MACscris there a static config i can use to set a default for amazon instance sizes? I only want to use tiny instances to start00:28
davecheneyMACscr: use bootstrap constraints00:28
davecheneyor deploy constriants00:28
MACscralso, just to verify, if i want to have 4 amazon instances in 4 regions, i have to have a controller in each region?00:28
davecheneyMACscr: not at this time00:30
davecheneywhat we call 'provider specific' constraints are being worked at this time00:30
davecheneythere is currently no way to say 'this unit must be in a specific availability zone'00:30
davecheneywe know it's a problem00:31
davecheneywe're working on fixing it00:31
sarnoldI think to have four regions under control at once, you have four separate environments in your environments.yaml -- and none of them know anything about any of the others, right?00:31
davecheneysarnold: regions are easy00:31
_mup_Bug #1160667: Expose regions and availability zones to users <juju-core:Triaged> <https://launchpad.net/bugs/1160667>00:31
davecheneyavaliability zones within a region areharder00:31
=== defunctzombie is now known as defunctzombie_zz
MACscrwell to be honest, i dont really need them to be aware of each other as they are just going to be dns servers with a little ping monitoring on them as well (smokeping)00:33
MACscrjuju would just be used to deploy ubuntu on them. I guess i could manually install mysql on them00:34
sarnoldMACscr: you either use --to to deploy a mysql charm onto them, or write a subordinate charm that installs mysql alongside the dns servers..00:35
MACscrok, so you and dave seemed to either have conflicting views or i just got a bit confused. Are having multiple regions easy or not? Basically I need to setup an instance in Asian, Europe and two in the US00:38
MACscrand the two in the US are going to be in two different amazon regions (east cost and west coast)00:39
=== weblife is now known as web-brandon
sarnoldMACscr: I think we both said the same thing from different angles :)00:41
sarnoldMACscr: you can do multiple environments and then juju -e asia deploy dns, juju -e europe deploy dns, etc... but they can't do relations between asia and europe00:44
MACscrsarnold: understandable. This might be an option in the future though? cross region relations?00:46
MACscralso, how do i setup multiple regions/environments for a single cloud? comma separate list?00:47
sarnoldMACscr: different environment declarations in your environments.yaml00:48
marcoceppiMACscr: there is talk on eventually having cross environment relations, not sure where on the roadmap that is ATM00:48
marcoceppiMACscr: you'll need to create a new environment for each region in your environments.yaml00:48
marcoceppiJust name them uniquely, and make sure each has a unique bucket-name. They can share the same credentials00:49
MACscrso do i just copy and paste the amazon group of configs and then just change the region? Im not sure how to setup multiple environments for the same provider00:50
davecheneyMACscr: yes, we call these cross environment relations01:00
davecheneybasically relations between services in different enironments01:01
davecheneyit's on the roadmap01:01
davecheneyi can't give you a solid idea when it will happen01:01
MACscrso not next monday?01:01
marcoceppiMACscr: here's an example01:03
marcoceppiof multiple juju environments01:03
* marcoceppi scrubs his environments yaml for MACscr01:10
marcoceppiMACscr: http://paste.ubuntu.com/6030850/01:12
marcoceppiYou just need to change the environments name (us-east, blah-lbah, whatever you want), the region, and the control bucket01:12
marcoceppiEverything else can be the same01:12
marcoceppiMACscr: and the region names, http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/using-regions-availability-zones.html#concepts-regions01:14
davecheneyMACscr: i have an environment in my environments.yaml for ever ec2 region01:26
davecheneyso I can do things like01:27
davecheneyjuju bootstrap -e ap-southeast-101:27
=== defunctzombie_zz is now known as defunctzombie
=== defunctzombie is now known as defunctzombie_zz
=== defunctzombie_zz is now known as defunctzombie
MACscrdavecheney: thanks!03:08
MACscrok, so i know juju-ui can be used for creating relations, etc, but what about managing them all after they are deployed? id like a single panel for all systems, vm or physical03:41
MACscrlandscape pricing is just to high imho03:43
x-warriorwhat is the parameters to create a micro instance? sorry to ask this again05:12
davecheneyx-warrior: for the ec2 provider, you'd use constraints05:14
=== freeflying is now known as freeflying_away
* davecheney does a test before recommending anyting05:14
x-warrioryeap it is constraints05:15
x-warriorbut I cant remember which ones05:15
x-warriorI asked this early today but I couldn't find a channel log :/05:15
davecheneyx-warrior: i'm just doing a demo now05:17
davecheneytwo secs05:17
x-warriorI tried with cpu-power=005:19
=== freeflying_away is now known as freeflying
x-warriorbut I get this error: "ERROR juju supercommand.go:235 command failed: cannot start bootstrap instance: cannot set up groups: cannot revoke security group: Source group ID missing. (MissingParameter)05:19
davecheneyx-warrior: ewww05:19
davecheneynice error05:19
davecheneywill try to reproduce that one05:19
davecheneystill working on your first question05:20
x-warriorI think that came from, starting a regular bootstrap... deleting s3, deleting ec2 machine and trying to create it again with constraints...05:20
x-warriormaybe it loose the id from groups and stuff like that to 'revoke' and 'recreate' or something05:20
davecheneyx-warrior: hmm, if you are rough with juju it probably won't respect you in the morning05:22
x-warriorremoving the juju-amazon groups05:25
x-warriorseems to solve the problem05:26
davecheneyx-warrior: I think it would be05:26
davecheneyjuju bootstrap  --constraints="arch=i386 mem=640M"05:26
davecheneybut I haven't been able to check yet05:26
davecheneyhaving some other problems at the moment05:26
davecheneyyou can't say type=t1.micro05:26
davecheneybecause we don't support what we call `provider specific constraints`05:26
x-warriorusing 'juju -v bootstrap --constraints="cpu-power=0"'05:27
davecheneyso you need to describe something that looks like a t1 micro05:27
x-warriorcreated the micro instance05:27
davecheneyyup, because there is nothing that is cpu-power005:27
davecheneyso the next largest is a t1.micro05:27
davecheneysame thing05:27
davecheneywe just did it a different way05:27
x-warriornow I see the instance running on my panel, but when I check juju status it gives me 'no instance running'05:28
davecheneyx-warrior: i can only recommend juju destroy-environment05:29
davecheneybecause you've probably damaged some invariants that juju was expecting05:29
x-warriorany other information I could provide? I'm able to connect to instance via ssh05:29
davecheneyjuju status -v05:30
x-warriorI can see '15088 ?        Ssl    0:00 /var/lib/juju/tools/machine-0/jujud machine --log-file /var/log/juju/machine-0.log --data-dir /var/lib/juju --machine-'  on server05:30
x-warrioron instance*05:30
davecheneywhat has probably happened is the control bucket provider-state file does not match the intsance id of your bootstrap node05:30
x-warrior2013-08-27 05:30:37 INFO juju ec2.go:128 environs/ec2: opening environment "amazon", 2013-08-27 05:30:43 ERROR juju supercommand.go:235 command failed: no instances found05:31
davecheneypop open the s3 console and get the contents of the provider-state file from your control bucket05:31
davecheneyi suspect it is missing or empty05:31
x-warriorshould i paste it to pastebin?05:32
davecheneyright, so does that instance number match the machine that is running ?05:32
x-warrioryes it does05:33
x-warriorrebooting the instance and checking juju status again, gives me the same result05:39
davecheneyx-warrior: can you paste the output of juju status -v05:41
davecheneyi supect it will error very early05:41
x-warriorthat is it05:42
x-warriortoo short I guess05:42
davecheneyx-warrior: so juju looks in the control bucket, gets the instance id of the machine05:44
davecheneyconverts it to an ip address05:44
davecheneyuses that ip to talk to mongodb running on the bootstrap node05:44
davecheneyfor whatever reason that instance id, or the yaml is invalid05:45
davecheneyso that is all she wrote05:45
davecheneyx-warrior: two secs, checking something05:45
x-warriorI'm not sure that I'm following, but ok05:45
x-warriorI'm trying to start without constraint option, to check if it is something related to micro instance05:46
x-warrioror something05:46
davecheneyx-warrior: short version is05:47
davecheneythat environment is broken05:47
davecheneyyou will probably have to delete it via the aws console and start again05:47
x-warriordelete ec2, security groups and s3 is enough?05:48
x-warrioror it writes to some other place?05:48
MACscrok, a controller is needed for each region/environment, right?  If so, is there any type of panel to keep track of everything as a whole?05:48
davecheneydelete the control bucket05:48
davecheneyand the instances you have lying around05:48
davecheneyMACscr: you need a bootstrap node per environment05:49
davecheneyand an environment can only cover one provider05:49
davecheneyso in effect you need one bootstrap node per ec2 region05:49
x-warriorwhat do you mean by control bucket? bootstrap instance?05:49
davecheneycontrol bucket is listed in your ~/.juju/environments.yaml05:50
davecheneyit is where juju records persistant state05:50
davecheneybootstrap instance is that machine that you have left running05:50
davecheneyit is the machine that juju spawns to host the mongodb05:50
MACscrdavecheney: right, a bootstrap, make sense. So while juju helps get things deployed and related, is there not a tool for managing things after the fact?05:50
davecheneyMACscr: you'd have to define after the fact05:51
davecheneywe have the gui where you can view your juju environment05:51
x-warriorthere is juju-gui05:51
davecheneyand commands like add-unit/remove-unit help you scale up and scale down the number of units of a particular service05:51
davecheneybut juju does not compete in the nagios/zabbix/xenoss space as a process monitoring tool05:52
x-warriorand you can deploy more then one service to the same machine now05:52
davecheneyx-warrior: --to should be used with care05:52
davecheneyit takes all the safety guards off05:52
MACscrok, but can juju-gui work with more than one environment? its restricted to a single one just like how a bootstrap is needed for each one. Correct?05:52
davecheneyMACscr: yes, correct05:53
davecheneyeach environment is separate and unrelated05:53
davecheneythe juju client and switch between environments with the -e, or juju switch commands05:53
davecheneybut the juju-gui, being a charm itself, is deployed into an environment05:53
MACscrright, so something is need to manage everything05:53
davecheneyso only controls that environment05:53
MACscrnot just a single environment at a time05:53
davecheneyMACscr: we have no product for that at this time05:54
x-warriors3 cleaned up, instance terminated, .juju deleted... starting from scratch now05:54
MACscrnot when it comes to relational stuff, but general instance management, etc05:54
davecheneyMACscr: juju talks about services and units of services05:54
MACscrdavecheney: yeah, seems like landscape is the only option of yours and its ridiculously priced05:54
davecheneythe fact that it creates machines to host them is sort of co-incidental05:54
davecheneyMACscr: i cannot comment on the price, but you have certainly interpreted our marketing message correctly05:55
MACscrhmm, well foreman with puppet can manage everything, but im trying not to have a bunch of overlap with tools05:55
MACscrand i really havent figured out puppet yet =P05:56
x-warriorafter bootstrapping should I wait a while, to zookeeper and stuff like that goes 'up'?05:57
x-warrioror get installed or something?05:57
davecheneyx-warrior: we don't use zookeeper anymore05:58
davecheneywe use mongodb05:58
davecheneybut the result is the same05:58
x-warriorah sweet05:58
davecheneyjuju status will block until the bootstrap node is up and running05:59
davecheneyyou can see tht with05:59
davecheneyjuju status -v05:59
davecheneyin fact, you should pass -v to everythig that you do with juju05:59
davecheneyotherwise you'll have to rerun the command with -v anyway05:59
x-warrioryeap I learned that05:59
davecheneywe are working on our logging05:59
davecheneyit needs fixing05:59
davecheneywe're not done yet06:00
x-warriorno problem :D06:00
x-warriorthat is all the outputs from a 'fresh' start... (deleted s3, groups, instance, .juju files)06:00
davecheneyx-warrior: something is very wrong with your setup06:00
davecheneywhich juju06:00
davecheneyit smells like you have both 0.7 and 1.12 installed06:01
x-warriorat least juju version gives me that06:01
x-warriorand this is the first time I install juju (like 1 hour ago)06:01
davecheneyhmm, i'm a bit stumped06:01
x-warrior(on this computer ofc, I was trying on a mac before, but I had the same issue... so I thought it was a mac os related problem... then I moved to ubuntu...)06:02
davecheneycan you confirm that i-2a4ed636 is running in the aws console06:02
x-warrioryes I can see it06:03
x-warriorI can connect to it as well06:04
davecheneyi cannot explain why this is not working for you06:04
davecheneythe logic is06:04
davecheneyget the instance id from the provider-state file in the control bucket06:04
davecheneylooking the ip that the instance points too06:04
x-warrioris there a -vv option or something?06:04
davecheneythen connect to mongodb on that ip06:04
x-warriorwhich gets more verbous?06:05
davecheneyx-warrior: there is --debug06:05
davecheneybut I don't think it will make it much more verbose06:05
davecheneyand it is fialing at the first step06:05
davecheneyit's rejecting your provider-state file in the control bucket06:05
davecheneyi do not know why06:05
davecheneyi have not seen this failure mode before06:05
davecheneyx-warrior: just for shits and giggles06:06
x-warrior17070/37017 are the corrects ports?06:06
davecheneycould you change the region: key in your enviornments.yaml to another region06:06
x-warrioryes I can try that06:06
davecheney(after deleting the current environemnt of course)06:06
davecheney37017 is the correct port06:06
davecheneybut you don't get that far06:06
x-warriorso that seems very weird I guess06:07
davecheneyi have not seen that failure mode before06:08
x-warrior deploying to another region06:09
=== tasdomas_afk is now known as tasdomas
x-warriordavecheney, changing the region 'fix' a little it goes a little bit further06:19
davecheneyfix ?06:19
kurt_davecheney: do you know this error? error: cannot create bootstrap state file: gomaasapi: got error back from server: 400 BAD REQUEST06:19
davecheneykurt_: i'm not a maas expert06:20
davecheneyi mainly do the public clouds06:20
davecheneylet me check06:20
kurt_googling it now...06:21
davecheneyrough guess there is a permission problem creating or reading your control bucket06:21
kurt_I believe it is this: https://bugs.launchpad.net/maas/+bug/120450706:22
_mup_Bug #1204507: MAAS rejects empty files <verification-done-precise> <verification-needed> <MAAS:Fix Committed by rvb> <MAAS 1.2:Fix Committed by rvb> <MAAS 1.3:Fix Committed by rvb> <maas (Ubuntu):Fix Released> <maas (Ubuntu Precise):Fix Committed> <maas (Ubuntu Quantal):Confirmed> <maas (Ubuntu Raring):Fix Committed> <https://launchpad.net/bugs/1204507>06:22
kurt_I think jcastro warned me about this this morning06:22
x-warriorwell it is not a 'bug fix' but it is going a little further... I have no idea what is different besides the region...06:22
kurt_but some of the other folks thought it was fixed06:23
x-warriorI was using sa-east-1 region which is São Paulo, BR region... maybe some inconsistence between regions? :S06:23
davecheneykurt_: it's a known bug in maas06:24
davecheneyif you switch your maas install to the daliy build06:24
davecheneyit is fixed there06:24
davecheneyi do not have a timeframe when the fix will be avilable in general06:24
kurt_ok, was just chatting with bigjools too06:25
davecheneyx-warrior: up, thatis normal operation for ec206:25
davecheneyit takes 3-5 mins for each instance to start up06:25
davecheneyonce it is ready status will succeed, that is why it is retrying06:26
x-warriornow it listed the bootstrap machine06:26
x-warriorshould the destroy-environment option06:27
x-warriordestroy the instance and do a correct cleanup?06:27
davecheneyx-warrior: which region did not work06:27
davecheneyand which region did work06:27
davecheneydestroy-environment will remove all regions and the control bucket06:27
davecheneyit might leave the security groups around06:27
davecheneythat is fine06:27
davecheneythey are not expected to be deleted and can cope with being reused06:28
x-warriorso, now I'm using the default region option06:28
x-warriorwithout setting it on environment.yaml file06:28
x-warriorand using region: sa-east-106:28
x-warriorit does not work06:28
davecheneyok, so there is something wrong with the sao paulo region atm06:28
davecheneyit happens06:28
davecheneyap-southeast-2 was broken for several months for me06:29
davecheneyx-warrior: if you would care to, you should log a but about this on juju-core06:29
davecheneyalthough ec2 will denyu it06:29
davecheneyeach region is subtly different06:29
x-warriorwhere should I log it?06:29
x-warriorok, I will log that later today06:30
x-warriorbtw I will keep joining this channel06:32
x-warriorif you guys need more help to trace it06:32
x-warriorI will be glad to help you06:32
davecheneyx-warrior: thanks for the offer06:32
davecheneypointing the finger at sa-east-1 is good enough for now06:32
x-warriordavecheney, ok, I will do that when I wake up later... need to get some sleep now, almost 4am here06:39
x-warriorthanks for all the help06:39
x-warriorhave a good one06:39
davecheneyx-warrior: ok, thanks06:40
=== defunctzombie is now known as defunctzombie_zz
=== defunctzombie_zz is now known as defunctzombie
=== defunctzombie is now known as defunctzombie_zz
=== defunctzombie_zz is now known as defunctzombie
=== defunctzombie is now known as defunctzombie_zz
=== defunctzombie_zz is now known as defunctzombie
=== defunctzombie is now known as defunctzombie_zz
vdsstub, hello, can I ask you how to use the persistent volume support of the postgres charm? That's how I changed the config http://paste.ubuntu.com/6030130/08:57
vdsthe volume exists already, of course08:58
stubvds: The bit of the charm I'm not familiar with :) I can try, or invoke our devops if needed.08:59
vdsstub, who's the one to blame? :)08:59
stubvds: You are modifying config.yaml, instead of passing configuration parameters to the charm?09:00
vdsstub, yes09:00
stubvds: I think you are supposed to use a config file looking like http://paste.ubuntu.com/5751886/, and then do 'juju deploy --config=myconfig.yaml cs:postgresql'09:04
gnuoyhi vds, whats the issue ?09:07
vdsstub, thanks I'll try09:07
vdsgnuoy, when I try to deploy postgresql changing the config this way http://paste.ubuntu.com/6030130/09:08
vdsgnuoy, I get this error http://paste.ubuntu.com/6030125/09:08
mthaddoner, you're changing the config directly in the charm rather than passing it as an option to the charm?09:09
gnuoylooks like the version param is missing ./09:09
gnuoyHOOK KeyError: 'version'09:09
vdsmthaddon, is that bad?09:20
mthaddonvds: yes, you shouldn't change the charm itself before deploying09:21
vdsmthaddon, ok, thanks.09:23
=== tasdomas is now known as tasdomas_afk
=== tasdomas_afk is now known as tasdomas
alejandrohola spanis?11:12
alejandropara que es esto?11:17
varudAnybody have thoughts on why an experimental 12.04 image I'm creating with juju has '    agent-state: down' after restart?11:18
varudFound this old stackoverflow - http://askubuntu.com/questions/218645/juju-instances-in-agent-state-down-after-turning-them-off-and-back-on-on-ec211:19
varudbut it's not relevant anymore (there's no juju-machine-agent.conf file)11:20
=== natefinch is now known as natefinch-afk
mthaddonjcastro: I can't find any docs on the upgrade-charm hook any more - is that expected?14:21
marcoceppimthaddon: no, not expected. Let me see if I can find them. If not what's your question, I'd be happy to answer it14:23
mthaddonmarcoceppi: no question besides wondering where the docs were - thx14:25
* marcoceppi makes notes about having documentation for all hooks in the author docs14:25
jcastromarcoceppi: can you add it to the doc sprint spreadsheet?14:36
jcastroso we don't forget?14:36
jcastroevilnickveitch: ^^^14:36
marcoceppijcastro: already did that14:36
m_3what the heck is that?14:37
m_3heart in a box?14:37
rick_hthat's 'marco'14:37
rick_hwe just smile and move on14:37
jcastrohelp me spread the word folks!14:53
m_3hmmm... getting intermittent 503's on manage.jujucharms.com again today14:54
=== freeflying is now known as freeflying_away
jcastro5 minute warning on the first UDS session14:59
jcastrowe're starting with a charm policy review15:00
kurt_jcastro: juju 1.12 did NOT work15:01
jcastroah, where did it fall over?15:01
kurt_jcastro: I ran in to this bug: http://pastebin.ubuntu.com/6032955/15:02
kurt_error: cannot create bootstrap state file: gomaasapi: got error back from server: 400 BAD REQUEST15:02
kurt_on bootstrap15:02
kurt_and, after talking with bigjools, there are no near term plans to put the fix in to maas in quantal15:03
kurt_fix has already made it's way to precise and raring (8/15)15:04
mattywmarcoceppi, I live here :)15:04
marcoceppiof course you do, just broadcasting for those who don't already live around here15:04
kurt_jcastro: for what I am trying to get done, do you suggest I start over on precise?  Raring is not LTS, right?15:06
marcoceppimojo706: anything juju goes here, even "off-topic" juju, whatever that might be15:06
marcoceppiCharm Policy review ongoing: http://summit.ubuntu.com/uds-1308/meeting/21897/servercloud-s-juju-charm-policy-review/15:09
mattywmarcoceppi, have you and jcastro already starting working towards the netflix cloud prize? or is it just planning at the moment?15:10
marcoceppimattyw: I'm not sure, I think m_3 was spearheading that, IIRC15:10
kurt_jcastro: would raring be a more viable option as long as my maas nodes are precise?15:39
jcastroI think going precise all the way is the way to go personally15:39
jcastrosorry I am unresponsive, we're doing UDS today15:39
kurt_no worries15:40
kurt_another acronym I don't know lol15:40
kurt_doesn't matter15:40
kurt_OK, I can try that.15:40
marcoceppikurt_: Ubuntu Developer Summit, http://summit.ubuntu.com/15:43
kurt_macroceppi: ah thanks! awesome!15:43
marcoceppikurt_: actually, I think this gives more information http://uds.ubuntu.com/15:43
jcastrokurt_: you might want to sit in on the openstack ones!15:46
kurt_I was just looking for those15:46
m_3mattyw: re netflix cloud-prize... how can I help?15:47
mattywm_3, just wondering if there's a way I can help with it?15:47
m_3mattyw: I'm doing some netflixoss charms, but not for the cloud-prize15:47
m_3mattyw: I'm disqualified for the netflix cloud-prize proper15:47
mattywm_3, how come?15:48
x-warriordavecheney, just filled that bug report :D15:48
=== x-warrior is now known as Guest28629
m_3mattyw: there're reciprocal prizes/judging between canonical and netflix15:49
m_3so canonical employees are excluded from both prizes15:49
=== Guest28629 is now known as X-warrior`
m_3mattyw: but there's lots to do... I'm getting recipes-rss working atm15:50
m_3lots of subs to be created15:50
m_3and I'm throwing around ideas about gradle/groovy hook impls15:50
=== tasdomas is now known as tasdomas_afk
marcoceppiStarting in about 2 minuts, http://summit.ubuntu.com/uds-1308/meeting/21896/servercloud-s-eco-messaging16:03
=== defunctzombie_zz is now known as defunctzombie
jcastrohey sinzui16:47
jcastrowanna attend or send someone so we can talk charm review queue stuff?16:48
kurt_silly question: is local provider support all of the lxc stuff?16:49
kurt_or is it *any* method in which juju is getting deployed locally?16:50
jcastrowhen we say local provider we mean LXC support16:50
jcastrocurrently. :p16:50
kurt_that appears to be hot topic16:51
AskUbuntuChecking my juju instance through Amazon's AWS console? | http://askubuntu.com/q/33798716:51
sidneihazmat: i think we need a new release of juju-deployer, and then to ping jamespage to upload to saucy. the one currently in saucy is missing some important fixes.16:51
jamespagesidnei, hazmat: let me know when and what16:53
kurt_jamespage: when you consolidate services on to fewer nodes (juju --to) , do you have a suggestion on which charms/services stack best together or some kind of dev blueprint you guys use?16:57
sinzuijcastro, that you for the reminder. I thought today was the 26.16:59
jcastroit's the 27th!16:59
jcastroyour name is Curtis and we have a session today.17:00
kurt_I was looking at this, but I don't believe this follows the tenant client you were suggesting I research https://github.com/mseknibilel/OpenStack-Folsom-Install-guide/blob/master/OpenStack_Folsom_Install_Guide_WebVersion.rst17:00
=== weblife is now known as web-brandon
hazmatsidnei, it is? i thought it got cut from the branch directly17:13
sidneihazmat: it's 0.2.1 :/17:14
hazmatsidnei, its possible we have a packaging version divergence against the same source.17:17
hazmatjamespage, so latest rev 114 / version 0.2.3 (just incremented to be sure) is stable17:17
sidneihazmat: it's missing the fix from r110 at least17:17
hazmatjamespage, if you could upload that we should be good for saucy. there are other changes inbound but they can land into the ppa for now.17:18
marcoceppihazmat: could you put that version in the stable ppa as well?17:22
jcastroarosales: ok details updated for the next session17:56
jcastroso we should be able to just start on time this time, heh17:56
arosalesjcastro, sorry about the conflicts last time17:57
arosalesjcastro, I am leaving getting session started in your capable hands :-)17:57
jcastrono worries, first day is always rough17:57
jcastrothat seems to have been our only problem today, I'll take it!17:57
arosalesjcastro, for sure17:57
jamespagehazmat, I'm cutting from the tarballs on pypi18:00
marcoceppijcastro: link?18:00
hazmatjamespage, aha, thanks18:04
hazmatjamespage, updated on pypi18:04
jamespagehazmat, sidnei: uploaded to saucy18:19
jamespagehazmat, is that compatible with juju-core 1.12.0 ?18:20
hazmatjamespage, yes18:20
jamespagekurt_, its probably easier to say what won't go together right now18:21
jamespagenova-compute, quantum-gateway, nova-cloud-controller will all conflict with each other in config files18:21
jamespagelikewise ceph, cinder, glance and nova-compute (around /etc/ceph)18:21
=== natefinch-afk is now known as natefinch
kurt_jamespage: thanks.  do you use a two or three node deployment for testing?  I would be curious to see what people typically cluster together on the same node18:47
jamespagekurt_, if you just want compute (i.e. no cinder) then you can get away with three nodes18:48
kurt_jcastro: is it normal I would have to "sync-tools" after fresh install of juju 1.12?18:48
jamespagekurt_, right now its tricky and kinda unsupported because of the conflicts in the filesystem18:48
jamespagekurt_, juju container support will help with taht18:48
jamespagea charm assuming it has control over the filesystem is not unreasonable18:48
kurt_juju container support = lxc = local support you guys were just talking about?18:49
jamespagekurt_, kinda18:49
jamespagelxc is used in the local provider right now18:49
jamespagebut a feature is being worked on to allow you to add LXC machines in other providers18:49
jamespageso you can slice up a server using LXC for deploying servers into18:50
kurt_right, but just forcing with --to is going to cause problems?18:50
kurt_that was the path I was going down18:50
sarnoldI wouldn't expect --to to work with every possible combination of charms18:52
sarnoldbut _many_ combinations might work fine18:52
kurt_sarnold: right.  If someone could share their working blueprint for a successful deployment, that would be awesome18:53
kurt_whether its 2, 3 or more nodes - I am just wondering what works18:54
kurt_juju 1.12 -> YARGH! LOL http://pastebin.ubuntu.com/6033776/18:54
* kurt_ beating fists, stomping feet, and rolling eyes18:56
marcoceppikurt_: destroying environment removes the bucket :\18:56
marcoceppidavecheney: ^ Might want to change that.18:56
kurt_marcoceppi: does syncing tools bootstrap too?18:56
kurt_that doesn't seem logical18:56
marcoceppiSo, juju destroy-environment; juju sync-tools; juju bootstrap18:57
sarnoldmarcoceppi: dunno. I could wanting the billing to end when the environment is destroyed...18:57
marcoceppikurt_: you need to sync-tools prior to bootstrap but after destroy18:57
kurt_marcoceppi: ok, the sync-tools is new for 1.12 for me18:58
kurt_never had to do that before18:58
kurt_marcoceppi: why does it say this then? "error: environment is already bootstrapped"19:00
* kurt_ confused19:00
marcoceppikurt_: because when you run juju bootstrap it creates a file in the bucket that says "this is bootstraped", even if an instance doesn't launch. It's to prevent two bootstraps from happening if the cloud provider takes a long time to start up the bootstrap node19:02
marcoceppiSo there's a bug in there, in that if no tools are matched, or if there is a general bootstrap error, it should clean up that file19:02
kurt_marcoceppi: but the tools are there after download, is the error being generated prior to the tools downloading?  If I destroy my environment, there is no bootstrapped node, so that error seems misleading19:04
marcoceppikurt_: there's a juju bootstrap --upload-tools option, however people keep telling me that it's more for development versions of Juju and that sync-tools is the right way to go.19:04
marcoceppikurt_: I'm not sure of the nuances with they sync-tools, I've not had the pleasure of using it much19:04
marcoceppihazmat: sync-tools and when to use it? ^19:05
kurt_marcoceppi: I don't think one has a choice when bootstrapping. you must have the tools19:05
marcoceppikurt_: Yes, most public clouds have the tools already sync'd somewhere, so it's effortless19:06
marcoceppifor private clouds and maas, I'm not sure of the proper procedure19:07
marcoceppiI'm pretty sure you want to follow: juju destroy-environment; juju sync-tools; juju bootstrap19:07
kurt_marcoceppi: I would hope it would function in nearly the same way. :)19:07
marcoceppiI think the sync-tools is a pre-step to the process19:07
kurt_that's exactly what I did and that error showed up19:07
marcoceppithe bootstrap error?19:07
marcoceppikurt_: that's not right19:08
marcoceppiwhat happens if you destroy-environment then bootstrap again?19:08
kurt_and in looking at my maas, I have no bootstrapped node19:08
kurt_same thing19:08
kurt_let me try once more - I will paste to paste bin for you19:08
marcoceppiso there's something else going on19:09
marcoceppikurt_:  use -v for both commands19:09
kurt_doing that...19:10
kurt_marcoceppi: ah… heh, in reviewing what I cut and pasted above, there was a slight error19:12
kurt_marcoceppi: notice: juju bootstrapdestroy-environment; juju sync-tools; juju bootstrap19:12
marcoceppisarnold: I understand the wanting to kill the billing, possibly a juju destroy-environment --preserve-tools would resolve that19:13
kurt_should be: juju destroy-environment;  juju sync-tools; juju bootstrap <- My bad for not picking that up19:13
marcoceppikurt_: np, give that a go19:13
kurt_working :)19:14
hazmatmarcoceppi, when your in a private cloud, and don't have a compile env, ie just want to use juju19:14
hazmatmarcoceppi, it will copy the latest release from public ec2 bucket into private cloud object storage (for the env/user)19:14
marcoceppihazmat: so it's really for those who don't have go-land installed and don't want to compile the tools themeslves? IE majority of users?19:14
hazmatmarcoceppi, yup majority of users in a private cloud.. public clouds should already have tools installed19:15
marcoceppihazmat: ack, gotchya19:15
marcoceppiWe'll need to update our docs to let people know about sync-tools and maas/private clouds19:15
hazmatthere's another critical /high bug for private cloud usage, re allowing invalid ssl certs that causes issues for juju-core19:15
hazmatatm, requires updating the client's os level certs to accept the private cloud ssl cert ca as trusted.19:16
hazmatbug 1202163 fwiw19:17
_mup_Bug #1202163: openstack provider should have config option to ignore invalid certs <papercut> <juju-core:Triaged> <https://launchpad.net/bugs/1202163>19:17
* marcoceppi +1s19:17
kurt_hazmat: yes I saw something about a cert error19:25
kurt_kurt@maas-cntrl:~$ juju status19:25
kurt_error: no CA certificate in environment configuration19:25
hazmatthat's a little different19:25
kurt_rebooting the node, destroying, and restarting seems to have fixed that anyways19:26
hazmatjuju also internally uses ca certs to secure communications with mongodb and the api server19:26
hazmatthose certs are kept in the JUJU_HOME (default ~/.juju)19:26
hazmatthe certs referenced in the bug are the underlying iaas provider ssl certs19:26
jcastromarcoceppi: kirkland has a bunch of bugs that your sentry interface autodocumenter should fix19:26
jcastrowhere can he file bugs?19:26
kurt_ah ok - but that's mostly for external provider scenarios, right?19:27
marcoceppijcastro: what kind of bugs?19:27
hazmatkurt_, it also applies to openstack using ssl19:27
hazmatin a private cloud19:27
marcoceppilp:amulet should suffice19:27
marcoceppijcastro: ^19:27
kurt_hazmat: ok, thnx19:27
kirklandmarcoceppi: okay, thanks19:27
X-warrior`so I'm creating a charm, and now I used deploy on it, but some stuff went wrong and I'm fixing it, how can I remove this one to try the new one? or is it possible to update?19:28
marcoceppiX-warrior`: you can either use juju upgrade-charm to upgrade the charm in place, you can deploy the charm again under a different alias (IE: `juju deploy --upgrade --repository /path/to/charm/repo local:charm-name now-your-alias) or you can destroy the environment and re-bootstrap19:29
kirklandmarcoceppi: thanks!  https://bugs.launchpad.net/amulet/+bug/121754019:30
_mup_Bug #1217540: Every interface defined by a charm should be documented with examples <Amulet:New> <https://launchpad.net/bugs/1217540>19:30
marcoceppikirkland: so I can do the first half of that, the documentation. The examples I might defer to another bug on another project. I'll let you know.19:31
marcoceppiwell, I can attempt to do the first part*19:31
kirklandjcastro: marcoceppi: okay, and now I'd like to file a bug, requesting that the squid-reverseproxy charm add support for https_port -- where do I file that?  against launchpad.net/charms?19:32
jcastrothat would be against the charm itself19:32
web-brandonIf I include a folder in my charm with files.  Where is it placed on the service instance server?19:34
X-warrior`marcoceppi, ty, does the upgrade-charm will update already deployed instance?19:34
X-warrior`or do I need to upgrade charm an then call deploy with --upgrade flag19:35
web-brandonOr is it only on the juju bootstrap instance19:35
marcoceppiX-warrior`: So, upgrade charm will update the charm contents, but that's it. If you want it to run hooks again (like run hooks/install or hooks/config-changed) you'll need to create a new hook in hooks/ called upgrade-charm and put that logic in there, for example: http://bazaar.launchpad.net/~charmers/charms/precise/wordpress/trunk/view/head:/hooks/upgrade-charm19:36
marcoceppiX-warrior`: nope, those are two different commands19:36
marcoceppione is just an upgrade charm, the other will deploy a fresh copy of the charm under a new service name, so it will follow the typical install and deploy as if it was freshly deployed19:36
X-warrior`and what about removing already deployed services?19:37
marcoceppiX-warrior`: you can run juju destroy-service to remove deployed services, but you can't, IIRC, deploy a service again with the same name in an environment19:37
marcoceppiSo you'll need to use the `juju deploy --upgrade --repository ... local:charm <alias>` syntax19:38
marcoceppiX-warrior`: So, in the event of mysql, if you've already run `juju deploy --repository ... local:mysql` and then you want to deploy fixes. You could run `juju upgrade-charm --repository ... mysql` OR if you wanted to start fresh, `juju destroy-service mysql` then run `juju deploy --upgrade --repository ... local:mysql db` that'll deploy mysql under the alias of "db" so you don't have it deployed again as "mysql"19:39
X-warrior`yeap, but if I would like to deploy fixes, my charm must have a upgrade-charm19:40
marcoceppiX-warrior`: It's not required, using upgrade-charm without an upgrade-charm hook works, a new version of the charm will be deployed19:41
marcoceppiHowever, that's all juju will do, is unpack the new version. It won't run any other hooks19:41
X-warrior`oh I get it19:42
marcoceppiX-warrior`: So, you could write an upgrade-charm hook right now, and then run upgrade-charm. Juju will unpack the new version then execute the new hook19:42
X-warrior`should the destroy-service command destroy the instance as well?19:42
marcoceppiBut all hooks are considered "optional" for juju, if it doesn't exist juju just skips it and moves on19:42
marcoceppiX-warrior`: no, instances will remain. It's juju's way of protecting data. You can remove them with `juju terminate-machine <machine-number>` if you're done with them19:42
X-warrior`let me try19:43
X-warrior`ERROR juju supercommand.go:235 command failed: no machines were destroyed: machine 1 has unit "test/0" assigned19:46
marcoceppiX-warrior`: is test/0 in an error state?19:47
marcoceppiX-warrior`: it probably says something like life: dying and agent-state is in error?19:47
X-warrior`I guess test/0 is an error state since the install hook failed and on log I can see "awaiting error resolution for "install" hook"19:48
marcoceppiX-warrior`: so, when a charm is in an error state it stops all other events and leaves them in an events queue19:49
marcoceppiIn this case, config-changed and a bunch of other hooks are all queued up, with the last event being the service destruction.19:49
marcoceppiX-warrior`: so you'll need to "resolved" the errors for the charm before it can get to those other events19:49
marcoceppiX-warrior`: `juju resolved test/0` should suffice19:50
marcoceppiX-warrior`: you may need to run that command multiple times if it hits any other errors19:50
marcoceppiX-warrior`: Also, for future reference, you can have juju retry a hook, using `juju resolved --retry test/0`19:50
X-warrior`what does the 0 stands for? install?19:51
marcoceppiX-warrior`: that's the unit number. So each service you deploy gets at least one unit. If you wanted to scale out you could run juju add-unit test and you'd get test/0 and test/119:51
marcoceppiif you notice in juju status they're listed under the "units" heading for the service19:51
X-warrior`so if I send the --retry flag it will rerun the failed hook19:52
X-warrior`so let's say in my case I had a problem with install hook, so it gets locked19:53
X-warrior`if I use upgrade-charm, it will get enqueued19:53
X-warrior`and if I use the --retry it will fail again19:53
marcoceppiX-warrior`: yes, so you could, for instance, juju ssh test/0 (to ssh in to the node), switch to the root user, go to /var/lib/juju/agents/unit-test-0/charm/, edit hooks/install, fix whatever the problem was, log out and then run juju resolved --retry test/0 to try again19:53
X-warrior`and then I need manually update the hook on the server19:53
marcoceppiRight, so at this point you're editing things on the server, it's just one possible workflow for writing charms. It's dangerous, because if you don't copy the fix to your local charm and destroy the service, you lose your changes19:54
marcoceppithe alternative is to destroy the service, fix it locally, re deploy, or to destroy-environment, fix it locally, re-bootstrap, then deploy19:54
marcoceppiIt's up to whatever way works best for you as an author19:55
marcoceppieach has their pros and cons19:55
web-brandonIf I include a folder in my charm with files.  Where is it placed? I cannot find it for the life of me19:56
X-warrior`let me try it again19:57
X-warrior`marcoceppi, I have to go now, will keep trying later, ty for your help20:23
X-warrior`have a good one20:23
jcastroFunnyLookinHat: is there a charm for beansbooks?20:29
marcoceppiweb-brandon: It's placed inside the $CHARM_DIR, typically /var/lib/juju/agents/unit-<service-name>*/charm20:32
web-brandonmarcoceppi: thank you so much.20:34
marcoceppiweb-brandon: each hook is executed at the $CHARM_DIR (and that variable is available to hooks)20:50
marcoceppiWe recommend not hardcoding paths when possible20:51
web-brandonjust ran through a test to spit out the path in to juju-log.  I am about to perform some 'cp' actions with that as the base dir.20:56
marcoceppiweb-brandon: if you want to be safe, you can just do $CHARM_DIR as the prefix to the path20:57
web-brandoni understand. it can change from server to server20:57
marcoceppiweb-brandon: not only server to server, but version to version. However, all hooks will _always_ be executed from the $CHARM_DIR20:57
web-brandonmarcoceppi: good to know.20:58
web-brandongonna make a irc-bot charm21:00
_mup_Bug #1217591 was filed: Had to manually add access to bootstrap node on 17070 and 37017 to the secgroup <juju:New> <https://launchpad.net/bugs/1217591>21:48
=== freeflying_away is now known as freeflying

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