=== thumper-gym is now known as thumper [00:45] marcoceppi, i wrote a charm localcharm/precise/mycharm i can deploy it in local now i want to deploy in AWS, I configure environments.yaml and make defaults : amazon now what will be my next command to deploy mycharm in AWS [00:45] bootstrap === dalek49_ is now known as pk [00:46] themonk, there's also juju switch for changing environments without modifying your environments.yaml [00:46] without args it tells your current env.. -l to list all.. [00:46] themonk, and then juju deploy --repository=localcharm local:mycharm.. ie exactly the same as local provider [00:47] hatch: thanks for getting back to me on askUbuntu. I'm having issues trying to debug the hooks [00:48] the debug-hooks command doesn't seem to do what it's purported to do [00:48] hazmat, so i need to bootstrap again for amazon env then juju deploy --repository=localcharm local:mycharm [00:48] themonk, yes [00:48] pk, how so [00:49] pk sorry just running off for supper... hazmat see http://askubuntu.com/questions/429541/why-are-my-juju-relation-hooks-not-running/429565 [00:50] hazmat: every command in the tumx session is followed by "The program 'relation-get' is currently not installed. You can install it by typing: \n apt-get install juju" [00:50] s/relation-get// [00:52] pk, you have to wait till a hook actually runs [00:52] pk, it will open up in a new window in tmux [00:52] pk, window 0 in tmux of a debug-hooks session is not actually a hook context/environment [00:53] pk, for example.. after doing juju debug-hooks in one terminal.. in a separate terminal do juju set myservice someconfig=value [00:53] and you'll see a config-changed window open in the tmux session of the first terminal [00:53] there you can do config-get / relation-get etc [00:54] oooooooh, thanks. I'll give that a shot [01:10] pk, the hook needs to be an executable [01:10] pk, chmod +x and your missing a shebang line at the top .. [01:10] pk, like #!/bin/bash [01:10] pk, re looking at https://github.com/peterklipfel/firesuit/blob/master/charms/flask_charm/hooks/amqp-relation-changed [01:21] yeah, I had changed the hook to executable, but haven't pushed. I suspect the shebang might be it [01:29] back.... pk hazmat got you all going? [01:32] hatch, I understand things a bit better, but I still have nothing working [01:32] I just got done testing the adding of the shebangs [01:32] pk so when you make the relation you get the debug-hook instance up? [01:34] basically I wan to make sure that the hook is actually getting fired when you make the relation [01:37] it seems like the hook is not firing. When I add the relation, I get no new windows in the tmux session, and the log output doesn't contain what I would expect [01:40] and the side effects aren't happening (touch /var/i_was_here) === CyberJacob is now known as CyberJacob|Away [01:41] hmm this is very odd [01:43] pk when you run a juju status after the add relation does it show the relation? [01:43] pk, also when your developing you have to deploy with -u or you get cached versions [01:44] yes, here's the full output http://pastebin.com/T8YBGdQu [01:44] you sanity check your charm from the tmux window 0 if your in debug-hooks already [01:44] pk, also you can turn up the default logging level [01:45] pk, juju set-env logging-config="=DEBUG" [01:48] hazmat, is there an INFO level? [01:50] pk, yes [01:51] aha! at the end of my flask log, I'm getting permission denied [01:51] see if that takes me anywhere [02:05] hazmat, the permission denied error is coming from the destroy-relation command. I have no hook for amqp-relation-departed or amqp-relation-broken, so I don't know if the permission denied is meaningful [02:18] pk, huh.. you don't use the client cli from charms.. there's a separate hook cli [02:20] hazmat, I don't think I understand. Are you talking about the 'relation-get' commands in the charms? [02:30] pk, no.. i'm just not clear where you getting the juju destroy-relation getting permission denied errors.. and seeing that in flask log [02:32] hazmat, when I run juju destroy-relation flask rabbitmq, the flask log (~/.juju/local/log/...) displays "2014-03-05 02:04:12 ERROR juju.worker.uniter.filter filter.go:117 permission denied" [02:41] shang: ping [02:41] pk, aha [02:41] pk, when its executing a hook? [02:42] interesting no hook [02:44] hazmat, right. I wasn't sure whether it was a meaningful error because there's nothing for it to be executing [02:55] nazmat, as an extra data point, I got the same error on the rabbitmq side when I added and destroyed the relation while the services were still pending [02:56] pk: can you post more of the log from the unit agent ? [03:15] davecheney, all I get after the relation joining is the following output from the flask log [03:15] 2014-03-05 03:12:23 DEBUG juju.worker.uniter.filter filter.go:314 got unit change [03:15] 2014-03-05 03:14:58 DEBUG juju.worker.uniter.filter filter.go:322 got service change [03:15] 2014-03-05 03:14:58 DEBUG juju.worker.uniter.filter filter.go:523 no new charm event [03:15] 2014-03-05 03:14:58 DEBUG juju.worker.uniter.filter filter.go:338 got relations change [03:17] pk: is that the wholelog [03:18] it looks like parts are missing [03:20] davecheney, that is all the output after the relation-add command. The full log is here http://pastebin.com/yugfcNYc [03:21] * davecheney looks [03:23] pk i can't see the original error in that log [03:25] davecheney, right, so the error is present when I do the relation-destroy. I get [03:25] 2014-03-05 03:25:08 DEBUG juju.worker.uniter.filter filter.go:338 got relations change [03:25] 2014-03-05 03:25:13 DEBUG juju.worker.uniter.filter filter.go:322 got service change [03:25] 2014-03-05 03:25:13 DEBUG juju.worker.uniter.filter filter.go:523 no new charm event [03:25] 2014-03-05 03:25:13 DEBUG juju.worker.uniter.filter filter.go:338 got relations change [03:25] 2014-03-05 03:25:13 ERROR juju.worker.uniter.filter filter.go:117 permission denied [03:25] and that's it [03:28] pk: this sholnd't be required, but are all your hooks executable ? [03:28] ie, chmod +x ? [03:31] davecheney, this was happening when there was no hook [03:31] Where can I find a list of the trusty charms? [03:32] so my question was primarily whether or not it was helpful to my ultimate question of why my hooks aren't running [03:33] * davecheney looks at the source [03:57] pk: still looking for where the permission deines are coming from [03:57] is your charm online somewhere ? [04:02] the charms I'm using are at https://github.com/peterklipfel/firesuit/tree/master/charms [04:06] pk: thanks [04:07] you should remove those dummy relation-name-* hooks [04:08] and all the other dummy hooks [04:11] davecheney, ok. Do you think those are causing the other deploy hooks to not run? [04:11] s/deploy/relation/ [04:14] pk: it shouldn't [04:14] but we are talkinga bout bugs here [04:14] so they will just cloud the issue [04:14] remove any hooks you don't need [04:14] any any that charm create added for you [04:17] davecheney, totally. I'm waiting on the environment to be up and running, but I'm testing now [04:26] davecheney, I got rid of all the dummy scripts, and redeployed and added the relation. As far as I can tell, the hooks didn't run [04:28] pk: ok [04:28] so is there still an issue [04:28] i'm a bit confused [04:29] davecheney, ok. The issue originally was that my hooks aren't running, and I don't know why. Juju status shows that the relation was joined, but none of my hooks seem to be running. I'm not seeing any output, and when I ssh into the lxc's, I'm not seeing the files that I put there (as a dummy side effect for testing) [04:30] pk: can you get the unit log for that unit [04:30] there will be lots of likes 'HOOK: ...' [04:30] that is the output of the hook [04:30] i see in your code you call a few other commands that will be outputting [04:30] the log that I posted before is that unit log [04:31] (http://pastebin.com/yugfcNYc) [04:31] There is no debug output that I would expect from my hooks [04:32] so I put the `touch /var/i_was_here` in the hook to see if maybe there was an issue with the logging. But when I ssh in, and check /var, the file isn't there [04:39] davecheney, I'll be back shortly, I have to switch venues [04:42] kk [04:42] pk: protip [04:42] set -xe in all hook scripts [04:42] then you get output [05:04] davecheney, are you still around? [05:10] pk: just [05:19] davecheney, ok, I'll go work on it and hopefully be back with better, more relevant questions later [05:19] thanks for your patience === CyberJacob|Away is now known as CyberJacob [07:57] http://www.rapp.org/archives/2014/03/passengers/ [07:57] oops, wrong window. apologies === rogpeppe is now known as 17SAAOELG === rogpeppe1 is now known as rogpeppe === CyberJacob is now known as CyberJacob|Away === wgrant_ is now known as wgrant [11:12] hi [11:13] i wrote a charm in which i need to put some fines in the hooks/ directory [11:13] and use those files [11:13] but when deploying the charm it says it cannot find the file [11:14] HOOK cat: queries.sql: No such file or directory [11:14] in my install hook i have executed the command [11:14] cat queries.sql | mysql -u root -pPa55w0rd [11:14] but its not able to find this file [11:15] i have also tried chmod +x queries.sql [11:15] can someone help me please [11:16] prathamesh875: heya - hooks run from the charm root directory - not from the hooks directory [11:16] ooh [11:16] prathamesh875: so cat hools/queries.sql | mysql ... [11:16] so i need to put that file in my charms directory? [11:17] i had also tried that [11:17] still it says no such file [11:17] prathamesh875: it needs to be in the root or a sub dir of the charm yes [11:17] prathamesh875: the usual conventions is to have a files/ directory [11:18] prathamesh875: and then from a hook you can do "cat $CHARM_DIR/files/myfile" or whatever [11:20] ok thanks bloodearnest [11:20] will try that [11:22] prathamesh875: to debug, you can also juju ssh /0 and examine the files in /var/lib/juju/agents/unit--0/charm to si if the file is actually present on the unit [11:22] s/si/see [11:25] ohh [11:25] thanks [11:25] http://paste.openstack.org/show/72552/ [11:26] i had this in the unit [11:26] but queries.sql was not in executable [11:26] somehow permissions not got pushed [11:27] http://paste.openstack.org/show/72555/ [11:27] now i have this [11:28] cat $CHARM_DIR/files/queries.sql | mysql -u root -pPa55w0rd [11:28] and i have this in my install hook [11:28] will deploy this again === amol_ is now known as amol === viperZ28_ is now known as viperZ28 === bodie_ is now known as Guest87187 [12:12] Wonder if anyone can help me, im trying to install juju on windows but keep getting "This app cannot run your pc" i am using windows 8 x64 am i doing something wrong? :) [12:15] snazy2000: not sure, doesn't sound like you are [12:16] hmm ive tried to install it on two pc's and says same thing [12:16] snazy2000: you may want to run the install in compatibility mode for Windows 7 [12:17] tried that, wont even let it get to admin mode [12:18] snazy2000: huh, I've not heard of people having issues before :\ [12:20] snazy2000: I don't have a Windows 8 machine, only Vista, so I can't really even help replicate [12:25] snazy2000: if it's still giving you grief, we have a Vagrant box you can use to try Juju out [12:26] Last thing im going to do it make sure file has downloaded correctly [12:35] are you running 32bit or 64? === JoshStrobl is now known as JoshStrobl|AFK [12:46] snazy2000: I've got a 64bit machine iirc [12:46] snazy2000: It's an older laptop that I keep around in case I need to test things on Windows [12:47] ok thanks. [12:52] snazy2000: I can certainly try, but it's Vista === smoser` is now known as smoser [14:52] When I run juju bootstrap -e local I see: ERROR cannot use 37017 as state port, already in use. [14:53] mbruzek: check for upstart jobs [14:53] mbruzek: sudo initctl list | grep juju [14:54] I see 2 upstart jobs running. I know how to stop them with the kill command, is there a "better" way to stop them? [14:56] mbruzek: sudo stop [14:57] mbruzek: then you'll want to rm -f /etc/init/juju-* [14:57] then try bootstrapping again [14:57] Thanks marcoceppi, I ran sudo initctl stop juju-db-mbruzek-local [14:57] That seemed to do the trick. I will re-run my clean script (which deletes that stufF) === Guest87187 is now known as bodie_ [15:39] jcastro: so juju talk went well last night, thanks for you help [15:39] nice, did you try the hadoop demo? [15:46] jcastro: I tried it locally but didn't get it working, so I decided not to in the end [15:46] but people seemed impressed with jujucharms.com despite it's fakery [15:48] jcastro: interesting contact - the new UK+RoI AWS evangelist was there (he's local), he was nodding sagely throughout the talk, and we had a good chat after about CloudFormation, and our plans for SaaS support via virtual/proxy charms [15:49] he twitched a bit at the "Democratization of the Cloud" slide :D [16:01] jcastro, I know "set -e" is recommended for Bash scripts, but I can't find it on the authors-charm-best-practice.html page. [16:01] I don't think we have that written down [16:01] add it perhaps? [16:01] jcastro, Is that recommendation located elsewhere? A google search only turns it up on our getting started. [16:01] I think it's only in marco's brain [16:01] jcastro, Yeah I will add that to my list. [16:02] Thanks jcastro [16:03] mbruzek: unless the author is checking $? of every command, it's pretty much a requirement for bash charms [16:03] marcoceppi, I know [16:03] just not really documented anywhere [16:03] marcoceppi, I was just checking for documentation to point him to [16:03] I found the document that I will update after the review. [16:03] cool [16:14] mbruzek, marcoceppi, jcastro, set -e is used and explained in authors-charm-writing.html page (the walkthrough one) [16:15] evilnickveitch, I saw that, but I think it should be a recommendation in best practices. [16:16] there is a Bash section and everything. [16:16] That mention is hidden, I had to do a google search to find it. [16:16] searching our docs for "set -e" did not turn the page up. [16:17] on the juju.ubuntu.com search, I mean. [16:17] mbruzek, cool, I think that page is more a style/process guide, but if you think it should be there, be my guest and add it [16:18] hey marcoceppi, I need to do this community weekly videocast thing; what's the TLDR on amulet and testing in general? [16:18] there is an open bug about search not working [16:18] I want to learn the new documentation process. [16:18] And I do think it should be in our best practices document as a great idea [16:18] jcastro: automated testing is in it's finaly stages, amulet had a big release which introduced post-deployment commands and bug fixes [16:20] mbruzek, I wrote down the new doc process in this readme: https://github.com/juju/docs [16:20] jcastro, after this review is done I am all over that [17:05] jcastro, When following the juju-docs process I encountered a problem [17:05] oh? [17:06] I searched github for juju-docs [17:06] it's juju/docs [17:06] I found juju/docs and when I forked it the command to clone juju-docs did not work [17:06] I will fix this in the README then. [17:07] Oh I see what you mean [17:08] actually, when you fork it you can call it whatever you want [17:08] but you're right, it makes sense to match it [17:11] * mbruzek is fixing that now [17:15] jcastro, the clone uri was not correct [17:16] huh [17:16] how did I mess that up? [17:16] git clone git@github.com:{yourusername}/juju-docs.git [17:16] It looks correct in other places, except that one [17:16] I will fix it in my commit [17:16] hey also [17:17] one thing I try to avoid is the inline editing in github [17:17] that way it forces me to learn the workflow [17:17] inline editing? Like from github.com's web site? [17:17] I still have to reference the README everytime I want to fix something, but I am getting better at it [17:18] yeah, you can just click on a file and do "edit" and fix it, then hit a button and it will PR it. [17:18] OK yeah I want to learn git too so I am following your READMe [17:18] right, better in the longrun === JoshStrobl|AFK is now known as JoshStrobl === marlinc_ is now known as Marlinc [20:38] hazmat, around? I'm looking at a race in deployer which is killing our automated testing [20:39] and I know time.sleep(20) is probably not the fix :-) === Guest18028 is now known as bodie_ [21:00] marcoceppi, dylan's use case to try an older revision worked [21:00] http://askubuntu.com/questions/430029/how-do-i-deploy-a-specific-revision-of-a-charm/ [21:00] so I wrote it down [21:01] hey jcastro [21:01] yay [21:01] hey jamespage [21:03] jcastro: sweet answer [21:05] personally I think it's silly to have 2 totally separate revisions of something, but whatever. :p [21:07] meh, every code thing has it. apache 2.4.7 isn't where you go get the changelog for that. It's in the vcs revisions before it [21:25] jamespage, where you waiting to ask me something? or just saying hello? [21:25] just saying hello [22:01] jamespage, around now [22:01] hazmat, right [22:02] hazmat, we're using deployer in the OpenStack lab to bootstrap and deploy test environments [22:02] hazmat, seeing two things [22:02] 1) dns-name not appearing straight away in status [22:03] 2) the websocket connection to the environment after bootstrap failing [22:03] guessing its not quite ready [22:05] jamespage, will you be around for 30m [22:05] hazmat, most likely in bed by then [22:05] but it can wait [22:06] I'm working around it with hacking loops and sleeps right now [22:07] jamespage, are you bootstrapping with deployer? or separately? [22:07] hazmat, with deployer [22:08] jamespage, ok meeting done.. [22:09] * hazmat src dives [22:11] jamespage, that's strange since the entire cli should be running through the api now... and we use the cli [22:11] jamespage, do you have a traceback on the timeout? [22:11] hazmat, yeah one sec [22:12] er.. use the cli to determine the api endpoint [22:12] hazmat, http://paste.ubuntu.com/7040964/ [22:12] jamespage, i can add some sort of loop around connecting to the env.. is sleep 20 working reliably for you? [22:12] the codebase uses deployer internals directly - we might be doing something wrong [22:13] hazmat, yeah - 20 secs seems to work OK [22:13] jamespage, except in the traceback you sent ;-) [22:15] jamespage, re use directly.. yeah.. that might account for it. depends on how your constructing env [22:16] hazmat, that was pre-hack [22:16] jamespage, if the api endpoint url isn't in the env.. deployer calls out to the cli, which blocks [22:16] till the api endpoint is ready.. and then deployer uses the api endpoint it calculates from that. [22:16] er.. from that cli output [22:17] jamespage, there's a sleep(20) in the traceback [22:17] hazmat, lp:openstack-ubuntu-testing [22:22] jamespage, yeah.. that usage should be fine [22:23] jamespage, easy fix though.. there's already a loop around connect that's catching timeout errors, i'll add in conn refused [22:24] great [22:52] is rackspace openstack still unsupported? === bradm1 is now known as bradm [23:16] maxcan, yes.. unless you use it with the manual/null provider (ie you do provisioning and point juju to the machine via ssh)