_mup_ | Bug #1195537 was filed: "type: local" doesn't work in host with kernel option 'ds=nocloud' <juju:New> <https://launchpad.net/bugs/1195537> | 03:03 |
---|---|---|
=== defunctzombie is now known as defunctzombie_zz | ||
=== efries_ is now known as efries | ||
Daviey | Anyone else seen this with juju-core? http://pb.daviey.com/WKch/ | 08:57 |
Daviey | evilnickveitch: Hey, I see the docs are updated.. but i noticed https://juju.ubuntu.com/resources/ the menu on the left doesn't work | 09:01 |
evilnickveitch | Daviey -ok, will look at that | 10:15 |
AskUbuntu | MAAS. Adding nodes | http://askubuntu.com/q/313790 | 10:32 |
Akira1 | good morning and happy friday all, anyone around? | 12:18 |
Akira1 | hmm perhaps not, I've got an issue, trying to update/upgrade a charm I have deployed due to additional config elements i've added to a config.yaml for that charm | 12:26 |
Akira1 | turns out the /var/lib/juju/units/<service unit>/charm contents are not being updated | 12:27 |
=== wedgwood_away is now known as wedgwood | ||
Akira1 | still crickets? | 13:05 |
paraglade | Akira1: did you the value in the <charm_home>/revision file? | 13:08 |
Akira1 | hmm actually no! | 13:10 |
Akira1 | however, everytime I attempt an upgrade, it is incrementing the internal revision | 13:11 |
Akira1 | i.e. this was earlier: 2013-06-28 08:01:54,406 unit:cdh4-master/0: charm.upgrade DEBUG: Replacing charm local:precise/cdh4-17 with local:precise/cdh4-18. | 13:11 |
Akira1 | 2013-06-28 08:01:54,415 unit:cdh4-master/0: charm.upgrade DEBUG: Charm has been upgraded to local:precise/cdh4-18. | 13:11 |
Akira1 | i'm honestly not entirely sure what is in the charm cache at this point, the hook scripts are missing things that should have been there in every revision | 13:12 |
Akira1 | is there a way I can workaround this and force update the hooks on each deployed service unit manually? | 13:12 |
paraglade | well I am not sure. Might be a good idea to wait for one of the more seasoned folks to jump in. :) | 13:14 |
Akira1 | fair enough ;) | 13:14 |
Akira1 | everything had been working swimmingly until last night =/ | 13:18 |
ehg | Akira1: my (go) juju upgrade-charm has a --force option, not sure what it actually does though :) | 13:18 |
ehg | i bump the number in the revision file sometimes to get it to update stuff | 13:19 |
ehg | (locally) | 13:19 |
Akira1 | well my understanding was they'll hash out most things | 13:19 |
Akira1 | basically what is happening here is we keep on needing to add config params | 13:20 |
Akira1 | so we keep updating config.yaml, and updating the hook to deal with the new params | 13:20 |
Akira1 | i've done it about 6 times now | 13:20 |
ehg | we had to use juju set --config=<file> <service> to get the new config values in | 13:20 |
Akira1 | but the params i added last night aren't getting picked up | 13:20 |
Akira1 | I inherited all this stuff from someone else so I didn't do this from scratch | 13:21 |
Akira1 | the hook script is missing a config-changed hook =/ | 13:21 |
ehg | Akira1: the unit isn't in an error state is it? | 13:25 |
Akira1 | it sure is now | 13:25 |
ehg | i think juju queues up hook runs until you resolve it | 13:25 |
ehg | so juju resolved <unit> may resume upgrades | 13:25 |
ehg | not completely sure :) | 13:26 |
Akira1 | im giving that a shot now too | 13:26 |
Akira1 | i was doing resolve -retries all night | 13:26 |
Akira1 | heh | 13:26 |
Akira1 | so that does make things run again | 13:28 |
Akira1 | and fail | 13:28 |
Akira1 | it is so weird, I'm actually on the service unit itself, blowing up teh charm dir | 13:28 |
Akira1 | forcing it to redownload it | 13:28 |
Akira1 | and it redownloads the wrong one everytime | 13:28 |
ehg | bit you increase the revision number in the revision file? | 13:29 |
ehg | s/bit/did | 13:29 |
Akira1 | you know | 13:31 |
Akira1 | you might be on to something here | 13:31 |
Akira1 | my revision # in the charm was out of sync with what juju thought it had | 13:32 |
Akira1 | when you do a juju upgrade-charm it will increment the number no matter what | 13:32 |
Akira1 | so I was on local:precise/cdh4-20 even though my revision file said 10 | 13:32 |
Akira1 | i just bumped up the value in revision to 30 | 13:32 |
Akira1 | and the juju dry-run says it will upgrade: 2013-06-28 08:31:29,783 INFO Service would be upgraded from charm 'local:precise/cdh4-20' to 'local:precise/cdh4-30' | 13:33 |
ehg | we have problems because they are multiple people on different machines doing stuff with the local juju repo | 13:33 |
ehg | *there | 13:33 |
Akira1 | similar type of stuff? | 13:33 |
Akira1 | well lets cross our fingers | 13:33 |
Akira1 | well you guys don't have to, but I sure as hell am | 13:33 |
ehg | yeah :) | 13:34 |
Akira1 | omg omg | 13:34 |
Akira1 | i think this worked! | 13:34 |
Akira1 | that sure is messed up | 13:34 |
ehg | woot | 13:34 |
ehg | so in the juju docs somewhere it mentions local repos causing weirdness if you have multiple people using them | 13:35 |
Akira1 | im going to have to dig that up | 13:35 |
ehg | imho it's something that juju should just support | 13:35 |
Akira1 | juju site got updated last night | 13:35 |
Akira1 | so a bunch of google'd links are 404s now | 13:35 |
ehg | although i was wondering if we could run our own juju repository | 13:35 |
ehg | ooh | 13:36 |
Akira1 | well anytime you have something like that, you could try and fake out the remote repo | 13:36 |
Akira1 | or maybe use launchpad? | 13:36 |
Akira1 | *shrug* | 13:36 |
Akira1 | it'd be pretty neat to be able to pull from git or something too | 13:36 |
Akira1 | since we obviously have our own git repo, svn, etc | 13:37 |
ehg | yeah, no bzr pls :) | 13:37 |
Akira1 | :) | 13:37 |
Akira1 | you guys know if anyone is doing any consulting in this space? | 13:37 |
Akira1 | i've got 6 days left with this team and I'm afraid juju is going to die with me when i leave | 13:38 |
ehg | maybe canonical? | 13:39 |
Akira1 | yeah im thinking of calling them on monday | 13:39 |
Akira1 | as with any services org, I'm sure they will SAY they can do something | 13:39 |
Akira1 | and then send someone out who never saw it before | 13:39 |
Akira1 | I saw that soooo many times with IBM | 13:39 |
Akira1 | "oh sure we're sending you an expert!, only $250/hr!" | 13:39 |
ehg | haha, well i think juju is pretty important to them | 13:40 |
ehg | so hopefully they can help | 13:40 |
Akira1 | yeah, we shall see | 13:41 |
Akira1 | while folks are feeling chatty... what platform are you running against, have you tried this stuff against MaaS? | 13:42 |
Akira1 | i'm using maas and things are going sort of swimmingly, but I get occasional http 401 errors from juju commands due to oauth failures | 13:42 |
ehg | ec2 for us | 13:42 |
ehg | was MaaS easy to set up? | 13:43 |
Akira1 | actually, yes | 13:44 |
Akira1 | I ran into a couple weird things | 13:44 |
jcastro | hey so we should have the doc 404 links fixed soon | 13:44 |
Akira1 | awesome :) | 13:44 |
Akira1 | with maas you can theoretically install it right from the OS boot media | 13:45 |
Akira1 | but it doesn't work | 13:45 |
ehg | using PXE? | 13:45 |
Akira1 | yeah | 13:45 |
Akira1 | but if you install that way you run into some chicken/egg scenarios with services coming up in the right order | 13:45 |
Akira1 | so I blew up the machine, just installed 13.04 on it | 13:46 |
Akira1 | THEN installed maas and it worked fantastically | 13:46 |
jcastro | waas the maas in 13.04 different than the one you used before? | 13:46 |
Akira1 | this was setting up a fresh cluster jcastro | 13:47 |
jcastro | also in today's news, Frank has an OSX build | 13:47 |
ehg | nice | 13:47 |
Akira1 | some folks had submitted a bug on what i saw, and the bug was supposedly fixed | 13:47 |
Akira1 | but alas, such was not the case | 13:47 |
ehg | goju compiles fine on OS X - pretty cool | 13:48 |
jcastro | I think all we need to do now is find out where to put it | 13:50 |
ehg | my OS X using colleagues would love it if they could go "brew install juju-client" or something similar | 13:54 |
jcastro | yep, that's what we should do | 13:56 |
jcastro | has homebrew pretty much taken over the package manager space on osx? | 13:57 |
ehg | i'm not the best person to ask that, but it seems homebrew is preferred to macports in our office at least | 13:58 |
jcastro | these homebrew formulas look pretty simple | 13:59 |
paraglade | has anyone used Tsuru - http://www.tsuru.io/ ? “juju” is the default provisioner used by Tsuru and claims that "It’s a extended version of Juju, supporting Amazon’s Virtual Private Cloud (VPC) and Elastic Load Balancing (ELB)." | 14:03 |
=== danilos__ is now known as danilos | ||
jcastro | hey EvilBill | 14:23 |
jcastro | I mean evilnickveitch | 14:23 |
jcastro | https://juju.ubuntu.com/docs/write-charms.html | 14:24 |
jcastro | do you think we could toss a 404 page up in the meantime? | 14:24 |
jcastro | something like "We've redone our docs, click here to get to the main page" | 14:24 |
evilnickveitch | jcastro, sure. well, we should have a redirects file | 14:27 |
jcastro | Yeah but I don't think Marco gets back to work until like Monday right? | 14:27 |
evilnickveitch | jcastro, yeah, I can put a dummy page in there to do a quick redirect and get it in before the next sync, that should hold us off until monday. | 14:32 |
evilnickveitch | jcastro - are there any others you can think of? | 14:32 |
Akira1 | sounds like someone needs to write some redirect rules! | 14:34 |
Akira1 | btw new layout looks nice, good work guys :) | 14:34 |
evilnickveitch | Akira1, thanks! We have the rules, but they are in someone's draft folder until they get back from holidays | 14:35 |
=== defunctzombie_zz is now known as defunctzombie | ||
evilnickveitch | jcastro - done that one, btw | 14:39 |
* jcastro nods | 14:41 | |
jcastro | evilnickveitch: ok, so what am I to do with getting-started? | 14:42 |
jcastro | you wanted that to redirect to the doc page right? | 14:42 |
evilnickveitch | jcastro - yeah, that would be best | 14:42 |
evilnickveitch | and the other getting started too | 14:42 |
jcastro | what other getting started? | 14:42 |
evilnickveitch | I thought there were 2 in different places on j.u.c | 14:43 |
evilnickveitch | ? | 14:43 |
evilnickveitch | or maybe I am mad | 14:43 |
jcastro | https://juju.ubuntu.com/get-started/ | 14:43 |
jcastro | you're probably thinking of the one on normal ubuntu.com | 14:43 |
evilnickveitch | ah, ok | 14:43 |
evilnickveitch | jcastro - can you change that link? | 14:44 |
=== 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 | ||
=== frankban_ is now known as frankban | ||
jcastro | juju charm school in 5 minutes! | 16:55 |
jcastro | follow along on http://ubuntuonair.com | 16:55 |
niemeyer | o/ | 17:01 |
jcastro | PLease post questions in this channel once we get started with the walkthrough | 17:07 |
jcastro | one sec | 17:12 |
arosales | jcastro, looks to be back now | 17:13 |
robbiew | charms *are* code...so document them well :) | 17:18 |
jcastro | the guy with the stable net connection ends up being the worst one | 17:19 |
negronjl | I guess I'm scaring the crowd :) | 17:19 |
jcastro | stand by please! | 17:19 |
henninge | where are the docs for the juju-tools? (unit-get etc.) | 17:20 |
robbiew | jcastro: it's fine for me | 17:21 |
arosales | jcastro, back now | 17:21 |
negronjl | http://jujucharms.com/charms/precise/mongodb/hooks/hooks.py | 17:21 |
niemeyer | Talking about docs, two quick plugs: | 17:22 |
henninge | jcastro: It's me ;-) Hi! https://launchpad.net/~henninge | 17:22 |
niemeyer | 1) The technical documentation for juju was *just* updated, and looks so much better: http://juju.ubuntu.com/docs | 17:22 |
henninge | niemeyer: Hi! Yeah, but I am missing stuff or not finding it. | 17:23 |
niemeyer | 2) I've you've missed it, last week I've published a long due high-level blog post about what juju is: http://blog.labix.org/2013/06/25/the-heart-of-juju | 17:23 |
niemeyer | s/I've you've/If you've/ | 17:23 |
henninge | The hooks docs seem to be gone as well. | 17:23 |
=== pretto is now known as Prett0 | ||
niemeyer | henninge: Fair point, we'll look into this and see if the new documentation has dropped content | 17:24 |
jcastro | Yeah I think that page is missing | 17:24 |
jcastro | I can't find it anywhere | 17:25 |
arosales | fyi, old docs are still available at | 17:25 |
arosales | http://jujucharms.com/docs/ | 17:25 |
niemeyer | The writing a charm document seems pretty light | 17:25 |
henninge | arosales: Great, thanks! | 17:25 |
arosales | so if you find a missing page please file a bug via https://bugs.launchpad.net/juju-core/+filebug | 17:26 |
jcastro | yeah it's still under construction | 17:26 |
jcastro | the write charms part | 17:26 |
arosales | jcastro, but all content should be in the new docs ported from the new docs | 17:26 |
arosales | specifically the old docs and new docs should have the same content | 17:26 |
arosales | if something is missing that is a valid bug | 17:26 |
arosales | henninge, thanks for trying out our docs :-) | 17:28 |
niemeyer | There we go again | 17:29 |
niemeyer | Oh, it's back | 17:29 |
henninge | There is also pages like this that now have dead links: | 17:29 |
henninge | http://askubuntu.com/questions/82683/what-juju-charm-hooks-are-available-and-what-does-each-one-do | 17:29 |
jcastro | the rewrites haven't been deployed yet | 17:29 |
jcastro | those will be fixed on monday | 17:30 |
arosales | redirects that is | 17:30 |
robbiew | jcastro: I'm seeing negronjl | 17:30 |
negronjl | robbiew, fixed ? | 17:30 |
jcastro | fixed | 17:30 |
henninge | Ah, good to hear ;-) Great work | 17:31 |
henninge | not fixed for me | 17:31 |
robbiew | nope | 17:31 |
robbiew | bingo! | 17:31 |
henninge | yeah | 17:31 |
henninge | No, it's fine now | 17:32 |
niemeyer | jcastro: Do we have a fallback deployment of the docs? | 17:32 |
niemeyer | jcastro: Having no docs whatsoever about a lot of those topics is a step backwards :( | 17:32 |
arosales | niemeyer, yes the old docs are still at http://jujucharms.com/docs | 17:32 |
niemeyer | arosales: Sweet, thanks | 17:33 |
niemeyer | arosales: We need a link from the new docs there | 17:33 |
arosales | niemeyer, please file any bugs you find for missing content | 17:33 |
niemeyer | arosales: Well, that sounds backwrads | 17:33 |
niemeyer | arosales: The content is there.. whoever is rewriting it should make sure that things are being dropped on the floor | 17:33 |
niemeyer | are not | 17:33 |
arosales | niemeyer, yup we'll udpate jujucharms.com when the docs have some more run time on juju.u.c | 17:34 |
henninge | I can read | 17:34 |
arosales | niemeyer, agreed | 17:34 |
arosales | niemeyer, I am saying we have tried to port all content over | 17:34 |
arosales | however if you notice missing content please let us know via a bug report | 17:34 |
jcastro | I don't think we ever documented the juju tools did we? | 17:34 |
arosales | niemeyer, the task before going live with the new docs was to make sure all "old" content was available in the "new" docs | 17:35 |
arosales | so all content ~should~ be there, but I am sure we will find some issues here and there as we use the new docs :-) | 17:35 |
henninge | Where does private-address get set? Is that automatic? | 17:36 |
niemeyer | arosales: Sorry, but that's not realistic | 17:37 |
niemeyer | arosales: Just look at https://juju.ubuntu.com/docs/authors-charm-writing.html | 17:37 |
niemeyer | arosales: and compare that with jujucharms.com/docs/charm.html | 17:37 |
niemeyer | arosales: A *ton* of important information was dropped | 17:37 |
hazmat | and formatting there is a bit off as well | 17:38 |
niemeyer | Yeah | 17:38 |
hazmat | niemeyer, that new doc is a straight copy of.. http://jujucharms.com/docs/write-charm.html | 17:38 |
arosales | so we are still polishing :-) | 17:39 |
jcastro | yeah I don't think Nick considers the Writing a Charm section complete | 17:39 |
arosales | but please file a bug even if it is saying content is missing from this page and we'll track it down | 17:39 |
arosales | https://bugs.launchpad.net/juju-core/+filebug | 17:39 |
jcastro | I think this first drop was to concentrate on the end user bits | 17:39 |
arosales | formatting, or any isses you see :-) | 17:39 |
jcastro | the docs are html5, if someone wants to just mass copy and paste and then mark up the old content and submit a MP that would be awesome | 17:40 |
henninge | What was the hook you were editing? | 17:40 |
hazmat | negronjl, i'm here huckleberry | 17:40 |
henninge | The one on the left | 17:40 |
henninge | -joined | 17:42 |
niemeyer | hazmat: Yeah, but it has been repositioned to be the only content about charms | 17:42 |
henninge | Thanks, I got it ;-) | 17:43 |
niemeyer | arosales: Sure, but that approach has some serious issues. Important content just went away | 17:43 |
niemeyer | arosales: and external links are now broken | 17:43 |
jcastro | the links will be fixed on monday | 17:43 |
arosales | niemeyer, yes redirects land on monday | 17:43 |
arosales | niemeyer, we'll review our process, and the feedback is much appreciated | 17:44 |
niemeyer | arosales, jcastro: That doesn't solve the problem if the content is different! | 17:44 |
arosales | niemeyer, what I am saying now is if you notice something let us know via a bug and we'll work to make it better | 17:44 |
niemeyer | arosales: For example, have a look at http://askubuntu.com/questions/82683/what-juju-charm-hooks-are-available-and-what-does-each-one-do and try to imagine how little sense will that response make with the new content that the links will point to | 17:45 |
jcastro | niemeyer: nick is working fulltime on the issues you're talking about | 17:45 |
jcastro | and unfortunately the guy who wrote the redirects went on vacation, so the links will be busted for 2 days, tops | 17:45 |
jcastro | let's discuss this after the charm school | 17:45 |
henninge | So, I just did something like that today but I put the lib (or rather a python package) inside the hooks dir. | 17:46 |
henninge | Since the hook is excuting in that directory, it will find it. | 17:46 |
henninge | w/o messing with pythonpath | 17:46 |
niemeyer | jcastro: I don't really have much else to say.. as I responded on juju@, I appreciate the new work, but suggest rolling back the new docs until they cover at least the most relevant portions of the old documentation | 17:46 |
niemeyer | henninge: The hook executes on the charm directory | 17:47 |
hazmat | henninge, CHARM_DIR is available to hooks | 17:47 |
henninge | oh, ok | 17:47 |
niemeyer | henninge: The documentation is misleading.. I sent a note about that to the list a couple of days ago | 17:47 |
hazmat | henninge, for adding libraries.. ie $PYTHONPATH=$CHARM_DIR:$PYTHONPATH | 17:47 |
hazmat | niemeyer, yeah.. that section part was wrong | 17:47 |
henninge | hazmat: A good, I'll try that | 17:48 |
henninge | I guess that's what I remembered | 17:48 |
robbiew | yep | 17:51 |
henninge | yes | 17:51 |
hazmat | jcastro, yes | 17:51 |
Akira1 | we've expanded the hadoop charm over here to work with the cloudera cdh4 distribution, btw | 17:54 |
jcastro | wait, what! | 17:54 |
jcastro | Akira1: dude, that is awesome! | 17:54 |
Akira1 | i haven't really looked at the OG one yet | 17:54 |
Akira1 | but i've been expanding the configs and stuff like that to make it work | 17:54 |
Akira1 | somehow the charm is missing the config changed hook | 17:54 |
Akira1 | it is pretty slick and works great | 17:55 |
Akira1 | one thing we can't figure out is how to deploy a secondary name node on another instance, right now the secondary namenode stuff runs on the same box as the primary namenode which is kind of a no-no | 17:56 |
Akira1 | yeah that'd be slick wouldn't it... | 17:57 |
Akira1 | sadly im not sure we did it that way... | 17:58 |
Akira1 | big paste inc: | 17:58 |
Akira1 | configure_sources () { | 17:58 |
Akira1 | source=`config-get source` | 17:58 |
Akira1 | juju-log "Configuring hadoop using Cloudera repo..." | 17:58 |
Akira1 | wget -q \ | 17:58 |
Akira1 | https://archive.cloudera.com/cdh4/ubuntu/precise/amd64/cdh/cloudera.list \ | 17:58 |
Akira1 | -P /etc/apt/sources.list.d | 17:58 |
Akira1 | wget -O - \ | 17:58 |
Akira1 | https://archive.cloudera.com/cdh4/ubuntu/precise/amd64/cdh/archive.key \ | 17:58 |
Akira1 | | apt-key add - | 17:58 |
Akira1 | apt-get update | 17:58 |
Akira1 | } | 17:58 |
Akira1 | sweetaction | 17:59 |
Akira1 | you guys going to run out of time before you talk about the monitoring interface | 18:03 |
Akira1 | ? | 18:03 |
m_3 | Akira1: yes | 18:03 |
m_3 | :) | 18:04 |
Akira1 | bah, that is one thing i actually want to hear about as i've been struggling with it :) | 18:04 |
Akira1 | may i suggest that as a topic for another one of these screencast thingies? | 18:05 |
m_3 | yes, absolutely!! | 18:05 |
Akira1 | :) | 18:08 |
Akira1 | yeah, do that | 18:09 |
Akira1 | :-P | 18:09 |
Akira1 | i've got nagios, nrpe, etc | 18:09 |
m_3 | lp:charm-helpers | 18:09 |
henninge | Thanks guys! | 18:11 |
Akira1 | yeah, epic presentation, thanks | 18:11 |
jcastro | \o/ | 18:11 |
jcastro | m_3: can you pass me along the youtube url for the recording? | 18:11 |
m_3 | hey | 18:11 |
Akira1 | i've inherited all this juju stuff from a predecessor, he was working on sweet stuff | 18:12 |
henninge | Are these sessions available for watching afterwards? I'll be on holiday in two weeks. | 18:12 |
Akira1 | integration with saltstack for one | 18:12 |
m_3 | Akira1: cool! | 18:12 |
jcastro | henninge: yep, I keep a list here: https://juju.ubuntu.com/resources/videos/ | 18:12 |
jcastro | Akira1: that is awesome man | 18:13 |
m_3 | henninge: yes... we're working on an actual youtube channel, but they're all available from juju.ubuntu.com/resources/videos | 18:13 |
m_3 | ah, jinx | 18:13 |
Akira1 | like i said WAY earlier, i've been left with a mess heh | 18:13 |
jcastro | well hopefully we can help! | 18:13 |
Akira1 | i started touching this stuff 4 weeks ago | 18:13 |
Akira1 | and now I'm moving on so i'm trying to leave the team with as much solidity as possible | 18:14 |
jcastro | m_3: I need the youtube url for the videos page please | 18:14 |
henninge | jcastro, m_3: thanks | 18:15 |
m_3 | henninge: np... thanks for hanging! | 18:16 |
m_3 | jcastro: dude, I'm looking for it | 18:16 |
jcastro | http://www.youtube.com/watch?feature=player_embedded&v=08dOs3eO04M | 18:16 |
jcastro | aha | 18:16 |
jcastro | it's linked from ubuntuonair of course | 18:16 |
henninge | I have a basic question: Is this all done using the new juju (juju-core)? I am still using 0.7 because I had some problems. | 18:16 |
Akira1 | i still use 0.6.1 | 18:17 |
Akira1 | so, *shrug*? | 18:17 |
jcastro | henninge: the one thing we don't have in -core is the local lxc provider | 18:17 |
jcastro | for everything else, you want to use juju-core | 18:17 |
m_3 | henninge: mixed actually... depends on which series you're on | 18:17 |
jcastro | Akira1: you'll want to upgrade to .7 | 18:18 |
m_3 | on precise, I run 0.7... on raring, I run 1.10.0 | 18:18 |
m_3 | Akira1: yes, definitely upgrade from 0.6.1 to 0.7! | 18:18 |
henninge | ok, my workstation is raring, the server is precise. | 18:18 |
m_3 | henninge: haha, yeah... that's pretty common | 18:19 |
henninge | but am i currently missing something using 0.7? Will it still get updates? | 18:19 |
m_3 | henninge: it's EOL'd, so we're trying to switch everybody to 1.x | 18:20 |
jcastro | all new development is happening in 1.x | 18:20 |
Akira1 | hmmmmmmmmmmm | 18:20 |
Akira1 | i'll have to check that out | 18:20 |
henninge | what I expected ;) | 18:20 |
Akira1 | i'll go look at a changelog | 18:20 |
m_3 | henninge: really it depends on what you're doing honestly... most things work fine in 1.x | 18:20 |
m_3 | no lxc support is the biggest gap atm | 18:20 |
m_3 | note that we now have a nice mac client for 1.x series! | 18:21 |
Akira1 | is 1.x still python or is that go? | 18:21 |
m_3 | 1.x is go | 18:21 |
Akira1 | cool | 18:21 |
Akira1 | that coming along ok? | 18:21 |
m_3 | yup... it's moving nicely | 18:21 |
Akira1 | sweet | 18:21 |
henninge | m_3: will 1.x get lxc support eventually? | 18:21 |
m_3 | biggest problem atm is packaging... otherwise the juju bits are working fine | 18:21 |
Akira1 | i've been giving a few presentations internally about juju, we're a kind of small company, only about 30k employees... | 18:21 |
jcastro | henninge: like this next month we hope | 18:22 |
m_3 | henninge: yup, it's on the nearterm roadmap... with us going "you done yet?" every day! | 18:22 |
henninge | :-D | 18:22 |
m_3 | haha | 18:22 |
henninge | You guys rock! | 18:22 |
m_3 | Akira1: let us know if we can help with presentation material or content | 18:22 |
Akira1 | too late! :-P | 18:22 |
henninge | I really like juju and it's exactly what I need here. So, thanks. | 18:23 |
m_3 | we can go over the monitoring stuff outside of regular charmschools too | 18:23 |
Akira1 | I gave my last presentation to mumbai this morning | 18:23 |
m_3 | henninge: awesome! | 18:23 |
Akira1 | I had a question way way way way earlier in the day, prolly before you guys were up | 18:23 |
m_3 | Akira1: dang, well we should totally hook you up | 18:23 |
Akira1 | which of you work for canonical? | 18:23 |
m_3 | well yes :) I'm on US pacific time atm | 18:23 |
Akira1 | I'm on monster energy drink time | 18:24 |
m_3 | several in the channel... I do as well as jcastro | 18:24 |
Akira1 | since my cdh4 charm blew up on me last night but i fixed it this am ;) | 18:24 |
Akira1 | are any of you affiliated with the consulting/services arm? | 18:24 |
m_3 | Akira1: I'm wondering what you did with the jdk version... | 18:24 |
Akira1 | I submitted an inquiry out to them via their shoddy web portal, we might be looking for some paid assistance | 18:24 |
Akira1 | right now everything is working with openjdk even if it isn't advised | 18:25 |
m_3 | Akira1: we're on the products side of the house... negronjl is canonical consulting though | 18:25 |
m_3 | Akira1: we can certainly put you in touch with the right folks | 18:25 |
Akira1 | in fact, my juju deployed cluster is actually humming along smoother/faster than our older cluster that was manually deployed | 18:25 |
Akira1 | that being said, i've made a bunch of hacks to get stuff working | 18:25 |
m_3 | Akira1: great news! I've been really dreading integrating the oracle versions | 18:25 |
Akira1 | tons of things in the charm aren't necessary best practices for config | 18:26 |
Akira1 | so my config.yaml is now... bigger... | 18:26 |
m_3 | yeah, that comes along | 18:26 |
Akira1 | and i have some issues due to maas too | 18:26 |
Akira1 | im sure i'll figure some of it out eventually | 18:26 |
Akira1 | sometimes this hadoop stuff causes you to do things that arent very... shall we say, service orchestration-y | 18:26 |
m_3 | ha! was just going to ask you which provider you were running on | 18:27 |
Akira1 | yeah i inherited about 170 legacy physical servers | 18:27 |
jcastro | hey so I don't know who you'd talk to wrt. consulting | 18:27 |
m_3 | Akira1: let us know what you had to do outside-of-paradigm... we can really cover a lot with config and constraints | 18:27 |
jcastro | negronjl: whom should Akira1 contact? | 18:27 |
m_3 | I think we made negronjl late to another meeting... soudned like he had to run | 18:28 |
Akira1 | m_3: the maas template doesn't mount all the local storage, so I do use maas tags to constrain to a few known nodes | 18:28 |
Akira1 | since my hadoop nodes have additional local storage | 18:29 |
Akira1 | but I actually mount the storage after provisioning | 18:29 |
Akira1 | it all works but it isnt' 100% automagical | 18:29 |
m_3 | right... gotcha | 18:29 |
Akira1 | and the big thing that could use fixing is the secondary namenode thing | 18:30 |
m_3 | there're some specs in place for more control of local storage in the 1.x go versions | 18:30 |
Akira1 | thing is im not sure what relation would define that | 18:30 |
m_3 | dunno where it landed in priority though | 18:30 |
jcastro | Akira1: pm me your contact info and I''ll have someone contact you asap | 18:30 |
henninge | bye! have a nice weekend (eventually) | 18:32 |
m_3 | that sounds like more of a pure constraint thing... | 18:32 |
Akira1 | well | 18:32 |
Akira1 | the way the charm works, as you guys talked about | 18:32 |
m_3 | could maybe specify mounts after-the-fact via relations or config.. .but the constraint is the key to getting the service positioned for that | 18:32 |
Akira1 | oh yeah the mounts thing is definitely outside of the box | 18:33 |
Akira1 | and i just have to make the right tags | 18:33 |
Akira1 | maas-tags | 18:33 |
m_3 | yup | 18:33 |
Akira1 | i got that working ok via hacky ways | 18:33 |
Akira1 | knowing machine serial numbers =/ | 18:33 |
m_3 | oh, sorry... so secondary namenode | 18:34 |
Akira1 | yeah | 18:34 |
m_3 | _should_ be able to do with relation "roles" | 18:34 |
Akira1 | ok | 18:34 |
m_3 | but we haven't tried b/c the main hadoop charm's using an older hadoop | 18:35 |
Akira1 | ah | 18:35 |
m_3 | it's worth trying to add multiple hadoop-master units and fixing the relations to work | 18:35 |
Akira1 | it is actually in here... | 18:35 |
Akira1 | but i haven't figured out how to get it to fire | 18:36 |
m_3 | but you could also just do something like... juju deploy hadoop hadoop-master; juju deploy hadoop hadoop-secondary-masters -n2 | 18:36 |
m_3 | ah, ok | 18:36 |
m_3 | cool, well it'd be cool if you could push the cdh changes into the current charm | 18:38 |
Akira1 | it is actually in the regular charm | 18:39 |
m_3 | that'd totally rock | 18:39 |
Akira1 | https://bazaar.launchpad.net/~charmers/charms/precise/hadoop/trunk/view/head:/hooks/hadoop-common | 18:39 |
Akira1 | er | 18:39 |
Akira1 | not the cdh4 stuff, the secondary namenode stuff | 18:39 |
Akira1 | i'll try and cook up a way, ours really looks almost exactly like the regular hadoop one | 18:39 |
Akira1 | but it isn't configurable to use base hadoop | 18:39 |
m_3 | oh, sweet! | 18:39 |
m_3 | jamespage rocks! | 18:39 |
Akira1 | he's got the secondary namenode stuff in there | 18:40 |
Akira1 | i'll see about cooking up a way to use one or the other, shouldn't be too hard | 18:40 |
Akira1 | you get neat issues with that and maas too | 18:40 |
m_3 | secondary namenode right there... so I don't know if that was working or if that was just planning for the future, but cool | 18:40 |
Akira1 | yup, i see it in there, dunno wtf can trigger that relation | 18:40 |
Akira1 | i mean | 18:40 |
m_3 | yeah... um... I bet you do | 18:40 |
Akira1 | i should make the charm reconfigure the squid proxy for maas hhe | 18:41 |
m_3 | ouch | 18:42 |
m_3 | ok, I'm gonna go grab some food... later gang | 18:43 |
Akira1 | adios | 18:44 |
Akira1 | thx again for the presentation earlier | 18:44 |
m_3 | sure.. .thanks for hanging! | 18:44 |
m_3 | Akira1: jorge set you up with canonical contacts huh? | 18:45 |
jcastro | Akira1: I'm going to have a guy mail you directly | 18:45 |
Akira1 | sounds good | 18:49 |
Akira1 | i can start a dialog with them about what we need and we can see if anyone can help :) | 18:49 |
jamespage | m_3, Akira1: secondary namenode stuff did work last time I tried it | 19:02 |
arosales | negronjl, m_3, jcastro thanks for the charm school! | 19:12 |
Akira1 | hmm really james? | 19:13 |
Akira1 | what was the relation you added to do it? | 19:13 |
Akira1 | we hve a running secondary namenode process, it just runs on the namenode for some odd reason | 19:14 |
=== defunctzombie is now known as defunctzombie_zz | ||
matsubara | hi there, does anyone know what's the alternative for juju debug-hooks on juju-core? | 19:35 |
=== defunctzombie_zz is now known as defunctzombie | ||
sidnei | matsubara: there's none yet | 19:48 |
matsubara | :-( | 19:49 |
=== wedgwood is now known as wedgwood_away |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!