Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2013-09-04

| Channels | #salt index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:03 Jahkeup_ joined #salt
00:07 jbunting joined #salt
00:12 [diecast] joined #salt
00:17 drags any tricks for globbing with cp.push or alternate methods to achieve the same?
00:19 kenbolton joined #salt
00:23 Katafalkas joined #salt
00:27 cocoy1 joined #salt
00:27 juanlittledevil joined #salt
00:30 dthom91 joined #salt
00:36 teebes joined #salt
00:44 jbunting joined #salt
00:46 jslatts joined #salt
00:49 mannyt joined #salt
00:50 mwillhite I'm building a box with salty-vagrant, how can I make sure everything is built with a user (other than root?)
00:53 azylman joined #salt
00:54 mannyt joined #salt
00:55 teebes Is there a way to only test one specific state declaration from the command line instead of highstate?
00:55 azylman I've set up a vagrant machine in order to test a new state that I'm adding to salt, but when I try to call highstate it only outputs three lines before saying "No Top file or external nodes data matches found". top.sls does exist, thoug: http://pastebin.com/yVhvUatZ
00:56 avienu joined #salt
00:59 mannyt joined #salt
00:59 juanlittledevil joined #salt
01:02 m_george|away joined #salt
01:23 Katafalkas joined #salt
01:24 mannyt joined #salt
01:33 dave_den teebes: yes, using state.sls http://docs.saltstack.com/ref/modules/all/salt.modules.state.html
01:37 Jahkeup joined #salt
01:39 lemao joined #salt
01:40 kenbolton joined #salt
01:40 liuyq joined #salt
01:41 dthom91 joined #salt
01:42 jbunting joined #salt
01:43 teebes dave_den: thx, but if within that state file I only want to test 1 specific named function?
01:45 dave_den teebes: state.single.
01:53 mwillhite joined #salt
01:59 torandu joined #salt
02:03 jbub joined #salt
02:04 sixninetynine joined #salt
02:06 xl joined #salt
02:12 rberger joined #salt
02:15 druonysus joined #salt
02:16 mannyt joined #salt
02:18 mafrosis joined #salt
02:19 mafrosis yo, can anyone give me a pointer on how to format an ENV var for a cmd.run state?
02:19 mafrosis I appear to be unable to figure it out
02:19 teebes joined #salt
02:20 mafrosis I suppose that could be rephrased as, how do I format a dict in yaml :)
02:24 Katafalkas joined #salt
02:24 racooper joined #salt
02:31 dave_den mafrosis: env:\n    ENV_VAR:  value
02:31 mafrosis aah right. Like context/defaults for jinja
02:32 mafrosis dave_den: thanks
02:32 dave_den np
02:36 baniir joined #salt
02:39 mgw joined #salt
02:44 jinrou joined #salt
02:49 HuangShaoyan joined #salt
02:51 ipmb joined #salt
02:52 HuangShaoyan left #salt
02:56 xl joined #salt
03:00 NV hrm, the key names as reported by salt-key thusfar have been fqdns, just added a new box and it's only the hostname, not the fqdn
03:00 littleidea joined #salt
03:00 NV hostname returns the fqdn on the machine, any idea why it would be different?
03:00 NV hasn't been specified manually or anything
03:01 logix812 joined #salt
03:02 joehh NV: depending on versions, some of that minion name code changed recently
03:02 lemao joined #salt
03:02 NV 0.16.3, same as all other machines
03:02 joehh NV: in nearly all cases now, we set our minion names
03:02 NV just thought to try socket.getfqdn() and that returns just the hostname, not the fqdn!
03:02 joehh odd
03:02 NV yeah...
03:02 joehh could the /etc/hosts file be different between the minions
03:02 joehh ?
03:03 NV ah sonofa...
03:03 NV that'd do it
03:05 NV cheers :)
03:07 Nexpro joined #salt
03:10 CheKoLyN joined #salt
03:14 drags is it common that cp.push takes over a minute to return, does not list all of the minions in its output, and in actuality has already recieved (or at least created placeholders for) all files from minions about 20s into that run time
03:20 mmilano left #salt
03:21 Furao joined #salt
03:24 Katafalkas joined #salt
03:31 rgarcia_ joined #salt
03:31 vbabiy joined #salt
03:36 VertigoRay joined #salt
03:37 dthom91 joined #salt
03:38 VertigoRay Hey guys ... been working on a dsconfigad wrapper for salt and and having an issue that I'm not sure how to troubleshoot:  https://github.com/VertigoRay/salt-osx-dsconfigad/issues/2  If you have any advice, please feel free to comment there.  I'm gonna run ang grab some grub, I'll try to check back in a bit.
03:39 Furao VertigoRay: is the file in your _states/ ?
03:39 VertigoRay What file?  I just made a module and put it in _module
03:40 VertigoRay err .. _modules
03:40 Furao Comment:   State dsconfigad.bind found in sls apps.dsconfigad is unavailable
03:40 Furao "State"
03:40 VertigoRay so here's the dilemma
03:41 Furao anyway, I don't know how that file can work
03:41 VertigoRay This works:  salt '*' dsconfigad.bind
03:41 Furao that's a module
03:41 Furao I mean, that's how you run a module
03:41 VertigoRay maybe I just don't know how to call the module from my top.sls
03:41 Furao module.run
03:42 Furao and "\t" -> " " * 4
03:43 Furao http://www.python.org/dev/peps/pep-0008/
03:44 Furao and you should not have that ret dict created at line 23, but create a different instance of it in each function
03:44 VertigoRay yeah, I've read pep8 ... I meant to convert it before posting to git.  My issue with spaces over \t is that preference is removed.   Python doesn't care what you do, just pep8
03:44 Furao if one module call an other you mess with it
03:45 middleman_ joined #salt
03:46 VertigoRay I see your point, but I took that method of doing the ret from salt code.  As it is, I return the ret to a new variable and if it needs to be saved globally, I save it globally.  Otherwise, the format is taken from global and modified privately.
03:47 Gifflen joined #salt
03:49 VertigoRay so, my sls should look like:  https://gist.github.com/VertigoRay/6432567  ?
03:50 VertigoRay nm, that's not it.  Tested it and got the same error, except this diff:  Name:  module.run
03:53 JaredR joined #salt
03:53 Furao no
03:54 Furao - name: dsconfigad.bid
03:54 Furao http://docs.saltstack.com/ref/states/all/salt.states.module.html#module-salt.states.module
03:54 wolfpack joined #salt
03:55 VertigoRay Grrr ... of course. Thanks.  Trying ...
03:56 Furao look at github.com/bclermont/states for plenty of example
03:57 kenbolton joined #salt
03:57 Teknix joined #salt
03:58 dave_den1 joined #salt
03:58 StDiluted joined #salt
04:00 pt|Zool joined #salt
04:00 djn joined #salt
04:03 Vasanth_ joined #salt
04:03 codysoyland joined #salt
04:05 VertigoRay Furao: Thanks, I've got a green response from "State: -module".  If I want to customize the output of that state, is that what the returner value is for?
04:09 tseNkiN joined #salt
04:09 gadams can I run commands from laptop to salt master to execute? or do I have to run everything from salt master?
04:10 jalbretsen joined #salt
04:14 CheKoLyN Hello All,  I have an error in a template; i'm trying to determine if a node is a hypervisor.  When I do   if salt['virt.is_hyper']       I get this error message: UndefinedError: 'dict' object has no attribute 'virt.is_hyper'
04:14 CheKoLyN but only if the node is not a hyper
04:14 CheKoLyN thank you for your time
04:16 Furao VertigoRay: depends of which kind of customization you mean :)
04:17 JaredR joined #salt
04:17 Furao CheKoLyN: https://github.com/saltstack/salt/blob/develop/salt/modules/virt.py#L39
04:17 StDiluted Evening all
04:17 StDiluted anyone using a database as an external pillar?
04:19 Furao StDiluted: indirectly, I wrote an inventory app in django and use a REST API to build some pillar from it
04:19 StDiluted hm
04:19 StDiluted not exactly what I'm thinking of
04:19 gadams Furao: is it on github?
04:19 StDiluted right now I have a python script which takes a few grains values and then returns some particular values based on those
04:19 Furao gadams: yes but a private repo, my client want it open-source but it's not yet released
04:20 Furao but it's used in prod
04:20 StDiluted I kind of want to do that in a database, rather than in a flat file
04:20 gadams does salt automagically add new hosts to the django app?
04:20 StDiluted I'm just not a programmer
04:20 Furao gadams: it's the other way around, It's a UI for salt-cloud as well, so you go into django admin interface and create new minions that ends created and managed by salt
04:21 gadams Gotcha
04:21 StDiluted that sounds fantastic. I eventually want to build something like that
04:21 StDiluted though I think it will eventually involve docker
04:21 gadams Furao: wanna add me to the private repo ;-) :-P
04:22 Furao StDiluted: client want to switch to docker too
04:22 Furao gadams: maybe, you want to contribute?
04:23 StDiluted wish it would go gold
04:23 VertigoRay Furao: I'll read up on returners and states next.  For now, I have the module running properly.  I just don't like the output -- but that's just my OCD. :) https://github.com/VertigoRay/salt-osx-dsconfigad/issues/2#issuecomment-23765097
04:23 gadams What are you looking for in way of contributions?
04:23 StDiluted wonder how rough it would be to set up dynamoDB as an external pillar
04:24 Furao it's quite complicated to setup as it requires many things to works, ext_pillars, custom modules, custom states. I basically wanted to fill all the missing parts of salt regarding inventory + automation of salt-cloud
04:24 StDiluted pillar is technically just a key/value store, right?
04:24 Furao I don't try to rewrite what salt already do
04:24 Furao StDiluted: yes
04:24 gadams I have a stupid question Furao
04:25 CheKoLyN Thank you Furao
04:25 Katafalkas joined #salt
04:25 Furao VertigoRay: well, if you want to just grab errors use my sentry returner :P
04:25 Furao gadams: ?
04:26 CheKoLyN I found the problem.  The issue was the missing python-libvirt package in the node
04:26 gadams Could I execute salt commands from my workstation that would push them to salt-master to be executed on minions and have data returned to my workstation ?
04:26 Furao CheKoLyN: yes, the __virtual__ returned False as it couldn't import libvirt
04:26 CheKoLyN before the virt.is_hyper was returning :     "virt.is_hyper" is not available.
04:27 Furao StDiluted: it's easy to write a ext_pillar for things such as DB, in my case I do many security check and other processing of data as well
04:27 StDiluted gadams: I believe salt-api is used for that
04:27 * gadams derpderp
04:27 Furao gadams: use salt-api for that and use salt-api to get jobs result using jobs.lookup_jid
04:27 StDiluted Furao: I'm not a programmer, so 'easy' is relative.
04:27 StDiluted I'm a sysadmin who can hack some shitty code together
04:28 StDiluted which is why I use salt to begin with
04:28 Furao gadams: I use salt-api in my inventory to ask salt to perform actions (run salt-cloud or state.highstate)
04:28 Furao StDiluted: I'm not a programmer too, maybe just write less shitty code :P
04:29 Vasanth_ left #salt
04:31 StDiluted hehe
04:34 Furao have you ever asked yourself, younger, even before your first job: "I'am going be a sysadmin or a developer?"
04:34 Furao probably all hackers did something in both worlds at some point
04:35 efixit joined #salt
04:35 CheKoLyN Furao: do you think is an error that is returning "virt.is_hyper" is not available  when libvirt couldn't be imported
04:35 CheKoLyN instead of False
04:38 Furao CheKoLyN: salt load ALL modules, frequently, to build the "salt" keys that hold all $module.$function (such as salt['virt.is_hyper'])
04:38 Furao as not everyone need libvirt and don't have python-libvirt installed
04:39 rmt StDiluted, All sysadmins should aim to be developers, imho.. There is a lot to be gained from adopting many of the practices, especially around quality control and repeatability (eg. build/deployment/etc. pipelines) .. Start by getting rid of the small scripts that you manually copy around and edit in place. ;-)
04:39 Furao you don't want this to be a fatal error
04:39 Furao rmt: true
04:39 Furao developer don't need to be sysadmin
04:40 Furao but they should understand where their code run (to prevent some basic mistake, such as hardcoded path)
04:40 Furao but sysadmin often need to compensate by also be a dev
04:42 rmt Furao, I like to make devs have shared responsibility for salt/puppet/chef configs.. use the same tools for their dev & test environments as for production, and have them maintain it.. and preferably as close as possible.. purposefully give them dev/test environments that are unique enough to require them to require some flexibility.
04:43 Furao sadly, in my past experience, most dev don't understand what an operating system is
04:43 Furao like it's some magic container that run their code
04:43 rmt ;-)
04:43 CheKoLyN Furao:  That's what I meant,  I had to install python libvirt on the node,  and restart the salt-minon service just to be able to evaluate  virt.is_hyper.  That seems rather extreme just find out if the node is a hypervisor.
04:44 Furao and too much time they just say "but it work on my workstation"
04:44 gadams I give my devs root access to the dev environments and staging
04:44 Furao CheKoLyN: you didn't had to restart minion
04:44 rmt Furao, it compiles, ship it! :)
04:44 StDiluted rmt: already got rid of most of those. I'm working on the dev side of things but I'll never enjoy programming.
04:44 gadams they don't realize I have their bash history automagically copied and backed up incase they change something in dev to get it working and decide to deploy to prod without making the same change
04:45 gadams i can go back and look at what they did
04:45 rmt StDiluted, I have a very short attention span ... most programming that I do is to make my life directly easier.  *then* it's fun. :-)
04:46 gadams StDiluted: I got a job on the spot because my employer asked "What's your primary goall as a sysadmin?"
04:46 gadams I stated, "To be as lazy as possible. When I've automated everything I can start working on value added services to the business"
04:47 CheKoLyN Furao:  I just did a test and I'm afraid that is salt-minion restart is required
04:48 robertkeizer joined #salt
04:48 StDiluted lol, that's pretty close to the answer i gave at this job.
04:48 StDiluted :)
04:48 StDiluted my goal is to make everything work without my intervention, as much as possible.
04:48 rmt StDiluted, And it's usually small programs (often CLIs) or re-usable libraries.  These can then often work together in more complex ways, ala Unix-tools.
04:49 StDiluted ok, off I go to sleep
04:49 StDiluted have a good night all
04:51 Ryan_Lane joined #salt
04:52 Furao gadams: if you know django and/or python + salt and looking for a job I might be interested :)
04:52 gadams I'm already maxed out for the next 5 or 6 months
04:53 gadams I work for 3 different corps and one has 3 agencies I managed
04:53 gadams s/managed/manage
04:54 Furao ok :(
04:54 Furao I already hired 2 guys here and looking for more
04:54 gadams I'll give your nick to a coworker
04:54 gadams PM me some company URLs or job descriptions
04:54 Furao that's a little more complicated :P
04:55 Furao where he's based
04:55 Furao ?
04:55 gadams we're all based in Florida
04:55 gadams He's on my red team, top notch guy
04:57 Furao US, hmm it's harder, but I know a mass company who's looking for someone
04:57 gadams Where are you based?
04:57 gadams Spain?
04:57 Furao furao is a portuguese word
04:58 gadams Ah
04:58 gadams okay
04:58 druonysus joined #salt
04:58 Furao but not it's complicated, but it's partially in Hong Kong
04:58 Furao and our team is spread everywhere
04:58 Furao and I'm in Malaysia right now (I'm a P.T.)
04:58 gadams heh
04:58 gadams my guy does japanese translations lol
04:59 Furao one of my django guy spent few years in .jp
04:59 Furao but now he's in argentina
05:04 Furao gadams: so, if your friend want to do some commute works for a .ma.us company, I can help. I can't hired in the US because of crazy laws
05:05 Furao I just filled some forms about that lately and US make everything so complicated. I'm glad I'm not a US citizen and I don't have to deal much with that tax system
05:06 druonysus joined #salt
05:06 druonysus joined #salt
05:19 xl joined #salt
05:21 rberger joined #salt
05:25 Katafalkas joined #salt
05:25 xl1 joined #salt
05:28 carmony joined #salt
05:39 robawt is this the part where we all get to complain about the docs being offline?
05:40 robawt i kid, but they are offline :\
05:40 Furao robawt: http://salt.readthedocs.org/en/latest/index.html
05:41 Furao just make html in salt repo
05:41 robawt Furao: you win at life!  Thank you
05:41 ahammond joined #salt
05:43 luminous joined #salt
05:44 xmj does salt use tcp or udp for internal comms?
05:45 luminous joined #salt
05:46 robawt xmj: i have to assume tcp since it keeps connections alive, but i can check.  it's all zeromq under the hood
05:46 robawt xmj: http://docs.saltstack.com/topics/tutorials/firewall.html  based on this it looks like tcp
05:46 xmj it doesn't matter much, i want to block connections to the salt master from outside :)
05:46 aantony tcp:4505
05:47 txmoose joined #salt
05:50 xmj ah.
05:51 xmj blanko-approving 192.168.0.0/24 but wondering why i can connect to 450* . . . /sips coffee
05:58 robawt hey for tar.gz files looking @ http://docs.saltstack.com/ref/modules/all/salt.modules.archive.html
05:58 robawt would i need to call both archive.gzip then archive.tar?
06:01 robawt ah n/m
06:02 robawt I see people are making sure the tarball is there with file.managed, then calling cmd.run tar on the file with -cwd set to the install directory
06:03 dthom91 joined #salt
06:34 Katafalkas joined #salt
06:34 TheCodeAssassin joined #salt
06:35 TheCodeAssassin joined #salt
06:40 ml_1 joined #salt
06:43 sfello joined #salt
06:45 aantony joined #salt
06:45 linjan_ joined #salt
06:49 balboah_ joined #salt
06:52 xmj how can I control the amount of master processes running on, well, master?
06:58 crazysim joined #salt
06:58 z3uS joined #salt
06:58 NV what is the correct way to set a sls to run on all hosts except one or two? I have a postfix_mailhost sls that configures postfix for use on all general hosts, but i want it to not run on my mail servers which have a more specialised configuration
06:58 drogoh joined #salt
06:59 jheise joined #salt
06:59 NV i tried using the exclude directive in the second sls, but it didn't seem to work as i expected it would
07:00 cetex NV: if else :)
07:01 cetex NV: or, if not "mail server-identifier": do stuff.
07:01 NV mhmm
07:03 NV might setup a grain or something
07:06 NV hrm, am i missing something or is the only way to define grains for a minion via its config?
07:07 NV (ie, /etc/salt/minion or /etc/salt/grains)
07:07 Ryan_Lane joined #salt
07:07 NV no way to go "node that matches <x> should have grain <y>"?
07:07 druonysus joined #salt
07:08 carlos joined #salt
07:08 NV i mean i could get salt to manage the salt-minion config to specify grains, but it seems a tad hackish and would require two invocations of highstate before becoming correctly configured, and may possibly get incorrect sts's applied before the grain is set
07:08 littleidea joined #salt
07:10 ronc joined #salt
07:11 cetex true true.
07:11 cetex hm
07:12 cetex can't you identify them by some default-grain? :)
07:12 cetex like hostname?
07:13 cetex we run multiple identical racks where the same host in each rack should have some special stuff installed
07:13 cetex {% if grains['host'] == 'srv1' %}
07:13 cetex might not be applicable for you though.
07:14 NV yeah, that's how i was going to define the grains tbh
07:14 NV just thought it would be nicer to have a "grains['role'] == 'mail'" or similar
07:15 NV so any logic based on hostnames is done all in one place
07:15 NV then apply configuration using the same grain
07:15 NV etc
07:15 NV rather than having hostname matches in multiple places
07:15 NV might feature request it :P
07:17 cetex ;)
07:19 fredvd joined #salt
07:19 cetex we use "role" in our ec2 environment. cloudformation sets the role for each service we deploy and we use it in salt to deploy the stuff needed for each service.
07:23 kenbolton_ joined #salt
07:25 dave_den joined #salt
07:25 crazysim joined #salt
07:29 NV yeah, our virtual infrastructure isn't anywhere near that fancy :P (see: completely manual setup by another department after a massive crap papertrail)
07:31 TheCodeAssassin i wish i would have the money to deploy my stuff in EC2
07:31 TheCodeAssassin * well we
07:31 blast_hardcheese joined #salt
07:31 TheCodeAssassin we are still running proxmox on a dedicated server, it's cheaper
07:31 jinrou joined #salt
07:32 vaxholm_ joined #salt
07:34 scott_w joined #salt
07:34 kleinishere joined #salt
07:34 NV we at least have a proper vm cluster setup, i mean the kit isn't bad, just badly managed
07:35 Furao problem with proxmox it's there is no good api to manage vm
07:36 tseNkiN joined #salt
07:36 Furao like, no salt-cloud provider for it
07:36 Furao at least the version I worked with
07:38 malinoff joined #salt
07:39 unicoletti_ joined #salt
07:39 malinoff Hi all
07:40 isomorphic joined #salt
07:49 goki joined #salt
07:56 cetex heh :)
07:57 cetex i'd think that running stuff in aws is the most useful if you have lots of "stateless" instances serving / crunching data.
07:58 cetex or are running services like cassandra.
07:59 TheCodeAssassin problem is the traffic costs
07:59 felixhummel joined #salt
07:59 TheCodeAssassin with ovh the traffic is unmetered
07:59 cetex that way you can (rather simply) just setup all those instances of the same type in an autoscaling group and just kill / and have the autoscaling group rebuild instances when something breaks.
08:12 backjlack joined #salt
08:17 TheCodeAssassin guys, i have the following logic in my state file: https://gist.github.com/thecodeassassin/4b9fa5d516828d2d56fb
08:18 TheCodeAssassin it works, but at runtime, the first runtime the if statement returns false because (at the time of running the state file) the file does not exist
08:18 TheCodeAssassin even though the require is completely valid
08:21 jinrou joined #salt
08:30 tomviner joined #salt
08:39 kleinishere joined #salt
08:46 Sypher joined #salt
08:59 Pate_ joined #salt
09:03 felixhummel joined #salt
09:07 Jason-AVST joined #salt
09:07 nocturn Can I use a not construct to assign salts to nodes with(out) a certain grain?  somehting like !os: Ubuntu
09:07 ml_1 joined #salt
09:10 bitmand joined #salt
09:11 Sypher joined #salt
09:25 xmj something like {% if not grains['os'] == 'Ubuntu' %}
09:25 xmj ?
09:27 TheCodeAssassin anybody here using pycharm ?
09:27 TheCodeAssassin grains['os'] != ?
09:29 austin987 joined #salt
09:31 p3rror joined #salt
09:36 N-Mi_ joined #salt
09:36 N-Mi_ joined #salt
09:39 jinrou joined #salt
09:40 Jason joined #salt
09:40 nocturn I found it
09:40 nocturn you cannot put not in the front
09:40 nocturn '* and not
09:40 nocturn that works
09:41 Guest32648 hi
09:42 beyond_ joined #salt
09:42 FengRen joined #salt
09:43 boole_ joined #salt
09:43 beyond_ 好像很多人的样子
09:43 test___ joined #salt
09:43 FengRen Hi, 骚年们
09:43 boole_ 不是国人吧,鬼佬多点?
09:44 piepie_ joined #salt
09:44 pengyao joined #salt
09:44 bjfengfeng joined #salt
09:44 bjfengfeng hi
09:44 boole_ hi
09:44 beyond_ echo "hello"
09:45 Zandy joined #salt
09:46 Guest32648 salt 'everybody' test.ping
09:49 jaja joined #salt
09:50 jaja left #salt
09:50 aa joined #salt
09:50 felixhummel joined #salt
09:51 ttst joined #salt
09:51 today joined #salt
09:51 ttst a ?
09:51 Guest98411 left #salt
09:52 today hello
09:52 halcyon joined #salt
09:52 isomorphic joined #salt
09:52 today 小伙伴们出来吧
09:53 ttst 我了个顶!~。这个好好玩儿噢!~。
09:54 huzi joined #salt
09:54 huzi y
09:54 huzi hello
09:55 MrTango joined #salt
09:55 pengyao joined #salt
09:55 huzi everyone
09:55 Clion joined #salt
09:55 zooz joined #salt
09:56 today 第一次用这个,很简洁
09:59 nocturn Does salt keep any backups when replacing config files?
10:00 halfss joined #salt
10:03 pengyao yes, http://docs.saltstack.com/ref/states/backup_mode.html
10:03 whiskybar joined #salt
10:04 puluto__ joined #salt
10:08 puluto joined #salt
10:12 deedubs joined #salt
10:13 shiin joined #salt
10:14 shiin is there a way to reduce the output of state.highstate so I only see what changed or went wrong?
10:14 shiin scrolling through all that unchanged state text to see what went wrong isnt really helpful
10:17 xl1 "state_verbose" and "state_output" configuration does the trick
10:17 shiin I guess its not. https://github.com/saltstack/salt/issues/2422
10:17 shiin well yes, in states
10:18 shiin Id like to determine that while running the salt command
10:18 oc joined #salt
10:22 shiin do I really have to set those variables in -every- state?
10:23 malinoff shiin: Try to play with --out option
10:23 fredvd joined #salt
10:26 shiin oh well. I might just try to grep-ignore the color code.
10:26 shiin thank you
10:27 xerxas joined #salt
10:32 sssslang joined #salt
10:34 goki joined #salt
10:37 ml_11 joined #salt
10:37 giantlock joined #salt
10:40 vaxholm_ joined #salt
10:42 gamingrobot joined #salt
10:44 pcarrier joined #salt
10:46 APLU joined #salt
10:46 akitada joined #salt
10:49 felixhummel joined #salt
10:54 az87c joined #salt
10:55 az87c_ joined #salt
11:03 jinrou joined #salt
11:06 emocakes joined #salt
11:10 isomorphic joined #salt
11:21 km` joined #salt
11:23 Jagan joined #salt
11:23 Jagan Hi All
11:23 bhosmer joined #salt
11:25 montuori hi -- i'd like to use pkg.installed to manage sudo-ldap (ubuntu); to do that successfully it seems the env var SUDO_FORCE_REMOVE must be set.  it appears that cmd.run will accept an env parameter; is there anything similar for pkg.installed (it appears not) or a recommended best practice in this case?
11:29 oz_akan_ joined #salt
11:30 Katafalkas Hey, I am using parted to make a file system. I got warning that you are planing to remove parted in later releases. Is anyone working on anything yet ?
11:32 mwillhite joined #salt
11:34 sssslang hi there. i'm trying use config management tools to manage our ~100 linux servers(centos/rhel/ubuntu). I know salt after I've read cfengine3's manual. It seems salt has a lot of runtime dependences. Is it worthy of using salt instead of cfengine? I want to hear some advice from people who had use both.
11:39 malinoff joined #salt
11:42 blee_ joined #salt
11:42 Furao sssslang: cfengine is the perl of config management
11:43 sssslang Furao: you mean the syntax of config?
11:49 lemao joined #salt
11:51 guardian_ joined #salt
11:51 guardian_ left #salt
11:53 kenbolton joined #salt
11:53 dthom91 joined #salt
11:56 aat joined #salt
11:58 vaxholm joined #salt
12:02 TheCodeAssassin sssslang: what runtime dependencies?
12:02 Furao sssslang: I mean cfengine is old, clunky and ugly :)
12:03 TheCodeAssassin i use salt-bootstrap, never had issues with this
12:03 sssslang TheCodeAssassin: solid python and lots of python modules, some needs compiled locally.
12:03 Furao compiled locally?
12:04 TheCodeAssassin i would recommend you using this sssslang: https://github.com/saltstack/salt-bootstrap
12:04 sssslang like M2Crypto, zeromq, pycrypto.
12:04 TheCodeAssassin well this takes care of all of that
12:04 TheCodeAssassin https://github.com/saltstack/salt-bootstrap#supported-operating-systems
12:05 sssslang TheCodeAssassin: some of our servers using quite old gcc & glibc. so I had problems installing salt.
12:05 TheCodeAssassin using the bootstrapper?
12:06 sssslang But environment install is not a big problem. I can install all the dependence into a big bundle, from openssl to python.
12:06 sssslang Then I'm care about the usage.
12:07 Drexx joined #salt
12:07 TheCodeAssassin all i can say is, i've used puppet and chef (haven't used cfengine though).
12:08 TheCodeAssassin but salt is far superior
12:08 TheCodeAssassin imho
12:08 TheCodeAssassin or at least it can be
12:08 sssslang I'm new to these tools, all I used is cfengine3, :-/
12:11 Jahkeup joined #salt
12:12 oc sssslang: I've used cfengine, chef, puppet, salt and ansible. Cfengine is the biggest hassle; chef and puppet are nice, but require more of a Ruby ecosystem than i'd like to install on all servers (and has a slightly too big footprint); python comes bundled on most servers anyway, so I favor the latter two for an all around infrastructure solution. (Either is a great choice).
12:12 TheCodeAssassin well, what got me excited sssslang was the ease of configuration and scalability. Take a look at these state files: https://github.com/kevva/states
12:12 TheCodeAssassin you'll see what i mean
12:13 TheCodeAssassin and the fact that the chef server ate resources like a java app
12:14 jslatts joined #salt
12:15 sssslang oc: do you mean cfengine2 or 3? can you show me some details about the problems cfengine introduced?
12:15 TheCodeAssassin and hosted chef costs a lot once you go past the 5 nodes
12:15 sssslang TheCodeAssassin: thanks, the syntax is so meaningful.
12:15 TheCodeAssassin sssslang: is it right? i was thinking the same thing when i looked at it
12:15 TheCodeAssassin or was the sarcastic ? hehe
12:16 sssslang TheCodeAssassin: i dunno, since i didn't read the salt manual.
12:16 TheCodeAssassin it requires a bit of research. i'd suggest googling for yaml and jinja as well as salt.modules and salt.states
12:17 TheCodeAssassin for configuration management i use salt states to put my machines in a certain state, based on a role that server should take on
12:17 oc sssslang: i've used both. v3 is a vast improvement, but salt/ansible are just so much more malleable
12:18 teebes joined #salt
12:19 sssslang oc: besides setup infrastructer, what else can salt do?
12:20 oc anything you want it to, but you shouldn't want it to ;)
12:20 shiin orchestrate clusters.
12:21 felixhummel joined #salt
12:25 sssslang Thanks guys, I decide to give salt a try.
12:27 oc sssslang: but it all boils down to your own preferences; if you like the cfengine syntax (and have plenty of time) it'll work fine too. If you're a developer and want to get things working and deterministic (and happen to know python) ansible/salt is the way to go ;)
12:29 sssslang oc: that's the point. I spend a lot of time reading the cfengine's manual. My team speak python, so I think salt may be a better choice.
12:35 Katafalkas sorry to repeat my question but: I am using parted to make a file system. I got warning that you are planing to remove parted in later releases. Is anyone working on anything yet ?
12:37 joehh sssslang: which os are you using salt on?
12:37 joehh ie how are the run time dependencies managed?
12:37 sssslang joehh: mainly rhel4, centos6 and ubuntu12.10
12:37 joehh or not?
12:38 joehh ok, I think I see now - from memory, cfengine is pretty much a smallish binary compiled from c
12:39 joehh in contrast, salt is python with deps on zeromq, python stuff, crypto stuff etc...
12:39 joehh Is that the reasoning behind the concern re dependencies?
12:41 sssslang joehh: It cost me 2 hours compiling openssl, python, M2Crypto, zeromq, etc.
12:41 juicer2 joined #salt
12:41 sssslang joehh: I can compile cfengine once, and copy to everywhere.
12:44 oc sssslang: hm; what system do you run it on? most distros bundle fairly stable versions of those libraries; otherwise -- perhaps try the virtualenv path; or build custom packages with i.e. FPM
12:45 oc rhel4, centos6 and ubuntu12 should be pretty simple to integrate with
12:45 oc use a conditional map for rpm vs. deb package names
12:46 sssslang oc: our system cannot use modern repos.
12:47 oc well, build packages for the target architectures using i.e. fpm then, it can build both rpm and deb
12:49 sssslang oc: is there a universal package manager?
12:50 [diecast] joined #salt
12:53 toddejohnson joined #salt
12:53 xmj packaged
12:54 Pate_ joined #salt
12:55 sssslang xmj: is it a package manager/
12:56 Furao joined #salt
12:58 oz_akan_ joined #salt
12:58 xmj nevermind
12:59 Jahkeup joined #salt
13:00 TheCodeAssassin http://unix.stackexchange.com/questions/23675/why-isnt-there-a-truly-unified-package-manager-for-linux
13:01 brianhicks joined #salt
13:09 Gifflen joined #salt
13:11 vbabiy joined #salt
13:13 sssslang TheCodeAssassin: It is for binary package. but I mean install from source. Some of our hosts are very old, and cannot use newer binary package, so we need to compile from source.
13:14 sssslang If there is a tool, use a universal compile description language(other than ebuild for gentoo, dh_ for debian), we can write once, compile package everywhere.
13:15 y0j joined #salt
13:15 pdayton joined #salt
13:20 mwillhite what does `salt-call highstate` with the --local flag do?
13:20 toastedpenguin joined #salt
13:20 mwillhite s/highstate/state.highstate
13:23 mwillhite I mean…I see the man pages, but I don't really understand:
13:23 TheCodeAssassin run salt locally
13:23 mwillhite Run salt-call locally, as if there was no master running.
13:23 TheCodeAssassin yes
13:23 unicoletti left #salt
13:23 TheCodeAssassin If you have a masterless minion
13:23 mwillhite I'm running it on a masterless setup…so would that be doing the same thing?
13:23 TheCodeAssassin yes
13:23 mwillhite ok
13:23 mwillhite another question…can I run a specific init.sls with this command?
13:24 mwillhite I'm debugging some things and I want to build this up a piece at a time if I can
13:24 mwillhite but I'm running into errors with services already being started
13:24 TheCodeAssassin you want to only run 1 sls?
13:25 TheCodeAssassin can't you comment out the rest in your top.sls?
13:25 Kholloway joined #salt
13:26 mwillhite yes…I just thought it would be cool if I could uncomment as I'm in the machine and run the states as I uncomment them
13:26 mwillhite rather than running everything over and over and again until I get my full box back
13:27 mwillhite maybe I'll just make this service a little smarter so it doesn't run if its already running :)
13:30 Ryan_Lane joined #salt
13:31 deedubs joined #salt
13:34 alunduil joined #salt
13:35 toastedpenguin left #salt
13:38 bhosmer_ joined #salt
13:41 Gifflen joined #salt
13:41 tuxIO joined #salt
13:42 littleidea joined #salt
13:43 Katafalkas sorry to repeat my question but: I am using parted to make a file system. I got warning that you are planing to remove parted in later releases. Is anyone working on anything yet ?
13:43 faldridge joined #salt
13:43 oz_akan_ joined #salt
13:44 mwillhite I have several apps that I'm setting up with this machine (I'm building a distributable dev environment) and I want to build up each app (pull from git, bundle gems, create db etc) as a specific user. Is there anyway to apply a user to the entire init.sls rather than setting runas on each state in the sls?
13:44 m_george left #salt
13:45 Gifflen joined #salt
13:53 TheCodeAssassin is that a problem four you mwillhite
13:53 TheCodeAssassin ?
13:54 TheCodeAssassin why don't you do {% set user = 'test' %} and use - runas: {{ test }} ?
13:54 TheCodeAssassin i mean user
13:54 TheCodeAssassin lol
13:54 mwillhite TheCodeAssassin some of these apps have a bunch of stanzas…so it would be nice if I could say something like…everything in this init.sls runas user
13:55 TheCodeAssassin dunno realy
13:55 TheCodeAssassin *really
13:55 TheCodeAssassin only use runas for git repo's
13:56 littleidea joined #salt
13:56 mwillhite yeah I can't seem to find anything
13:58 imaginarysteve joined #salt
14:01 Jahkeup joined #salt
14:01 mannyt joined #salt
14:05 lineman60 joined #salt
14:06 xinkeT joined #salt
14:10 Katafalkas When writing a runners, is there some way of getting target string inside the script ? lets say "salt cookies* runner_file.run_func" how do I get "cookies*" inside the script ?
14:11 mattmtl joined #salt
14:13 swa is there a way to retrieve the name of a role.. I'd like something like this in the motd : these roles are applied to this server : sshd, snmpd
14:16 TheCodeAssassin grains.get role ?
14:17 TheCodeAssassin does this make sense swa?
14:17 swa TheCodeAssassin: should :) didn't know that one
14:17 swa thx
14:17 Furao swa: sshd is a role for you? for me it's more: frontend, backend, bus, logging, etc
14:17 TheCodeAssassin check this: http://docs.saltstack.com/ref/modules/all/salt.modules.grains.html
14:17 TheCodeAssassin it will make sense to you i'm sure
14:18 Furao pillar can also be a good place for roles, if it's if defined centrally
14:18 swa Furao: i mean state, but as my colleagues are french speakers, I say "role" instead of states.. it makes more sense to them
14:18 Furao je parle français et je n'utilise pas le mot rôle pour des states :P
14:19 Furao role is a concept on top of the states
14:19 swa j'essaie desesperement d'inculquer le vocabulaire de salt à mes collègues (dans la limite de mes connaissances), mais ils ne s'impliquent pas vraiment
14:19 Furao but it's not a core concept of salt, it's a good practice
14:20 gordonm joined #salt
14:20 swa i'll stick with salt vocabulary actually, it's the only way for my colleagues to grasp it :)
14:20 TheCodeAssassin swa:  do you add roles on the minion? what i do is something like this: https://gist.github.com/thecodeassassin/596997363f181a7db06d
14:21 swa TheCodeAssassin: there's a misunderstanding on the "role" term in my motd.. i meant states
14:21 TheCodeAssassin oooh
14:22 TheCodeAssassin haha
14:22 swa i'm not that far into salt yet
14:22 swa still need to learn python :)
14:24 hazzadous joined #salt
14:24 vbabiy joined #salt
14:25 tuxIO_ joined #salt
14:26 jinrou joined #salt
14:27 TheCodeAssassin swa: thats a good idea
14:27 mwillhite TheCodeAssassin I found how to call individual states: salt-call state.sls pep8
14:27 TheCodeAssassin whats pep8?
14:28 alexandrel python enhancement proposals
14:28 alexandrel like RFCs, but python centric.
14:29 alexandrel oh wait, was that rethorical?
14:30 vbabiy joined #salt
14:30 opapo joined #salt
14:32 StDiluted joined #salt
14:32 jbunting joined #salt
14:35 JaredR joined #salt
14:38 mgw joined #salt
14:38 mwillhite its just an example
14:39 mwillhite …from the sls docs
14:39 cnelsonsic joined #salt
14:40 TheCodeAssassin n1
14:40 mwillhite I'm creating a db, I'd like to use the vagrant user (I'm using salty-vagrant), but my database only shows up under `sudo mysql`, rather than `mysql`…I'm not sure how to fix that as runas: vagrant under mysql_database.present doesn't seem to make a difference
14:40 dthom91 joined #salt
14:42 bhosmer joined #salt
14:43 bhosmer_ joined #salt
14:44 tyler-baker joined #salt
14:45 anteaya joined #salt
14:45 mgw joined #salt
14:46 mwillhite ah…runas is not an option
14:46 mwillhite maybe that should be handled in grants?
14:47 ronc joined #salt
14:51 abe_music joined #salt
14:51 ggoZ joined #salt
14:52 lazyguru joined #salt
14:55 Brew_away joined #salt
14:59 kaptk2 joined #salt
15:02 abele joined #salt
15:03 ckao joined #salt
15:12 chrisgilmerproj joined #salt
15:18 forrest joined #salt
15:21 jdenning joined #salt
15:26 UtahDave joined #salt
15:26 organization-woe joined #salt
15:26 mgw joined #salt
15:26 organization-woe greetings
15:26 organization-woe I've got an open question about best practices
15:27 nbari joined #salt
15:27 organization-woe how should I organize my salt directories?
15:27 organization-woe right now I've got salt/{states,pillars,configs}
15:27 organization-woe for a particular project
15:27 nbari cedwards: does your minions lost conectivity after couple of hours ?
15:28 organization-woe But now I want to add environemnts
15:28 cedwards nbari: no
15:28 nbari I have both master/minions runing on FreebSD but notice that minions lost conectivity randomly
15:28 cedwards nbari: in my setup i have masters on the host and minions in jails
15:28 nbari I have to restart the minions in order to connect
15:28 nbari oh ok
15:29 cedwards nbari: i wonder if it's because FreeBSD still uses zeromq 2.x
15:29 nbari inded is 2.x
15:29 forrest organization-woe, so you're adding the environments to the top file?
15:29 cedwards 3.x is in the -devel package, but i've never really worked on tying tho two together
15:30 nbari editing the minion config 'tcp_keepalive_intvl: 75 ' help me to keep the minion arround 12 hours
15:30 organization-woe yeah forrest
15:30 organization-woe Targeting them is fine, but they're all in different directories correct?
15:31 organization-woe would it be as simple as salt/{env1,env2}/{states,pillars,configs} ?
15:31 forrest yea
15:31 forrest Did you look at http://docs.saltstack.com/ref/states/top.html
15:31 organization-woe indeed
15:31 forrest if you scroll down about 50% of the way, it shows using multiple environments like what you're doing
15:32 organization-woe but it's not clear about how it loads files for different environments
15:32 teskew joined #salt
15:32 organization-woe I understand how to target different environments
15:32 forrest ok
15:32 forrest so when you set the file_roots
15:33 organization-woe but what if I want to share some files across environments but not others
15:33 forrest oh I see.
15:33 organization-woe Like, most of my state tree will be the same
15:33 organization-woe there are just a few states that I want to have different environments
15:33 forrest If it were me I'd just copy pasta them between
15:33 forrest but I don't know what the official best practice is
15:33 organization-woe :/
15:33 forrest It's not much work to copy them over, just text files :P
15:33 organization-woe I was thinking about symlinks
15:33 forrest why
15:33 organization-woe to make it slightly more manageable
15:33 forrest so think of it like this though
15:34 forrest you have your base apache config right, it's the same across all 3 envs
15:34 organization-woe Copying the files means that every time one gets updated, the developer has to remember to update the other
15:34 organization-woe never duplicate files if you don't have to
15:34 forrest so what happens when your developers need a change to base apache in dev to test something? Now your symlinks are screwed
15:34 nbari zmp should be on master or also in minions ?
15:34 cedwards nbari: maybe we should do some testing toward the 3.x series
15:34 forrest Usually what I do in that case organization-woe is use a repository
15:34 forrest because if you have to test something in dev, you don't want it pushed to prod
15:34 organization-woe wait what?
15:34 nbari I am talking about version 3 the one in theory fixe the connectivity issues
15:35 forrest for some poor dude to run and screw the environment
15:35 organization-woe We have tests to prevent that kind of thing
15:35 forrest so I'd have a git repo
15:35 nbari I will try with port zmq-dev
15:35 organization-woe but it's worse if you change something in dev and it doesn't get to production
15:35 forrest and then the dev, test, and prod branches
15:35 organization-woe because they didn't make the change in two places
15:35 forrest then you push to dev, if everything looks good, onwards to test, then onwards to prod.
15:37 organization-woe alright so how about this
15:37 organization-woe instead of asking about best practices, how does salt actually handle multiple states in different environments
15:38 organization-woe maybe it's best iwth an example
15:39 bhosmer_ joined #salt
15:41 forrest actually organization-woe
15:41 forrest look at this line
15:41 forrest Then only place state trees in the dev, qa and prod environments, leaving the base environment open for generic file transfers. Then the top.sls file would look something like this:
15:41 forrest I don't know what the term 'generic file transfer' really refers to
15:42 organization-woe forrest: right.  That's what I'm confused about
15:42 organization-woe it's just not clear how environments work at all
15:43 organization-woe is there precedence? how does salt decide which environment to use based upon the environment
15:43 organization-woe for example
15:43 organization-woe http://pastebin.com/F4PD8vGJ
15:43 organization-woe which does salt pick?
15:44 littleidea joined #salt
15:46 forrest http://docs.saltstack.com/ref/states/top.html#how-top-files-are-compiled
15:47 forrest so in your example it just uses the base web.sls
15:47 malinoff organization-woe: Read this http://docs.saltstack.com/ref/states/top.html#how-top-files-are-compiled - when you run highstate salt discovers automatically what env it should use
15:47 malinoff Otherwise, you can explicitly specify env='myenv' in state.sls command
15:48 montuori joined #salt
15:48 racooper joined #salt
15:48 gildegoma joined #salt
15:49 Lue_4911 joined #salt
15:50 Pakdel joined #salt
15:51 organization-woe Right, so I don't have multiple top files.  Perhaps it's just better to target things in the minion config
15:51 organization-woe since that seems like the most sane way to be sure what you're actually getting
15:52 mwillhite joined #salt
15:53 organization-woe Can you target multiple environments?
15:53 organization-woe here, let me make another example
15:54 wolfpack_ joined #salt
15:55 devinus joined #salt
15:55 organization-woe Would this work? http://pastebin.com/FBLHibte
15:55 organization-woe or would it only take the base env and ignore the reset?
15:55 organization-woe rest*
16:03 UtahDave organization-woe: The minion will take each SLS file that it matches and compile them all into one big dict
16:04 UtahDave organization-woe: remember that each environment has its own directory  so "application.sls" might be a different file for dev and prod environments in your example
16:05 abe_music What would cause salt to never finish a highstate? The minion log shows nothing, top shows no substantial activity, and it's been running for 20 minutes….any way to debug into this to see what's going on?
16:06 organization-woe UtahDave: right.  So you can target multiple environments then
16:06 organization-woe Thanks!
16:06 organization-woe ha, derp.  I just realized that I targeted 'web*prod*' for dev and prod
16:06 organization-woe that first one was supposed to be dev
16:07 organization-woe was just an example anyways
16:07 mgw joined #salt
16:07 UtahDave organization-woe: I figured that's what you meant to have there.
16:09 middleman_ joined #salt
16:10 billh joined #salt
16:11 KyleG joined #salt
16:11 KyleG joined #salt
16:12 troyready joined #salt
16:13 abe_music salt returned, but the job is still running…not sure what to do beyond killing that job and trying again
16:13 forrest abe_music, did you run it in debug mode?
16:13 forrest when you ran the highstate?
16:13 forrest and do you run any commands?
16:14 abe_music forrest, yes i did….i actually ran a state.sls to test a particular state i'm working on, but should be the same
16:14 UtahDave abe_music: you can check to see the status of the job
16:14 forrest because we noticed a few days ago that if you set the timeout on a command to say, 120 seconds, and it only takes 5 seconds to run, it will just hang out while the timeout finishes for some reaosn.
16:15 abe_music UtahDave, forrest: i re-ran the same job, but it complained that it was already running….so doing saltutil.find_job 20130904153833232590 shows that it's still going, but no activity on the minion
16:16 forrest abe_music, what does your state do?
16:16 UtahDave abe_music: can you pastebin your sls?  Is it possible an command you've set it to execute is waiting for input?
16:16 aptiko_ joined #salt
16:18 bitz joined #salt
16:20 ProT-0-TypE joined #salt
16:20 jalbretsen joined #salt
16:20 abe_music forrest, UtahDave: sure, i'll give a gist in a sec
16:21 ProT-0-TypE joined #salt
16:21 ProT-0-TypE joined #salt
16:22 ProT-0-TypE joined #salt
16:23 jrpurdy left #salt
16:23 ProT-0-TypE joined #salt
16:23 abe_music UtahDave, forrest : https://gist.github.com/abemusic/6439387
16:24 abe_music i'm attempting to get my CDH4 hadoop states running in CentOS after I got them working in Ubuntu….maybe this is something to do with CentOS?
16:24 abele joined #salt
16:26 forrest abe_music, can you see if on the minion that hdfs process is still running?
16:26 forrest or if any of the process that run from these commands are hung up
16:27 abe_music sure
16:27 middleman_ joined #salt
16:28 abe_music forrest: no hadoop related packages have been installed, thus no processes running, no related users, etc...
16:28 abe_music checked my yum log and nothing relevant in there either except for the RPMs from salt bootstrap
16:29 forrest interesting, so we aren't even getting that far. Did the repo get created?
16:29 abe_music nothing in my yum cache, so the packages haven't been downloaded either
16:29 abe_music the repo did get created, but i actually ran "state.sls cdh4.repo" first to make sure my yum repo would get created
16:29 abe_music then proceeded to run the namenode state
16:32 mwillhite joined #salt
16:33 forrest so what does this cdh4_refresh_db do?
16:34 abe_music i think for ubuntu it runs apt-get update, while on redhat/centos it runs a yum clean all
16:34 abe_music just wanted to keep them both the same regardless of OS
16:35 forrest yea I understand
16:35 abe_music so, i just fired up a couple more machines using Ubuntu 12.04 and it went through just fine….hmm
16:35 forrest so the issue only happens on the cent machine?
16:35 forrest are there any rpm/yum processes running?
16:35 abe_music apparently…i'll bring them down and try again to see if they same issue occurs
16:35 abe_music let me check
16:36 forrest ok
16:36 abe_music no rpm or yum process on the centos box
16:36 forrest weird
16:36 forrest if you can catch it, maybe try to stacktrace the minion process when you start the run
16:36 forrest then we can see what system calls it is hanging up on
16:37 abe_music i can try that now if you think that'll help…or should i do it on a clean run?
16:37 forrest try it real quick
16:38 abe_music k
16:39 aptiko_ left #salt
16:40 abe_music forrest, hmm…apparently a kill -3 is not what you mean for me to run :)
16:41 abe_music we do a lot of java and that's how we get our thread dumps to see what's going on….that actually killed the python process
16:44 dthom91 joined #salt
16:47 tyler-baker joined #salt
16:47 tyler-baker joined #salt
16:48 forrest oh no abe
16:48 forrest do strace -p {pid} if you just want to hop on
16:49 forrest if you want to output to a file do strace -p {pid} -o /root/strace.txt or something
16:49 forrest or you could get really fancy
16:49 forrest and do strace -e trace=open,read -p {pid} -o /root/strace.txt
16:50 forrest then you'll get the open and read system calls
16:50 forrest so that MIGHT return something worthwhile
16:50 ronc joined #salt
16:51 forrest so next time :P
16:51 robertkeizer joined #salt
16:51 forrest try to hop on the process and see what's occurring. If you want to trace the actual process on the master do strace -o /root/strace_master.txt {command here}
16:53 abe_music forrest, gotcha…alright, i just brought up a new machine so we'll see what happens…thanks
16:54 forrest yea np, I've got a meeting in a few minutes, if you get any info back from that let me know
16:54 abe_music will do
16:54 forrest cool
16:56 abe_music forrest: this time i got an error :-/
16:56 abe_music https://gist.github.com/abemusic/6439387#file-state-sls-cdh4-hadoop-namenode-error
16:57 forrest better than the system hanging :P
16:57 abe_music definitely haha
16:57 forrest intereting
16:57 forrest *interesting
16:57 abe_music the command i ran: salt -ldebug centos-nn-abe-1 state.sls cdh4.hadoop.namenode
16:57 forrest can you see what happens when you manually try to install that? I don't see why it gives the broken pipe error
16:57 forrest anyways meeting time, hooray.
16:58 abe_music will do, ttys
16:58 abe_music thanks for the help, btw
16:59 philip741 joined #salt
17:00 philip741 I have used puppet mostly in the past and I am really liking saltstack. I have probably a stupid question but how do you maintain a file like in puppet.
17:02 lazyguru joined #salt
17:03 Ryan_Lane joined #salt
17:04 philip741 Like the puppet daemon checking every so many defined minutes to see if a file has changed content. I know I could set up a cron job and do like salt '*' state.sls or is there other ways to do it?
17:05 StDiluted philip741: http://docs.saltstack.com/topics/jobs/schedule.html
17:05 philip741 ah perfect thanks
17:06 dthom91 joined #salt
17:07 spacedentist joined #salt
17:07 Jahkeup joined #salt
17:08 duck_ joined #salt
17:09 Jahkeup_ joined #salt
17:10 jpeach joined #salt
17:10 rberger joined #salt
17:10 faldridge joined #salt
17:11 karlgrz joined #salt
17:12 karlgrz hey all, I'm having trouble installing rails using rvm as the rvm user on ubuntu 13.04. Anyone around to help?
17:12 karlgrz To clarify, I mean installing it using salt
17:12 jpeach in a jinja template, should I be able to use a nested dictionary for pillar data?
17:13 jpeach as shown in http://apaste.info/35Pp
17:13 jpeach I can get the right pillar data with a manuaal pillar.get, but the jinja template always renders the default value rather than the pillar data value
17:15 karlgrz It keeps asking me for rvm password
17:16 karlgrz I am running salt-call --local state.highstate -l debug
17:16 karlgrz would be happy to provide the states I am using as well
17:17 karlgrz it appears that rvm and ruby get installed properly
17:17 karlgrz but the following comman:
17:17 mwillhite joined #salt
17:17 karlgrz err, following state, is failing:
17:18 karlgrz http://pastebin.com/QNiEVkiU
17:19 devinus joined #salt
17:20 aat joined #salt
17:21 jpeach ok, I see what happens:
17:21 jpeach this works: salt['pillar.get']('trafficserver:config:proxy.config.http.cache.allow_empty_doc')
17:21 dave_den jpeach: in your state,s pilar is a python dict
17:21 karlgrz here's ruby/init.sls: http://pastebin.com/gpXc1LU3
17:21 jpeach this doesn't: pillar.get('trafficserver:config:proxy.config.http.cache.allow_empty_doc')
17:21 dave_den karlgrz: are you running salt-call as root?
17:22 karlgrz and rvm/init.sls: http://pastebin.com/3SnC0Qqm
17:22 karlgrz yes
17:22 karlgrz should i not be?
17:23 dave_den karlgrz: show the output where it asks for your rvm password
17:23 karlgrz k, uno momento please, i need to restart that instance
17:23 dave_den jpeach: when you use salt['pillar.get'], you are calling the pillar.get module like you do with salt-call. when you do pillar.get, you are invoking python's .get() on a dict
17:24 jpeach dave_den: oh, that explains the behaviour
17:24 jpeach thanks
17:25 jinrou joined #salt
17:32 ml_1 joined #salt
17:32 ml_1 joined #salt
17:32 karlgrz ok, here's the output:
17:32 karlgrz http://pastebin.com/EUDG4agk
17:38 karlgrz i can provide more output if necessary
17:39 cjh joined #salt
17:39 cjh anyone running salt in lxc containers?
17:41 lemao joined #salt
17:44 karlgrz looking at the output it appears that the rvm user is correctly being used to actually try to install the package, which is what I want (I think)
17:44 karlgrz I'd rather not have to add rvm to the sudoers list
17:47 Katafalkas joined #salt
17:49 karlgrz when I run the command /usr/local/rvm/bin/rvm install ruby-1.9.3 as rvm, the error I get mentions adding rvm to the sudoers file
17:51 EWDurbin joined #salt
17:52 EWDurbin howdy
17:52 EWDurbin so i have this silly little state
17:52 EWDurbin https://gist.github.com/ewdurbin/e6b3fab473cf22b230f2
17:53 EWDurbin to generate an ssl cert for LB -> backend server https
17:53 EWDurbin so i just need it to exist.
17:53 EWDurbin and now i'm nitpicking, because it executes every time and instead of all green
17:53 EWDurbin i get 99% green and one blue
17:54 EWDurbin any tips to make this only execute if the cert/key do not already exist?
17:54 honestly is there a tutorial or something anywhere the gives the low-down on salt-flavoured jinja?
17:54 UtahDave cjh: I know that quite a few people are using salt in lxc containers
17:55 toastedpenguin joined #salt
17:55 dave_den EWDurbin: add an 'unless' to that state that tests for the filename
17:55 EWDurbin dave_den: tried that, it didn't seem to care
17:55 EWDurbin i don't see anything in the module.run state code that accepts an unless either
17:56 EWDurbin salt-0.16.0-1.el6.noarch
17:56 EWDurbin salt-minion-0.16.0-1.el6.noarch
17:56 EWDurbin btw
17:56 UtahDave EWDurbin: try module.wait   and have it wait on hte pkg: pyOpenSSL
17:57 dave_den karlgrz: i'd have to check further, but it seems like a bug. sorry, gotta run
17:58 EWDurbin @UtahDave: i'm not sure I see what that accomplishes.
17:58 sixninetynine left #salt
17:59 forrest abe_music, any luck with that error?
18:01 cjh UtahDave: ok awesome.  I'd like to start up a lxc container and have a salt minion install/configure apache or nginx
18:01 cjh shouldn't be too hard i think to adapt my current sls files to lxc
18:02 Ryan_Lane joined #salt
18:03 Ryan_Lane joined #salt
18:03 TheCodeAssassin joined #salt
18:03 druonysus joined #salt
18:03 druonysus joined #salt
18:05 karlgrz k, thanks
18:05 dthom91 joined #salt
18:06 TheCodeAssassin hey utahcon
18:06 TheCodeAssassin UtahDave:
18:10 Psy-Q left #salt
18:11 stevejohnson joined #salt
18:13 abele joined #salt
18:14 karlgrz left #salt
18:23 pdayton joined #salt
18:24 faldridge joined #salt
18:28 littleidea joined #salt
18:28 Lue_4911 joined #salt
18:30 naemon joined #salt
18:32 sfello joined #salt
18:32 Teknix joined #salt
18:33 Lue_4911_ joined #salt
18:37 robawt aloha again!
18:37 robawt what is the best practice for updating config files served via salt?
18:37 robawt update it and highstate?
18:37 robawt or only  run the state for that saltstate?
18:40 robawt based on the not so basic tutorial they highstate the machine to update index.html
18:40 robawt http://docs.saltstack.com/topics/tutorials/states_pt2.html
18:40 robawt but that seems prone to being clobbered
18:46 Katafalkas joined #salt
18:49 faldridge joined #salt
18:54 Lue_4911 joined #salt
18:59 JaredR joined #salt
19:02 naemon any ideas why this line 'unless: mongo local --eval "printjson(rs.conf())" | grep {{ grains['mongodb_replicaset'] }}' is turning into this on the minion 'Executing command 'mongo local --eval "printjson(rs.conf())" | grep [\'localmongoreplica\']' in directory '/tmp''
19:03 naemon specifically, why it's including the '[]'  brackets, which is causing the logic to fail
19:03 EWDurbin left #salt
19:03 naemon grain is this on /etc/salt/minion
19:03 naemon mongodb_replicaset:
19:03 naemon - localmongoreplica
19:04 naemon joined #salt
19:09 naemon robawt:  There's always 'test=True' to the highstate...
19:10 dthom91 joined #salt
19:10 p3rror joined #salt
19:11 ahale looks like your {{ grain }} is a list? what does 'mongodb_replicaset: localmongoreplica' do ?
19:12 robawt naemon: good point
19:14 naemon ahale:  Thanks for the kick in the head… I see now.  this doesn't need to be a list...
19:15 ahale np :)
19:18 jpeach joined #salt
19:19 aat joined #salt
19:21 bhosmer joined #salt
19:22 robawt ok another annoying question
19:23 robawt is "watch" file a two way street?  if i modify it on the minion side it will restart the service identified with it, will it do the same if i modify it on the master's end?
19:23 backjlack joined #salt
19:23 jinrou joined #salt
19:24 UtahDave robawt: I'm not sure what you mean.
19:25 robawt UtahDave: i've got a httpd.config let's say
19:25 robawt so i highstate a minion and now it has my httpd.config
19:25 robawt if I want to make a change on that file, what is the 'salt way' to push the changes to the minion?
19:26 UtahDave robawt: yeah, always make your changes on the master's file server
19:27 robawt UtahDave: so that part I agree with
19:27 robawt so how do I do it most effectively with salt?
19:27 robawt just a highstate?
19:27 robawt or will "watching" a file in a saltstate look for changes on the master?
19:27 robawt I'm writing out some documentation so I want to be as close to the "salt preferred" method as possible
19:27 UtahDave no, "watch" only happens when you execute a highstate.
19:28 robawt ah interesting ok
19:28 UtahDave Salt never does anything unless you tell it to, or unless you schedule it to
19:28 robawt so i call highstate on the minion in question, it'll go get the modified file off the master
19:28 UtahDave yep
19:31 honestly to whom it may concern: https://github.com/saltstack-formulas/users-formula/pull/3
19:33 UtahDave honestly: did you look at the pillar example in the repo?
19:33 honestly yes
19:33 honestly oh, did that change?
19:34 robawt UtahDave: you're the man, thank you Sir
19:34 * robawt highfives UtahDave
19:34 honestly just looked over it again - still think my change makes sense
19:34 honestly pillar example defines login shell per user
19:35 UtahDave honestly: Ah, true and at the top of the file it's setting that pillar user to "user"
19:35 honestly yep
19:35 UtahDave so, indeed, you're correct.
19:35 * honestly bows
19:35 zooz joined #salt
19:35 UtahDave ok, sorry. I apparently needed to talk it through out loud first.  :)
19:36 honestly since I got you here - are you aware of a document somewhere that explains the down-low of salt-flavoured jinja?
19:38 UtahDave honestly: have you seen this yet?  http://docs.saltstack.com/ref/renderers/all/salt.renderers.jinja.html
19:38 honestly might have skimmed it at some point weeks ago, but functionally, no.
19:38 honestly thanks!
19:39 UtahDave you're welcome!  Also, the regular jinja docs can be very helpful, too
19:40 littleidea joined #salt
19:40 pt|Zool joined #salt
19:40 kevinbrolly joined #salt
19:40 honestly hmm, I'm too tired to process that.
19:40 clone1018 joined #salt
19:40 * honestly bookmarks
19:41 defunctzombie_zz joined #salt
19:41 kamal_ joined #salt
19:41 robawt1 joined #salt
19:41 joonas joined #salt
19:42 utahcon joined #salt
19:43 robawt joined #salt
19:43 hjubal joined #salt
19:44 m_george|away joined #salt
19:45 robawt UtahDave: i have to apologize, I was looking at this from the wrong view
19:45 robawt file.managed does exactly what I was thinking
19:46 UtahDave ah, ok.  good!
19:47 robawt loss of brain I guess
19:47 robawt :)
19:47 robawt UtahDave: have you guys considered online training?
19:48 robawt put a couple videos out (other than saltair) and charge people $$$?
19:49 UtahDave I think we've discussed it a few times, but I don't know where those discussions went.  :)
19:50 robawt cool
19:50 robawt I'll just wait for Salt Conf :)
19:50 forrest hah
19:51 auser joined #salt
19:51 UtahDave :)
19:54 m_george left #salt
19:54 * xmj predicts a Pepper Gate.
19:59 forrest a pepper gate?
20:02 forrest oh the thing at UC Davis
20:05 KyleG1 joined #salt
20:07 KyleG joined #salt
20:07 Lue_4911 joined #salt
20:08 SpX joined #salt
20:08 cbloss Can files be stored in Pillars as well? or only in states?
20:08 robawt cbloss: what kind of file would you need to store in pillar?
20:09 robawt pillar is more like a collection of data, vs a file
20:09 cbloss robawt: ssl certificates
20:09 robawt someone chime in if i'm screwing this up
20:09 honestly UtahDave: https://github.com/saltstack-formulas/users-formula/blob/master/users/init.sls#L12 - wouldn't it be more "idiomatic" to have group.present there and strike the "- present" two lines down?
20:10 forrest hey cbloss, check out this issue from a while back
20:10 forrest https://github.com/saltstack/salt/issues/1543
20:11 cbloss forrest: thanks, looking at it now
20:11 forrest That issue has a lot of workarounds, and UtahDave opened https://github.com/saltstack/salt/issues/3790 to look at what you're trying to do
20:11 forrest cool
20:11 cbloss Yeah, Pillars make a lot of sense to use for SSL/SSH keys
20:12 robertkeizer joined #salt
20:12 akoumjian This is an awesome deck by Peter of LL: https://speakerdeck.com/ipmb/getting-started-with-salt
20:15 forrest nice akoumjian
20:18 UtahDave honestly: It's fine how you have it.  group.present is just a shortcut.
20:19 honestly UtahDave: okay, I'll just sneak it into the next merge request then :P
20:20 honestly that's not from me originally btw
20:21 honestly I didn't create users-formula
20:21 UtahDave ok, gotcha
20:21 honestly it wouldn't be like that if I had
20:21 TheCodeAssassin joined #salt
20:22 rberger joined #salt
20:24 honestly I'm currently working out a special twist on the users-formula that you might want to include in salt-formulas
20:24 honestly stay tuned!
20:24 whit joined #salt
20:29 stevejohnson left #salt
20:31 littleidea joined #salt
20:32 ingwaem joined #salt
20:35 ingwaem greetings everyone
20:38 UtahDave hey, ingwaem!
20:39 ingwaem hows things Dave ?
20:41 ingwaem quick question about SLS files and repos…I understand that the first line is usually the address of the repo…is there any way I can replace this with a name first, and then define the host? Reason being is I have 2 branches in the same repo that I want to pull into 2 separate directories but am getting messages from salt saying there is a conflict on the id's if i execute them together…if I execute 1 straight after the other it associates the job of th
20:41 ingwaem first execution with the 2nd so only 1 job actually kicks off
20:45 UtahDave ingwaem: yeah, the first item is called the ID Declaration.
20:46 UtahDave Change each of those ID declarations to something arbitrary like   repo1  and coolrepo2
20:46 faldridge joined #salt
20:46 UtahDave then use   - name: path/to/repo
20:46 UtahDave underneath each one to give the actual address to the repo
20:46 ingwaem sweet I'll give that a bash :)
20:49 UtahDave cool, let me know how that goes
20:50 Ryan_Lane joined #salt
20:51 ingwaem UtahDave: will do :)
20:51 Ryan_Lane joined #salt
20:52 hannesstruss joined #salt
20:54 juicer2 Hi all,
20:54 devinus joined #salt
20:54 ingwaem greetings
20:54 juicer2 trying to install sun java jre 1.7.0.25 on centos 6 with salt 0.16.3,
20:54 juicer2 getting http://pastebin.com/MbiPhxin
20:55 juicer2 I've tried couple diff package names, jre, jre-1.7.0_25_fcs, etc
20:56 SunSparc joined #salt
20:58 ingwaem hmm, looking at some documentation you could use yum install java7…I wonder calling the package just java7  would work
20:59 ingwaem JRE comes with the JDK so would install both
20:59 hannesstruss Hi, I'm having a problem with Pillar not using one of my sls files. This is my /srv/pillar/top.sls: https://gist.github.com/hannesstruss/8ce2382ffd797a3644ea
20:59 juicer2 ingwaem: hmm, that'd work, installs a bunch of x libs I don't really need is the only caveat
21:00 VertigoRay Yah!  First pull to salstack/salt accepted. :D  I guess it's official ... I'm a part of the community.
21:00 ingwaem Whoo VertigoRay welcome!
21:00 hannesstruss say, the minion is named 'web1-staging', only the settings-staging.sls is used, not the settings-global.sls
21:00 ingwaem juicer2: yea, it's the only caveat…I guess you could try downloading the bin file, then running from that via command line or rpm but it would be a longer route I think .
21:01 ingwaem juicer2: I could be wrong so please don't hold my word as gospel…if someone else has experience with JRE please do chime in with your findings
21:01 hannesstruss salt-call -l all pillar.items hints that the file is parsed (and the correct json is produced), but the values don't show up in pillar.items. Any pointers?
21:03 nliadm anyone have any luck parsing the json out from salt?
21:04 juicer2 ingwaem: yep, thanks
21:04 abe_music|away forrest: got sidetracked earlier, but i eventually ran "yum install hadoop-hdfs-namenode", it presented the list of dependencies, I selected yes, and it installed fine
21:05 ingwaem nliadm: yea I always use json … where are you hitting a snag?
21:06 nliadm it's not valid json to just have multiple objects
21:07 forrest abe_music, aww man that sucks. What happens now when you try to run the state?
21:08 abe_music forrest: hah, same issue, but when installing hadoop-0.20-mapreduce-jobtracker package
21:08 UtahDave hannesstruss: have you run  salt \* saltutil.refresh_db
21:08 ingwaem nliadm: to get json output use --out=json as a parameter of your command. However you might also want to use --static. As quoted in the documentation "If using --out=json, you will probably want --static as well. Without the static option, you will get a JSON string for each minion. This is due to using an iterative outputter. So if you want to feed it to a JSON parser, use --static as well."
21:09 nliadm aaaah
21:09 abe_music forrest: i'm whittling down my state to see if there's something going on in setting up the repo that might be causing this
21:09 nliadm thanks, ingwaem
21:09 forrest yea I was gonna suggest to start at the very beginning, and go through each item to see where it fails (or if it does)
21:09 ingwaem nliadm: I don't usually hit this problem because I use the Salt-api and I have my php talking to it…get everything in json by default
21:09 ingwaem nliadm: yw :) hope it helps
21:10 Ryan_Lane joined #salt
21:10 hannesstruss UtahDave: I'm running masterless, salt-call saltutil.refresh_db gives me a "Function saltutil.refresh_db is not available"
21:10 forrest abe_music, it seems odd to me the that the installation would fail on that. You know what also might be worth a shot, is listing those deps in your sls, and then having that package require them, JUST to test it out.
21:11 forrest and have you confirmed that the gpg key is getting installed properly first when running the SLS?
21:11 UtahDave hannesstruss: sorry,  saltutil.refresh_pillar
21:12 abe_music forrest: re: gpg, yeah, if i don't it will warn and fail unless i override the gpg check
21:12 hannesstruss UtahDave:
21:12 forrest Hmm.
21:12 hannesstruss doesn't help :/
21:12 UtahDave hannesstruss: do those pillar sls files have some of the same key names?  key names can't conflict
21:13 jalbretsen joined #salt
21:13 honestly salt needs a bundler for state files the way pathogen for vim works.
21:14 vbabiy Can you using salt-formulas with out a master?
21:14 forrest yes vbabiy
21:14 forrest masterless minion
21:14 honestly I suppose you could just do that with git...
21:14 forrest Check out the masterless quickstart: http://docs.saltstack.com/topics/tutorials/quickstart.html
21:15 vbabiy forrest where do I configure the minion to use gitfs?
21:15 vbabiy honestly I think was getting at it.
21:15 vbabiy Do I just need to check them out?
21:15 honestly vbabiy: I think so
21:16 honestly you don't actually want to use gitfs
21:16 vbabiy honestly why?
21:16 forrest gitfs is specifically for the management of files.
21:16 honestly well
21:16 forrest sls or whatever
21:16 forrest at least according to the example
21:16 honestly augh, I'm too tired to brain
21:16 ingwaem I use a combination of git SLS files to pull into a central location and then cp.get_dir to move them to minions
21:17 forrest instead of using gitfs ingwaem?
21:17 ingwaem yes
21:17 forrest any particular reason?
21:17 honestly vbabiy: the problem here is that git doesn't nest well
21:17 vbabiy Ah I was thinking that the formulas should be added to your gitfs
21:17 forrest you can vbabiy
21:17 forrest the example shows having the sls files in there
21:17 forrest I just don't know about doing so on a masterless minion
21:17 Ryan_Lane1 joined #salt
21:17 forrest never tried to configure that on mine
21:17 vbabiy ah
21:18 vbabiy For how long does the master cache the repo
21:18 vbabiy And can you lock it in to a version
21:18 ingwaem it was just the easiest way to get it up and running. By defining my git repos in SLS files I can just call that state file anytime I want to update…it logs into git checks for changes and does updates as necessary…as if using git checkout. Then using cp.get_dir I get the whole benefits of the salt file management which is fast, supports gzip etc
21:18 ingwaem master will cache it each time you execute the SLS files….if there are no changes nothing gets updated.
21:19 ingwaem you can also lock it into branches, version or tags
21:19 vbabiy Nice
21:19 vbabiy thanks ingwaem
21:19 ingwaem yw :)  if you want to search for some documentation on it, just search git.latest
21:20 Ryan_Lane joined #salt
21:20 ingwaem http://docs.saltstack.com/ref/states/all/salt.states.git.html?highlight=git.latest#salt.states.git.latest
21:20 CheKoLyN joined #salt
21:22 ingwaem to easily maintain my git repos I have a mysql db with php. I enter all the params for a repo and then when I save the record, I use salt to inject the correctly formatted contents of the SLS file into the correct location using salt. I then execute that state file and bingo it's done. Makes changing branches or tags very easy too, just update the record using the little manager and the state file is regenerated before executing
21:23 hannesstruss UtahDave: ah, that seems to be it, thanks! I thought the tree structures would merge somehow… Is that documented anywhere? What would be the canonical way to implement something like that?
21:24 UtahDave hannesstruss: Right now they don't merge. There's been some work to allow for that, but right now the keys have to all be unique
21:25 lazyguru left #salt
21:26 mezgani joined #salt
21:26 hannesstruss ok, thanks nonetheless :)
21:27 UtahDave you're welcome!
21:27 honestly aaand I've exhausted all my brain capacity for today. This is the result: https://github.com/duk3luk3/nfs-user-formula
21:27 abe_music forrest: ARGH, so this problem occurs on *any* package
21:27 forrest that comes from that specific repo?
21:27 honestly needs to be tested and then eventually merged with the users-formula
21:27 forrest or even on stuff like apache?
21:27 rberger joined #salt
21:27 * honestly collapses
21:28 abe_music forrest: the simplest sls, with just a ack: pkg.installed fails with the broken pipe error
21:28 forrest lol
21:28 forrest what version are you on?
21:29 abe_music centos 6.4
21:30 forrest no I mean what version of salt sorry
21:30 abe_music ah
21:30 abe_music salt-master = salt 0.16.0-2741-g80ae64e
21:30 abe_music salt-minion = salt-minion 0.16.3
21:31 ingwaem ooooh you may want to update your master
21:31 abe_music ingwaem: looks a bit behind :)
21:31 ingwaem yea just a bit :) about 3 sub versions
21:32 ingwaem once you update just make sure you stop your master service then restart it incase it is stuck in the old state
21:32 forrest can you update your master abe_music, and then try to run your simple sls again
21:32 forrest good point ingwaem
21:33 ingwaem :)
21:33 abe_music forrest, ingwaem: will do…i'm running the develop branch, think that's alright?
21:33 whit joined #salt
21:33 forrest this is in a test environment right?
21:33 forrest lol
21:33 abe_music yes, testing
21:33 forrest lets do it then
21:33 ingwaem you want to check out the tag v0.16.3
21:33 forrest yup
21:33 abe_music there are some pull requests that i added a week or so ago that i'll need for my environment
21:34 ingwaem abe_music: try this when checking out:  git checkout v0.16.3
21:34 ingwaem abe_music: then cd into the salt dir and run this: python setup.py install --force
21:36 abe_music ingwaem, i'm using pip for installation, so i added the tag to my requirements file: -e git://github.com/saltstack/salt.git@v0.16.3#egg=salt
21:36 ingwaem ahh ok :)
21:36 jxqz joined #salt
21:37 whit joined #salt
21:37 jxqz well
21:37 jxqz I interviewed at salt yesterday
21:37 smcquay joined #salt
21:37 jxqz not sure I made the callback :(
21:38 forrest That's pretty much every interview ever I believe.
21:38 jxqz haha
21:39 abe_music does salt require onsite developers?
21:39 jxqz or on the ones you think you made it, just after you leave, they say 'man, that guy was lame'
21:39 forrest abe_music, they have an office, but a bunch of people work remote
21:39 auser joined #salt
21:39 abe_music that's cool
21:39 whit joined #salt
21:40 abe_music not many python gigs in the nashville area
21:40 forrest it happens jxqz, worst case you study up and apply again right?
21:40 jxqz yeah
21:41 jxqz but by then, they'll be so big with facebook style meta office space that they'll only take stanford grads
21:41 forrest lol
21:41 forrest have you ever been interviewed by an intern?
21:41 forrest that's a blow to your pride when the dude who is supposed to interview you can't make it and they hand it over to the intern :D
21:41 jxqz I about told Tom I'd be happy parking a banana chair in the corner
21:42 hazzadous joined #salt
21:42 emocakes joined #salt
21:42 ingwaem one of the most important things I do when an interview is coming to an end is to declare outright that I want the job and when can I start…that way there are no assumptions…if you seemed a little off in the interview they may think "perhaps he doesn't want the job" but they don't know because you didn't say so…just my 5c :)
21:43 forrest The other good thing there ingwaem, is then you get an immediate response, which is awesome when it's a company that just likes to take forever.
21:43 whit joined #salt
21:43 cedwards jxqz: if they grow any more they're going to need more office space
21:43 jxqz I'm not one of those guys who demands to be pampered with luxurious furniture, hors de oeuvres, and game consoles
21:43 cedwards jxqz: it's been a while since i've been over there, but they seemed pretty full then
21:43 giantlock joined #salt
21:44 ingwaem mmmmhm hordes durverers ! nom nom nom
21:44 jxqz cedwards: indeed
21:44 ingwaem should have said horses* durverers…crazy take on pronunciation :)
21:44 abe_music forrest: both the v0.16.3 and develop branches exhibit the same issue…sounds like a bug. i'm wondering if it has anything to do with centos 6.4
21:44 forrest that sounds really weird.
21:44 cedwards I worked at my current company almost a year before I even discovered we had a game room lol
21:45 ingwaem cedwards lol :)
21:45 forrest I worked at my old job for 2 years knowing we had a game room (couches, Wiis and such), and never touched it once. The only time I went over there was to fill a water bottle up
21:45 forrest no time
21:46 UtahDave holy crud, cedwards, that basketball gym you guys have hanging right over the freeway is awesome looking!
21:46 jxqz businesses like that kind of repel me
21:46 cedwards UtahDave: guess how many times I've used it :)
21:46 cedwards UtahDave: ..or the rock wall, or the gym
21:47 forrest abe_music, I'm running on rhel 6.4 for some VMs, are you using a master/minion setup? Or masterless minion?
21:47 forrest 0
21:47 abe_music forrest: master/minion
21:47 UtahDave jxqz: We have corndogs in the freezer!
21:47 jxqz yeah!
21:47 jxqz :)
21:47 abe_music forrest: and i can run the same state against my ubuntu machines fine
21:47 forrest yea I remember you saying that
21:47 forrest hang on
21:47 forrest you're on 0.16.3-1?
21:48 ingwaem I work in some gaming studios at times…they always have snack like candy chips peanuts, sodas etc. gaming rooms are not that common but consoles on each desk…Yea!
21:48 ingwaem and crunch time always = at least 1 square meal per day
21:48 forrest abe_music, does it fail running the install straight from the command line as well?
21:50 abe_music forrest: salt-minion is 0.16.3 (installed from bootstrap) and salt-master is 0.16.0-2948-g61605a2 (latest develop branch as of a few minutes ago)
21:50 ingwaem abe_music did you restart the salt-master service after that?
21:50 abe_music not sure why it's coming up 0.16.0 though
21:50 abe_music ingwaem: yup, killall salt-master, followed by salt-master -ldebug -d
21:50 forrest it should be 0.16.3
21:51 abe_music not sure if there's anything i need to do on the salt-minion side
21:51 ingwaem does your master also run a minion?
21:51 z3uS joined #salt
21:51 ingwaem as in master and minion being on the same machine
21:51 ronc joined #salt
21:52 abe_music ingwaem: nope, just a master
21:52 ingwaem that's strange
21:52 abe_music forrest: yeah i'm not sure why salt-master is reporting that….my salt source is checked out and the last commit in the log is 61605a2a5fa13b6dca1ac27583b6f23a005a52ec, which github confirms in the develop branch
21:52 forrest weird
21:52 forrest it's gonna take me a minute here abe_music, I updated to the 0.16.3 RPM and now the damn service won't start
21:53 abe_music no worries…i might have to pick back up on this later this evening or tomorrow. gonna have to leave soon to make sure i get home by 6 with traffic
21:54 jinrou joined #salt
21:54 forrest there we go, odd when I stopped the service it didn't drop the ports for some reason so it couldn't rebind
21:54 smcquay so I have a vm image I am prepping to be a template for future vms. good avenue of thought to think get salt-minion installed but not started, place enough /etc/salt/minion in place to talk to master, make sure salt-minion would come up after powercycle? is that sufficient?
21:54 forrest ok, so did you try to install a package from the command line with salt abe_music?
21:55 abe_music forrest: i did: salt centos-nn-abe-1 pkg.install 'ack'
21:56 abe_music and got the broken pipe on the minion log, master didn't really dump anything useful
21:56 forrest I assume that's available for you right? For me it just shows it's coming from the epel repo
21:56 ingwaem smcquay: you may want to look into salt-cloud or salty-vagrant as they can spin up the vm, install salt and get it running
21:57 smcquay ingwaem: I noticed those projects, but didn't dig deep. thanks for the suggestion.
21:57 ingwaem smcquay: you're welcome…hope they help.
21:57 abe_music forrest: yeah, i can run yum install ack on the minion and it will ask if i want to install it…comes from epel
21:57 forrest ok
21:57 forrest so I just ran that same command you gave me
21:57 forrest this is on rhel 6.4, salt version 0.16.3
21:57 forrest (the rpm version)
21:57 smcquay they may, but I may need to roll my own salt-non-cloud-not-virtualbox :D
21:57 forrest and it worked just fine
21:58 forrest abe_music, can you try to install the RPM from epel for 0.16.3?
21:58 forrest to see if maybe something about what you're pulling in from the dev branch is breaking it
21:58 abe_music forrest: minion or master?
21:58 ingwaem smcquay salt-cloud also works with things like vmware I believe.
21:58 forrest your minion is already on 0.16.3 right?
21:58 forrest lets try JUST the master to start
21:58 abe_music forrest: oh, right…yeah so master
21:58 forrest to try and eliminate as many variables as possible
21:59 jpadilla joined #salt
21:59 forrest if it still bombs on the master, let's revert to the 0.16 version you've got from git (which should be 0.16.3) and then try installing the RPM on the minion
22:01 abe_music forrest, yeah that worked
22:01 abe_music hmm
22:01 forrest I think you're actually pulling in 0.16 then :P
22:01 forrest or someone busted something on the develop branch
22:02 forrest so can you try to run your full sls before you go?
22:02 forrest I just want to know if it works or not
22:03 abe_music forrest: alright
22:03 forrest awesome thanks
22:03 forrest then leave before traffic gets bad
22:03 abe_music forrest, yeah it's moving right along now...
22:03 forrest Great
22:04 abe_music forrest, i checked the version.py in the develop branch and it's default is 0.16.0, so maybe that hasn't been updated in a while?
22:04 forrest Oh that could be it
22:04 forrest ingwaem, have you been able to pull 0.16.3 from git?
22:05 abe_music forrest, thanks a ton for the help…on the right track at least. i'll get back in the morning and see what i can do about maybe tracking down why the develop branch is busted
22:05 ingwaem let me try on a clean vm forrest
22:05 abe_music ingwaem: ty too
22:05 forrest Yea np abe_music, just glad we got it figured out!
22:05 aleszoulek joined #salt
22:05 forrest awesome thanks ingwaeum
22:05 forrest *ingwaem
22:06 kermit joined #salt
22:08 MrTango joined #salt
22:11 ingwaem sorry it's coming…had to clone and start up the vm…won't be long
22:11 forrest It's all good man
22:14 forrest ingwaem, what OS are you spinning up?
22:14 ingwaem ubuntu I believe…want Centos?
22:14 forrest no
22:15 forrest I don't think it realy matters
22:15 forrest *really
22:15 dthom91 joined #salt
22:15 ingwaem apt-get install python-zmq
22:15 ingwaem oops
22:15 forrest we just want to see what the pull looks like
22:15 ingwaem wrong window lol
22:15 forrest lol
22:15 forrest manual installs??
22:15 ingwaem I cloned the repo then ran the setup…setup ran fine but then when I try salt-master --version it says zmq isn't found
22:15 forrest bummer
22:16 ingwaem I can try from bootstrap if that's better?
22:16 forrest so abe_music was seeing salt-master is 0.16.0-2948-g61605a2
22:16 forrest no no, do git please
22:16 forrest since that's how abe was doing it
22:16 linjan_ joined #salt
22:17 ingwaem kk
22:23 ingwaem after manually installing the required packages (trail and error, watching errors for zmq, msgpack, jinja etc) salt-master --version returns 0.16.3
22:23 ingwaem salt-minion --version returns the same
22:23 forrest weird
22:24 ingwaem after cloning I did force it to use the tag by checking that out
22:24 forrest I wonder if he needs to fetch the tags
22:24 ingwaem seems the master branch stays back a few versions so yes, I would say its' a must
22:24 forrest yea I'll propose that when he logs in tomorrow
22:24 ingwaem I can try it again without the specific checkout and see if that returns the same versions he had
22:25 forrest if you have time sure
22:25 rberger joined #salt
22:26 ingwaem I'll re-checkout and do an install force…should revert back to default install
22:26 ingwaem Ahh! yes that's the issue
22:26 ingwaem salt-master -.16.0-2950-g71a15e3
22:26 oz_akan_ joined #salt
22:27 ingwaem oops 0 instead of the 0
22:27 ingwaem instead of the -
22:27 forrest I gotcha
22:27 ingwaem danged flangees
22:27 forrest so that was when you didn't specify the tags?
22:27 ingwaem yes
22:27 forrest awesome
22:27 forrest thanks for checking on that ingwaem
22:27 ingwaem once cloned cd into salt and just run git checkout v0.16.3 and that will force the new version through
22:27 ingwaem yw
22:28 forrest yea it probably wasn't working for him because he had that script defined
22:28 ingwaem and then run install with --force tags
22:28 forrest right, so that must have been the issue
22:28 forrest cool cool
22:28 forrest the dev branch isn't broken, woo
22:28 ingwaem yea my assumption is he's pulling from master…master is always a few versions behind…especially with the minor versions being released
22:29 ingwaem and also master is defacto when you clone
22:30 forrest It's odd though, because he said this was in his requirements file:  -e git://github.com/saltstack/salt.git@v0.16.3#egg=salt
22:30 forrest It must have been because he didn't specify the dev branch
22:31 ingwaem could be…I've never used pip to check out code before
22:32 ingwaem I use tried and tested methods that work for all systems including no python being available
22:32 forrest Yea, I haven't used it in that fashion either
22:32 cewood joined #salt
22:32 forrest but at least we've got the solution now
22:32 ingwaem :)
22:35 dthom91 joined #salt
22:42 StDiluted joined #salt
22:43 chrisgilmerproj left #salt
22:47 mgw joined #salt
23:06 Pate_ joined #salt
23:08 dthom91 joined #salt
23:11 oz_akan_ joined #salt
23:22 mafrosis joined #salt
23:23 alunduil joined #salt
23:23 robertkeizer joined #salt
23:29 Thiggy joined #salt
23:32 Thiggy I'm using gitfs for my states, but I also have a file_root under the base environment I'm trying to get to serve some a file "special_file" I need to keep out of the states repo. when I do `cp.get_file salt://secure/special_file` it downloads fine, but if I specify an env `cp.get_file salt://secure/special_file env=production` it doesn't show up. I thought the base environment was supposed to be flattened into all the other environments so
23:32 Thiggy those files are always available. Am I wrong?
23:32 Thiggy (sorry for super long question)
23:32 aat_ joined #salt
23:38 dthom91 joined #salt
23:40 Thiggy Neeeeevermind… Just learned about salt://secure/special_file?env=base
23:41 aat joined #salt
23:42 fxhp Thiggy - could you expand that?
23:42 Thiggy How do you mean?
23:42 fxhp is "secure" in git?
23:42 Thiggy oh no, secure was the local file_root
23:42 Thiggy so it was base: file_root: /srv/secure/
23:43 Thiggy and everything else was in gitfs
23:43 fxhp I see
23:43 fxhp ok
23:43 fxhp does gitfs support SSH yet?
23:43 goodwill gitfs?
23:43 Thiggy and in my git repo in the production branch, I was trying to grab a file in secure (in the base env)
23:43 Thiggy gitfs does support ssh
23:44 Thiggy Though the syntax I have probably spent at least an hour debugging over the course of ~4 diff times because I can't seem to get it right.
23:44 fxhp Thiggy: hmm last time I looked at it, it only supported github via HTTP
23:44 Thiggy http://docs.saltstack.com/topics/tutorials/gitfs.html#gitfs-remotes-over-ssh
23:44 Thiggy It's new, and I couldn't get it working in .16.0 but it works in .16.3 with a few bumps
23:45 fxhp goodwill - gitfs allows a system administrator to setup git repos as salt file_roots and pillar_roots
23:45 NV so in my setup i'd like to manage roles centrally, which doesn't appear easily possible with grains, so pillar would be the next best bet - has anyone done anything like that before that might have some suggestions on how best to architect such a pillar-based role database type jigger?
23:45 fxhp goodwill - so that a local checkout is not needed on the salt-master
23:45 goodwill perverts
23:45 goodwill PERVERTS!!!
23:45 * whiteinge slaps goodwill
23:46 goodwill awww
23:46 * goodwill hugs whiteinge
23:46 fxhp goodwill - I wish gitfs could be tethered to a specific revision.  that would be magic
23:46 * goodwill tugs on whiteinge's red beard
23:46 * goodwill hums ... do you believe in magic!!!
23:46 whiteinge :D
23:47 fxhp whiteinge - can gitfs be tethered to a specific revision?
23:47 whiteinge i don't know.
23:47 Thiggy @fxhp you can tie it to a specific branch. Does that not suffice?
23:48 Katafalkas joined #salt
23:48 Thiggy branches map to environments, and the master branch maps to the base environment
23:49 fxhp Thiggy : tieing to a branch might work
23:49 Thiggy what's your use case
23:49 fxhp Thiggy : going to need to harass my teamates to help me with branches
23:50 fxhp Thiggy - use case, freezing a salt-master to a specific "release" till its reviewed
23:50 Thiggy Ah yeah totes. I'm doing that now. I have 2 diff environments/branchs production and staging, we do our fiddling in staging, then push to prod when it's happy (theoretically… it's a work in progress)
23:51 Thiggy I think UtahDave has a workflow over 3 branches, dev, qa, and prod
23:51 Thiggy you'd have to talk to him about that though
23:51 fxhp Thiggy - we don't currently have different envs in our implementation.  I might tests them out on my lab salt-master though
23:52 Thiggy We didn't either until about 2 weeks ago. If you have questions let me know. It was a little mind-bendy at first.
23:53 fxhp Thiggy - we are using just base right now, cool I'll give it a test-drive at some point
23:53 NV fxhp: you could have two git repos
23:53 NV one for development, then once approved, pull changes from the dev git repo to your prod git repo
23:53 fxhp NV - we currently have 4 ...
23:54 NV so what's the problem then?
23:54 fxhp NV - hmmm no problem just trying to hear about other peoples solutions
23:55 kula NV: i had an idea once for something like that, never got around to implementing it. i tend to think in terms of "machine_a has roles a, b, c" and i was going to write a pillar plugin that translated that to creating pillars called role_a, role_b, etc. with appropriate data in them
23:55 moondawg joined #salt
23:55 NV just discovered node groups, might be the solution i'm after
23:56 kula do you still have to restart the salt-master to update node groups? that's the bit that always got me.
23:56 kula or is that pluginable now?
23:58 fxhp NV - Where do you define nodegroups?  master conf file?
23:58 forrest joined #salt
23:59 moondawg So, I have some that I am testing that extends both the file module and state. It safely allows someone to create character, block, and fifo for chrooted environments. Would this feature be accepted by the salt dev team?
23:59 moondawg code

| Channels | #salt index | Today | | Search | Google Search | Plain-Text | summary