22:00:32 #startmeeting reddwarf 22:00:33 Meeting started Tue Feb 5 22:00:32 2013 UTC. The chair is hub_cap. Information about MeetBot at http://wiki.debian.org/MeetBot. 22:00:35 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 22:00:37 The meeting name has been set to 'reddwarf' 22:00:37 PARTAY time! 22:00:44 lol 22:00:47 #link 22:00:50 #link http://wiki.openstack.org/Meetings/RedDwarfMeeting 22:00:55 #link http://eavesdrop.openstack.org/meetings/reddwarf/2013/reddwarf.2013-01-29-21.59.html 22:01:02 so lets start w/ the action items 22:01:05 #topic action items 22:01:14 hello 22:01:34 i have forwarded joes email to vipul, so he can distribute to the team. and we will chat about the api spec later as its on the list 22:01:43 got it 22:01:46 so has _everyone_ added content to the wiki? 22:01:46 thanks! 22:01:48 np dude 22:01:52 sounds good. 22:01:58 It's been growing. 22:02:00 yup, checkout the wiki wiki.openstack.org/Reddwarf 22:02:04 The Wiki, I mean... 22:02:13 looks great whoever wrote that 22:02:21 #link http://wiki.openstack.org/Reddwarf 22:02:40 very nice team 22:02:45 #kudos 22:02:52 oh wait thats not a action 22:02:59 haha, we need to make it one. 22:03:03 SlickNik: youre up! 22:03:10 #agreed good job on wiki 22:03:11 looks like u did doc it 22:03:17 hence the nice pretty wiki 22:03:28 Yup, added the install and testing info to the wiki 22:03:28 can u link to your devstack/redstack stuff on the wiki 22:03:41 #link http://wiki.openstack.org/reddwarf-installation 22:03:43 we should add a pretty picture of the arch to the doc 22:03:51 thx SlickNik 22:03:52 #link http://wiki.openstack.org/reddwarf-redstack-testing 22:04:07 ok lets skip the kaganos bit since there is a percona topic 22:04:09 sound good? 22:04:10 Still a work in progress, so will keep refining it 22:04:26 we can talk about percona in a few 22:04:48 sounds good, kaganos is here too, I think... 22:04:50 cool 22:04:52 grapex: link your sqlite BP 22:05:08 https://blueprints.launchpad.net/reddwarf/+spec/in-memory-sqlite 22:05:28 #link https://blueprints.launchpad.net/reddwarf/+spec/in-memory-sqlite 22:05:35 thx sir 22:05:58 dj was supposed to give an us update but hes playing ping pong 22:06:00 :P 22:06:11 ill add repose to the topics 22:06:24 as a sub of the quotas/limits 22:06:50 ok so root remove 22:06:54 lets do it. 22:07:00 i'm here 22:07:02 heh.. 22:07:02 its valid, and we agree it should be in the api 22:07:16 i laugh cuz i got some pushback from intrenally as well 22:07:25 #info disable/delete root is good to have in the api 22:07:26 LOL 22:07:42 but if we wanna move forward.. let's leave the Blueprint in 22:07:42 so then we shouldnt have it? maybe we should discuss offline vipul :) 22:07:42 kaganos: we are going to talk about the percona topic in a little bit, so stick around :) 22:07:49 yes... 22:07:52 kk 22:07:53 needs further discussion 22:08:01 ok the next one... i cant type 22:08:07 so we will never learn 22:08:09 doh 22:08:11 err i will never learn 22:08:11 see 22:08:18 djohnstone: u have a subtopic in quotas stick around 22:08:23 ok 22:08:41 vipul: BP on additional roles 22:08:57 was going to file done but didn't find an empty dummy 22:08:59 did u get that squared away? the whole policy thing 22:09:00 #action vipul to file BP on additional roles in reddwarf (user, superuser, admin) 22:09:05 vipul: ill create u one now 22:09:15 Yep, it will be based on policies 22:09:20 thanks hub_cap 22:09:44 ok well done team 22:09:51 10 min, all of the action items done 22:09:52 WOOT 22:09:52 sweet. 22:09:58 that's some sort of a record 22:09:59 Keep movin' :) 22:10:02 yup it is 22:10:06 awesome! 22:10:11 way what? did we talk about quotas? 22:10:11 #topic Quotas / Limits Updates 22:10:11 yeah, I think it is. 22:10:21 We're talking about it now. 22:10:21 steveleon: lol we arent done w/ teh whole meeting ;) 22:10:40 ok do tell. then djohnstone will talk about repose 22:11:19 we have decided to implement the quota feature... and adapt the nova's implementation of it 22:11:21 https://github.com/openstack/nova/blob/master/nova/quota.py 22:11:37 the quota feature will be done as a filter/middleware... 22:11:42 #link https://github.com/openstack/nova/blob/master/nova/quota.py 22:12:01 so that rax can disable it and use their own (repose) 22:12:11 steveleon.. was going to mention this morning.. filter approach may not be complete correct 22:12:17 since the instance is actually created by taskmanager 22:12:20 is async 22:12:21 by doing it as a filter/middleware, we dont touch business logic 22:13:05 reservation will be made prior the action is performed and commit (or rollback) will be done after 22:13:28 yes vipul... im aware 22:13:50 and if it breaks in the taskmgr... then its broken (i think servers is the same way WRT quotas) 22:13:51 steveleon: with the being separate from the business logic this won't be able to assign different quota's to different customers? Is that correct? 22:14:57 ok cool, if we have basically guarantee that it will be created 22:15:18 we can assigned quota based on a tenant_id.... or a class 22:15:26 at least that is how nova does it 22:15:39 cool 22:15:40 steveleon: k 22:15:40 a class quota can be applied to a set of customers.. 22:15:59 class: api_abusers :P 22:16:16 vipul: i was thinking that... if the reservation is made and committed and the taskmanager comes back as error. The resources will still be committed 22:16:24 class: people_who_cant_type :P 22:16:30 DUDE THATS ME 22:16:32 so the user will have to remove the instance (for example) so that the resouces are released 22:16:47 ya thats no diff from nova... if a server fails launch isint that ths same functionality? steveleon vipul 22:16:48 hub_cap never learned 22:16:52 nope... 22:16:54 failed his action item... 22:17:04 oh wait we are not there yet 22:17:18 hub_cap yea sounds ok now that you mention how it's implemented in onva 22:17:21 nova 22:17:29 word. 22:17:47 thas great. have u started working on it steveleon? 22:17:53 or still analyzing so to speak 22:17:56 so it seems that we might not even use rollback 22:18:06 itll be nice to have that in the app 22:18:17 #info steveleon implementing quota feature based on nova's quota implementation 22:18:36 yes. hub_cap. vipul put me up to it :P 22:18:47 hehe nice :) 22:18:48 do it ! :) 22:18:50 good man vipul 22:19:09 ok we ready to move to repose? 22:19:11 word 22:19:29 sorry in a meeting concurrently.. may be a bit slow today 22:19:29 sounds good, lay out the plan djohnstone... 22:19:36 vipul: daz cool 22:19:44 K. So I got repose installed and configured on a local dev VM. 22:20:04 It seems kinda heavy on the configuration side. Lots of config files to tweak. 22:20:21 right, and the sun is "kind of" bright at noon 22:20:27 ha 22:20:30 (heh) 22:20:56 lol datsun180b 22:20:56 rate limiting is really easy to get going. and can be done on URI level for GET/POST... 22:21:29 I'm just starting to look at the Quota's part, so I haven't delved into that too deep. 22:21:44 datsun180b: sun is not usually a problem in Seattle :) 22:22:05 esp1: Today it is.. I had to close the blinds 22:22:10 cool lets have a update next meeting too djohnstone 22:22:11 datsun180b: There's a lot of config files, but when you dig into Repose it makes sense why (kind of like how in Java there are dozens more files than simialrly sized Python projects). 22:22:15 lol 22:22:24 #action djohnstone to update us with Quotas in repose 22:22:25 first world problems 22:22:25 I've noticed that there is allot of documentation but that sometime it's lacking in the exact part that you are looking for. So as one of the other devs put it. It's really easy, but has a steep learning curve. 22:23:05 That sounds like a canidate for an OpenStack project! 22:23:07 ok we all done w/ quotas i assume? 22:23:11 :) 22:23:15 :-) 22:23:26 lulz grapex 22:23:31 heh 22:23:36 I think we're good with quotas 22:23:38 #topic Percona Image Updates 22:23:44 kaganos: batter up 22:24:00 i'm here ... 22:24:07 so, we've been having some issues ... 22:24:10 tell us all your dark percona image secrets 22:24:29 well, as of now, still not working :) 22:24:42 we have some issues with the differences between the two deployments 22:24:53 like my.cnf and mysql.conf differences, as well as 22:25:07 some specific code assuming it's stock mysql installed 22:25:20 ya that is somewhat annoying, the guest apt checks i assume 22:25:20 we're going through it one step at a time 22:25:24 kaganos: GREAT! 22:25:31 looking forward to seeing it 22:25:38 same here ... 22:25:39 ;) 22:25:40 hehe 22:25:43 looks like percona doesn't have upstart 22:25:57 also doesn't behave the same way on first boot 22:25:57 like it doesnt come w/ a upstart script? boo 22:26:00 no 22:26:01 yeah, that was the mysql.conf kaganos mentioned earlier. 22:26:06 I think. 22:26:08 so that might be something we have to stick in reddwarf and copy 22:26:10 yep 22:26:16 the upstart is actually add ed by the folks at Ubuntu 22:26:17 sure vipul 22:26:22 not mysql 22:26:34 dkehn: thats a good point yall arent using percona from a apt repo via ubuntu right? 22:26:41 do they have their own apt repo? 22:26:45 yep 22:26:49 They have their own apt-repo.. 22:26:53 don't think they have one through ubuntu at all 22:27:01 they don't 22:27:05 there is this procedure with the gig keys 22:27:13 yup debian ftw ;) 22:27:18 that's also part of the problem why the current design is difficult for percona 22:27:23 for stock mysql, 22:27:35 there is an assumption that the image might be clean, 22:27:41 and then the code will set mysql on the fly 22:28:00 with percona, it's not just a matter of using a different apt-get install command 22:28:11 there is a bunch of stuff that needs to be taken care of 22:28:13 for sure 22:28:15 so basically, 22:28:21 we just were elaborating on the init script wrt that :) 22:28:23 either we won't support that flow with percona 22:28:45 or we revise the whole mechanism to use some sort of extesability 22:29:04 so you pass in interface implementing a provider 22:29:10 so you'd have all there: 22:29:18 is_installed() 22:29:22 is_running() 22:29:29 install_yourself() 22:29:34 config_yourself() 22:29:35 etc. 22:29:51 that would take longer of course ... 22:29:55 for sure 22:30:08 so currently we're trying to take the shortest path 22:30:10 but would be nice once the entire oss community abandons stock mysql for maria for instance :P 22:30:12 for right now, I think the plan is to assume the image is pre-baked with percona (in the percona-case). 22:30:13 possible just to make percona run 22:30:22 cool 22:30:32 we assume that, but the code doesn't .... 22:30:35 hub_cap, http://www.itwire.com/opinion-and-analysis/open-sauce/58571-monty-has-last-laugh-as-distros-abandon-mysql?goback=.gde_3177671_member_211032261 22:30:41 so, the code needs to be modified somehow ... 22:31:01 * hub_cap reads the article 22:31:13 quicker than you thinbk 22:31:28 kaganos: for sure, i dont doubt that. it was coded for 1 type, not assumign > 1. but i welcome the code changes to make it more extensible 22:31:31 ;) 22:31:47 we'll do our best 22:31:47 #info percona work still in progress 22:32:02 good good. time to move on? 22:32:07 thx for the update kaganos 22:32:12 #action kaganos still working on getting percona build working with guest agent. 22:32:22 sure 22:32:28 #agree 22:32:34 #aprove 22:32:37 #happy 22:32:37 Thanks kaganos…good work! 22:32:42 lol 22:32:43 #topic Security Groups feature feedback 22:32:45 heh 22:32:51 yes 22:32:53 I put that in there 22:32:55 #tag happy 22:32:55 so... ive thought about this a bit 22:32:57 since I sent an email out friday 22:33:13 hub_cap: go for it 22:33:21 …and what are your thoughts? 22:33:25 since its not in the core openstack api, id say we welcome security groups, but i dont htink we can put them in our core api... we can keep them an extension tho 22:33:53 does that make sense? it seems _in general_ a bit of a stretch to have something in our core api that depends on a openstack extension 22:34:09 Hmm yea, that's fair 22:34:19 We can go with extensions.. 22:34:26 since not every deployment will need it 22:34:29 wait, you can install Openstack without security groups? 22:34:34 ok cool. but i think the api makes sense 22:34:39 SlickNik: technically yes 22:34:42 since its not in the core api 22:34:53 Wow, didn't know that. News to me… 22:35:05 If so, I guess that makes sense. 22:35:06 i mean, it comes w/ those extensions 22:35:36 but they arent guaranteed 22:36:03 so we good w/ that? can someone link the bp? 22:36:07 #info Security Groups will be extension API 22:36:15 or wait was it just a email? did u bp it vipul? 22:36:20 I'm not sure we have a bp yet 22:36:23 no bp yet, can you create nother dummy 22:36:32 #action hub_cap to create bp_dummy 22:36:33 sounds like i'm calling you dummy lol 22:36:35 LOL 22:36:38 i am 22:36:55 lol, I added an underscore to make it clearer. :P 22:37:04 hahaha 22:37:06 nice 22:37:35 LOL 22:37:48 #link http://en.wikipedia.org/wiki/Eats,_Shoots_%26_Leaves 22:37:57 punctuation woes... 22:38:06 ok moving on then? i feel like we are all good w/ the api since it mirrors the openstack one 22:38:10 nice esp1 22:38:34 #topic API Spec General Update 22:38:39 sounds good 22:38:47 ok there is not much to report here. weve gotten together and got thru like 3 api calls 22:38:59 ill be starting a markdown to keep track of them 22:39:06 ok cool 22:39:10 a la 22:39:11 #link https://github.com/openstack/identity-api/blob/master/openstack-identity-api/src/markdown/identity-api-v3.md 22:39:26 just wanted to give you another heads up.. we're going to be submitting an API in the BP for snapshots.. 22:39:27 and so we can start commenting on them now rather than having ot consume them _all_ at once 22:39:35 vipul: very nice 22:39:39 that BP will also be modified a bit to implement via xtrabackup 22:39:57 we agree snapshots are core? 22:40:09 DEF 22:40:12 :) 22:40:13 k 22:40:25 hopefully have something to review next week 22:40:31 so i dont really have much more to talk about on that... but i will action me 22:40:36 so is there any issues from rax using xtrabackup? 22:40:41 #action hub_cap to start markdown on api 22:41:17 well.... its not exactly standard :) 22:41:29 and.... 22:41:36 its free tho rigth? 22:41:45 yes 22:41:50 if its the right tool for the job i have no issue w/ it 22:41:58 k 22:42:03 …and open source too, I believe. 22:42:17 ok im fine w/ it 22:42:23 I thin dkehn you may want to call out the pluggability aspects in the BP 22:42:41 because we may do xtrabackup now, but nothing should prevent another impl 22:42:48 ya that would be smart, a pluggable interface. nice vipul 22:43:14 confirmed 22:43:17 that would be neat. 22:43:27 #action dkehn to modify Snapshots BP 22:44:06 ok so im done w/ items, gen discuss time if thats a-ok 22:44:43 sounds good.. 22:44:44 #topic Open Discussion 22:44:45 open issues? 22:44:49 GO! 22:45:05 Point to note…. 22:45:36 haha 22:45:36 dkehn got devstack-vm-gate script running, but hit a snag when he added swift back in... 22:45:41 * Esmute doesnt like to have his real name in the internet 22:46:01 then def dont /whois you Esmute 22:46:10 we are close but there are a lot of hard coded things in there 22:46:20 crap.. i have to change that too 22:46:22 there's likely some conflict with the local apt server on the host on 8080 and swift, so we're looking at that. 22:46:26 hhe 22:46:39 we being dkehn and I 22:46:49 I'm actually testing a fix at present 22:46:54 fingers crossed 22:47:19 Esmute: it's the just the interweb what could possibly happen? 22:47:21 sweet, dkehn…I'll probably have some time to look into it this evening, if that doesn't work. 22:47:30 datsun180b weren't you working with esp1 on a grants fix? 22:47:30 k 22:47:32 vewy cool guys 22:47:38 will let u know 22:47:38 do you guys have that sync'd up? 22:47:48 esp1: lol 22:48:10 vipul: I think I have way I need from datsun at the moment. It was very helpful to see what he's working on! 22:48:32 I updated the BP as well. 22:48:45 #link https://blueprints.launchpad.net/reddwarf/+spec/create-restricted-root-account 22:48:55 thanks for the link, esp1 22:49:26 that's pretty much all I had. 22:49:53 cool 22:50:09 anything else? 22:50:20 If not, we might be poised for another record :) 22:51:00 WOOT!!!! 22:51:08 im gonna end it in 1 min if no one has anythign to say 22:51:11 do it... 22:51:31 lol 22:51:31 w00t! 22:51:36 #endmeeting