20:00:09 #startmeeting trove 20:00:10 Meeting started Wed Jul 10 20:00:09 2013 UTC. The chair is hub_cap. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:11 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:13 The meeting name has been set to 'trove' 20:00:15 heyo lets give stragglers some time 20:00:19 \m/ 20:00:32 but id like to confine to 1 hr 20:00:43 #link https://wiki.openstack.org/wiki/Meetings/TroveMeeting 20:00:44 \o^ 20:00:51 o/ 20:01:23 cp16net: what is that?! 20:01:38 he broke his arm 20:01:48 haha yeah 20:02:07 cp16net: u broke your arm?!?!? 20:02:19 no 20:02:21 o/ 20:02:24 coo 100% core is here lets get this rolling 20:02:29 o/ 20:02:39 but my ascii art did 20:02:41 \o 20:02:41 https://wiki.openstack.org/wiki/Meetings/TroveMeeting <-- already linked in #link so i wont do it 2x 20:02:49 * esmute is left-handed 20:02:55 wow 20:03:03 wrong window 20:03:05 esmute: but thats your right hand 20:03:05 o/ 20:03:11 #topic Update to Action items 20:03:27 cp16net: good catch lol 20:03:29 vipul: did u do any link stuff? cuz i sure didnt 20:03:37 cp16net: I didnt say where i was facing :P 20:03:37 yes cp16net: for esmute his left hand is his right hand. 20:03:38 hub_cap: Yes, i fixed most of them 20:03:42 GOOD! 20:03:45 lol 20:03:49 did u move pages to /Trove/blah ? 20:03:53 https://wiki.openstack.org/wiki/Trove 20:03:58 yep, a few remaining at the bottom 20:04:00 nice 20:04:06 oh did u just fix the ones on that page? 20:04:10 there is a whole slew of reddwarf references 20:04:13 if u search for em 20:04:27 Yea just started with the jump off page 20:04:30 ok cool 20:04:35 probably need to look closer 20:04:36 so u gonan work on it ongoing? 20:04:45 #action vipul to keep cleaning up the wiki 20:04:47 #action Vipul to continue to update reddwarf -> trove 20:04:47 :) 20:04:50 LOL 20:04:54 #undo 20:04:55 Removing item from minutes: 20:04:58 u like that!!?!? 20:04:59 woah 20:05:00 BOOYA 20:05:05 nioce 20:05:07 #undo 20:05:08 Removing item from minutes: 20:05:10 w00t! 20:05:11 #action Vipul to continue to update reddwarf -> trove 20:05:14 there now its proper 20:05:22 yours was nicer than mine anyway 20:05:24 so you can keep undoing eh 20:05:28 yup 20:05:32 prolly undo all of time 20:05:37 undo big-bang 20:05:52 SlickNik: ci job for docs? 20:05:53 #redo 20:05:55 hows that goin 20:05:59 lol SlickNik 20:06:20 I looked into it. 20:06:30 The job already exists. 20:06:36 :o 20:06:38 It's just a conf change once we have the docs. 20:06:46 sweet 20:06:52 so lets get the docs :D 20:06:55 SlickNik: Awesome! 20:07:08 I need to write the docs. 20:07:11 omg im freezing here 20:07:27 Basically the job builds them using sphinx. 20:07:32 weak 20:07:37 i gotta go back in my house, vipul talk about the next action after SlickNik is done. brb 2 min 20:08:06 k 20:08:19 The format is .rst (same as docutils) 20:08:47 I need to convert our redstack docs to .rst files to have some initial trove dev docs. 20:09:03 cool.. which repo do we push them to 20:09:10 brr back 20:09:26 what is brr back? 20:09:32 trove/trove-integration? 20:09:45 oh ok.. it's part of our stuff 20:09:50 brr is the onomonopia of me being cold 20:09:53 and back is im back 20:10:05 Yeah, most projects have it in a doc folder in the main project. 20:10:12 it's probably warmer up in seattle than where you at hub_cap 20:10:13 thats quite nice SlickNik 20:10:29 vipul: im not in the sun. in the sun its perfect 20:10:41 anyhoo... lets try to write some docs! 20:10:50 So I was thinking of adding it under trove/docs. 20:11:00 id say spend some time getting it started 20:11:06 so we have a jumping off point 20:11:10 #action SlickNik write initial dev docs. 20:11:14 and its a smaller barrier to have other people do it 20:11:19 docs good 20:11:22 ++ 20:11:34 Yeah, I can get it started. 20:11:36 so moving on to the next AI 20:11:48 And that's good, because it'll let me test the CI job's configuration as well. 20:11:54 Cool, moving on. 20:11:56 erxactly SlickNik 20:12:14 So I gave the BPs a quick look, and i think we got them all marked 20:12:18 vipul: grapex SlickNik and hub_cap need to look @ the blueprints 20:12:19 at least the ones that are in h1 20:12:20 h2 20:12:36 i didn't check trove-integration or troveclient 20:12:56 ya lets spend some time on em too 20:12:59 I checked trove-integration and cleaned up a bunch of bugs (there were no bps) 20:13:00 i have let them rot too 20:13:12 as a core group we need to keep on em 20:13:44 ok moving on 20:13:59 seems like thats the rest of the AI's 20:14:15 #topic h2-blueprints 20:14:17 #link https://launchpad.net/trove/+milestone/havana-2 20:14:42 we have < 7 days till h2 is cut 20:14:49 i should have mine done 20:14:59 is there anything _not_ on that list that shoudl be? 20:15:15 h3 will have some code freeze but it wont affect us till we are out of incubation 20:15:38 tap tap tap... is this thing on? 20:15:41 Nothing that I can think of that's not on that list that should be. 20:15:49 Add backup volume size to database is anyone looking at this? 20:15:52 SlickNik: as for your comment to my review on the config stuff, i totally agree 20:16:04 https://blueprints.launchpad.net/trove/+spec/backup-volume-size 20:16:09 if not, probably should be out of h2 20:16:11 i dont know vipul but i remember it being a problem in us doing impl'ing 20:16:25 grapex: do u know if anyone is lookin @ that? 20:16:36 robert mightve been.... 20:16:44 robertmeyers is out, his baby was born 20:17:13 another one bites the dust :) 20:17:17 lol 20:17:18 hub_cap: I'm not sure off hand, which probably means no. :( 20:17:22 poop 20:17:47 cool then we should move it to h3 20:17:48 hmm…anyone else who might be able to take this on? 20:18:02 i think we are stretched thin @rax currently 20:18:07 grapex: would make that call 20:18:19 hub_cap SlickNik: Yeah, lots of people are out at the moment. 20:18:34 I'm okay with h3 if it's okay with you grapex / hub_cap. 20:18:36 grapex: cool w/ movin it to h3? 20:18:43 And we're dealing with some internal matters... trust me, it sucks. We want to be back out here. In the sunlight. :) 20:18:45 yep, don't think it'll make it in 7 days 20:18:47 as in, are u confident no one will touch from rax for a wk 20:19:08 Right away, one issue I see with that BP I don't think anyone thought of 20:19:08 grapex: yea you guys have been quiet, do come out :) 20:19:22 vipul: im never quiet 20:19:26 when i have electricity 20:19:26 The disk utilization may not be the size required by the backup 20:19:48 grapex: plz add comments to the bp to help make it mo-better 20:19:53 ill move it to h3 20:19:56 Because I've been testing backup stuff lately, you can fill up a volume adding a huge table filled with rows, then drop the table but you don't get that disk space back. 20:20:00 hub_cap: Ok 20:20:11 grapex: the issue is that you have no way to know how big the volume needs to be 20:20:12 <3 20:20:31 demorris: Yeah, even when we're done with the backup there's no good way to check 20:20:37 ugh grapex that sounds....fun....... 20:20:38 unless xtrabackup can figure it out for us somehow 20:20:39 either you provide something in the DB that says how big the DB on the instance was, or you have the restore process override and rightsize the volume during the restore 20:20:50 but I am talking about the size on disk before the backup 20:20:50 ok lets add thsi stuff to the BP 20:20:53 thats what matters 20:20:57 backups are compressed, etc. 20:21:02 so the size of the backup is meaningless 20:21:06 otherwise itll fall in to the aether 20:21:12 right now, its a guessing game 20:21:16 demorris: Size on disk though may be more than you need if you've deleted a lot of stuff. 20:21:17 unless the original instance is around 20:21:18 so is that it for them? 20:21:20 yea but we could store the uncompressed size though right? 20:21:32 http://stackoverflow.com/questions/5983814/reclaim-disk-space-after-drop-database-in-mysql 20:21:54 demorris / grapex: is the scenario to be able to move from a bigger volume to a smaller one? 20:21:55 grapex: sure, but either way, how do you know how big of a volume you need? 20:21:56 Maybe we should talk about this particular aspect after the meeting? 20:21:57 initially we can require that the new size > than original (reclaimed or not) 20:22:29 #topic disk space for backups 20:22:41 yeah, might need further discussion. 20:22:43 vipul: uncompressed size wouldnt work you need size on disk 20:22:48 the scenario is, I take a backup of my database and it gets stored in cloud files 20:22:48 later I want to restore to a new instance... 20:22:48 how big should my volume be? 20:23:11 can we put up a meta data file for the backup on how large it was on disk at time of backup? 20:23:20 did the topic not get changed? 20:23:24 am i even online?? 20:23:29 you is 20:23:30 hub_cap: i see ya 20:23:32 hmm thats odd 20:23:33 I see you 20:23:35 openstack is not on 20:23:36 i dont see the topic change 20:23:42 there was a netsplit 20:23:42 netsplit al day 20:23:46 it didn't get changed. 20:23:53 geez netsplit 20:24:04 the openstack chan op is not here 20:24:08 we also have to consider the behavior here 20:24:10 ya i see that hes gone 20:24:14 ok so shoudl we discuss this now? 20:24:23 we dont ahve a ton of stuff to discuss honestly in the emeting this wk 20:24:27 we can make this an official topic 20:24:33 we allow you to specify a volume, so lets say you save the original disk util size in the DB and you use that when doing a create 20:24:37 If we're ok with restores not being able to reclaim space perhaps on a smaller volume, that seems like an easy thing to add 20:24:54 aww hell who needs topics lol we are already discussing 20:25:02 hub_cap: exactly :) 20:25:08 freenode services are disabled today due to some sort of attack 20:25:29 when *isn't* freenode under attack? 20:25:33 It's because the attackers hate our free(dom)node 20:25:36 lol 20:25:48 i think grapex is talking xml now 20:26:00 I can implement that one bk blue print 20:26:03 when is the deadline? 20:26:13 next tue 20:26:19 hub_cap: Sign me up! 20:26:35 so the BP is missing something 20:26:43 demorris / grapex: If you don't need to support the downgrade scenario, the simplest thing to do would be to save the volume size with the backup and ensure that restore happens on a volume at least as big. 20:26:45 demorris: that why i said put nots on the bp 20:26:47 *notes 20:27:01 if we store this value in the DB, when do you use it? What if I specify a volume size on the create? who wins? 20:27:02 #link https://blueprints.launchpad.net/trove/+spec/backup-volume-size 20:27:04 https://blueprints.launchpad.net/percona-xtrabackup/+spec/compact-backups 20:27:11 ^ ^ sign yourself up grapex thre are like 3 tim simpsons 20:27:15 looks like teh --compact option might work? 20:27:22 SlickNik: yeah, that could work 20:27:24 SlickNik: Agreed. That's how I'll fix it. If we can think of a fix for that downside issue before next Tue I'll add it 20:27:24 demorris: would you store it in a db or with the backup? 20:27:35 imsplitbit: I don't care where you store it 20:27:43 lies demorris u do care 20:27:47 do u want it on the list backups 20:27:51 if so it needs to be in the db 20:27:58 we had this talk already 20:28:02 ok 20:28:26 i think it would need to be in the db 20:28:31 hub_cap: i see, well I did not realize initially that was limiting that 20:28:35 so you could get the info for restore 20:28:41 I was more curious from a resilience/dr situation as an ops person how would I know how big the backup was if the db got nixed 20:28:44 what would you call it? Minimum volume size? 20:28:51 sizeOnDisk? 20:28:56 we can figure it out 20:28:59 lets not get too into it now 20:29:03 np 20:29:08 okay 20:29:12 lets let tim get into impl'ing it and he can propose names 20:29:25 i prefer size_in_grains_of_rice personally 20:29:26 I <3 grapex's anems. 20:29:31 lol 20:29:36 ++ SlickNik 20:29:55 so just a reminder 20:29:59 as long as we understand the scenario…today we have a guessing game where potentially the users gets to guess on how big the volume needs to be until they get it right... 20:30:15 #topic next replication meeting Thursday, 11th July 1 p.m. PST/3 p.m. CST 20:30:19 that is all 20:30:24 CMON THIS SUX 20:30:29 sounds good 20:30:38 what sucks about it? 20:30:42 I know, no bot :( 20:30:51 oh 20:30:56 bots make everthing better 20:30:58 woot 20:31:01 yeah 20:31:05 he's back? 20:31:06 #agree 20:31:06 #topic next replication meeting Thursday, 11th July 1 p.m. PST/3 p.m. CST 20:31:12 so it's likely this meeting notes are not being captured? 20:31:19 vipul: i would say no 20:31:19 yea prolly not 20:31:20 very likely. 20:31:22 vipul: correct 20:31:22 :( 20:31:25 he prolly died 20:31:25 actually 20:31:27 they might be 20:31:33 well only 1 way to find out!!!!! 20:31:34 i think the bot just doesn't have ops to set the channel 20:31:35 someone can check 20:31:37 topic 20:31:53 which meeting is this? 20:31:55 ya anyway we have to update that blueprint w/ the stuff we talked about 20:31:58 trove jeblair 20:32:07 http://eavesdrop.openstack.org/meetings/trove/2013/ 20:32:13 yeah http://eavesdrop.openstack.org/meetings/trove/2013/trove.2013-07-10-20.00.log.txt 20:32:16 it's being recorded 20:32:24 cool 20:32:24 i think if anyone knows whats goin on w the blueprint, grapex does, so its all good 20:32:48 so i didnt have anythign to say about that topic 20:32:51 jsut bringing it up 20:32:52 moving on 20:33:06 #topic Mission stmt finalized 20:33:08 no one has sent me any updates 20:33:19 * datsun180b end-zone dances 20:33:21 so im going to send that mission statement out 20:33:21 has it been sent out to ML? 20:33:24 not yet 20:33:26 oh I must have missed this 20:33:28 i wanted us, as a gropu to update it 20:33:36 did we still get no change to topic 20:33:38 BOOOOOOOOO 20:33:46 https://wiki.openstack.org/wiki/Trove 20:33:49 jeblair: its broken again me thinks 20:33:50 we have a mission statement now? 20:33:54 it's there 20:33:56 sweet 20:34:03 thx for the update jeblair 20:34:06 So wait are we now a Mission or a Program? 20:34:14 Program w a mission 20:34:21 I'm holding out until we're a Strike Force. 20:34:33 I'm with grapex on that one 20:34:37 u should call your internal team that grapex 20:34:39 grapex: Agent of S.H.I.E.L.D. 20:34:43 stroke force delta tacos 20:34:46 LOL 20:34:47 strike 20:34:54 wow talk about a bad misspel 20:34:57 lol 20:35:08 and yes this meeting _is_ being recorded 20:35:30 #topic extensions 20:35:40 ps topic changes arent being recorded but thats ok 20:35:52 so, we discussed last wk that they were a big ole fail 20:36:06 i think that grapex needs to discuss how they are broken currently as well 20:36:22 we all know that we need to use stevedore and junk like that eventually 20:36:28 but hes got a intermediary problem 20:36:36 grapex: care to divulge 20:37:14 hub_cap: Sure! Is this extensions? 20:37:23 yes 20:37:28 #topic extensions 20:37:32 boo 20:37:36 imsplitbit: i already did that 20:37:37 its nto working 20:37:41 #topic extensions 20:37:43 maybe it is now 20:37:47 So we've noticed that in trove/extensions/ there are a series of Python files which have the same names as the directories 20:37:47 jeblair: is on it 20:37:48 woot! 20:37:48 I saw openstack joined 20:37:54 thought I'd try 20:38:08 imsplitbit: i think only meetin chair can change topic 20:38:09 imsplitbit: only the person who started the meeting can change the topic. 20:38:10 so if you try to import the modules in those directories from anywhere other than those particular files it won't work 20:38:14 ah 20:38:22 well fooey 20:38:27 So for example 20:38:46 In many contexts, "from trove.extensions.account import models" fails 20:38:57 because it finds "trove/extensions/account.py" 20:39:31 I was thinking, the several files in trove/extensions could be moved to somewhere else, and then we could just make that the default place for extensions 20:39:33 like maybe 20:39:47 contrib/extensions/account.py, contrib/extensions/mgmt.py etc 20:40:05 Then point the extensions to contrib/extensions, since the current (yucky) way extensions works is file based 20:40:41 Then, we could put any useful classes that are in those bad files such as trove/extensions/mgmt.py's "Mgmt" class into the __init__.py sitting in the folder of the appropriate name 20:40:41 yea that would work.. do we have stuff that needs to go into trove/extensions that is not part of the exisitng API extensions? 20:41:20 Well at Rax we do. We actually have some billing code which we need a MGMT API extension for. It's all stuff that talks to the Rax system. 20:41:57 So we went to make an extension directory of our own and figured out if we did that we wouldn't be able to import any of the modules sitting in the current extension directory. 20:42:08 Ok, yea i'm fine with it.. 20:42:18 I know the extension classes have changed upstream in oslo quite a bit. 20:42:19 we would jsut module extensions in trove/extensions 20:42:51 SlickNik: they dont exist anymore in oslo lol 20:42:55 SlickNik: would it be worth pulling in latest to see if it satisfies what grapex wants to do? 20:42:56 Wonder if there are any changes to address this problem that you mention grapex... 20:43:04 hub_cap: oh 20:43:19 hub_cap; I think they have plugins now. 20:43:20 #link https://github.com/openstack/oslo-incubator/tree/master/openstack/common 20:43:21 vipul: The issue isn't entirely the Oslo extension feature as much as that we have import failures. 20:43:30 I don't know how similar or different they are to extensions. 20:43:39 whatever they have, its not in oslo as i se it 20:43:47 so this is 2 things 20:43:58 1) tim needs to import stuff and cant -- needs to get fixed 20:44:01 ah, maybe they took that out too... 20:44:10 2) we need to update extensions, period, and use stevedore 20:44:17 he cares less about #2 now 20:44:18 hub_cap: Yes 20:44:36 so there will be a small extension refactor i think coming down the line for #1 20:44:40 and then later a big one for #2 20:44:45 I'm okay with moving extensions so that grapex is unblocked. 20:44:46 just wanted to make everyone aware 20:44:54 grapex: plz BP it ;) 20:44:55 hub_cap: Good summary. 20:44:57 Will do 20:45:11 ok thats really it... 20:45:19 #topic open discussion 20:45:23 we have 15 min now 20:45:39 i want to reiterate that h2 is < 1 wk away from being cut 20:45:54 id love to get the config stuff merged by then *wink wink* 20:46:07 and possibly the rpm stuff as well 20:46:14 which is still a WIP 20:46:25 hub_cap: I'll try to look at that 20:46:26 well hurry up 20:46:31 :) 20:46:31 lol right imsplitbit?!? 20:46:36 plz do grapex 20:46:42 hub_cap: when's the cut date? 20:46:45 im going to use a __metaclass__ thing 20:46:51 #link https://launchpad.net/trove/+milestone/havana-2 20:46:56 under Expected 20:47:05 hub_cap: you forgot my topic :P 20:47:06 https://blueprints.launchpad.net/trove/+spec/admin-create-restore-backups 20:47:13 i wanted to run this by you guys... 20:47:19 oi sry vipul 20:47:20 hub_cap: thanks 20:47:32