Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2015-05-04

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

All times shown according to UTC.

Time Nick Message
00:17 Berty_ joined #salt
00:19 aquassaut joined #salt
00:38 FeatherKing joined #salt
00:40 scbunn joined #salt
00:44 thayne joined #salt
00:46 otter768 joined #salt
01:00 alexhayes joined #salt
01:01 bhosmer_ joined #salt
01:03 alexhayes joined #salt
01:05 alexhayes joined #salt
01:06 alexhayes joined #salt
01:09 hasues joined #salt
01:13 brianfeister joined #salt
01:13 hasues left #salt
01:27 JDiPierro joined #salt
01:36 brianfeister joined #salt
02:08 rojem joined #salt
02:15 JDiPierro joined #salt
02:20 sunkist joined #salt
02:23 mosen joined #salt
02:29 scbunn joined #salt
02:30 ageorgop joined #salt
02:30 iggy syed: you can just put it in /srv/salt/_grains and it'll get sync'ed to the minions
02:31 cedwards iggy: is it auto-synced or does it still require running saltutil.sync_grains?
02:33 iggy it gets synced on a highstate, sync_grains, sync_all, amongst other things
02:35 favadi joined #salt
02:41 catpig joined #salt
02:43 JDiPierro joined #salt
02:45 cmcmacken joined #salt
02:48 michelangelo joined #salt
02:54 syed thanks iggy
03:02 bhosmer_ joined #salt
03:11 TyrfingMjolnir joined #salt
03:15 rdas joined #salt
03:42 GeorgeHahn joined #salt
03:51 fxhp joined #salt
04:03 malinoff joined #salt
04:07 ageorgop joined #salt
04:09 thayne joined #salt
04:14 sunkist1 joined #salt
04:22 TyrfingMjolnir joined #salt
04:26 GeorgeHahn left #salt
04:29 linjan joined #salt
04:37 cztanu hey guys, i'm having a bit of a wrestle with my states.... I've changed a for loop in a file.managed jinja template, but it doesn't seem to actually update! Which is weird because I have a {% set %} clause above it, which is updating no problem :\ any ideas?
04:43 thayne joined #salt
04:45 writtenoff joined #salt
04:48 ChesFTC cztanu: Have you tried using state.show_highstate, state.show_lowstate, or state.show_sls to see what's happening?
04:55 Furao joined #salt
04:59 ramteid joined #salt
05:03 bhosmer_ joined #salt
05:05 brianfeister joined #salt
05:16 tkharju joined #salt
05:25 steffo joined #salt
05:32 cztanu didn't even know about those commands, but checking them shows that everything is normal when it comes to states.
05:32 cztanu It's almost as if my file has been cached somewhere, and isn't updating properly for that reason.
05:32 cztanu I'm removing and adding things, and it's behaving like it's still using the old jinja template.
05:33 cztanu oh no......... it was gitfs the entire time
05:36 brianfeister joined #salt
05:40 steffo joined #salt
05:40 joeto joined #salt
05:42 markm_ joined #salt
05:49 ndrei joined #salt
05:55 stoogenmeyer joined #salt
05:59 krelo joined #salt
06:05 otter768 joined #salt
06:07 JayFK joined #salt
06:11 vstoniest joined #salt
06:11 colttt joined #salt
06:17 jhauser joined #salt
06:22 jbirdman joined #salt
06:23 flyboy joined #salt
06:25 AndreasLutro joined #salt
06:28 Furao joined #salt
06:29 st_iron joined #salt
06:29 st_iron hello
06:43 evle1 joined #salt
06:44 stoogenmeyer_ joined #salt
06:49 Furao joined #salt
06:53 trikke joined #salt
06:56 Auroch joined #salt
07:01 kawa2014 joined #salt
07:04 bhosmer_ joined #salt
07:09 eseyman joined #salt
07:11 totte left #salt
07:12 Furao joined #salt
07:13 lb1a joined #salt
07:21 CeBe joined #salt
07:22 CeBe1 joined #salt
07:23 teogop joined #salt
07:26 ph8 joined #salt
07:29 KermitTheFragger joined #salt
07:31 fbergroth joined #salt
07:34 stooj joined #salt
07:37 linjan joined #salt
07:41 Chadk joined #salt
07:43 dingo joined #salt
07:46 _JZ_ joined #salt
07:49 funzo_ joined #salt
07:52 c4urself joined #salt
07:56 siciliansun joined #salt
07:58 wpot joined #salt
07:59 Chadk This is a little bit of a stupid question. But can somebody explain the difference between tne OS Salt, and Enterprise Salt? The enterprise page really doesn't seem to offer anything other than support. :-/
08:00 hemebond Chadk: I've never looked but I'm not aware of any difference.
08:01 Chadk Odd!
08:01 hemebond I think you just buy support from Salt Inc.
08:01 hemebond As far as I know the code is the same everywhere.
08:02 Chadk Yeah, I don't quite get it. The enterprise page mentions a GUI, but it's not clear to me if that's just Halite(Which is deprecated?). Very confusing, as I'm currently trying to figured out which CM system to go with :-/
08:02 Chadk All the other guys have very clearly defined enterprise offerings, but this is very muddy, which is a shame.
08:03 c10 joined #salt
08:03 CeBe joined #salt
08:05 otter768 joined #salt
08:06 hemebond Wow. The enterprise page looks quite full of info to me.
08:06 Grokzen joined #salt
08:06 hemebond Saltstack itself is completely open source.
08:07 hemebond It looks like they have enterprise stuff that runs on top.
08:07 hemebond Quite a lot of stuff in fact.
08:08 Chadk I had a hard time parsing anything from it, due to the marketing speak that's stuffed in there. To me, it seems like it's just describing the OS stuff.
08:09 Chadk But I don't know, to be fair. I'm simply trying to narrow down the options to POC for my company. And while Salt looks pretty awesome since I'm a python guy, the Enterprise pricing/feature set being not very clear kinda is a shame. :(
08:23 lothiraldan joined #salt
08:23 MatthewsFace joined #salt
08:24 favadi joined #salt
08:25 c10 joined #salt
08:25 stoogenmeyer__ joined #salt
08:25 slav0nic joined #salt
08:25 slav0nic joined #salt
08:26 slav0nic left #salt
08:34 N-Mi joined #salt
08:34 N-Mi joined #salt
08:47 keimlink joined #salt
08:50 favadi joined #salt
08:55 N-Mi joined #salt
08:55 N-Mi joined #salt
08:56 bemehow joined #salt
09:01 ITChap joined #salt
09:04 bhosmer joined #salt
09:08 chiui joined #salt
09:16 catpig joined #salt
09:19 dariusjs joined #salt
09:26 bemehow joined #salt
09:35 k_sze[work] joined #salt
09:35 lothiraldan joined #salt
09:38 st_iron I used the user state to create a user, and I made a typo in the password argument, I wrote passwd, and it ran, but made no effect
09:38 st_iron is it the expected behaviour?
09:41 peters-tx joined #salt
09:46 ashb Chadk: enterprise is just a 'pinned' version of OS salt but they will backport security and bug fixes to it, and possibly do custom coding for you on it
09:47 saffe joined #salt
09:49 julienlavergne joined #salt
09:49 julienlavergne hi all
09:49 julienlavergne I need some information about user_data usage
09:50 julienlavergne from an example i get "${file(\"files/nginx/cloud-init.txt\")}"
09:50 julienlavergne is the syntax correct ?
09:52 stooj joined #salt
09:53 zpotoloom joined #salt
09:53 julienlavergne what means     user_data:                   "" => "ced5632d2ef1254e8b3ece87477d14a2bf7566a9" ?
09:58 froztbyte hmm
09:58 ssm left #salt
09:59 froztbyte do I get to structure pillar like I would structure a state? I want to segment up my users into groups
09:59 froztbyte so that I can do `- users.dev`, `users.support`, etc
09:59 froztbyte julienlavergne: I don't recognize what you're talking about. in which section of salt are you looking at?
10:00 froztbyte offhand none of that syntax really looks like salt
10:01 AndreasLutro froztbyte: the same way you would states, just create directories
10:02 froztbyte okay, so that works
10:02 froztbyte sweet
10:05 CedNantes joined #salt
10:05 CedNantes hello
10:06 otter768 joined #salt
10:13 lothiraldan joined #salt
10:29 dariusjs anyone from the Netherlands hangs out here?
10:36 steffo joined #salt
10:38 AndreasLutro joined #salt
10:42 KoFFiE dariusjs: from .be
10:42 KoFFiE ^^
10:46 stoogenmeyer__ joined #salt
10:55 bhosmer joined #salt
10:57 iggy Chadk: the enterprise gui is not halite... nothing like it... it's actually quite nice
10:58 illern_ joined #salt
11:03 babilen (it is what halite should have been)
11:04 babilen iggy: Is there a demo somewhere?
11:04 iggy there's a video from saltconf iirc
11:04 iggy and I think SSE will do demo's for prospective clients
11:07 _JZ_ joined #salt
11:08 dkrae joined #salt
11:08 st_iron how can the password of a user set with the user state? I use a pillar for it, but the user cannot authenticate
11:08 st_iron the pillar contains the passwd in plain text
11:12 MatthewsFace joined #salt
11:14 einyx joined #salt
11:14 einyx i'm using salt-master on a debian wheezy box
11:15 einyx and salt-minion on a ubuntu 14, both use the saltstack repo
11:15 bemehow joined #salt
11:15 einyx but when i try to run salt I get this error: https://gist.github.com/einyx/78b1d024953f79e9f46e
11:15 einyx any clue?
11:17 AB_ joined #salt
11:18 AB_ Hello There. I want to know if we can do NAT on minion using SALTSTACK?
11:18 EvaSDK hi there
11:18 EvaSDK I'm trying to run salt-cloud from master
11:18 EvaSDK to see if I can fix issues I have with lxc provider
11:19 EvaSDK it appears after some amount of commands to the lxc container, the temporary directory where the bootstrapping happens "disappears"
11:20 EvaSDK like it gets up to copying minion.pem, gets its md5sum and second md5sum call cannot find the file anymore. Subsequent commands then fail because the folder does not exist anymore and makedirs being false, it cannot be created at that point
11:22 dariusjs KoFFiE: ok, cool was wondering to startup some user group around amsterdam
11:23 fusionx86 joined #salt
11:23 dariusjs st_iron: user account for what OS?
11:25 AB_ I have gone through the modules list here http://docs.saltstack.com/en/latest/ref/states/all/
11:25 AB_ And there is no hyperlink to iptables module
11:25 st_iron sorry, Linux
11:26 st_iron I think I figured it out
11:26 st_iron I just generated a hash instead of the plain text, now it works
11:27 st_iron I can confirm, it is solved with using the hash in the pillar
11:29 favadi left #salt
11:32 CeBe1 joined #salt
11:35 bhosmer joined #salt
11:38 joehh einyx: can you paste your gitlab state?
11:40 fredvd joined #salt
11:45 EvaSDK oh god, the logic changed again
11:49 yggdrasi1 joined #salt
11:49 yggdrasi1 hi
11:50 azeazrazrza joined #salt
11:51 hojgaard joined #salt
11:52 iggy einyx: would have to see the state that caused that (protip: gist's can have multiple files per paste and can be updated in place)
11:52 hojgaard Hello. For some strange reason i salt-master always outputs info twice for one of my hosts: Look at this: http://pastebin.com/e8SEuzcZ
11:53 hojgaard mailscanner2 feedback always prints twice.. Cant seem to find the reason..
11:53 iggy hojgaard: probably have 2 minions running on it... stop the minion service, then check/kill any other python/salt-minion processes
11:55 hojgaard iggy, simple solution, thanks :D
11:55 yggdrasi1 anyone can give a hint? I have some issue on one of my custom module. Trying to create vz containers, the function managed of my script is not recognized. The running function instead load successfully. when trying to highstate all of my network confuguration is pushed (for the hypervisor) but none of my VMs pop-up...
11:56 yggdrasi1 I have seen on the git forum that we have to redefine the module_dirs conf but nothing happen
11:57 yggdrasi1 I have a "vz.managed in sls vms unavailable"
12:00 _Cyclone_ joined #salt
12:02 Chadk iggy, interesting. I guess I'll have to request a demo to get a full walk-through of the enterprise version before I make a decision on CM platform to go work :)
12:04 iggy Chadk: as for your other question, I think it was already answered, but there isn't really a difference between Salt and SaltStack Enterprise (aside from the gui)
12:05 iggy unlike chef/puppet/etc where you only get certain functionality from the enterprise versions
12:05 iggy or have artificial limits
12:07 iggy I think there are some addon modules for some high end hardware that you get access to as well, only heard something about it briefly in thatch's talk at SaltConf
12:07 ndrei joined #salt
12:07 iggy yggdrasi1: it would be easier to debug if we saw code, commands being run, the output from said commands, etc.
12:10 MaliutaLap joined #salt
12:13 lothiraldan joined #salt
12:14 llb joined #salt
12:15 llb left #salt
12:15 makouser joined #salt
12:16 yggdrasi1 iggy :http://pastebin.com/8KDGW9H7
12:17 s0lar joined #salt
12:20 aquassaut joined #salt
12:21 iggy that's one part of what I asked for
12:24 KoFFiE dariusjs: hmm amsterdam is a bit too far to go to a usergroup :p
12:26 saffe joined #salt
12:26 yggdrasi1 iggy : http://pastebin.com/a4AUtgfN
12:27 yggdrasi1 here is the code can't share more sorry ;)
12:28 iggy missing a virtual function
12:28 iggy although I'm not sure exactly where that is required
12:30 MaliutaLap left #salt
12:32 saffe_ joined #salt
12:36 losh joined #salt
12:38 dariusjs KoFFiE:  yeah I thought it would be :) I'll try twitter or something else
12:39 dariusjs st_iron: good, avoid plaintext if you can, but also be aware you can also hide things to be available to certain hosts
12:40 dariusjs I have a github gist here where I manage openssh host keys with if statements in pillars, https://gist.github.com/dariusjs/eeb83e1c4774a85f5265
12:45 makouser joined #salt
12:51 subsignal joined #salt
12:57 pcdummy joined #salt
12:57 pcdummy joined #salt
12:58 bhosmer joined #salt
12:59 makouser left #salt
13:00 JDiPierro joined #salt
13:01 MatthewsFace joined #salt
13:02 FeatherKing joined #salt
13:02 JDiPierro joined #salt
13:04 llb42 joined #salt
13:05 cpowell joined #salt
13:06 llb42 hello folks, I have a problem with salt 2014.7.5
13:06 llb42 if I set the default renderer in the master file to yaml_mako salt starts to break in a strange manner:
13:06 llb42 I run highstate, all is well, my one state completes sucessfully
13:06 llb42 I run highstate again, it fails with: Comment: Unknown yaml render error
13:06 llb42 I run highstate again, it fails with: ID info in SLS data is not a dictionary
13:06 llb42 at this point, I can see in /var/cache/salt/minion/files/base that the pillar file data.sls is present there
13:06 llb42 which seems wrong
13:06 llb42 any suggestions greatly appreciated!
13:10 _prime_ joined #salt
13:10 dyasny joined #salt
13:11 fxhp joined #salt
13:12 st_iron dariusjs: thanks, I check it
13:13 llb42 I forgot to mention that if I leave the renderer as jinja_yaml, then there are no issues
13:16 thayne joined #salt
13:18 jdesilet joined #salt
13:18 Furao joined #salt
13:22 scbunn joined #salt
13:23 iggy llb42: I'd open an issue
13:24 babilen llb42: I would be quite surprised if switching the default renderer is actually possible, given that there are numerous SLS files in the wild that do *not* have a shebang for jinja_yaml. This should, however, only affect you if you use any of those SLS files ... Could you paste your actual files to, say, http://refheap.com ?
13:24 babilen (one of the many things that make using mako trickier than it should be. I would have personally *strongly* preferred mako)
13:25 babilen And yeah, please open an issue about that.
13:26 llb42 ack, I'll open an issue with all the details, thanks for the feedback
13:26 mapu joined #salt
13:27 llb42 I ran in to the issue when attempting to update a large install from a previous salt version
13:27 llb42 now I have stripped it down to just one state and one pillar definition
13:28 fusionx86 joined #salt
13:29 jerematic joined #salt
13:30 primechuck joined #salt
13:30 bastion1704 joined #salt
13:34 ndrei joined #salt
13:35 bhosmer_ joined #salt
13:38 perfectsine joined #salt
13:39 XenophonF joined #salt
13:40 XenophonF with salt-cloud and ec2, how do i override the volume type?  i have a bunch of dev windows instances that all use gp2 (provisioned iops), when they just need to be on standard storage
13:40 babilen llb42: You *should* be able to switch the default renderer. Unfortunately nobody uses a shebang when using jinja (I am guilty of that too). In a way it would be great if every SLS file had to explicitly specify the renderer
13:41 XenophonF i tried adding "{ device: /dev/sda, type: standard }" to volumes list in my profile, but it didn't work
13:41 XenophonF this is salt 2014.7.5 in case that matters
13:45 perfectsine joined #salt
13:48 matthew-parlette joined #salt
13:50 c10 joined #salt
13:51 Tyrm joined #salt
13:51 lietu joined #salt
13:53 timoguin joined #salt
13:55 murrdoc joined #salt
13:56 murrdoc man this vault thing from the consul people is pretty damn dope
13:57 dariusjs most of the hashicopr products are
13:57 dariusjs murrdoc: well all even
13:57 dariusjs they made vagrant and terraform is also super interesting
13:57 bmccormick_ joined #salt
13:57 rojem joined #salt
13:57 dariusjs the salt beacons look interesting but I havent tried them yet
13:58 murrdoc vagrant is nice, packer too
13:58 murrdoc but the thing with vault is now instead of having to manage a gpg key and so on
13:58 murrdoc i just need to manage auth creds on a salt master
13:59 murrdoc and can query the key store for almost anything, minion keys ? certs, so on
14:00 babilen Is it integrated into salt?
14:00 murrdoc nope
14:00 murrdoc its new
14:00 babilen How do you use it then?
14:00 babilen (given that this is #salt)
14:00 dariusjs not out of the box I think, hashicorp are only aware of puppet and chef mostly, they have low exposure to salt
14:00 murrdoc pillar babilen
14:00 murrdoc specifically ext_pillar
14:00 babilen Yes, how do you use it as pillar in salt if it is not integrated?
14:01 murrdoc well it has an http front end
14:01 murrdoc or would u rather we didnt talk about non salt thing at all
14:01 murrdoc cos thats ok
14:01 babilen No, I'm fine with that
14:02 CK__ joined #salt
14:02 FeatherKing joined #salt
14:02 rosenfs_ joined #salt
14:02 babilen It's just that I have been looking for a solution to the "initial password" problem and vault looks as if it might fit the bill. That does, however, require a way to access it as ext_pillar and I don't see it listed in http://docs.saltstack.com/en/latest/ref/pillar/all/
14:03 murrdoc its not
14:03 murrdoc gonna have to write something for it
14:03 murrdoc https://www.vaultproject.io/docs/http/
14:04 murrdoc is what i was looking into
14:05 Berty_ joined #salt
14:05 mackstick joined #salt
14:05 bbradley hello
14:05 bbradley can i get the log from a git repo managed by gitfs?
14:07 murrdoc do u mean the git clone output ?
14:07 timoguin bbradley: yes. the repos are under hashed paths in /var/log/salt/master/gitfs/<hash>
14:07 timoguin you can 'git log' in any of them
14:08 timoguin they just aren't checked out
14:08 bbradley yea git log 'branchname' didn't work for me.
14:08 bbradley (in a repo directory there)
14:08 otter768 joined #salt
14:08 bbradley so i can check out code in that folder?
14:09 pdayton joined #salt
14:11 iggy murrdoc: ext_pillar on it's way
14:11 bbradley i think you mean /var/cache/salt/master/gitfs but yea.
14:11 andrew_v joined #salt
14:11 iggy that's exactly what we've been looking for
14:13 murrdoc iggy are u gonna use the cli ?
14:13 murrdoc or the http api
14:14 iggy haven't decided yet, but I am on 2015.2, so salt.utils.http is fairly feature complete
14:14 murrdoc i couldnt find anything on acquiring an auth token on the website
14:14 murrdoc for vault
14:15 bbradley oh!
14:15 rigor789 joined #salt
14:16 bbradley since it's not a mirror, i have to do git log remote/branchname.
14:16 ndrei joined #salt
14:17 murrdoc iggy:  lmk what u end up doing, i ll test it out or contribute too
14:19 iggy murrdoc: auth token... as in what you need to actually talk to the api/etc?
14:19 murrdoc yeah
14:20 iggy from the site it sounds like oauth2 or cert based
14:20 murrdoc yeah, you can even use github auth tokens
14:20 iggy haven't gotten that deep yet though
14:21 murrdoc thing is, if you use the cli tool, then all minions would need the vault binary
14:21 steffo joined #salt
14:21 iggy which is not much different than some modules today
14:21 iggy about half of them require some cli or python module to function
14:22 iggy but yeah, if it can be avoided...
14:22 murrdoc yup , if its all http based, then the ext_pillar can acquire lease, then acquire info
14:22 murrdoc if its using cli too, then install tool, authenticate, then acquire info
14:23 murrdoc tool*
14:24 StDiluted joined #salt
14:24 pdayton joined #salt
14:24 iggy I'm not real fond of us having the database passwords in files in pillar
14:24 iggy or our ssl certs
14:25 iggy with GitHub's "permissions" scheme it's impractical to have that stuff kept away from people that don't need to see it
14:26 murrdoc yeah
14:26 markm__ joined #salt
14:27 iggy although, I guess at some level there's going to be some key/etc necessary to get the master/minions to talk to vault... which we'll still have to keep in pillars I imagine
14:27 murrdoc u dont have to
14:27 murrdoc u could drop the user/pass in a file anywhere on the server, shit even an environment variable
14:28 ALLmightySPIFF joined #salt
14:28 iggy it would be awesome if I could have it email me everytime a secret was accessed
14:29 debian112 joined #salt
14:30 murrdoc vault ? or salt ?
14:30 iggy vault
14:30 timoguin Reading the shit out of the Vault docs. Looks very interesting.
14:30 iggy or salt... either way
14:30 timoguin I've been messing with Amazon KMS with credstash.
14:31 timoguin But like a cloud-portable approach.
14:31 iggy oh crap, gotta get into the office, bbl
14:31 lietu- joined #salt
14:32 clintberry joined #salt
14:32 ALLmightySPIFF joined #salt
14:33 murrdoc the biggest upside to vault for me is, i can have our current, 'we have to use this as our data store' mssql system, push to consul or become a mount within vault itself
14:33 murrdoc huge win especially cause i cant touch the existing datasore or make a new one
14:33 murrdoc :D
14:35 lothiraldan joined #salt
14:36 murrdoc documentation has 404's though
14:36 markm_ joined #salt
14:37 bluenemo joined #salt
14:37 bluenemo joined #salt
14:37 pressureman joined #salt
14:38 matthew-parlette joined #salt
14:39 lietu joined #salt
14:40 andrew_v_ joined #salt
14:40 rhodgin joined #salt
14:41 murrdoc brb
14:43 smcquay joined #salt
14:44 murrdoc joined #salt
14:44 conan_the_destro joined #salt
14:44 _prime_ joined #salt
14:46 lietu- joined #salt
14:47 fyb3r joined #salt
14:50 [7hunderbird] joined #salt
14:50 MatthewsFace joined #salt
14:51 markm joined #salt
14:51 c10 joined #salt
14:54 ksj hi, I want nginx to restart when apache has to be stopped. I have the following: http://dpaste.com/1TRCQ02 but it's not restarting nginx.
14:56 murrdoc can u try onchanges ?
14:57 murrdoc or better yet, listen_in
14:57 Brew joined #salt
14:57 beneggett joined #salt
14:58 iggy it would be "- service: start nginx service"
14:58 st_iron is that possible to write a state file to check if a user exists on the system, if yes, then add to several groups?
14:58 iggy it uses the id, not the name
14:58 ksj iggy: I tried that initially, but it didn't work....let me try again. I thought the docs said it was the name that counted, not the id
14:59 iggy if it took the name, it would be different than _every_ other requisite
14:59 TyrfingMjolnir joined #salt
14:59 ksj iggy: according to the docs it can be either: Requisites match on both the ID Declaration and the name parameter.
14:59 murrdoc listen_in instead on onchanges_in will work
14:59 ksj thanks, I haven't looked at listen in...I'll take a look
15:00 murrdoc scale of 1 to omg i hate it, hows the redis formula
15:00 jalbretsen joined #salt
15:00 iggy ksj: odd, I never noticed that
15:00 babilen murrdoc: What are your specific issues?
15:01 mpanetta joined #salt
15:01 murrdoc no issues, i havent even looked at
15:01 murrdoc it*
15:01 murrdoc looking now
15:01 babilen It should be quite alright - I'd maybe move the defaults into default.yaml (rather than keeping it in map.jinja), but it should be usable
15:02 babilen There's support for multiple versions and you can set everything in pillars
15:02 murrdoc i ll do the defaults.yml thing
15:02 babilen ta
15:02 ksj murrdoc: listen_in works great thanks....now just to figure out what the difference is between that and onchanges_in
15:02 murrdoc as a thank you for the rest
15:03 murrdoc ksj:  listen_in uses mod_watch
15:03 nesv joined #salt
15:03 murrdoc i have no clue how onchanges_in works, havent needed it yet
15:03 murrdoc will read code when i does
15:04 scbunn joined #salt
15:06 WesleyTech joined #salt
15:06 racooper joined #salt
15:10 terratoma joined #salt
15:18 andrew_v joined #salt
15:21 bhosmer joined #salt
15:24 Eugene joined #salt
15:25 SeeDickCode joined #salt
15:25 Grokzen joined #salt
15:30 bastion1704 I just manually install salt-minion and the /etc/salt/minion was not created
15:30 bastion1704 is it a new behaviour ?
15:30 bastion1704 version : 2014.7.5+ds-1ubuntu1
15:30 bastion1704 manually = apt-get install salt-minion
15:31 thayne joined #salt
15:31 lietu joined #salt
15:31 subsigna_ joined #salt
15:33 rideh joined #salt
15:35 CeBe1 joined #salt
15:36 bhosmer_ joined #salt
15:38 denys joined #salt
15:38 murrdoc try like a dpkg -C of the deb
15:38 murrdoc it should list out what files it is supposed to add/remove
15:39 Heartsbane joined #salt
15:39 larkir joined #salt
15:41 kunersdorf joined #salt
15:48 martoss joined #salt
15:50 martoss Hey Folks, I have https://github.com/martinhoefling/rsyncd-formula ready for forking into salt-formula organization
15:54 iggy ping basepi whiteinge ^
15:56 joeto Hi all, I have an issue with 2014.7.5 and AWS, I am creating Master server with role to read from specific S3 bucket, all sls are there. BUT Salt can't read from this bucket for some time and after several reboots it start works without any modification in config files.
15:56 joeto do I need to remove something from cadche or?
15:56 joeto I am out of ideas
15:56 joeto it hapens aways when I create new environment with new S3 Buckets and new masters.
15:57 joeto after few hours ewverithing start works :( , I can retrive ls from aws cli without problem
15:57 joeto during this time
16:04 XenophonF can anyone point me to salt-cloud docs that show how i can change the default block device mappings for ec2 instances?
16:05 iggy joeto: don't know specifically about aws/s3, but when it's failing, try running "salt-run -l debug fileserver.update"
16:05 markm_ joined #salt
16:06 iggy XenophonF: if you find anything, I'd like to know too... when we looked at salt-cloud last, it had very limited block device capabilities
16:07 XenophonF iggy: i naively assumed that one could use the volumes key in the cloud profile to do that (after looking up the root device name in the image)
16:08 XenophonF but "{ device: /dev/sda1, type: standard }" isn't working - the volume still gets created as gp2 (provisioned iops)
16:08 iggy oh, specific to ec2... figure
16:08 * iggy on GCE... can't help
16:09 armguy If you want all files to be a specific perm in a dir can I just do a /dir/* ? Or is this bad practice?
16:09 XenophonF bummer
16:09 c10 joined #salt
16:09 mirko joined #salt
16:09 otter768 joined #salt
16:09 murrdoc armguy:  file.directory has a recurse param u can use
16:10 XenophonF iggy, on ec2 (and maybe with other providers, too) there's a flag called "del_root_vol_on_destroy"
16:10 adelcast left #salt
16:10 XenophonF salt/cloud/clouds/ec2.py appears to use block device mappings to alter this setting
16:10 armguy murrdoc~ Thanks for the pointer :)
16:11 XenophonF sorry gp2 isn't provisioned iops, it's just general purpose ssd
16:11 conan_the_destro joined #salt
16:11 XenophonF still, i only need standard (magnetic) storage for my dev instances
16:11 TooLmaN joined #salt
16:16 XenophonF i'm reading through "def create_attach_volumes" at the moment
16:18 XenophonF ah here we go!
16:18 XenophonF block_device_mappings (duh) in the profile
16:19 XenophonF http://docs.saltstack.com/en/latest/topics/cloud/aws.html#optional-settings
16:19 mirko joined #salt
16:19 XenophonF and then to get the root device, i can use "aws ec2 describe-images --image-id ami-032b4b74" (or whatever the windows ami id is)
16:20 * XenophonF crosses figners
16:21 * XenophonF crosses fingers, too
16:21 XenophonF it works!
16:23 conan_the_destro joined #salt
16:23 murrdoc yay
16:23 XenophonF https://gist.github.com/xenophonf/8df49e830ee89a8d421c
16:24 XenophonF that's been bugging me for a week
16:24 murrdoc gist doesnt hvae your key this time
16:24 murrdoc i miss the old times
16:24 bhosmer_ joined #salt
16:24 murrdoc gotta admit thou
16:24 XenophonF hah
16:24 XenophonF i've probably read and re-read section 7.5.4 a dozen times by now
16:25 murrdoc looking at a server as a yaml config is pretty awesome
16:25 writtenoff joined #salt
16:26 XenophonF it sure is!
16:27 XenophonF too bad one can only extend a cloud profile once
16:27 iggy ^
16:27 murrdoc u have the power of jinja
16:27 murrdoc TM
16:27 iggy I hate that (and the fact that it's not documented more)
16:28 XenophonF very true murrdoc
16:28 XenophonF for a long time i've had this idea for what i though of as object oriented system administration
16:28 XenophonF where i can have a base class that defines some kind of common config all my servers have
16:28 iggy yeah, I had at one point added better profile/provider support to the salt-formula
16:28 XenophonF and then i can sub-class, or create instances, etc.
16:28 iggy but we stopped using salt-cloud, so I never pushed it
16:29 KyleG joined #salt
16:29 KyleG joined #salt
16:29 clintberry joined #salt
16:30 VR-Jack2 I'm doing something similar, but writing it out between kickstart and virt-install commands.
16:30 VR-Jack2 Still, it's nice to define the server in yaml
16:31 murrdoc iggy:  need to add this to the mongodb formula https://github.com/rueckstiess/mtools
16:31 murrdoc cos its awesome
16:31 murrdoc its a pip install at most
16:31 murrdoc but i didnt want to put it in where ever, figured i d ask u before i rewrote it all
16:32 iggy mongodb.mtools ... go
16:32 iggy I'm not on mongo this week
16:32 iggy this week, I'm the data scientist
16:33 scbunn joined #salt
16:34 _prime_ anyone ever get salt.modules.cmdmod.retcode to actually work in a grain?
16:35 _prime_ I want to detect a return of 1 and set a grain value accordingly, but the  python code craps out with [salt.modules.cmdmod][ERROR   ] Command 'test -n blah' messages, so I can never use the return value of '1' accordingly. (salt 2014.7.5)
16:37 ndrei joined #salt
16:38 pdayton joined #salt
16:38 JDiPierro joined #salt
16:38 ckao joined #salt
16:38 aparsons joined #salt
16:39 Springwater joined #salt
16:39 _mel_ joined #salt
16:42 Springwater I have a noob question. Hopefully this is the right place for it. I'm using Salt to deploy a Java WAR file. The flow is 1) stop tomcat, 2) use file.managed to install the new file if necessary 3) start tomcat. However I don't want to stop tomcat if the WAR file is already up to date. I don't see any obvious way to simply check to see if a managed fill will be replaced ahead of time. Any advice?
16:43 spookah joined #salt
16:43 nesv Springwater: I believe the onchanges requisite might be what you're looking for
16:43 iggy do you have to stop it before putting the war file in place?
16:44 Springwater It's a best practice to stop tomcat when doing deploys otherwise you can run out of permgen
16:44 iggy I don't know what permgen is
16:44 rm_jorge joined #salt
16:45 Springwater It's just part of the java memory model
16:45 murrdoc iggy:  whats a data scientist
16:45 murrdoc and doe sit pay better
16:45 iggy prereq?
16:45 gladiatr joined #salt
16:46 iggy basically runs the prereq state with test=True if changes were to happen, it would run the prereq, then the main state
16:46 spookah pergem errors haunt my dreams
16:46 _prime_ heh, user error.  had a trailing strip() from when I was using cmdmod.run, once that was removed retcode worked fine. nothing to see here....
16:47 iggy murrdoc: I don't know, and I wish
16:48 murrdoc did u try new relic
16:48 murrdoc its the best for monitoring stuff in cloud both servers and websites
16:48 adelcast joined #salt
16:48 Springwater prereq looks promising. I'll try that. Thanks.
16:48 stoogenmeyer__ joined #salt
16:48 joet0 joined #salt
16:49 iggy murrdoc: yeah, I mentioned it, they said no
16:49 iggy not sure if cheap or paranoid about that info going somewhere else
16:50 iggy they are very paranoid about what they think are trade secrets getting out
16:50 penguinpowernz joined #salt
16:50 iggy only my cohort and I really know that we are really like 5 years behind the tech curve
16:51 murrdoc k
16:51 iggy I had to get my SaltConf slides approved by 8 different people
16:51 murrdoc NIH time
16:51 MatthewsFace joined #salt
16:51 iggy that's almost 10% of the company...
16:51 linjan joined #salt
16:51 murrdoc u know u got a job in la waiting for your ass
16:52 murrdoc fixing my code
16:52 MatthewsFace joined #salt
16:53 _JZ_ joined #salt
16:54 iggy I'm not sure many companies would put up with me being the top talker in #salt (by almost 2x the amount of anybody else ;)
16:54 murrdoc i am here too
16:54 murrdoc finish your work no one cares
16:55 Dev0n joined #salt
16:56 troyready joined #salt
16:57 hasues joined #salt
17:01 MaliutaLap joined #salt
17:02 Ahlee iggy: heh, tell me about it.
17:03 hasues left #salt
17:04 spookah joined #salt
17:09 desposo joined #salt
17:09 linjan joined #salt
17:11 SheetiS :D
17:13 bemehow joined #salt
17:13 dopesong joined #salt
17:15 enarciso joined #salt
17:16 julienlavergne joined #salt
17:16 ageorgop joined #salt
17:19 fbergroth joined #salt
17:19 bemehow joined #salt
17:20 ajw0100 joined #salt
17:34 denys joined #salt
17:37 bhosmer joined #salt
17:41 tmclaugh[work] joined #salt
17:41 stoogenmeyer__ joined #salt
17:42 ajw0100 joined #salt
17:43 spookah joined #salt
17:47 lothiraldan joined #salt
17:47 ndrei joined #salt
17:51 theologian joined #salt
17:54 rojem joined #salt
17:58 druonysus joined #salt
17:58 druonysus joined #salt
18:01 baweaver joined #salt
18:07 joeto iggy: thank you for your sugestion, I used salt-master -l debug jsut to check what happens (output is the same) so error is 307 and after that 403 (access denied ?!?) which kill me :(. Without any change from my site same config start working after a while :(. Already deleted /var/cache/salt/master - nothing happens :(
18:10 otter768 joined #salt
18:16 yetAnotherZero joined #salt
18:18 wendall911 joined #salt
18:19 Edgan basepi: ping
18:21 iggy joeto: and if you try the aws/cli tools it works from that same node at the time the master is failing?
18:21 basepi Edgan: pong
18:24 Edgan basepi: I am trying to track down a new issue in relation to PR #23105. If the host can't be connected(times out) to somewhere in the client/ssh/* it is returning an exit code of 0, but ssh on the command line returns 255. Can you help, or point me at someone who can?
18:26 basepi Edgan: that pull request isn't actually merged, so you won't see its changes yet. Have you tested with that pull request? At first glance I feel like salt-ssh should be returning an exit code of 1 for any error (including timeouts)
18:26 zugzwang joined #salt
18:26 basepi But I could be wrong
18:26 zugzwang how to install salt-cloud in centos7? the docs only show how to install in ubuntu
18:27 zugzwang i thought the bootstrap script would install it automatically
18:27 Edgan basepi: it doesn't in all cases. I am the author of the PR.
18:27 basepi Edgan: oh, heh, I hadn't connected you as the author. =P
18:28 basepi In the case of a timeout, have you checked what the `ret[host]` looks like?
18:28 basepi Is the 255 retcode not being returned?
18:28 basepi We may be intercepting it to make it look nicer
18:28 basepi Or something
18:29 Edgan basepi: yes, I am getting back 0. But there are so many layers between handle_ssh and the actual ssh command I am getting lost.
18:30 basepi Edgan: you're going to want to focus on handle_routine()
18:30 basepi Once you get to that layer you're only working on a single host.
18:30 basepi It creates a Single object, which is what handles all of the work for that single host
18:30 subsignal joined #salt
18:31 Edgan basepi: Is the actual ssh call for the initial connection in Single, or somewhere else?
18:31 basepi The run() method in that object is where the work is done, and it's actually fairly straightforward
18:31 basepi The actual ssh call is in salt/client/ssh/shell.py
18:32 basepi exec_cmd()
18:32 Edgan basepi: yeah, I was just looking at that
18:32 basepi or, more precisely, _run_cmd()
18:32 basepi Most of the work is handled by salt.utils.vt
18:32 basepi And I don't know that code
18:33 basepi But I doubt the problem is in the vt code
18:33 basepi I'm guessing we're consuming the retcode somewhere else
18:33 basepi But I could be wrong
18:33 rojem joined #salt
18:33 Edgan basepi: I am going to look at exec_cmd
18:33 basepi Sounds good. Ping me if you hit any blockers
18:33 Fiber^ joined #salt
18:41 nzero joined #salt
18:48 zugzwang why does my salt-cloud says: "There was an error listing images: There are no cloud providers configured." ?
18:48 zugzwang donest salt-cloud come with a lot of default providers?
18:49 pannon joined #salt
18:54 j-saturne joined #salt
18:55 amcorreia joined #salt
18:55 Edgan basepi: I think I found the issue. run in client/ssh/__init__.py  has an if statement, and the middle part, "elif self.fun in self.wfuncs:" only returns stdout, and not the return code. In my case that seems to be the code path it takes. I suspect it would work if I defined raw_shell
18:55 iggy zugzwang: it doesn't come with them configured by default (i.e. keys/passwords/etc)
18:55 murrdoc1 joined #salt
18:56 baweaver joined #salt
18:58 rhodgin joined #salt
18:59 rhodgin joined #salt
19:01 rhodgin joined #salt
19:01 iggy kind of wish python's get worked like Salt's (i.e. with : separator)
19:02 pannon Can a jinja variable include another variable? example: {% set node = grains['nodename'] %} {% set nodeid = salt['pillar.get']('app:nodeid:node') %}
19:04 pannon not sure how to tackle the variable "node" in 'app:nodeid:node'
19:04 iggy sure
19:04 iggy {% set nodeid = salt['pillar.get']('app:nodeid:' ~ node) %}
19:04 iggy or even {% set nodeid = salt['pillar.get']('app:nodeid:' ~ grains['nodename']) %}
19:05 iggy one-liner ftw
19:05 pannon iggy: nice thank you!
19:06 pannon will dive in deeper into jinja
19:06 pannon impressed by salt templating options!
19:07 joeto iggy: yes, If I use aws cli on master it is able to list buckets ... I will try to get files ... shame on me I just list bucket :(. Maybe get files not works ... will check this tomorrow
19:08 joeto iggy: do you know where salt save temp/cache/all data? only /var/cache/salt or somewhere else?
19:08 baweaver joined #salt
19:08 murrdoc1 yes
19:08 mapu joined #salt
19:09 murrdoc1 run salt 'minionname' config.get cachedir
19:09 murrdoc1 it will tell u where it is on the minions
19:09 hybridpollo joined #salt
19:10 druonysuse joined #salt
19:14 cansis joined #salt
19:26 ndrei joined #salt
19:32 babilen iggy: It is ridiculous that Python's .get doesn't support key sequences, yeah
19:33 babilen (or .get_in(...) if they insist)
19:33 babilen https://clojuredocs.org/clojure.core/get-in is what I have in mind there
19:34 farcaller left #salt
19:38 bhosmer__ joined #salt
19:39 murrdoc1 what are u trying to do with .get
19:40 murrdoc1 cos you could use the same get that pillar.get uses
19:40 murrdoc1 so instead of doing .get it would be like salt.util.get something
19:41 babilen I simply think that something like Clojure's get-in (and assoc-in) would be better than what is currently in the Python stdlib
19:41 murrdoc1 yes
19:41 murrdoc1 it would be
19:41 murrdoc1 heres the 'salty' way
19:41 murrdoc1 https://github.com/saltstack/salt/blob/v2014.7.2/salt/modules/pillar.py#L57-L67
19:41 jdesilet joined #salt
19:42 Berty_ joined #salt
19:42 babilen I also dearly miss http://clojuredocs.org/clojure.core/merge-with whenever I have to use Python's .update(...)
19:42 murrdoc1 at this point maybe formulas need to ship with a _module python scrtip that handles that sorta get and .merge like in salt.utils.dict
19:49 wnkz joined #salt
19:51 Edgan basepi: Yeah, that is it. It is function run_wfunc. It returns json, and no return code. The easiest solution for me is salt-ssh -r for raw_shell.
19:52 basepi It should only be running a wfunc if you'
19:52 basepi you're running a wrapper function (like state.*, pillar.*, etcZ)
19:52 murrdoc1 babilen:  https://gist.github.com/anonymous/42203948dd06440a7c1c
19:52 basepi (Can't type today)
19:53 druonysuse joined #salt
19:53 basepi Edgan: For some wfuncs, it makes sense for them to not be returning a retcode, because not all wfuncs even hit the minion. Which wfunc were you using in your testing?
19:54 Guest89 joined #salt
19:54 Edgan basepi: I am, salt-ssh -i #{type}*.#{env}* state.highstate   I do this for deployment.
19:55 basepi Ya, we should definitely see if we can't make highstate return the retcode, at least when things fail.
19:57 scbunn joined #salt
19:57 sdm24 joined #salt
19:57 Edgan basepi: nod
20:02 babilen murrdoc1: ah, nice
20:03 babilen i.want.formula.dependencies.
20:03 murrdoc1 just do it eh
20:04 iggy I'm not working on Salt today
20:04 murrdoc1 basepi:  is it possible to expose the salt.utils functions in jinja por fa vor
20:04 iggy think it'd be a pita to copy out pillar.get's code
20:04 murrdoc1 https://gist.github.com/anonymous/42203948dd06440a7c1c
20:04 murrdoc1 its 1 line
20:05 iggy I repeat that I'm not working on Salt today
20:05 iggy I do actually hack on python that isn't Salt ;)
20:06 murrdoc1 dont we all
20:06 iggy babilen: take your dependencies elsewhere
20:06 murrdoc1 if salt would provide access to salt.utils in jinja
20:06 murrdoc1 we dont need no dependencies
20:09 iggy I'm slowly warming up to the idea of some sort of trust/popularity based central repo of salt related code, but don't call it formulas
20:10 iggy and quit trying to shoehorn weird shit into jinja
20:11 otter768 joined #salt
20:11 elfixit joined #salt
20:12 murrdoc1 its not wierd shit fool
20:12 murrdoc1 go science some data
20:13 cmcmacken joined #salt
20:13 sandah joined #salt
20:13 aparsons joined #salt
20:15 xintron_ joined #salt
20:17 synical joined #salt
20:18 iggy when you've got like 40 lines of set's/if's/etc and 1 line of actual template, you're doing something wrong
20:24 baweaver joined #salt
20:24 notnotpeter joined #salt
20:25 yetAnotherZero joined #salt
20:25 bbhoss joined #salt
20:25 mikepea_ joined #salt
20:25 grepory joined #salt
20:27 primechuck Is there a way to create a job from the websocket?  It looks to be listen only.
20:29 notnotpeter joined #salt
20:31 yetAnotherZero joined #salt
20:33 codekobe joined #salt
20:33 ashb joined #salt
20:33 bbhoss joined #salt
20:33 LinuxHorn joined #salt
20:33 JonGretar joined #salt
20:34 tr_h joined #salt
20:34 antonw joined #salt
20:35 iggy primechuck: it is listen only
20:35 zipkid joined #salt
20:35 Tritlo joined #salt
20:35 iggy make a second connection if necessary
20:35 clintberry joined #salt
20:35 ramishra joined #salt
20:35 lkannan_ joined #salt
20:37 murrdoc1 u didnt read the code
20:38 murrdoc1 u drama mama u
20:38 murrdoc1 using this module … you can do simeple gets like default_setting:pkg:name
20:38 murrdoc1 instead of jinja ing it
20:39 akitada joined #salt
20:39 iggy I'm not talking about you... talking about things like zookeeper-formula
20:40 * murrdoc1 wont touch it
20:42 * iggy goes back to data dumpster diving
20:45 murrdoc1 is there a file.empty
20:45 terminus_ joined #salt
20:45 murrdoc1 like if file exists, make it empty ?
20:45 murrdoc1 dont delete
20:45 certaindoom joined #salt
20:45 nzero joined #salt
20:45 certaindoom Is it possible to obtain the minion_id that is being acted upon in a state sls file?
20:46 sdm24 grains['id']?
20:46 iggy murrdoc1: file.managed -contents: "" ?
20:47 murrdoc1 ah yes
20:47 murrdoc1 might be better to leave in a comment
20:47 murrdoc1 so gonna do that
20:47 sdm24 or alternatively, couldn't you have a blank template as the source for file.managed?
20:48 druonysuse joined #salt
20:48 druonysuse joined #salt
20:49 iggy unnecessary inode use... what are you a reiserfs user?
20:50 sdm24 or even use jinja to set a variable to be blank and then have the template print the variable!
20:50 sdm24 yay salt
20:50 Tyrm joined #salt
20:54 sine_nitore joined #salt
20:57 spookah joined #salt
21:04 murrdoc1 is nicoel from saltstack in this chat ?
21:04 iggy neg
21:04 murrdoc1 too bad
21:05 murrdoc1 i <3 side bands
21:05 murrdoc1 side bars ?
21:05 murrdoc1 what is the correct word for that
21:06 certaindoom Is it possible to store the states that are applied to a minion as grains on that minion?
21:07 iggy no(t easily)
21:08 certaindoom iggy: not sure what that means
21:08 iggy it means you could probably rig something up, but there's no telling how reliable it would be
21:09 iggy and it probably wouldn't be easy
21:10 certaindoom So there is no way for a host to be aware of what states are being applied to it?
21:12 iggy what's the actual problem you're trying to solve
21:13 certaindoom We want to report in a web interface what states were applied in to each minion.  Having that minion send data to a system we develop that can then display to us what states are being applied.
21:14 iggy best bet is to probably returner everything to some database and pull that info from there
21:17 hemebond iggy: Would show_highstate be useful to certaindoom?
21:18 iggy ask them?
21:18 * iggy predates fancy things like show_highstate
21:18 hemebond I was just wondering if my understanding of the function was correct.
21:18 hemebond Ah okay.
21:18 cornfeedhobo left #salt
21:18 hemebond I think it just shows you what states are applied during a highstate,.
21:18 yggdrasi1 joined #salt
21:19 rojem joined #salt
21:20 iggy certaindoom: sudo salt-call state.show_highstate
21:20 iggy see if that shows you what you want to know (although it's more of "what would be applied" rather than "what has been applied"
21:20 certaindoom Ah, I found salt 'hostid' state.show_top
21:22 certaindoom Another question, could I call something like "salt 'hostid' state.show_top from a state file for use during compilation?
21:22 zugzwang I am successfuly using salt-cloud to create a Droplet on Digital Ocean, but it seems salt is not installed after the vm is created. Any ideas?
21:23 hemebond certaindoom: You should be able to use a grain for that.
21:24 XenophonF joined #salt
21:24 iggy certaindoom: for another host or the same host?
21:25 certaindoom hemebond: can you elaborate?  That was my original question, but iggy told me I could not use a grain to get the list of states that would be applied from inside a state file
21:25 hemebond certaindoom: Sorry, I thought you just wanted to know the minion id inside the state.
21:25 certaindoom iggy: I think I could run salt '{{ grains['id']}}" state.show_top to get the show_top output for the current minion
21:25 hemebond It sounds to me like you need to do your reporting/data collection from the master.
21:26 murrdoc joined #salt
21:26 certaindoom But can I execute that piece of code inside a state file and get the list of states to be applied to a minion
21:26 iggy certaindoom: {% set topdata = salt['state.show_top'] %} would get you the data for the current host
21:26 certaindoom hemebond: I have the reporting stuff covered, this was a new question
21:26 iggy and no, don't run "salt foo ..." from a state
21:26 iggy salt-ception
21:27 certaindoom salt-ception indeed
21:27 amcorreia joined #salt
21:30 nzero joined #salt
21:33 certaindoom iggy: When I try to set the variable topdata = salt['state.show_top'] %}, is that outputting as a dictionary, list, something I can iterate over?
21:35 pdayton joined #salt
21:37 pdayton joined #salt
21:37 athros joined #salt
21:38 conan_the_destro joined #salt
21:38 athros Question: has there been any resolution to https://github.com/saltstack/salt/issues/10948 that might not have made it into the issue? (pkg.install and pkg.available_version on windows minion to SL master do not seem to behave according to docs.)
21:38 iggy {% set topdata = salt['state.show_top']() %}
21:38 iggy forgot the ()
21:39 bhosmer joined #salt
21:39 rhodgin joined #salt
21:39 iggy certaindoom: it'd be similar to what you get on the command line (nested dicts and lists)
21:39 baweaver joined #salt
21:39 certaindoom iggy: Awesome, I've got a dict now.  Thanks so much!
21:40 murrdoc {% set topdata = salt['state.show_top']() %} is huge
21:40 murrdoc bad idea maybe ?
21:40 iggy show_top shouldn't be that big
21:40 iggy show_highstate would be massive
21:40 primechuck iggy: thanks.  So jobs can only be created via the rest interface, but I can listen for the job ID from the websocket event or HTTP stream.
21:41 certaindoom show_top is not huge.  It's just a list of states for that minion
21:44 certaindoom iggy: That's going to be very useful, now we can make decisions in different state files based on what states are being applied.  Thanks again.
21:46 murrdoc show_top parses all states included and figure out all the states to be executed
21:46 murrdoc no  ?
21:46 iggy you should really use the top file for targeting, not a bunch of jinja if's all over your state files
21:46 murrdoc yes
21:46 JayFK joined #salt
21:47 iggy show_top is the compiled/combined top file for the host
21:47 iggy which here is only about 20 lines long
21:50 gladiatr joined #salt
21:52 racooper pillar question: is it possible to set grains from pillars based on hostname? I know I can do it in states but can't find any examples of pillar/state for doing it.
21:54 thayne joined #salt
21:55 iggy I think someone gist'ed a custom grain module that pulled some data from pillars the other day
21:57 fintler joined #salt
22:05 ajw0100 joined #salt
22:06 notnotpeter joined #salt
22:09 [vaelen] joined #salt
22:10 ek6 alright my missing the obvious moment of the day...ive always done schedules on the minion...can you have a schedule on the master and if so how do you specify targets?
22:11 murrdoc yah
22:11 murrdoc make a minion on the master foo
22:12 otter768 joined #salt
22:14 pdayton joined #salt
22:15 pdayton joined #salt
22:15 ageorgop joined #salt
22:16 pdayton joined #salt
22:17 [vaelen] joined #salt
22:17 pdayton joined #salt
22:19 Edgan basepi: I realized that -r doesn't work for me for deployment. You mentioned turned on return codes just for state.highstate. Do you think that is the best approach, or disable it only for certain functions? Aka whitelist or blacklist?
22:20 basepi I haven't done a ton of investigation, but I feel like we should probably have a retcode for most things in salt-ssh. Not sure what effect that will have, though
22:21 murrdoc Gareth:  ek6 has crazy but sensible ideas of how to use the scheduler
22:21 Gareth murrdoc: Is it going to lead to dogs and cats living together in perfect peace harmony?
22:22 ek6 only if the cats make it out of the blender alive
22:22 Gareth Don't hate on cats.
22:22 murrdoc Gareth:  as it should
22:22 Gareth They'll cut you.
22:22 ek6 once they are in the sack on the way to the river i have no problem with cats
22:23 murrdoc did u know this exists http://www.reddit.com/r/cathighfive
22:23 pdayton joined #salt
22:24 Gareth anyway, so yeah. you can schedule stuff via the master but only runners.
22:24 ek6 gareth: basically i want to run a scheduler on the MASTER with schedule config like it is today..except with tgt: or other param to have it execute towards those minions
22:24 ek6 gareth: boo..why only runners?
22:25 ek6 gareth: cant just give me a tgt: option and we can call it a time based reactor? :)
22:26 thayne joined #salt
22:26 Gareth That's the way it's written today, no one has had a compelling reason to have it another way and there are better ways to target scheduled jobs at minions, eg. use schedule items in pillar or use the execution or state execution modules to add the schedule items to the minion.
22:26 baweaver joined #salt
22:28 ek6 err...id argue different...and the second one is reasonable.....just in some cases i like to have that schedule at the top of the stack without just having it in cron
22:29 whytewolf ek6: sooo orchestrate runner + scheduler?
22:30 mosen joined #salt
22:31 murrdoc ay caramba
22:31 ek6 whytewolf: sure....many different ways to skin the cat...i guess i just looked at the scheduler options and was like 'yup that has everything i need except tgt option'
22:32 whytewolf poor cats in this channel
22:33 Gareth ek6: explain to me again how this tgt would work.  So you would have the schedule item defined in the master config and it would essentially publish the commands out to the minions each time it's supposed to run?
22:35 ek6 yup...basically no different then if you ran  salt   'tgt'  function args      on the master
22:36 Gareth And what advantages do you see of that approach over using pillar or the existing modules or state modules to push the scheduled jobs out to the minions?
22:36 kaictl joined #salt
22:37 viq joined #salt
22:38 scbunn joined #salt
22:42 ek6 its just different gareth...i havent looked into using the pillar to drop schedule config in...think that may give me the same net effect to keep the data at the top of the stack..since minions when they pop up will pull that..
22:43 Gareth ek6: Not against the idea :)  Just curious.
22:44 ek6 but still not as centralized as id like it for some of the folks around here..   again...plenty of hammers in the drawer to get the job done a different way.   I was just trying to remove the cron component on the master
22:45 ek6 it just annoyed me because again i stare at the scheduler and say 'thats perfect but where is the tgt: option'
22:45 notnotpe_ joined #salt
22:46 bfoxwell joined #salt
22:47 baweaver joined #salt
22:48 ajw0100 joined #salt
22:49 Tyrm joined #salt
22:50 nzero joined #salt
22:55 Gareth Using pillar or state files for schedule items makes everything centralized.  And you can target the pillar data or state runs, etc.
22:59 ek6 its ok..this is now an 'over a beer' argument sometime
23:00 Gareth SaltConf 2016 :)
23:01 pannon joined #salt
23:03 TBAU joined #salt
23:13 murrdoc thing is ek6, the tgt'ing is done by assigning a pillar to a minion
23:13 murrdoc so you could do {% if grains.get(id') … %}
23:13 murrdoc schedule:
23:13 murrdoc
23:13 murrdoc {% endif %}
23:14 ek6 oh my...and then the cage drops on the rat?
23:19 ek6 yeah through the pillar will work...ill just have to really watch for pillar naming clashes..which is a devil you guys dance with lately that to this point ive been able to avoid
23:20 tcolvin joined #salt
23:20 ek6 lately=daily
23:23 iggy use descriptive IDs and you should be fine... don't use job1 everywhere ;)
23:23 _Cyclone_ joined #salt
23:25 Gareth One job to rule them all!
23:25 ek6 really? i thought it had to be named job with a single digit int after that...was wondering why that klod gareth put so low a cap in
23:25 ph8 joined #salt
23:26 cmcmacken joined #salt
23:26 Gareth stupid klod.  hate that guy.
23:26 dopesong_ joined #salt
23:26 murrdoc kneel before KLOD
23:27 ek6 be afraid gareth....murrdoc came up with this awesome plan to deal with you...just need to find 10 skunk bladders and a snipe skin and your in for it
23:28 Gareth The line forms to the right.
23:28 ek6 think the skunk bladders were to deal with the line...but unclear at this point
23:33 baweaver joined #salt
23:37 murrdoc joined #salt
23:39 bhosmer_ joined #salt
23:40 cmcmacken joined #salt
23:43 clintberry joined #salt
23:44 patrek joined #salt
23:45 ajw0100 joined #salt
23:53 baweaver joined #salt
23:54 spookah joined #salt
23:55 dopesong joined #salt
23:59 subsignal joined #salt

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