/srv/irclogs.ubuntu.com/2015/02/17/#juju.txt

stubtvansteenburgh, marcoceppi : I broke your Jenkins05:37
weblifeThought someone here may appreciate this: https://mastersproject.info/blog/54de76fb03b198870a6daa1206:04
traveanyone around here connect juju into digitalocean, using kapilt/juju-digitalocean ? i feel that i am close, just having trouble with it timing out.06:12
traveafter setting up a free-tier account on AWS, things work like a charm. thanks anyway :]07:51
=== rcj is now known as Guest33360
lazyPowertrave: i have successfully use/am-using digital ocean w/ the plugin08:55
lazyPowerif you're still on that path. the AWS account provider is however one of the more stable/tried/tested providers in the toolbox08:56
travelazyPower, yea, i was likely pretty close with digitalocean, just wasnt picking up my ssh_key identity .pub name or something.. but this AWS path seems pretty stable. I take it the m1.small is about $18/mo.? Just to get my feet wet, how many charms can I throw at one of those instances? I want to toy with IO.js and Redis.09:09
lazyPowertrave: give or take - Digital Ocean is fairly straight forward once you get the flow - and i've found that often i have to manuallya dd the machines myself since the provisioning of VM's takes longer than their highly touted 60 seconds09:09
lazyPowerjuju add-machine ssh:root@host09:10
lazyPowerthen its in the pool of available machines - however - to answer your aws question - 2 machines (1 for bootstrap, 1 for your apps/containers)09:10
travemaybe thats all im missing there, is it doing a timeout when im boostrapping the first instance... now that ive spun up and destroyed a few attempts on AWS, i think i might go back to trying D.O. again09:10
lazyPoweri'll be around if you want help troubleshooting09:11
travecool, thanks man :]09:11
lazyPowertheir provisioner gets wonky late at night09:11
lazyPowermaintenance windows and all that09:11
traveyea, i might catch some zzzs soon, still got work in the a.m. but feel accomplished having waded my way this far, so far.09:12
lazyPowerhttp://blog.dasroot.net/juju-digital-ocean-awesome.html09:12
lazyPoweri did a fairly decent writeup of the flow to get moving w/ juju on DO09:12
lazyPowerbut that also was submitted for the official docs - so not much to do here other than boost my analytics :)09:13
traveOh yes, I did watch that, which is what got me as far as I did. :D thanks for putting that together, maybe i'll absorb more the second time watching it09:13
lazyPowertrave: allright - i'll be off on another work station so ping me if you need me for anything.09:14
travelazyPower: probably a dumb question, but heres where i get stuck when spinning up a new digitalocean-0 bootstrap instance... it asks for the root password, even though ive specified an ENV_VAR for DO_SSH_KEY, (do I need to add .pub to the end of that value?) Since I dont yet know what that instances generated root password is, do I let it hang there for a bit while I use the DO UI to go reset that to something else which emails it to me?09:28
lazyPowerah, no - you probably need to add your SSH key to DigitalOcean's admin web ui so it loads that into the instance(s)09:28
traveIts listed in there, strange.09:29
lazyPowerhttp://i.imgur.com/dBakNhu.png09:29
lazyPowerwhat version of the plugin do you ahve installed with pip? 0.5.1 assuming?09:30
traveyea, ive got it in there, and ive used it successfully before scp'ing it to my other instance by hand before.09:30
travechecking..09:30
travejuju-docean==0.5.109:32
lazyPowerjuju version 1.21.1?09:32
trave1.18.4-unknown-amd6409:32
lazyPowerwhoa, thats crusty09:32
trave;]09:32
lazyPowersudo add-apt-repository ppa:juju/stable09:32
traveim on osx09:33
lazyPowerah09:33
lazyPowerwe're shipping 1.18 on mac still? wat09:33
* lazyPower makes note to follow up on that later today09:33
lazyPowerit should still work regardless - the manual provider code hasn't changed.09:34
lazyPowerlet me check the bug tracker on hazmat's repo real fast, 1 sec09:34
travei'll try the --devel flag09:34
lazyPowerehh09:34
lazyPowerlets stick with 1.18 for the time being until i can verify09:34
travek :]09:34
lazyPowerah09:35
lazyPoweris the key you're using to authenticate against docean ~/.ssh/id_rsa?09:35
travei named it something else09:35
travei can make an id_rsa09:35
lazyPoweri bet thats why its having a derp moment09:35
lazyPowerhttps://github.com/kapilt/juju-digitalocean/issues/2909:35
traveaha, you da man.09:36
travek, making new key, trying again.09:36
travewell, it did something different at least: juju_docean.exceptions.ProviderError: Failed to get running instance digitalocean-0 event: {u'status': u'OK', u'event': {u'droplet_id': 4177198, u'percentage': u'90', u'event_type_id': 1, u'id': 43954384, u'action_status': None}}09:47
traveERROR exit status 109:47
lazyPowerit timed out - thats in relation to this: https://github.com/kapilt/juju-digitalocean/issues/2709:47
lazyPowerit gets pokey now and again unfortunately09:48
travek, trying agin :]09:48
lazyPowerhit paydirt yet trave?10:01
traveno, just what keeps seeming like timeouts10:02
traveyou think the --devel build is in a unstable state?10:02
lazyPoweri dont know10:06
lazyPowerits worth a go10:06
lazyPowerare tyou tracking the vm launch in a browser while you bootstrap?10:07
traveor, is 512MB memory/20GB disk the smallest tier that DO does, would spinnging up a smaller instance be faster at all to be under that 3min window?10:07
traveyea, i can see it spinning up a digitalocean-0 new instance, the CLI throws the error before its done provisioning that box10:08
travemy next feat will be to try out the --devel build, i bet it will work like a charm10:08
lazyPowereh10:09
trave;]10:09
lazyPowerits not a juju problem :( its a DO problem atm10:09
traveyea10:09
lazyPowerthe only thing i can suggest is either keep trying until the APi gets un-pokey, or do a pure manual bootstrap10:09
lazyPowerwhich is slightly more inconvenient10:09
lazyPowerbut it doesn't have any timing issues, you spin up the machine, specify the bootstrap host in your environments.yaml and you're off and running10:09
travegotcha. yea, i'll try through the day tomorrow, maybe their performance is up during the peak hours10:10
lazyPowershould be better then yeah.10:10
lazyPowerits that or try a different DC10:10
traveyep, several more avenues to try, this is good for me to broaden my skillset and learn10:10
travethank you so much for the help, youve been great10:11
lazyPowerI'm sorry it hasn't been as smooth as cheddar tonight however10:11
travei'll be around :]10:11
lazyPowerif only everything were as reliable :)10:11
trave:D10:11
travelazyPower: hey, a simple: brew uninstall juju; brew instal juju; now gives me version: 1.21.1-yosemite-amd6410:35
lazyPower\o/10:35
travetrying yet again ;]10:36
travewoot! bootstrap complete.10:40
traveis there any reason the examples show 2g memory for the bootstrap machine? this time I just used a 512m box, that should be fine, right?10:40
lazyPoweryeah thats fine10:54
lazyPowerthe example was with constraints - and thats why you see the higher mem limit10:55
traveagain, thanks. :]10:57
lazyPowerhazmat: thanks for the reply, thats basically what i had figured. Juju isn't doing anything special outside of using the agent and falling back to that specified key12:48
hazmatlazyPower: its very badly documented in juju12:49
hazmat search authorized-keys over here https://juju.ubuntu.com/docs/config-general.html12:49
hazmatand afaics thats the only doc for it.12:50
lazyPowerhttps://github.com/juju/docs/issues/26812:51
lazyPowerleft a TODO to follow up on. cheers12:51
hazmatlazyPower: thanks12:52
hazmatlazyPower: for the other issue i have a timeout option sitting in my working copy...12:53
* lazyPower fist pumps12:53
lazyPowerso we can see that branch land soon, very nice12:53
lazyPoweri was considering pulling the repo this weekend and working on the v2api support if you haven't already looked into it and getting something shot over for review12:54
lazyPoweri got sidetracked with my tinfoil hat, i must admit12:54
=== mhilton is now known as mhilton-lunch
lazyPowerhazmat: it may be prudent to close #29 as thats a juju issue and not a plugin issue.12:58
lazyPowerunless you plan on supporting upload ssh key to DO and setting config to leverage that key12:59
=== kadams54 is now known as kadams54-away
=== kadams54-away is now known as kadams54
=== Guest33360 is now known as rcj
=== rcj is now known as Guest49948
tvansteenburghstub: i killed that jenkins job13:48
stubtvansteenburgh: Ta.13:51
stubERROR failed to bootstrap environment: cannot start bootstrap instance: no OS images found for location "West US", series "trusty", architectures ["amd64" "arm64" "armhf" "i386" "ppc64el"] (and endpoint: "https://management.core.windows.net/")13:53
stubI think the next one is doomed too13:53
stubStill, am interested in how things go with the other providers. I had some code paths not running under lxc that failed with the other providers.13:54
tvansteenburghstub you want me to kill this one too?13:56
stubtvansteenburgh: Only if you think a retry will fix that error. 'no OS images found for location' doesn't fill me with confidence of it working today.13:56
tvansteenburghstub: i'll let it run i guess. azure is the environment most often broken. it may work later though13:58
stubtvansteenburgh: How many environments are there btw? About 10?13:59
tvansteenburgh513:59
tvansteenburghlxc, hp, awc, azure, joyent13:59
=== kadams54 is now known as kadams54-away
=== kadams54-away is now known as kadams54
stubI might have told bundletester to run the unittests twice then.14:00
tvansteenburghstub: well let me know if you need stuff killed or whatever, happy to help14:01
stubta. This run looks happy for the next hour in any case.14:01
tvansteenburghstub: also next time you want to start one, let me know, i want to try it on the new jenkins setup i've been working on14:02
stubok14:02
hazmatlazyPower: definitely not uploading people's public keys.. to much responsibility.. i'll close it out.14:13
=== jhobbs_ is now known as jhobbs
=== mhilton-lunch is now known as mhilton
jcastrojose, what day do you get to SCALE?15:12
josejcastro: tomorrow 3pm local time15:12
joseI'm getting ready to fly out later today15:12
=== kadams54 is now known as kadams54-away
jcastrooh excellent, me and marcoceppi as well, we can just sit down and bash out the details of our talk15:12
jcastrowhich isn't until saturday anyway15:12
joseyeah, sounds good to me15:13
=== kadams54-away is now known as kadams54
lazyPowermarcoceppi: mbruzek, tvansteenburgh - whats our stance on incoming tests that only stand up the service? Enough to satisfy the requirement or do we gently guide them to write an actual functional test?  https://code.launchpad.net/~jose/charms/precise/quassel-core/add-tests/+merge/246001 <- is the test in question16:00
lazyPowerI'm inclined to say the latter, but its not just my opinion here.16:01
mbruzekgently guide16:01
mbruzeklazyPower: even if he did a ps aux | grep quassel that would make me happy16:02
lazyPoweryeah, we have enough weak tests i feel16:02
lazyPower+116:02
jose+1 to that, nice tests are easy enough to make16:03
jose(I made those in a hurry since Paul wanted the charm approved and had no tests, no results yet)16:03
mbruzekack16:04
lazyPowerI promise this wasn't a poke @ your effort(s) jose :)16:04
josequestion, though16:04
joseshould 'simple' tests be renamed to 10-deploy, or should I keep them as 99-autogen?16:04
lazyPowerif you're not doing anything to actually validate, the 99-autogen gives me a clear indicator of what's been done16:05
joseok16:07
marcoceppijose: yeah, leave them as 99-autogen, we'll be sniffing that out in the future to find charms that need better tests and it makes it so we can track unmaintained charms16:29
joseok16:29
stubtvansteenburgh: Do you recall what constraints the test vms have? I can install the charm on our local OpenStack, but with little RAM the install hook fails due to the OOM killer.17:28
stubtvansteenburgh: I don't think I'm going to get anything more useful out of this build, so you can kill it. I might have hung it again too sorry.17:29
stubtvansteenburgh: Happy to have it kicked off in your new Jenkins if you want to give it a workout.17:29
tvansteenburghstub: will do17:30
tvansteenburghstub: in the new setup we bootstrap with mem=2G by default17:30
stubI'll test that here. I know it fails with 1G, and installs with 8G17:31
stubI may be able to tune the requirements down at my end if it is needed.17:31
tvansteenburghstub: http://juju-ci.vapour.ws:8080/computer/charm-bundle-slave/17:32
stokachujcastro: lazyPower do you guys know if juju http proxy settings are exposed in charms?17:32
stokachufor things like wget to make use of17:33
lazyPowerI'm not sure tbh17:33
jcastroyeah, that's a good question17:33
tvansteenburghstub: new tests running, the main job kicks off one job per substrate17:33
stubtvansteenburgh: Oh, cool. No more interleaving of output.17:33
lazyPowertvansteenburgh: is my docker test still in the queue or did it get wiped?17:33
stublazyPower: Its running now17:33
* lazyPower does a dance17:33
lazyPowerta17:33
stublazyPower: oh, it *was* running17:34
tvansteenburghstub: yeah, much better for several reasons. will be the default soonish17:34
lazyPowerstokachu: i'm reasonably certain that if its set as an env variable it will be, however it would be best to confirm with someone like wwitzel317:34
tvansteenburghlazyPower: pfft i thought that was stub's job and killed it17:34
lazyPowertvansteenburgh: >.>17:35
jcastrohttp://askubuntu.com/questions/430865/juju-http-proxy-and-no-proxy-settings17:35
tvansteenburghlazyPower:  what was the test url17:35
jcastrostokachu, ^^^17:35
jcastro"The proxy options are exported in all hook execution contexts, and also available in the shell through "juju ssh" or "juju run"."17:35
lazyPowertvansteenburgh: i didnt get one, i kicked it off from the revq17:35
tvansteenburghlazyPower: if it's any consolation, it had just started17:36
stokachujcastro: ah nice find!17:36
tvansteenburghlazyPower: no i mean the url of the thing being tested17:36
lazyPowerhttps://bugs.launchpad.net/charms/+bug/141377517:36
mupBug #1413775: New Charm - Docker <Juju Charms Collection:New> <https://launchpad.net/bugs/1413775>17:36
tvansteenburghlazyPower: is this it? lp:~lazypower/charms/trusty/docker/trunk17:38
lazyPoweryep17:38
stubmanual deploy to openstack vm with 2G seems ok, no oom17:38
tvansteenburghlazyPower: queued up behind stub's jobs in the new queue17:39
lazyPowerta17:39
stub@    WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED!     @17:43
stubtvansteenburgh: You may need to have the authorized_keys scrubbed between jobs, like http://paste.ubuntu.com/10276641/17:45
stubor just nuke both of them entirely I guess17:45
stuboh, here it goes...17:46
* stub watches his job on hp17:46
tvansteenburghstub: yeah good point17:51
stubDEBUG:runner:2015-02-17 17:37:21 Error getting env api endpoints, env bootstrapped?17:52
stubThe lxc job bombed17:52
* tvansteenburgh looks17:53
stubgrr... my install hook failed on joyent, but I can't reproduce locally or on our openstack :-/18:02
tvansteenburghstub: no idea what happened on lxc, it's running again here http://juju-ci.vapour.ws:8080/job/charm-bundle-test-lxc/8/console18:03
stubta18:04
tvansteenburghstub: we're capturing all-machines.log at the end of the tests now, that might yield a clue re joyent18:05
stubtvansteenburgh: It should, ya.18:05
stubI think Cassandra is crashing and we are waiting for the install hook to timeout and fail, rather than slow bootstrapping.18:06
sh00kI have what seems  like a pretty dumb question that I'm hoping someone in here might be able to help with me. I already have an environment configured and a machine bootstrapped in EC2. I am now on a different client machine and want to connect to the existing environment/bootstrap. I've already tried migrating the entire 'juju' folder to the other client machine, but I am getting an error back when trying to run 'juju status' of: WARNING discar18:07
sh00k70/environment/d4c68949-2e62-4604-8aa2-xxxxxxxxxxxxx/api"18:07
sh00kERROR Unable to connect to environment "aws".18:07
sh00kPlease check your credentials or use 'juju bootstrap' to create a new environmen18:07
sh00kt.18:07
sh00kError details:18:07
sh00kno instances found" any thoughts?18:07
lazyPowersh00k: did your $JUJU_HOME/environments/aws.jenv make it during the copy?18:09
lazyPowerif it did, are you able to ping the ip of the state server?18:09
sh00kYup, i double checked that - as well as the file permissions18:09
sh00kstand by let me try that18:09
sh00kDamn, so i'm guessing that's it. Ping requests timed out :/18:11
lazyPowerthat would be a key cause, if you rnetwork reachability isn't there.18:11
lazyPowerits bit me a couple times too - dont feel like the lone stranger sh00k :)18:11
sh00khahah thanks a lot18:12
marcoceppitvansteenburgh: what do I need to change in the review queue for the new testing stuff?18:16
tvansteenburghmarcoceppi: new jenkins endpoint, and you'll get one callback per substrate now18:17
marcoceppiDocs for this?18:18
tvansteenburghmarcoceppi: HAHAHAHAH18:18
marcoceppiROFL18:19
tvansteenburghmarcoceppi: i figured we could just bang it out at the sprint18:19
marcoceppitvansteenburgh: I plan to make a bunch of changes this week to revq18:19
marcoceppiThat works too18:19
tvansteenburghmarcoceppi: oh ok, i'll type something up18:19
tvansteenburghsooner would be better i guess18:20
tvansteenburghi'll document it18:20
marcoceppiEither way is fine. Does the old style work?18:20
tvansteenburghthe old one still works for now18:20
marcoceppiRough docs are fine. Even a bunch of cmd outputs18:20
tvansteenburghi haven't changed it, just built the new thing on the side18:21
tvansteenburghack re docs18:21
marcoceppiCool18:21
tvansteenburghlunch time18:21
=== tvansteenburgh is now known as tvan-lunch
=== kadams54 is now known as kadams54-away
lazyPowermarcoceppi: the endpoint/json stuff you sent me will largely remain the same yes?18:42
marcoceppilazyPower: yes, if it changes I'll let you know18:44
marcoceppiWill also be exposing more data18:45
lazyPowernice18:45
=== kadams54-away is now known as kadams54
lamontwill juju destroy-env reomve volumes that may have been attached to instances?19:03
lamontor simply detach?19:03
lazyPowersimply detach19:04
lazyPowerlamont: you're referring to extraneous volumes, such as those that are provided by block-storage-broker correct?19:04
lazyPoweror additional block devices allocated to the node, not just the baseline disk that comes with the VM?19:04
lamontlazyPower: correct19:05
lamontspecifically one taht our scripts euca-attache-volumes/19:05
lazyPoweryeah, those are not deleted on machine termination19:06
* lamont does hisdestroy-env then19:06
lazyPowerlamont: take a snapshot to be safe :)19:06
lamontnow you tell me.19:06
lamontvolume still tehre,arked avaialble19:06
lamontwoo typing!19:06
* lazyPower grins19:07
lazyPowernothing like a dose of doubt right after you send a command no?19:07
lazyPowersorry about that - just happened to think its always better to err on the side of caution19:07
lamontyep19:08
lamontyou were about 8 seconds after I said 'y'19:08
lazyPoweri wont steer you wrong... on purpose.19:09
lamonthehe19:09
lamontalls well that ends well19:09
lazyPowerhowever with storage landing in core soon - you might want to be a bit more persnickety about snapshotting before destroying things.19:09
lazyPoweri'm not 100% sure how it will work, i'd need to fish up some planning docs19:10
=== kadams54 is now known as kadams54-away
=== kadams54-away is now known as kadams54
=== kadams54 is now known as kadams54-away
=== kadams54-away is now known as kadams54
=== roadmr is now known as roadmr_afk
=== _thumper_ is now known as thumper
=== roadmr_afk is now known as roadmr
=== kadams54 is now known as kadams54-away
=== kadams54-away is now known as kadams54
lazyPowertvansteenburgh: did you kick off that test manually?20:45
lazyPoweri ask because the bug hasn't been updated, and i assume that is a byproduct of how the test was run20:45
tvansteenburghlazyPower: correct20:46
tvansteenburghlazyPower: i expected to see the results here though, not sure why they're not http://reports.vapour.ws/charm-summary/docker20:49
tvansteenburghjog might know20:49
* jog looks20:49
travnewmatichello all, i know there's a separate maas channel but i'm not getting any response on there21:10
travnewmaticcan anyone help me with enlisting a node?21:10
travnewmaticthe node boots from pxe, and ends up at an ubuntu login prompt, but it doesnt appear as a node in my maas contron panel on either the region controller or cluster controller21:11
travnewmatici can ping the pxe'd node21:12
tvansteenburghlazyPower: http://reports.vapour.ws/charm-summary/docker21:35
=== menn0_ is now known as menn0
marcoceppitvansteenburgh: that's awesome dude21:51
tvansteenburghmarcoceppi: what's awesome?21:53
marcoceppitvansteenburgh: that URL21:53
tvansteenburghah, yeah, thank jog :)21:53
marcoceppijog: sweet url21:54
marcoceppiawesomeee http://reports.vapour.ws/charm-summary/mysql21:56
tvansteenburghmarcoceppi: yeah, mysql has a long history of passing proof and lint!21:56
marcoceppi\o/21:56
tvansteenburghlol21:57
=== kadams54 is now known as kadams54-away
=== thumper is now known as thumper-dogwalk
mwako/22:44
=== thumper-dogwalk is now known as thumper
=== kadams54-away is now known as kadams54

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