#chef

countdigi is there a way to set a default mode for a directory provider? i could wrap it as a def but curious if there is another way 01:50
hoover_damm countdigi, man umask 01:52
countdigi heh 01:52
hoover_damm countdigi, the default mode i believe should be specified by umask 01:52
and it's possible to monkeypatch the provider easily with libraries (but not endorsed) 01:53
it's ruby fairly easy to monkey :) 01:53
countdigi yeah umask is what it says in docs 01:53
right its not a big deal but i was used to in puppet (forgive me) being able to easily set defaults at top of manifest 01:53
its cool not a big deal 01:53
hoover_damm nothing wrong with monkey patching that in really 01:54
but I understand where your coming from 01:54
countdigi yeah i might do it in like a site cookbook and include that in the other recipes - with a little convenience syntax helpers :-) 01:54
but i know that stuff can get out of hand if u aren't careful 01:54
looking in 3 files to find the damn mode of a dir :-) 01:55
hoover_damm I would say it's part of the differences of puppet and chef 01:55
as there are many 01:55
can't be that hard 01:57
;) 01:58
countdigi, and drop your patch in a library and have it depended upon so it's loaded 01:59
;) 01:59
Majost_ Is it possible to do the following in an erb: <% require 'uuid'; uuid = UUID.generate; puts uuid -%> 02:13
I tried this; but it didn't seem to do what I was hoping 02:13
Is there a different syntax I should use? 02:14
countdigi u need <%= UUID.generate %> i believe 02:15
that runs the code and replace return value inline 02:15
Majost_ ah 02:22
Let me give that a shot 02:22
countdigi real quick syntax overview is here: http://en.wikipedia.org/wiki/ERuby 02:24
Majost_ countdigi, thanks for the eruby link -- I didn't realize what .erb was until now honestly. heh 02:33
countdigi, your suggestion worked too. =) 02:33
countdigi cool 02:33
hoover_damm Majost_, wrong pattern 04:22
Majost_, require uuid in your recipe and then use variables( :uuid => UUID.generate ) 04:22
wrap the require in a ruby block of course 04:22
avoid using logic in templates completely 04:23
easier to manipulate in a recipe then in a template 04:23
stormerider is there a good tutorial on JSON for chef? never really dealt with JSON much before 05:57
BryanWB_ ohai chefs! 06:01
jondot1 hi guys. im trying to figure out a workflow with librarian and a hosted chef for managing cookbooks 06:41
with chef-solo its clear to me, but with hosted chef there seem to be many redundant steps 06:42
i.e. add cookbook to cheffile, install, commit the resulting update to source control, upload cookbooks to hosted chef 06:42
and then i may want to have many different cheffiles 06:42
jerem` when i'm using knife command, I get "ERROR: Server returned error for http://blabla..&quot; but the command is executed right 08:26
Guest4420 hi, is there a way to set a ruby variable (in a recipe) to the output of a linux command? like in linux: lines=$(cat file.txt | wc -l) 09:35
Mithrandir foo = `command`? 09:37
Guest4420 thank you, works great 09:47
BryanWB_ damnit, my chef-server doesn't recognize any client keys anymore 09:53
every client including my own admin client gets 401 unauthorized 09:54
gale BryanWB_: that happened to me a few days ago - turned out to be a couchdb issue underneath 10:01
BryanWB_ gale: yeah, it turns couchdb had died 10:02
now everything ok 10:02
gale had to recover the couchdb 10:02
yup 10:02
cpasmoi Hi ! 10:07
cnf pompom 10:22
anyone here using chef to configure jenkins? 10:24
gale there's a cookbook on the community site, haven't used it though but intend to 10:25
masterkorp just ask 10:25
cnf gale: i am using the cookbook, it just installs jenkins 10:25
gale cnf: well that's a start :) 10:25
cnf a start, yes 10:26
doesn't seem like you can do any actual configuration with it, though 10:27
gale you can manage jobs and nodes and what-have-you 10:29
beyond that your jenkins setup is likely to be quite custom though I would have thought? 10:29
cnf i don't know, i have never used jenkins before 10:29
gale there's your problem 10:30
cnf it seems unaware of my java and maven installs 10:30
and wants me to set that up through the web interface 10:30
now isn't that just the bit i want to automate :P 10:30
gale haha 10:30
well set them up through the web interface and see if it writes any config would be my next step 10:31
cnf hmz 10:35
Prateek hello everyone 11:17
m new to chef 11:18
SuburbaSs Hi guys... Is possible put an IF after "notifies :restart, "service[myservice]"" I would execute the notifie only if the IF statement return true 11:24
Prateek anyone tell me about how to create custom cookbbos 11:27
cookbooks 11:27
railsraider i can't get chef-server to start i try bootstrapping again and i get Error: unable to connect to node 'rabbit@chef-server': nodedown 11:47
rabbitmq is running rabbitmq 1814 0.0 0.0 10712 420 ? S 10:29 0:00 /usr/lib/erlang/erts-5.7.4/bin/epmd -daemon 11:48
ani idea? 11:48
mayank1 Hi, any ideas how can I make use of not_if and only_if together within a single resource in a chef recipe ? 11:52
cpasmoi I'm on a new server, I've installed like advised by my compagny rubygems-1.3.7 , installed chef-client 0.9.16 (chef server 0.9.12), created /etc/chef/client.rb, filled client.pem with the correct key but still, when running 'chef-client' I got the "Can not find config file: /etc/chef/client.rb, using defaults." I'm running on CentOS 11:53
I've seen the FAQ about it but chef-client -c /etc/chef/client.rb won't deal with this issue 11:53
mayank1 cpasmoi: so chef-client -c /etc/chef/client.rb doesn't work either ? 11:54
railsraider cpasmoi: are you running chef-client as root? 11:54
cpasmoi yep, and yes 11:54
in fact 11:55
i do have at the end "Report handlers complete" 11:55
but the warning remains 11:55
railsraider that's normal, for a new machine that can't find client.pem and register 11:55
it's just telling you i can't find client.pem (On the remote node) 11:56
cpasmoi client.pem exists, I created the node on chef server before 11:56
railsraider client.pem exist on your end 11:56
cpasmoi on the server ? 11:56
railsraider you are trying to run chef-client on what? 11:56
on the server? 11:56
cpasmoi on the client 11:57
a new machine 11:57
railsraider then again 11:57
mayank1 railsraider: cpasmoi has mentioned error as "Can not find config file: /etc/chef/client.rb" ... this seems to be different from warning about missing client.pem ... no ? 11:57
railsraider your new machine is virgin, it does not have a client pem, being you are using knife to bootstrap the new node, there isn't any client.pem 11:58
so it's registering the new node on the server 11:58
passing on client.pem from the server to the new node 11:58
i think it says unable to find /etc/chef/client.pem 11:58
mayank1 ok 11:59
cpasmoi oh, in fact in my company internal doc, it says to add the node on chef server web interface, to copy/paste the private key on client.pem 11:59
I also have a client.rb given by the internal doc, which I pasted, so the two files exist 12:00
railsraider cpasmoi: don't really have to, if you bootstrap a new node it will register automatically with the client.pem from the server 12:01
it's fine that's not a problem 12:01
workmad3 railsraider: unless the chef server has been configured to not allow automatic registration 12:01
railsraider workmad3: true 12:01
by default it is 12:01
workmad3 yes, but a company that has internal documentation about setting up nodes is more likely than most to have messed around with defaults :) 12:02
railsraider agreed, however chef-run completed so auto reg is enabled 12:03
workmad3 he said he had created the client.pem file 12:04
it's just the warning 'can not find config file, using defaults' that seems to be an oddity... 12:04
railsraider can't get chef-server-webui to start 12:12
STDERR: Error: unable to connect to node 'rabbit@chef-server': nodedown 12:15
mayank1 guys any ideas how can I use not_if and only_if together ? 12:19
cpasmoi workmad3, railsraider thank you, indeed I had to put a role on my new node, to see changes. As far as I understand, the warn message isn't a big issue and disapear at the second chef-client. 12:23
geekbri Has anybody gotten any strange ArgumentError: invalid byte sequence in UTF-8 errors from chef-client before? 12:37
railsraider geekbri: i believe i have seen that before, but not on chef, it is related to ruby 1.9 12:38
geekbri railsraider: What I find completely strange about it, is that its being thrown in a recipe that hasn't changed in 2 months (but is actively used) 12:39
railsraider what recipe? 12:40
geekbri custom one. 12:40
:) 12:40
I'm poking around I suspect maybe something in a cache is corrupted, deleting things. thanks for the tip 12:40
ill poke around more on google if this doesnt work 12:40
mayank1 I am trying to use both only_if and not_if within a single resource ... Hope it works ! isn't there logical operators like 'and', 'or' available ? 12:41
railsraider mayank1: why not use a simple ruby_block resource for that 12:43
geekbri: what r u trying to install? 12:43
geekbri railsraider: not trying to install anything, we have a flume cookbook for a bunch of our flume related stuff. Looks like chef-client suddenly started failing. nothing in the system changed 12:43
mayank1 since cron was already a resource meant for setting up cron jobs so I found tht to be more useful ... 12:43
railsraider mayank1: still can use back ticks with ruby for any bash related stuff 12:44
mayank1 aah .. ok 12:44
railsraider geekbri: chef-client -v ? 12:45
mayank1 lemme see if tht works 12:45
railsraider pastie 12:45
geekbri railsraider: i'm on 0.10.8 12:45
jondot1 im having trouble with AWS knife ec2 and keypairs. 12:46
looks like even though i created a keypair in AWS, i'm getting "The key pair '..' does not exist" from my knife ec2 12:46
anyone, any idea? 12:46
railsraider geekbri: try upgrading chef-client to latest 12:46
jondot1: you need the name for the key pair not the path , set it in your knife.rb knife[:aws_ssh_key_id] = 'key pair name' 12:48
geekbri railsraider: I can't yet, We have a cookbook that higher than chef 0.10.8 breaks, I haven' thad time yet to dig into that bug though 12:48
jondot1 railsraider, thanks thats what i did though. 12:48
does it have anything to do with the AWS zone? 12:48
railsraider nopw 12:48
geekbri railsraider: however, there are other servers using this same cookbook that is throwing the invalid byte sequence error and they are not experiencing the same issue. 12:48
railsraider geekbri: sounds like you're using ruby 1.9 with a string that is not compatibly encoded with 1.9 maybe this will help, http://blog.grayproductions.net/articles/ruby_19s_string 12:51
geekbri: what i mean is maybe you are using gems that are not the right version for the ruby you are running 12:52
you can test the machine with rvm 12:53
jondot1 railsraider, is it possible that because none of these keypairs were used to make new instances yet, it cant find them? 12:53
railsraider jondot1: chef-client is looking at your knife.rb and checking for that name of key pair on your ec2, did you create the key pair on a VPC rather on the EC2? 12:54
jondot1 railsraider, i think it has to do with the region 12:55
railsraider if theres a separate key per region 12:55
geekbri: try adding this to the recipe Setting Encoding.default_external = Encoding::UTF_8 if RUBY_VERSION > "1.9" on the recipe 12:56
jondot1 looks like it. 12:56
railsraider jondot1: sorry didn't have so much experience with regions and the way keys are 12:56
geekbri: this looks like a template issue - http://tickets.opscode.com/browse/CHEF-3304?page=com.atlassian.jira.plugin.system.issuetabpanels%3Aall-tabpanel 12:57
jondot1 railsraider, its ok im still trying to figure out.. will update when resolved 12:57
railsraider p.s the problem i had with the rabbitmq not running was resolved by removing rabbitmq and re-apply bootstrap to the chef-server 13:00
for the error of: STDERR: Error: unable to connect to node 'rabbit@chef-server': nodedown 13:00
geekbri railsraider: Thanks for that link. What is strange about it though is that the template hasn't changed for 2 months (according to our git history) and it is running fine on other servers. Very strange. 13:01
railsraider geekbri: only thing i can think of is the bootstrap, sometimes the bootstrap runs a different template then you're usually use (from diff computers) this will install a diff version of ruby which will break your recipe 13:02
anyone knows what the hell is going on: ERROR: Connection refused connecting to chef-server:4000 for /clients, retry 1/5 13:03
geekbri railsraider: nah, they were all run with the same bootstrap, We are using ruby 1.9.2 on all our machines as the system ruby. 13:03
railsraider then one of the gems is broken and does not run on 1.9 13:03
geekbri railsraider: but I have the exact same cookbook running on an identical server, and it doesn't throw the encoding error. I tried deleting chefs cache and running again, same error. It's very strange. 13:04
jondot1 railsraider, yep. looks like AWS separates regions with their keypairs. 13:05
railsraider jondot1: thanks good to know 13:05
jondot1 railsraider, and since knife defaults to us-east and zones us-east-1b, things seem to be confusing 13:06
railsraider i can' t get any client to talk to chef server! 13:06
jondot1 now what i'm getting is "Failed to authenticate root - trying password auth" although i have 'ssh_user' set to 'ubuntu' anyone? 13:07
railsraider could that be the problem? tcp 0 0 0.0.0.0:4000 0.0.0.0:* LISTEN 15542/merb : chef-s 13:17
i can't connect from any client to chef-server 13:18
i figured out the problem 13:21
firewall 13:21
geekbri ssd7: Any idea when you'll start merging changes into knife-ec2 :) 14:30
ssd7 geekbri: :) Basically I only have 1 more to do and then a release will be coming unless something important makes it in quickly. Any in particular you have a concern about? 14:35
geekbri: Apologies for the delay, I had schedule to do this last week but various other projects took priority unfortunately 14:36
geekbri ssd7: Oh, no apologies necessary, I do not envy what you had to do :) 14:36
ssd7: it was purely self serving, because I am so lazy and want to be able to knife instead of bundle exec knife ;) 14:37
ssd7 :) Indeed, I know that feeling. 14:43
geekbri ssd7: rebasing now ;) 14:49
ssd7 geekbri: Great, thanks! 14:51
jondot1 trying to use https://github.com/edelight/chef-mongodb results in a 'no version specified' when going for the 10gen repo recipe.. anyone have an idea why? 15:24
ashb jondot1: have you customized the code at all? and what is the full output log? 15:24
jondot1 ashb not at all. let me gist the output 1 sec 15:24
ashb, https://gist.github.com/3164228 15:26
ashb hrm 15:28
jondot1: oh it seems that the notifies :run, "execute[apt-get update]", :immediately isn't working right 15:31
so it's added the repo but not done an apt-get update 15:31
jondot1 yea i was looking at that, assumed its working right though =\ 15:31
ashb i think the problem is that there is already an `apt-get update` resource that has run once that run 15:32
so it doesn't happen again 15:32
jondot1 ashb, any good way to work around that? 15:33
ashb try changing the name on line 28 and 39 of https://github.com/edelight/chef-mongodb/blob/master/recipes/10gen_repo.rb 15:34
and add a command 'apt-get update' inside the execute block 15:34
jondot1 do i even need the initial apt-get update? 15:35
ashb jondot1: on line 28? 15:44
jondot1 yes 15:45
ashb yeah - thats what is notified 15:46
you will also need to remove the /etc/apt/sources.d file to get it to run/fail in the same way 15:46
jondot1 i see.. let me try that 15:47
afterglow does anyone use chef-jenkins to sync their server and repo? 15:47
jondot1 ashb, not sure i'm getting why execute apt-get update is even needed 15:49
ashb you just added a new custom repo to apt - it hasn't yet queried it to see what packages are available in it 15:49
apt-get install just looks at what is in the locally cached list of packages 15:50
jondot1 ctrl-w'd the wrong window :) 15:50
erratic lawl 15:50
jondot1 well i thought apt_repository would refresh the cache for me 15:50
chiel that's why you run irssi in a screen :p 15:50
erratic I have a tablet dedicated to irc 15:51
and hten I have my other tablet that I never use 15:51
and my 900$ phone, it makes good phone calls 15:51
jondot1 ashb, tried the renaming trick, doesn't seem to work properly 15:54
ashb gist the new log and the new recipe 15:56
jondot1 looks like the same log 15:58
however, i can now be almost certain apt_repository will call apt-get update on itself 15:58
looks like running apt-get update manually makes everything work.. crap =\ 16:00
thank god for vagrant. 16:01
erratic "Be nice. :-) Use Pastie for good, not evil. " 16:01
how the hell would I use something for pasting for evil 16:01
bunch of crack smokers 16:02
http://www.securixlive.com/barnyard2/ guess what the name of this project reminds me of 16:06
__sjh lo all, just started to look at chef .. i'm managing multiple projects for multiple clients at an agency and trying to figure out if chef would be a good fit, are there any guides/docs that deal with chef usage in an agency environment, thinking about isolation between groups of servers & projects 16:12
geekbri __sjh: as in you will be responsible for hosting their chef-server for them? 16:13
erratic geekbri better question would be are you willing to pay for cloud chef or will you be running your own chef-server 16:13
__sjh geekbri, that's what i'm thinking atm, i'd prefer to have one server, but it would seem like a chef server per project (could be 5-20 servers per project) would seem to be the way to do it 16:14
erratic I think there's chef solo too but I'd never suggest that to anybody 16:14
geekbri erratic: Yes well that was my lead in. I would provide the client the option. 16:14
__sjh using chef solo seems to have the worst of both worlds ;) 16:14
geekbri __sjh: if you are hosting it yourself FOR them, I would give them each their own chef server. 16:14
__sjh which would leave me having a chef server for the configuring the chef servers... 16:15
geekbri __sjh: alteratinvely, you could pay opscode to host it, and give them each their own account 16:15
__sjh: where they pay the billing, or maybe you do 16:15
__sjh geekbri, that is tempting, just exploring the problem space atm 16:16
geekbri __sjh: it certainly will make your life easier, then you won't have to worry about backing up / hosting a ton of chef-servers 16:16
But then your client might ask... hey if we are paying somebody else to host this... what are we paying these darn consultants for :-P 16:17
jondot1 ashb, this related? http://tickets.opscode.com/browse/COOK-1372?page=com.atlassian.jira.plugin.system.issuetabpanels%3Achangehistory-tabpanel 16:18
ashb jondot1: yes 16:18
jondot1 dang! 16:18
ashb or it looks like it mightb e 16:18
jondot1: though only superficially 16:19
the notifes command in that cookbook is meant to deal with it 16:19
but its obviously not 16:19
geekbri ssd7: just FYI, I rebased my pull request against master. Should merge easy peasy now 16:19
jondot1 ashb looks like my installed version is also chef 10.10 16:20
jondot1 ashb, looks like, if i move the execute apt-get update downwards, and do 'action :run' everything works 16:31
ashb it would yes - it would also needlessly run apt-get update every time chef runs 16:32
its up to you if that is a problem or not 16:32
jondot1 so what was the purpose of the 'nothing' action? 16:32
ashb did you ever post the version of the receipe you tried with changing the names? 16:32
_rc so you can desclare it and then send it notifies 16:32
ashb nothing says don't run anything unless you get signalled to 16:32
jondot1 ashb, no, it had the same log output really 16:33
ashb its possible it is related to the issue you linked then 16:33
jondot1 i see, so those guys wanted to disable the auto update run that was supposed to run as part of the apt_repository LWRP 16:33
but then explicitly signaling it didn't work either 16:34
well looks like someone just updated this http://tickets.opscode.com/browse/COOK-1372 a minute ago 16:38
pointing to another bug.. i hope to update to 10.12 16:38
erratic I came into work to find one of the servers I was benchmarking with sql-bench in the mysql distribution with a cpu load of 2000.0 16:53
was like really 16:54
doomviking heh 16:54
miah imagine that.. 16:58
erratic miah it ran for two days 17:08
probably just like that 17:08
never checked it over the weekend 17:08
timsmith Has there been any talk of another Chef Summit up in Seattle? 17:14
Zenigor When I spinup and deploy a new server, and then doing a cap deploy:setup it fails on "Host key verification failed." when trying to clone my repo to the server. If I ssh into the server and do "ssh github.com" then accept the host, cap deploy:setup works just fine. My question is, is there a good way to add github.com as a known host using chef? this manual step of intervention seems unnecessary 17:20
erratic can you just add the public key to the known hosts file on every machine you want to ssh from? 17:22
armarco question: does anyone have a way of getting replication going using couchDB between different instances? 17:23
spox Zenigor: yeah, you can have chef manage the known hosts file, or you can setup agent forwarding in your cap settings 17:23
armarco ok, how does that work? 17:24
I don't actually want to pay for a service, but it seems like there's a command for couchDB that does replication, I'm just having trouble with getting the curl command right 17:26
if anyone has tried that 17:26
jackl0phty zenigor: sshkey gem can also help. have a look at gitlab::default on community site for an example. 17:31
erratic one more hour 17:51
then its lunch time 17:51
http://chzlolcats.files.wordpress.com/2012/07/funny-cat-pictures-i-must-be-in-soviet-rushya.jpg 17:51
newb hi chefs 17:51
erratic ohai2u 17:52
newb :-) 17:53
what info does chef tracks via AMQP? 17:53
mattray the queue of stuff to be indexed by solr 17:53
not very fancy 17:53
yfeldblum newb, every time you save a node, role, environment, client, cookbook-metadata, or data-bag item to the chef-server, the chef-server will save that item to couchdb and also put the item onto the queue to be indexed by chef-solr and chef-solr-expander 17:56
erratic thats whats I thought too 17:56
anyone ever started rabbit mq from command line 17:57
yfeldblum newb, that way, after a few seconds normally, you can ask the chef-server to run a search (as opposed to a direct fetch by id) for any of those items, and chef-server can just go to chef-solr to do a search against the search index 17:57
erratic it barfs out this disgustingly ugly looking ascii rabbit 17:57
newb yfeldblum, in other words, the queue could have been in redis as well? 17:58
mattray erratic: gist? 17:58
erratic mattray lol one sec 17:58
mattray newb: yes. At some point, it may go away completely 17:58
erratic this is such a waste of time but for this I'll make an exception 17:58
mattray erratic: emacs user? gist-mode is the bomb 17:58
erratic oo 17:58
yeah I donno everytime someone mentions emacs to me I like it more 17:59
I'm a vi users 17:59
user* 17:59
mattray I run my shell inside of emacs, so I can highlight anything, gist it and the url gets sent to my copy/paste 17:59
erratic https://gist.github.com/3165032 18:00
newb @mattray, I am working with spiceweasel atm, and noticed that there seems to be no way to pass some knife options, for example "-o" : do you have something planned? 18:00
essentially I want some more stuff in options['knife_options'] 18:00
mattray erratic: I was hoping for something more Donnie-Darkoesque 18:00
newb so, that this: 18:00
@create += "knife cookbook#{options['knife_options']} upload #{cb}\n" 18:00
reads like this: 18:00
erratic mattray my profile pic on g+ and fb was set to that creepy bunny mask for the longest time 18:01
newb options['knife_options'] -> all passthrough options to knife 18:01
erratic I want a mask like that 18:01
I love masks 18:01
I guess what I dont understand is whats the difference between expander and solr 18:02
yfeldblum newb, depends on what redis supports 18:02
mattray newb: what options do you want knife cookbook to use? 18:02
newb: I guess I could add support for more stuff 18:02
yfeldblum erratic, chef-expander is what reads the newly-saved object from couchdb (based on the id sent in the queue message) and "expands" it into a flat document that lucene can then be indexed, and it is that flat document that is indeed then indexed 18:03
newb @mattray, how much work would it take for options['knife_options'] to passthrough options to knife? 18:03
yfeldblum erratic, chef-expander takes {"key1": {"key2": "value2"}} and turns it into {"key1_key2": "value2"} and passes that to the index 18:03
newb you can assert that the yaml/json have validly formatted passthrough option strings 18:04
yfeldblum erratic, so that you can then run a search, in knife or in your recipe, for "key1_key2:value2" and find that object 18:04
erratic gotcha 18:04
newb yfeldblum, erratic: my idea was that redis displaces both rabbit and couch 18:05
but perhaps couch + ruby already displaces 18:05
yfeldblum newb, why? 18:05
erratic interesting 18:05
mattray newb: that's what knife-config is for, you want to pass them from the CLI instead of a file? 18:05
newb: couch is gone in Chef 11 18:05
newb @mattray, exactly - right now I am having to write a knife.rb and pass it through knife-config 18:06
mattray as are the ruby API endpoints 18:06
newb it would be "easier to pass it via cl? what do you think?" 18:06
mattray I'd probably take the patch if it's documented. knife-config was a patch I accepted as well 18:06
geekbri mattray: I'm guessing that a chef-client upgrade will be necessary to interact with the new backend? 18:06
mattray no 18:06
geekbri: no 18:07
geekbri oh well, excellent. 18:07
mattray chef-client stays the same, talks to APIs 18:07
newb @mattray: rabbit - don't you guys think it is a bit more high maintenance than say redis for what we use it for? 18:07
mattray APIs stay the same, just different back-end 18:07
geekbri Yeah i've noticed it has a lot of couchdb references in it. 18:07
So that was my concern 18:07
mattray newb: redis is used in Hosted/Private Chef for caching 18:07
I don't know if that's a long-term thing 18:07
newb you guys have all the good stuff in the Hosted/Private Chef x-( 18:08
mattray newb: it's open sourcing in Chef 11 18:08
newb aha 18:08
when does it come out? 18:08
mattray at least, the erlang/sql replacement of ruby/couch 18:08
"soon" 18:08
newb ha ha 18:08
mattray I'm not in the loop, being a remote non-dev and all 18:09
but I'd say this year 18:09
newb @mattray: what do you think about the options['knife_options'] passthrough? or would the best way out be to specifiy a custom knife.rb? 18:09
mattray well the custom knife.rb already works 18:09
newb that 's some good tool right there you wrote 18:09
hh 18:09
hhm 18:09
ah ok 18:10
mattray so passing additional stuff from the cli seems extraneous 18:10
at some point spiceweasel needs to become a knife plugin 18:10
newb I agree 18:10
mattray and then you'd get it all for free 18:10
geekbri I basically wrote my own knife plugin that does a lot of the same stuff as spiceweasel for us here 18:10
newb would you have time to code it? 18:10
geekbri: gimme a URI 18:11
mattray newb: not for awhile 18:11
newb * looking at geekbri 18:11
geekbri newb: ok, but its probably going to need major modification to suit your needs. 18:11
newb lemme check it out 18:11
geekbri newb: I haven't had enough time to enhance and build the rest of it, like I had planned 18:11
newb <happens to all of us> :-D 18:11
geekbri newb: https://github.com/brianbianco/knife-nodefu 18:11
it is very rough right now, although we currently use it to launch a lot of instances. 18:12
Dev time will be dedicated to it eventually. We just have other more pressing issues. 18:12
It's a little too tailored to a lot of our specific conventions at the moment. Like I said, if you want it to work for you, you will have to fork and modify heavily. 18:13
It needs some refactoring and reorg, and things pulled out and made nicer, so I make no promises about the current code quality :) 18:14
newb I saw a git plugin, but I forget where, but it allowed me to fetch cookbooks from a git repo and then upload them to the chef server etc 18:28
*knife plugin 18:28
not this one: https://github.com/websterclay/knife-github-cookbooks 18:28
miah librarian 18:28
berkshelf 18:28
look that way. knife-github == pain 18:28
geekbri both of those are good 18:32
newb miah, thanks, heard a lot about librarian 18:33
knuckolls i realize the answer is probably zookeeper or it's brethren, but is anyone out there dealing with needing to wait for things to happen on a group of nodes before doing something else in the architecture. i need like a distributed lock or something adequate. 18:44
i'm assuming this should happen within the context of the chef-run, but i'll rip it out if necessary. just adds more complexity. 18:44
grantr knuckolls, that's a common need but there's no turnkey solution. one attempt was https://github.com/lusis/noah 18:45
knuckolls i haven't looked at noah since lusis was building it and talking about it often 18:45
it's a SPOF right? 18:45
just a registry a-la zk? 18:45
grantr yes i dont think its actively developed 18:45
zookeeper would be a more stable choice 18:46
i suspect you could take noah's lock and barrier resources and rewrite them for zk 18:46
then share them because i need them too :) 18:46
knuckolls i may. i'd be surprised if someone out there hasn't encountered this yet and build that for zk 18:46
haha i see ;) 18:46
atomic-penguin knife-github worked for me, its less of a pain than git submodules 18:46
knuckolls grantr: so i guess what i'm hearing from you is, doesn't exist: build it. 18:47
which you know, i kind of knew to be true 18:47
grantr knuckolls, yes :) 18:47
yfeldblum knuckolls, is there some way you could write the recipe such that you can be tolerant of edge cases? 18:47
knuckolls yfeldblum: i'd likely have to do some heavier design work before i could answer that question ;) 18:48
but maybe, i kind of need it. so i'll have time to think about tit. 18:48
it. 18:48
yfeldblum knuckolls, even just failing the chef-client run early? 18:48
knuckolls earlier i was convinced that wouldn't work 18:49
but let me chew on it for a second. 18:49
doomviking so glad my previous boss decided it was a good idea to outsource dns services. 18:52
yfeldblum doomviking, ? 18:53
doomviking ah. DDOS. 18:53
DDOS against zerigo.com 18:53
just bitching. carry on. 18:54
miah smart admins host primary, secondary, and tertiary dns name servers on different providers.. 19:05
jeffg i kind of get the impression that it's considered poor form to create a new top-level databag for an application, that chefs should nest their apps' data under a key in the top-level 'apps' bag instead. is this accurate? 19:06
doomviking miah: smart? whats that? 19:08
this isn't prodction, anyway. annoying. 19:09
hoover_damm teritary dns used more often now? 19:09
doubt it 19:09
gotta havethe glue set right now no? 19:09
yfeldblum jeffg, what gave you that impression? 19:10
doomviking pretty rare to see a tertiary 19:10
jeffg yfeldblum, looking at the radiant recipe from the hosted quickstart, along with enough programming experience to have developed an allergy to doing things that risk polluting a namespace 19:11
yfeldblum jeffg, i think of data bags as if they were mysql tables, and data bag items as if they were rows in a table, and the id field (or filename) of the data bag item as if they were the primary key values of those rows 19:13
jeffg you lost me at "mysql" ;) 19:13
yfeldblum jeffg, postgres? 19:13
jeffg that's better :) 19:13
yfeldblum jeffg, mssql? oracle? 19:13
jeffg, text file? 19:13
jeffg, does that way of looking at it help? 19:14
jeffg your analogy makes sense, i'm just a DB snob and a smartass 19:14
yfeldblum, thanks! 19:15
yfeldblum jeffg, right there with you on the preference for postgres 19:16
jeffg, however i find that the dev-null db system is much faster for writes 19:16
jeffg yfeldblum, what is dev-null? does it have btrees?!? are there many examples for its use on stackoverflow.com? 19:18
yfeldblum jeffg, it's kind of like redis, but more ephemeral 19:19
jeffg badass, i know where i'm putting all my data from now on 19:19
geekbri dev-null is so fast for reads and writes, even faster than mongodb! Same data resiliency. 19:19
;) 19:20
yfeldblum geekbri, turn on --journal in mongo then? 19:20
geekbri yfeldblum: Wait you mean I have to examine what I want from a product the configure it? 19:20
I can't do that sorry. 19:20
yfeldblum geekbri, do writes with safe mode? 19:20
jeffg geekbri, then you should just put everything in The Cloud. 19:21
yfeldblum geekbri, indeed! 19:21
geekbri If I put everything in the cloud it will reside in the same general vicinity as my head. 19:21
yfeldblum jeffg, dev-null is a true cloud db, because it is made of gaseous dihidrogen-monoxide formations 19:21
geekbri I don't store my data in databases, I just remember it all. It's faster that way. 19:21
jeffg <isaacwang> jeffg, mysql is a network-attached spreadsheet used by pr0n sites and other dodgy entities across the Internet. 19:21
geekbri i _AM_ the endpoint to the APIs 19:22
jeffg yfeldblum, ooh, i've heard about this dHmO thing. it's the new GaAs. 19:23
newb https://twitter.com/devops_borat new tweet needs to be "dev-null is so fast for reads and writes, even faster than mongodb! Same data resiliency" 19:23
jeffg +1 19:23
geekbri oh devops borat, how he captures my heart. 19:24
newb he captures the truth 19:24
lol 19:24
geekbri Eventually consistent is not say when eventually is. 19:24
jeffg lulz 19:25
yfeldblum geekbri, sometime after the partition has been fixed? 19:25
geekbri yfeldblum: Or sometime after the partition is destroyed! 19:25
newb I play kraftwerk's boing boom tschak everytime new tweet from devops_borat 19:26
geekbri I want to know who the man behind the curtain is 19:26
man/woman 19:27
yfeldblum geekbri, dev-null is a linearly scalable high-throughput low-latency immediately-consistent always-available partition-tolerant distributed database 19:27
geekbri because he/she is very funny 19:27
yfeldblum: That had enough buzzwords to sell me, who do I pay to implement it??? 19:27
yfeldblum geekbri, 10gen 19:27
jeffg might be too many buzzwords for a single tweet 19:28
atomic-penguin 140 chars, its legit, can't tag it with OH though 19:29
jeffg yow! echo 'dev-null is a linearly scalable high-throughput low-latency immediately-consistent always-available partition-tolerant distributed database' | wc -m 19:29
140 19:29
yfeldblum "I beat CAP with dev-null!" is short enough to fit on a t-shirt 19:29
atomic-penguin jeffg, I did the same thing 19:30
yfeldblum https://twitter.com/yfeldblum/status/227485722209292288 19:30
newb +1 19:33
geekbri damnit well now I can't retweet it bryanwb already did ;) 19:36
only one bwb is allowed to retweet at a time 19:36
miah heh 19:38
geekbri yfeldblum: I beg you to change your twitter avatar off that horrid blue egg :) 19:40
newb hoi bryanwb, where the fresh logstash content on showz 19:41
geekbri I'm not foodfights bryanwb, I'm just normal existence Brianwb 19:41
miah when people don't change the default icon it always irks me 19:42
geekbri I especially just hate that icon. 19:42
miah default icon = probably a spammer 19:42
geekbri I dont know what it is about it, maybe some suppressed traumatic childhood memory. 19:42
miah icon of some hot chic = probabl a spammer 19:42
geekbri hah, maybe thats it. 19:42
bytheway Has there been an omnibus sever 10.10 release for redhat yet? 19:48
Looking at the s3 repo, not seeing it, just making sure I'm looking at the right place 19:49
newb bytheway, there's a knife server plugin 19:51
bytheway That only works for Debian/Ubuntu, does it not? 19:52
mattray bytheway: probably not until Chef 11 19:55
bytheway Ok, from the chefconf intro I thought it was going to be earlier than that 19:57
mattray I don't think there's a lot more 10.x releases planned 19:58
10.14 has why-run and logging enhancements 19:58
jtimberman 10.14 is beta! 19:58
try it 19:58
mattray it's super-dope! 19:58
bytheway Yeah, if I'm going to have to brute force install it, I'll go for the latest 19:59
erratic http://sphotos-b.xx.fbcdn.net/hphotos-ash4/376310_332272673527109_563587752_n.jpg 19:59
bytheway Wish I didn't have to use redhat, dang large corporations 20:01
anthroprose truth to that 20:02
ssd7 hrmm. I can see that someone in this room pinged me, but can't find it for some reason 20:08
geekbri ssd7: It was just me :) 20:08
ssd7: I was just letting you know that i rebased 20:08
erratic dude redhat is ghetto 20:12
why the fuck 20:12
would anybody want to use it 20:12
I dont get it 20:12
bytheway Because some bean counter won't let us run anything that we can't buy support for. 20:12
erratic saying you use it because you're a large corporation is like saying you want aids just because you dont want to have to worry about whether or not you'll ever get aids.... 20:12
ohhh support 20:12
well then why not just use microsoft then 20:13
bytheway I don't make the call, unfortunately 20:13
erratic lol 20:13
bytheway Hey, don't give them any ideas 20:13
We actually do have a lo 20:13
erratic no I get that 20:13
bytheway A lot of MS 20:13
erratic support makes sense 20:13
bytheway It would make more sense if we actually used it... 20:14
erratic yeah I mean 20:14
thats the second half of that 20:14
who can't support linux on their own 20:14
besides people who don't know what they're doing 20:14
what a waste of money to hire people who do know what theyre doing and pay for support 20:15
yfeldblum erratic, paying insurance premiums =/= wasting money 20:17
erratic, but of course a policy that says you have to buy an insurance premium for every pencil, desk, and ceiling tile that you buy ... is madness 20:18
erratic yfeldblum are we talking about like 20:19
um 20:19
ssd7 geekbri: cool, thanks :) 20:19
erratic compliance bs 20:19
geekbri ssd7: no no no, thank YOU 20:19
erratic because I dont know anything about compliance :( 20:19
yfeldblum erratic, bytheway saying "some bean counter won't let us run anything that we can't buy support for" 20:20
erratic yeah but does this have to do with like pci dss compliance or something. 20:20
is that why you buy support? 20:20
cheeseplus bytheway: I feel for you on that one, we had to go RHEL at the University I was working at for similar reasons 20:20
bytheway I'm at a healthcare company 20:21
Compliance is everything 20:21
cheeseplus on my side it was meeting FERPA compliance which is exactly although not quite as bad as healthcare compliance issues 20:21
close though 20:21
erratic HIPAA too I know a little tiny bit about it 20:22
geekbri a healthcare company??? 20:22
erratic logistical nightmare 20:23
geekbri quick, somebody start an angry mob! 20:23
bytheway Lol 20:23
erratic I second that 20:23
pitchforks out 20:23
anthroprose haha... I used to work in insurance leadgen <shudders>.... All the same exact private health information.. but none of the regulation! 20:23
bytheway It's a pretty good place overall, can't complain too much 20:24
geekbri Sweet, so that is how the world knows about my count choculitis 20:24
anthroprose attorney general of mass decided we didn't fall under hipaa 20:24
cheeseplus erratic: when HIPAA and FERPA meet it's frigging terrifying 20:25
anthroprose I filled out a mortage refi info request at a local credit union... then went the drivethrough at my big bank and they asked me why I wasn't looking to refi with them... that was my latest one 20:25
cheeseplus we _had_ to use Oracle for a long time just to meet certain compliance standards 20:25
it took like 3 years to get MySQL certified 20:26
erratic anthroprose lol 20:32
Majost How does one define an order of operations for a recipe? 20:39
chip- Top-to-bottom 20:40
Majost chip-: Thats not working in this case 20:41
chip- Some things get queued until the end (service actions, for instance) 20:41
Majost some actions happen before others in the sequence no matter where they are, which is breaking my cookbook 20:41
geekbri Majost: can you gist it for us? Are they resources that are firing out of order, or is it some ruby code being evaluated before you want it to be 20:41
Majost eg: rabbitmq is installing before the application rabbitmq uses is installed, no matter where I place it in the recipe. 20:41
miah Majost: make your cookbook smarter 20:43
newb if I do a "knife role from file base.rb", then update the .rb and do "knife role from file base.rb" again, will uploading the same role overwrite the old role? 20:43
miah newb: yes 20:43
Majost miah: https://gist.github.com/3166106 20:44
miah not reading gists today 20:44
geekbri Majost: keep in mind that include_recipe actually runs the recipe you pass it. 20:44
benblack #lazyweb: what version of chef-client is currently in the ubuntu 12.04 apr repo? 20:45
apt 20:45
jtimberman: ^^ 20:46
miah seems like you could just query the repo and find the answer rather than bothering people 20:46
benblack oh, i'm sorry, miah, i thought this was the #chef channel where you asked questions about chef. 20:46
Majost /topic 20:46
heh 20:46
benblack i didn't realize i had joined #butthurt 20:47
doomviking oh dear 20:47
miah sure ask chef questions, but asking something that would literally take you <5seconds to figure out on your own is really frustrating 20:47
benblack you poor thing. 20:47
erratic come on yall 20:47
benblack i encourage you to avail yourself of /ignore 20:47
then you needn't concern yourself with my questions at all. 20:48
miah ok. i gladly wont. but i wonder if i'll see you asking the same question in an hour 20:48
benblack you're precious. 20:48
miah <3 20:48
anthroprose ben: I would avail yourself of installing from gems instead of the ubuntu repo... ruby dependency nightmare sometimes 20:48
geekbri My mellow is being harshed so hard right now. 20:48
anthroprose geekbri: more coffee 20:48
benblack anthroprose: ok, thank you, was debating the various options and there was no clear right answer. 20:49
geekbri anthroprose: I do not consume caffeinated beverages! 20:49
anthroprose ben: I don't know if there is... I've just had issues in the past 20:49
behemphi *erratic I am sorry to be somewhat off topic, but could you explain how /ignore works. This is my first foray into IRC 20:49
benblack i have, too 20:49
miah benblack: /ignore <person> is pretty much all you need to know 20:49
er, behemphi 20:49
yay tab complete fail =) 20:50
behemphi ah, thx miah 20:50
anthroprose ubuntu 12.04 wasnt/isnt in the platform list from the repo... but the gems have always worked 100% for me 20:50
yfeldblum miah, i believe the term is "gistignore" 20:50
benblack just assembling a little script and didn't know whether it would be simpler to install ruby and then chef-client via gems or just chef-client from repo 20:50
geekbri wouldn't it be I'm gistnoring ? 20:50
miah generally when i want to ignore irc, i just detach my tmux session 20:51
benblack ruby and chef-client it is, thanks for the help 20:51
behemphi *benblack I have tried the apt install on 12.04 once (about 2 months ago) and it was dodgy (there were two gems I had to go find). I have did not have a problem with "gem install", "knife bootstrap" or "knife rackspace server create ..." 20:51
anthroprose if its on a newly minted server, I would suggest bootstrapping... 20:51
miah omnibus installer ftw. 20:51
behemphi So I second the "better to install from gems anyway" 20:51
benblack is there knife gce server create yet? 20:51
evilantnie is there an easy way to check within a recipe which type of chef client is currently running, specifically solo vs server? 20:52
behemphi @anthroprose that begs a question for my not-so-sysadmin self. 20:52
If I have existing infrastruction and want to roll out chef 20:52
yfeldblum benblack, gce? 20:52
behemphi do I just insure I have it right through copious testing then roll it over the existing stuff? 20:52
miah evilantnie: Chef::Config[:solo] 20:52
evilantnie thanks miah 20:52
benblack yfeldblum: google compute engine 20:52
behemphi (I get there are alot of ways to approach, and I am asking in general) 20:53
anthroprose behemphi: I am fortunate enough to be completely greenfield... however when back-applying I have a tendency to pre-create all my creds and then just dump the .pem and update the client.rb on the clients/nodes 20:53
not that that addresses your running infrastructure new code in existing prod issue 20:54
behemphi That does, however help me start thinking about the problem some 20:54
I was greenfield too 20:54
anthroprose I will be at the meetup thursday... 20:55
behemphi hey, what is your @ctual name? 20:55
anthroprose Alex Corley... we met at the mysql meetup... talked for a bit out front 20:55
you find a python guy yet? 20:55
behemphi yep, I remember 20:55
geekbri Boston? 20:56
bawt_ Boston is naughty 20:56
geekbri haha 20:57
I didn't expect that. 20:57
behemphi @anthroprose, not yet. We are delaying that in the face of a couple other opportunities. Ah startup life :-) 20:57
geekbri behemphi, anthroprose: are you both boston based? 20:57
anthroprose geekbri: Austin 20:58
Austin Massachusetts.. 20:58
behemphi yep, ATX man 20:58
geekbri I suspected as much! 20:58
perhaps ill see you guys at a meetup 20:58
behemphi not as smart as boston but we got groovy :) 20:58
geekbri Grails 4 life ? 20:58
behemphi what is your @actual name then? 20:58
geekbri @brianwbianco 20:59
just like my twitter claims! 20:59
behemphi great, see you Thursday. 20:59
geekbri There is a meetup on thursday? 20:59
behemphi behemphi is a play on my name 20:59
geekbri tis not on me calendar. 20:59
behemphi sec, I will get you a link 20:59
http://www.meetup.com/Austin-DevOps/ 21:00
we are just getting started 21:00
freerobby I'm able to deploy fine when logged into a box, but when I run "knife ssh" as the same user I get "ec2-23-20-158-139.compute-1.amazonaws.com [Mon, 23 Jul 2012 20:58:47 +0000] DEBUG: STDERR: Bundler::GemNotFound: Could not find rake-0.9.2.2 in any of the sources". Any thoughts on what else could cause this discrepancy? 21:00
anthroprose hows parking down there? 21:00
behemphi I am hoping to attract studs like Matt Ray 21:00
ON a thurs night it should be good 21:00
that early anyway :-) 21:01
but the attraction part comes after we are established 21:01
anthroprose cool... wasnt sure if there was a garage or something 21:01
freerobby Here's my knife command (i login as "ubuntu"): "knife ssh "role:rails_app_server" "sudo chef-client" -x ubuntu -a ec2.public_hostname -C 1" 21:01
cheeseplus behemphi: fyi, there is probably a bit of overlap with the Austin Web Arch meetup 21:01
which is poorly named since we end up talking infrastructure and devops more often 21:01
behemphi thx @cheeseplus that is part of the "agenda" this week is to pick the right night. That is hard. Maybe we need to get the groups combined? This is not specifically Chef (also Puppet, CFEngine and self rolled) 21:02
I have no skin in "hosting" this thing just filling a percieved need 21:03
cheeseplus It may be beneficial, web arch is still small and I know the guys already running it 21:03
brandonvalentine google and the chef docs seem to be failing me, but does anyone know of a way i can get the path of the currently executing cookbook from inside a recipe? i have a file that's part of the cookbook i need to read into a ruby object, and i'd rather not have to install it somewhere first using cookbook_file or the like 21:03
cheeseplus there are other devops meetups but they are up northwest 21:03
behemphi <consternated> hmm, need to do better research then and work to consolidate </consternated> 21:04
cheeseplus behemphi: I'll sign up and hopefully make it, my impression is that there is plenty of interest 21:04
mattray behemphi: at some point you'll probably be double-booked against ACUG and OpenStack Austin, both on Thursdays 21:04
cheeseplus it's just getting the word out that is the hard part 21:05
behemphi If location is the only issue, Jason Cohen has offered the wpengine offices as a bigger venue 21:05
anthroprose cheese: I'm NW 21:05
and would prefer to stay here ;) 21:05
behemphi I am Pflugerville 21:05
freerobby Has anybody else seen inconsistency between ssh vs "knife ssh" environments? Does knife do anything funk with the environment that I should be aware of? 21:05
anthroprose anderson mill here.. 21:05
mattray behemphi: I'm doing the Agile Austin DevOps SIG Wednesday at lunch 21:05
cheeseplus so Canada 21:05
anthroprose South Dallas 21:05
or as the classic FB map coined us ^^ "Who Cares" ^^ 21:06
mattray and tomorrow's Agile Austin Architecture SIGN is DevOps-related too 21:06
(also at lunch time) 21:06
cheeseplus I had a friend do some great meetup research that seems to prove out downtown being better for meetups generally 21:06
anthroprose mattray: are the lunchtime meepups worth it? 21:06
mattray anthroprose: free lunch ;) 21:06
anthroprose haha 21:06
im down near bazzarvoice, so I was going to do the velocity one 21:07
mattray yeah, I'll be there 21:07
behemphi My experinence is that the room makes the meetup if it is to be more than a social. 21:07
mattray yup 21:07
behemphi So I am inclined to go with the facility 21:07
So let me ask this question then b/c Matt gets me thinking 21:08
mattray behemphi: you want it downtown? 21:08
behemphi is there really a need to have a dedicated Devops meetup? 21:08
if there are a number of groups focusing on it under differentt titles? 21:08
mattray behemphi: I don't know. The lunch ones get 20ish people 21:08
cheeseplus web arch is about the same number 21:09
mattray OpenStack is 30-50 21:09
behemphi @mattray I don't really care personally. MySQL is getting 10 to 13 consistently but the space at Pervasive is great for presentations 21:09
mattray the Linux meetups seem to be getting anemic 21:09
behemphi I have preferences for the facility and consistency (same time/place each month) 21:09
erikh add more iron to the food? 21:10
mattray too many of them 21:10
behemphi I should say if you don't know Pervasive is NW out by Apply 21:10
s/Apply/Apple 21:10
mattray yeah, I've been there 21:10
I'd probably make about half of them, same as the OpenStack meetups 21:11
between travel and needing to be home with my kids, nights are hard 21:11
btm benblack: https://github.com/opscode/knife-google 21:11
behemphi Well the goal is to shake all this out Thursday night. Lunch seems a great suggestion 21:11
mattray lunch I would make a point to make 21:11
much easier 21:11
behemphi I will be sure to bring it up! 21:11
mattray (and I gotta eat anyway) 21:11
benblack btm: thanks! 21:11
btm benblack: btw, from what I've heard, you probably won't like the gce 'api' 21:15
benblack btm: don't know why that is in quotes, it's a really clean REST api and i like it a lot 21:17
btm benblack: maybe things have gotten significantly better and all is good then. I haven't looked at it since before launch. 21:20
benblack before the private beta? 21:20
https://developers.google.com/compute/docs/reference/v1beta12/ 21:20
btm benblack: when I looked at it last we had to download a cli tool and wrap it because the API wasn't stable. I'd have to search my email to recall how long ago that was. 21:21
benblack that is still the case, but the api is nice 21:21
i'm using gcutil for everything, though 21:22
freerobby Is anybody who works on the knife core utility around by chance? 21:22
btm freerobby: yes, there are knife developers around. 21:24
benblack: glad you think so. I anxiously await the future where we use it directly/via fog. 21:25
freerobby btm: I'm trying to figure out why a "knife ssh" command runs with a different environment than the same ssh command on the box. 21:27
wasn't sure if this was the best place 21:27
anthroprose define different environment.... chef environment, console environment? 21:28
freerobby anthroprose: shell environment 21:28
for instance 21:28
on box i'm able to see "JBOSS_HOME=/opt/torquebox/current/jboss" 21:28
but that whole variable is missing when run through knife ssh 21:28
benblack is knife ssh running as the same user? 21:28
freerobby both running bash shell for same user 21:28
yes. 21:28
btm freerobby: what script defines JBOSS_HOME? 21:30
freerobby benblack: the only variable showing in both that is different (besides ssh ports) is TERM 21:30
anthroprose free: possibly source your .bashrc or whatever is populating the environment variables from the knife ssh run? 21:30
paulmooring freerobby: where is JBOSS_HOME set? 21:30
benblack freerobby: that's a pretty good hint at the problem 21:30
freerobby btm: paulmooring: It's set in a torquebox init script. I'm just trying to figure out why one new ssh session would have it and a knife session wouldn't, and if it implies anything about the environment that knife creates. 21:31
workmad3 freerobby: knife ssh would be the equivalent of doing ssh with a command, which doesn't do a login shell 21:34
benblack ding 21:34
btm freerobby: something else is going to be executing that script when you login, anthroprose offered bashrc. you'll need to source the same login scripts if you want those variables set (or just set them as part of the command) 21:35
freerobby btm: I was assuming that knife ran a standard bash session in interactive mode. Is that errant? 21:37
workmad3: Just saw that, thanks. 21:37
leifmadsen not sure if this is a silly question, but is there a way to tell chef-client to attempt to connect to the chef server via a specific interface? 21:42
cakehero anyone know why this role wouldn't be recognized: http://cl.ly/image/040N3X151r09 21:43
btm freerobby: knife ssh uses net::ssh, which does not provide a login shell. it's very low level. you could use the screen or tmux options to run a login shell if you wanted. 21:43
cakehero i have the 'name' attribute set in the roles/<role_name>.rb file ('api_server') 21:43
behemphi @cakehero 21:43
check the case 21:43
cakehero sup 21:43
behemphi roles are case sensitive 21:43
cakehero all lower all the time 21:43
behemphi well damn that was supposed to be easy! 21:43
freerobby btm: I will look into that, thank you. 21:43
cakehero could i be overwriting the name somehow on accident? 21:44
I'm a r00bie newbie 21:44
btm freerobby: if you need an env via knife ssh I would do something like 'knife ssh foo:bar JBOSS_HOME=/dev/null /opt/jboss/start_something' 21:44
behemphi my other suggestion would be to insure you have uploaded it to the server if it is a new role 21:44
cakehero behemphi chef solo 21:44
behemphi no server 21:45
behemphi I am not a hero and will not be having cake tonight #sad 21:45
cakehero I put an "exit" int he roles file and it didn't exit, perhaps it's not being read? 21:45
behemphi sorry man 21:45
cakehero any intricacies w/the roles_path? 21:45
behemphi I don't use files for my roles yet (#shame) so I am not versed there 21:46
will have to defer to the chef gods 21:46
cakehero any way to debug the roles path to make sure it's being set properly? 21:46
workmad3 cakehero: can you pastie the roles file too pls? 21:46
cakehero workmad3 sure 21:46
workmad3 https://gist.github.com/c086230aeb2278fb6be0 21:49
thanks 21:50
workmad3 behemphi maybe something to do w/ruby versions or chef versions? 21:50
workmad3 cakehero: hmm... actually, it could be because you're doing ~/chef/roles as your roles path... 21:51
cakehero I used to have it as 'roles' then './roles' and then '~/chef/roles' 21:51
I tried all three 21:51
but i'll try a different one if you think I should 21:51
workmad3 cakehero: try File.expand_path("~/chef/roles") 21:51
behemphi I guess my only input is that the hard code just "feels" wrong 21:52
cakehero workmad3 ok that makes it absolute? 21:52
behemphi maybe shoudl say "absolute path" 21:52
cakehero behemphi whats hard code? heh 21:52
workmad3 behemphi: it's a config file 21:53
behemphi: once you're into config files, you tend to need to push in absolute paths :) 21:53
cakehero so I'm czeching out chef/role.rb 21:54
and it looks kosher 21:54
workmad3 I will try the absolute 21:54
workmad3 nice - it thinks the file doesn't exist, I bet your solution is right 21:56
newb @mattray: https://github.com/mattray/spiceweasel/blob/master/lib/spiceweasel/cookbook_list.rb#L72 22:11
will this turn "depends 'apt'" into a dependency called '' instead of 'apt'? 22:12
btm freerobby: please remember to follow up with the mailing list thread you created as to what you learn. 22:14
freerobby btm: Will do. Still trying to figure out exactly what's different. 22:18
erratic https://github.com/opscode-cookbooks/application_php is this getting rewritten? 22:20
flaccid_ i hope not coz i'm using it atm 22:21
erratic: where did you hear that? 22:21
erratic flaccid_I heard something about an application cookbook being rewritten at the first 206 devops meetup I went to a few months ago 22:22
I don't remember exactly how long ago it was 22:22
I seem to want to remember that this application cookbook was a unified cookbook and now its several different application cookbooks but I could just be smokin crack 22:23
4 months seems too long ago though 22:23
so I thought I might check 22:24
flaccid_ also don't worry too much about whether or not it does 22:24
all that amtters is that what you have suits your purpose *shrug* 22:24
I'm not in a hurry to start a cookbook like this tbqh 22:24
but it couldn't hurt 22:24
you know what would help me alot is to find out if there's anything going on with this mysql cookbook for ubuntu 12.04 22:25
I looked into it today and it seems like the last one they had support for was 10.0 22:26
so I'm wondering if I should hold off for someone else to work on that or if I should wait 22:26
or if I should just do it myself 22:26
erratic so 22:35
are there any good opscode chef books yet 22:35
rava erratic: the wiki's pretty good. from there i just look through community cookbooks 22:36
anthroprose erratic: dont bother with the test driven infrastructure book 22:36
its like f'n 70 pages or something 22:37
rava anyone using the users community cookbook? 22:37
i'm trying to figure out how to tell it to skip users that are logged in so it doesn't puke half way through a run 22:37
erratic anthroprose I seem to remember I had a pirated copy of the test driven book 22:45
and that it was just a bunch of pseudo intellectual nonsense about agile this and that 22:45
and not much about what I needed to do to get the shit working 22:46
I could just be smoking crack though 22:46
anthroprose sounds about right 22:46
how about we publish the chat logs & make a killing? 22:46
erratic hehe 22:47
anthroprose there has to be some sort of form you can fill out that produces an oreilly book 22:47
yfeldblum it's also a very early work whose purpose was to introduce the idea into the minds of people who have never heard of it and to propose the way forward 22:48
erratic I was thinking about that a second ago and I decided not to say anything in favor of my own motivation but I'll bet you anything they're picky about what they'll take and who they will take it from 22:48
yfeldblum one usually does not see the comprehensive treatise on a subject until that subject has been somewhat developed 22:48
anthroprose yfeld: its just frustrating to people looking for something substantive... and that little tomb is literrally the only volume on the market 22:48
yfeldblum anthroprose, indeed; newton had the same complaint on the subjects of the derivative calculus and the phenomenon of gravitation 22:49
erratic lol 22:50
gettin nerdy up in here 22:50
anthroprose and look where it got us.... we're still living in a newtonian driven understanding way past its time... 22:50
yfeldblum anthroprose, to begin getting technical, newton has not been superceded; he has only been extended 22:51
anthroprose teehee 22:51
agreed... just frustrated at my public schooling 22:51
I was taught atomic theory as marbles wizzing around a globe ya know... 22:52
at least we talked about location probability... 22:52
yfeldblum anthroprose, that's what amazon is for 22:54
anthroprose, and here we thought that amazon was for hosting websites ... 22:54
anthroprose and thats how I ended up with test driven infrastructure :( 22:54
hehehe 22:54
I find short format presentation is better for us ADAHAD whatever kinda people 22:55
so I tend to do most of my learning through forum/blog discussions that include references 22:55
metafilter is good for such things 22:55
rava anyone know how i can check if a user is logged in in a provider? 22:57
yfeldblum rava, how would you do it from the shell? 22:57
rava basically: $(ps -ef | grep USER | grep -v grep | wc -l) > 0 && skip_this_user 22:58
using the users cookbook, but the recipes puke half way through if a user's logged in 22:59
yfeldblum rava, you can use that same technique in a provider then 22:59
rava, the basic ruby technique for shelling out is `` ... e.g.: result_from_stdout = `ls -A` 23:00
rava yfeldblum: right, i was looking to see if there was a method in the User lwrp that users uses, but it doesn't seem to be the case 23:00
yfeldblum rava, but using the method `shell_out`, which should be available in provider instances, is often more preferable 23:01
rava yfeldblum: oh? interesting. i'd think one would want to grab from the lwrp methods first before hitting the shell 23:01
rava yfeldblum: Chef::ShellOut.new() is the method you were talking about? 23:17
newb is there a tool that given a cookbook name, will output a complete, recursive list of dependencies? 23:18
I am using Berkshelf, but when I want to upload a cookbook to a remote server, I am getting dependency errors as my "knife upload" commands are out of order 23:18
or is it possible to ask Berkshelf create such a list 23:20
newb . 23:31
stormerider so im a little confused about one thing. im using the user cookbook and i have data bags for the user-specific data (like uid, etc). some of the folks have left since i started taking this over. so i put in "action": "remove" for those users, but when i did a chef-client run, it didn't actually remove those users. 23:45
rava so if you look at the data bag with knife show users username, the remove action is there? 23:46
erratic http://wiki.basho.com/Concepts.html 23:47
so awesome 23:47
stormerider if i do "knife data bag show users <username>", yes, it's listed there. 23:48
rava hrm, and no errors spit by the client run? 23:48
stormerider no, it actually updated the SSH keys for the user in question, heh. 23:48
rava and 'recipe[users]' is in the node's run list? 23:48
anthroprose erratic: there is a youtube video where they talk about the algorithm implementation.... 23:49
rava anthroprose: want 23:49
stormerider the node list has a role which includes 'role[server]' which has 'recipe[users]' 23:49
rava those are the 2 things that buggered my use of that cookbook 23:50
stormerider rava: didnt quite follow that, could you elaborate? 23:52
rava stormerider: meant that those were the 2 gotchas i ran into. first, i don't know why the data bag didn't update when i ran "from file". secondly, having 'recipe[users]' in the run list, since when i first used it i had made a few other recipes that i had in my run_lists without the default recipe. 23:55
stormerider: try running a search for "action:remove" and see if you get the users you want deleted in the list 23:55
then make sure that the :remove action is using the right field (u['id'] on mine. which is the user's username. userdel (what the lwrp uses) won't work against the uid if that's what is being returned by the u['id']) 23:56
stormerider bit of a chef newbie, what's the easiest way to run that search? 23:58