Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2015-05-11

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

All times shown according to UTC.

Time Nick Message
00:06 joshin joined #salt
00:07 ALLmightySPIFF joined #salt
00:08 pdayton joined #salt
00:11 catpig joined #salt
00:18 kusams joined #salt
00:20 VR-Jack hmm. weird. I didn't used to get Detected conflicting IDs in pillars. :(
00:23 otter768 joined #salt
00:36 aarontc joined #salt
00:37 VR-Jack and a master restart fixes it. heh
00:49 aarontc joined #salt
00:51 subsignal joined #salt
01:02 pviktori_ joined #salt
01:03 c10b10 joined #salt
01:04 ChesFTC Anybody know of a way or existing output module to give play recaps like ansible does? e.g. Ansible will output lines like:
01:04 ChesFTC 127.0.0.1                  : ok=2    changed=0    unreachable=0    failed=0
01:04 ChesFTC For each host
01:05 iggy the base of pi... our only link to SSE
01:07 mapu joined #salt
01:11 ChesFTC Essentially, I'm just after a summary table at the end
01:12 pviktori joined #salt
01:12 cozby_ VR-Jack: yeah, thats precisely whats happening
01:12 cozby_ argh..
01:14 iggy ChesFTC: state_verbose: false is helpful
01:15 VR-Jack ChesFTC: I think salt has flags that limit it's output to more summary type results. Of course, make sure you're timeouts are good enough.
01:16 ChesFTC iggy, VR-Jack: The issue is that I'd like to use --state-output=changes with an additional summary at the end
01:16 ChesFTC We wish to log all of the changes, but also have a quick table at the end that we can refer to in order to see a summary
01:16 iggy not possible I think
01:17 ChesFTC Yeah, I think I'll have to write a patch
01:17 * ChesFTC shrugs
01:17 iggy the way that salt returns results per minion doesn't lend itself to that kind of thing
01:17 iggy it won't be a easy patch
01:17 iggy the output layer is pretty complicated
01:18 ChesFTC Yeah, it looks like salt.state.output.highstate.output is called for one or more hosts at a time
01:18 ChesFTC Which would probably give us a bunch of different tables with one or more hosts in it scattered throughout the output
01:18 VR-Jack ChesFTC: Ryan Lane I think had a blog post on a combination that basically did a summary except for diffs for changes.
01:19 VR-Jack but good luck funding it. hehe
01:20 ChesFTC Yeah, I can't find it in his saltstack articles :-/
01:20 VR-Jack I think it was an early ansible vs saltstack article actually
01:21 VR-Jack I think it was ryan who went from puppet to salt, trying ansible on the way. Been awhile. I've read more than I can hold. lol
01:21 ChesFTC I'm probably still more experienced in ansible, but salt is definitely the right fit for where I'm working now
01:22 ChesFTC It's interesting how similar the tools are, but I still think that they have very different optimal use-cases
01:22 VR-Jack definitely. I was torn between them, but the call home function is useful
01:22 ChesFTC Where I used to work, all hosts were emphemeral
01:22 VR-Jack and I worried about ssh logs with ansible
01:23 ChesFTC So the call-home thing didn't matter
01:23 ChesFTC If an ec2 instance disappeared, no worries, spin up another
01:24 ChesFTC On the other hand, salt's templated sls's are way more powerful, but we're already having discussions here on how to decrease the complexity of one of our state files
01:24 iggy you could just do a returner and pull what you need from those results
01:25 VR-Jack I like simple, though iggy would argue my top file is anything but. lol
01:25 ChesFTC iggy: Thanks, that looks like the right approach
01:25 iggy it's hideous
01:25 ChesFTC VR-Jack: We're auto-building VPC security groups
01:25 VR-Jack of course, I found a bug in salt thanks to my fun code
01:25 Antiarc I'd personally suggest just piping the output to a custom parser/reporter :)
01:25 Antiarc Probably easier than hacking up salt itself
01:26 ChesFTC Antiarc: You know, that's pretty sensible approach that I bloody well should have thought about
01:26 ChesFTC Apparently I'm doing the 'all I've got is a hammer' approach today :-/
01:26 VR-Jack ChesFTC: nice. I autobuild ISP servers for telcos. One reason I did my hideous top file. let's me add customers to my own master without jacking with top.sls much.
01:30 bhosmer_ joined #salt
01:31 pdayton joined #salt
01:33 c10b10 joined #salt
01:49 mapu joined #salt
01:50 froztbyte ChesFTC: building them from salt, or something else?
01:50 yexingok joined #salt
01:52 ChesFTC froztbyte: Building them from salt, but also from ec2 metadata (which goes into the salt mine)
01:52 subsignal joined #salt
01:53 froztbyte I'd be curious to read a succinct summary of that if you're posting stuff sometime
01:54 froztbyte Do you do it via a nice subset of IAM that's limited to just the vpc you want?
01:55 pdayton joined #salt
01:58 alexhayes joined #salt
01:59 seev salt is the best
02:00 seev it's also guaranteed job security since no one knows it, like Chef or Puppet
02:14 otter768 joined #salt
02:25 iggy we're kind of hoping that doesn't last forever
02:28 VR-Jack a little more stability would be nice. I still get inconsistent results sometimes
02:31 iggy says the guy with the most hideous top file know to man
02:31 VR-Jack lol, at least that part is working right. :P
02:37 favadi joined #salt
02:48 ITChap joined #salt
02:53 subsignal joined #salt
02:53 donmichelangelo joined #salt
02:55 mens joined #salt
03:22 c10b10 joined #salt
03:22 TyrfingMjolnir joined #salt
03:28 andrej How do I make file.recurse actually make symlinks on the far end if they exist on the master inside a directory structure?
03:29 andrej I end up with files only, where I expect 8 files and 9 symlinks
03:31 bhosmer_ joined #salt
03:32 VR-Jack andrej: make sure you set keep_symlinks: True
03:34 andrej VR-Jack , that's set ... as well force_symlinks: True
03:37 VR-Jack hmm. try salt-call with a -l debug?
03:37 andrej http://pastebin.com/R3BCQKgw
03:37 andrej OK, I'll try that
03:41 VR-Jack ahh, sweet. works even in my pillar jinja. muahaha. {% if salt['file.file_exists'](value ~ '/hosts.sls') %}
03:41 VR-Jack and appears to obey the environment path
03:42 mosen joined #salt
03:47 favadi joined #salt
03:50 snaggleb joined #salt
03:50 snaggleb joined #salt
03:51 ramaseshan joined #salt
03:54 subsignal joined #salt
03:58 slav joined #salt
04:00 slav hi does anyone know how to provision a rackpace vm with a custom root password using salt-cloud?
04:10 andrej Hmmmm ... this makes no sense at all
04:10 andrej running salt-call with -l debug I get this at the end
04:10 andrej The function "state.highstate" is running as PID 20499 and was started at  with jid req
04:10 Freddy joined #salt
04:10 andrej after a local: data failed to compile
04:10 andrej but the state as such (w/o salt-call) works fine
04:12 Not_ joined #salt
04:12 iggy usually means you have 2 minion processes running
04:12 Garo_ I have a multi-master syndic setup with one master-of-masters. When I run test.ping on the master-of-masters to '*' I will see among the replies also lines from "master_id" which reply is the name of the master-of-masters. What and why does the master_id appear in the output?
04:14 andrej iggy ... maybe I don't quite get how to use salt-call
04:14 andrej salt-call -l debug state.sls minicom ... with minicom being the state I pastebinned earlier
04:14 iggy sounds like you're using it right
04:15 iggy service stop salt-minion
04:15 andrej ok
04:15 iggy then look to see if there are any other minion processes running
04:15 iggy kill if necessary, then start the minion back up
04:16 andrej Meh.
04:16 andrej Did that
04:16 andrej same outcome
04:17 andrej and after the salt-call I have two minions on the master
04:17 andrej Or do I need to do this on the respective minion?
04:18 ramaseshan joined #salt
04:19 andrej right
04:20 kitplummer joined #salt
04:21 kitplummer joined #salt
04:25 hemphill joined #salt
04:38 linjan joined #salt
04:40 VR-Jack ahh, my beautiful top.sls is getting more beautiful. lol
04:42 VR-Jack salt['file.find'] just made my day. heh
04:48 mosen mines just an explosion of random modules at the moment :)
04:49 VR-Jack ahh, I'm seriously having fun with pillar top.sls
04:50 ramteid joined #salt
04:50 evle1 joined #salt
04:51 rdas joined #salt
04:53 c10b10 joined #salt
04:55 VR-Jack bleh. another yaml rewrite. didn't like the organization
04:56 joeto joined #salt
05:06 kitplummer joined #salt
05:19 catpigger joined #salt
05:28 ramteid joined #salt
05:32 bhosmer_ joined #salt
05:36 dimeshake joined #salt
05:40 kusams_ joined #salt
05:55 subsignal joined #salt
05:57 stoogenmeyer_ joined #salt
05:57 saifi joined #salt
05:59 bdf joined #salt
05:59 JayFK joined #salt
06:06 kitplummer joined #salt
06:14 AndreasLutro joined #salt
06:22 bluenemo joined #salt
06:24 bluenemo joined #salt
06:34 teogop joined #salt
06:36 otter768 joined #salt
06:36 flyboy joined #salt
06:37 soren_ joined #salt
06:42 c10b10 joined #salt
06:46 VR-Jack Interesting. Sometimes when the pillar loads, it's at /srv/pillar, and other times it's at /.
06:46 VR-Jack hmm, or I should say the trip to salt['file.find'] is.
06:48 yuhl_work__ left #salt
06:48 yuhl_work___ joined #salt
06:50 jalaziz joined #salt
06:50 Auroch joined #salt
06:54 viq joined #salt
06:56 subsignal joined #salt
06:56 stoogenmeyer__ joined #salt
06:58 joeto joined #salt
07:02 keimlink joined #salt
07:06 evle1 joined #salt
07:06 stanchan joined #salt
07:08 favadi joined #salt
07:11 yuhl_work___ left #salt
07:12 yuhl_work____ joined #salt
07:14 plindgren joined #salt
07:17 kawa2014 joined #salt
07:19 fbergroth joined #salt
07:21 kitplummer joined #salt
07:22 Eliz_ joined #salt
07:22 saifi joined #salt
07:23 thehaven joined #salt
07:26 bezaban joined #salt
07:26 noway__ joined #salt
07:32 yggdrasi1 joined #salt
07:33 bhosmer_ joined #salt
07:33 soren_ joined #salt
07:42 iwishiwerearobot joined #salt
07:50 rofl____ is 2015.5 released?
07:51 babilen The answer to that question depends on what you mean by "released"
07:53 lb1a joined #salt
07:53 babilen But, given that you have been asking this question for a while now, I'd go with "no"
07:53 rofl____ babilen;)
07:54 PI-Lloyd babilen, any word on when 2015.5.x will be available in the Ubuntu repos? Getting tired of the bootstrap script now.
07:54 rofl____ the definition of released is not so easy to understand i guess
07:54 rofl____ tagged < released
07:54 babilen PI-Lloyd: I haven't heard anything about that, but I wouldn't expect too long a delay
07:54 plindgren im so sleepy
07:55 PI-Lloyd ok cool, our new infrastructure will be needing it very soon ;)
07:55 plindgren took me 1 hour to realize i could git files to a server directly instead of gitting files to the salt master and then move them from there
07:55 babilen rofl____: The main problem is that saltstack is not a normal upstream that simply releases software, but also one that maintains the packages itself. A release is not considered to be "released" before packages for all supported operating systems and distributions have been provided.
07:56 rofl____ babilen: i think debian/ubuntu packages are community maintained by one guy as well
07:56 subsignal joined #salt
07:57 babilen Is any of you guys working with the salt provisioner in packer? I'd like to create boxes based on the current salt configuration in GitFS, but the provisioner only supports masterless. Not sure how to get all the various salt repositories into /srv/salt for that to work.
07:58 babilen rofl____: They are packaged by Joe Healy for saltstack, Debian and Ubuntu. The main difference is that saltstack doesn't simply say "Here, we've released the new version now package it as you see fit" (like any other upstream), but asks packagers to have their packages ready *before* a release is announced.
08:00 AndreasLutro odd approach
08:02 babilen The idea is, I guess, that people who deal with heterogenous infrastructures can be sure that they can roll out new salt versions to all supported systems at the same time.
08:02 keimlink joined #salt
08:02 ThomasJ An approach I appreciate :)
08:05 maciek joined #salt
08:05 denys joined #salt
08:06 keimlink joined #salt
08:07 CeBe joined #salt
08:11 linjan joined #salt
08:11 losh joined #salt
08:15 pf_moore joined #salt
08:17 KermitTheFragger joined #salt
08:18 chiui joined #salt
08:21 ninkotech joined #salt
08:26 Xevian joined #salt
08:28 markm_ joined #salt
08:31 c10b10 joined #salt
08:35 RobbieAB|laptop joined #salt
08:37 otter768 joined #salt
08:39 Amit_ joined #salt
08:39 Amit_ Hello Group. I wanted to know if anyone has integrated Redis with SaltStack for publishing
08:50 kusams joined #salt
08:50 yawniek joined #salt
08:51 mpanetta joined #salt
08:51 yawniek should'nt this:  '- watch:\n file: /etc/nginx/conf.d/*'  work? it fails
08:52 AndreasLutro yawniek: only if you have states with IDs starting with /etc/nginx/conf.d/
08:57 udiabon joined #salt
08:58 yawniek AndreasLutro: ah, my state is called /etc/nginx/conf.d, so i guess / needed
08:58 udiabon Hi, how to set the configuration option “docker.version” of the module dockerio?
08:59 yawniek yep that did the trick
09:02 fredvd joined #salt
09:10 kitplummer joined #salt
09:15 hebz0rl joined #salt
09:16 wnkz joined #salt
09:20 inad922 joined #salt
09:25 izak joined #salt
09:28 Auroch joined #salt
09:29 johndeo joined #salt
09:31 izak Hi all. I have a (hopefully simple) question:
09:31 izak is Salt capable of monotiring and keeping alive (restarting a service)?
09:31 izak I know I can do this in state.highstate
09:31 izak by using "service.running"
09:32 izak but state.highstate does not get called continuously, only on occation
09:32 johndeo If I have a windows exe file with an install file (exe -f installconfig for example), how to put this into the install_flags in conjunction with the saltmaster filesystem?
09:32 johndeo izak: set up a schedule task
09:33 johndeo izak: Hang on, let me take a quick look.
09:33 izak johndeo: Thanks for the answer. Is this "good practice" in salt?
09:33 lothiraldan joined #salt
09:33 izak Or would it be better to let systemd handle this by specifing a "Restart=" in my systemd service files?
09:34 bhosmer_ joined #salt
09:34 Auroch joined #salt
09:35 johndeo Well, then you would have 2 places to check for things running, I would try to keep as much as possible in salt.
09:36 izak yes, my thoughts exactly. Also I want to avoid editing or wrapping third party *.service files
09:36 izak But, I am worried that trying to do this in salt is somehow not what it was designed for?
09:37 johndeo You can have a specific schedule for this service, with salt it doesn't have to be a system wide schedule, like puppet.
09:39 izak well, basically I want it to be monitoring the service continually and restart it when it fails. Check every few seconds. (That is what systemd can do)
09:40 izak The examples at http://docs.saltstack.com/en/latest/topics/jobs/schedule.html seem to use longer time periods, so not sure if this usecase is intented?
09:42 johndeo I've done a 3 minute interval, seconds, it seems, are also feasible. However if you have a lot of those services, I think it becomes a bit of a burden on your system, I'd say. Maybe setup systemd with satlstack then to manage them.
09:43 peters-tx joined #salt
09:43 matthew-parlette joined #salt
09:43 danielcb joined #salt
09:44 babilen izak: I wouldn't use salt for that. You can either configure systemd in a way you want or use tools like consul
09:45 izak Thanks johndea and babilen , that confirms my suspicion. Just wanted to double check before I go and create wrapper systemd *.service files for my services
09:48 babilen izak: I mean you can implement that on top of salt, but it simply isn't the best tool for that.
09:49 johndeo Does anyone have an answer to my question. I have 1 solution, which is putting it on the remote filesystem with a highstate and copying it over, but is there another / better solution?
09:54 udiabon left #salt
09:56 Rockj joined #salt
09:56 jhauser joined #salt
09:57 supersheep joined #salt
09:58 izak babilen: Yes, I understand. Thanks.
09:58 izak Johndeo: sorry I could not be of help to your question.
09:58 julez joined #salt
10:01 johndeo izak: that's ok.
10:01 c10b10 joined #salt
10:01 * babilen hasn't touched MS Windows in more than a decade
10:01 favadi joined #salt
10:03 johndeo Hmm.. lucky you! Because on some fronts it still sucks, windows that is. But I've managed to deal with it in the best way possible through saltstack.
10:08 VSpike johndeo: are you using the windows repository to install it?
10:08 johndeo VSpike: yes.
10:10 VSpike oh wait, sorry. had to read the question a few times before I got what you meant.
10:10 VSpike Not sure of any easy way to solve that
10:11 babilen I have no idea what "put it into install_flags" means and simply concluded that it is some Windows specific SNAFU that he doesn't want to understand
10:11 VSpike Apart from wrapping the whole thing in a custom installer. For example, are there any tools to create self-extracting archives which allow you to run a custom command line on extraction?
10:12 ninkotech joined #salt
10:12 VSpike babilen: I read it to mean you need an exectutable installer and a file containing config data in order to install this program. The current windows repo system doesn't allow for that, AFAICT
10:14 JayFK joined #salt
10:14 babilen Why wouldn't that data be part of the installer to begin with?
10:14 ildiroen joined #salt
10:19 Segfault_ joined #salt
10:22 ninkotech joined #salt
10:23 bluenemo joined #salt
10:25 tmh1999 joined #salt
10:25 ninkotech_ joined #salt
10:28 martineg_ joined #salt
10:28 tmh1999 left #salt
10:28 tmh1999 joined #salt
10:29 Berty_ joined #salt
10:33 ninkotech joined #salt
10:35 Auroch joined #salt
10:38 otter768 joined #salt
10:43 paolo hi, i have a problem running highstate. the first time it finishes with an error (State 'pip.installed' found in SLS u'circus' is unavailable), but launching the highstate a second time the error disappears and everything goes fine. Can anyone have a look? http://pastebin.com/NEh2HWXK
10:47 martineg_ left #salt
10:48 markm_ joined #salt
10:51 dr4Ke joined #salt
10:54 denys joined #salt
10:55 dr4Ke Hi all, I'm new there. I'm working on the grains module, fixes and addings, and I'd like to discuss some changes I'd like to propose.
10:57 babilen okay
10:57 babilen You may proceed
10:57 bhosmer_ joined #salt
10:59 subsignal joined #salt
10:59 kitplummer joined #salt
11:01 dr4Ke Thanks. I opened a ticket https://github.com/saltstack/salt/issues/23446 to add a grains.set function that could be used for setting a value inside a nested grain.
11:02 factor joined #salt
11:03 babilen Looks good
11:03 dr4Ke I'm writing the function right now, for my own needs, and there are some caveats I'd like to avoid.
11:03 dr4Ke These are related to the way grains.get works.
11:06 dr4Ke If I have a grain like this {list: [elt1, elt2]}, the function grains.get 'list:elt2' returns with a None object.
11:10 dr4Ke I have some trouble exposing my concerns, sorry. I need to think a little more about it, to be clear.
11:12 babilen Yeah, Python is a little ... well .. limited in that it doesn't allow you to provide key sequences for map access and writes
11:15 markm_ joined #salt
11:16 joehh babilen, rofl____: working on packages now - not expecting any issues so should have them available for testing in next few hours
11:17 favadi joined #salt
11:18 CeBe1 joined #salt
11:18 dr4Ke Let's say I want to set this (nested) grain : 'list:elt3' to some value 'val3'. But I have an existing grain (as specified above). How should we behave?
11:19 dr4Ke 1. append a new element to the list : '{elt3: val3}'
11:19 dr4Ke 2. fail
11:21 dr4Ke The dilemn is getting worse if I want to set this grain: 'list:elt2' to 'val2'. Do I append a new element {elt2: val2}? Do I replace the existing element 'elt2' with a new element {elt2: val2}? Do I fail?
11:23 AndreasLutro if list is not a dictionary it should error out, in my opinion
11:23 AndreasLutro because the nested syntax with : assumes everything is dicts
11:24 babilen I would implement it the same way Clojure implements assoc-in: You can provide a sequence of keys that creates nested dicts and sets the last "leaf" to the provided value. If you want to extend a list at that point you should simply get it before, extend it and then set the new value. Otherwise you could use it like "update-in" http://clojuredocs.org/clojure.core/update-in
11:25 babilen The problem is that Python is not really uniform in its behaviour across data structures. (i.e. There is no .get() for lists that returns the n-th element)
11:25 ndrei joined #salt
11:26 babilen http://clojuredocs.org/clojure.core/assoc-in would be the behaviour I had in mind
11:26 rofl____ joehh: excellent! :)
11:27 ALLmightySPIFF joined #salt
11:29 mage_ is there a way to set an ENV variable in an SLS ?
11:29 joehh rofl____: which os/releases are you most interested in? - i'll let you know when it is availble
11:29 JDiPierro joined #salt
11:29 rofl____ joehh: debian wheezy/debian jessie :-)
11:30 babilen mage_: environment variables have no meaning in the context of SLS files. What are you trying to do?
11:30 babilen +1 for Debian :)
11:30 evle joined #salt
11:31 mage_ babilen: I'm building virtual env with rbenv but I have some old version of ruby which require a specific version of gcc (so I need to set CC=gccxxx before)
11:32 babilen mage_: How exactly is that related to salt?
11:32 c10b10 joined #salt
11:32 babilen (provide states et al. on http://refheap.com )
11:32 mage_ I'm using http://docs.saltstack.com/en/latest/ref/states/all/salt.states.rbenv.html#module-salt.states.rbenv
11:35 eSgr joined #salt
11:38 bhosmer__ joined #salt
11:39 johndeo VSpike: Thanks for your answer. I was away for lunch.
11:40 johndeo Babilen: It is part of the installer, but apparently not part of the saltstack windows repo with the install_flags options as of yet.
11:43 babilen mage_: I see no support for setting environment variables in that state -- Given your constraint to use "an old version of Ruby" (why?!) you might have to implement it using cmd.run or simply for states.rbenv by placing the state in _states and the corresponding module in _modules. You could then override whatever needs changes in there and expose the state as something like states.rbenv_why_are_we_so_old_st
11:44 babilen ill
11:44 babilen (or another, less intuitive, name)
11:44 dr4Ke AndreasLutro: should we avoid using lists in grains? In contrast, the state system is using lists with dict values everywhere.
11:46 Grokzen joined #salt
11:47 AndreasLutro dr4Ke: I don't really know, I know that I avoid lists in pillars most of the time because it's easier to merge/manipulate dicts but it really depends on your use case
11:47 favadi left #salt
11:49 tmclaugh[work] joined #salt
11:51 andrein joined #salt
11:53 bhosmer_ joined #salt
11:54 andrein Hi guys, I'm looking for a way to automate the config generation fo our keepalived load balancers. I'm trying to figure out a way to get the IP addresses of all my web servers when highstate is run on the loadbalancers, any ideas how to do this?
11:55 Twiglet Check out salt mine
11:55 Twiglet http://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.mine.html
11:56 dr4Ke OK AndreasLutro. I'm trying to think globally here. I would like the function to work as most would expect it to. So I think I must consider that grains could be any form of dicts and lists. I'm updating the github ticket with new behavior: appending values in list or adding
11:56 dr4Ke in dicts when appropriate
11:57 andrein Twiglet: thanks a lot for the link :) The Haproxy example looks good :)
12:00 tmclaugh[work] joined #salt
12:00 mage_ babilen: I think I'll go with the cmd.run way :)
12:03 wwwBUKOLAYcom joined #salt
12:07 KennethWilke joined #salt
12:09 markm__ joined #salt
12:09 saifi joined #salt
12:13 jonatas_oliveira joined #salt
12:16 huddy joined #salt
12:25 markm_ joined #salt
12:30 joehh rofl____, babilen and others: debian packages available at debian.saltstack.com in unstable-testing, jessie-testing and wheezy-testing
12:30 grrrrr joined #salt
12:31 joehh well - almost - files uploading now - should be there in a minute or two
12:35 cmcmacken joined #salt
12:36 joehh up now - now to ubuntu
12:39 otter768 joined #salt
12:40 TooLmaN joined #salt
12:41 dendazen joined #salt
12:45 subsignal joined #salt
12:46 kitplummer joined #salt
12:51 kawa2014 joined #salt
12:51 ndrei joined #salt
12:52 babilen +1
12:52 catpig joined #salt
12:57 djinni` joined #salt
12:57 emaninpa joined #salt
13:00 numkem joined #salt
13:03 hasues joined #salt
13:03 jakubm hi, I'm trying to come up with a solution for mutual restart between cron and salt-minion - what would be the best command to check connectivity between minion and master on minion host?
13:03 hasues left #salt
13:04 jdesilet joined #salt
13:05 cpowell joined #salt
13:05 JDiPierro joined #salt
13:05 cpowell joined #salt
13:06 leszq joined #salt
13:11 julez joined #salt
13:11 markm_ joined #salt
13:12 _JZ_ joined #salt
13:14 dyasny joined #salt
13:18 tkharju joined #salt
13:18 ferbla joined #salt
13:19 jdesilet joined #salt
13:20 c10b10 joined #salt
13:21 iwishiwerearobot joined #salt
13:22 benvon joined #salt
13:23 Ahlee is there a way to apply state_output=Terse to the jobs runner? or am i gonna have to write more code
13:24 fxhp joined #salt
13:25 XenophonF joined #salt
13:25 mpanetta joined #salt
13:25 Tyrm joined #salt
13:35 racooper joined #salt
13:36 markm_ joined #salt
13:37 lothiraldan joined #salt
13:37 rideh joined #salt
13:38 losh joined #salt
13:38 cpowell joined #salt
13:39 mapu joined #salt
13:40 saifi joined #salt
13:40 kawa2014 joined #salt
13:41 Deevolution joined #salt
13:42 JDiPierro joined #salt
13:43 kawa2014 joined #salt
13:45 dkrae joined #salt
13:48 kitplummer joined #salt
13:51 bhosmer joined #salt
13:54 amcorreia joined #salt
13:55 subsigna_ joined #salt
13:55 edrocks joined #salt
13:55 timoguin joined #salt
13:55 rideh joined #salt
13:56 rideh joined #salt
13:58 ligthert_ jakubm: a netstat to see if you see a TCP session?
14:03 masterkorp joined #salt
14:03 masterkorp hello everyone
14:04 masterkorp with salt-cloud, its possible to run highstate right after the salt-cloud bootstrap
14:06 ligthert_ I am not sure, but either salt-cloud does this for you or the minion does this once its bootstrapped.
14:08 masterkorp http://docs.saltstack.com/en/latest/topics/cloud/reactor.html
14:08 masterkorp looking at this
14:10 faliarin joined #salt
14:17 iggy PI-Lloyd: salt-bootstrap is the only way to get a known version (without running your own repo server)
14:17 XenophonF something's wrong with my salt-cloud config: i can watch it push keys and possibly a configuration file to a new EC2 instance, but it immediately deletes them without trying to install salt-minion or start the salt-minion service
14:17 andrew_v joined #salt
14:18 iggy babilen: masterless supports gitfs in 2015.5
14:18 kusams joined #salt
14:18 thayne joined #salt
14:20 masterkorp is it possible to have this file in  top file in the repo ?/etc/salt/master.d/reactor.conf
14:20 iggy top file is strictly for targeting, you could target a state that installs that file though
14:21 masterkorp https://github.com/saltstack-formulas/salt-formula/blob/master/salt/files/master.d/reactor.conf
14:21 masterkorp aparently I can
14:22 ksj hey, I'm looking into setting grains directly from the master. I have a working set up that looks something like this: http://dpaste.com/11PBHF0 . The problem is that in production we have about a hundred machines, so either I have to list them all singly, or use a jinja loop (which is somewhat ugly). I'd prefer to use the powerful targetting you can use in the top file, but can't figure out how to use that to
14:22 ksj apply grains, rather than reading from it...
14:23 iggy A. if you are going this route, you'd be better off putting your roles in a pillar... it's more secure (a minion can change it's own grains)
14:24 iggy B. There are very powerful matchers built-in to SaltStack... there are also some ext_pillar's that are quite powerful (reclass, etc.)
14:24 babilen iggy: Yay, that's great news.
14:24 ksj iggy: can change it's own? you mean if someone has root and they edit the file....yeah...but it shouldn't really be a risk right? assuming everything else is locked down
14:25 iggy ksj: depends on your environment... it's not for me because it's just me and one other person, but if you have a lot of users...
14:25 ksj iggy: yeah, I like the targetting of the top file....oh wait....you mean I should be targetting using the top.sls pillar? let me go have an experiment. my head's spinning
14:25 premera joined #salt
14:26 ksj iggy: in the end, we want to lock it down so only one or two people have access to the boxes. everyone else will, theoretically, do all work via salt
14:26 iggy ksj: it was a general message as we really don't have enough info to go on
14:26 ksj how well that will work in practice, I don't know
14:28 rojem joined #salt
14:28 XenophonF how does salt-cloud deploy the minion on new instances?
14:28 XenophonF does it use salt-ssh?
14:28 vstoniest joined #salt
14:28 XenophonF salt-bootstrap?
14:28 XenophonF some other third thing?
14:28 iggy not exactly salt-ssh, but some of the same code + salt-bootstrap
14:30 masterkorp XenophonF: it calls salt-bootstrap by default
14:30 masterkorp you can change the bootstrap script
14:31 XenophonF thanks all - looking through that part of the salt-cloud troubleshooting docs now
14:33 ksj iggy: at it's simplest, I want to be able to say 'server[0-30] should have the grain "role: \n  - webfrontend"'
14:33 ksj and have salt add the approriate grain in /etc/grains on the correct minions
14:34 ksj it almost feels like a small db to keep track of which machines will perform what functions, but I don't want to overcomplicate it
14:35 ksj and I know salt already has excellent tools for matching in the top files....I just can't figure out the right way to go about it
14:36 c10b10 joined #salt
14:36 markm joined #salt
14:37 XenophonF ksj: i use pillar to make role assignments instead of grains - https://github.com/irtnog/salt-states/blob/master/top.sls#L106
14:37 hackel joined #salt
14:37 clintberry_1 joined #salt
14:37 XenophonF i also use pillar to assign the salt environment (dev/test/stage/prod)
14:38 ksj XenophonF: wow thanks, that's really helpful, seeing a realworld example
14:39 XenophonF ksj: in .../pillar/environment/template.sls, I have the following code:
14:39 XenophonF environment: {{ sls.split('.')[1] }}
14:39 masterkorp One questiont
14:39 XenophonF and so .../pillar/environment/development.sls, testing.sls, et seq. are all symlinks pointing at template.sls
14:39 kusams joined #salt
14:40 otter768 joined #salt
14:40 XenophonF and the same setup in .../pillar/role/
14:40 markm joined #salt
14:41 masterkorp the salt event salt/cloud/<vm>/created is fired after the salt/cloud/<vm>/deploy_script
14:41 ksj XenophonF: ok...that confused me...
14:41 ksj I'm gonna get coffee...
14:41 masterkorp coffffeeeeeeeee"
14:41 XenophonF so in .../pillar/top.sls, each minion gets assigned its environment and role, so "uxedevbsdsalt01.irtnog.net" gets the environment.development and role.salt-master pillars
14:41 masterkorp the salt event salt/cloud/<vm>/created is fired after the
14:41 masterkorp | salt/cloud/<vm>/deploy_script ?
14:42 masterkorp or before ?
14:42 XenophonF ksj: ping me after the coffee break and i'll give you a more detailed example
14:44 iggy masterkorp: you'll probably have to test or read code
14:44 mage_ stupid question, is salt supposed to run things one after each other ? for example I have sls to create venvs is it not possible to create them in // ?
14:45 gladiatr joined #salt
14:45 masterkorp iggy: yeah i am applying TIAS methodology
14:45 amcorreia joined #salt
14:46 sandah joined #salt
14:48 davisj so, returners.... The minions must connect directly? or does the master do the forwarding to addition returners?
14:49 ksj XenophonF: ok, I have caffeine inside me...
14:49 * davisj needs to dump job info to a db that minions won't have access to.
14:51 debian112 joined #salt
14:53 ksj XenophonF: if possible, can you give me an example of the pillar file you're matching against in your github example. I'm struggling a little bit to see how it fits, especially since I thought grains were the "correct" way to do this stuff
14:54 XenophonF ksj: yeah - just a second i'm importing some stuff from my actual configs
14:54 XenophonF it will be a little incomplete as some of my states don't exist on github yet
14:55 julez joined #salt
14:55 fyb3r joined #salt
14:55 julez joined #salt
14:56 sdm24 joined #salt
14:56 XenophonF ksj: consider my minecraft server
14:57 XenophonF ksj: here's the state assignment - https://github.com/irtnog/salt-states/blob/master/top.sls#L285
14:57 ferbla_ joined #salt
14:57 Ahlee so.  Apparently pillar.item can return a different value than pillar.get: https://gist.github.com/jalons/f70ba03182eebf9635d7
14:58 XenophonF ksj: that means if I assign a server the "minecraft" role, it will have the spigotmc SLS sent to it
14:58 XenophonF which (use your imagination here a little) installs java, builds and installs spigot, generates a world, etc.
14:59 rojem_ joined #salt
14:59 XenophonF ksj: the environment and role assignment happens in pillar - https://github.com/irtnog/salt-pillar-example/blob/master/top.sls#L10
14:59 ksj XenophonF: sure, that makes sense, but how exactly are you assigning the role via pilar?
14:59 thayne joined #salt
14:59 ksj hold on, I sent that the same time I received your last message, let me read
15:00 conan_the_destro joined #salt
15:00 XenophonF ksj: environment/production.sls sets "environment" to "production" (with a little jinja wizardry)
15:00 masterkorp http://docs.saltstack.com/en/latest/ref/states/all/salt.states.slack.html#module-salt.states.slack
15:01 schristensen joined #salt
15:01 XenophonF ksj: it would probably be simpler if you imagined that environment/production.sls had as its contents "environment: production"
15:01 ksj ok, but you still have to list each server separately, no? my hope was that I could target using the top file targeting (e.g. bracket sequences) to "tag" a set of machines with a role
15:01 masterkorp this is only available is 2015.5 ?
15:01 ksj it's a lot clearer to see [0-100] than all explicitly listed
15:01 XenophonF ksj: sure! you can do that with a different kind of match - let me give you an example
15:02 primechuck joined #salt
15:03 primechuck joined #salt
15:04 XenophonF ksj: here's another way to target - https://github.com/irtnog/salt-pillar-example/blob/master/top.sls#L10
15:04 XenophonF ksj: that just uses a simple glob, and thankfully my naming conventions will cooperate
15:04 ndrei joined #salt
15:05 XenophonF you might still need to send minion-specific configs
15:05 XenophonF perhaps minecraft is a bad example in that you can't set up a fleet of identical minecraft servers, and then round-robin load balance among them
15:05 XenophonF you could with apache or nginx though
15:05 ksj XenophonF: thanks, you've been really helpful. Let me have an experiment and try to get my head round it, but pillar seems to be the way to go in preference to grains
15:06 XenophonF ksj: you can do the same thing with grains, but i don't think that's secure
15:06 XenophonF grains are set in the minion config, so a hacker could change the "role" or "environment" grain for a particular server and gain access to settings you might not want them to have
15:07 VR-Jack hmmm. doing salt[file] commands can be weird, since there's no guarantee of what your cwd is.
15:10 bcvrjqruqqau joined #salt
15:11 Captain joined #salt
15:16 ferbla joined #salt
15:21 JayFK joined #salt
15:21 JayFK joined #salt
15:27 kitplummer joined #salt
15:31 jalbretsen joined #salt
15:32 Laogeodritt joined #salt
15:34 jimklo joined #salt
15:36 JayFK joined #salt
15:37 Auroch joined #salt
15:38 bhosmer joined #salt
15:39 VR-Jack iggy: my pillar top is very custom to my needs, but I finally have it auto-generating hostgroup stanzas based on the pillar data itself.
15:40 synical joined #salt
15:41 Georgyo joined #salt
15:42 Laogeodritt joined #salt
15:43 smcquay joined #salt
15:43 writtenoff joined #salt
15:47 ndrei joined #salt
15:47 ProT-0-TypE joined #salt
15:49 johndeo babilen: and because of the software and variables that differ for every environment, like in my case this windows NTP executable.
15:51 jakubm ligthert_: hah. good idea - thanks
15:51 kunersdorf joined #salt
15:51 sdm24 joined #salt
15:52 c10b10 joined #salt
15:52 lothiraldan joined #salt
15:58 VSpike When doing a state.highstate call which takes a while, is salt meant just to give up and exit with no message?
15:58 c10b10 joined #salt
15:59 iggy VSpike: yes
15:59 Fiber^ joined #salt
15:59 iggy you can adjust that timeout or check the job cache later to see what happened
15:59 VSpike OK. Seemed odd to me but I wasnt sure if it was intended. I would have expected a message to say it had timed out, and perhaps a job id output too
16:00 iggy VSpike: new versions have exactly that ;)
16:00 iggy think 2015.5 will have it, if not, it'll be in -next
16:01 pdayton joined #salt
16:01 VSpike aha. great! that will be less surprising
16:02 JayFK joined #salt
16:03 the_lalelu joined #salt
16:04 bhosmer joined #salt
16:06 bhosmer_ joined #salt
16:08 wendall911 joined #salt
16:10 jespada joined #salt
16:11 JDiPierro joined #salt
16:13 ksj joined #salt
16:16 enarciso joined #salt
16:18 bastiandg joined #salt
16:18 martintamare joined #salt
16:18 martintamare Hello guys, just move two year of puppet code on saltstack in two weeks :) feel happy !
16:18 martintamare But
16:18 martintamare :)
16:18 martintamare i'm missing the concat module of puppet a lot.
16:19 markm joined #salt
16:19 martintamare file.accumulated helps a lot, but it would be better to pass a dictionnary instead of just the text
16:19 martintamare does anyone find a workaround ?
16:19 martintamare can i create a request ?
16:19 thayne joined #salt
16:20 Ahlee You can certainly create a feature request on github
16:20 Ahlee just need to find somebody that knew how puppet's concat worked :)
16:20 Ahlee s/worked/works/
16:21 martintamare ahah ^^
16:21 bluenemo joined #salt
16:21 bluenemo joined #salt
16:24 joeto1 joined #salt
16:26 VSpike This is an odd one. I add a few Ubuntu PPAs using pkgrepo.managed and they are all working except one. The state doesnt fail, but installing the packages does because it hasn't added the key
16:26 VSpike I've run this on two machines now with the same result.
16:26 monkey661 joined #salt
16:27 kitplummer joined #salt
16:28 iggy martintamare: other people have asked for something similar
16:28 dh__ joined #salt
16:28 kitplummer joined #salt
16:28 iggy don't know if they actually opened FRs for it though
16:28 iggy VSpike: you'll want to pastebin the code
16:29 martintamare Just did, will look into the code in the next to see if i can make this a pull request ;)
16:29 KyleG joined #salt
16:29 KyleG joined #salt
16:29 VSpike iggy: log output https://bpaste.net/show/e99f37afed29 and code https://bpaste.net/show/d4905006b44b
16:30 VSpike the included state for completeness https://bpaste.net/show/57e5c381e4f3
16:30 dh joined #salt
16:31 iggy VSpike: all those requires shouldn't be required... salt works top-down
16:31 pfallenop joined #salt
16:31 iggy unless you have other explicit dependencies elsewhere that is reordering things
16:32 VSpike iggy: hm, is it good practice to depend on ordering?
16:33 pcn Is there a procedure for installing salt + modules in a way that doesn't affect the rest of the system, like an omnibus install?  I'm wondering about installing modules like deps for gitfs.
16:33 VSpike I always thought it would be better to be explicit where a dependancy exists
16:33 VSpike The odd thing is that running apt-add-repository -y ppa:ondrej/php5 fixes it
16:33 VSpike Which is exactly what the minion says it's doing too
16:35 terinjokes joined #salt
16:35 iggy VSpike: there was a fairly compelling presentaion at the saltconf that said so
16:35 spookah joined #salt
16:36 Tecnico1931 joined #salt
16:37 VSpike :) fair enough
16:37 iggy pcn: there are directions for installing to a virtualenv
16:38 iggy pcn: the problem being, aside from gitpython, you actually need some outside compiled libraries
16:38 iggy VSpike: try setting the keyid/keyserver on that ppa stanza?
16:39 adelcast joined #salt
16:39 sudosteph joined #salt
16:39 iggy VSpike: and did you make sure the minions have all the correct deps installed (and if installed via pip, did you use reload_modules: True?)
16:39 jasonrm joined #salt
16:40 murrdoc joined #salt
16:40 VSpike Aha, a smoking gun https://bpaste.net/show/df4358f466aa
16:40 otter768 joined #salt
16:41 VSpike I think it might be something to do with the name of the author
16:41 VSpike Ondřej Surý
16:42 mattrobenolt joined #salt
16:43 xeryuz joined #salt
16:46 murrdoc On Rage Surrey
16:48 desposo joined #salt
16:49 rojem joined #salt
16:51 aparsons joined #salt
16:53 masterkorp Can I reboot on the middle of a formula ?
16:53 ksj ok, I'm stil fighting pushing grains out to minions. I've got a pastebin here http://dpaste.com/1N1VD4Y. If anyone can point me in the right direction it would be much much appreciated
16:58 VSpike Dashed underhand of the fellow to have such an un-British sort of name
16:59 masterkorp is there a state to run a dist-upgrade ?
17:00 Diep joined #salt
17:01 iggy masterkorp: there's a module function (salt.modules.aptpkg.upgrade)
17:02 iggy and you can use salt.states.module.* to run that
17:03 iggy but a dist-upgrade is probably better just run from the command line as a one-off kind of thing
17:03 subsignal joined #salt
17:04 sdm24 ksj: Is there a reason you are using pillars to push grains to push roles? Why not just use the pillars to push roles?
17:05 ksj sdm24: because I couldn't think of a sane setup for that. I want to have a clean pillar like the one in the example
17:05 masterkorp iggy: yeah it to upgrade freshly bootrapped machines
17:05 ksj i.e. where I organise things by a hierarchy of roles, and add machine names to each role
17:05 sdm24 ok. Have you tried using grains.present and having the for loop surround the whole ID?
17:07 ksj sdm24: yeah, I did, but that was about an hour ago and I can't remember exactly what happened. I'm doing it manually now, with no jinja in the state file, and still getting the same result - it adds an extra dash to the first entry
17:07 masterkorp what are the best practices for rebooting a machine withing a formula ?
17:07 iggy don't
17:08 theologian joined #salt
17:08 ksj sdm24: oh wait, no, I looked at doing it with present, rather than list_present, but that only works for single values. I want multiple roles
17:08 ksj so I have to use list_present
17:08 ksj and if I try looping round the whole thing with list_present it was giving me errors
17:08 masterkorp iggy: so what is the best way to reboot a machine with salt ?
17:09 sdm24 oh, would doing grains.present in a for loop just update to the last value?
17:09 iggy masterkorp: if this is a "on startup" kind of thing, use a reactor+orchestrate to run the upgrade, then reboot after everything is done
17:09 ksj sdm24: I think so, haven't tried
17:09 masterkorp iggy: this formula will be called by the reactor
17:09 masterkorp only
17:09 sdm24 ksj: the docs make it sound like thats what happens
17:11 ksj sdm24: either way, if you try the example list_present from the docs (the second example with web and dev roles), it cocks up the formatting, as in my pastebin
17:11 sdm24 https://gist.github.com/sdm24/c5fbdac17b5c9a5b68d8
17:11 sdm24 what about that?
17:11 sdm24 I  believe it will still add each role to the grain list
17:11 sdm24 and maybe it was just a strange formatting issue adding the extra -
17:12 ksj http://dpaste.com/3YXFR6Y
17:13 CeBe1 joined #salt
17:14 ksj sdm24: no, I get a lot of "not a dictionary" errors when using your method
17:15 stanchan joined #salt
17:18 Auroch joined #salt
17:23 keimlink joined #salt
17:23 masterkorp How can I emit an event inside a formula ?
17:23 forrest joined #salt
17:24 iggy event.send_master ?
17:25 masterkorp thanks
17:25 iggy http://docs.saltstack.com/en/latest/ref/states/all/salt.states.event.html#module-salt.states.event
17:25 sdm24 ksj: What if you just remove the - before {{ roles }}?
17:25 masterkorp are the events on the reactor run sequentially ?
17:25 speedlight joined #salt
17:25 speedlight joined #salt
17:25 ksj sdm24: doesn't work. I'm thinking this is a bug
17:26 aboe joined #salt
17:26 sdm24 if you do that, try removing the dashes inside {%, as that will eliminate white space in between the brackets, and you need that whitespace
17:26 tempspace joined #salt
17:27 basepi Gareth: ping
17:27 ksj I'm not working with jinja templates any more. I'm doing it manually. see here for clarification: http://dpaste.com/0J5SEZ3
17:27 * Gareth hides
17:27 basepi Gareth: actually, I'll ping you in #salt-devel
17:27 basepi =)
17:27 aboe Gareth, lol
17:28 VR-Jack ksj: but jinja templates happily drive you crazy!
17:28 Laogeodritt joined #salt
17:29 ksj VR-Jack: yeah, well I'm foregoing them for now while I get to the bottom of this issue. can anyone try that paste (http://dpaste.com/0J5SEZ3) and confirm that's the behavious they're getting - i.e. it's not just me and my setup
17:29 ek6 so what is considered the 'best' way to separate minion specific information in a pillar....yes i can do it in the pillar top or in the init files at the second layer or or.. but for like feeding formulas I want to make sure my layout is as KISS as possible
17:30 VR-Jack ek6: to my knowledge, top.sls is the way
17:30 joeto joined #salt
17:31 sdm24 ksj: I'm getting the same output as you, weird
17:31 hal58th joined #salt
17:32 bregalad joined #salt
17:32 ksj sdm24: that's valid yaml (according to http://yaml-online-parser.appspot.com/)....but it's not how grains are supposed to look
17:33 sdm24 ksj: Maybe doing a file.managed to /etc/grains with the correct format would fix it, but then I believe that requires a minion restart
17:33 ksj I'm surprised if a bug like that has remained there for a year and a half though...unless no-one uses list_present
17:33 ksj I've tried wiping the grains file completely. I always get the same result
17:33 hal58th There are some very outstanding bugs that have been there for a long time ksj
17:33 ek6 vr-jack: yeah ok that gives me high level pointers as to which 'trees' they get...and gets you to the init.sls of that tree where you what...carve them by id towards id specific sls?  or one huge fat init carved up with if's  ?
17:33 ksj you think I should raise a bug report? I don't really know how to do that
17:34 hal58th Just search https://github.com/saltstack/salt/issues for your issue. Then click on New Issue and write a detailed report. Include your salt version in your info
17:35 VR-Jack ek6: depends on how you do it. my pillar divides things into groups/types etc, and has lists to match minion names and push them to those sls
17:35 sdm24 ksj: check https://github.com/saltstack/salt/blob/develop/Contributing.rst for guidlines on what to include in your issue. It helped me post my first issue a few days ago
17:36 bregalad aboe: this is jfindlay, sorry about not installing salt-minion and salt-master
17:36 slav joined #salt
17:36 bregalad of course that makes sense :-)
17:36 VR-Jack I often do hostgroup/<hostgroupfile>.sls for mine. so hostgroup/ has hosts.sls and users.sls, etc.
17:36 ek6 vr-jack:  sure like i said...many ways to skin the cat..like how im currently doing it or you are...just didnt know if there was any kind of stated BCP or the like that I just wasn't finding
17:36 aboe no problem bregalad,
17:37 aboe please do test it, if it's all good,
17:37 fbergroth joined #salt
17:37 bregalad yeah, I'm going to retest right now
17:37 slav hi, I want to  use salt-cloud and proobision a vm to rackspace. I got my setup running however I don;t know how to set up a root password in that rackspace vm
17:37 VR-Jack ek6: how I'm doing it would be considered not BCP. Most that I have seen just carve everything in top.sls and send them to the correct modules.
17:38 ek6 vr-jack: and so you just do your gets prefaced with hostgroup type
17:38 aboe bregalad, I was about to push the changes to the devel:languages:python repo to night
17:38 VR-Jack ek6: check out my top.sls in this one https://gist.github.com/vr-jack/d53733b554fb15b12bfd
17:38 ek6 vr-jack: yeah think im just allergic to having huge top files
17:38 druonysus joined #salt
17:38 druonysus joined #salt
17:38 rojem joined #salt
17:38 ksj just to confirm before I file a bug report, this behaviour is not expected, right: http://dpaste.com/0J5SEZ3
17:39 VR-Jack ek6: you can do similar just making jinja lists at the top of top.sls and create your loops to fill it out. Can even do a salt['file.file_exists'] check to make sure the file is there first.
17:40 VR-Jack in that particular case, the file had to be there, or it wouldn't be in the hgroup
17:40 bregalad ksj: that output looks silly
17:41 VR-Jack for my "it's in this hgroup and is of this server type" the file might not exist, so I'll have to do file_exists to avoid errors.
17:41 ek6 vr-jack: ill take some time to really stare at that...but its far more then I need right now and certainly not KISS but interesting all the same
17:41 ek6 yeah i gotcha on catching the file must exist
17:41 bregalad ksj: I would file a report on that
17:42 VR-Jack if you use file.* in the master context, always do full paths. depending on how salt is called, you're working directory changes
17:42 VR-Jack ek6: mine is complex. it uses the pillar data itself to generate lists so that pillar can load. chicken-egg
17:42 murrdoc joined #salt
17:42 VR-Jack ek6: but a simple jinja list at the top of the file to list hostnames works well.
17:43 ksj bregalad: thanks
17:43 VR-Jack or even a jinja string with a salt list context L@
17:43 ek6 yeah have to do endswith because im not listing out 8000+ but i hear you
17:43 ajw0100 joined #salt
17:44 VR-Jack ek6: always the issue. one method is to make minion names descriptive so you can glob on them
17:46 VR-Jack of course, if 8k hosts have 8k different users, you're going to need to generate the user list using something other than standard pillar file, but I hear writing external pillars is easy
17:46 ek6 all of that I can do..i do thats my extensive use of endswith...but thanks im sure ill be able to steal something from your example....still kinda puzzled why nobody has written a BCP on this though
17:47 VR-Jack ek6: be aware that currently you cannot use jinja import/include in pillar top.sls. it will break
17:47 markm_ joined #salt
17:47 ek6 vr-jack: you have custom code to make sure yours doesnt shatter i take it?
17:47 VR-Jack I'm patched locally with the fix. hopefully someone else will do the git pull
17:48 VR-Jack my last comment on this report gives the fix. https://github.com/saltstack/salt/issues/23524
17:49 murrdoc nice
17:49 VR-Jack dont' have time to learn git, so can't make the patch for them
17:50 murrdoc go here https://github.com/saltstack/salt
17:50 murrdoc click Fork
17:50 murrdoc commit in your code
17:51 nahamu if it's a simple change to a single file, the gihub web interface allows you to "edit" a file which will let you modify the file in the web browser and create a pull request all from the browser...
17:51 nahamu just need to be logged in to github.
17:51 murrdoc then go https://github.com/saltstack/salt/compare/develop...develop
17:51 murrdoc and thats it
17:53 bhosmer_ joined #salt
17:55 VR-Jack hmm. i'll try that.
17:55 VR-Jack never even forked a project
17:55 forrest VR-Jack: You don't use git?
17:55 forrest VR-Jack: https://docs.saltstack.com/en/latest/topics/development/contributing.html
17:55 VR-Jack forrest: nah. still on cvs. looking at switching to hg at some point
17:56 forrest once you fork there are additional details
17:56 forrest VR-Jack: cvs?? How do your devs work :(
17:56 VR-Jack me, myself, and I. :P
17:57 forrest VR-Jack: How do your devs work ;P
17:57 forrest If someone offered me a gig and said they used CVS, I'd probably turn it down, lol.
17:57 iggy ^
18:00 VR-Jack forrest: it's legacy and haven't needed to switch yet. on my long term todo list, but since I don't employ devs, hasn't been an issue
18:00 forrest It's pretty easy to move repos from CVS to HG/git, had to do it for a gig a while back
18:00 VR-Jack okay. changes made in web int. do I do the new branch and start pull request or commit directly?
18:01 forrest VR-Jack: You can commit directly to your fork, then open a PR on your repo.
18:01 forrest should auto-open against saltstack:develop
18:03 rojem joined #salt
18:04 VR-Jack it did. do I just past the link in the comment of the open issue?
18:06 forrest uhh no, so you made the commit on your fork right? Then on your repo you create the PR, and it should have a file diff or whatever. Then you can add a comment noting the issue number it fixes and such
18:06 clintberry joined #salt
18:06 fyb3r Good god, the documentation on the reactors and events has got to be the most vague thing I've ever read. Aside for Openstack's Neutron documentation.
18:06 murrdoc VR-Jack:  where did you push up the change
18:06 murrdoc like the github url
18:07 VR-Jack https://github.com/saltstack/salt/pull/23549 and linked it to the issue
18:07 iggy fyb3r: PRs welcome ;)
18:07 murrdoc yay VR-Jack
18:07 murrdoc whoooooo
18:07 murrdoc you go girl
18:07 forrest looks good to me VR-Jack
18:07 VR-Jack very simple, though considering I don't know python.... lol
18:07 fyb3r iggy: I really am considering it lol
18:07 forrest should run a bunch of tests and if everything looks good then someone from the salt team will merge it in.
18:07 iggy I imagine they don't pass it on purpose
18:08 forrest or discuss further.
18:08 VR-Jack forrest: I'm running it on 2014.7.5 without any issues
18:08 forrest VR-Jack: For sure, you never know if it's tied to something else though so the salt devs might have additional notes.
18:08 murrdoc oh
18:09 murrdoc 2014.7.5
18:09 murrdoc hmm
18:09 murrdoc should have entioned that
18:09 iggy fyb3r: I for one would have appreciated it about 6 months ago... I ended up just trial&error'ing it until it did what I needed
18:09 VR-Jack forrest: definitely. I knew jinja import worked right in pillar sls files, just not top. This was the diff
18:09 forrest Gotcha
18:09 VR-Jack forrest: I've used it on current dev as well, but I don't work extensively in dev.
18:10 murrdoc should have made it like so then , https://github.com/saltstack/salt/compare/2014.7...vr-jack:develop
18:10 forrest Yeah, hopefully the tests will all pass and it will be good to go.
18:10 ndrei joined #salt
18:10 murrdoc so it gets into the next 2014.7 release
18:10 forrest wtf is that murrdoc
18:10 murrdoc well
18:10 murrdoc also should have pushed your change to 2014.7
18:10 murrdoc so that this shows the diff
18:10 murrdoc https://github.com/saltstack/salt/compare/2014.7...vr-jack:2014.7
18:10 forrest What he has is perfectly fine
18:11 murrdoc uh forrest hes working in 2014.7
18:11 murrdoc hows it gonna make it back to 2014.7 ?
18:11 VR-Jack it's up to devs if they want to backport
18:11 murrdoc cherry pick ?
18:11 forrest it won't unless they backport
18:11 forrest I'd be interested to see if it's still an issue in the latest release.
18:11 murrdoc thats all i trying to show
18:11 forrest gotcha
18:11 VR-Jack and given that I found the issue, there's probably not the interest to backport
18:11 forrest murrdoc: I thought you were saying he should do a full checkout/pull then merge in all the most recent stuff
18:12 forrest was like 'man too much work'
18:12 murrdoc nah
18:12 murrdoc i meant pull against 2014.7
18:12 murrdoc so u get it in the next 2014.7 release
18:12 iggy like I said, I doubt they'll see it as a bug... probably desired
18:12 signull is 2015.5 now stable?
18:13 VR-Jack iggy: if desired, it should fail properly when an import is done, not break the file_roots
18:13 murrdoc and the fact that dude only had to add a param
18:13 murrdoc to get it work
18:13 murrdoc means they didnt not implement it
18:13 iggy signull: it's tagged, but not
18:13 iggy signull: it's tagged, but not "released"
18:13 bhosmer_ joined #salt
18:13 murrdoc what about 2015.5.2
18:13 murrdoc or 2015.2
18:14 iggy 2015.2 4lyfe
18:14 VR-Jack murrdoc: was renamed to 2015.5
18:14 signull alright, thats what I was thinking. I was confused by basepi's email
18:14 VR-Jack I think
18:14 murrdoc VR-Jack:  i know
18:14 murrdoc i was trying to trigger iggy reflex
18:14 VR-Jack lol
18:16 Tyrm joined #salt
18:16 julez joined #salt
18:19 denys joined #salt
18:19 Auroch joined #salt
18:20 * iggy reflexively kicks murrdoc
18:20 murrdoc k shits normal
18:22 sdm24 ksj are you still here? I might've found a work-around for your grains issue
18:23 _JZ__ joined #salt
18:23 seev_ joined #salt
18:24 bluenemo joined #salt
18:24 bluenemo joined #salt
18:25 eclectic_ joined #salt
18:25 utahcon_ joined #salt
18:26 elektrix joined #salt
18:26 bigmston1 joined #salt
18:26 fxdgears joined #salt
18:26 jmccree joined #salt
18:26 retr0h_ joined #salt
18:26 x58 joined #salt
18:27 joehh joined #salt
18:27 qpst joined #salt
18:27 SaveTheRbtz joined #salt
18:27 Ixan joined #salt
18:27 mephx joined #salt
18:27 Ludo- joined #salt
18:27 etw joined #salt
18:27 murrdoc joined #salt
18:27 hal58th joined #salt
18:27 aparsons joined #salt
18:27 debian112 joined #salt
18:27 factor joined #salt
18:27 ckao joined #salt
18:27 heise joined #salt
18:27 SpX joined #salt
18:27 toastedpenguin joined #salt
18:27 pmcg joined #salt
18:27 seb` joined #salt
18:27 ecdhe joined #salt
18:27 babilen joined #salt
18:27 Jimlad joined #salt
18:27 jY joined #salt
18:27 TOoSmOotH joined #salt
18:27 bVector joined #salt
18:27 paha joined #salt
18:27 emid_ joined #salt
18:27 7GHAAC66H joined #salt
18:27 msciciel joined #salt
18:27 dec joined #salt
18:27 ropes joined #salt
18:27 drags joined #salt
18:27 tmmt_ joined #salt
18:27 Micromus joined #salt
18:27 Chadk joined #salt
18:27 Chadk joined #salt
18:27 xnaveira joined #salt
18:27 GnuLxUsr joined #salt
18:27 MK_FG joined #salt
18:27 MK_FG joined #salt
18:28 Ssquidly joined #salt
18:28 ph8 joined #salt
18:28 bash124512 joined #salt
18:29 saurabhs joined #salt
18:31 AbyssOne joined #salt
18:32 baweaver joined #salt
18:34 aba joined #salt
18:35 hal58th joined #salt
18:38 aba Hi. I want to manage ssh authorized_keys with salt, but (a) depending on which host and its features different, and (b) remove any key which is not explicitly allowed. ssh_auth seems not be fit for (b). I tried to do (a) with a file, and an include variable via pillar, but I failed to add values to an variable via pillars. What would be recommended?
18:40 kermit joined #salt
18:41 otter768 joined #salt
18:44 sdm24 what were you doing to include the variable via pillar? if you have a pillar with information targeted to the minion, use ['pillar.get']
18:44 forrest aba: Check out: https://github.com/saltstack-formulas/openssh-formula
18:44 aba forrest: thanks
18:44 forrest np
18:50 VR-Jack ugh. the problem of testing pxe/bootstrap on bare metal is the extremely long wait for boot.
18:51 VR-Jack should have just bought a mini-pc instead of testing on a server class that takes forever to get through proms
18:51 ageorgop joined #salt
18:53 soren_ joined #salt
18:57 Auroch joined #salt
19:01 iggy use a VM?
19:01 ajw0100 joined #salt
19:01 aparsons joined #salt
19:02 keimlink joined #salt
19:04 rojem joined #salt
19:05 clintberry joined #salt
19:06 forrest VR-Jack: I have a demo app up with a vagrant file you can use with virtualbox
19:06 forrest https://github.com/gravyboat/demo-app-1
19:06 forrest VR-Jack: ^
19:06 forrest if you want to go the VM route.
19:06 forrest just steal the vagrantfile.
19:06 thekabal joined #salt
19:07 VR-Jack forrest: thanks. what I'm doing now is just the annoyance of handling baremetal
19:07 forrest fair enough
19:07 forrest bare metal is quite lame
19:07 thekabal Hello! In a SLS file, how do I do a conditional based on fqdn, like “if fqdn contains london…”
19:08 Deevolution Anyone successfully run a salt-master via pudb?  Trying to debug an issue and can't seem to get the master to run properly via the debugger.
19:09 txomon|home joined #salt
19:09 catpig joined #salt
19:10 hybridpollo joined #salt
19:10 aba left #salt
19:10 sdm24 thekabal: http://docs.saltstack.com/en/latest/topics/tutorials/states_pt3.html#using-grains-in-sls-modules is an example
19:11 thekabal Sorry, I should have been more specific. I know how to use grains, I don’t know how to do a python/jinja(?) conditional for contains/location-independent-match
19:12 babilen {% if foo in bar %}
19:12 lionel joined #salt
19:12 thekabal Its seriously that easy? Geez, I feel very small now. :( Thanks, Babilen!
19:14 mnguyen joined #salt
19:15 babilen thekabal: Well, depending on what you want to do it might be a bit more tricky. Do you have some logic to your fqdn? I mean that simple check might be all you need, but then it is quite, well, broad.
19:15 babilen http://jinja.pocoo.org/docs/dev/templates/#other-operators
19:15 thekabal That was exactly what I needed. To elaborate, I have hosts in different regions (london, NYC, etc), and they have an airport code in their name. Based on the location, I need different NTP servers.
19:16 saltnewbie joined #salt
19:16 thekabal So literally all I needed was the if statement so I could choose a different jinja.
19:16 saltnewbie Hi,does anybody have experience installing salt in a disconnected from the internet cluster?
19:17 fyb3r so is there any way to have a reactor fire off for a new minion if auto auth is enabled?
19:17 forrest saltnewbie: You'd need to have all the install packages available to install from some sort of local repo
19:17 fyb3r nvm I figured out a work around for my issue
19:18 eliasp joined #salt
19:18 mnguyen I have a question about the salt "extension_module" setting in the minion conf file.  Does anyone know why salt blows everything away that does not start with a "_" in the "extension_modules" directory every time highstate is ran?
19:18 babilen thekabal: I personally end up writing an "parse minion name" execution module if my minions follow a strict naming scheme. I would then call it via salt['foo.parse_minion_id'](grains['id']) and match on the resulting tuple ...
19:18 mnguyen It seems like salt recreates those files every time it runs high state. The files and dirs not starting with "_" I mean
19:18 murrdoc joined #salt
19:19 pcn I have a documentation question/comment: http://docs.saltstack.com/en/latest/ref/states/top.html
19:19 babilen thekabal: The problem is that "foo in bar" might overgenerate if you have overlaps with other parts of the minion id/fqdn
19:19 c10b10 joined #salt
19:19 saltnewbie forrest: thx...what i figured...though that means cant use salt-cloud etc.
19:20 pcn For some reason, it uses the term "top file" but doesn't say "top.sls" at the beginning.  Is this because there's some abstraction for the top file (e.g. there would commonly be some other name?)
19:20 thekabal Thankfully, I don’t, but thanks for the heads up babilen!
19:20 thekabal left #salt
19:20 forrest saltnewbie: You can still install/configure salt-cloud if you wrote a provider for local stuff, but otherwise yeah.
19:21 forrest pcn: Not that I can think of
19:21 VR-Jack I use orchestrate with a server connected to port 4 of the bare metal to bring mine online. all packages are on the local repo
19:21 rojem_ joined #salt
19:22 VR-Jack if I wasn't carrying a portable vm on my surface pro, I might do it differently, though.
19:22 pcn forrest: OK, because here: http://docs.saltstack.com/en/latest/topics/tutorials/starting_states.html it introduces "top.sls", but without that link, the naming of the file is an open question on that page.
19:23 Not_ joined #salt
19:23 VR-Jack pcn: you can use something other than top.sls in a config option
19:23 forrest pcn: http://docs.saltstack.com/en/latest/topics/tutorials/starting_states.html#the-top-file ?
19:24 forrest VR-Jack: Yeah but the doc introducing it shouldn't go by a different name.
19:24 VR-Jack not sure what version that was introduced in, though
19:24 forrest pcn: I've created a PR to note this: https://github.com/saltstack/salt/pull/23555
19:25 spookah joined #salt
19:25 VR-Jack I suck at docs. I'm not even gonna try a PR on the pillar doc for my issue request.
19:26 pcn forrest: thanks, I was looking to find the docs, you are far faster.
19:26 VR-Jack I was shocked that none of them mention the master's reference id for top
19:27 forrest pcn: a couple thousand lines committed to the docs will help you with that ;)
19:31 dalexander joined #salt
19:33 VR-Jack ugh. orchestrate fails because state.sls is still running on minion from last command. that doesn't sound right
19:36 andrew_v joined #salt
19:37 jalaziz joined #salt
19:38 baweaver joined #salt
19:38 ckao joined #salt
19:38 adelcast on the project I am currently working on, we use Salt on embedded devices, which has opkg as it's package manager
19:38 badon_ joined #salt
19:39 adelcast I am working on an opkg execution module, using the aptkg module as a template
19:39 adelcast I think it shouldn't be too hard to get it into decent shape, but wanted to ping the community to find out if there are similar efforts around (from Saltstack or the community)
19:40 adelcast so...does anyone is currently working on an adding opkg support for salt?
19:40 rem5 joined #salt
19:40 hobakill joined #salt
19:42 iggy adelcast: none that I've heard of, have you checked the mailing list and github?
19:44 adelcast nothing definitive...I found a reference to opkg and salt here: http://comments.gmane.org/gmane.comp.sysutils.salt.user/14380
19:45 iggy do you actually have salt-minion running on the devices yet?
19:45 adelcast and I know there is not a lot of people using Salt on Embedded, so I might be on uncharted territory
19:45 adelcast yeah, I created an OpenEmbedded recipe for Salt
19:45 rideh Anyone know why salmon (https://github.com/lincolnloop/salmon) died?
19:46 adelcast https://lists.yoctoproject.org/pipermail/meta-virtualization/2015-May/001089.html
19:46 iggy rideh: it says right there on the main page... "My need for simple monitoring quickly became a need for more complex monitoring"
19:46 iggy adelcast: be a trend-setter!
19:47 adelcast hehe, ok....my main concern was to avoid duplication in case there was an ongoing effort already
19:47 pcn Is there a "saltstack for users of chef" page out there?
19:48 pcn I've been doing some internal mappings, but if it's not out there, I may start gisting stuff...
19:48 rideh iggy sorry should ahve been more specific, has a new package superceded it or why didn't the community pick it up?
19:48 adelcast but since opkg is so close to dpkg, shouldn't be too hard to get support added (*I think*)
19:49 iggy rideh: because everybody uses collectd?
19:49 rideh suppose thats what i was looking for,
19:51 hobakill left #salt
19:51 Gareth adelcast: I've been kicking around the idea of adding general support for running Salt on openwrt.
19:52 iggy should be a lot less cumbersome with newer versions with the lazy loader
19:54 adelcast Gareth: ah, interesting....we are not using OpenWRT (we use our own distro), but with OpenWRT support, it would be trivial to add support for our distro
19:56 adelcast Gareth: I will be working on adding opkg support during the next few weeks, that should make it easier to add OpenWRT support
19:56 Gareth Nice.
19:56 numkem joined #salt
19:57 VR-Jack hmm. guess I can try gather_job_timeout, although debug reported a success. :(
20:05 forrest pcn: Nah there aren't any 'saltstack for users of X' as far as I am aware. I thought about writing them, but I haven't used puppet or chef in so long I don't think I could do them justice.
20:09 spookah joined #salt
20:11 soren_ joined #salt
20:12 throwawayj joined #salt
20:13 monkey661 left #salt
20:14 bhosmer__ joined #salt
20:16 throwawayj Any one every had the issue "The procedure entry point ConvertLengthToIpv4Mask could not be located in the dynamic link library IPHLPAPI.DLL" this is on a server 2003 service pack 2. Trying to run a salt-call
20:17 throwawayj Trying to start the salt-minion also causes the service to go directly to a paused state
20:17 Lyfe joined #salt
20:21 fbergroth joined #salt
20:22 spookah joined #salt
20:24 iggy pcn: http://ryandlane.com/blog/2014/08/04/moving-away-from-puppet-saltstack-or-ansible/ might be useful for puppet users
20:26 supersheep joined #salt
20:27 c10b10 joined #salt
20:27 learning joined #salt
20:29 pcn I'm starting this: https://gist.github.com/pcn/2a4be1b36539927ff7e7 but I don't know how much I'm going to add to it
20:29 learning using salt-cloud to spin up minions in aws. I understand salt-cloud calls salt-bootstrap to actually install salt on the new minion. Is there a way to tell salt-cloud called salt-bootstrap what version of salt to install on the minion?
20:29 Berty_ joined #salt
20:30 VR-Jack yeah, adjusting get_iter_returns timeout definitely seemed to help orchestrate not fail; especially with network setup
20:31 thayne joined #salt
20:33 aparsons joined #salt
20:36 looprock joined #salt
20:36 rsimpkins joined #salt
20:37 looprock Hey guys, sorry if I'm doing something stupid, I'm just learning salt and trying to manage a supervisor service. I cannot for the life of me figure out how to handle the 'supervisorctl reread' side if I add a new config
20:38 looprock consul:
20:38 looprock supervisord.running:
20:38 looprock - update: True
20:38 looprock - restart: True
20:38 looprock - require:
20:38 looprock - pkg: supervisor
20:38 looprock - watch:
20:38 looprock - file: /etc/consul.d/agent/config.json
20:38 looprock ----------
20:38 looprock ID: consul
20:38 looprock Function: supervisord.running
20:38 looprock Result: False
20:38 looprock Comment: consul: ERROR (no such process)
20:38 looprock Started: 22:30:44.090379
20:38 looprock Duration: 602.59 ms
20:38 looprock Changes:
20:38 iggy looprock: use gist.github.com
20:38 sdm24 Please use https://gist.github.com f
20:38 VR-Jack https://gist.github.com is your friend
20:38 looprock sry
20:39 looprock https://gist.github.com/looprock/1f025e98750c771a50fd
20:39 iggy at first glance, I'd guess you're missing some consul file in /etc/supervisord or something
20:40 keimlink joined #salt
20:40 looprock it's there, it just hasn't be 'added' via reread
20:41 ildiroen joined #salt
20:41 iggy use states.module to run salt.modules.supervisord.reread
20:42 looprock no, nm, I'm an idiot
20:42 looprock :)
20:42 looprock user perms error
20:42 otter768 joined #salt
20:43 looprock all good, problem between the keyboard and chair
20:44 mpanetta joined #salt
20:46 baweaver joined #salt
20:48 kunersdorf can I pass pillar data to a minion without adding it the pillar top file?
20:49 CeBe1 joined #salt
20:50 sdm24 I don't believe so, and if you can, it's not the best practice
20:50 moloney joined #salt
20:51 VR-Jack even pillar modules simulate the data; just in a more variable way
20:54 kitplummer joined #salt
20:54 VR-Jack *sigh* kickstart, net, pkgs, reboot, highstate. repeat
20:54 baweaver joined #salt
20:55 meteorfox_ joined #salt
20:57 VR-Jack reboot takes awhile. Duration: 162396.099 ms
21:00 linjan joined #salt
21:05 iggy kunersdorf: some modules/states allow it, check the module docs for which ones
21:05 iggy f.ex. sudo salt '*mqdb*' state.sls postgres pillar='{"postgres": {"pg_hba.conf": "salt://postgres/pg_hba.conf", "acls": ["host","mqdb","mqdb","172.16.0.0/16","md5"]}}'
21:09 moloney iggy: I was trying to install a package on minion start to handle my custom grain dependencies like you suggested last week. I got to the point where no errors are logged running the master with "-l debug" but the install doesn't happen.  I setup a reactor state with "cmd.state.pip.installed" and I set the "arg" to "python-hwinfo".  Any ideas on what I am doing wrong or how to debug?
21:14 iggy I don't think you can do it like that
21:15 iggy at least all mine are modules... cmd.<module>.<function>
21:15 iggy you can do cmd.state.sls tgt: {{ data['id'] }} arg: some.slsfile
21:18 murrdoc joined #salt
21:19 spookah joined #salt
21:21 moloney iggy:  then the file "som.slsfile" is located with the rest of my regular state files, not the reactor ones?
21:24 iggy right
21:25 iggy although I keep my reactor files under file_roots too
21:25 WildPikachu joined #salt
21:26 murrdoc me too
21:26 murrdoc namespace under file_roots
21:26 murrdoc its the lazy way
21:26 iggy https://gist.github.com/iggy/b7599a97c18ddd6e6ca4 that's a reactor->orchestrate->state(s) example
21:26 ek6 its not lazy..its easy to find
21:27 thayne joined #salt
21:28 murrdoc ek6actly
21:28 murrdoc <so proud, stinky joke, i know>
21:29 ek6 for you? nah..thats pretty much average
21:29 hal58th_ joined #salt
21:31 Berty_ joined #salt
21:34 mnbbrown joined #salt
21:36 mpanetta_ joined #salt
21:36 IBMan joined #salt
21:37 IBMan does anyone know how to remove a rpm with salt while ignoring depends?  so like remove java-1.5.0-gcj without removing tomcat
21:38 mnbbrown Hi, am running salt-call in a masterless environment and it keeps getting 'Killed' - help would be appreciated. sls and stdout are here: https://gist.github.com/mnbbrown/02ed1167fa523c37ccc2
21:42 VR-Jack mnbbrown: have you tried running the command manually?
21:42 iggy mnbbrown: check dmesg to see if there's any reason why it was killed?
21:45 mnbbrown VR-Jack: I have - and it works fine - I think iggy has pointed me in the right direction
21:46 murrdoc #ssceapproved
21:48 mnbbrown iggy: looks like it was the OOM killer doing its job. salt-master appears to be using crazy amounts of memory. is that normal?
21:50 iggy it can
21:50 iggy depends what you've got going on
21:50 iggy I used to see it with older versions and large gitfs trees (like >1G of objects)
21:51 ek6 doctor doctor it hurts when i do 'this'!
21:51 mnbbrown iggy: I'll remove my gitfs remotes and see if that fixes the problem.
21:51 iggy normal gitfs remotes won't cause problems
21:52 iggy we were just storing a bunch of big blobs in one of ours
21:53 julez joined #salt
21:54 kitplummer joined #salt
21:55 nyx_ joined #salt
21:55 VR-Jack only 440m usage on my master currently
21:55 ek6 iggy: did you find a comfortable zone doing that?  as in....<400m gitfs has no issue >600m always fails?
21:57 mnbbrown iggy: htop looks like this http://imgur.com/dPAuSCg
21:58 mnbbrown VR-Jack: how is your master being utilised?
21:59 baweaver joined #salt
21:59 VR-Jack mnbbrown: I run an orchestrate on it to boot some baremetal
22:00 VR-Jack minimal usage
22:01 iggy ek6: our other trees are all standard formulas and our state/pillar repos are both under 5M
22:02 iggy so... didn't try anything inbetween 5M and 800M (it was originally over 1G, I went back and force pushed to get rid of a bunch of garbage)
22:02 ek6 yeah alright..just didnt know if you had any experience with large but not insane git objects
22:02 mnbbrown VR-Jack: mine is sitting at ~250mb.. and this is less than 30mins after I installed it - and ran salt-call --local state.highstate -l debug
22:02 lempa joined #salt
22:02 ek6 no worries..thx
22:02 iggy it didn't handle 800M well either (so after the cleanup, I still had to go back and remove it from our gitfs)
22:03 iggy it works fine as a normal git.latest
22:03 iggy + file_roots
22:04 iggy mnbbrown: holy... what do you have worker_threads set to?
22:04 VR-Jack if like me, probably running default config
22:05 iggy that's definitely not the default (5)
22:05 VR-Jack 2014.7.5 seems to be memory hungry by default
22:05 rojem joined #salt
22:06 iggy there's 29 entries in that list... so if it's set at the default, somebody has some serious problems
22:07 pcn forrest: Is there an example of what a top file should look like when there is only a single environment?
22:08 forrest pcn: https://github.com/gravyboat/hungryadmin-sls/blob/master/salt/top.sls is what mine looks like with just one box
22:08 murrdoc joined #salt
22:08 forrest pcn: Here's one that just matches everything: https://github.com/gravyboat/demo-app-1/blob/master/salt/roots/salt/top.sls
22:08 forrest and has some roles stuff in it
22:09 pcn So base, in a single environment, will get its states from /srv/salt/<whatever>?
22:09 Topic for #salt is now Welcome to #salt | 2015.5.0 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/
22:10 forrest pcn: Yep depending on how your system is set up. If you are just messing with default stuff, then yep.
22:10 VR-Jack joined #salt
22:11 forrest basepi: Knocking out those later afternoon releases! Is 2015.5.0 berylliumthituhmimnjtn? I can't remember.
22:11 basepi 2015.5.0 is Lithium
22:11 basepi Renamed from 2015.2.0
22:11 basepi Beryllium will (likely) be 2015.8.0
22:11 forrest basepi: Damn I wanted to mess with the masterless reactor!
22:12 basepi =D
22:12 iggy lol, 2015.8
22:12 forrest can you ask Tom what he meant by https://github.com/saltstack/salt/issues/15265#issuecomment-88347584 ?
22:12 forrest basepi: ^
22:12 iggy I know you guys are trying very hard, but I don't see that
22:12 mnbbrown joined #salt
22:12 forrest iggy: Did you look at the roadmap?
22:13 iggy why would I do that
22:13 basepi iggy: hopefully we will surprise you. ;)
22:13 forrest I think it's feasible. Salt has been hiring on a good amount lately as well for QA and other stuff to take quite a bit of load off the engineers.
22:13 moloney iggy:  I am trying to replicate your reactor->orchestrate->state example without much luck.  In the master log I can see the reactor being rendered.  But no errors on the client and no sign of the orchestration state running.
22:13 basepi And we're not trying to get a GUI out this time around...
22:13 mnbbrown iggy: looks like worker_threads was set to the default 5
22:13 mnbbrown I've set it to 1
22:14 iggy mnbbrown: you don't want to do that (unless you ahve a very small number of minions... like <5)
22:14 mnbbrown it has no minions - its masterless
22:14 iggy oh, then why are you running a master at all?
22:15 Georgyo joined #salt
22:15 bhosmer__ joined #salt
22:15 mnbbrown iggy:  was under the impression that you need to run salt-master to salt-call.. I have only been using salt for <24 hrs.
22:15 Georgyo joined #salt
22:16 iggy no, I think you want salt-call --local (or something) for masterless... and then you can less the master
22:16 mnbbrown iggy: I just ran salt-call without salt-master running and it worked!
22:16 spookah joined #salt
22:18 forrest thanks basepi!
22:18 mnbbrown iggy: thanks for you help!
22:19 pcn forrest: so here's what I'm missing - I'd like to but a state in /srv/salt/emacs/install.sls, and refer to that in the top.  That doesn't seem to be working?
22:20 forrest can you gist your top.sls pcn?
22:20 iggy moloney: paste stuff
22:21 pcn forrest: what I have here: https://gist.github.com/pcn/2a4be1b36539927ff7e7
22:22 baweaver joined #salt
22:23 forrest looks okay to me pcn, What do you get as an error?
22:23 scoates joined #salt
22:24 iggy salt-call cp.list_master
22:24 iggy did you change file
22:24 iggy did you change file_roots
22:24 pcn So I run: salt '*' state.highstate, and get failed to compile with "No matching sls found for 'emacs' in env 'base'
22:24 iggy I hate this keyboard
22:25 ndrei joined #salt
22:25 forrest pcn: That seems like it might be a directory issue? Can you do ls -laR /srv/salt/
22:25 pcn Hmm... it's listing base/emacs/install.sls instead of emacs/install.sls from an earlier experiment
22:26 pcn And now it's better?
22:26 iggy file_roots is wrong then
22:26 forrest pcn: Yup seems like a file roots issue, if you change stuff up make sure you restart the service so it picks it up
22:26 iggy you probably set the base file_roots to /srv/salt instead of /srv/salt/base
22:27 iggy which is confusing the docs do either/or in different places
22:27 blast_hardcheese joined #salt
22:27 iggy (always hated that)
22:27 pcn file_roots is default
22:27 pcn commented out in /etc/salt/master
22:27 pickledpig 4/j #mysql
22:27 pickledpig Whoops, sorry.
22:27 pcn I made the directory because of the error message, so that was my problem
22:27 forrest pcn: ahh
22:28 pcn I'm using /srv/salt
22:28 dfinn joined #salt
22:32 pcdummy joined #salt
22:33 thayne joined #salt
22:34 moloney iggy: here is some code: https://gist.github.com/moloney/9451b96468576d206fc3  It does work on the command line if I do "salt-run state.orchestrate orchestration.prerequisites"  so I guess it is something with the reactor setup...
22:35 iggy moloney: what version of salt?
22:36 moloney iggy: 2014.1.13
22:37 iggy ouch
22:37 iggy I had issues with reactor/orch on 2014.1
22:38 moloney do you recommend latest?
22:38 iggy well, 2015.5 is out now...
22:38 iggy that's what I'm using (ish)
22:38 * iggy 2015.2 4lyfe
22:39 Ahlee careful, you'll end up like me.
22:39 iggy 2015.2.iggy
22:39 iggy got a nice ring to it
22:40 murrdoc heh
22:40 murrdoc we are gonna stick with 2014.7.5 for now
22:41 murrdoc 2015 once beacons is useful and sdb is a thing
22:42 Ahlee sdb?
22:43 otter768 joined #salt
22:45 murrdoc salts db, rest interface to a mariadb backed service
22:46 Ahlee wat
22:46 VR-Jack I'll probably try 2015.5 pretty quick and see if it handles resources better
22:46 Ahlee a returner?
22:46 mosen joined #salt
22:47 Ahlee I hate windows so much.
22:48 pickledpig +1 Ahlee
22:48 iggy VR-Jack: it's heavier
22:48 murrdoc yeah its a returner
22:48 murrdoc can be default
22:48 Ahlee I'm ready to jus tnot use salt any more if it gets me away from trying to make it work on windows.
22:48 murrdoc u ever use puppet db
22:48 iggy murrdoc: I'd really like to see some docs on sdb... that's not what I got out of it at all
22:48 murrdoc i read the code
22:49 markm__ joined #salt
22:49 mosen murrdoc: thats some heavy stuff
22:49 murrdoc docs is an apertif mah friend
22:49 primechuck joined #salt
22:50 diegows joined #salt
22:50 Ahlee i thought there already was a mysql returner? mariadb can't be that different?
22:50 baweaver joined #salt
22:51 smcquay joined #salt
22:52 Berty_ joined #salt
22:54 cheus joined #salt
22:55 kitplummer joined #salt
22:56 VR-Jack iggy: dev appears to run faster and more efficient than release. Was hoping some of that made it into 2015.5
22:56 iggy there are sdb modules for couchdb, etcd, memcache, generic rest, sqlite3... no mention of mariadb
22:56 mosen maridb is often 1:1 compatible with MySQL
22:57 mosen you can use python-MySQL against it fine
22:57 pickledpig it should be trivial to make it happen
22:57 bemehow joined #salt
22:57 iggy no mention of mariadb or mysql
22:57 mosen strange
22:58 iggy and I just read 4 of the sdb modules and still have no friendly clue what use it fills
22:59 iggy aside from a magical new sdb:// uri
22:59 murrdoc mariadb implementation is the one that comes with saltstack gui
23:00 c10b10 joined #salt
23:00 iggy like, I see no mention of if this is a minion side thing or a master side thing
23:00 _JZ_ joined #salt
23:00 murrdoc its a minion side thing
23:01 iggy no usage examples in jinja (not sure if that's an oversight or if it means it's not meant to be used in jinja)
23:01 murrdoc and a master side thing
23:01 bastiandg joined #salt
23:01 iggy so it's not per minion like pillars
23:02 iggy I guess I'll just have to wait and see some of the actual use cases in... 6-12 months
23:02 Ahlee optimistic
23:04 funzo joined #salt
23:04 murrdoc basically you know how u do source: salt:/path/to/stuff
23:04 murrdoc u can now do sdb:/path/to/varname
23:04 murrdoc and it will get it from a db
23:04 iggy can do the same with ext_pillar
23:05 Ahlee indeed
23:05 iggy looks like maybe it's just easier to set
23:05 Ahlee if you had come to my talk at saltconf, hyou'd have seen how we did it
23:05 VR-Jack From what I gathered, it sounded like they were going to make it more dynamic, and support pushing return data and stuff into it. but that's just going by what people said in irc
23:05 Ahlee <3
23:11 iggy basepi: like the mention of the cmd.* changes, thanks
23:12 balltongu_ joined #salt
23:13 throwawayj left #salt
23:16 * basepi nods
23:17 aparsons joined #salt
23:17 forrest iggy: link? Wasn't aware of any changes
23:18 iggy it's in the release email and notes
23:19 iggy and it was neutered a lot from it's initial version (which was... "break a shit ton of people's states and a lot of the docs even")
23:19 markm_ joined #salt
23:19 forrest Ahh was that on the mailing list? I had to start filtering
23:19 iggy salt-announce
23:19 iggy I'm not actually on -users
23:20 KyleG joined #salt
23:20 KyleG joined #salt
23:20 murrdoc whats the eta on the first bug fix release /
23:21 murrdoc also uh congratulation on 2015.5 based pi
23:21 keimlink_ joined #salt
23:21 forrest oh I am probably not on salt-announce
23:22 totte joined #salt
23:23 bfoxwell joined #salt
23:24 murrdoc basepi:  do u know this colton meyers guy
23:24 murrdoc i bought his book
23:24 hal58th_ big mistake murrdoc
23:24 murrdoc hahah
23:24 murrdoc damnit man
23:24 basepi No, but I do know Colton Myers. ;)
23:24 hal58th_ just kidding, haven't had a chance to read it yet.
23:24 murrdoc i read it on the flight over
23:24 murrdoc ok read
23:24 hal58th_ Oh yeah, Meyers is how you spell my last name.
23:24 murrdoc definitely good starter
23:25 basepi Glad you liked it! :)
23:25 markm__ joined #salt
23:26 murrdoc so do u know the guy
23:26 * murrdoc wont let a bad joke die
23:27 murrdoc so basepi sdb
23:27 murrdoc can we get a doc
23:27 iggy forrest: where's this mythical roadmap of which you speak?
23:27 forrest iggy: https://github.com/saltstack/salt/milestones/Beryllium
23:27 Gareth murrdoc: you'll get nothing and like it!
23:28 iggy that's not a roadmap
23:28 VR-Jack GH is that way. Taking pull requests. ----->
23:28 basepi Thats not a hard and fast roadmap btw.
23:28 forrest iggy: It's kind of a roadmap...
23:28 forrest basepi: It's something at least! ;)
23:28 basepi We haven't evaluated that issue list yet, it's preliminary.
23:28 basepi True! :)
23:29 iggy a roadmap traditionally has ties together dates with release goals
23:29 iggy that's more of a release plan
23:29 iggy -has
23:29 arif-ali joined #salt
23:31 nyx_ joined #salt
23:36 murrdoc Gareth:  u right, basepi skipped over the question
23:36 murrdoc i ll ask again on the morror
23:36 Gareth murrdoc: https://www.youtube.com/watch?v=0f6l1QljpMo
23:36 VR-Jack Always important to save questions for another day.
23:37 murrdoc Gareth:  steve linns joining
23:37 basepi Sorry, what question?
23:37 iggy 19:27 < murrdoc> so basepi sdb
23:37 murrdoc can we get some documentation and use cases for the sdb stuff
23:37 Gareth murrdoc: who? what? where?
23:37 iggy (would be my gues)
23:37 murrdoc similar to what we got for beacons
23:38 murrdoc Gareth: mah bad, i assumed u knew him, friend of our mutual friend
23:38 Gareth murrdoc: nope.  dont think so. bad with names though :)
23:38 * basepi is not familiar with the sdb stuff, will look tomorrow
23:40 murrdoc thanks based pi
23:40 murrdoc Gareth:  kk
23:40 murrdoc still sucks u live too far
23:40 baweaver joined #salt
23:41 * iggy adds salt-formula contributors guidelines to his TODO list
23:42 iggy if I ask if someone has tested the PR and someone else merges it without an answer -> not-happy-iggy
23:42 julez joined #salt
23:42 iggy if you commit to master -> not-happy-iggy
23:42 forrest iggy: Is the person committing to master a repeat offender from last time?
23:43 murrdoc the arrogance of committing to master
23:43 * murrdoc dont get it
23:43 forrest if so, I'll just remove them from the org that can do that
23:43 forrest problem solved.
23:43 forrest You get one warning, then that's it
23:43 forrest sorry, make a fork
23:43 murrdoc do it forrest
23:43 murrdoc do it
23:43 forrest I will if they are a repeat offender
23:43 forrest if it's someone new, I can't really blame them since we don't have guidelines, and github's permissions are shitty
23:44 forrest iggy: PR link?
23:44 iggy yeah, hence the adding writing guidelines to my list
23:44 iggy nothing new that I've seen in a bit... but that one time thoroughly pissed me off (and I wasn't even a contributor then)
23:45 forrest iggy: okay
23:46 iggy plus you aren't automatically supposed to update copyright years when nothing changes except the year
23:47 dendazen joined #salt
23:55 * iggy no longer slacking on -formulas reviews
23:56 mosen you volunteering for QC iggy? :)
23:58 forrest The iron fist of iggy
23:58 forrest no man shall smile, no man shall PR.
23:58 forrest - Iggy the iron
23:58 VR-Jack and my pillar top is hideous. :)
23:58 primechuck joined #salt
23:59 bhosmer joined #salt
23:59 mosen iggy + your code = rejected, thats the formula
23:59 VR-Jack glad someone else accepted my code. haha

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