Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2015-04-07

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

All times shown according to UTC.

Time Nick Message
00:01 shaiguit1r joined #salt
00:04 Guest70 joined #salt
00:07 lothiraldan joined #salt
00:07 sunkist joined #salt
00:10 amcorreia joined #salt
00:17 _JZ_ joined #salt
00:18 bhosmer joined #salt
00:28 JDiPierro joined #salt
00:30 SeeDickCode joined #salt
00:36 elektrix_ so when doing version control on multiple environments (e.g. /srv/salt/prod /srv/salt/dev), I'm assuming you should put each environment under version control, then just clone the dev environment to prod once you're done testing/developing?
00:37 elektrix_ and not just version control /srv/salt, and manually copy files between /srv/salt/dev and /srv/salt/prod
00:37 elektrix_ ?
00:38 elektrix_ so for a full dev/test/prod env, you'd have 3 different git repos, and another 3 for /srv/pillar/{dev,test,prod}?
00:40 keltim we use just one, and target environments in sls files
00:40 keltim via grains
00:40 keltim but our devs have no desire to use a salt server
00:41 keltim I guess it really depends on who and how many are developing states
00:45 jerematic joined #salt
00:54 quintinadam joined #salt
01:00 timoguin joined #salt
01:01 jonlangemak joined #salt
01:07 pahko joined #salt
01:07 lothiraldan joined #salt
01:11 CeBe1 joined #salt
01:15 iggy elektrix_: you can also use gitfs' ability to render branches aas environments
01:18 desposo joined #salt
01:20 jonlangemak joined #salt
01:25 iggy speedlight: use - creates: probably
01:26 iggy XenophonF: ouch, yeah, that bit me too... you should file a bug about that (and let me know what # it is so I Can follow too)
01:31 ITChap joined #salt
01:33 pahko joined #salt
01:35 jessie_ joined #salt
01:35 jessie_ is there anyone around ?
01:36 Vynce well, i am, but i am unlikely to be able to answer a question. ( :
01:37 Vynce i've got one to ask, though — what's the in-sls format of, say, -C 'G@role:webserver and primary-*' ?
01:39 jessie_ sdorry dont klnow
01:40 robawt Vynce: in the top.sls file?
01:40 Vynce robawt: sure
01:43 robawt Vynce: http://docs.saltstack.com/en/latest/ref/states/top.html#other-ways-of-targeting-minions
01:49 robawt see the last one about compound matching in the example
01:49 otter768 joined #salt
01:49 Vynce ah, perfect, thanks
01:49 jessie_ Vynce: do you use pycharm or any other APP to access VCS
01:49 jessie_ robawt: do you use pycharm or any other APP to access VCS
01:51 ilbot3 joined #salt
01:51 Topic for #salt is now Welcome to #salt | 2014.7.2 is the latest | Please use https://gist.github.com for code, don't paste directly into the channel | Please be patient when asking questions as we are volunteers and may not have immediate answers | Channel logs are available at http://irclog.perlgeek.de/salt/
02:00 boubou joined #salt
02:06 scbunn joined #salt
02:09 ITChap joined #salt
02:11 iromli joined #salt
02:15 Furao joined #salt
02:21 I3olle joined #salt
02:21 pahko_ joined #salt
02:24 evle joined #salt
02:26 iggy that's what branches are for
02:27 iggy make a branch, test the branch, merge the branch to master, bam
02:33 michelangelo joined #salt
02:34 ITChap joined #salt
02:35 jessie_ iggy: your back !!!
02:37 jessie_ iggy: do i have to test the branch first before i can grab the lot ?
02:44 Vynce man, i wish i could understand salt's color coding. i mean, i think i udnerstand what it's supposed to mean, but it doesn't seem to always mean that
02:45 jessie_ Vynce: think of its as a massive puzzle and that the docs are a seperate puzzle :)
02:45 Vynce sadly accurate
02:45 jessie_ Vynce: all fun though. :)
02:46 jessie_ Vynce: almost makes you want to go to chef
02:46 Vynce don't know anything about chef, but i'm pretty resistant to a technology that uses the term "databag"
02:46 jessie_ if it works then i dont care what they call it :)
02:47 iggy jessie_: don't follow
02:47 jessie_ a hooker is a hooker but still gets the job done :) lol
02:47 jessie_ iggy: hi dude, assumed you wernt back till thursday.
02:48 iggy I left Thursday
02:48 jessie_ iggy: having some poo issues with pycharm, bitbucket in grabbing repos.
02:48 iggy got back today
02:48 stanchan joined #salt
02:48 iggy don't know anything about pycharm
02:48 favadi joined #salt
02:49 jessie_ iggy: do you use any 3rd party VCS, ? pycharm is the only one am looking at
02:49 iggy github?
02:50 jessie_ iggy: yeah i can use bit bucket or github but i wanna grab the branches / projects locally to then edit and then push back to VCS
02:52 iggy my workflow: edit a local copy, commit to dev, test, merge to qa, test, merge to prod
02:52 jessie_ hence using pycharm...
02:52 iggy I don't even know what pycharm is... so continuning to mention it isn't going to change anything
02:53 jessie_ iggy: hehe fair enough :)
02:53 jessie_ iggy: wb anyway
02:54 iggy I told you how I work, that's the best I can tell you
02:55 Vynce (i think the basic workflow isn't what jessie_'s trying to establish, but i am not sure what the actual question is.  of course, i also don't use pycharm — maybe the question would be clearer to someone who did.)
02:57 scbunn joined #salt
02:57 mosen im using IDEA so basically the same
02:58 jessie_ pycharm is a mass python based editing tool where it auto pushes to bitbucket commits etc. just handles things better in that sense. grab files from VCS to local, edit locat, then commit then push to VCS, then salt stack grabs data automatically ready for execution
02:58 jessie_ thats what am lookin at i think as a process.
02:58 mosen I dont use any kind of auto commit anything so im not sure how that works :)
02:59 jessie_ mosen: what do you use ?
02:59 mosen jessie_: IDEA
02:59 mosen jessie_: its the big brother of pycharm
02:59 jessie_ mosen: you using bitbucket ?
02:59 mosen jessie_: nah just github
02:59 mosen jessie_: but if youre using git, should be able to pull branches the same
03:00 iggy that sounds terrible... hundreds of little bullshit commits to get one feature working
03:00 jessie_ iggy: just means am working clearer.
03:01 iggy don't ever work on formulas... we won't accept them like that
03:01 jessie_ mosen: you on mac or win
03:01 mosen jessie_: mac
03:01 mosen iggy: its possible to use a sane workflow with the IDE
03:02 jessie_ mosen: i would really like to borrow you for a few mins if possible to guage a proper understanding of how you working with it
03:02 iggy that doesn't sound like one (personally, but whatever works)
03:03 XenophonF hey iggy - which issue did you say you wanted to follow?
03:03 XenophonF was it the permissions of the cloud providers?
03:03 iggy the one you file about multiple level inheritance
03:04 XenophonF ah gotcha
03:04 __number5__ jessie_: yoo need this http://rogerdudler.github.io/git-guide/  then gitref.com
03:04 __number5__ sorry, gitref.org
03:04 XenophonF man i am so close to having all the salt-cloud stuff working
03:05 jessie_ XenophonF: i am quite good with salt cloud so any issues gimmi a shout happy to help
03:05 iggy I had some salt-formula changes that did the inheritance
03:05 XenophonF jessie_: which operating systems (and which cloud offerings) are you deploying?
03:05 jessie_ i deploys with AWS azure and DO
03:06 XenophonF OK, I'm using AWS, too, and soon vSphere.
03:06 XenophonF Are you using instance credentials in your AWS provider configs, or are you using API keys?
03:06 XenophonF I tried to use the instance role credentials or whatever they're called, and it works with the AWS CLI, but it doesn't work with salt-cloud
03:07 jessie_ vsphere... is lame as ...
03:07 jessie_ api keys for testing purposes
03:09 XenophonF i thought i had everything set up properly for my centos7 deployments, too, but it looks like none of the minions got installed
03:09 XenophonF do i have to set up salt-ssh at the same time i set up salt-cloud?
03:10 jessie_ XenophonF: check your configs again dude. some stuff is NOT in the docs...
03:10 XenophonF that's what i was afraid of
03:10 jessie_ important vars are missed
03:11 jessie_ i was pissed when i found out
03:11 iggy open issues
03:13 pahko joined #salt
03:14 XenophonF the AWS docs conflate EC2-Classic and EC2-VPC, too
03:15 XenophonF i'm using EC2-VPC everywhere, so setting "securitygroup" in the provider is wrong
03:15 XenophonF that's definitely going into the issues list
03:17 ajw0100 joined #salt
03:18 __number5__ XenophonF: you should look at those boto_* states/modules. salt-cloud won't help you much if you are using VPC
03:18 XenophonF __number5__: i'll do that
03:19 XenophonF i really want the provisioning stuff, though
03:20 iggy the salt-cloud maps aren't wildly extensible
03:20 __number5__ just for your reference, we are using packer + salt-masterless to build AMI, then throw then together in VPC either via Boto or terraform
03:21 iggy (it's something I'd like to fix at some point)
03:23 XenophonF for windows images, can win_installer point to a salt:// or s3:// or http:// (https://) URL?
03:24 XenophonF or does the windows installer package have to live on the salt master?
03:24 XenophonF or rather the same server as salt-cloud?
03:24 murrdoc joined #salt
03:25 Vynce anybody happen to know if the cron.present yaml can take a yaml list for, say, hours, or if it's expected to just pass a comma-delimited string?
03:26 murrdoc example ?
03:27 Vynce -hour: 2,4,10 vs -hour: \n -2 \n -4 \n -10
03:27 iggy I've only ever seen it as csv
03:27 murrdoc yup
03:27 iggy doesn't mean it won't work... give it a shot
03:27 Vynce no time for testing edge cases, sadly.
03:28 murrdoc you can use state.show_sls on both versions to see if the state renders the same in both cases
03:28 Vynce just need A Way That Works.
03:28 murrdoc tm
03:28 Vynce yup
03:29 viq joined #salt
03:37 perfectsine joined #salt
03:44 otter768 joined #salt
03:48 dalexander joined #salt
03:54 StDiluted joined #salt
03:59 pdayton joined #salt
04:02 murrdoc iggy:  to be fair , in https://github.com/saltstack-formulas/sysstat-formula/pull/9, the fact that there were multiple top level keys like config_location and pillar.lookup is pretty ghetto
04:02 iggy no doubt
04:02 iggy that's why I didn't kline the change
04:02 murrdoc i ll put a note in the Disclaimer section of the README
04:03 iggy also why I said we need to start documenting these changes in a repeatable fashion
04:03 Vynce i get really nervous about what feels liek the very implicit grammar of topfiles using " - someword" meaning "apply the state from someword.sls" — is there a more explicit syntax that means the same thing?
04:03 Furao kernel: [423225.702913] xen_netfront: xennet: skb rides the rocket: 19 slots
04:03 iggy other formulas (indeed the salt-formula) is going to need to break backward compat at some point... we need a consistent way of letting people know
04:03 Furao those create so many errors and performance issue in ec2 :(
04:04 iggy Vynce: you mean you want to - someword.sls?
04:07 lictor36 joined #salt
04:08 rdas joined #salt
04:08 beneggett joined #salt
04:08 Vynce well, ideally something even more explicit, but whatever i can get
04:09 iggy how does it get more explicit than that?
04:10 Vynce well, teh fact that we're applying a state contained in a file is completely unsaid but meant
04:10 Vynce like, i read '*' - someword and my mind does not immediately leap to "there's another file named someword.sls; apply all teh stuff from that"
04:10 iggy there's not a top way to do that, but in newer salt's you can do that on the command line
04:11 murrdoc iggy:  updated
04:12 murrdoc https://github.com/saltstack-formulas/sysstat-formula/pull/9/files
04:12 Vynce so i presume too, then, that there's not a way for a .sls file other than top.sls to do that at all?
04:12 murrdoc no any file can include another file
04:12 Vynce i think fundamentally one of the more surprigint th… really? in the same way?
04:12 murrdoc not in the same way
04:13 murrdoc useing the 'include' construct
04:13 murrdoc only the top.sls is environment aware
04:13 Vynce but you can't use the include construct in top.sls?
04:13 murrdoc its not needed
04:13 Vynce i don't knwo what you mean by "environment aware"
04:13 murrdoc the top.sls is parsed differently
04:13 Vynce not needed… but allowed, or not allowed?
04:13 Vynce i know.  that drives me nuts.
04:13 Vynce ( :
04:13 iggy sudo salt-call state.sls_id apache http
04:14 murrdoc well its not weird perse
04:14 murrdoc its an entry point so it has more logic
04:14 murrdoc the top.sls is the entry point for the highstate
04:14 murrdoc and all the state files are 'classes' of work
04:14 iggy modules in python speak
04:15 murrdoc so the 'map' of classes to apply to 'environment and other compound matching' should be different form a class
04:15 murrdoc from*
04:15 Vynce (iggy: was that salt-call line to me?  if it was, i don't understand it)
04:15 Vynce murrdoc: your'e over my head again
04:15 Vynce please to be answering straightforward question: is it allowable to use "include" syntax in top.sls?
04:15 iggy Vynce: yes, it lets you run a single state stanza (from a sls file)
04:16 iggy why would you use include in top.sls?
04:16 Vynce to be more explicit.  becaus ehte implictness of - someword bothers me
04:16 Vynce explicit is better than implicit
04:17 Vynce (as a langauge i don't love likes to say)
04:17 murrdoc the part u are missing the part where they arent the same thing
04:17 iggy include doesn't buy you any more explicit-ness
04:17 iggy include takes the same format of module name (that gets converted to a file)
04:18 Vynce um.  maybe we mean different things by explicit?  ahving the verb actually present is pretty definitely more explicit, to me.
04:18 Vynce "include someword" including someword.sls is still more explicit than "- someword" including someword.sls
04:19 iggy that sounds very pedantic (and overly repetitive)
04:19 Vynce "they aren't the same thing" — well, sadly, the docs on include are not terribly clear to me… hwo are they different?
04:20 iggy I didn't understand what you wanted
04:20 Vynce i like pedantic.  it's clearer.
04:20 iggy I (personally) dislike your idea... if I had a top file with the word "include" in it 120 times I'd shoot myself
04:21 iggy (I don't even like having - match: as many times as I do)
04:23 Vynce i agree that - match: is terrible. i thinkt eh yaml grammar for sls fiels is pretty counter-intuitive and messy. and maybe after i've been using salt for a year, i'll want to drop the "include" myself.
04:23 Vynce but for a new user, i think "include file" is more clear that it's including a file than "file" is.
04:23 iggy it isn't including it though (not the same way include works)
04:24 Vynce OK, as i said, i don't understand the docs for include.
04:24 iggy it's saying "assign X module to a host and determine the required dependency tree"
04:24 iggy (the top file)
04:26 Vynce i don't understand the difference
04:27 Vynce for reference, everything i (don't) know about include is based on this URL: http://docs.saltstack.com/en/latest/ref/states/include.html … it's pretty vague.
04:27 iggy include the designated module in the current modules context (it doesn't necessarily run anything)
04:28 Vynce i think so far i have that there is a keyword "include" that invokes a verb "include" but no understanding of what is meant by the verb.
04:28 spookah joined #salt
04:29 Vynce anyway, it's not important tha ti understand this right now; the questionwa s just "is ther ea more explicit way to write  ' - someword'?" and it looks like the answer is "no"
04:32 iggy so yeah, the answer to your question... no, and include doesn't work the same way top mapping does
04:33 iggy so if you try to talk about this again, don't use include in the top file context... it's confusing
04:34 Vynce if you check my original question, i didn't, initially. i started using tha tword when someone brought it up. i am now back to believing that i have no idea what "include" means.
04:34 Vynce and i agree that it's confusing. ( :
04:35 rdas joined #salt
04:36 iggy one of the big problems is (in my eyes) that top was called top.sls
04:36 Vynce (though perhaps not as confusing as the fact that top.sls and antyhingelse.sls are both yaml files with completely different semantics despite having the same 'extension')
04:36 Vynce yes
04:36 scbunn joined #salt
04:37 iggy but if you know how the software evolved, it makes a lot more sense
04:37 malinoff joined #salt
04:38 Vynce i believe that. just another stumbling block in the salt startup curve.
04:41 iggy most people don't put that much thought into the why... they just believe
04:43 Vynce i can't do things i don't understand well enough to hang a memory on.
04:45 beneggett joined #salt
04:50 pdayton joined #salt
04:53 cuonglm joined #salt
04:53 TyrfingMjolnir joined #salt
04:54 cuonglm Does anyone have trouble with setting root password when using salt-formula mysql
04:55 cuonglm I used Debian, set root password in pillar file, running state mysql.server
04:56 cuonglm the result is ok but root password was not set
04:58 bhosmer_ joined #salt
04:58 Vynce joined #salt
04:59 chandankumar joined #salt
05:01 catpig joined #salt
05:07 tkharju joined #salt
05:09 tkharju joined #salt
05:16 benegget_ joined #salt
05:26 aranhoide joined #salt
05:45 otter768 joined #salt
05:46 cuonglm joined #salt
05:48 stoogenmeyer joined #salt
05:51 catpigger joined #salt
05:55 hebz0rl joined #salt
05:58 colttt joined #salt
06:01 catpig joined #salt
06:05 zircote joined #salt
06:09 tzero joined #salt
06:10 loz-- joined #salt
06:10 TristanMatthews joined #salt
06:16 rhodgin joined #salt
06:18 TristanMatthews Not sure if anyone is around at this time of night, but I have a question about salt best practices. I'm re-writing my initial bootstrap scripts for right after an aws instance comes up. My old system handled formatting and mounting ebs volumes before salt was called. I always thought this was a kludge, so was going to move everything into salt. salt.states.mount seems to do most of what I want, but I'm not sure I understand th
06:24 AndreasLutro joined #salt
06:28 rdas joined #salt
06:34 KermitTheFragger joined #salt
06:37 illern joined #salt
06:49 jeddi joined #salt
06:50 trikke joined #salt
06:53 OnTheRock joined #salt
06:54 echo joined #salt
07:01 flyboy joined #salt
07:08 kawa2014 joined #salt
07:15 lb1a joined #salt
07:17 Romlok joined #salt
07:20 wicope joined #salt
07:21 babilen TristanMatthews: Your IRC client unfortunately truncated your message at "I understand the ..."
07:29 andreyst joined #salt
07:30 wicope joined #salt
07:30 andreyst1 joined #salt
07:36 vstoniest joined #salt
07:38 TyrfingMjolnir joined #salt
07:43 grrrrr Anyone else experiencing high memory usage + high load with salt-master on Ubuntu? I have to restart the service from time to time just to keep the load on reasonable levels on that box
07:44 Furao joined #salt
07:45 eseyman joined #salt
07:46 Hydrosine grrrrr: my master process running 2 months now, nothing strange on load or mem. also ubuntu
07:46 otter768 joined #salt
07:47 fredvd joined #salt
07:47 grrrrr it only takes about a week to become like this and i only have few minions (less than 20)
07:49 grrrrr 14.04 ?
07:50 _ale1_ joined #salt
07:50 babilen grrrrr: How beefy is your master?
07:52 Hydrosine grrrrr: yes 14.04, 65 minions.
07:53 grrrrr well it's running on a VM but 2 cores & 2048 megs
07:54 Grokzen joined #salt
07:55 grrrrr i'll test do i just run out of memory ^^
07:55 o5k_ joined #salt
07:57 wicope joined #salt
07:57 hojgaard joined #salt
08:08 markm joined #salt
08:12 iwishiwerearobot joined #salt
08:12 Furao from all manually installed VM in DO to fully installed infrastructure of 30 hosts in EC2 in 6 days with monitoring, self hosted sentry, metrics graphite, centralize logs in graylog, docs, vpn with office
08:13 Furao surprised it took only so few days
08:14 Furao oh i forgot backup too
08:15 malinoff Furao, do you have glassfish or a customer with ssh access that breaks everything every day? :)
08:17 Furao malinoff: nope :) one of them do have a ssh key that grant them access, but they never logged so far
08:17 malinoff Furao, that's why it took only few days
08:17 malinoff my previous project was all about coming to the office in the morning and realizing that I lost my VPN
08:17 malinoff or ssh access
08:18 malinoff or hostnames were changed
08:18 malinoff or ips
08:18 malinoff or whatever you can imagine (and also that you can't)
08:18 Furao :( why they did that?
08:18 malinoff because they could, i guess
08:19 malinoff they had to explain spendings for two devops engineers
08:22 Furao nasty humans
08:22 Furao only trouble
08:22 malinoff true
08:23 malinoff Furao, we have been asking them to give us some time to make everything to shine as their lovely new car, but they have been refusing us all the time
08:24 malinoff Furao, so we ended up with 'task-do-block-repeat' mode
08:24 ckao joined #salt
08:27 wicope joined #salt
08:28 glyf joined #salt
08:28 wicope joined #salt
08:29 wnkz joined #salt
08:34 jrluis joined #salt
08:36 ITChap joined #salt
08:36 N-Mi_ joined #salt
08:40 Furao https://github.com/example42/puppet-orientdb/blob/master/templates/orientdb.init.erb#L25
08:40 Furao why they all do that
08:40 Furao it’s worst than salt-formulas repos
08:42 pf_moore joined #salt
08:45 fredvd joined #salt
08:48 ITChap joined #salt
08:53 Romlok so salt-formula's salt.master state deletes all my custom /etc/salt/master.d/ config files
08:53 ange hi
08:53 Romlok gee, thanks
08:53 wicope joined #salt
08:54 ange salt-cloud : any way to push/update grains from the map & profile onto already running hosts?
08:55 seev I don't use any formulas
08:55 seev other than as abstract inspiration
08:57 signull joined #salt
08:57 Romlok ahahaha, and since salt-formula wiped out my config files, salt can no longer find the salt-formula
08:58 babilen \o/
08:59 seev I have found them useful for patterning my own states though
08:59 seev much help, very full
09:00 Romlok I'm still at the stage of just throwing stuff together until it seems to do what I wanted
09:00 seev I think I have about 2600 lines of states so far
09:00 yawniek when you want  tear up/down machines via salt-cloud but you want to have them to attach specific volumes, where do you handle this? e.g. mounting in a database dir
09:00 seev not counting the templated configs and raw files
09:01 Furao find . -type f -name '*.sls' | xargs cat |  wc -l
09:01 Furao 26847
09:01 Furao ouch
09:01 seev damn, Furao, how many contributors?
09:01 Furao 5
09:02 seev over what span of time?
09:02 Furao > 3 years
09:02 seev OK, I've been at mine ~6 months with just me
09:02 dRiN joined #salt
09:02 seev it's somewhat comparable, I think
09:02 Furao find . -type f | xargs cat | wc -l
09:02 Furao 199073
09:02 Furao oups
09:03 Furao https://doc.robotinfra.com/doc/intro.html
09:03 Furao our formula generate their own doc
09:04 armguy joined #salt
09:04 seev impressive,
09:05 Furao we have a pillar.get proxy function that make sure that every pillar key are documented
09:05 Furao default value is always the same
09:05 jollyroger joined #salt
09:05 Furao similar things for metrics and monitoring checks
09:06 Furao and also I wrote a formula testing framework running the entire test run take ~ 2 days for 1000 units
09:06 Xevian joined #salt
09:06 seev are you Bruno?
09:06 Furao yes
09:07 seev so you're a generic managed IT services company, essentially
09:07 seev I am just running ops for a small startup
09:08 Furao i started as that but my formula become so big it became a product
09:08 jollyroger Hi. I'm experiencing a weird issue when I can get the correct data from pillar using pillar.get but cannot obtain the same data in state sls file. Tried running saltutil.sync_all - nothing changed. Details on files and commands being run: https://gist.github.com/jollyroger/802d80cf664aa5a232de
09:11 yawniek jollyroger: so salt server1 pillar.items   does show all the items right?
09:12 yawniek jollyroger: check http://docs.saltstack.com/en/latest/topics/pillar/ pillar.get() vs salt['pillar.get']()
09:12 jollyroger yawniek: yes, I can get all the correct data from pillar using "salt server1 pillar.items" as well as "salt server1 pillar.get ..."
09:13 echo joined #salt
09:13 yawniek it seems that the : delimiter functionality is not supported for pillar hash
09:14 jollyroger yawniek: right. somehow I missed that note in the docs. Thank you.
09:19 babilen yawniek: You want salt['pillar.get']('nginx:control_vhost:auth_file', '') rather than pillar.get('nginx:control_vhost:auth_file', '')
09:20 babilen The latter simply calls the normal .get() method on the 'pillar' Python dictionary, while salt['pillar.get']() is salt's speshul-super .get
09:20 rdas joined #salt
09:20 babilen Why Python still doesn't support a method to get and set nested values is beyond me though
09:22 clmsy joined #salt
09:23 yawniek babilen: the question was asked by jollyroger, but yeah, python...
09:23 babilen Ah, sorry ... ENICK
09:24 losh joined #salt
09:26 faust joined #salt
09:27 ITChap joined #salt
09:28 favadi joined #salt
09:28 faust left #salt
09:31 mdupont joined #salt
09:39 ujjain joined #salt
09:46 wicope joined #salt
09:47 wicope joined #salt
09:47 otter768 joined #salt
09:47 jollyroger yawniek: babilen: thank you, everything works now. I'm looking forward on merging pillar data (and a sane way to merge python dicts) as well.
09:48 wicope joined #salt
09:55 CeBe joined #salt
09:57 o5k joined #salt
10:04 wicope joined #salt
10:06 ltsampro` hey guys
10:06 zircote joined #salt
10:06 ltsampro` does the python setup.py bdist_rpm works for you?
10:07 marnom joined #salt
10:08 marnom Anyone using iptables state with a different protocol than UDP/TCP? I keep having issues... It seems to create the wrong command line
10:08 sysrich joined #salt
10:18 iwishiwerearobot joined #salt
10:20 __ale__ joined #salt
10:21 XenophonF has anyone else noticed that the saltstack windows minion installers don't match the posted md5 checksums?
10:22 babilen jollyroger: Take a look https://github.com/saltstack-formulas/collectd-formula/blob/master/collectd/map.jinja for one way of handling merging of pillar data and default handling
10:22 babilen brb
10:22 giantlock joined #salt
10:22 XenophonF neither Salt-Minion-2014.7.4-x86-Setup.exe nor Salt-Minion-2014.7.4-AMD64-Setup.exe match the posted checksum
10:23 marnom I just downloaded from http://docs.saltstack.com/en/latest/topics/installation/windows.html and both 32 and 64bit md5sum checks out
10:23 XenophonF i'm looking at http://docs.saltstack.com/downloads/
10:24 marnom ah oke, let me check
10:25 marnom XenophonF, yeah, just checked and they don't match
10:25 Romlok ditto
10:26 XenophonF ok i'm going to file an issue on github - i just noticed that the md5 checksum files are a day older than the installer exe files, too
10:28 marnom XenophonF, yeah probably cause it's not on the website yet it might mismatch because of older version
10:28 marnom The md5 sums for the posted files on http://docs.saltstack.com/en/latest/topics/installation/windows.html work for me
10:28 marnom brb lunch break :)
10:32 matthew-1arlette joined #salt
10:33 echo joined #salt
10:37 AirOnSkin joined #salt
10:39 AirOnSkin Hey guys. How does salt.states.apache_module work on RedHat based systems where there's only a configuration file for HTTPD? Does ist uncomment a line in httpd.conf or is there some kind of command?
10:39 AirOnSkin I don't really get it since the only argument the state requires (or has) is the name of the module...
10:39 jhauser joined #salt
10:46 bluenemo joined #salt
10:47 aquassaut joined #salt
10:50 ltsampros question: can i implement and place custom wheel modules under /srv/_modules/ ?
10:52 XenophonF ah so 2014.7.4 hasn't been released yet, marnom?
10:53 XenophonF i'll hold off deploying these, then :)
10:56 ujjain joined #salt
10:56 marnom XenophonF, As far as I'm aware 2014.7.2 is the latest stable release
10:57 XenophonF i was hoping there was a new version for windows, because some things are broken (like states that download things to the minion over https)
11:00 bhosmer_ joined #salt
11:02 QiQe joined #salt
11:02 SpX joined #salt
11:04 analogbyte is it possible to order state files, so that one file is executed before the other starts its jinja rendering? or do all states render at the same time?
11:05 marnom analogbyte, require?
11:06 analogbyte marnom: requires are parsed only after the jinja in the state is done rendering
11:06 marnom analogbyte, oh oke :)
11:06 analogbyte marnom: afaik at least, but I would be happy to learn different :)
11:07 ktosiek joined #salt
11:08 QiQe guys, Im trying to use this formula
11:08 QiQe http://pastebin.com/gWDE7MyP
11:09 QiQe do you know why user and group of the files extracted arent kibana:kibana
11:09 QiQe -user: and -group: part of archive state seems to be not working
11:10 QiQe any idea?
11:10 QiQe maybe a bug?
11:10 babilen Hmm, you are the second person who seems to have problem with that in archive.extracted
11:10 bhosmer_ joined #salt
11:10 babilen Which hints at the fact that this is undesired behaviour
11:11 keltim joined #salt
11:12 QiQe I've tried also archive_user:
11:12 QiQe http://docs.saltstack.com/en/latest/ref/states/all/salt.states.archive.html
11:12 QiQe it's not working neither
11:13 favadi left #salt
11:16 analogbyte QiQe: there is bug #19260 which you might want to look into/comment on
11:17 analogbyte my problem from before comes down to one question: can I execute stuff before the states jinja is rendered? does anybody know a way?
11:18 denys joined #salt
11:21 babilen analogbyte: What are you really trying to achieve?
11:22 peters-tx joined #salt
11:31 AbyssOne joined #salt
11:32 CeBe joined #salt
11:34 Negher joined #salt
11:34 rome_390 joined #salt
11:34 AirOnSkin How does salt.states.apache_module work on RedHat based systems where there's only a configuration file for HTTPD? Does ist uncomment a line in httpd.conf or is there some kind of command?
11:34 AirOnSkin I don't really get it since the only argument the state requires (or has) is the name of the module...
11:37 buxy joined #salt
11:43 Eureka_ joined #salt
11:43 AndreasLutro AirOnSkin: looks like it doesn't at all: https://github.com/saltstack/salt/blob/develop/salt/states/apache_module.py#L23-L27
11:45 Ixan joined #salt
11:46 teepark_ joined #salt
11:47 bdf joined #salt
11:48 otter768 joined #salt
11:53 ixs joined #salt
11:55 mirko joined #salt
11:57 wnkz joined #salt
12:00 evle1 joined #salt
12:01 Romlok hmm, it would be nice if the salt docs noted in what version features were added
12:01 Romlok since all google searches seem to go to the latest in-development docs
12:05 AndreasLutro Romlok: some of them are noted, you can make a pull request or issue about it
12:05 pdayton joined #salt
12:14 zwi joined #salt
12:15 monkey66 joined #salt
12:16 monkey66 left #salt
12:22 faust joined #salt
12:25 faust left #salt
12:27 cmcmacken joined #salt
12:27 robsavino joined #salt
12:32 keekz joined #salt
12:44 jerematic joined #salt
12:45 glyf joined #salt
12:46 faust joined #salt
12:47 faust left #salt
12:47 dendazen joined #salt
12:54 Romlok hmm, any thoughts as to why my VM master sometimes timeouts trying to highstate its own minion?
12:55 Romlok afaict it does actually perform the highstate, but never reports back
12:58 Xevian joined #salt
12:59 yawniek how do you guys work with ebs volumes to be attached. define them in salt-cloud or attach with some special states?
13:02 numkem joined #salt
13:02 subsignal joined #salt
13:03 racooper joined #salt
13:03 racooper joined #salt
13:04 JDiPierro joined #salt
13:04 jdesilet joined #salt
13:05 ponpanderer joined #salt
13:07 ponpanderer Hello, anyone aware of any known causes that will lead to a salt master (2014.7.1) process to consume 100% cpu and become unresponsive to any salt commands until the salt master is restarted? It seems to reoccur entirely randomly, sometimes after days, other times after weeks. this is on a test salt-master with only a dozen or so minions
13:08 ninkotech joined #salt
13:09 I3olle joined #salt
13:12 fusionx86 joined #salt
13:13 wicope joined #salt
13:13 wicope joined #salt
13:14 wicope joined #salt
13:14 Tecnico1931 joined #salt
13:15 justyns joined #salt
13:15 llua have you tried tracing it when it happens?
13:20 Andre-B joined #salt
13:20 AirOnSkin AndreasLutro: Sorry for my late reply. I was distracted. I thought it might work with the Debian command... in that case I'll work with Pillar and file.replace... thanks :)
13:28 eliasp joined #salt
13:30 ponpanderer @llua: tracing with strace on the running process yields nothing unfortunately
13:30 ponpanderer what's interesting is the it's all system and not user
13:33 notnotpeter joined #salt
13:33 llua need something to trace kernel space then, like systemtap(i've never used it).
13:37 mpanetta joined #salt
13:37 * llua could've worded that a bit better.
13:38 cmcmacken joined #salt
13:40 dyasny joined #salt
13:46 hasues joined #salt
13:47 Tyrm joined #salt
13:47 hasues left #salt
13:47 niseak joined #salt
13:47 slav0nic joined #salt
13:49 otter768 joined #salt
13:51 CeBe joined #salt
13:52 danee hi guys, i'm trying to use salt-ssh with the REST API - is there a way to provide a roster at request time rather than having to specify it on the server?
13:52 StDiluted joined #salt
13:52 danee trying to keep it all as dynamic as possible and i'd really like to be able to specify the servers to run against in the request to the API
13:53 murrdoc joined #salt
13:53 JDiPierro joined #salt
13:53 mpanetta_ joined #salt
13:54 GreyGnome joined #salt
13:54 XenophonF left #salt
13:56 pdayton joined #salt
13:56 murrdoc morning
13:57 ek6 joined #salt
13:58 babilen afternoon, murrdoc
13:58 andrew_v joined #salt
13:58 murrdoc afternoon babilen
13:58 murrdoc where is it after noon ?
13:58 murrdoc london ?
14:02 pdayton joined #salt
14:03 bhosmer__ joined #salt
14:04 echo joined #salt
14:06 babilen Berlin in this case, but most parts of the planet are "after noon" right now.
14:06 renoirb_ joined #salt
14:07 murrdoc :)
14:07 LotR isn't it always about 50/50?
14:07 vieira_ joined #salt
14:07 murrdoc babilen:  want to come work in london ?
14:08 murrdoc we could use some smart peoples in our london office :)
14:09 babilen LotR: That really depends on how you define it, but I had adopted a "the planet is eating up the day" sort of view
14:10 LotR babilen: but it's already 2am in new zealand for example
14:10 babilen murrdoc: I'm always open for interesting ideas, but also not really unhappy here ... feel free to tell me more in private though
14:11 murrdoc :)
14:11 andreyst joined #salt
14:11 babilen LotR: Sure, but that is the next day. There simply aren't that many places that still have to make it "after noon" of 2015-04-07. I completely agree with your 50/50 analysis though
14:11 iggy murrdoc: you never addressed the sources/ vs files/ change
14:12 iggy either by changing the commits or by explaining where that decision came from
14:12 andreyst1 joined #salt
14:12 rvankleeck joined #salt
14:12 rvankleeck any word on the stable date for 2015.02?
14:13 murrdoc iggy:  the original templ sat in the root dir
14:13 iggy my guess of February 65th didn't come to pass, so nfc
14:13 rvankleeck lol @iggy
14:13 murrdoc so i just put the flle in sources directory
14:13 iggy murrdoc: I saw that... that's not what I was asking about
14:13 murrdoc k
14:14 iggy every other formula uses files/ ... for some reason you decided to use sources/
14:14 iggy indeed even the upstream docs use files/ fairly regularly
14:15 murrdoc sure
14:15 murrdoc ton of them use templates too
14:15 murrdoc so i just picked sources
14:15 murrdoc in house i am using sources
14:15 iggy please comment on why you chose that over one of the (multiiple) existing standards
14:16 iggy preferrably on the PR so it's at least somewhere someone can find later on
14:18 murrdoc because files interfere with the tab complete
14:18 iggy ON THE PR!
14:18 murrdoc I AM JUST SAYING THATS WHY I DID IT
14:18 murrdoc SO U CAN SHAME ME HERE FIRST
14:18 murrdoc GAWD
14:19 * murrdoc goes edits the pr
14:19 iggy I'm actually not opposed to changing files/
14:19 yawniek is it possible to specify multiple pip packages in one pip.installed state?
14:19 babilen Yeah, it should be "files/" throughout. I don't like it that we have some with "templates/" but more because it breaks consistency rather than due to the fact that I think templates/ is a bad choice (which it is not)
14:19 zwi joined #salt
14:20 iggy but we need to be consistent, we need to update the existing stuff (not in a commit to master loop), and we need to update the docs
14:20 murrdoc i prefer sources
14:20 * babilen grabs tar and a pitchfork
14:20 murrdoc or templates
14:20 murrdoc files isnt clear to what it is cos its has files and templates
14:20 murrdoc sources is more encompassing
14:21 murrdoc imho
14:21 * murrdoc braces for impact
14:21 murrdoc srsly think about it though
14:21 murrdoc babilen:  the pull request for context https://github.com/saltstack-formulas/sysstat-formula/pull/9
14:22 Brew joined #salt
14:22 babilen ta!
14:22 lietu- joined #salt
14:22 iggy yawniek: doubt it, usually there would be a "names" option (like pkg.installed)
14:23 murrdoc - names is global
14:23 iggy is it now?
14:23 iggy when did that happen?
14:23 murrdoc i *think* 2014.7.2
14:23 murrdoc but i know for sure it works in pkg, file, hosts
14:25 dyasny joined #salt
14:27 iggy the docs don't say (murderous rampage awaits all that don't version annotate their code
14:27 iggy but yes, it appears to be global (as of at least 2014.7)
14:28 iggy yawniek: so try - names:\n  - foo\n  - bar ... etc
14:28 iggy http://docs.saltstack.com/en/latest/ref/states/highstate.html#names-declaration
14:28 murrdoc yeah its good stuff
14:28 elektrix_ iggy and keltim - thanks for the possible directions
14:28 murrdoc and i have no adopted the credo of my russian counterparts
14:28 murrdoc code is doc
14:29 yawniek ooh nice! thanks iggy murrdoc
14:29 iwishiwerearobot joined #salt
14:29 yawniek whats the pkgs option for in pip.installed then ?
14:29 _ale1_ joined #salt
14:29 murrdoc probably the same thing
14:29 murrdoc i d recommend using - names
14:30 LotR even sid only has 2014.1.13...
14:30 iggy it's undocumented... probably for a good reason
14:30 iggy LotR: yeah, don't rely on distro packages to keep up with salt's release "schedule"
14:30 debian112 joined #salt
14:31 yawniek it's undocumented... probably for a good reason  <- lol. good one
14:32 MZAWeb joined #salt
14:32 MZAWeb left #salt
14:33 LotR iggy: getting random packages from uncountable 'upstreams' defeats the purpose of having a distro at all
14:33 murrdoc iggy:  updated pr description and commented
14:34 TooLmaN joined #salt
14:36 mpanetta_ Is there any module/state to update the limits (say number of open files) in sysctl and limits.conf?
14:36 murrdoc yeah
14:37 murrdoc sysctl.present with the config option
14:37 mpanetta_ murrdoc: Thanks!
14:37 murrdoc there is also the sysctl formula
14:38 murrdoc which is fully pillar driven
14:38 murrdoc that u should use
14:38 murrdoc cos its the bees knees
14:38 mpanetta_ haha ok
14:38 murrdoc https://github.com/saltstack-formulas/sysctl-formula/
14:38 murrdoc look at the pillar
14:38 murrdoc thats how easy it is now to add to sysctl
14:38 murrdoc i mean damn
14:39 ekristen joined #salt
14:39 iggy LotR: if that's your position, then I suggest you also get your salt support from upstream... you aren't going to find  whole lot of people here equipped/prepared to support old versions (and in fact 2014.1 is already desupported by SaltStack Ent)
14:40 murrdoc #ssceapproved
14:40 iggy #sscenotdissapproved
14:41 favadi joined #salt
14:41 kaptk2 joined #salt
14:42 sandah joined #salt
14:43 Aikar i have similar question - any saltier way for /sys/kernel/mm/transparent_hugepage/enabled ?
14:43 murrdoc sysctl works with keys in sysctl and files
14:43 Aikar sysctl is for /proc though
14:44 murrdoc ah damnit
14:44 Aikar this one is /sys
14:44 Aikar i currently have rc.local as file.managed
14:44 Aikar but was curious if another way
14:44 murrdoc yeah dont use rc.local :D
14:44 murrdoc sorry
14:44 Aikar thats what everyone recommends for this :P
14:44 Aikar unless theirs another salt way
14:45 murrdoc checking salt code
14:45 murrdoc hold up
14:45 Nils- joined #salt
14:46 LotR iggy: you
14:46 LotR iggy: you're confusing upstream and downstream :)
14:46 pass_by_value joined #salt
14:47 wm-bot4 joined #salt
14:48 mpanetta joined #salt
14:48 murrdoc Aikar:  can u run salt '*' sysctl.get '/sys/kernel/mm/transparent_hugepage/enabled'
14:48 wm-bot4 joined #salt
14:48 murrdoc actually no
14:49 favadi joined #salt
14:49 murrdoc so i cant find a way to do it
14:49 murrdoc now i am sad
14:50 Aikar its ok, its a fringe case (MongoDB)
14:50 ltsampros what a nice database
14:51 wm-bot4 joined #salt
14:51 Aikar so not like im going to need it on tons of servers
14:51 murrdoc hahah
14:51 murrdoc why are u enabling hugepage
14:51 Aikar disabling transparent hugepage
14:51 Aikar echo never > /sys/kernel/mm/transparent_hugepage/enabled
14:51 Aikar something mongo recommends doing
14:52 murrdoc yeah
14:52 Aikar finally upgrading from mongo 1.8 to  3.0
14:52 wm-bot4 joined #salt
14:53 Aikar thankfully mongos design made that so super easy
14:53 Aikar and its a low impact product too
14:53 murrdoc u are already upgrading ?
14:53 murrdoc i am waiting for 3.2
14:53 ltsampros why disable hugepages?
14:54 ltsampros is there proper documentation from mongodb on how it affects them?
14:55 timoguin joined #salt
14:55 ja-s joined #salt
14:55 Aikar transparent hugepages
14:55 moos3 is there a better way for minions to talk to the master besides the default ? because i'm ending up with a ton of minions that salt is say is down that aren't down
14:55 Aikar they use Memory Mapped files, so i dont think HP works for that
14:56 murrdoc the problem is they limit their file sizes
14:56 murrdoc and hugepage throws that off
14:57 Aikar murrdoc, were moving infrastructure, and gotta get everything off old servers asap. its a very low impact backend service, so i doubt w/e 3.2 has will help us any
14:57 murrdoc i am a pessimist when it comes to mongo aikar
14:58 Aikar its just been on my to do list to get everything with this product updated, so this infrastructure move gave the incentive for it (since the code needed updates for newer versions in general)
14:58 Aikar lol why
14:58 murrdoc i havent installed a .0 or a .x.0 ever
14:58 moos3 anyone running a infra with 500 minions with one master ?
14:58 Aikar oh worried about bugs?
14:58 murrdoc yeah
14:58 Aikar did 2.0 have alot of issues?
14:59 echo joined #salt
14:59 Aikar were not doing anything intense or complex at all, so hopefully will be fine
14:59 murrdoc :) it should be
14:59 Aikar just, I had unstructured data to store that I needed ability to also query, so mongo was the right tool for the job
15:00 _ale1_ joined #salt
15:00 vieira_ how do you guys manage your ssh keys with salt? I was trying to structure it with pillar but I think I will have problems because it's not possible to merge keys, right?
15:00 Aikar this product really was meant to be a simple monitoring process that kinda evolved into something useful for production, but has always ran fine. but trying to get its support level up to standards and getting it behind salt
15:01 vieira_ I would like to give a group of users access to all machines, then some other groups may have access to some machine groups, etc
15:01 dfelke joined #salt
15:02 vieira_ any advice?
15:03 murrdoc install consul
15:04 murrdoc put keys in consul
15:04 murrdoc totally kidding
15:04 murrdoc look into yamlex renderer
15:04 murrdoc Aikar:  got it
15:04 murrdoc vieira_:  yamlex renderer will let you merge pillars
15:05 desposo joined #salt
15:05 jonlangemak joined #salt
15:06 vieira_ murrdoc: do you think it is a good fit for managing ssh keys of groups of users across groups of machines?
15:06 vieira_ where a group of machine may have more than one group of users
15:06 mpanetta joined #salt
15:06 vieira_ or is there a better, simpler way to achieve this?
15:07 murrdoc maybe u should look at the users formula
15:07 murrdoc it lets u add/remove users using pillars
15:07 murrdoc that my different machines can specify the users they want in pillars
15:10 dfelke Quick question - I have a yaml configuration file for another product on my servers that needs to be edited by multiple states. What's the best way to handle this? All I can think of is using the file.replace state with a regex but I'm imagining I've missed something for handling yaml natively
15:10 clintberry joined #salt
15:11 StDiluted joined #salt
15:11 ltsampros has anyone successfully wrapped wheel.key functions under _modules/ ?
15:12 dalexander joined #salt
15:12 Aikar erm, surely salt has a yaml state lol?
15:12 scbunn joined #salt
15:13 vieira_ murrdoc: hhmm, I will look into it! thanks!
15:14 dfelke Aikar: Surprisingly not, at least as far as I've seen
15:15 Aikar build everything around yaml but no yaml modifying state, that surprises me
15:15 bmac2 joined #salt
15:16 peters-tx Well I just discovered OpenLMI in RHEL7...
15:16 ajw0100 joined #salt
15:17 bastion1704 joined #salt
15:17 bastion1704 hello, is possible to set IAM role with salt-cloud ?
15:19 stoogenmeyer joined #salt
15:20 elfixit joined #salt
15:20 echo joined #salt
15:22 conan_the_destro joined #salt
15:25 _ikke_ joined #salt
15:29 mpanetta Hmm, I don't suppose there is a recommended way to restart the salt minion is there?
15:29 mpanetta From a state
15:29 murrdoc reboot the machine
15:29 mpanetta heh
15:29 mpanetta Really?
15:29 murrdoc no
15:29 murrdoc its really easy actually
15:30 murrdoc salt-minion-service:
15:30 murrdoc service.running:
15:30 murrdoc - enable: True
15:30 murrdoc - name: salt-minion
15:30 Eureka_ @mpanetta This offers some suggestions. http://humankeyboard.com/saltstack/2013/how-to-restart-salt-minion.html
15:30 murrdoc setup a state like that
15:30 murrdoc and then whenever u need the minion restarted at the end of a state run
15:30 murrdoc throw a
15:30 murrdoc - listen_in:
15:30 murrdoc - service: salt-minion
15:30 murrdoc in the state that needs to tell the minion to reboot
15:31 peters-tx http://docs.saltstack.com/en/latest/faq.html#what-is-the-best-way-to-restart-a-salt-daemon-using-salt
15:31 murrdoc or that
15:31 murrdoc i like my way
15:32 mpanetta_ joined #salt
15:32 mpanetta_ GRR stupid connection
15:33 mpanetta_ Eureka_: peters-tx thanks for the links
15:33 bluenemo_ joined #salt
15:33 Eureka_ =)
15:33 mpanetta_ murrdoc: Hmm, that sounds interesting too
15:33 murrdoc listen_in lets the state run finish before rebooting the minion
15:33 murrdoc if u need instant gratifiation use require or onchanges
15:34 mpanetta joined #salt
15:35 mpanetta Eureka_: The link you sent I know of :)  We already do it that way.  I was hoping that there was a better way that didn't kill the states running
15:35 Xevian joined #salt
15:35 rhodgin_ joined #salt
15:35 mpanetta murrdoc: Does that listen_in work in 2014.1.x?
15:35 murrdoc dont remember
15:36 mpanetta_ joined #salt
15:37 denys joined #salt
15:38 smcquay joined #salt
15:39 SeeDickCode joined #salt
15:42 vieira_ If I have a pillar in test/example.sls and inside that pillar have test: example: A and also have B (a sibling of A), is it possible to, in top.sls, target like 'group1': test.example.A ?
15:42 echo joined #salt
15:42 vieira_ or do I have to create a folder example and have two files A.sls and B.sls?
15:43 mpanetta Hrm it is 2014.7 :(
15:44 Tyrm joined #salt
15:45 funzo joined #salt
15:45 robothands vieira_: you don't need a folder example, you can just have A.sls and B.sls, then target like 'group1': A
15:45 amcorreia joined #salt
15:46 robothands but i dont think you can have 2 pillar values in one file and then use them independently
15:46 robothands maybe im misunderstanding
15:47 vieira_ robothands: it confirms my experience, I thought that maybe it could be achieved in other way
15:47 smcquay joined #salt
15:49 MatthewsFace joined #salt
15:50 otter768 joined #salt
15:53 aw110f joined #salt
15:54 luca__ joined #salt
15:55 favadi left #salt
15:56 CeBe joined #salt
15:57 monkey66 joined #salt
15:58 aw110f_ joined #salt
15:58 iggy LotR: either way... 2014.1 is desupported by salt (and thus by us unless you find some helpful soul)
16:01 iggy moos3: shouldn't be a problem (well aside from the obvious bugs)
16:01 CheKoLyN joined #salt
16:01 moos3 iggy yeah out of my 500 nodes manage.down at any given point in time claims 100 of them down
16:01 iggy mpanetta: listen is 2014.7+
16:02 mpanetta iggy: Yeah I see that, thank you :)
16:02 iggy sorry, was skimming the backlog
16:02 mpanetta iggy: Nothing to be sorry about!
16:03 iggy and by skimming, I mean reading every 3/4 lines
16:03 mpanetta haha
16:03 teskew joined #salt
16:04 echo joined #salt
16:05 rhodgin_ joined #salt
16:06 murrdoc mpanetta
16:06 murrdoc upgrade
16:07 murrdoc its your only option
16:07 mpanetta We will eventually
16:07 iggy 2015.2!
16:07 ipmb joined #salt
16:07 murrdoc or u know event.fire to a reactor
16:07 murrdoc and let teh reactor restart the minion
16:07 mpanetta iggy: That one is out now?
16:07 iggy moos3: what's your network look like?
16:07 iggy mpanetta: no :/ They totally lost me money on February 65th too
16:08 mpanetta haha
16:08 moos3 a mix of 1G and 10G host, pretty flat, no vlans if thats what your asking
16:08 iggy moos3: so not like a bunch of routing or NAT or etc... There goes my idea
16:09 rm_jorge joined #salt
16:10 desposo joined #salt
16:11 quintinadam joined #salt
16:11 moos3 nope
16:12 rm_jorge buenas
16:12 dalexander joined #salt
16:13 iggy moos3: do you have the ability to test 2015.2 in a test env? It has some connectivity improvements and it also has the ability to run using raet vs zeromq
16:13 iggy moos3: I suppose you've done all the normal stuff of upping open files, etc?
16:15 moos3 yeah i have upped the open files
16:17 smcquay joined #salt
16:19 wendall911 joined #salt
16:20 KyleG joined #salt
16:20 KyleG joined #salt
16:21 aparsons joined #salt
16:22 cmcmacken joined #salt
16:22 rhodgin_ joined #salt
16:25 scbunn joined #salt
16:26 echo joined #salt
16:27 micko joined #salt
16:28 JDiPierro joined #salt
16:28 aparsons_ joined #salt
16:30 enarciso joined #salt
16:31 briantrice joined #salt
16:31 aparsons joined #salt
16:32 aparson__ joined #salt
16:37 _JZ_ joined #salt
16:38 ek6 someone asked me the other day when i thought 2015.2 would be released... i responded with 'well 2015.2 means the second season of 2015 so sometime in the spring'
16:38 sxar joined #salt
16:39 bhosmer_ joined #salt
16:39 murrdoc joined #salt
16:40 spookah joined #salt
16:41 MatthewsFace joined #salt
16:42 rhodgin_ joined #salt
16:45 amcorreia joined #salt
16:47 echo joined #salt
16:48 stanchan joined #salt
16:49 writtenoff joined #salt
16:50 noway__ joined #salt
16:55 theologian joined #salt
16:56 zircote joined #salt
16:57 Aikar why does ssh known hosts print an error even though it works successfully: [ERROR   ] Command 'ssh-keygen -F "bitbucket.org" -f "/root/.ssh/known_hosts"' failed with return code: 1
16:57 zircote joined #salt
17:01 mdupont joined #salt
17:02 Grokzen joined #salt
17:03 rhodgin_ joined #salt
17:04 stoogenmeyer joined #salt
17:05 debian112 joined #salt
17:09 davet1 joined #salt
17:09 Guest70 joined #salt
17:09 echo joined #salt
17:13 dude051 joined #salt
17:15 helderco joined #salt
17:16 neolastic joined #salt
17:17 druonysus joined #salt
17:19 giantlock joined #salt
17:23 Vynce joined #salt
17:23 rickflare so I have a map.jinja file http://pastebin.com/CKw80fkT here and when I run my formula I get the following error.  Rendering SLS 'base:firewalld' failed: Jinja variable 'None' has no attribute 'server' Can someone help me troubleshoot what I am doing wrong?
17:24 Vynce1 joined #salt
17:25 Vynce1 where's the rest?
17:25 jalbretsen joined #salt
17:25 rickflare one sec
17:26 rickflare http://pastebin.com/Fr4cLSQq is my init.sls
17:26 rickflare service sls http://pastebin.com/FBwYBkNA
17:27 rickflare defaultconfig.sls http://pastebin.com/txYeLLAm
17:27 Andre-B joined #salt
17:27 rickflare thats all that make up the formula
17:28 Aikar Can someone tell me what I'm doing wrong: https://gist.github.com/aikar/82142585406c9dddf5a2 - we broke out the fpm/cli deps so we can selective run other states that depend on them w/o also running the php modules state every time (git pull states). I'm trying to make the FPM mcrypt module only activate in setups that use fpm, but thise code results in neither of the mcrypts being symlinked. why is the fpm and cli files not setting variable correctly?
17:30 Aikar trying to install only cli on a server and it blew up on the fpm mcrypt, but adding the if block isnt working
17:30 Aikar and i really dont want to break that out more
17:31 echo joined #salt
17:31 stoogenmeyer joined #salt
17:31 Aikar the file referencing both does load cli first
17:31 stoogenmeyer joined #salt
17:32 ona_matt joined #salt
17:33 iggy rickflare: protip: gist.github.com allows you to paste multiple files at once
17:33 rickflare ahh
17:33 rickflare ok
17:33 ona_matt Is there a way to use .saltrc to specify the username and the authentication system?  I would like to not have to specify —username and -p pam each time.
17:33 rickflare ill use that next time
17:33 nzero Aikar, do the variables get carried over from one sls file to the next? maybe the variables are only visible in the file they are defined in
17:33 rickflare iggy do you know what is causing that error?
17:33 Aikar nzero, that might be my issue, so i guess the question is how do i declare a global variable?
17:34 iggy rickflare: I'm still waiting for the 18 pastebin tabs to open
17:34 iggy and all the ads... zomg
17:34 nzero Aikar, try looking at import
17:34 denys joined #salt
17:34 nzero http://docs.saltstack.com/en/latest/ref/renderers/all/salt.renderers.jinja.html
17:35 tomh- joined #salt
17:35 nzero Aikar, or use Pillar
17:36 nzero http://docs.saltstack.com/en/latest/topics/pillar/
17:36 Aikar goal really is for 1 state to be aware of other states heh
17:37 iggy rickflare: is it Centos or CentOS?
17:37 nzero Aikar, don't have experience with that goal. maybe someone else can comment about how others handle that
17:37 elektrix_ another question about environments and best practices...say I have directories /srv/salt/{base,dev,prod} and have a formula /srv/salt/base/foo. If I want to make significant changes to foo, should I (manually) copy it into /srv/salt/dev, get it working, then copy back to base?
17:38 nzero iggy, CentOS
17:38 rickflare ahh
17:38 rickflare dam
17:38 rickflare well
17:38 rickflare i set it CentOS
17:38 rickflare same error
17:39 elektrix_ trying to figure out the cleanest way to have things under version control, make changes to things in base w/o messing w/ production, and not have to modify sls files when moving them between environments
17:39 iggy the other shitty thing about pastebin is it jacks up whitespace
17:39 rickflare ok
17:39 rickflare let me put it in that other format for you
17:39 iggy so I can't even really tell if your states are valid yaml
17:39 ajw0100 joined #salt
17:39 TooLmaN joined #salt
17:40 evle1 joined #salt
17:41 iggy Why are you trying to use map.jinja on a service that only runs on RH based systems?
17:42 iggy map.jinja is generally used for packages that run on multiple distros but have differences (service name, package name, etc.)
17:43 rickflare https://gist.github.com/bjames301/b03d43dbe60f31e39da6
17:43 rickflare well
17:43 rickflare i am going to add defs
17:43 rickflare for ubuntu
17:43 rickflare but for now
17:43 rickflare i was just testing on Centos
17:43 rickflare but yes I will be supporting multiple distros
17:45 iggy but everything is the same
17:45 iggy the package name, the service name... all the same
17:46 murrdoc use the defaults.yml way
17:46 murrdoc check out the sysctl formula for the goodness
17:46 jhauser joined #salt
17:46 iggy or don't even bother with it
17:46 rickflare I know
17:46 rickflare this is more of me learning formulas
17:46 iggy the aptly-formula doesn't have a map.jinja... because it doesn't need it
17:46 rickflare when I do have different names
17:47 rickflare right
17:47 rickflare but if i had ubuntu
17:47 iggy (and if anybody tries to add it without good reason, I'll go out of my way to block it)
17:47 rickflare and gentoo in there
17:47 rickflare it would
17:47 rickflare I know its not best practice
17:47 rickflare i need to understand
17:47 rickflare why its failing
17:47 rickflare with that error
17:48 rickflare its not very discriptive
17:48 kunit joined #salt
17:48 iggy it's fairly descriptive... firewalld is None... which means your map isn't working
17:49 Aikar iggy, you always have the answers, do you know a way to solve my issue?
17:49 iggy try salt-call -l debug and see how it's rendering stuff
17:49 murrdoc or salt state.show_sls
17:50 rickflare when I run salt-call do I need to pass the name of the formula?
17:51 murrdoc sate.sls formulaname
17:51 otter768 joined #salt
17:51 murrdoc or state.apply if u l33t like iggy
17:51 elektrix_ since defaults.yaml vs map.jinja came up, is there a good reason to use one over the other? or specific use cases for each?
17:51 murrdoc #ssceapproved
17:51 iggy elektrix_: they go together... check examples
17:51 elektrix_ k, I'll dig around
17:51 murrdoc sysctl formula
17:51 murrdoc ftw
17:52 murrdoc right ?
17:52 rickflare oh geeze I dont know what im doing
17:52 iggy salt, collectd, sysctl, etc.
17:52 rickflare im such a newbie
17:52 murrdoc salt too
17:52 rickflare so salt-call -l debug init.sls firewalld?
17:52 echo joined #salt
17:52 iggy salt-formula still has some cleanup left
17:52 iggy but it's a good start
17:53 iggy rickflare: salt-call -l debug state.sls firewalld
17:53 rickflare what does state.sls do?
17:53 rickflare esp since I dont have a state.sls file
17:54 rickflare [CRITICAL] Rendering SLS 'base:firewalld' failed: Jinja variable 'None' has no attribute 'server'
17:54 evle3 joined #salt
17:55 iggy it's a module... and you have a circular dependency... service includes init, init includes service
17:55 dyasny joined #salt
17:55 rickflare ahhhh
17:55 iggy and Idon't think you want CentOS there
17:56 rickflare well I will
17:56 rickflare because im going to add gentoo
17:56 rickflare debian
17:56 rickflare arch
17:56 rickflare suse
17:56 rickflare next
17:56 iggy it's the output of sudo salt-call grains.item os_family
17:56 iggy which I think for CentOS is still something about redhat
17:56 rickflare ahh
17:56 Aikar is there a global I can safely tack on a variable to?
17:56 rickflare so I should just use Redhat?
17:57 iggy Aikar: no
17:57 iggy rickflare: run the command I mentioned and use that
17:57 rickflare its Redhat
17:57 jerematic joined #salt
17:58 Aikar iggy, do you have any suggestions for me then? I want a shared php/modules state to share between cli and fpm, and i want to run the rpm module symlink only if the fpm state has been included too, and vice versa for cli.
17:58 Aikar fpm*
17:58 DaveQB joined #salt
17:58 iggy set a grain?
17:58 Nazzy joined #salt
17:58 iggy need more context... but meeting
18:00 Aikar im namely trying to avoid doing an fpm modules vs cli modules, or also avoid moving the modules symlink into the fpm/cli states. I want to keep all php module management into 1 state, but for example symlinking the FPM mcrypt module when FPM isn't included, fails.
18:00 Aikar i thought i could set a variable in the fpm file and cli files, then read them in the module file, but it doesnt see them
18:02 baweaver joined #salt
18:06 sc` left #salt
18:06 aparsons joined #salt
18:06 lietu joined #salt
18:06 stoogenmeyer joined #salt
18:06 rickflare iggy
18:06 rickflare thank you very much that seems to have fixed my issue
18:06 stoogenmeyer joined #salt
18:07 murrdoc joined #salt
18:08 * SheetiS has a php + php-fpm module that allows for quite a bit of flexibility with modules.  It's currently only build against Amazon Linux as I haven't had a chance to expand it past that, but I could share it if you wanted, Aikar.  The pillar lets you set things for stuff that is used in /etc/php and /etc/php.d for cli stuff as well as /etc/php-fpm.d or whatnot for fpm configurations.
18:10 o5k__ joined #salt
18:14 echo joined #salt
18:15 o5k joined #salt
18:15 Aikar sounds a bit more complex than were going for, but i could look at it. I just fixed my issue by simply asdding - unless: test ! -d /etc/php5/fpm
18:16 Aikar and ensuring fpm and cli have require_in: set for modules
18:18 SheetiS yeah this would have 100% of the configuration in the pillar like so: https://github.com/rfairburn/salt-php-formula/blob/master/pillar.example.  It'd do requirements for modules kinda like this: https://github.com/rfairburn/salt-php-formula/blob/master/php/modules.sls
18:19 nk joined #salt
18:19 SheetiS I'm not sure any of that is a 'good' way to do it, but it looked to work universally for me.
18:19 murrdoc its ugly
18:20 SheetiS It would require a bit of love to do the debian style folders with symlinks vs the redhat-style configs that you see in RHEL/CentOS/Amazon Linux, but mostly the same.
18:20 murrdoc please rewrite that
18:20 murrdoc u should be ashamed of yo self
18:20 murrdoc gawd
18:20 * murrdoc rage quits
18:20 Aikar so salt created home dir owned by root, surely the default behavior should be to chown it to the user being created?
18:20 murrdoc that looks pretty good actually
18:22 robawt Aikar: you shoudl always specify a user/group/mode in any directory you manage with any config tool
18:23 Aikar yeah majority of php config is something youd really never care to change, wouldnt want pillar so full of stuff you know wont change.
18:24 Aikar robawt, i just expect it to behave like useradd -m
18:25 Aikar so you have to do another state just to fix the ownership :/ ?
18:26 Aikar think i rather use useradd with an unless clause
18:26 sjwoodr joined #salt
18:26 TaiSHi joined #salt
18:26 sjwoodr Hey all... i'm migrating from 2014.7.1 -> 2015.2 rc2 and am encountering a strange error when running any state.sls that includes a particular sls file
18:26 sjwoodr Requisite types must not contain dots, did you mean 'file'?
18:27 sjwoodr i looked at salt/state.py in 2014.7.1 -vs- the 2015.2.0 rc2 and it seems this is a new check
18:27 sjwoodr can you no longer have dots in the ID ?
18:27 SheetiS sjwoodr: if you do a - require:\n and then have something
18:27 SheetiS it shouldn't have a . in it
18:27 SheetiS you don't require - file.managed, just - file:
18:27 sjwoodr so i can have a file that is simply:   include:\n  - packages.installers
18:27 sjwoodr its just included ... not a require:  and its giving this error
18:29 njlg joined #salt
18:29 SheetiS can you share the state in a pastebin/refheap/gist?  packages.installers might be good to look at as well since it is included (assuming that packages.installers refers to /path/to/packages/installers.sls).
18:29 Vynce left #salt
18:30 sjwoodr yeah. let me double check these files and then i'll give a link to pastebin or something.  stand by.  :)
18:31 dyasny joined #salt
18:31 f8a316f8 joined #salt
18:32 timoguin joined #salt
18:34 sjwoodr ok @sheetis: http://pastebin.com/UrpqL8Fp
18:34 SheetiS - require:\n - file.directory: /root/.ssh
18:34 SheetiS those lines
18:34 SheetiS 14 and 15
18:34 SheetiS make that just - file:
18:35 SheetiS instead of file.directory
18:35 sjwoodr doh.
18:35 njlg is there a good way to test all changes from a git branch without affecting the salt master? i want to try to trigger something from a pull request that will run 'salt '*' state.highstate test=True' against all of our environments for that particular branch and post the results to the PR. anyone else do something similar?
18:35 sjwoodr is this a change in 2015.2 ?  because this is an old state we've had for ages
18:35 SheetiS It's been correct to do as just - file: since I've used salt (2014.1) and documented as such even in 0.17 from what I can see.
18:36 echo joined #salt
18:37 sjwoodr yeah i see the check for dot in the key was added after 2014.7.1 .... so i'll update my state files.
18:37 sjwoodr thank you @sheetis !!  (again)
18:41 mpanetta joined #salt
18:44 elektrix_ iggy: figured out correct usage of map.jinja and defaults.yaml after looking at the salt example. thanks
18:44 JordanTesting joined #salt
18:44 Aikar oh i think my vagrant file share created the dir owned by root
18:44 baweaver joined #salt
18:45 jhauser joined #salt
18:47 babilen elektrix_: You don't necessarily have to keep your defaults in a yaml file, check out: https://github.com/saltstack-formulas/collectd-formula/blob/master/collectd/map.jinja for a different approach
18:48 elektrix_ babilen: yeah, that's how I originally had it
18:49 f8a316f8 left #salt
18:49 babilen ah, right
18:49 elektrix_ trying to figure out the "best" way to do a lot of things, but it looks like there is a lot of "best" ways
18:49 murrdoc there is only one way
18:50 murrdoc the #ssceapproved way
18:50 Darod joined #salt
18:50 elektrix_ I initially did my map.jinja based on this: https://github.com/saltstack-formulas/influxdb-formula/blob/master/influxdb/map.jinja when is how the collectd formula does it as well
18:51 elektrix_ figuring out how to go from dev -> test -> prod using version control
18:51 Darod Greetings, I am trying to create a comma sepparated file using pillar and a jinja template but i cannot get it to work: https://gist.github.com/anonymous/9c2850eaf2749236c409
18:52 elektrix_ I read somewhere to put the top.sls in the base (and that's it), then put all other formula in the dev,test,prod directories
18:52 elektrix_ might try that approach unless I find something better
18:53 castro joined #salt
18:53 castro Hi, has anyone ever had a problem with ioflo.base?
18:53 castro I can't start salt-master.service because I can't import nonblocking
18:54 nzero Darod, with that for loop you are calling join on a single string not a list
18:55 nzero if you want it on one line, off the top of my head, try something like salt['pillar.get']('names').join(',')
18:57 Tyrm_ joined #salt
18:58 echo joined #salt
18:58 AirmanAirman joined #salt
18:58 niseak_ joined #salt
18:59 echo_ joined #salt
19:03 Darod @nzero, that does not work, gives error message: Unable to manage file: Jinja variable 'list object' has no attribute 'join'; researching
19:03 nzero try the pipe symbol instead of the dot
19:04 niseak joined #salt
19:04 nzero http://jinja.pocoo.org/docs/dev/templates/    Filters that accept arguments have parentheses around the arguments, just like a function call. For example: {{ listx|join(', ') }} will join a list with commas (str.join(', ', listx)).
19:06 ajw0100 joined #salt
19:06 babilen Darod: You'd have to make that ','.join(some_list) in Python, but then there is also the jinja filter that you can use right awaay
19:10 perfectsine joined #salt
19:16 Darod Reading and trying to understand this; have not yet gotten the aha! moment
19:18 babilen Darod: There are essentially two ways of calling things in jinja. On the one side you have jinja filters that are documented on http://jinja.pocoo.org/docs/dev/templates/#list-of-builtin-filters
19:18 nzero filters: salt['pillar.get']('names')|join(',')
19:18 babilen On the other hand you are given just 'normal' Python objects and you can, naturally, simply call their methods directly
19:18 nzero python: ','.join(salt['pillar.get']('names'))
19:18 babilen |join(',') is the correct way to call the join *filter* on the object before the |
19:19 babilen (as wonderfully exemplified by nzero)
19:19 nzero (and wonderfully explained by babilen )
19:19 dendazen I am using this formula in salt master: https://github.com/saltstack-formulas/resolver-formula/blob/master/resolver/files/resolv.conf
19:19 dendazen this line search {% for searchpath in searchpaths %}{{ searchpath }} {% endfor %} in jinja template
19:20 dendazen finds correct prefix.
19:20 dendazen but line in the file is created like this “search s e c 1 . v e l i o s y s t e m s . c o m”
19:21 dendazen why is it doing that?
19:21 Darod ok so having tried |join(',') as output i get comma separated names; but every character is on its own line..
19:21 babilen dendazen: You seem to pass a single string and not a list, could you paste your pillar? I guess that you are simply missing a "-" in front of sec1.veliosystems.com
19:22 babilen Darod: Are you sure that you are joining a list?
19:22 dendazen i mean you can see the whole formula in the link above
19:22 KyleG Does anyone know how I can specify which interface to pull the ip_interfaces grains from? For example, I want to return the IP for igb0
19:22 dendazen it is public
19:22 babilen dendazen: Do you have resolver:searchpaths defined in your pillar?
19:23 nzero darod, are you still doing a for loop? you dont need that
19:23 nzero maybe reshare your code
19:23 dendazen searchpaths: {{ salt['pillar.get']('resolver:searchpaths', salt['grains.get']('domain')) }}
19:23 dendazen that is what i have.
19:25 babilen dendazen: Do you define a pillar? That entry looks wrong as salt['grains.get']('domain')) would return a single string and not a list, so looping over that obviously loops over the characters in the string
19:25 baweaver joined #salt
19:26 Darod nzero: Ok i changed my jinja template to "{{ salt['pillar.get']('names')|join(', ') }}" only. Output is now "buffy, willow, faith". Im new to salt: what i am not understanding yet why i did not need the loop anymore?
19:26 niseak joined #salt
19:27 vieira_ hello, it is a known problem that ssh_auth.absent does not work with source:
19:27 vieira_ ?
19:28 nzero darod, the join function joins a list already. by using the for loop you are breaking the list into individual items
19:28 nzero if you join a string, then it breaks the string into a list and tries to join it
19:28 vieira_ I have found this closed report https://github.com/saltstack/salt/issues/3817
19:28 vieira_ but I would assume this did not make into 2014.7.2
19:29 Pixionus joined #salt
19:30 Darod ok and this join function is that salt, jinja or python; as in where could i read up more on this :)
19:31 babilen dendazen: Could you file a bug?
19:34 I3olle joined #salt
19:34 markm joined #salt
19:34 vieira_ seems that the patch did not make into 2015.2 either :(
19:38 ja-s joined #salt
19:39 timoguin joined #salt
19:41 chandankumar joined #salt
19:43 CeBe joined #salt
19:44 leapingDeere joined #salt
19:45 leapingDeere Some states in my highstate are not running and I can't figure out why.  Any suggestions?
19:46 KyleG leapingDeere: Up the logging to debug on the minion and/or master
19:46 iggy vieira_: no, and it shouldn't... 2014.7 shouldn't be getting new features (although it sometimes does)
19:46 KyleG It'll tell you everything it's doing or running into
19:47 iggy Darod: check jinja docs first and python docs next
19:48 germs_ joined #salt
19:48 Darod Ok thank you. You too nzero, babilen.
19:48 leapingDeere KyleG: I've tried that. Logs don't show anything about the states that are missing.  If I run state.show_highstate the states in question don't get listed either, but they are indeed in the top.sls.
19:49 raygunsix joined #salt
19:49 KyleG weird
19:49 dendazen do you think it is a bug?
19:49 babilen yes
19:50 dendazen ok i will can i fix it temporary in any way?
19:50 dendazen in jinja join.(‘ ‘) or some sort.
19:51 iggy dendazen: pass a list in a pillar
19:51 dendazen this is my pillar
19:51 dendazen searchpaths: {{ salt['pillar.get']('resolver:searchpaths', salt['grains.get']('domain')) }}
19:52 dendazen salt['grains.get']('domain')) is the string?
19:52 otter768 joined #salt
19:52 iggy you can't pillar.get in a pillar
19:52 vieira_ iggy: ok thanks! any hope for it to be in 2015.2?
19:53 iggy dendazen: searchpaths:\n  - salt['grains.get']('domain')
19:53 I3olle joined #salt
19:53 babilen dendazen: That isn't your pillar, that is what is being used in the state
19:53 dendazen i am not, i did not write this formula it is public in master.
19:53 babilen Yes, so pass your own pillar
19:54 * iggy gets the clue-by-four
19:54 JDiPierro joined #salt
19:55 iggy but that is a bug
19:56 Guest70 joined #salt
19:56 fusionx86 joined #salt
19:57 dendazen Ok i will file the bug
19:57 iggy I already submitted a pull request
19:57 murrdoc ext_pillar shits broke
19:57 murrdoc in 2014.7.2
19:58 dendazen K
19:58 iggy murrdoc: works fine in 2015.2
19:58 * iggy runs
19:58 baweaver joined #salt
19:59 * murrdoc steals iggys  sniper gun , lines up shot
19:59 ek6 dont forget to lead him a bit
19:59 Guest70 Darod, join is in both python and jinja template filters. you can look in both places. if you are making jinja templates, probably more traditional to use filters
20:00 zmalone joined #salt
20:01 JDiPierro Is there any way I can define bash aliases that would be available for remote execution with salt? Right now I've got salt appending some aliases to /bash/bash.bashrc and that makes them available to every user, including root.. but when I do a cmd.run through salt it doesn't have the aliases.
20:01 babilen iggy: merged
20:01 babilen dendazen: Could you try again?
20:01 aparsons joined #salt
20:01 dendazen yeah i am
20:01 murrdoc self.ext_pillars = salt.loader.pillars(ext_pillar_opts, self.functions)
20:02 murrdoc that broked
20:02 murrdoc i wish we could shit simple
20:02 dendazen great
20:02 dendazen search sec1.veliosystems.com
20:02 dendazen worked
20:02 dendazen Thanks, guys.
20:03 khris joined #salt
20:04 iggy that was fast turnaround
20:04 zmalone Is anyone using Salt on Solaris?  The most recent version available from OpenCSW (as recommended in http://docs.saltstack.com/en/latest/topics/installation/solaris.html ) is 0.14.
20:04 iggy (and I'm glad it actually worked... wasn't sure if I needed that trailing comma or not
20:05 iggy zmalone: sadly... probably not many
20:05 iggy zmalone: tried pip install instead?
20:05 nahamu zmalone: someone at SaltStack is working on builds of it.
20:05 nahamu I use it on SmartOS quite happily.
20:06 JDiPierro Even tried manually sourcing the alias file during remote execution.. still can't use the aliases: http://pastebin.com/MAwPAFSa
20:06 Negher left #salt
20:06 babilen iggy: You don't, but it doesn't hurt either
20:06 iggy sweet
20:06 zmalone SmartOS has a decent toolchain though, I'm trying to squeeze Sol9/10 into a Salt project.  I guess I'll need to try it myself then.
20:06 nahamu zmalone: my sense is that they have some paying customers who want to run it on S10 and S11. I don't know how far along they are in getting official builds out.
20:06 zmalone Thanks!
20:07 nahamu Sol9, huh?
20:07 nahamu good luck with that one... :)
20:07 iggy what version of python does that have?
20:09 nahamu if you have enough tooling to bootstrap a pkgsrc build environment you might be able to get yourself all the way up to some decent tooling...
20:09 anotherZero joined #salt
20:09 andrew_v joined #salt
20:10 StDiluted joined #salt
20:11 signull hey quick question. what is the syntax to target a role? salt 'G@role:<something>' test.ping
20:11 signull tried that, but it didnt work
20:12 babilen You are missing "-C" to say that it is a compound matcher
20:12 murrdoc salt —help
20:12 signull thanks got it
20:12 iggy or just salt -G 'role:foo' test.ping
20:13 signull salt -I 'role:web' test.ping
20:13 signull yeah, no idea why i drew a blank
20:13 giantlock joined #salt
20:13 iggy -I is pillar G@ is grain
20:15 XenophonF joined #salt
20:16 JDiPierro So no love for defining aliases to be used by cmd.run?
20:17 sjwoodr i'd be interested in a solution for that.
20:17 sjwoodr don't have a need for it today.... but sounds useful ;0
20:17 iggy use a script
20:20 nzero dirty hack, write a script that loads your aliases and then executes whatever you passed as the arguments
20:20 sjwoodr thats what we do now - just have a wrapper script.   (shrug)
20:20 sjwoodr i have a git repo for devops tools that i deploy to the minions and i can cmd.run stuff out of there.
20:20 sjwoodr but cmd aliases in pillar data would be interesting..
20:21 XenophonF so i have this salt-cloud profile definition - https://bpaste.net/show/defb6e3af8e6
20:22 nzero salt would have to call commands differently so it could load the commands via a shell that itself loaded the aliases. essentially the same thing. you could extend cmd state to something like cmd.shell so it loads a full shell
20:22 XenophonF yet when try to deploy using the windows profile, it doesn't appear to have the powershell script set in the userdata
20:23 XenophonF so it never opens port 445/tcp in windows firewall
20:25 XenophonF what am i doing wrong?
20:25 iggy using windows?
20:26 iggy sorry, couldn't resist
20:26 ek6 iggy: the joke that never gets old
20:26 XenophonF heh
20:29 Karlthane Getting started with salt and configuration management, is it save to put the ssh authorized_keys file up on github for all my minions to pull down?
20:29 WonderSlug joined #salt
20:30 ek6 pub keys are pub keys...worst your doing is telling folks is which private keys they need to steal and where to use them
20:30 zmalone authorized_keys would be safe, private keys are not.
20:31 Karlthane Ok thank you. I thought it would be safe, but wanted to make sure.
20:33 ek6 karlthane: insert blob here about how you can always use bits like the gpg renderer if you get paranoid
20:34 swazman joined #salt
20:37 MTecknology When I run something like this  salt -v -b 8 -C '* and not boothost.* and not ldps.*' cmd.run 'telinit c', I end up with a server that shows up twice.   "mysql1.domain.tld Detected for this batch run"  will be there twice and because it won't return twice, I'm left with batch runs never completing. I'm confused. :(
20:38 MTecknology It shows up twice every time, for things like test.ping as well
20:39 vstoniest joined #salt
20:40 pipeep joined #salt
20:42 MTecknology heh... Apparently it's sent twice for non-batch jobs as well, just to that one minion
20:43 vstoniest joined #salt
20:45 nzero is the agent running twice somehow?
20:46 MTecknology heh.. apparently it was
20:47 nzero =)
20:47 mpanetta joined #salt
20:47 nzero funny how these things tend to do exactly what we tell them
20:47 XenophonF here's a transcript of the salt-cloud call - https://bpaste.net/show/3ce7a20b8cf4
20:47 MTecknology yup, weird stuff
20:49 XenophonF the call to RunInstances is at line 93
20:49 XenophonF but i don't see it including the userdata in the call
20:49 XenophonF what do?
20:49 XenophonF anyone else using salt-cloud with windows ec2 instances?
20:50 baweaver joined #salt
20:50 zer0def joined #salt
20:51 theologian joined #salt
20:52 XenophonF oh man - i can on see a reference to userdata_file in salt.cloud.clouds.nova and salt.cloud.clouds.openstack
20:52 XenophonF "i can only"
20:54 XenophonF dammit, it's only in the 2015.2 branch
20:57 * murrdoc slides XenophonF a beer
20:58 XenophonF thanks murrdoc
20:58 rhodgin_ joined #salt
20:59 ndrei joined #salt
20:59 murrdoc should saltutil.clear_cache and saltutil.sync_grains be part of a state that restarts a minion
20:59 murrdoc and why is the answer yes
20:59 murrdoc because eventual consistency
21:00 murrdoc brain is no fried
21:00 * murrdoc snatchers another beer
21:05 glyf joined #salt
21:08 MTecknology Heh.. this is interesting. "[ERROR   ] Got a bad pillar from master, type str, expecting dict: "
21:10 fredvd joined #salt
21:13 swazman joined #salt
21:16 murrdoc how does one 'touch' files in salt
21:16 seev very gently
21:16 murrdoc :D
21:17 nkuttler file.touch
21:18 opnsrc joined #salt
21:19 Vye seev, very quickly :)
21:20 opnsrc i'm attempting to debug *random* "Source file salt:// - not found errors"   - I know the master exists and the file(s) are available, but randomly a minion fails with these errors.  Any ideas?
21:21 opnsrc salt version 2014.07.02, btw
21:21 _ale1_ joined #salt
21:22 ndrei joined #salt
21:26 signull joined #salt
21:28 jhauser joined #salt
21:30 murrdoc Note that if you just want to create a file and don't care about atime or
21:30 murrdoc mtime, you should use ``file.managed`` instead, as it is more
21:30 murrdoc feature-complete. (Just leave out the ``source``/``template``/``contents``
21:30 murrdoc arguments, and it will just create the file and/or check its permissions,
21:30 murrdoc without messing with contents)
21:32 baweaver joined #salt
21:37 pravka joined #salt
21:41 igorwidl joined #salt
21:42 Tyrm joined #salt
21:42 lnr joined #salt
21:42 lnr joined #salt
21:42 XenophonF OK so linux deployments via salt-cloud aren't working either
21:43 lnr joined #salt
21:43 XenophonF it looks like it finished installing a centos 7 ec2 instance, and salt-key lists an accepted key for it, but test.ping doesn't return anything
21:43 lnr joined #salt
21:43 lnr joined #salt
21:44 lnr joined #salt
21:44 lnr joined #salt
21:45 lnr joined #salt
21:45 lnr joined #salt
21:45 lnr joined #salt
21:46 lnr joined #salt
21:46 lnr joined #salt
21:47 __ale__ joined #salt
21:47 XenophonF here's the debug output
21:47 XenophonF https://bpaste.net/show/7e1196931ee4
21:47 KyleG Did you wait a minute or two before trying test.ping ?
21:49 XenophonF yes
21:49 XenophonF i just logged into the guest
21:49 XenophonF the salt-minion isn't installed
21:49 XenophonF nothing's running
21:49 XenophonF what am i missing?
21:50 XenophonF i see in the debug log where it connected to the minion over ssh
21:52 otter768 joined #salt
21:52 opnsrc i am using file.managed, however with `source`
21:53 PintoBean joined #salt
21:53 opnsrc i though `source` was needed as the file being managed differs in name on the salt master
21:53 jalaziz joined #salt
21:53 opnsrc for ex:
21:54 opnsrc `/etc/some.conf:`
21:54 XenophonF opnsrc: i've only ever used the source or contents_pillar arguments
21:54 opnsrc `file.managed:`
21:55 opnsrc ` - source: salt://salt/dir/to/desktop-some.conf`
21:55 XenophonF this is so strange - salt-cloud uploads a minion key pair and then immediately deletes it
21:56 opnsrc in the salt master log, i see 'Source file salt://salt/dir/to/desktop-some.conf not found' even though it exists
21:58 PintoBean joined #salt
21:58 _ale1_ joined #salt
22:04 subsignal joined #salt
22:05 ipmb joined #salt
22:09 _ale1_ joined #salt
22:09 notnotpeter joined #salt
22:09 XenophonF opnsrc: does desktop-some.conf live in /srv/salt/dir/to/?
22:12 mitsuhiko joined #salt
22:12 mitsuhiko joined #salt
22:13 alexbst joined #salt
22:15 jalaziz joined #salt
22:16 _ale1_ joined #salt
22:19 itru joined #salt
22:21 _ale1_ joined #salt
22:22 __ale__ joined #salt
22:24 mosen joined #salt
22:28 jchen joined #salt
22:28 jchen left #salt
22:30 _ale1_ joined #salt
22:31 andrew_v joined #salt
22:34 bhosmer_ joined #salt
22:36 _ale1_ joined #salt
22:36 I3olle joined #salt
22:40 cmcmacken joined #salt
22:42 jhauser joined #salt
22:42 signull joined #salt
22:43 sunkist joined #salt
22:44 __ale__ joined #salt
22:44 hasues joined #salt
22:45 hasues left #salt
22:47 aquassaut joined #salt
22:47 baweaver joined #salt
22:47 MaliutaLap joined #salt
22:47 lnr joined #salt
22:47 MaliutaLap left #salt
22:48 raygunsix joined #salt
22:50 echo joined #salt
22:52 norii joined #salt
22:52 _ale1_ joined #salt
22:54 boubou joined #salt
22:54 spicyWith joined #salt
22:56 echo_ joined #salt
22:59 spookah joined #salt
23:01 jalaziz joined #salt
23:02 sunkist joined #salt
23:04 ek6 joined #salt
23:04 glyf joined #salt
23:07 ChesFTC joined #salt
23:17 _ale1_ joined #salt
23:17 timoguin joined #salt
23:27 _ale1_ joined #salt
23:34 _ale1_ joined #salt
23:36 _ale1_ joined #salt
23:42 _ale1_ joined #salt
23:43 theologian joined #salt
23:49 _ale1_ joined #salt
23:50 badon joined #salt
23:51 murrdoc joined #salt
23:53 otter768 joined #salt
23:54 pass_by_value joined #salt
23:58 pass_by_value joined #salt
23:58 KyleG Hi #salt…..I'm running into a problem. It appears that my salt minions are caching system-level python packages?
23:59 KyleG I have a salt module, that basically acts as an intermediary between salt and this pip package we have built for our deploy process
23:59 KyleG If I upgrade the pip package with a newer version, I have to restart my salt minion for the new code to be in play

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