negronjl | m_3: I'll look into the mongodb/juju/precise thing a bit later ... Can't get juju to work with precise yet. | 00:06 |
---|---|---|
imbrandon | hazmat: where does the code for charmworld live on lp ? i peeked at https://code.launchpad.net/charmworld | 00:29 |
imbrandon | hazmat: never mind, i'm an idiot | 00:30 |
imbrandon | err maybe not, i thought i found it | 00:31 |
imbrandon | i guess not though | 00:32 |
hazmat | imbrandon, its not published | 01:03 |
hazmat | SpamapS, sounds reasonable, is there a way to identify the active series on lp? | 01:03 |
imbrandon | ahh that would explain it, just not published or closed ? i dont mind either i was just gonna try and lend a hand with the sorting and maybe upgrade it to bootstrap 2.0 with responsive-ness and make sure it was all in like with the new branding updates like i did on the commmunity sites | 01:05 |
imbrandon | :) | 01:05 |
imbrandon | s/like/line* | 01:05 |
imbrandon | hazmat: so if i can give a hand with it just ping | 01:08 |
m_3 | negronjl: I've got 531 localling bringing up 531 ppa on local provider for precise... seems to be running ok so far | 01:53 |
negronjl | m_3: Can you pastebin a copy of your environments.yaml file ? | 02:06 |
negronjl | m_3: and if you used any odd commands or switches for juju, I'll take those too :) | 02:07 |
negronjl | m_3: I'll look at it a bit later tonight | 02:07 |
SpamapS | hazmat: yes the distro object will show you | 02:12 |
m_3 | negronjl: http://paste.ubuntu.com/933424/ is what I was using from that precise instance to launch other local precise instances | 02:24 |
=== _mup__ is now known as _mup_ | ||
=== almaisan-away is now known as al-maisan | ||
huwshimi | Hi, I'm trying to deploy the openstack-dashboard charm locally (http://charms.kapilt.com/charms/precise/openstack-dashboard). I have set up my environment for local use, but when I do 'juju deplou openstack-dashboard' it says "Error processing 'cs:precise/openstack-dashboard': entry not found". Do I need to do something special to get it to find that charm? | 10:24 |
imbrandon | try cs:oneiric/openstack-dashboard instead | 10:33 |
huwshimi | imbrandon: Same problem :( | 10:34 |
imbrandon | is there and openstack-dashboard listed ni the charmstor ? /me looks | 10:35 |
huwshimi | This is the full output: http://paste.ubuntu.com/933781/ | 10:35 |
imbrandon | bot sure, you'll have to bug hazmat m_3 or SpamapS when one of them gets in this morning | 10:37 |
imbrandon | not* | 10:37 |
huwshimi | imbrandon: No problems. Thanks :) | 10:37 |
imbrandon | try mysql cuz i know that know works, to make sure your env is right | 10:37 |
imbrandon | afk a few | 10:37 |
huwshimi | yeah, mysql worked | 10:38 |
imbrandon | ok yea its breakage from the switch to precise then likely yesterday and one of those 3 will fix it up here in a few hours i'm sure, they are normally active here pretty soon | 10:39 |
huwshimi | No problems, I can wait | 10:40 |
=== al-maisan is now known as almaisan-away | ||
=== almaisan-away is now known as al-maisan | ||
=== matsubara-afk is now known as matsubara | ||
hazmat | imbrandon, its likely because its not 'officially' released yet.. ie been reviewed. it can still be installed from cs:~charmers/precise/openstack-dashboard | 12:43 |
imbrandon | hazmat: ahh | 13:46 |
=== al-maisan is now known as almaisan-away | ||
negronjl | m_3: ping | 16:30 |
m_3 | yo | 16:39 |
m_3 | negronjl: yo | 16:39 |
negronjl | m_3: mongodb works :) | 16:39 |
m_3 | cool... thanks for checking it out! just wanted to make sure we explicitly went through the handout/demo examples after promoting everything to precise | 16:40 |
negronjl | m_3: no worries ... glad it works | 16:41 |
jcastro | negronjl: you did test cloudfoundry on precise right? | 16:42 |
* jcastro runs | 16:42 | |
negronjl | jcastro: not yet ... | 16:42 |
m_3 | it looks like parts of it are still failing for some reason... probably packages | 16:42 |
negronjl | m_3: parts of cloudfoundry ? | 16:43 |
m_3 | http://ec2-23-20-58-1.compute-1.amazonaws.com:8080/ (only a tmp url) | 16:43 |
m_3 | unfortunately, it wasn't capturing logs on these initial precise runs | 16:43 |
negronjl | m_3: I'll work on some troubleshooting as I get time ... | 16:44 |
m_3 | I was going through yseterday knocking them one-by-one to green | 16:44 |
m_3 | negronjl: cool... thanks | 16:44 |
=== matsubara is now known as matsubara-lunch | ||
SpamapS | FYI, pending approval of FFE, I'll be uploading bzr531 to precise later today or early tomorrow. Please give it a good vigorous testing everybody. | 17:21 |
=== matsubara-lunch is now known as matsubara | ||
jcastro | SpamapS: hey, busy? | 17:38 |
jcastro | krish is at openstack and he wants to do a live G+ over the air about juju | 17:39 |
jcastro | want to hop in with me? | 17:39 |
jcastro | and/or m_3 | 17:39 |
robbiew | jcastro: right now? | 17:41 |
jcastro | yeah he just pinged me on twitter | 17:41 |
robbiew | jcastro: does he know you and clint will be there Thu? | 17:41 |
jcastro | yeah | 17:41 |
robbiew | interesting | 17:42 |
jcastro | wanna hop in? | 17:42 |
jcastro | I want to talk about the charm school and stuff | 17:42 |
robbiew | nope | 17:42 |
robbiew | lol | 17:42 |
* robbiew has to start blueprint wrangling for 11.10...and do some I-am-leaving-my-wife-with-you-young-boys-this-week chores :P | 17:43 | |
robbiew | /s/-you-/-two- | 17:43 |
SpamapS | jcastro: sorry Iw as grabbing an early lunch | 18:04 |
jcastro | no worries he hasn't responded back | 18:04 |
jcastro | he says he'll ping me with a suitable time when he gets there | 18:04 |
jcastro | SpamapS: I was thinking we could just talk about the problems we're trying to fix, etc. Basically, the stuff we talk about every week anyway | 18:04 |
SpamapS | sure | 18:05 |
SpamapS | I should like.. shower so I don't look like a bum | 18:05 |
jono | jamespage, hey, FYI - I had to get all this data of my 1TB drive which is taking a while, and then I can kick off the mirror | 18:16 |
jono | the drive was formatted for a Mac (it was used to record our album), so I will get the data off, format it, and then download the mirror so we have a backup | 18:17 |
m_3 | jcastro: hey... I was in a meeting | 18:23 |
m_3 | jcastro: any update on time? | 18:23 |
jcastro | no worries, he hasn't given me a time yet | 18:23 |
m_3 | jcastro: cool | 18:23 |
jcastro | try to stay open today. (hah!) | 18:23 |
m_3 | :) | 18:23 |
m_3 | dude, that was a rude awakening this morning! | 18:24 |
jcastro | yeah sorry | 18:24 |
* m_3 open eyes... crawl outta be... check irc... charmschool in 20mins! | 18:24 | |
m_3 | I'm gonna go get some breakfast... bbiab | 18:26 |
jcastro | SpamapS: and then we should probably talk about charmschool ODS again | 18:27 |
SpamapS | jcastro: indeed, I have rsyslog written though it has a bug that I'm fighting with | 18:30 |
jamespage | jono, great - I think the guys are pulling one locally as well for redundancy | 18:30 |
jono | cool | 18:30 |
jcastro | SpamapS: I know you're in the zone but gitolite promulgation would be swell as well. Looks like shazzner's fixed all your issues | 18:31 |
SpamapS | jcastro: promulgation needed for demo? :) | 18:32 |
jcastro | that would be a cool demo | 18:34 |
yagey | could someone suggest a charm that installs from a cvs source that I can use as a template please? | 18:38 |
marcoceppi | yagey: like an actual CVS source, or just installs from some form of version control? | 18:39 |
yagey | cvs | 18:39 |
yagey | to install the virgo kernel: http://www.eclipse.org/virgo/download/ | 18:40 |
marcoceppi | I don't think we have any that install from cvs source, but it would be similar to if you were doing an install, with regards to the commands used to branch the repo | 18:40 |
marcoceppi | yagey: why not just grab the zip download instead of cvs? | 18:40 |
yagey | ok | 18:40 |
yagey | any examples of a zip based install? | 18:41 |
marcoceppi | yagey: sure, quite a few do | 18:42 |
marcoceppi | let me find one | 18:42 |
marcoceppi | yagey: owncloud, statusnet, roundcube, and a few others grab some form of archive from a remote server for installation | 18:44 |
yagey | thx | 18:44 |
marcoceppi | They all make use of a charm-helper called "ch_get_file" which takes two parameters: a url, and a hash of the payload ( in order to do payload validation ) | 18:45 |
m_3 | yagey: node-app installs from git... shouldn't be too different to change that to cvs | 18:45 |
yagey | k | 18:45 |
m_3 | yagey: but there are more examples out there of ch_get_file | 18:46 |
SpamapS | yagey: cvs, *wow* | 18:48 |
yagey | ? | 18:48 |
SpamapS | shocking to hear that anybody has cvs anywhere anymore | 18:49 |
SpamapS | svn is well over 10 years old | 18:49 |
SpamapS | bzr is better than svn | 18:49 |
SpamapS | git is more popular than bzr | 18:49 |
SpamapS | so many options | 18:49 |
SpamapS | cvs is.. just so bad | 18:49 |
SpamapS | oh | 18:50 |
SpamapS | yagey: they seem to be on git :) | 18:50 |
SpamapS | yagey: https://github.com/eclipse/virgo.apps ... | 18:51 |
yagey | seems they are spread over git/svn/cvs | 18:51 |
SpamapS | yagey: I'd recommend targetting your charm at a released version to start with. You can add a config option to switch to git later. | 18:52 |
yagey | yes, doing that. thx | 18:52 |
yagey | ch_get_file is helpful | 18:52 |
yagey | sigh, my connection to ec2 has started just hanging, any troubleshooting advice please? | 19:02 |
SpamapS | yagey: are you sure its the ec2 connection? | 19:02 |
SpamapS | yagey: sometimes doing 'juju -v ...' will reveal more | 19:02 |
yagey | ah, thanks. claims it's still initializing. but the host logs show it's finished | 19:04 |
yagey | the juju-admin is failing, unrecognized constraints | 19:05 |
SpamapS | yagey: that is a version mismatch between whats on your machine and what your 'juju-origin' is set to. | 19:05 |
yagey | "juju-admin: error: unrecognized arguments: --constraints-data=" | 19:05 |
SpamapS | yagey: most likely you have 'juju-origin: distro' | 19:06 |
SpamapS | yagey: or juju incorrectly guessed that you were using the distro version. | 19:06 |
SpamapS | yagey: dpkg -l juju | 19:06 |
SpamapS | yagey: I suggest putting an explicit 'juju-origin: ppa' in your environments.yaml | 19:07 |
yagey | on my localhost: 0.5+bzr519-1ju on the EC2 boostrap 0.5+bzr398-0ubuntu1 | 19:08 |
SpamapS | yagey: yikes! | 19:09 |
SpamapS | yagey: that 398 version is the distro version from oneiric | 19:09 |
SpamapS | yagey: I suggest making sure you have the latest from the ppa, and then adding a 'juju-origin: ppa' to the environment | 19:10 |
yagey | ok | 19:10 |
SpamapS | yagey: also you should consider deploying on precise. :) | 19:10 |
yagey | what is precise? | 19:10 |
SpamapS | yagey: Ubuntu 12.04 | 19:10 |
SpamapS | oneiric == 11.10 | 19:10 |
yagey | ah | 19:10 |
marcoceppi | Anyone ever seen this before? http://paste.ubuntu.com/934414/ | 19:11 |
SpamapS | marcoceppi: no .. weird | 19:11 |
SpamapS | marcoceppi: newer juju has a more robust zookeeper though, so its possible it would have exploded much more spectacularly before | 19:12 |
marcoceppi | Running the command again appears to hang | 19:12 |
m_3 | marcoceppi: nope... I'm actually excited to see it... usually a connection loss has to be inferred :) | 19:12 |
* marcoceppi checks ec2 status page | 19:12 | |
* marcoceppi runs with -v | 19:13 | |
marcoceppi | http://paste.ubuntu.com/934420/ | 19:14 |
marcoceppi | it just keeps going | 19:14 |
SpamapS | marcoceppi: zk not started yet | 19:14 |
SpamapS | or failed somehow | 19:14 |
marcoceppi | SpamapS: this environment has been bootstrapped for about 5 hours | 19:14 |
SpamapS | marcoceppi: slight chance also that your local ssh is bombed out somehow | 19:15 |
marcoceppi | *it is the current omgubuntu environment* | 19:15 |
marcoceppi | Though, this might explain why wp can't connect to mysql which is running on that bootstrap | 19:15 |
SpamapS | marcoceppi: ssh directly in? | 19:15 |
marcoceppi | aye, I can | 19:16 |
SpamapS | zk running? | 19:16 |
marcoceppi | I guess this is bad: /dev/xvda1 7.9G 7.5G 0 100% / | 19:16 |
marcoceppi | SpamapS: it's not | 19:17 |
SpamapS | ooohhh fuuuuuudddddddgggge | 19:17 |
jcastro | hah | 19:17 |
jcastro | http://www.sadtrombone.com/ | 19:17 |
marcoceppi | There shouldn't be 7.5G of MySQL databases on there | 19:18 |
marcoceppi | yeah, /var/lib/mysql is only 567M | 19:18 |
SpamapS | http://www.youtube.com/watch?v=Yii2mcNq2Zw | 19:18 |
SpamapS | marcoceppi: I bet its the juju agent logs or something silly like that | 19:19 |
SpamapS | or ZK | 19:19 |
yagey | so with ppa and precise, now it refuses to connect: 579: Socket [127.0.0.1:42204] zk retcode=-4, errno=111(Connection refused): server refused to accept the client | 19:19 |
jcastro | m_3: I have an idea. | 19:19 |
jcastro | m_3: feel like experimenting with something? | 19:19 |
marcoceppi | 5.9G./log | 19:19 |
* marcoceppi moves log to /mnt | 19:19 | |
jcastro | daddy needs centralized logging! | 19:20 |
marcoceppi | it's actually /var/log/mysql | 19:20 |
marcoceppi | has a crap ton of 101M files mysql-bin.XXXXXX | 19:20 |
marcoceppi | all have like, json in them | 19:21 |
SpamapS | binary logging | 19:21 |
SpamapS | marcoceppi: flush logs; | 19:21 |
SpamapS | marcoceppi: if you're not going to use replication, you can turn that off | 19:21 |
SpamapS | with log-bin | 19:21 |
SpamapS | marcoceppi: and you can pretty much delete all of them | 19:22 |
m_3 | jcastro: hey... sure | 19:23 |
jcastro | http://googledevelopers.blogspot.com/2012/04/add-spdy-support-to-your-apache-server.html | 19:23 |
jcastro | We totally should just toss that in a charm right now and tell people about it | 19:23 |
marcoceppi | SpamapS: Okay, that's fixed. For ZK, just `restart zookeeper`? | 19:23 |
jcastro | "I read something cool and I want to try it right now." | 19:24 |
SpamapS | marcoceppi: yeah that should be fine | 19:24 |
SpamapS | marcoceppi: service zookeeper restart btw .. don't call restart directly | 19:24 |
marcoceppi | good to know | 19:24 |
jcastro | m_3: they have packages, etc. And the byobu-classroom charm you just used uses mod_ssl already afaict. | 19:24 |
SpamapS | marcoceppi: and actually .. for oneiric.. stop/start is better | 19:24 |
SpamapS | marcoceppi: in 12.04 I fixed 'service' to do stop/start .. in 11.10 and earlier it uses the broken headed upstart 'restart' command | 19:24 |
SpamapS | which has a few bugs and does weird stuff | 19:24 |
marcoceppi | sweet juju status works again | 19:25 |
marcoceppi | whew! | 19:25 |
marcoceppi | that was fun | 19:25 |
SpamapS | marcoceppi: time to create a troubleshooting.rst | 19:25 |
jcastro | m_3: so I was thinking, find a charm that people would want to play with, toss spdy on it, demo! | 19:25 |
marcoceppi | yup | 19:25 |
SpamapS | what are the available options for spdy? | 19:26 |
m_3 | jcastro: are you trying to show this at ods? | 19:26 |
SpamapS | mod_spdy? nginx? | 19:26 |
jcastro | no, this just got published today this mod_spdy snapshot from google | 19:26 |
marcoceppi | SpamapS: will MySQL just keep on chugging with binary logging, or can I set like a ceiling? | 19:26 |
SpamapS | marcoceppi: its designed to be an audit log.. so you should turn it off | 19:26 |
SpamapS | marcoceppi: it will increase write performance by about 5x ;) | 19:27 |
jcastro | my idea is basically tell the "I saw something neat on the internet and want to mess with it, modify a charm, done." | 19:27 |
jcastro | it just so happens this came out right now | 19:27 |
marcoceppi | SpamapS: too bad the database rarely gets touched :) | 19:27 |
SpamapS | marcoceppi: enough to fill the disk ;) | 19:29 |
marcoceppi | SpamapS: it's weird because we don't have this problem on the previous environment | 19:30 |
marcoceppi | Though, we just fixed an issue with apc that was causing php to go apeshit | 19:30 |
m_3 | jcastro: dude... I've really gotta get automated tests working on ec2. lemme see what I can put together in between deployments this afternoon | 19:33 |
SpamapS | marcoceppi: did you set tuning on the mysql service? | 19:33 |
SpamapS | marcoceppi: I believe "fast" turns off log-bin | 19:33 |
jcastro | m_3: I'm not saying do it, I'm just thinking out loud! | 19:33 |
marcoceppi | SpamapS: perfect I'm going to use that instead | 19:33 |
m_3 | jcastro: testing got bumped over the weekend to upgrade charms to precise | 19:33 |
SpamapS | marcoceppi: yeah for them, the super fast tuning is the way to go. | 19:33 |
jcastro | m_3: actually let me back up, do you think it'd be a cool demo? | 19:34 |
m_3 | jcastro: yeah, of course it's a good story | 19:34 |
marcoceppi | SpamapS: <3 the MySQL config options | 19:34 |
m_3 | faster on the heels of the tool announcement the better too | 19:34 |
marcoceppi | It was rather cool, the database for omgubuntu was down for about 20 mins, and the site just kept serving from cache | 19:34 |
jcastro | marcoceppi: never let em see you sweat | 19:34 |
marcoceppi | I don't think anyone knew, I only knew because I just happened to try to log in to the admin panel | 19:35 |
m_3 | marcoceppi: that's awesome | 19:35 |
SpamapS | marcoceppi: hrm, no.. fast doesn't turn it off | 19:35 |
SpamapS | marcoceppi: but actually, it shouldn't be on by default anyway | 19:35 |
m_3 | jcastro: the openstack thing is a whole panel? | 19:36 |
marcoceppi | SpamapS: I just turned it off manually. I can add a patch to the charm in a few mins. Should it be a config option like log-bin true|false then by default put the bins on /mnt ? | 19:36 |
SpamapS | ahhh.. $JUJU_RELATION_ID is quite handy | 19:36 |
jcastro | m_3: it's going to be more like shooting the breeze with Krish. | 19:36 |
m_3 | jcastro: email sounded like an actual panel | 19:37 |
jcastro | what email? | 19:37 |
SpamapS | marcoceppi: I think it should be a config option, string, default to "", and if you have 'safest' tuning, then it sets it to /var/log/mysql/mysql-binlog or whatever it is now as the default... | 19:37 |
m_3 | can't see who else was copied... was a g+ mail | 19:38 |
SpamapS | marcoceppi: and if you have 'fast' or 'unsafe' then it should be left off | 19:38 |
m_3 | jcastro: PM | 19:38 |
marcoceppi | SpamapS: then the config option would be to what? over-ride the default path? | 19:38 |
jimbaker | SpamapS, glad to hear that about $JUJU_RELATION_ID | 19:39 |
yagey_ | still can't run juju status with ppa and precise, and suggestions please? ^C2012-04-17 12:39:17,262:9336(0x7f4e3126c700):ZOO_ERROR@handle_socket_error_msg@1603: Socket [127.0.0.1:57177] zk retcode=-4, errno=112(Host is down): failed while receiving a server response | 19:40 |
yagey_ | interesting, the host halted itself and refusing connections | 19:43 |
SpamapS | yagey_: I'd guess that ZK did not start properly | 19:43 |
SpamapS | jimbaker: yeah, its useful because its consistent through the entire relation lifecycle and can be used to run relation-get outside of hooks. :) | 19:44 |
SpamapS | jimbaker: so for instance in the rsyslog-forwarder charm I just wrote, I use it to control the name of the per-remote-host config file I add.. broken can just 'rm -f /etc/rsyslog.d/80-$JUJU_RELATION_ID.conf' | 19:44 |
SpamapS | jimbaker: before, broken hooks were always hard to write because they had no context available. | 19:44 |
SpamapS | jcastro: speaking of charms that need reviewing.. I have submitted 3 in the last 2 days, and I have 2 more | 19:48 |
jcastro | \o/ | 19:48 |
jimbaker | SpamapS, sounds like it worked as intended. very nice indeed! | 19:57 |
bkerensa | jcastro: you did not require signature on the UPS shipments did you? | 20:09 |
SpamapS | 2012-04-17 13:08:06,922 INFO Charm deployed as service: 'mod-spdy' | 20:09 |
* SpamapS waits patiently while his instances start up | 20:09 | |
jcastro | don't forget to push! | 20:10 |
jcastro | I want to see how you did it | 20:10 |
jcastro | so I can learn | 20:10 |
jcastro | since I see the first one, and then yours. | 20:10 |
jcastro | hah, can you believe I work at an engineering company? I should be killed. | 20:10 |
SpamapS | lp:~clint-fewbar/charms/precise/mod-spdy/trunk | 20:10 |
bkerensa | =o | 20:11 |
jcastro | \o/ | 20:11 |
jcastro | SpamapS: ah I like how you did the arch thing there | 20:13 |
SpamapS | jcastro: they should have an apt source.. or a PPA :) | 20:13 |
jcastro | SpamapS: I think what we do is on the very first instance we see where their package puts the sources.list for their thing | 20:14 |
jcastro | and just put that in the charm | 20:14 |
nteon | okay, so I'm new to juju. I'd like to use it to deploy our cluster of machines | 20:16 |
bkerensa | nteon: What would you like to know? | 20:16 |
nteon | including a couple of postgres machines | 20:16 |
nteon | bkerensa: can I tell juju about things like which disks/set of disks make up a software raid array, and have them attached and then mounted on an ec2 box? | 20:17 |
nteon | I'm not super clear on where juju falls into the stack I guess | 20:17 |
bkerensa | I dont think juju does that yet but I could be wrong | 20:18 |
jcastro | nteon: I have some videos you might want to check out | 20:18 |
bkerensa | ^ | 20:19 |
jcastro | http://www.brighttalk.com/webcast/6793/41933 | 20:19 |
nteon | jcastro: awesome, I will check it out in a min | 20:19 |
bkerensa | jcastro is the juju guy | 20:19 |
bkerensa | :D | 20:19 |
jcastro | nteon: we're likely above the disk/array level | 20:20 |
_mup_ | juju/apt-proxy-support r532 committed by kapil.thangavelu@canonical.com | 20:21 |
_mup_ | update cloudinit renderer for apt proxy url | 20:21 |
SpamapS | hrm, open-port from a subordinate might not work :-/ | 20:22 |
SpamapS | 2012-04-17 13:19:58,221 unit:mod-spdy/1: unit.hook.api DEBUG: opened 443/tcp | 20:22 |
SpamapS | Yeah, doesn't work | 20:22 |
SpamapS | opens the port for the subordinate, but the subordinate has no watch for that so it never gets opened for the unit | 20:23 |
SpamapS | bcsaller1: ^^ thoughts? | 20:23 |
bcsaller1 | SpamapS: it should be better documented, the principal service controls the networking for the subordinate as its really running the container, but some adjustment there is most likely needed. | 20:24 |
hazmat | marcoceppi, re the connection loss, we guard against that for the agents but not for the cli which is short lived connection | 20:24 |
nteon | jcastro: jcastro: watching now | 20:24 |
SpamapS | bcsaller1: indeed, subordinates will need to open ports | 20:24 |
SpamapS | I appreciate teh conservative take on it.. but subordinates are adding to the primary, so they should definitely be allowed to open ports | 20:25 |
bcsaller1 | SpamapS: but exposing a port for the subordinate from the principal works | 20:25 |
bcsaller1 | thats just counter intuitive | 20:25 |
hazmat | yeah.. the port stuff is setup to watch against the service unit assigned to the machine, in a subordinate case that distinction is never made | 20:25 |
SpamapS | the use case is quite simple to imagine. I want to add a management daemon that gets queried from outside the network. | 20:25 |
hazmat | bcsaller1, its also defeats the purpose of subordinates.. | 20:26 |
hazmat | ie. adding out of band functionality to a principal | 20:26 |
SpamapS | thats.. the entire point of subordinates | 20:26 |
SpamapS | I get why it doesn't do it today | 20:27 |
SpamapS | but its going to be desired | 20:27 |
SpamapS | already is actually :) | 20:27 |
bcsaller1 | easily done | 20:27 |
hazmat | SpamapS, i'd like to turn down the logging for the ssl stuff, its a bit inane in its verbosity 4 messages for the same thing | 20:27 |
bcsaller1 | SpamapS: mind adding a bug on it? | 20:27 |
=== zyga is now known as zyga-afk | ||
hazmat | SpamapS, this is patch to clean it up http://paste.ubuntu.com/934535/ | 20:28 |
SpamapS | hazmat: its being clear about each vulnerability. -l ERROR gets rid of it. | 20:28 |
SpamapS | bcsaller1: I will definitely | 20:28 |
bcsaller1 | SpamapS: great, thanks | 20:29 |
SpamapS | hazmat: I'd like it to log 20 messages :) | 20:29 |
SpamapS | but I only had enough text for 3 | 20:29 |
SpamapS | ugh | 20:36 |
SpamapS | lodgeit is just broken | 20:36 |
SpamapS | ok.. lets try statusnet | 20:36 |
SpamapS | jcastro: lodgeit seems broken in precise, should be investigated | 20:36 |
jcastro | k | 20:38 |
jcastro | ditto byobu-classroom | 20:38 |
nteon | heh. I don't really see the point of LTS releases if you're going to deploy a bunch of unpackaged software on top of it. in that case what is left that is 'stable'? libc and the kernel? | 20:38 |
nteon | (that wasn't in response to anything in particular) | 20:39 |
SpamapS | hazmat: btw, thanks for the 'open-port' script in juju-jitsu. Totally works perfectly! | 20:39 |
jcastro | nteon: this is additive, the charms have an option of doing that if you want to | 20:39 |
jcastro | nteon: so like, if you want to use the upstream version of wordpress instead of what's packaged you can do that, etc. | 20:40 |
SpamapS | jcastro: https://ec2-107-22-22-168.compute-1.amazonaws.com/ | 20:40 |
SpamapS | *SPDY* | 20:40 |
nteon | jcastro: well this video talks about how charms can help keep your software up to date when you're running on an LTS release... in that case why not just use whatever the latest ubuntu release is? | 20:41 |
jcastro | I see it's served via spdy, but it's a 404 | 20:41 |
jcastro | nteon: you could do that if you want, but people do want a stable OS that is very stable and long lived but then have their application on top of it be new | 20:42 |
SpamapS | jcastro: cs:~clint-fewbar/precise/mod-spdy *should* work :) | 20:43 |
SpamapS | jcastro: well, once the store picks it up | 20:44 |
SpamapS | I have to run now | 20:44 |
jcastro | <3 | 20:44 |
jcastro | is the add-relation normal? | 20:44 |
SpamapS | jcastro: included a README on how to make it work | 20:44 |
jcastro | ah | 20:44 |
jcastro | thanks! | 20:44 |
SpamapS | jcastro: also open a bug against lodgeit.. its totally busted | 20:44 |
jcastro | k | 20:44 |
nteon | jcastro: yea, I guess I just disagree with that line of thinking :P but thats neither here nor there | 20:45 |
nteon | is there a way of having a new juju environment pick up a copy of the DB from somewhere? | 20:45 |
jcastro | nteon: it's ok, it's up to you how to run your cloud. :) | 20:45 |
jcastro | which db? | 20:45 |
nteon | any? right now mysql, in a month or two, postgres | 20:46 |
jcastro | oh, well right now both of those charms just apt-get install mysql or postgres | 20:46 |
SpamapS | nteon: sure, thats how omgubuntu does it, they dump to s3 and new deployments import that dump | 20:46 |
SpamapS | ok gone | 20:46 |
jcastro | oh I see what you mean | 20:46 |
nteon | SpamapS: thanks, let me look at the omgubuntu stuff | 20:47 |
jcastro | http://jujucharms.com/~marcoceppi/oneiric/omgubuntu-wp | 20:47 |
m_3 | jcastro: nice, when did we get jujucharms.com? | 20:48 |
jcastro | a few weeks ago, kapilt'ed | 20:48 |
m_3 | that rocks | 20:48 |
nteon | yes, I'm looking at db_relation_changed right now | 20:49 |
imbrandon | heya robbiew , I got juju working on OSX , i need to do a bit more cleanup on the pkg and such but it should be ready for semi public testing in the next 24 hours i'd say, i just had to do some very very wrong things to get zkpython to play nice that i need to work out | 20:49 |
imbrandon | but overall its "working" :) | 20:49 |
m_3 | imbrandon: cool | 20:49 |
imbrandon | m_3: it will be avail via "brew install juju" too :) charm-tools and such are next :) | 20:50 |
robbiew | imbrandon: nice! | 20:50 |
hazmat | the old charms.kapilt.com redirects transparently to the new domain | 20:51 |
nteon | okay, so I have 2 things I'm interested in: having postgres on a raid, and having a new postgres instance be initialized from an existing db dump. it looks like the answer is that I would do both with hooks? | 20:52 |
imbrandon | the existing db dump will for sure | 20:52 |
imbrandon | the raid , it _might_ be better to start with a specific ami ? not sure | 20:53 |
nteon | imbrandon: okay, so I would roll my own ami in that case I guess | 20:54 |
imbrandon | nteon: cloud servers are "cheap" in that sense though, i think the more natuaral way to ensure raid levels of service would be full server replication though now a days | 20:54 |
imbrandon | e.g run 2 or 3 postgres instead of one monster | 20:54 |
nteon | imbrandon: its not space, its speed. individual ebs disks are too slow | 20:55 |
imbrandon | that way it can be scaled with juju to :) | 20:55 |
imbrandon | nteon: sure i understand that, read replicas are fast | 20:55 |
imbrandon | nteon: and i'm not 100% sold on the idea, just was the first thing that came to mind | 20:56 |
nteon | imbrandon: writes are too slow (and unpredictable) on individual ebs volumes | 20:56 |
imbrandon | nteon: true at high service levels but at that point i'm looking at memcache persistant to the db or RDS like services | 20:57 |
imbrandon | nteon: but yea a raided ami would likely work | 20:57 |
nteon | imbrandon: well, RDS does striping over several ebs volumes in the background for you, you just don't have access to a shell on that machine | 20:57 |
imbrandon | exactly, thus i was talking about doing that too :) | 20:58 |
imbrandon | not trying to sell ya on it though, just was making sure you had thought about the other options ;) | 20:58 |
hazmat | SpamapS, the top link on the browser ('charms') goes straight to the current series now, hopefully that suffices | 20:59 |
imbrandon | and juju deploy posgress , to scale out more is much sexier than playing with raid imho :) | 20:59 |
imbrandon | but yea, infact __one__ way to use an existing dump is the way marcoceppi has setup for when we deploy omgubuntu.co.uk , you can find the charm here to see it as an example, there is alot of omg specific stuff in there just ignore it, and its mysql but i'm sure you can translate the diffs http://jujucharms.com/~marcoceppi/oneiric/omgubuntu-wp | 21:01 |
imbrandon | nteon: ^^ | 21:01 |
imbrandon | nteon: it grabs the latest backup from s3 and uncompresses it and loads it up if the db is currently blank ( new deploy etc ) | 21:02 |
imbrandon | but just adding webheads it checks and skips it | 21:02 |
imbrandon | jcastro: btw when i went to the doc's today i stopped at microcenter and got me a "real" keyboard :) lol | 21:04 |
hazmat | nteon, we don't have a good way to expose that at the moment re ebs raid volumes, ideally volume management should be in the core | 21:06 |
hazmat | given the available tech, it can either be done inline to the charm at the cost of pushing ec2 creds there | 21:07 |
hazmat | or via an ebs charm that would allow the creation mapping of the volumes | 21:08 |
hazmat | its really a provider specific functionality so ideally the framework would be to abstract that from a charm perspective | 21:09 |
marcoceppi | ahhhh, I can't turn off binary dumping for some reason | 21:10 |
=== almaisan-away is now known as al-maisan | ||
=== al-maisan is now known as almaisan-away | ||
imbrandon | marcoceppi / m_3 : added you both to the "juju-hackers" group on github so you should both have direct commit access to the juju brew formula ( and later charm-tools ) etc , if you have other juju related brews feel free to put the forumlas in with the others | 22:26 |
imbrandon | and i'll make sure they are easy to install etc just like with juju | 22:27 |
imbrandon | if anyone else has a github acct and wants to help with the OSX juju hacking just ping me and i'll add ya | 22:27 |
imbrandon | ( marcoceppi and m_3 are full admins too and can add ppl as needed as well ) | 22:28 |
jono | jamespage, you there? | 22:34 |
m_3 | imbrandon: cool... thanks, I'll try it out when I get home and have access to an osx box | 22:47 |
imbrandon | m_3: kk, the instructions at http://juju-hackers.github.com/homebrew-juju/ are a little ahead of themself | 22:48 |
imbrandon | e.g the pull req isnt done yet,but you can just use brew install ./Libraries/Formula/juju.rb | 22:49 |
imbrandon | for tonight | 22:49 |
jamespage | jono: am now - wassup? | 23:30 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!