=== sidnei_ is now known as sidnei
_mup_ensemble/expose-ec2-provider r271 committed by jim.baker@canonical.com00:30
_mup_Merged trunk00:30
=== koolhead11|afk is now known as koolhead17
=== daker_ is now known as daker
SpamapShazmat: I have the ec2->common stuff ready, will push soon now11:02
kirklandhazmat: awesome, thanks for working with negronjl 12:28
kirklandhazmat: glad to see we're cranking ;-)12:28
ixxvilhey sladen , kim0 12:34
etnegsladen: ping12:36
etnegkim0: ping12:36
kim0wonder what time is it for sladen 12:37
kim0cool then12:37
kim0etneg: how's it going 12:37
sladenetneg: are you the same as ixxvil?12:38
sladenetneg: (one disappeared, as the other appeared)12:38
etnegnot too bad12:39
etnegoh ye12:39
etnegdid you guys see some rocky guy's logo?12:39
etnegthat looked like an atom12:39
etnegnobody commented on it, just thought i'd give you a headsup12:40
etnegi haent done any new stuff, kinda got tied up but ill do a few today and upload12:42
koolhead17etneg, capital E would be much better :P12:48
etnegkoolhead17: for which one?12:49
etnegthe one i just posted?12:49
etnegthat's not my design, it's a robbie williamson's concept12:49
koolhead17etneg, hmm. okey12:50
etnegbut ye leave comments12:51
etnegmore options on the table12:51
koolhead17he comes here too i suppose12:52
etnegah ok12:53
etnegwell either way we have more people working on it,good sign12:54
koolhead17etneg, indeed12:54
* koolhead17 looks at SpamapS 12:54
etnegi sitll need someone to start coloring mine:/12:55
_mup_Bug #809319 was filed: zookeeper finding should be the machine provider's responsibility <Ensemble:New> < https://launchpad.net/bugs/809319 >13:23
niemeyerhazmat: Yo!13:44
niemeyerhazmat: How did it go yesterday?13:47
hazmatniemeyer, pretty good, got some good momentum on the security policy stuff13:50
hazmatniemeyer, negronjl and i were able to work through some of the deploy issues, afaics it was just an update issue13:50
niemeyerhazmat: Hmm.. how did the issue go away?  It felt like it was reproduced after a clean bootstrap.. was that not the case?13:51
hazmatniemeyer, not after updating and doing a clean bootstrap13:51
hazmatthe issue went away13:51
niemeyerhazmat: Ah, sweet13:51
hazmatniemeyer, re robbiew's email on weekly releases.. i'm a little wary atm, but i'm also unclear on how often packages get pushed to the distro repo for ubuntu-next (oneiric)13:52
niemeyerhazmat: Probably not too often13:53
niemeyerhazmat: Weekly integration releases sounds sensible13:53
hazmatto me the question would be if we're hitting backwards compatibility issues13:53
niemeyerhazmat: We have some time to allow for wild backwards incompatible changes in terms of deployments13:54
hazmatniemeyer, what's involved in making that happen, can the daily build recipe be adjusted to a weekly?13:54
niemeyerhazmat: We'll soon have to put that in control, though13:54
niemeyerhazmat: I suggest we address this on the Austin sprint13:54
hazmatniemeyer, sounds good13:54
niemeyerhazmat: From then on, it has to handle the situation cleanly13:54
niemeyerhazmat: Even if that means saying to admin "You can't use that version" in some tricky situations13:55
niemeyerhazmat: We'll need a hand from real Ubuntu Developers there :-)13:55
koolhead17hi robbiew 14:03
niemeyerfwereade: Can you please have a look at this branch when you have a sec: https://code.launchpad.net/~hazmat/ensemble/security-principal-def/+merge/6761814:22
fwereadeniemeyer, ofc14:22
robbiewhazmat: so with this treadmill desk thing you got setup...will we need to move our sprint from the hotel conference room to the gym? :P14:32
hazmatrobbiew, i'm game. while the view might be nicer, the logistics are a bit harder to get a good roundtable setup or whiteboard usage. we might as well just sprinting from the pool treading water ;-)14:36
robbiewhazmat: hmm...they make waterproof laptops?14:37
SpamapSthey do14:37
hazmatrobbiew, toughbooks probably have a model for that14:38
robbiewhazmat: on the serious side...about how many calories do you burn a day now...or miles walked14:38
hazmatrobbiew, about 6 miles a day walked, i take desk breaks where i work/sit at a desk, i've been trying to start the day with a mile run to kick things off.14:40
robbiewhazmat: nice14:42
niemeyerhazmat: Would you mind to have a look at fwereade's branch which is a follow up on the previous conversation: https://code.launchpad.net/~fwereade/ensemble/zkaw-in-provider/+merge/6770314:45
hazmatniemeyer, sure14:55
fwereadebtw, what's the protocol for merging to trunk?15:12
fwereaderequest reviews from hazmat and niemeyer, wait for both, go?15:12
fwereadewell, wait for both to *approve*, I guess ;)15:12
SpamapSfwereade: btw, welcome. :)15:18
fwereadeSpamapS, thanks :)15:19
fwereadeSpamapS, very glad to be here (and especially *here*)15:19
hazmatfwereade, basically, the second reviewer is also supposed to set the merge proposal to approved, for work done at the sprint, we're going with one review, to try and keep things a bit more fluid. i typically do a bzr-pipeline when i've got multiple related things in review, so i can keep forward progress on development15:25
hazmatfor related things15:25
* SpamapS is going to have to look into pipelines15:26
* fwereade is looking into the right now :)15:27
fwereadehazmat, anyway, I'm good to merge now then?15:28
hazmatfwereade, can you add a separate test for the new provider method15:30
kim0SpamapS: Howdy Clint. So did you check out Ryan's email from mediawiki. I see this as a good opportunity, and I imagine either you or m_3 should get involved. Any thoughts ?15:30
hazmatelse it lgtm, +115:30
fwereadehazmat, ofc15:30
SpamapSkim0: I nearly spit my turkish coffee out when I read that.. *very* excited about it.15:31
kim0SpamapS: wow very good :)15:31
kim0so the ball is in our playground 15:31
kim0I think we'll need someone to help convert that infrastructure15:32
SpamapSkim0: I know some ex-wikia people and have met Ryan personally. I'll reach out to them soon and maybe go up there and play with it a bit with them.15:38
kim0SpamapS: Awesome :) so we're in good hands15:38
SpamapSRyan and I think very much alike.15:38
kim0good stuff indeed15:39
fwereadehazmat, niemeyer: re tests: a lot of the stuff I want to test is currently in the tests for EC2Connect, and (almost certainly) EC2LaunchMachine and EC2Bootstrap15:43
jcastroHi everyone, here's the weekly report so far, anyone have anything to add? http://pad.ubuntu.com/ensemble-report15:43
fwereadewhat's the local consensus on where such things should be?15:43
jcastrokim0: I don't know how to describe william's ec2zookeeper aware thing.15:44
fwereadejcastro: it's just some preliminary refactoring we think we need before we can do a cobbler provider15:44
fwereadejcastro: there have been no actual changes in functionality15:45
fwereadehazmat, niemeyer: on one hand, duplication bad: if we end up changing the precise operations performed we need to change tests in several places15:46
kim0fwereade: Glad you explained it :) indeed I had not idea what it was doing15:46
fwereadekim0: heh, sorry :)15:46
fwereadeon other hand, redundant verification good: if we didn't have this style of test it would have been much harder to refactor in this specific instance15:47
fwereade...but I'm still reluctant to introduce *more* duplication until someone tells me it's a good idea :p15:47
kim0SpamapS: I can't wait for the storage formulas nfs/ceph/...etc This is gonna rock so hard ;)15:48
hazmatjcastro, i'm doing some of the initial security implementation work15:50
hazmatjcastro, else most of what i've been landing has been following stuff up from last week15:51
kim0hazmat: is the config-set work still in review?15:52
hazmatkim0, yes.. it hasn't landed yet.15:53
jimbakerkim0, SpamapS - any thoughts on what's the planned mapping of ceph to the cloud infrastructure? that sounds quite interesting15:55
=== serge_af1 is now known as serue
SpamapSjimbaker: ceph is a ring.. it should work a lot like Cassandra actually.16:07
SpamapSjimbaker: the holy grail is to be able to just say 'requires: storage\ninterface: shared-mount' in any program that needs a shared mount point.16:08
koolhead17hey all16:08
SpamapSjimbaker: so ceph and nfs and gluster are interchanagable16:08
SpamapSchangeable even16:08
* SpamapS wishes irssi had spell check.. maybe time for a foray into X-chat again.16:08
kim0koolhead17: hey 16:09
koolhead17apt-cache showpkg moodle  shows mysql-client as deps but does not install it. Using dbconfig-common for installation throws error saying mysql-client needs to be installed before this process to finish. do i need to file a bug for this?16:09
* kim0 wishes unity had bell support so he can hear irssi's bells16:09
koolhead17i asked the same on u-server16:09
m_3SpamapS: do we have an interface list yet?16:09
kim0r u falling in love with dbconfig-common16:09
m_3SpamapS: I chose "filestore" and "mount"... we should standardize16:10
SpamapSm_3: no we *definitely* should standardize though16:10
m_3storage and mount/shared-mount are great16:11
m_3does mount need 'master' or is that something for 'shared-mount'?16:11
SpamapSm_3: the relation name should always be a reference to what function inside the formula it performs..16:11
m_3SpamapS: right, but filestore and storage would both work16:12
SpamapSm_3: I forget, did I hand off the NFS formula entirely to you or did we agree I should complete it?16:13
m_3SpamapS: I do like storage better16:13
* SpamapS feels his head wanting to turn hard right, starting the spin...16:13
koolhead17kim0, well indeed i am16:14
m_3SpamapS: ha!  I've been working on it, but was expecting to merge... I can finish if you want16:14
SpamapSm_3: Just push up what you've got into a branch.. I wanna see. :)16:15
m_3yup, will do... I've been having problems with legal branch names in lp... just using +junk now16:15
m_3is 16:16
m_3'master' there for a reason? or is that just thinking about a 'shared-mount' interface?16:16
koolhead17SpamapS, so i forget dbconfig-common while writing ensemble formulas :D16:17
SpamapSkoolhead17: well you can use it!16:17
SpamapSkoolhead17: since it knows how to configure the service for database access, its not totally useless. Its just *confusing* is all.16:18
koolhead17SpamapS, am stuck at phpmyadmin because a dbconfig-common bug which was reported in debian too16:18
SpamapSthat reminds me16:19
SpamapSadam_g: did you hack the mysql formula to work w/ shared db's yet?16:19
adam_gSpamapS: yes!16:26
koolhead17hey obino 16:26
adam_gSpamapS: i was gonna clean it up just a bit and send a proposal.16:26
adam_gSpamapS: https://code.launchpad.net/~gandelman-a/ensemble/mysql   ive been using it a bunch. works well16:27
SpamapSadam_g: *sweet* .. I think it will be useful for phpmyadmin as well16:28
adam_gSpamapS: cool. ill send a merge proposal before lunch hopefully16:28
SpamapSadam_g: no rush.. just wanted to see how it was going16:29
adam_gone thing ive noticed, tho16:29
adam_gmaybe this is the only formula that has multiple peers relating to it? but, the list of related peers for mysql unit in 'ensemble status' only lists the last related peer, not a list of all peers related.. if that makes sense16:30
SpamapSadam_g: I noticed it too16:30
SpamapSadam_g: I think the status output is just masking the actual list of relations16:30
adam_gobviously not a huge deal, but it would be convinient to have them listed there as well16:30
SpamapSadam_g: I think I saw a bug on that, did you file that?16:30
adam_gSpamapS: no, i haven't16:30
SpamapSadam_g: status' output is critical actually.. for a number of reasons.16:31
m_3adam_g: dude, the shared-db-relations look great16:31
adam_gcool thans16:31
adam_give also got a rabbitmq formula that needs a bug worked out and ready to submit, and a buncha nova stuff16:32
adam_g(nova-compute, nova-cloud-controller, glance)16:32
SpamapSadam_g: swift?16:32
adam_gno swift yet, was gonna try to tackle that this week16:32
SpamapSseems swift should be pretty straight forward as a formula.. the consumers just have to list and connect to the nodes right?16:32
jimbakerSpamapS, i take it that ceph would simply use the instance storage?16:33
adam_gSpamapS: well, yeah. but constructing the swift proxy and all of the storage nodes is a bit less trivial. its almost like building a raid array16:33
SpamapSjimbaker: yeah I think thats the simplest way16:33
SpamapSadam_g: sounds like I need to re-read the swift architecture page again16:34
jimbakerSpamapS, yeah, that does keep it simple. i wonder how that works out costwise in comparison to EBS16:34
SpamapSjimbaker: Actually I think I'd recommend we use EBS root instances.16:35
SpamapSjimbaker: at least until we have a better storage management story.16:35
jimbakerSpamapS, yeah, that's why i was curious :)16:36
SpamapSRoAkSoAx: re the orchestra-provider branch, I'm playing with it now.. can you try running './test ensemble.ftests.test_orchestracontrol' ? does that work with the environments.yaml I pastebinned earlier?16:37
RoAkSoAxSpamapS: let me check16:40
=== koolhead17 is now known as koolhead17|afk
RoAkSoAxSpamapS: ZOOKEEPER_PATH what should I set in the environment variable16:46
SpamapSRoAkSoAx: hm shouldn't need that.. 16:46
SpamapSRoAkSoAx: do you have zookeeper installed?16:47
RoAkSoAxSpamapS: yeah working now16:49
kim0SpamapS: m_3 You are invited to add Ensemble sessions to https://wiki.ubuntu.com/UbuntuCloudDays/Timetable16:49
jcastroso are we keeping the principia name or not? I'd like to just rename the wiki page to /Formulas16:51
koolhead17|afkrobbiew, saw the ensemble logo idea of yours. Capital E would be better :P16:52
robbiewkoolhead17|afk: heh...that's pretty easy to switch...but I thought the lower case looked better16:54
kim0reminds of IE though :)16:55
SpamapSsomebody should maybe grab this https://launchpad.net/formulas16:56
SpamapSrobbiew: Oh! yeah, the logo is *sweet*. As much as I like the oxygen reference with the number of electrons.. I think it might be too busy ;)16:57
jcastroSpamapS: I'll snag that on launchpad16:57
SpamapSRoAkSoAx: http://pad.ubuntu.com/orchestra-setup-for-ensemble .. working on it now16:58
jcastrolp.net/ensemble-formulas perhaps?16:58
robbiewSpamapS: yeah...had that thought too16:58
robbiewthat's easily reduced though16:58
robbiewmaybe just three16:58
robbiewto hint at the heads in the ubuntu logo16:58
RoAkSoAxSpamapS: awesome, thank you!16:59
robbiewthat's actually how I originally had it16:59
kim0m_3: thanks for the session \o/17:00
smoserRoAkSoAx, so is the pad doc there working for you?17:00
RoAkSoAxsmoser: yes17:00
smosercool. i will reproduce here too.17:01
m_3kim0: np... lemme know if you'd like another topic (i.e., if we need one for data instead)17:06
m_3kim0: I figured maybe the etherpad-lite node app would be cool to see17:07
jcastrowhat's the preferred license of a formula? Ideally whatever the upstream project's license is I would assume?17:12
m_3jcastro: maybe more open than that... formulas won't nec be written by the people maintaining the upstream project17:19
jcastrom_3: right, so in general we don't really care as long as it's open? Like we don't have an overarching guideline?17:20
smoserRoAkSoAx, i'm guessing its just user error17:22
smoserbut /etc/apache2/conf.d/dav.conf doesnt work for me17:22
smoserSyntax error on line 5 of /etc/apache2/conf.d/dav.conf:17:22
smoserInvalid command 'Dav', perhaps misspelled or defined by a module not included in the server configuration17:22
SpamapSsudo a2enmod dav17:26
smoserok. so that was fixed.17:28
smosernow Unknown DAV provider: filesystem17:28
m_3a2enmod dav_fs?17:29
m_3jcastro: we want to promote formula re-use... don't know what else17:31
SpamapSjcastro: the formula has to be redistributable under a free license so that we are in the clear and users know they can use it w/o any problems.17:32
SpamapSsmoser: thats printing when you restart apache?17:35
smoserworks now.17:35
smoserneeded dav_fs17:35
smoserupdated etherpad17:35
SpamapSNote that I just updated the pre-seed template to point at the right 'repo' instead of 'pkgs'17:37
=== daker is now known as daker_
_mup_ensemble/security-node-policy-def r271 committed by kapil.thangavelu@canonical.com18:15
_mup_implement security policy, returns acls for new nodes by path.18:15
niemeyerI love robbiew's logo: http://ubuntuone.com/p/148H/18:33
m_3niemeyer: +118:35
m_3goes right along with the physics theme... "canonical ensemble"18:36
m_3we really do need some options that allow for "grand canonical ensemble" or "micro canonical ensemble"18:36
SpamapSEnsemble: Now w/ less electrons. ;)18:40
SpamapSniemeyer: +1 from me too.. I think we should adopt it.18:40
* SpamapS begins work on creating an "Ensemble Man" out of The Simpsons' Radioactive Man18:41
m_3like little ladder operators... from the vacuum state to an excited state in two shakes18:41
m_3SpamapS: rofl18:41
m_3SpamapS: set to the tune of 'triangle man' from theymightbegiants18:42
SpamapShaha ensemble man beats cranky-ops man. ensemble wins18:44
SpamapScranky ops man hates ensemble man. they have a fight. ensemble wins. ensemble man.18:44
m_3ok, we'll have a recording session in Austin18:46
SpamapSyou bring the accordion18:46
m_3tub-bass is more my talent18:47
_mup_Bug #809498 was filed: intended providers.ec2 submodule visibility is unclear <Ensemble:New> < https://launchpad.net/bugs/809498 >18:48
robbiewapparently my only key contribution to ensemble will be the logo18:52
SpamapSrobbiew: don't forget you also gave us a theme song18:58
SpamapSOr, I guess, its a mission statement more than a song. ;)18:59
SpamapSRoAkSoAx: any progress?19:02
m_3negronjl: the masters/slaves aren't set by the package, but it evidently doesn't matter... working fine so far19:02
zulensemble queries the uec-images.ubuntu.com by default right? how do we get to use our own images19:02
_mup_ensemble/expose-ec2-provider r274 committed by jim.baker@canonical.com19:03
_mup_Merged trunk19:03
SpamapSzul: right now ensemble only uses ec219:03
SpamapSzul: oh sorry I misunderstood your question19:03
SpamapSzul: there's a config parameter in environments.yaml19:03
SpamapSzul: where you set the ami19:04
zulyeah i tried that i didnt get any further 19:04
zulSpamapS: http://pastebin.ubuntu.com/642836/19:05
RoAkSoAxSpamapS: just got back from lunch19:05
RoAkSoAxworking on it now19:05
zulSpamapS: and my environment http://pastebin.ubuntu.com/642837/19:06
SpamapSRoAkSoAx: I updated the kickstart template again19:07
SpamapSzul: that looks more like a problem connecting to the ec2-uri than the ami19:11
SpamapSif not, then thats a crappy error message19:11
SpamapSzul: give it '-v' to get a full backtrace19:11
zulSpamapS: yeah i figured out what is wrong with that error message different ip now trying on the bucket stuff19:11
SpamapSas in ensemble -v bootstrap19:11
zulnow it cant find the s3 bucket19:12
RoAkSoAxSpamapS: k19:13
jcastroSpamapS: when you have a few minutes like when you take a break or something can we skype or google hangout? I have some dumb questions for ya19:13
m_3kim0: that 'getting started with hadoop on ubuntu' blog post simplifies a bunch19:18
SpamapSjcastro: still trying to get out the door to eat.. how about in about 1 hour?19:26
jcastroSpamapS: sure, whenevs19:26
serueSpamapS: go eat!19:26
serueforget that, give in and call dominos19:27
_mup_ensemble/expose-ec2-provider r275 committed by jim.baker@canonical.com19:28
_mup_Refactored _ensure_groups19:28
RoAkSoAxSpamapS: zookeeperd should also be installed and running right?19:28
kim0m_3: simplifies? it sounded like it needed simplification :)19:36
_mup_ensemble/expose-provider-ec2 r276 committed by jim.baker@canonical.com19:40
_mup_Fix callback refactoring19:40
m_3kim0: yeah, dude... rocks19:44
kim0m_3: did you have time to write that readme file you mentioned19:44
smoserSpamapS, http://pad.ubuntu.com/orchestra-setup-for-ensemble . 'ensemble-branch' should probably match the orchestra branch, rigth?19:44
m_3kim0: just ran through the demo and am writing it up now19:47
m_3kim0: still a little ugly... references +junk repos, and lots of steps that won't be there after release19:48
SpamapSsmoser: yes but thats actually not important as that functionality is currently disabled. :-P19:54
SpamapSsmoser: the way to achieve the same thing is to put that branch's packages in the repo19:56
SpamapSwhich also has me thinking that there's work to be done in making ensemble deploy itself more cleanly.19:56
* niemeyer mumbles something about cloud-init19:57
SpamapSRight, I think cloud-init may be the right way to do what I'm thinking. I'm more concerned with making sure ensemble can install without access to ppa.launchpad.net or bazaar.launchpad.net.20:02
SpamapSI was just thinking that the provisioning agent should be able to encapsulate the desired ensemble version and deliver it to the machines.20:03
SpamapSProbably by setting a repo address and serving said repo itself... and then it can just build the .deb or .egg or whatever and serve it up.20:03
niemeyerSpamapS: I believe hazmat has already fixed part of that porblem in trunk20:08
niemeyerSpamapS: In revision 266, more specifically20:10
niemeyer"porblem" is an awesome typo, btw20:12
SpamapSSounds delicious20:12
SpamapSniemeyer: right, so the private cloud story is to also setup a bzr server available to your machines. I'm thinking that it should be part of the provisioning agent to just do that automatically.20:14
niemeyerSpamapS: I don't think I understand what you mean thre20:14
niemeyerSpamapS: The private cloud story should work the same way as the public cloud story20:15
niemeyerSpamapS: As far as protocols and procedures go20:15
niemeyerSpamapS: One talks to public services, the other doesn't20:15
SpamapSniemeyer: it has no access to lp:ensemble or ppa:ensemble/ppa .. just a local mirror of ubuntu.20:15
niemeyerSpamapS: Yes, and we won't need that.. trunk is able to deploy Ensemble without bzr branches available20:16
niemeyerSpamapS: It needs packages, of course, and we'll need to ensure there is a repository available20:16
niemeyerSpamapS: But that's somewhat of a need anyway.  Or at least that's my understanding20:16
SpamapSniemeyer: yeah, the story is handled.. I'm just not entirely happy with it at the moment. Still too dependent on ppa/lp ..20:17
niemeyerSpamapS: We don't depend on it.. we depend on an archive20:17
SpamapSare you guys going to EOD at 17:00 or 18:00 ? I still haven't eaten.. so much to handle after 2 weeks away from home. :p20:18
niemeyerSpamapS: It's trivial to implement support for looking at a different archive20:18
niemeyerSpamapS: Oh, 18+, please go for it20:18
SpamapSok.. bbiab20:18
niemeyerSpamapS: Enjoy20:18
jcastrokim0: alright! our first community question, http://askubuntu.com/questions/52840/differentiator-between-ensemble-and-front-runners-puppet-and-chef20:29
_mup_ensemble/expose-provider-ec2 r277 committed by jim.baker@canonical.com20:33
_mup_Remove default inbound fw rules20:33
SpamapSjcastro: wanna chat now?21:20
m_3jcastro: first question cuts right to the chase21:20
SpamapSjcastro: btw I'm going to answer that question with, hopefully, a short answer21:22
jcastroSpamapS: heh21:23
jcastroSpamapS: yeah I can chat now21:23
SpamapSjcastro: I have yet to try google's video chat.. and with Skype being under the scourge of new ownership... I'm anxious to give it a whirl21:24
m_3kim0: here's my first pass at a hadoop getting started post http://blog.markmims.com/ensemble/2011/07/12/hadoop-on-ubuntu21:43
m_3feedback from all pls... kim0 lemme know if you want something done differently21:43
jimbaker m_3, nice21:58
m_3jimbaker: thanks... trying to walk the balance between short/sweet and too many 'ensemble status' snippets22:07
jimbakerm_3, you may also want to show ensemble status using a filter22:09
jimbakerm_3, we probably should take an arg for add-unit so you don't have to do that list. i think this has been discussed before22:11
jimbakerm_3, lastly - you can do add-relation as soon as the services on each side of the relation have been deployed. no need to wait22:12
SpamapSm_3: have you tried the ensemble status svg output?22:15
m_3SpamapS: is that the dot one?22:15
m_3jimbaker: awesome thanks... that saves a step of waiting/checking22:16
SpamapSjcastro: btw, here is the draft docs for how service configs should work https://ensemble.ubuntu.com/docs/drafts/service-config.html22:16
_mup_Bug #809599 was filed: ensemble add-unit should support adding more than one unit at a time <Ensemble:New> < https://launchpad.net/bugs/809599 >22:16
* SpamapS also wants a "provision-machines" command so he can just tell the provision agent to start 10 ahead of time.22:17
m_3I also _so_ want a --stack arg to deploy22:18
m_3enseble deploy --stack svc1 svc2 svc322:18
jimbakerm_3, the other thing that comes to mind is that ideally the formula for hadoop should leave it in a good state such that it can always have jobs run against it w/o observing ensemble status by a user22:19
m_3but I'm so lazy I've got two character aliases for all of this22:19
m_3jimbaker: I don't follow22:20
jimbakerm_3, i have my own deploy-stack alias ;) - basically you should be able to do everything in that listing all at once, as long as the commands are run sequentially22:20
jimbakerm_3, that is, alias deploy-stack='ensemble status && ensemble deploy --repository=examples mysql && ensemble deploy --repository=examples wordpress && ensemble add-relation mysql wordpress && ensemble status'22:20
m_3the 'ensemble status' observations are to wait until the service is related before trying to ssh in and use the thing22:20
jimbakerm_3, yeah, i do wonder about using ensemble ssh in this way. it's really meant for debugging a system22:21
m_3jimbaker: there are lots of times that status reports the service started and the relation up, but there's still another several minutes (literally) of stuff to go before the relation is really up22:22
jimbakerm_3, maybe the right way to do this is have some sort of job submission service... not certain22:22
SpamapSisn't there a web interface for hadoop?22:23
m_3yup, but that's fer sissies22:23
m_3yeah we can show that too22:23
jimbakerm_3, precisely. but if you have a service that depends on hadoop being in a good state, there's no need to poll ensemble status. that might be the web interface SpamapS mentions, i don't know22:23
m_3I was just trying to reproduce the seminal posts on the topic... only changing ensemble22:23
jimbakerm_3, definitely it would be nice to have the web interface demonstrated22:23
_mup_Bug #809607 was filed: EC2-specific code is mixed in with generic machine-launching code <Ensemble:New for fwereade> < https://launchpad.net/bugs/809607 >22:24
SpamapSIf ssh is the normal way to use it.. then I think ssh is fine.22:24
jimbakerSpamapS, certainly a reasonable point22:25
SpamapSBut maybe ssh is the normal way its used because it has always been a PITA to get more structured access methods setup22:25
m_3yeah, I'd imagine the best way to use it would be from your laptop... letting other tools do the ssh-ing for you22:26
m_3BTW, these use cases are crying out for osx ensemble tools... lots of people use ubuntu only in the cloud22:27
jimbakerm_3, as far as i know, ensemble works fine on os x22:28
SpamapSYeah I'd be surprised if it didn't.22:28
jimbakerm_3, although i haven't tried for a while22:28
SpamapSpython setup.py22:28
SpamapStho maybe there are 2.7'isms now22:28
m_3was just thinking of the friends I would send the post out to22:28
SpamapSI think Lion may have 2.722:28
m_3good deal of them develop on mac but use ubuntu for cluster deployments22:28
jimbakerdefinitely depends on 2.622:28
SpamapSjimbaker: Snow Leopard has 2.622:29
jimbakerSpamapS, cool. as you mentioned, it might also be at 2.7. not sure. i should get out my MBP and try out the test suite again22:30
SpamapSWe actually should setup jenkins tests and have a mac slave running somewhere22:31
jimbakerSpamapS, that would be very nice22:31
SpamapSHmm.. I wonder if the test suite will run in a buildd's isolation..23:45
_mup_Bug #809634 was filed: Ensemble should allow additional user-defined userdata/cloud-config <Ensemble:New> < https://launchpad.net/bugs/809634 >23:47

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!