Tugwhen adding a relation to 2 services, events "joined" and "changed" are not synchronized between the 2 services. For instance my hooks execute in that order :00:38
Tugservice1-relation-joined > service1-relation-changed > service2-relation-joined > service2-relation-changed, but what I was expecting would be to have both "joined" hooks completed before running "changed" hooks.00:38
Tugthe issue here is that "joined" hooks execute 'relation-set' commands so we have missing entries in the relation list when running "service1-relation-changed"00:42
TugI'll try to ask you guys this during the day ;)00:42
tvansteenburghTug: this is covered in the docs here: https://juju.ubuntu.com/docs/authors-relations-in-depth.html00:43
tvansteenburgh"In one specific kind of hook, this is easy to deal with. A relation-changed hook can always exit without error when the current remote unit is missing data, because the hook is guaranteed to be run again when that data changes -- and, assuming the remote unit is running a charm that agrees on how to implement the interface, the data will change and the hook will be run again."00:43
tvansteenburghso in your relation-changed hook, do a relation-get, and if the value is empty (not there), just exit 0, knowing that your hook will be run again when the data /is/ there00:44
Tugthx tvansteenburgh, yeah I skipped some parts of the docs00:44
tvansteenburghTug: no worries00:45
Tugwell I did that (actually the mongodb charm does that) but I think it fails00:46
TugI mean it is not run again00:46
Tugthe mongos/0 unit is in error state 'hook failed: "mongos-cfg-relation-changed"'00:48
Tugrunning resolved -r indeed works00:48
tvansteenburghTug: yeah ok, if a hook fails all events are paused until the failure is resolved00:48
tvansteenburghdo you know what caused the failure?00:49
Tugyeah the relation-set00:49
TugI mean get00:49
Tugmongos_relation_changed: relation data not ready.00:49
Tugmongos_relation_changed returns: False00:49
TugERROR juju.worker.uniter uniter.go:490 hook failed: exit status 100:50
tvansteenburghTug: if you pastebin the code i can try to help00:50
tvansteenburghbut in general, don't use `relation-get` in a config-changed hook00:51
tvansteenburghb/c you can't be sure that the relation is established00:52
tvansteenburghonly use it in relation hooks00:52
TugI did not change too much the original script00:55
Tugbut what we see here is that configsvr_relation_joined() can happen after mongos_relation_changed()00:56
Tugthus causing relation_gets to return None00:56
Tugline 1337 (^^)00:56
tvansteenburghTug: yep, that is a normal scenario00:57
sarnoldwould a correct fix be to add a new line after 1338 that sets retVal = True  ?00:58
tvansteenburghsarnold: yes00:58
Tugnice :)00:58
Tugreally ?00:58
tvansteenburghTug: yep00:59
TugI have to try !00:59
tvansteenburghTug: it's not an error for relation data to not be ready. if that it the case you must return True (exit 0) so that juju will continue processing events, and eventually run your hook again when it does have the data01:00
Tug(exit 1 you mean ?)01:02
Tugalright, I'm running the fix :)01:02
tvansteenburghTug: no, exit 001:04
tvansteenburghexit 0 = success, exit > 0 = fail01:05
Tugreally ? I always type exit 1 with debug-hook to say it ended correctly01:05
tvansteenburghsee lines 1660 - 166301:05
Tugok I was wrong then, good to know :)01:05
Tugah yes01:06
tvansteenburghTug: glad to hear you've discovered the joy of debug-hooks though!01:06
Tugyeah thanks to lazyPow3r a few weeks ago01:07
lazyPowertvansteenburgh: how do i get root path system independently? i thought os.path.abspath() would return it...01:08
Tugbtw, I just spent a few days improving the mongodb charm, it might help the community later ;)01:08
lazyPowerbut its including the CWD01:08
lazyPowerwhich i do not want.01:08
tvansteenburghlazyPower: i don't understand what you want it to return01:09
tvansteenburghTug: that's awesome!01:09
lazyPowertvansteenburgh: if i say os.path.join('foo','bar') using abs path, iw ant '/foo/bar' to be the return path.01:09
tvansteenburghlazyPower: yeah, you'd have to os.path.join('/', 'foo', 'bar')01:10
lazyPowerthat kind of defeats the idea of using os.path.join though...01:10
tvansteenburghthere must be a better way, that's not portable01:10
tvansteenburghwell do abspath on foo first01:10
Tugthanks tvansteenburgh, sarnold it worked01:11
lazyPoweri think abspath always returns from CWD01:11
joselazyPower: now that you're on review, I've got a charm that got a +1 from a charm-contributor and a non-reviewing charmer01:12
tvansteenburghTug: great!01:12
lazyPowerjose: offduty :P01:12
lazyPoweri've been on duty when when not duty'ing01:12
* lazyPower takes a vacation to work on other projects01:12
sarnoldTug: nice :)01:12
* jose flips table01:13
* lazyPower dangles carrots in front of jose01:13
* tvansteenburgh laughs and points01:13
lazyPowerjose: which charm? i'll look at it tomorrow01:13
lazyPoweri'm assuming seafile?01:13
joseyou got it right01:13
lazyPoweri figured, i saw extra work in that one this week01:13
joselooks like you've read your emails01:13
lazyPowerhalf hour in the morning, half hour at the end of the day01:14
tvansteenburghlazyPower: i still don't get what you're trying to achieve. are you making a path to a dir that doesn't exist?01:14
lazyPowertvansteenburgh: yep01:14
lazyPoweri want to make /vagrant/charms/precise01:14
lazyPowerwhcih will not exist by default01:14
lazyPoweri mean, i could get cheap, and just make it on the host... but thats not a fair assumption to make. the script should drive all actions and make assertions.01:15
josesudo mkdir /vagrant01:15
lazyPowerthis is the equivalent of a nose test for testing the juju vagrant image.01:15
tvansteenburghi guess that's the heart of what you want01:15
lazyPowertvansteenburgh: look at the second answer01:15
* lazyPower edited that like, 5 minutes ago01:15
tvansteenburghhey, nice01:16
lazyPowerand its such a hack. you're finding the path to the interpreter, which may or may not be correct on windows.01:16
lazyPowerwhat if vagrant mounts on C:\\ but the python interpreter is on d:\\01:16
tvansteenburghyour answer isn't a hack01:16
lazyPowerits not solid though. i just want the root of the current filesystem.01:16
lazyPowerthats the only safe assumption i'm willing to give with windows01:17
lazyPoweri guess this works :|01:17
* lazyPower resigns to using his own hack01:17
lazyPowera hack, using hacks, to produce hacky scripts01:17
* lazyPower hackety hack hacks01:17
lazyPowertvansteenburgh: thanks though, appreciate the extra braincells @ the problem.01:18
* tvansteenburgh was not much help01:19
* tvansteenburgh wanders off to eat pie, waving as he goes01:19
sarnoldI thought on windwos you just shoved everything into C:\windows\system32\ and called it a day? :)01:19
lazyPowersarnold: duh01:20
lazyPoweronly real lusers put stuff elsewhere01:20
lazyPowermaking it easy to remoev01:20
sarnoldor backup01:20
lazyPoweri mean, software is so great on that platform why wants to remove it?!01:20
lazyPowerhey sarnold01:48
sarnoldevening lazyPower :)01:48
lazyPowerhttp://askubuntu.com/questions/465544/what-is-the-reason-that-i-see-cron-session-opening-and-closing-every-hour-in-va <- this is a good question. Why DOES this happen?01:48
sarnoldlazyPower: heh that is a decent question :)01:49
lazyPoweroh man, did i stump you?01:49
sarnoldheh, no, I'm just saying that for a beginner it'll be utterly impenetrable with no clue where to go for finding the answer :)01:50
sarnoldhis or her guess is utterly adorable :)01:50
lazyPoweri keep hoping i'll find an area of grey knowledge and stump you, its become quite the fun game to play.01:50
lazyPowerbetter know the day it happens i'm pooping the cork on the champagne.01:51
lazyPowerwhat a typo wow01:51
lazyPowercontext... it is everything.01:51
sarnoldhaha, look closer to home -- I know nearly nothing about Go. I spent two hours just trying to figure out how to do cscope-like things in it that didn't involve "Step 1: install a Java Servlet Container"01:51
lazyPowerbut, go isn't home here01:53
lazyPoweri work with pretty much everything *but* go01:54
sarnoldwell, I guess if you're just using the API of juju, it wouldn't be your regular stomping grounds either..01:54
lazyPoweruntil there's a go charm in the store.01:54
lazyPowerthen i'll be like "yo dawg"01:54
lazyPower"whats up with this go code?"01:54
sarnoldand I'll guess my way through it :)01:55
sarnoldthe answer there isn't too bad, but if a better one isn't posted when I'm done with dinner, I'll write a -good- one :) hehe01:56
lazyPowerlooking forward to it01:56
=== axw is now known as axw-lunch
lifelessmkdir: cannot create directory �/var/run/rabbitmq�: Permission denied04:02
lifeless/usr/lib/rabbitmq/bin/rabbitmq-server: 80: /usr/lib/rabbitmq/bin/rabbitmq-server: cannot create /var/run/rabbitmq/pid: Directory nonexistent04:02
lifelessbah, echannel04:03
cruisibesareshey guys im testing out juju and maas. So far things are pretty awesome. I also have an aws cluster and i have set that up with a different name in my environments file. Do i need to run an instance of juju for each envionrment with juju bootstrap or is there a way that I can have one gui/juju server that will manage both providers? If you need one juju node per provider can it be colocated with maas? I found this http://askubuntu.04:20
cruisibesarescom/questions/181880/does-each-juju-environment-specified-require-its-own-master-node but im wondering if anything has changed at this point in juju's development. I think that it would be really nice to be able to join all my clouds with one orchestration tool. If its not there now is on the roadmap or am i missing something fundamental?04:20
=== CyberJacob|Away is now known as CyberJacob
=== CyberJacob is now known as CyberJacob|Away
=== CyberJacob|Away is now known as CyberJacob
=== vladk|offline is now known as vladk
=== zer0 is now known as Guest93246
=== vladk is now known as vladk|offline
=== vladk|offline is now known as vladk
CuyHi! I'm looking into Juju for service orchestration and the planned deployment of an OpenStack cloud (possibly in combination with Saltstack). Could anyone tell me how resource hungry Juju is? As in: How many servers can I realistically expect to steer from one master? And is there some infrastructure size you would consider a hard limit of Juju's capabilities?11:33
CuySorry for asking here, but I didn't find any information on these topics anywhere on the net (and I've been looking into service orchestration and configuration management for a few weeks now ;) )11:36
=== magicrob1tmonkey is now known as magicrobotmonkey
gnuoyIs it possible to write amulet tests that interrogate the relation sentry for charms which are not yet in the charm store ? The reason I ask is that I'm getting a "request failed with: 404" and it seems to be trying to query  https://manage.jujucharms.com/api/3/charm/...14:16
lazyPowergnuoy: you can specify a launchpad branch to deploy from.14:29
lazyPowerare you using amulet 1.5?14:29
gnuoyI am14:29
lazyPowerbueno, that *should* work.14:29
gnuoybranch: lp:~blah14:29
lazyPowerif it doesn't let me know and i'll ping the parties working on amulet.14:30
lazyPowerthat behaviour should have been triaged in 1.4.x of amulet14:30
gnuoylazyPower, thanks, I'll give it a whirl14:30
cruisibesaresHey all i have asked this question last night be im guessing everyone was asleep. I have a physical infrastructure that im running with maas and a cloud in aws. I would like to manage both of these providers with one juju server. Im looked over all the questions concerning juju and maas on ask ubuntu and i found this http://askubuntu.com/questions/181880/does-each-juju-environment-specified-require-its-own-master-node which seems14:43
cruisibesares to sugest that you need one juju machine per enviornment. Im wondering if that is still true or if what i want to do is possible using manual provisioning14:43
cruisibesaresi think my main concern is that juju isn't ha and i would like to keep my single point failure on aws14:46
cruisibesaresit has a higher uptime guarantee and more tools14:47
cruisibesaresthat and the point of the physical hardware im deploying is to be cheap and expendable14:47
lazyPowercruisibesares: we don't have cross environment relationships yet. You can do it with the manual provider, but as the provider name implies, there's manual effort involved in it.14:49
cruisibesares<lazyPower> thanks14:50
lazyPowerIts on the roadmap, but i don't have an ETA for X-environment implementation. So at best, you'll want to do 2 bootstrap nodes, one for your maas cluster one for aws, or go manual and do enlistment manually.14:50
lazyPowercruisibesares: sorry i don't have a better answer for you than 'its coming' :(14:51
cruisibesaresok great thats really helpful i will consider both of those options thanks so much14:51
cruisibesaresno thats totally fine i get that there is a lot on the roadmap14:51
lazyPowerYep, features galore are coming in this iteration. I think we have it up for the iteration after this - but that's uncertain. It's still mid to high priority though.14:52
lazyPowerare you on the mailing list? You'll get notices of what lands in teh changelog on the mailing list.14:52
cruisibesaresso now i will just have to gamble on which setup will have cleanest update path14:52
cruisibesaresno im not on the mailing list yet14:53
cruisibesaresthats on the main page for juju?14:53
lazyPoweryou may want to ping the list with your question for other community members that have taken one route vs another - and get feedback. Kind of a straw poll to speak - and see if their experiences align with your goals.14:54
lazyPowerI'm running a manual provider setup between Do and Softlayer that has been pretty solid.14:54
lazyPowerbut I'm a minority in that aspect.14:54
lazyPowerand working with a true bootstrap node in the environment that does enlistment for me - is sorely missed. Its not *that* big of a deal to manually enlist but if i were scaling > 10 machines, i'd want this all to be automated for me.14:55
cruisibesaresgreat idea14:55
lazyPowerGiven that you're running with maas and aws - both of which have full providers, it would probably be best to go that route and put some glue around it.14:55
lazyPowerdepending on your scaling future :)14:56
cruisibesaresalright well its good to know people are doing it manually14:56
cruisibesaresalright cool i'll try and send something to the mailing list soon14:56
cruisibesaresthanks for your ideas and help14:56
cruisibesareswill give me something fun to play with while i wait :)14:56
gnuoylazyPower, fwiw I've raised Bug#1319437 for the amulet issue15:03
_mup_Bug #1319437: Amulet breaks when inspecting the relation-sentry regarding charms not in charmstore <Amulet:New> <https://launchpad.net/bugs/1319437>15:03
lazyPowerthanks gnuoy! i'll poke the fellas and let them know15:04
gnuoynp, thanks for the help15:04
lazyPowergnuoy: from the creators mouth - <marcoceppi> it's doing the right thing, it assuems OH YOU'RE NOT A LOCALCHARM, LETS USE THE API HERP DERP15:07
lazyPowerso you found a valid corner case, it'll get addressed soon.15:07
gnuoylazyPower, it breaks in the same way if the charm is in a local directory rather than lp15:07
lazyPowergnuoy: should land in 1.5.115:08
marcoceppiyay, being quoted verbatium from another channel15:14
pindongahi, anyone around to help me with running juju locally? I'm running on trusty, but after an upgrade path from saucy15:34
pindongaI can't get juju to create the machines with the local provider properly15:34
pindongakeep getting: WARNING juju.worker.instanceupdater updater.go:231 cannot get instance info for instance "": no instances found15:34
pindongaI had lxc already set up previously, so I assume this must be an issue with lxc (mis)configuration15:34
cjohnstonwallyworld_: I'm a little confused.. is bug #1306537 fixed in 1.18.3 in the PPA?15:34
_mup_Bug #1306537: LXC local provider fails to provision precise instances from a trusty host <deploy> <local-provider> <lxc> <juju-core:Fix Released by wallyworld> <juju-core 1.18:Fix Released by wallyworld> <juju-quickstart:Fix Released by frankban> <juju-quickstart (Ubuntu):New> <juju-quickstart (Ubuntu Trusty):New> <https://launchpad.net/bugs/1306537>15:35
=== ted is now known as tedg
lazyPowersinzui: looks like apparmor is going to be a troublemaker this time around - https://bugs.launchpad.net/ubuntu/+source/apparmor/+bug/129638415:51
_mup_Bug #1296384: LXC apparmor profile broken w/recent trusty update <amd64> <apport-bug> <trusty> <AppArmor:Confirmed> <apparmor (Ubuntu):Triaged> <https://launchpad.net/bugs/1296384>15:51
sinzuilazyPower, looks like Ubuntu is still ignoring bug 130528015:54
_mup_Bug #1305280: apparmor get_cgroup fails when creating lxc with juju local provider <apparmor> <armhf> <local-provider> <lxc> <packaging> <regression> <juju-core:Invalid> <apparmor (Ubuntu):Confirmed> <https://launchpad.net/bugs/1305280>15:54
avoinepindonga: can you tell me what is your version of lxc + juju-core?15:55
pindongaavoine, lxc==1.0.3-0ubuntu3 , juju-core==1.18.1-0ubuntu115:56
cjohnstonpindonga: I wonder if your having bug #1306537 too15:57
_mup_Bug #1306537: LXC local provider fails to provision precise instances from a trusty host <deploy> <local-provider> <lxc> <juju-core:Fix Released by wallyworld> <juju-core 1.18:Fix Released by wallyworld> <juju-quickstart:Fix Released by frankban> <juju-quickstart (Ubuntu):New> <juju-quickstart (Ubuntu Trusty):New> <https://launchpad.net/bugs/1306537>15:57
avoinepindonga: what lxc-ls gives you?15:58
pindongaexit 0 (no output)15:59
pindongacjohnston, it's possible, I'm trying to deploy a charm for testing with default-series: trusty, and it looks like something is happening (taking it's time though)16:02
pindongaand it just failed with: (error: error executing "lxc-start": command get_cgroup failed16:02
pindonga      to receive response)16:02
cjohnstonthere's also bug #1317197 but I'm not sure that'd be it if your getting a cgroup issue16:04
_mup_Bug #1317197: juju deployed services to lxc containers stuck in pending <oil> <juju-core:Fix Committed by axwalk> <https://launchpad.net/bugs/1317197>16:04
=== jam2 is now known as jam1
pindongathx cjohnston the title looks promising, will take a deeper look after lunch16:15
_mup_Bug #1319474 was filed: Juju uses hard-coded regions <pyjuju:New> <https://launchpad.net/bugs/1319474>16:35
cjohnstonlazyPower: I have an ansible question for you if you have a moment16:36
_mup_Bug #1319475 was filed: Juju should support new signing format <pyjuju:New> <https://launchpad.net/bugs/1319475>16:38
didrocksjcastro: https://github.com/juju/docs/pull/97 mind having a look?16:49
jcastroon it16:57
jcastrodon't know how I missed that, thanks!16:57
lazyPowercjohnston: shoot - sorry about the delay was away from my desk.17:02
didrocksyw man ;)17:02
cjohnstonlazyPower: np..17:02
lazyPowersinzui: tyhicks in #ubuntu-server has confirmed the apparmor bug is duplicated by one that was supposed to be fixed in reprelease.17:03
lazyPowercjohnston: but fire when ready :)17:03
cjohnstonlazyPower: I have http://paste.ubuntu.com/7463732/ ... however, it's failing that 'user' isn't found.. I added the when because it was failing for the same reason when I didn't have the when..17:03
cjohnstonI'm wondering if there is something else I should be doing, or maybe if I should add some sort of repeat type something?17:04
cjohnstonIt seems like I'm just not getting back the relation data quick enough for ansible17:04
lazyPowercjohnston: "'user' doesnt' seem like it would expand. Ansible playbooks expand jinja2 variable syntax as of 1.617:04
lazyPowerso wouldn't it be {{user}}17:04
lazyPowerthey depreciated the older style of variable notation using $'s, and i think what you're seeing is related. but i'm not positive.17:05
cjohnstonI tried that.. but I got an error.. let me try it again and see if I can figure the error17:05
lazyPowercapture the error for me and i'll take a look.17:06
cjohnstonlazyPower: http://paste.ubuntu.com/7463825/17:27
lazyPowercjohnston: looking, 1 sec17:29
lazyPowercjohnston: where's your line defining user?17:29
lazyPowerthere sh9ould be a register play in your playbook that defines the username.17:29
lazyPoweror is it a config option on the charm?17:29
cjohnstonlazyPower: AIUI charmhelpers automatically makes everything in the /etc/ansible/host_vars/localhost available.. is that not the case?17:31
avoinecjohnston: have you tried without all the quotes single and double?17:36
cjohnstonavoine: based on http://docs.ansible.com/playbooks_conditionals.html#applying-when-to-roles-and-includes when: "'reticulating splines' in output" <-- is how I was operating17:38
cjohnstonIs that not correct?17:41
avoineI think when you use a variable name instead of plain text you don't need the single quotes17:41
cjohnston"{{ user }} in current_relation" <-- is what I just tried that gave the traceback I pasted17:42
lazyPowercjohnston: thats correct17:43
avoineI would try: user in current_relation17:44
cjohnstonlazyPower: which part are you saying is correct17:45
lazyPowerThjat the ansible helpers make all config values global keys in the playbook.17:45
lazyPoweri didn't know if that was coming from a config value or from your play17:46
=== vladk is now known as vladk|offline
lazyPowercjohnston: thats strange that its undefined though...18:21
lazyPoweri'm nto sure what to recommend here.18:21
lazyPowersyntax looks fine18:22
balloonswwitzel3, ah-hah, I've found you after all. heh. Glad I didn't bike, would have been a wet ride18:39
cjohnstonlazyPower: well.. avoine's suggestion didn't cause an error18:39
=== CyberJacob is now known as CyberJacob|Away
tvansteenburghballoons: o/ (Tim from lunch)18:48
=== CyberJacob|Away is now known as CyberJacob
=== vladk|offline is now known as vladk
onezeroTrying to deploy using the "manual" environment via which my bootstrap node will be in an already existing "juju-server" machine.  How do I change the port.  I see bootstrap-host: however there is no "bootstrap-port:" and appending the port at the end of the hostname doesn't work.  Any ideas?19:20
marcoceppionezero: it's not a good idea to bootstrap the same server twice19:20
onezeroI'm using a docker container.19:21
marcoceppionezero: because, juju isn't designed to host more than one bootstrap on a single machine19:21
marcoceppiyou can deploy services to the bootstrap node, but a single node can't have more than one bootstrap running on it19:21
marcoceppionezero: I think there might be a communciation issue between bootstrapping and deploying19:23
onezeroIn my case a vm is entirely dedicated as a juju server and that's the only function it serves.  It seems like a waste to create a whole new vm for each additional environment...19:23
marcoceppionezero: well, you can deploy services to that VM, but that one VM also controls any and all other VMS you wish to spin up in that environment19:24
marcoceppiit's the orchestration service for that environment19:24
onezeroSo basically, if I want to orchestrate multiple environments... I need multiple actual servers.19:25
onezeroOne juju server per environment19:25
marcoceppionezero: you need atleast one server per environment19:26
marcoceppian environment can have 1+ severs which can have n+ charms deployed on to it (either using the whole machine or using containerization kvm, lxc, etc)19:27
marcoceppion to it, being on to any of the servers in the environment19:27
marcoceppiwith manual provider you can enlist an additional machine by running juju add-machine <user>@<machine>19:28
marcoceppiwhich will make it available in that environment19:28
onezeroI just wanted to keep the orchestration (ie juju bootstrapped node) separated from the actual servers that are being orchestrated.  Basically I have a maas cluster which is my first environment for which I bootstrapped a single vm to serve only that singular purpose of juju orchestration.  Next I have some servers NOT part of maas that I would like to manually deploy charms to... but in order to do that I created a new manual environment in my yaml config.  I19:32
marcoceppionezero: yeah, so you want something like cross environment relations, which is on the roadmap but probably won't happen in the next few months19:34
marcoceppionezero: you can create a KVM or LXC on the maas bootstrap node19:34
marcoceppiand use that as your manual provider bootstrap node19:34
marcoceppithe main problem is there's not isolation, so juju-db will stomp all over itself19:35
marcoceppibut if you put the bootstrap node in a container inside an existing bootstrap node, there's no real collison19:35
onezeromarco... THANKS a bunch.  Glad to get confirmation of that.19:42
rick_h_mbruzek: around?19:56
rick_h_hey, see PM19:57
l1lOk folks, could use some help..20:12
l1lThis is a error I get when trying to view a juju status -e maas20:12
l1lERROR state/api: websocket.Dial wss://tngek.maas:17070/: dial tcp: lookup tngek.maas: no such host20:13
josel1l: afaik that's because bootstrap hasn't finished20:16
l1ljose; The machine is already booted, and I can ssh into it.20:16
josel1l: bootstrap does some additional things as it needs some tools to be the master :)20:17
josedid your 'juju bootstrap' finish?20:17
l1lYes, it finished with no errors and the node booted up. However, when I try to check the juju status I get that error repeating20:19
josehmm, that's weird20:21
josemaybe someone else would be able to help20:21
l1lI have googled till blue in the face and have found similiar bugs, but they see a "connection refused" instead of the "no such host". It's apparently related to DNS.20:22
l1lThanks though!20:22
josenp :)20:22
andreas__l1l: add "nameserver <maas-ip>" to the top of your /etc/resolv.conf temporarily20:31
=== andreas__ is now known as ahasenack
ahasenackthere might be a way to tell your local resolver to only use that DNS for the .maas domain20:32
ahasenackdnsmasq has the --server option which seems to suit well, but I haven't used it20:35
l1lahasenack; Adding that to the resolv.conf fixes it. So time to point the finger at maas-dns ?20:37
ahasenackl1l: no20:37
ahasenackl1l: maas is controlling that zone, so you need to use it for dns when talking to machines in that zone20:38
ahasenackit's as simple as that20:38
ahasenacknow, that change you just made should be temporary, because that means all your other name resolution queries will go to maas, even the ones that have nothing to do with maas20:38
ahasenacklike google.com, gmail, etc20:38
l1lYea, wonder how I can make maas just use that zone.20:39
ahasenacksee man dnsmasq, look for the -S option20:39
ahasenackand I think you can add a similar option to /etc/dnsmasq.conf, but I haven't tried20:40
ahasenackwould be a way to tell your local resolver, assuming you are on ubuntu and using dnsmasq (the default), to only use the maas dns for resolving names in the .maas domain20:40
ahasenackthat would be on your machine, where you are issuing juju commands, btw20:40
ahasenackthe maas nodes are already using the right dns20:41
l1lodd, I dont have that dnsmasq.conf20:41
ahasenackI think ubuntu works with snippets in dnsmasq.d20:43
ahasenackthere might be ubuntu specific documentation about this20:43
ahasenackI also see a /etc/dnsmasq.d-available/20:43
l1lhmm, I don't see anything todo with dnsmasq in the /etc dir20:44
ahasenackI'm on trusty, if that makes a difference20:47
ahasenackoh, and on a desktop20:47
ahasenack$ dpkg -S /etc/dnsmasq.d20:47
ahasenacknetwork-manager: /etc/dnsmasq.d20:47
ahasenackyou might not have network-manager20:47
l1ldoes a server install not get the network-manager?20:51
=== jeremydei_ is now known as jeremydei
=== vladk is now known as vladk|offline
=== CyberJacob is now known as CyberJacob|Away

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