=== CyberJacob is now known as CyberJacob|Away [06:03] Hi there [06:03] I am getting some errors with juju deply commands [06:03] specifically no matter what I try to deploy I always get the error "ERROR cannot get latest charm revision: charm not found: cs:saucy/juju-gui" [06:04] juju status is reporting information about bootstrap node, however I am not able to deploy anything with saucy salamander. Any help would be greatly appreciated [06:05] Luca__: I don't think there are any saucy charms so far: http://manage.jujucharms.com/charms [06:08] This is what I noticed, however I dont see charms from raring either, which surprises me a lot [06:10] sarnold:I have seen documents deploying mysql on raring, however I cant see the charm available there. I am lost === freeflying is now known as freeflying_away [06:11] Luca__: well, now that's funny. I thought I saw some raring charms there too... [06:13] sarnold: yes, and it looks strange to me there are no charms at all for the 13.04, which is 6 months old [06:14] Luca__: I do know that raring and quantal charms were rare, since most people deployin juju were using 12.04 LTS instead of the newer releases [06:15] sarnold: funny thing about is that 13.10 should apparently make the life easier for an openstack deploymnent, however I cant even start with a simple mysql installation... [06:16] sarnold: therefore I was wondering if there is something that I am not considering or escaping from my understanding [06:16] Luca__: I'm no expert, but I do wonder if you'd run openstack on 13.10 and then run the charms on precise instances -inside- openstack... [06:17] Luca__: (though 13.10 might not have the support length you'd really want for a cluster...) [06:20] sarnold: I agree for the support, though this is only a PoC waiting for the 14.04 in april. I am trying to build openstack using juju charms. I know this has been done on raring, however I cant find anything for saucy [06:23] sarnold: beside on the ubuntu server page they specifically talk about charms http://www.ubuntu.com/server [06:29] jcastro: ^^^ what am I missing? :) [06:35] sarnold: I bootstrapped using precise and it is working. However I was pretty sure would not be a problem using charms on Saucy [06:35] Luca__: having the saucy charms around would probably be useful for folks, like you, planning for 14.04 LTS... [06:36] What is the point then installing 13.10 for Openstack is no charms for this release are available? Could anybody clarify this? [06:36] you _can_ install openstack without juju, even if it is a pain in the butt.. :) [06:37] sarnold: As you said it is a pain without an orchestration tool [06:37] I did a year ago and I was crying [06:38] ow :/ [06:38] beside I need an HA configuration, therefore I was hoping on Ubuntu [06:38] heh, setting up 28 machines by hand doesn't sound fun.. [06:38] Rackspace just come out with a new coobook for quantum support among them [06:39] sarnold: right, this is another point that I would like to get clarified. I am sure many services can be deployed on the same machine, reducing the # of needed servers [06:40] Luca__: i certainly hope so. === freeflying_away is now known as freeflying [06:50] sarnold: http://www.jorgecastro.org/2013/10/30/a-very-spooky-charm-status/ [06:55] Luca__: interesting, jcastro linked to http://manage.jujucharms.com/recently-changed but there's only precise charms.. === freeflying is now known as freeflying_away === freeflying_away is now known as freeflying [07:48] sarnold: Indeed. I may raise a question in the Ask Ubuntu and see if I get some answers [07:49] They might have been pulled off from the repository and will be later pulled in [07:49] Luca__: that's a decent idea, I know the juju team follows up on questions there pretty well; the mail list might also work well [07:50] mail list? [07:51] Luca__: ah! here we go: https://lists.ubuntu.com/mailman/listinfo/juju [07:53] sarnold: Thank you! === freeflying is now known as freeflying_away [09:03] What are the steps to deploy OpenStack in a VM using juju? | http://askubuntu.com/q/377353 [10:04] any ideas what could be wrong on a saucy local bootstrap, status shows forever pending, ssh to machine 0 gives connection refused, deployed a service, machine 1 also in pending, can see in syslog the IP given from DHCP but ssh connection reset, ping works .. last log line "2013-11-15 09:52:49 INFO juju.provisioner provisioner_task.go:367 started machine 1 as instance jaywink-local-machine-1 with hardware " [10:05] tools 1.16.0.1 [10:06] (and UFW is off, tried that already :)) === rogpeppe1 is now known as rogpeppe [10:17] pastebin: http://paste.ubuntu.com/6420446/ [10:17] (new attempt) [10:28] ok now it worked, I deleted server image (oct 24) from /var/cache/lxc/cloud-precise/ as instructed https://juju.ubuntu.com/docs/troubleshooting-local.html -> No machines start ... === 3JTAADOM8 is now known as wallyworld === wallyworld is now known as Guest37360 [11:07] Is it possible to use Juju on MAAS nodes not connected to Internet? | http://askubuntu.com/q/377397 [11:41] mgz, if you do win the internal lottery and end up working on Bug#1241674 let me know if I can be of any help, I have an environment we can test in. [11:41] <_mup_> Bug #1241674: juju-core broken with OpenStack Havana for tenants with multiple networks

[11:48] gnuoy: thanks!... I guess [11:48] you guess right ! [11:49] Is this just bad timing (related to uploading new juju tools perhaps?), bootstrap failed with "error: Get https://juju-dist.s3.amazonaws.com/tools/juju-1.16.3-precise-i386.tgz: EOF". More at http://paste.ubuntu.com/6420792/ [12:41] how do I get rid of a "life: dying" situation - a failed nfs service install does not want to be destroyed :( juju local does not seem very stable :( === gary_poster|away is now known as gary_poster === kentb is now known as kentb-afk [13:36] jaywink: I haven't updated in the past few weeks, but yes, the local stuff is quite new, and last time I was playing with it, I needed the following to destroy the environment properly: http://paste.ubuntu.com/6421224/ [13:37] (obviously don't use if you've got other lxc containers you care about there) [13:38] noodles775, doing "juju resolved nfs/0" (in this case) solved it :P maybe it wasn't destroying because of the failed install? [13:39] jaywink: great (sorry, I thought you were trying to destroy/recreate your local environment). [13:59] jaywink: running resolved is something you'd have to do in any environment [14:00] marcoceppi, ok, I think the docs could be clearer - I saw no mention while skimming through for help :) [14:01] (or just a hint from destroy-service that there is an error would be golden - will check if this is not reported yet..) [14:03] jaywink: it should have been placed in the destroying services page as a caveat [14:03] but it wasn't [14:03] * marcoceppi updates the docs [14:07] thanks marcoceppi, added a note to a bug I had already commented on the same issue to :) (https://bugs.launchpad.net/juju-core/+bug/1219902) [14:07] <_mup_> Bug #1219902: Cannot destroy service when install hook failed

[14:47] rogpeppe: I'm still new to mongo, but I think a query like this will do what you want: {field: {$size: {$mod: [2, 0]}}} [14:48] (where "field" is the name of the element that holds the array) [14:48] benji: thanks [14:48] * rogpeppe tries it === zz_paulczar is now known as paulczar [14:50] benji: doesn't seem to work unfortunately [14:50] benji: i think the argument to $size must be a number [15:02] rogpeppe: darn === kentb-afk is now known as kentb === paulczar is now known as zz_paulczar === zz_paulczar is now known as paulczar [16:40] hey all, how do I file bugs against juju installed from the cloud-archive tools pocket on precise? [16:40] apport tells me that Juju is not an official Ubuntu package :/ [16:40] should I file it on Launchpad against juju, or is there a special place for the cloud-archive packages === paulczar is now known as zz_paulczar === TheFreeMue is now known as TheMue [17:04] Can a hook call another hook? Like if I wrote a start and stop hook can the installer call those or is it better to have the code to start and stop in the install hook? [18:10] mbruzek, hello [18:10] sorry, missed that message earlier [18:10] mbruzek, is the question still hook can call another hook? [18:10] we have a few folks at a conference this week so little late on the replies [18:10] Hi Antonio [18:11] * hazmat looks for the irc logs [18:11] Yes, hazmat can I call start from install? [18:11] Or should the hooks not know about each other? [18:11] hazmat, thank for chiming in :-) [18:12] I read the hook documentation and I did not see anything about calling each other. [18:12] mbruzek, so they can, but they don't nesc have the same access to information [18:12] I see that very often people write one hook and make symbolic links to the one hook. [18:13] mbruzek, when you say installer, it looks like that is the application installer not the hoook install, correct? [18:13] mbruzek, yes.. they have the exec name as dispatch [18:13] to internal functions doing the hook, mostly its code orgranization technique [18:13] mbruzek, so order of hooks on a charm starting up is.. install, config-change, start [18:14] mbruzek, you can call out to other hooks .. their just other executables. [18:14] OK thanks. I understand the all in one hook, but I am trying to write them separately [18:14] mbruzek, relation hooks in particular have additional context [18:14] That answers my question, thanks. [18:15] like REMOTE_UNIT_ID, etc. for those you have to setup some additional env variables if they want to use the relation cli api (relation-get/relation-set/relation-list) etc. [18:15] ie. config-changed needs to modify a relation value, it will use relation-ids to fetch the applicable relation and then set that before calling out to the rel hooks [18:15] k [18:16] OK let me see if I understand this correctly. [18:16] mbruzek, so to your question, "Can a hook call another hook" [18:16] * arosales lets mbruzek type [18:16] I have a database-relation-changed hook that needs to stop and restart a service. [18:17] I have that code in the stop and start hook. So I was wondering if I can simply call the hook or if I need to duplicate the start and stop logic in the database-relation-changed hook. [18:17] My understanding is that I DO NOT need to duplicate. [18:17] And that one hook can call another., [18:18] Is there anything wrong in my assertion hazmat / [18:18] ? [18:19] mbruzek, nothing wrong with that [18:19] mbruzek, ie. call out to start/stop in your rel-changed hooks [18:20] mbruzek, which db you working with? [18:21] Thanks, that was the basic question thank you. [18:21] mysql === CyberJacob|Away is now known as CyberJacob [19:12] ok, something that is confusing me: there is LXC (which I've used) and I've heard of the local provider. Are they the same thing? Or is local just configuring what it is run on, without containers? === zz_paulczar is now known as paulczar [19:22] hi stokachu. DO you have a few minutes to talk about Bug #1222671 ? [19:22] <_mup_> Bug #1222671: Using the same maas user in different juju environments causes them to clash

[20:20] Azendale, local provder uses lxc [20:21] Azendale, local provider just automates the creation and setup of lxc containers as though they were machines in an environment.. effectively translating.. add-unit -n 3 wordpress into creating lxc containers and setting up workloads on them [20:21] Azendale, there's also separately the notion of adding lxc containers to machines in other environments .. ie add-unit --to=lxc:2 wordpress ... which will create an lxc container on machine 2 in the env and deploy wordpress to it === paulczar is now known as zz_paulczar [20:39] hazmat: thanks! I was wondering if you could get better density when using a MaaS environment by using LXC :) [20:41] Azendale, indeed.. that's why that capability is there. it works for most charms.. the only exceptions to date have been some of the openstack ones which want either direct device access or kernel modules (ceph and neutron) [20:53] What would be the correct command to copy down the mysql charm to my machine? [20:54] bzr branch lp: /mysql/trunk [20:54] Something like that right? [20:57] Sorry I found the answer on my own, for future reference it was: [20:58] bzr branch lp:~charmers/charms/precise/mysql/trunk mysql [21:19] Do we have any mysql experts in the room? [21:23] marcoceppi, it looks like you wrote the charm. Are you there? [21:24] I am writing a charm that uses the database relation and I don't understand why. The error message is: [21:24] ???Error executing sql: create database if not exists `?` default character set utf8 - Could not create connection to database server. Attempted reconnect 3 times. Giving up.??? [21:24] Unable to create the database. The password might be incorrect or the database is not started. [21:25] I retrieved the root mysql password as it says in the README, by going to /var/lib/mysql/mysql.passwd [21:26] And I entered that into openmrs DB wizard as the mysql root password. But when it tries to create the database I get the error. [21:31] * arosales reads backscroll [21:32] marcoceppi is traveling atm . . . /me looking into error [21:34] mbruzek, our tech writer just took a first whack at documenting the mysql charm interface @ https://docs.google.com/a/canonical.com/document/d/1WTws_k5K__NAsfZwxf-oKHYqZnZYvrg7K58MvOHFpAY/edit# [21:35] mbruzek, do you check if the db has been set up before accessing [21:35] mbruzek, take a look at the doc under "# Check to see if 'database' has been set, and loop until it is" [21:36] Yes it is up and I can even log in using the root password that I gave openmrs [21:36] mbruzek are you setting the mysql port to 3306 in openmrs? [21:37] Yes that too. Here is the Database Connection string before OpenMRS attempts to create the database [21:37] jdbc:mysql://localhost:3306/@DBNAME@?autoReconnect=true&sessionVariables=storage_engine=InnoDB&useUnicode=true&characterEncoding=UTF-8 [21:37] well maybe it should not be localhost... since this is running on tomcat [21:38] let me dig a little bit more [21:38] mbruzek, also bzr branch lp:charms/precise/mysql [21:38] hazmat, Do I need the ~ before charms? [21:38] or a bit more ambiguous but also valid atm .. bzr branch lp:charms/mysql [21:38] mbarnett, nope [21:38] whoops [21:38] OK thanks. [21:39] mbruzek, ping here if changing out local host helps establish the connection [21:39] mbruzek, the fully qualified name includes the ~charmers, but when a charm is officially approved (ie, gone through qa by charmer reviewers) and promulgated those aliases are valid. [21:39] I don't know where the localhost is coming from, the property file that my charm builds properly puts the host ip address in there, let me grab that [21:39] mbruzek, i wrote the original version of it (mysql).. what's the issue? [21:40] OpenMRS tries to create a database and I get the following error: [21:40] ???Error executing sql: create database if not exists `?` default character set utf8 - Could not create connection to database server. Attempted reconnect 3 times. Giving up.??? [21:40] Unable to create the database. The password might be incorrect or the database is not started. [21:40] the jdbc conn string which looks invalid. you need to pull relation-get private-address from the db-relation-changed hook when forming the jdb conn string [21:41] mbruzek, your connection string looks invalid [21:41] yes I am pulling that info in the charm., but it does not look like open [21:41] openmrs is not getting my property file [21:42] This is the URL my charm is generating [21:42] connection.url=jdbc:mysql://10.0.3.3:3306/openmrs?autoReconnect=true [21:42] mbruzek, you can use debug-hooks to try and debug the issue.. add an exit 1 into your relation-db-changed hook, so it fails, and then use debug-hooks to get into it, and do juju resolved --retry to drop into the debug shell [21:42] mbruzek, you can juju ssh openmrs/0 to get into the machine and use mysql client to verify the connection string, where's the username password? [21:43] my database-relation-changed hook outputs a properties file which OpenMRS is supposed to read and use. [21:43] That value is what the charm wrote in the file. [21:43] * hazmat nods [21:43] # OpenMRS Runtime Properties file. [21:43] connection.username=iochahmezahhaip [21:43] connection.password=quiethoovuyiphi [21:43] connection.url=jdbc:mysql://10.0.3.3:3306/openmrs?autoReconnect=true [21:44] So it looks like OpenMRS is making up their own connection URL, I will have to look if I can disable that. [21:44] k, that looks good [21:44] Or change it [21:44] Thanks hazmat [21:47] OK I have found a way to enter the database connecton url, and changed it to what should be right jdbc:mysql://10.0.3.3:3306/openmrs?autoReconnect=true [21:47] I still get the could not create the connection to the database error [21:47] Do I need to open port 3306 on the mysql server? [21:48] So my charm is running as a subordinate to tomcat, and the mysql server is a different machine 10.0.3.3 [21:50] mbruzek, https://answers.openmrs.org/questions/795/openmrs-configuration-error suggests so [21:50] on the mysql side [21:51] Yeah but I think the mysql charm should have that one open [21:51] That is how normal connections are made [21:52] Reading Antonio's document helped me, I guess the database is created when the relation is made. Now that I fixed the URL it tries to create the database that is already there. [21:53] I just tried the option to not create the database and that looks like it worked. [21:53] woot! [21:53] mbruzek, so you got past "Could not create connection" error? [21:53] Thanks hazmat and arosales for your assistance [21:53] Yes [21:53] \o/ === gary_poster is now known as gary_poster|away [22:24] hazmat, Are you still here? [22:33] mbruzek, hello [22:33] mbruzek, could you state your question again? [22:34] In the database-relation-changed hook I am able to call `relation-get user` for the database user name. [22:34] I was wondering if there was a way to get that database user name using the juju command line [22:35] The hook gets these three values. [22:35] db_user=`relation-get user` [22:35] db_db=`relation-get database` [22:35] db_pass=`relation-get password` === zz_paulczar is now known as paulczar [22:36] In this case my charm is named openmrs, and when I create a relation the openmrs database is created [22:36] mbruzek, can openmrs be foreced to reread the properties file [22:37] No, the more I read about it here it only writes the propertis file. [22:40] * arosales is also new to openmrs [22:40] mbruzek, does the mysql creds have to be typed in via the config wizard? [22:41] if not perhaps post config wizard you can invoke a config-changed to pass in the mysql creds [22:41] * arosales not sure what methods openmrs allows for passing in creds post set up wizrd though [22:42] The only way I got it to work was by doing the advanced set up, where I could specify the DB url, and then tell the wizard NOT to create the database and then yes I had to give it the db user and db pass [22:44] Reading the mysql README there was a complicated way to get the mysql root password. You have to ssh to mysql instance and cat a file. [22:44] mbruzek, and in that case you needed to query the db user and pass, correct? [22:44] Is there an easier way to get the root password from mysql that I missed? I may be able to work with that [22:45] Yes the user who is setting up the charm would need a way to get the database user and password. [22:46] mbruzek, and you are querying that from the tomcat charm, correct? [22:46] Well the OpenMRS is a subordinate to the Tomcat charm [22:47] but yeah the hook is getting that info properly, I just want to see if the user can get that information easily? [22:49] bcsaller, hazmat ^ ? [22:49] mbruzek: you don't want to use the root password. [22:49] Once the hook has that information, can I put it somewhere the user can get it out easyis? [22:50] Hi marcoceppi, I would only need the root password if I had to create the table. [22:50] I found out that the table openmrs is created when I create the relation [22:50] So I need the user and pass that are authorized to that table and they are not natural like openmrs/opemrs [22:51] connection.username=iochahmezahhaip [22:51] connection.password=quiethoovuyiphi [22:51] connection.url=jdbc:mysql://10.0.3.3:3306/openmrs?autoReconnect=true [22:51] The user who deploys this charm will not know that iochahmezahhaip is the username [22:52] and they need to enter that in a wizard, I am just trying to figure out how I can provide this to the user in an easy way [22:52] mbruzek: one second, getting my laptop [22:52] config opt on the openmrs charm? [22:53] arosales, As far as I know the user/pass for the DB is not something I can specify. I am only getting it once the relation to the DB is created [22:54] I could set the config option once I have them couldn't I ? [22:54] you could via juju set, but that may not be the most elegant solution talking about outloud now [22:55] marcoceppi, hello btw :-) [22:56] what happened to all the != precise charms? a chap in here last night couldn't find any charms from any other versions and wondered if juju was more or less dead or dying.. [22:57] sarnold, was he looking in trusty? [22:57] precise is where we maintain the charms atm, [22:57] I would point him there, and he can force a precise charm onto a later ubuntu version if need be. [22:58] mbruzek: you can't set configuration options from the charm [22:58] arosales: he was mostly interested in saucy, e.g. http://www.ubuntu.com/server promised interesting juju charms for 13.10 and better openstack... [22:58] OK [22:58] mbruzek: what you need to do, is find out the fields that the wizard expects, then POST that data to it from http://127.0.0.1/path/to/wizard [22:58] marcoceppi, not even in a hook? [22:58] mbruzek: there is no config-set from within a hook [22:59] configuration is explicitly done by the user [22:59] The wizard is a multipage wizard [22:59] I am not sure POST will work... but that is a great idea [22:59] The wizard has like 3 or 4 pages [22:59] marcoceppi, well from one charm to another you can't, yet . . . [23:00] mbruzek: you can probably still mock out each step in the wizard, it's a bit of work and a little tricky, but it's how I would approach the problem [23:00] arosales: o/ re:invent is over [23:00] time to de-compress ;) [23:00] mbruzek, well there is the advanced option though and just forget the wizard [23:01] arosales: charms can't set their own config options. There's charm feedback from a charm on the roadmap but it hasn't been specified [23:02] sarnold, ah I see. The charms still can run on other ubutu versions and the client is of course tested on 13.10, but the workload itself is tested on the precise branch [23:02] marcoceppi, yup [23:02] marcoceppi, why do I not want to get the root pass on mysql? You mentioned that I should not do that right away [23:02] marcoceppi, mysql-root relation appropriate here? [23:03] mbruzek: that's the root mysql password, arosales the mysql-root relation creats a new superuser account, not the root [23:03] mbruzek: same reason we use sudo accounts over giving our root passwords [23:03] out* [23:04] arosales: makes sense, you've only got so much time so you've got to pick your priorities. It just was jarring for a user to see 13.10 promoted so heavily and then be told that nearly all the work was being done on a system three releases back :) he wanted to get ready for 14.04.. [23:04] OK. Well the wizard wants the root mysql password to create the table, but will accept the user/pass that has access to the table too [23:05] sarnold, understood and we need to start looking at testing charms against 14.04 [23:06] hopefully when the testing starts we'll be able to better field those questions [23:06] arosales: thanks for catching me up :) [23:08] the charm branch conundrum is not an easy one to follow :-/ [23:08] sarnold, thanks for pinging here with the feedback [23:08] no, it isn't. I thought I had even authored one or two charms for newer-than-precise in the charmstore, so was surprised to not find them. :) hehe [23:09] ("I _did_ do that, right? right?" :) [23:09] sarnold: they're still there, just not promoted [23:10] sarnold, if you had a charm in 13.04 it doesn't automatically get promoted to 13.10 and 14.04 [23:11] Thus, as of today there are no charms in 14.04 [23:11] arosales: sure, I wouldn't expect _that_, that's asking for magic :) but I thought I had a 12.10 charm or two written.. [23:11] as they need to be tested and ack'ed into 14.04 [23:12] sarnold, hmm the charmstore may not look at other series branches [23:12] sinzui could confirm [23:12] I don't see him online atm [23:12] sarnold, you said you had charms in quantal? [23:13] arosales: let me go look .. [23:15] arosales: hrm. all I can find now are my own precise versions. Color me confused. :) [23:16] sarnold, you looking in lp or in jujucharms? [23:16] arosales: local hard drive, I figured I'd still have the trees laying around.. [23:16] arosales: .. and launchpad agrees, again just precise. sorry for the wild goose chase. [23:17] sarnold, no worries. I think there are a few, but currently we concentrate on LTS'es for the charms [23:17] and we will be working on the 14.04 charm story [23:17] veryveryveryvery soon [23:17] which I think was the root of the person's queries for no charms in 13.10 (getting prepped for 14.04) [23:18] he -mostly- wanted a way to deploy the newest version of openstack with less pain :) [23:18] sarnold: he can do that already [23:18] marcoceppi: I spent half an hour trying to find charms to help :( [23:18] sarnold: all versions of openstack are available in the openstack charm, for the record [23:19] marcoceppi: wow. got a pointer? :) [23:19] sarnold: as part of our promise for LTS, we make all versions of openstack available via the cloud-archive [23:19] arosales: the charm readmes [23:19] config opt in the charm to point the the cloud archive [23:19] https://wiki.ubuntu.com/ServerTeam/OpenStackHA [23:20] https://wiki.ubuntu.com/ServerTeam/OpenStackHA?action=AttachFile&do=view&target=local.yaml [23:20] source: 'cloud:precise-updates/grizzly' [23:20] s/grizzly/newest-openstack [23:20] you can change cloud:precise-updates/grizzly to havanna or folsom, etc [23:20] we prefer LTS, given the support lifespan, etc [23:21] makes sensefor a server, not nessisiarly on a desktop though [23:21] marcoceppi: I saw the OpenStackHA page last night but figured it was Too Old when I saw references to juju 0.7 .. [23:21] sarnold: almost all charms are compatible between juju versions [23:21] now that _is_ magic. :) [23:23] sarnold: it's a promise the core team has made [23:23] think of the pain of updating over 300 charms if juju changes :) === paulczar is now known as zz_paulczar [23:24] wow, 300? nice work :) [23:24] well, we've got 140 "charm store" charms [23:24] also impressive [23:24] but a ton more outside the store [23:43] Has anyone gotten juju connected to Rackspace yet? I see some have looked at doing it, but nothing successful. === CyberJacob is now known as CyberJacob|Away [23:48] flickerfly: i heard yesterday that it should work now