | |
| *** benner has quit IRC | 12:15 | |
| *** eglynn_ has quit IRC | 12:16 | |
| *** flaviamissi has joined #openstack-dev | 12:16 | |
| *** eglynn_ has joined #openstack-dev | 12:18 | |
| *** benner has joined #openstack-dev | 12:22 | |
| notmyname | justinsb: ping | 13:07 |
| *** justinsb_ has joined #openstack-dev | 13:13 | |
| *** termie_ has joined #openstack-dev | 13:13 | |
| *** benner_ has joined #openstack-dev | 13:13 | |
| *** yamahata_ has joined #openstack-dev | 13:13 | |
| *** Kiall_ has joined #openstack-dev | 13:17 | |
| *** benner has quit IRC | 13:19 | |
| *** dachary has quit IRC | 13:19 | |
| *** Kiall has quit IRC | 13:19 | |
| *** justinsb has quit IRC | 13:19 | |
| *** termie has quit IRC | 13:19 | |
| *** villep has quit IRC | 13:19 | |
| *** fesnel has quit IRC | 13:19 | |
| *** yamahata has quit IRC | 13:19 | |
| *** Kiall_ is now known as Kiall | 13:19 | |
| *** villep has joined #openstack-dev | 13:20 | |
| *** dachary has joined #openstack-dev | 13:20 | |
| *** fesnel has joined #openstack-dev | 13:20 | |
| *** mjfork has joined #openstack-dev | 13:27 | |
| *** ncode has joined #openstack-dev | 13:52 | |
| *** roge has joined #openstack-dev | 14:00 | |
| *** dolphm has joined #openstack-dev | 14:08 | |
| *** ywu has joined #openstack-dev | 14:09 | |
| *** wwkeyboard has joined #openstack-dev | 14:23 | |
| *** dolphm has quit IRC | 14:26 | |
| *** dolphm has joined #openstack-dev | 14:28 | |
| *** dolphm has quit IRC | 14:32 | |
| *** sdake has joined #openstack-dev | 14:40 | |
| *** ncode has quit IRC | 14:43 | |
| *** ywu has quit IRC | 14:47 | |
| *** dolphm has joined #openstack-dev | 14:49 | |
| *** eglynn__ has joined #openstack-dev | 14:56 | |
| *** eglynn_ has quit IRC | 14:56 | |
| *** roge has quit IRC | 15:08 | |
| *** ywu has joined #openstack-dev | 15:34 | |
| jeblair | bcwaldon: thanks. cleaned them up. devananda has a fix in progress for that. | 15:45 |
| *** dolphm has quit IRC | 15:52 | |
| *** dolphm_ has joined #openstack-dev | 15:55 | |
| *** Mandell has joined #openstack-dev | 16:02 | |
| *** dolphm_ has quit IRC | 16:09 | |
| *** roge has joined #openstack-dev | 16:11 | |
| *** dolphm has joined #openstack-dev | 16:20 | |
| *** dolphm has quit IRC | 16:24 | |
| *** danwent has joined #openstack-dev | 16:25 | |
| *** roge has quit IRC | 16:27 | |
| *** danwent has quit IRC | 16:27 | |
| *** dneary has joined #openstack-dev | 16:29 | |
| *** dneary has quit IRC | 16:29 | |
| *** dneary has joined #openstack-dev | 16:29 | |
| *** rgoodwin is now known as rgoodwin_away | 16:44 | |
| *** rgoodwin_away is now known as rgoodwin | 16:46 | |
| *** Mandell has quit IRC | 16:52 | |
| *** byeager has quit IRC | 17:15 | |
| *** akshayms_ has joined #openstack-dev | 17:20 | |
| *** dolphm has joined #openstack-dev | 17:22 | |
| *** ywu has quit IRC | 17:22 | |
| *** ywu has joined #openstack-dev | 17:25 | |
| *** dneary has quit IRC | 17:35 | |
| *** btorch_ has joined #openstack-dev | 17:42 | |
| eglynn__ | jeblair: still around? | 17:47 |
| eglynn__ | jeblair: could you re-enable GLANCE_TEST_S3_CONF & GLANCE_TEST_SWIFT_CONF for the glance builds in Jenkins? | 17:47 |
| eglynn__ | (now that https://review.openstack.org/5466 has landed) | 17:47 |
| *** andrewsmedina has joined #openstack-dev | 17:47 | |
| eglynn__ | jaypipes / bcwaldon: https://review.openstack.org/5487 should address the temp dir leakage that was borking /tmp on the Jenkins slaves | 17:48 |
| *** dubsquared has quit IRC | 17:54 | |
| *** Ryan_Lane has joined #openstack-dev | 18:00 | |
| *** roge has joined #openstack-dev | 18:01 | |
| *** dolphm has quit IRC | 18:12 | |
| *** akshayms_ has quit IRC | 18:14 | |
| *** ywu has quit IRC | 18:17 | |
| *** jakedahn is now known as jakedahn_zz | 18:18 | |
| *** dolphm has joined #openstack-dev | 18:18 | |
| *** novas0x2a|laptop has quit IRC | 18:28 | |
| *** roge has quit IRC | 18:30 | |
| *** novas0x2a|laptop has joined #openstack-dev | 18:31 | |
| *** ncode has joined #openstack-dev | 18:31 | |
| *** ncode has joined #openstack-dev | 18:31 | |
| *** novas0x2a|laptop has quit IRC | 18:37 | |
| *** novas0x2a|laptop has joined #openstack-dev | 18:40 | |
| *** danwent has joined #openstack-dev | 19:41 | |
| *** dolphm has quit IRC | 19:47 | |
| *** novas0x2a|laptop has quit IRC | 19:53 | |
| *** novas0x2a|laptop has joined #openstack-dev | 19:56 | |
| *** ywu has joined #openstack-dev | 20:03 | |
| *** roge has joined #openstack-dev | 20:18 | |
| *** roge has quit IRC | 20:25 | |
| *** dolphm has joined #openstack-dev | 20:46 | |
| *** dolphm has quit IRC | 20:49 | |
| *** asalkeld has joined #openstack-dev | 21:09 | |
| *** roge has joined #openstack-dev | 21:40 | |
| *** zigo has quit IRC | 21:43 | |
| *** roge has quit IRC | 21:44 | |
| *** anotherjesse_zz is now known as anotherjesse | 22:48 | |
| *** ncode has quit IRC | 22:55 | |
| justinsb_ | notmyname: (Very delayed) ack | 23:12 |
| *** Mandell has joined #openstack-dev | 23:19 | |
| notmyname | justinsb_: still here? | 23:21 |
| notmyname | justinsb_: I haven't looked at your code yet, but I wanted to run a related idea by you | 23:23 |
| notmyname | (the swift ring changes) | 23:23 |
| notmyname | justinsb_: if I don't catch you tonight, ping me sometime tomorrow | 23:29 |
| *** hazmat has quit IRC | 23:33 | |
| justinsb_ | notmyname: Stepped away for a minute, but I'm back! | 23:37 |
| *** martine has joined #openstack-dev | 23:37 | |
| justinsb_ | notmyname: I'm working on figuring out how to actually test changing the # replicas | 23:39 |
| notmyname | justinsb_: have you heard of the idea to make replica placement in the zones "as unique as possible"? | 23:41 |
| justinsb_ | notmyname: I think chmouel pointed me to it, but I didn't really see how to apply it | 23:42 |
| notmyname | justinsb_: the idea is to keep N replicas (because you really do want that), but you can have as few as one zone. you then grow your cluster by adding zones (not replicas) | 23:43 |
| notmyname | right now, swift places each replica in a unique zone and requires that you have at least as many zones as replicas. | 23:44 |
| notmyname | this works fine from the medium to large case, but isn't very elegant for smaller clusters | 23:44 |
| notmyname | (none of this should be taken as me opposed to exploring dynamic replica counts) | 23:45 |
| *** hazmat has joined #openstack-dev | 23:45 | |
| notmyname | justinsb_: I'm curious if you think that sort of change (the as unique as possible one) would work in your use case | 23:45 |
| justinsb_ | notmyname: What I'd like to solve is the packaging problem - I'm trying to figure how we install the first server and then grow nicely to a properly redundant system | 23:45 |
| justinsb_ | notmyname: It sounds like I'd still have to install multiple swift instances on my first box | 23:45 |
| notmyname | no, you wouldn't | 23:45 |
| notmyname | well, perhaps multiple processes, but just one of each (proxy, account, etc) | 23:46 |
| justinsb_ | notmyname: So it would only store it once if there was only one zone? | 23:47 |
| notmyname | what would happen is that the 3 replicas would go to the same zone/box/process | 23:47 |
| notmyname | all 3 replicas would go to the first zone | 23:47 |
| notmyname | of course, at that point the replicas aren't really doing much for you except taking up extra space | 23:47 |
| notmyname | but it's a starting point for a larger cluster | 23:48 |
| justinsb_ | notmyname: But as I added more machines it would rebalance, I presume | 23:48 |
| notmyname | ya | 23:48 |
| justinsb_ | notmyname: That does solve the packaging issue then | 23:48 |
| *** deshantm has joined #openstack-dev | 23:48 | |
| notmyname | and if you aren't planning on growing past the capabilities of one box, perhaps swift isn't the right storage solution (just use RAID and NFS or something similar) | 23:48 |
| justinsb_ | notmyname: I'm coming at this from the point of view of e.g. what should apt-get do. | 23:49 |
| justinsb_ | notmyname: I think you should be able to apt-get install swift and get a single node install that works | 23:49 |
| notmyname | http://joearnold.com/swift-in-the-small/ | 23:49 |
| notmyname | exampel of how to do small clusters with the current design ^ | 23:50 |
| justinsb_ | notmyname: But then you should also be able to grow it from that point to a full swift install | 23:50 |
| *** martine has quit IRC | 23:50 | |
| justinsb_ | notmyname: I definitely agree that swift on one node isn't the best example of swift in action | 23:50 |
| justinsb_ | notmyname: But it is nice to be able to use the same APIs etc | 23:51 |
| notmyname | justinsb_: I don't really buy the argument that packaging should dictate design changes. swift is intended to be used in a distributed system. apt-get on one box doesn't really make sense | 23:51 |
| notmyname | I do like the argument of starting from a small cluster and growing from there, though | 23:51 |
| justinsb_ | I just want to see it being used, and a complicated installation process is a big barrier to that. | 23:52 |
| justinsb_ | Even if we're not using the full capabilities on one box, I think it should be possible to get something up and running on one box | 23:52 |
| notmyname | I'm a big fan of lowering barriers to entry :-) | 23:52 |
| justinsb_ | Both incrementing the replica count and the 'as unique as possible' approach seem good to me | 23:53 |
| notmyname | I don't like changing replica counts as much because it doesn't make sense in existing clusters of most any size | 23:53 |
| Lumiere | packaging is meant as a configuration management tool as much as anything else | 23:54 |
| notmyname | justinsb_: the replication bandwidth needed would be tremendous (I haven't read your code to see how you handle this yet) | 23:54 |
| justinsb_ | notmyname: I can certainly see that you don't want to change replica count too often | 23:55 |
| justinsb_ | notmyname: I haven't even tried it yet. I just reconfigure the ring and rely on the repair to fix things. I'm working on trying it now and seeing where it breaks! | 23:55 |
| justinsb_ | notmyname: The 'as unique as possible' isn't going to be available for essex though, right? | 23:56 |
| notmyname | correct | 23:56 |
| notmyname | it's been attempted as a side project by one of the swift devs, but it hasn't been integrated into swift yet. | 23:57 |
| notmyname | I'd give it better than 50% odds of being in folsom, though | 23:58 |
| justinsb_ | notmyname: So are we suggesting that swift be packaged? You can apt-get it, but it won't work unless you have at least 3 machines and configure them together? | 23:58 |
| justinsb_ | notmyname: Oops... "so how are we" ... | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!