Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2016-05-25

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

All times shown according to UTC.

Time Nick Message
00:01 ElasticApps joined #salt
00:12 Eugene LDAP(or other directory) is the general-case way to make sure you have the same UID across systems. I don't think enforcing it in a Salt state is a good idea, unless that's how the users are initially created
00:13 yetAnotherZero users are indeed created in salt Eugene
00:13 Eugene I think you'd need to extend the user state to havea chown=True argument to do what you want
00:13 yetAnotherZero we deploy new servers each week and want to have our current employee base configured on each
00:14 quasiben joined #salt
00:14 Eugene If you don't do any user creation/deletion on the server itself it should be fine. Just make sure you don't duplicate UIDs, and consider starting with a high number like 10000
00:16 yetAnotherZero I suppose i can just write a manual script to chown the current users' files for a one-time uid change we are about to implement
00:16 djgerm if I wanted to install 2015.8.8.2 from salt repo… would that be possible?
00:16 Eugene Yeah, legacy sucks.
00:18 amcorreia joined #salt
00:23 Nahual joined #salt
00:43 justanotheruser joined #salt
00:57 Ahlee oh man. does anybody remember the name of first cut of the web ui from salt?
00:58 dyasny joined #salt
01:08 edrocks joined #salt
01:09 estahn joined #salt
01:09 mosen_ Ahlee: before halite?
01:09 Ahlee halite!
01:09 Ahlee that was it.
01:09 Ahlee thanks mosen_
01:09 mosen_ no probs
01:11 SheetiS joined #salt
01:37 subsignal joined #salt
01:39 djgerm is file.line the best way to remove a single line from a file?
01:39 djgerm if so… how does it work? :)
01:39 djgerm https://docs.saltstack.com/en/latest/ref/states/all/salt.states.file.html is a bit vague
01:46 djgerm1 joined #salt
01:46 quasiben joined #salt
01:47 ilbot3 joined #salt
01:47 Topic for #salt is now Welcome to #salt! | Latest Versions: 2015.5.10, 2015.8.10, 2016.3.0rc3 | Support: https://www.saltstack.com/support/ | Logs: http://irclog.perlgeek.de/salt/ | Paste: https://gist.github.com/ (please don't multiline paste into channel) | See also: #salt-devel, #salt-offtopic | Ask with patience as we are volunteers and may not have immediate answers
01:49 racooper joined #salt
01:53 catpigger joined #salt
02:00 hasues joined #salt
02:01 hasues left #salt
02:25 k_sze[work] joined #salt
02:28 iceyao joined #salt
02:29 ageorgop joined #salt
02:44 yetAnotherZero joined #salt
02:46 west575 joined #salt
02:46 quasiben joined #salt
02:49 tuxx joined #salt
02:55 DEger joined #salt
03:07 writtenoff joined #salt
03:09 brianfeister joined #salt
03:10 yetAnotherZero sorry if this is ignorant, but is there a way to match all files in a directory for a pillar?  ie 'my-epic-minion': - dirname.*
03:23 joshin joined #salt
03:23 joshin joined #salt
03:29 tracphil joined #salt
03:38 west575_ joined #salt
03:52 quasiben joined #salt
04:04 west575 joined #salt
04:05 CeBe1 hi, I am trying to add files from a directory to pillar but from the docs it is not clear to me, where to put the example configuration: https://docs.saltstack.com/en/2015.5/ref/pillar/all/salt.pillar.file_tree.html#module-salt.pillar.file_tree
04:05 CeBe1 if I put this into the top.sls, salt complains with "Error encountered while render pillar top file."
04:06 CeBe1 if I put it into a file in a subdir, it will be added as simple pillar data and not include the files
04:06 nmadhok joined #salt
04:12 thejrose1984 joined #salt
04:13 DEger joined #salt
04:14 thejrose1984 joined #salt
04:17 favadi joined #salt
04:32 zenlot joined #salt
04:33 krymzon joined #salt
04:33 CeBe1 okay, seems to belong into salt-master config...
04:33 CeBe1 problem solved
04:35 aqua^c_ joined #salt
04:41 stooj joined #salt
04:58 brianfeister joined #salt
05:03 ivanjaros joined #salt
05:06 tristianc_ joined #salt
05:11 Ayo joined #salt
05:12 sauvin joined #salt
05:22 brianfeister joined #salt
05:23 cyborg-one joined #salt
05:25 Sylvain31 joined #salt
05:28 brianfeister joined #salt
05:30 jxm_ joined #salt
05:33 keimlink joined #salt
05:36 slav0nic joined #salt
05:43 sjorge joined #salt
05:49 aw110f joined #salt
05:54 DEger joined #salt
05:56 shiriru joined #salt
06:04 lvf joined #salt
06:10 mosen joined #salt
06:18 mTeK joined #salt
06:18 punkoivan joined #salt
06:18 Sylvain31 hi, up… not yet service.running ;)
06:22 DEger joined #salt
06:24 ivanjaros3916 joined #salt
06:31 dmaiocchi joined #salt
06:33 Sylvain31 working with powerDNS state I noticed that: https://github.com/PowerDNS/pdns/issues/3896
06:33 saltstackbot [#3896]title: add a "reload" for systemd | under debian jessie:...
06:35 Sylvain31 may be not a good idea, I was derivating my work from the old (2013) powerdns formula, which I dropped.
06:36 Sylvain31 do you have more on this? reload or restart or the same or something else?
06:39 favadi joined #salt
06:39 Sylvain31 restart works for me: http://paste.debian.net/701902
06:39 scarcry joined #salt
06:42 Rumbles joined #salt
06:42 Elsmorian joined #salt
06:43 fracklen joined #salt
06:49 josuebrunel joined #salt
06:49 haam3r joined #salt
06:51 dmaiocchi joined #salt
06:52 rdas joined #salt
06:57 tristianc_ joined #salt
07:03 fracklen joined #salt
07:04 fracklen joined #salt
07:05 haam3r Hey! Does anybody know what the status of ubuntu 16.04 repos are? When might they be coming?
07:05 deniszh joined #salt
07:07 dgutu joined #salt
07:08 TyrfingMjolnir joined #salt
07:10 Miouge joined #salt
07:10 Sylvain31 haam3r: you mean what version of salt will you get if you: apt-get install salt-master?
07:15 iggy they are working on it
07:16 dkrae joined #salt
07:21 hrumph i have a pr ready for tommorrow i think
07:21 babilen For Ubuntu 16.04 repositories?
07:21 hrumph so you can create volatile windows registry keys
07:21 babilen Ah ..
07:23 k_sze[work] joined #salt
07:25 iceyao joined #salt
07:27 haam3r No, I mean when will a repo be made available in repo.saltstack.com for Ubuntu 16.04LTS
07:27 lero joined #salt
07:28 KermitTheFragger joined #salt
07:30 babilen https://github.com/saltstack/salt/issues/32761
07:30 saltstackbot [#32761]title: Ubuntu 16.04 Xenial Xerus Support | Ubuntu 16.04 is out. http://releases.ubuntu.com/16.04/...
07:30 babilen https://github.com/saltstack/salt-bootstrap/pull/852
07:30 saltstackbot [#852]title: Add support for forthcoming Xenial repo binaries. | ### What does this PR do?...
07:30 babilen (and so on .. read the bugs!)
07:31 babilen haam3r: And official Ubuntu/Debian packages are available in xenial.
07:31 babilen It comes with 2015.8.8
07:33 dkrae joined #salt
07:36 krymzon joined #salt
07:39 slav0nic joined #salt
07:40 viq joined #salt
07:40 kawa2014 joined #salt
07:44 manji joined #salt
07:45 Sylvain31 in http://paste.debian.net/701960, I would like to log else where. So I'm configuring rsyslog for that. how do I tell to restart rsyslog, the file is passed, but no trace that rsyslog has restarted. I don't manage rsyslog, I just drop a file for him. because powerdns is logging via syslog…
07:45 VSpike joined #salt
07:46 Sylvain31 line 41
07:46 ravenx joined #salt
07:46 VSpike I seem to be having fun with the issue that never goes away - adding Ubuntu PPA's with non-ASCII characters in the name, particularly Sury's PHP ones
07:47 ravenx i'm using grains os to get the saltversion, and i see something like 2015.8.10
07:47 ravenx i want to se that to my advatange in my jinja state to compare
07:47 ravenx "do this, if the version is less than that"
07:47 ravenx but this isnt' exactly a floating point number so i can't use > or < operators
07:47 ravenx how will i go about doing something like "if smaller than this version"
07:48 babilen ravenx: Split it into major, minor and point and compare each individually
07:48 ravenx thanks babilen
07:48 iggy anybody think of a way to set env variables for salt.client.LocalClient() ?
07:48 VSpike ravenx: there's a saltversioninfo grain which is already split
07:48 ravenx split within the jinja, or is there grains command where i can just grab the minor?
07:48 babilen VSpike: Hell yeah!
07:48 ravenx VSpike: oh god i love you
07:48 AndreasLutro in my experience, python compares version strings just fine, which is easier to read and maintain
07:49 VSpike salt-call grains.get saltversioninfo --out=raw -> {'local': [2015, 8, 10, 0]}
07:49 ravenx that returns a list...o_O
07:49 AndreasLutro actually I take it back
07:49 ravenx but within jinja, how do i extract only the 8 and the 10?
07:49 AndreasLutro just disproved myself
07:49 Mowens im so glad you sleep at the same time i do. Never. :)
07:50 iggy ravenx: saltversioninfo[0] saltversioninfo[1] etc
07:50 VSpike ravenx: grains['saltversioninfo'][1:2] I guess
07:50 rem5 joined #salt
07:50 babilen ravenx: "major, minor, point = grains['osversioninfo'][:3]"
07:52 babilen Adjust as necessary (and you might want to do some checking before running into a ValueError)
07:53 ravenx thanks
07:53 ravenx wait, this is on python though
07:53 ravenx not so much for jinja, no?
07:53 babilen Works in jinja
07:53 babilen {% set .... %}
07:54 viq joined #salt
07:54 manji joined #salt
07:56 VSpike https://bpaste.net/show/e288e7c157d7
07:56 hoonetorg joined #salt
07:57 VSpike This is using 2015.8.10 and the latest PHP formula
07:57 iko_ joined #salt
07:57 iko_ hi all
07:57 VSpike That workaround they are using is well known and usually works AFAIK.
07:57 babilen VSpike: Workaround for what?
07:57 VSpike I checked, and I do have the language-pack-en-base package installed
07:58 babilen ENOTENOUGHCOFFEE
07:58 VSpike babilen: there's an issue with PPAs with non-ASCII names. I think it's possibly both a Salt issue and an Ubuntu issue
07:58 evidence joined #salt
07:58 babilen This is completely unrelated to what we have been discussed before, isn't it?
07:59 babilen Ah, found your earlier comment in the backlog.
07:59 Electron^- joined #salt
08:00 VSpike babilen: yes, sorry ... relating to an problem I mentioned before the version numbers https://github.com/oerdnj/deb.sury.org/issues/56
08:00 babilen Why am I not surprised that apt-add-repository is broken?
08:00 * babilen never understood why you need that tool
08:00 Rumbles joined #salt
08:00 VSpike babilen: there are a few duplicates but https://github.com/saltstack/salt/issues/23543
08:00 saltstackbot [#23543]title: Adding PPA with pkgrepo.managed fails silently if owner name contains non-ASCII | Using 2014.7.5 with this state code:-...
08:01 iko_ hi all..
08:01 babilen That's a bug in softwareproperties/ppa.py though, isn't it?
08:01 iko_ learning salt here... LD
08:01 VSpike That supposedly was fixed, although IIRC when I tried removing the similar workaround I had in my own states (before recently switching to the formula) it still broke
08:01 iko_ this is my top.sls pillar file
08:01 iko_ http://pastebin.com/ENEuuv7B
08:02 babilen VSpike: Just "echo 'deb http://foo-ppa....' > /etc/apt/sources.list.d/foo-ppa.list" ;)
08:02 iko_ i get this error when executing
08:02 iko_ http://pastebin.com/TTKwWJwj
08:02 VSpike babilen: Actually, it's creating that file. What it's not doing is adding the key
08:03 babilen Ah, Mr. Ondřej Surý
08:04 VSpike This workaround was what I was using in my states https://github.com/saltstack/salt/issues/23543#issuecomment-102448757
08:04 saltstackbot [#23543]title: Adding PPA with pkgrepo.managed fails silently if owner name contains non-ASCII | Using 2014.7.5 with this state code:-...
08:04 babilen VSpike: Does it work if you configure it as a normal repository and do not rely on that apt-add-repository code? (i.e. just give the complete URL and "key_url" in a pkgrepo.managed state)
08:04 Sylvain31 how do I restart rsyslog from a state when I add a file in /etc/rsyslog.d ?
08:05 babilen VSpike: It seems to be the apt-add-repository code that is broken rather than salt or python-apt.
08:05 VSpike babilen: don't you love it when you replace your own working code with the easy to manage, drop in, standardised non-functioning formula :)
08:05 tharkun joined #salt
08:06 babilen Well, this has little to do with salt, but with the (unnecessary IMHO) apt-add-repository codebase .. This would fail if you run it manually also, wouldn't it?
08:06 iko_ does not make sence, yaml seems to be fine
08:09 VSpike babilen: no, it works if I run it manually
08:09 VSpike It probably depends on the executing process's locale settings
08:09 babilen https://www.refheap.com/119575 ← repasted iko_'s pastebin.com
08:10 babilen VSpike: That might be, still broken then. How does salt behave if you manage that repo like "normal" (i.e. non-ppa) repositories?
08:10 mage_ with some pillar data is there an easy way to put that in a state file (yaml) and maintain indentation ?
08:11 s_kunk joined #salt
08:11 Sylvain31 iko_: how do you run the state?
08:11 keimlink joined #salt
08:12 babilen iko_: That looks fine .. Is that your entire top.sls ? Could you remove the bit after line 3 and check if it renders then?
08:12 iko_ Sylvain31: salt '*' state.sls users_and_ssh
08:12 VSpike iko_: almost looks like you have an odd character in there
08:12 mage_ for example in https://gist.github.com/silenius/99f4efd2d4766ca05f37cecc99f30759 at line 19 I'dl like the complete webapps:items:ckan_bbpf:ini section
08:13 iko_ babilen: yes that is all
08:13 VSpike iko_: can you do hd top.sls | curl -F 'sprunge=<-' http://sprunge.us
08:13 babilen iko_: Could you run "cat -v /path/to/your/pillar/top.sls" and paste the result to one of http://refheap.com, http://paste.debian.net, https://gist.github.com, http://sprunge.us, … ?
08:14 * babilen high-fives VSpike
08:14 VSpike heh :)
08:14 babilen Same idea, different approach :)
08:14 edrocks joined #salt
08:15 iko_ babilen: VSpike http://sprunge.us/TPJA
08:15 VSpike Sylvain31: There's an example at the bottom of this section https://docs.saltstack.com/en/latest/ref/states/requisites.html#watch
08:16 VSpike you can either use "watch" or "watch_in", depending on which way you want to point the dependency
08:17 VSpike iko_: Hm, looks ok for the characters I think
08:17 babilen iko_: And "cat -v .." ? I find that considerably easier to read :)
08:17 babilen But it looks fine so far
08:19 iko_ here is pillar and state
08:19 iko_ https://www.refheap.com/119576
08:20 mage_ is there some YAML jinja filter to format a structure in YAML ?
08:20 mage_ tried {{ webapp.ini|yaml_encode }} but doens't seems to work
08:21 babilen iko_: That looks perfectly fine? I take it that "cat -v /path/to/your/pillar/top.sls" doesn't show anything untowards? What happens if you remove line 5-7 from your pillar top.sls and render again?
08:21 babilen mage_: That's it. Why doesn't it work?
08:22 dgutu joined #salt
08:23 mage_ babilen: https://gist.github.com/silenius/51d61aa9ab07f8b9847213ad32ed2564
08:23 iko_ wtf...... # cat top.sls base:   '*':     - core
08:23 iko_ and i still get the same error
08:23 iko_ i removed second env
08:23 iko_ ...
08:23 iko_ is there some cacheing or smth
08:24 iko_ does nor make sence
08:25 dgutu joined #salt
08:26 iko_ restarted salt-master service... still the same
08:26 krymzon joined #salt
08:26 babilen mage_: Ah, yeah. yaml_encode is for scalars such as nts, floats, timestamps, ... You want |yaml
08:27 N-Mi joined #salt
08:27 N-Mi joined #salt
08:27 babilen iko_: What does "cat -v /path/to/your/pillar/top.sls" give you? Could you paste the content of "cat -v /path/to/your/pillar/core.sls" ?
08:27 mage_ babilen: then I get https://gist.github.com/silenius/4ee0d8192695c757127f9623e531603e  .. :)
08:28 babilen And how are you refreshing your pillar? Are you using GitFS or something like that?
08:29 iko_ https://www.refheap.com/119577
08:29 babilen mage_: What is webapp.ini?
08:29 mage_ a dict, it's returned by a custom function
08:29 babilen iko_: That looks perfectly fine. How are you refreshing your pillar?
08:30 iko_ saltuitil.refresh_pillar
08:30 babilen mage_: Does it work if you take it out of webapp?
08:30 babilen iko_: And these are local files?
08:30 iko_ yep
08:30 iko_ going by the basepi's book
08:30 babilen iko_: Okay, remove all of that .. refresh your pillar. Then create a "foo.sls" and target it in a new top.sls
08:30 iko_ should be pretty somple :(
08:31 mage_ babilen: yep, it's something like: https://gist.github.com/silenius/df2f7f52148f42908f3d0b00f13a3211
08:31 babilen Restart the master for good measure. I can't spot anything wrong with what you've pasted
08:31 iko_ hm
08:32 iko_ now it works
08:32 GreatSnoopy joined #salt
08:32 iko_ after refreshing pillar again
08:32 babilen mage_: You have many dots in there .. maybe they are interpreted as property access?
08:32 babilen (and : also)
08:32 babilen iko_: Well, at least .. I can't tell you what was wrong though. It looks perfect.
08:33 mage_ maybe the app:main:
08:33 babilen mage_: Maybe simplify that data to a simple "ini: foo: bar" and then add in things step by step?
08:33 mage_ yep that's another solution
08:34 babilen Yeah, the "app:main" and "ckan.bbpf_theme.publications_mendeley_group.name" both incorporate some form of nested collection access/property access that might break some, well, suboptimal code that doesn't simply treat that as strings
08:34 Sylvain31 how do I restart rsyslog from a state when I add a file in /etc/rsyslog.d ?
08:36 rrei joined #salt
08:37 Sylvain31 can I add salt.states.cmd.run and linkit some what to the file.mananged? I missed something, I can't get how it should be…
08:39 babilen Sylvain31: Just think about the relationship between states and requisites
08:39 babilen eom
08:39 jacksontj joined #salt
08:41 Sylvain31 hum, that way it worked… http://paste.debian.net/702134 line 42
08:44 Sylvain31 but I'm pretty sure it could be added in the file.managed line 34
08:45 babilen It can
08:54 barmaley joined #salt
08:55 guest2007 joined #salt
08:55 guest2007 hi
08:56 Elsmorian joined #salt
08:57 guest2007 does salt work with active directory
08:57 guest2007 ?
09:02 estahn hi guys ... it appears i cant use compound in a pillar top.sls file, is that correct/
09:03 Rumbles joined #salt
09:03 estahn never mind https://github.com/saltstack/salt/issues/26727
09:03 saltstackbot [#26727]title: KeyError: 'pillar' when using I@ compound match in pillar top.sls | With the following pillar files:...
09:07 babilen estahn: Pillars can't match themselves
09:08 JohnnyRun joined #salt
09:08 estahn there is an option for the master to evaluate external pillar first. i guess then you would be able to use those
09:09 guest2007 can salt replace sccm ?
09:09 estahn anyways, they didnt dismiss the bug report so it seems to be a valid problem @babilen
09:10 babilen estahn: Well, the underlying problem is that standard pillars can't be self-referential (i.e. they are not being recursively evaluated) and you therefore cannot refer to pillar data when targeting new pillar data. In the end: What would you want to happen if the data you are targeting with I@ allows the targeting to succeed in the first place?
09:11 JohnnyRun Is there someone of the saltstack docs here? windows installer link are broken here: https://docs.saltstack.com/en/latest/topics/installation/windows.html good one are here: http://repo.saltstack.com/windows/
09:11 estahn @babilen my current problem is that pillar_roots and environment mapping is not working correctly
09:12 estahn i have the usual env config https://www.dropbox.com/s/z1t01x77hv2cjlb/Screenshot%202016-05-25%2019.11.57.png?dl=0
09:12 mosen hiya saltines
09:12 babilen estahn: We therefore have to introduce some concept of pillar hierarchy in order for this to work. That way "earlier" pillars can be referenced when targeting/referencing "later" pillars. Look into external pillar and, in particular, pillar stack for that. https://github.com/saltstack/salt/issues/23910 should also be noted.
09:12 saltstackbot [#23910]title: Please implement static pillars | Hi,...
09:12 guest2007 -_- I'm ignored !
09:12 estahn @babilen yep, saw that a while ago... i agree that pillars should stay simple
09:12 babilen estahn: It's typically preferred if you could just paste the text to a pastebin such as http://refheap.com, http://paste.debian.net, https://gist.github.com, http://sprunge.us, …  :)
09:13 babilen guest2007: I guess that you would have to compare the features of both and factor in your needs
09:13 estahn @babilen https://gist.github.com/estahn/1734e7b93694dce3f73586204ab5c7b7
09:14 cyborg-one joined #salt
09:14 estahn in this case the production pillar is never read
09:14 estahn because staging is defined before
09:15 estahn i'm going to do some dynamic saltmaster configuration voodoo now
09:21 toanju joined #salt
09:24 estahn @babilen any ideas how to solve environment pillars differently?
09:24 babilen estahn: You include the same set of minions in both staging and production by targeting to "roles:saltmaster", aren't you?
09:25 babilen I stopped using environments in salt as they are simply annoying and just have different salt masters for each
09:25 estahn i see
09:25 babilen In particular the "git branch == environment" bit was horrible
09:25 estahn doesnt help me though :)
09:25 babilen No, it doesn't .. sorry for that
09:26 estahn yeah i have everything in master, just in different subdirs
09:26 babilen Maybe if you are to target different minions for staging and production?
09:27 estahn compound doesnt work, see above bug request
09:30 babilen estahn: That's for I@ isn't it?
09:34 Kurisutian joined #salt
09:37 estahn yes, G@ works... i will take that as workaround
09:38 estahn a shame really, because i already have the info in ec2 tags pillar
09:38 estahn and its working well in state top.sls 'I@ec2_tags:Environment:staging and G@roles:saltmaster'
09:38 VSpike Hm. The nginx formula helpfully sorts the config items from the pillar in order to build the file. The problem is, the order in the file is significant
09:39 estahn @VSpike why is the order significant
09:40 VSpike estahn: well, I think that's the problem anyway. https://bpaste.net/show/43a7e6c479a2 is the generated config, and the error is nginx: [emerg] unknown log format "with_site" in /etc/nginx/nginx.conf:15
09:41 VSpike Yep, move the access_log line after the log_format ones and it's fine
09:41 estahn if you move the access_log manually underneath it works?
09:41 VSpike Yeah
09:41 estahn ah ok
09:42 VSpike I may have to move that to an include file
09:42 estahn you could move the access_log config into conf.d/...
09:42 estahn as workaround
09:42 VSpike Yeah, I was thinking the same
09:42 estahn but i agree the reordering is annoying ... its the same with php-fpm formula
09:43 estahn but there its just asthaetics
09:43 VSpike This is one of the things about salt. When you see newcomers doing stuff in their own code that a formula exists for, it's tempting to say "use a formula instead"
09:44 VSpike but that way you don't learn much, and also using the formulas is really, really hard sometimes compared to just doing it yourself
09:44 VSpike A lot depends on the quality of the formula
09:44 estahn i think chef in that case is miles ahead
09:44 estahn or lightyears to be specific
09:44 estahn :D
09:45 estahn e.g. supermarket, automatic cookbook testing etc
09:45 VSpike I've still not had a chance to try chef properly. It would probably do my Windows setup a lot better since it uses DSC
09:46 estahn well chef has its disadvantages too
09:46 estahn but the supermarket is just awesome https://supermarket.chef.io/
09:46 VSpike My powershell scripts to set up IIS from Salt are hair raising :) And they repeat stuff every time because anything else would be insanely hard
09:46 aw110f joined #salt
09:46 estahn you have version, dependency control etc etc
09:48 VSpike I did start reading about DSC to see if I could mod my powershell scripts to use it, but I don't really like Windows or Powershell or IIS and it made me feel a bit queasy
09:50 VSpike Comment here from the Head of IT the other day: "We [i.e. him, the CEO & CFO] can't really see the value of this Salt stuff. It seems to create a lot of work for something we'll only use once. We were thinking perhaps we should go back to documents with step-by-step instructions instead for some cases"
09:50 zer0def a quick question - when evaluating jinja templates, does salt (or jinja) evaluate them recursively or just once?
10:01 Kurisutian Hi guys! I currently use salt-ssh for agentless operation and was wondering if there is any way to use privilege escalation with it similar to Ansible. Unfortunately for different doings I need to log in with one account and switch to another priviliged user locally. Is there any way to do something like that?
10:02 jhauser joined #salt
10:03 shiriru joined #salt
10:07 linjan__ joined #salt
10:13 evidence joined #salt
10:14 Miouge_ joined #salt
10:22 honestly It supports using sudo
10:22 honestly Not sure if it supports sudo to a user other than root because I haven't needed that
10:38 Kurisutian The sudo I found, but not how to specify the user to sudo to if possible....
10:41 _JZ_ joined #salt
10:43 evle joined #salt
10:46 dunz0r Is there something like a "recommended" way to handle IPtable-rules with Salt?
10:49 Rumbles salt state iptables?
10:49 aw110f joined #salt
10:49 Rumbles https://docs.saltstack.com/en/latest/ref/states/all/salt.states.iptables.html
10:49 Rumbles or google search "salt stack iptables"
10:50 AndreasLutro I wouldn't use the iptables states, it's sorely missing a stateful iptables.managed function
10:50 JohnnyRun joined #salt
10:50 AndreasLutro as it is you either have to flush all rules every time you make a change or there will be duplicate rules
10:54 amcorreia joined #salt
11:00 Qlawy Guys, anyone can exmplain me why salt-call works and when I try to do salt 'minion' state it fails with: state is not available?
11:00 Qlawy in debug of minion I found: Error loading module.ssh_service: The ssh_service execution module failed to load: only works on an ssh_sample proxy minion.
11:02 jenastar joined #salt
11:03 dunz0r AndreasLutro: Yeah, that's what I felt to.
11:04 dunz0r I /could/ just tell it to put stuff in the 'correct' place I guess.
11:05 fracklen joined #salt
11:11 debian112 joined #salt
11:11 Ayo joined #salt
11:15 fracklen joined #salt
11:16 quasiben joined #salt
11:16 fracklen joined #salt
11:18 netcho what is the best way to install list of core packages
11:19 netcho put them all in one i.e. core.sls state> and then install them using jinja rendering?
11:20 cyborg-one joined #salt
11:21 netcho is it possible to select them from state like it can be done from pillar?
11:23 Biopandemic joined #salt
11:25 scoates joined #salt
11:26 rdas joined #salt
11:28 netcho i guess not :D
11:30 Biopandemic joined #salt
11:39 eseyman joined #salt
11:45 mowntan joined #salt
11:45 mowntan joined #salt
11:45 mowntan joined #salt
11:45 babilen netcho: "select them from state like it can be done from pillar" what does that mean?
11:45 mowntan joined #salt
11:45 netcho babilen: nvm
11:45 netcho :)
11:51 auwooo-0x71 joined #salt
11:52 babilen You can pass a list from pillars as "pkgs:" argument to pkg.installed .. but then .. you seem to be happy
11:53 ivanjaros joined #salt
11:53 Biopandemic joined #salt
11:59 west575 joined #salt
12:11 Sylvain31 babilen: pkgs: doesn't provide expansion as names: does, so I discorvered that pkg.require: doesn't work on pkgs but do on names…
12:12 Sylvain31 I found manual note, spreads arround about this…
12:14 babilen Sure, requirements reference the state ID or name argument
12:15 zenlot6 joined #salt
12:17 edrocks joined #salt
12:17 Sylvain31 Is there a deep requirement doc I miss a lot of knowlegde on this, I was hopping everything was quite intiutive as python tend to emphasis himself, but for salt, it is not obvious most of the time…
12:18 babilen "this" ?
12:18 babilen There's https://docs.saltstack.com/en/latest/ref/states/requisites.html which you have surely read already
12:21 oeuftete joined #salt
12:23 Sylvain31 I met this one already, yes. But I will take an hour to reread carefully. It didn't helped me with file.managed: + restart a service I wanted…
12:23 babilen Why not?
12:23 babilen What would you change about the documentation now that you have solved that issue
12:24 babilen Was it simply not obvius that you cannot define requisites on things for which you do *not* have a state? (e.g. like restarting rsyslog when you don't manage the service?)
12:24 rgrinberg joined #salt
12:26 Sylvain31 I don't know. I should have skipped some fondamental step about salt, hopping to step into in by deducing be seeing example. Mainly, for salt it is not the case. You cant skip step an hopping to understand from your pass knowlegde… Quite strange it worked well with all previous things I've learnt…
12:26 TooLmaN joined #salt
12:26 ravenx if i have a local instal of bower, not system, can i still use the bower.bootstrap tool?
12:26 ravenx same with npm
12:26 ravenx and if so, how?
12:29 babilen Why does it make a difference if you have a local install or not?
12:29 ravenx cuz the bower binary is else where
12:29 babilen ... and ... ?
12:29 ravenx which bower keeps showing the one in /usr/bin
12:30 DEger joined #salt
12:30 babilen So, adjust your PATH to match your expectations?
12:31 linjan__ joined #salt
12:31 ravenx so i can just remove /usr/bin right?
12:31 ravenx but if i do that, won't the other thing sthat depend on it, break?
12:31 babilen Which other thing?
12:32 ravenx other things in my /usr/bin
12:32 babilen But sure, *iff* something depends on /usr/bin/bower being present then you shouldn't remove that file
12:32 ravenx ah
12:32 ravenx what i meant was removing /usr/bin from my $PATH
12:32 ravenx cuz i would need to adjust my path
12:33 babilen How is this related to salt btw? I have the feeling as if I'm missing something :)
12:33 ravenx because bower state is callign the wrong bower
12:33 ravenx i have a bower in /home/user/.nvm/v0.10.32/bin/bower
12:33 ravenx and i woul dlike to use that one
12:34 babilen Ah, so *that* is the issue :)
12:34 babilen The module essentially calls "which bower" and uses whatever that returns
12:35 babilen Not sure if you can override that .. is the system one broken or outdated?
12:35 Sylvain31 babilen: I think the main confusing thing about salt, is that state "look-like" command to be performed, but they aren't. As you said, it is not imperative. But on the other side, it is. I expect that the task I program will be effectively executed.
12:36 ravenx babilen: the system one is too new lol
12:36 ravenx babilen: yeah so that's my problem.  i was thinign about remove /usr/bin from my $PATH
12:36 ravenx until i realized that i may get fired for doing soemthing like that.
12:36 babilen Sylvain31: What do you mean by "state "look-like" command to be performed" and why are "programs" not effectively executed?
12:37 babilen ravenx: Well, one very hacky solution would be a symlink from /usr/bin/bower to the one you want to use, but that is sooo fugly that I feel ashamed to even mention it ;)
12:37 Sylvain31 babilen:  in state for example "require: - pkg: vim" look-like : check if package vim is installed. Which is not what it is…
12:38 babilen ravenx: Or another approach: Fix whatever problems arise from /usr/bin/bower being too new?
12:38 Sylvain31 babilen: it is: require that the other state named vim is executed.
12:38 ravenx babilen: hahaha, thanks.  i do like the symlink one.  at the moment, i'm using cmd.run to solve this.
12:38 ravenx babilen: i will see if i can downgrade that package we have
12:38 ravenx babilen: thanks :)
12:39 Biopandemic joined #salt
12:40 babilen Sylvain31: Yeah, requisites are between states. That's something you have to know or infer it from various example as you will notice that the leading bit is *always* the name of the corresponding state module.
12:40 Sylvain31 babilen: and it is written (for me badly) "In the example above, the file /etc/vimrc depends on the vim package."
12:41 babilen You assumed that "pkg" has a meaning outside the context of the salt state system which is simply a misunderstanding
12:41 Sylvain31 babilen: which not the real!
12:41 babilen Yes, that is misleading
12:41 babilen It should have referred to the *state* rather than the *package*
12:41 Sylvain31 babilen: of course…
12:44 Sylvain31 babilen: in state, "file.managed: - soucre: filename" looks like function call file.managed(source=filename)… it confuse me. But I may have missed fondamental tutorial. That may be one reason…
12:44 babilen That's exactly what it is
12:45 babilen https://github.com/saltstack/salt/blob/develop/salt/states/file.py#L1104
12:46 Sylvain31 Yes I do that too, read the code when the doc seems obscure… ;)
12:48 babilen Well, you wanted to clarify something about salt internals and the way states are being called. I see no other way than looking into the code for that. It doesn't really matter in the way you use states though.
12:48 gh34 joined #salt
12:49 numkem joined #salt
12:50 aw110f joined #salt
12:53 tracphil joined #salt
12:54 rbjorklin joined #salt
12:54 ssplatt joined #salt
12:54 rbjorklin What problems remain with the 2016.3 version? Anything I can help test to get it released?
12:55 Electron^- hi guys... in an sls, can I initialize a variable like a folder name with an if statement to use it in a state?
12:58 subsignal joined #salt
12:58 iceyao joined #salt
12:59 DammitJim joined #salt
12:59 amcorreia joined #salt
13:01 Ayo joined #salt
13:01 toastedpenguin joined #salt
13:02 Rumbles Electron^-, I think you can do that with Jinja and set
13:02 Rumbles stackoverflow.com/questions/3727045/set-variable-in-jinja
13:02 Rumbles jinja can be used in state files
13:02 Rumbles more jinja info: http://jinja.pocoo.org/docs/dev/templates/#comparisons
13:03 subsigna_ joined #salt
13:03 Electron^- yep, I solved using "{% set var_name = 'val' %}"
13:05 Rumbles cool :)
13:09 komputes joined #salt
13:10 Electron^- thanks anyway :)
13:11 komputes joined #salt
13:11 Rumbles I'm just happy I got the answer right :)
13:11 Electron^- :)
13:12 TOoSmOotH joined #salt
13:15 mage_ if in my foo.sls I have a {% set bar = ... %} and that I'm using include: - anothersls how can I use {{ bar }} in anothersls ?
13:16 mage_ should I use some macro?
13:17 babilen http://jinja.pocoo.org/docs/dev/templates/#import-context-behavior
13:19 edrocks joined #salt
13:19 mage_ yep.. but {% include "anothersls" %} and include: - anothersls don't have the same behavior
13:22 DEger_ joined #salt
13:24 ravenx because of this issue with pip==8.1.2, packages with "." in them breaks. https://github.com/pypa/pip/issues/3666
13:24 saltstackbot [#3666]title: Package names containing a dot don't work under DevPI | * Pip version: 8.1.2...
13:24 ashmckenzie joined #salt
13:25 ravenx so i'm using virtualenv.managed, and it installs the latest pip,a nd as a result, a package which has a dot, fails my entire salt state.
13:25 ravenx anyone know of a fix, or a way to pass in a pip version for virtualenv managed?
13:27 mpanetta joined #salt
13:28 barmaley joined #salt
13:32 DEger joined #salt
13:34 mage_ babilen: you'll get conflicting ids: when two files are {% include %} but not with include: I'd like to avoid that (:
13:35 babilen from foo import bar
13:35 babilen But I don't tend to do this .. a single foo_settings where everything is to be found is good enough for me :)
13:35 racooper joined #salt
13:37 perfectsine joined #salt
13:38 mage_ mmh
13:40 debian112 joined #salt
13:42 mpanetta_ joined #salt
13:45 mpanetta joined #salt
13:47 fracklen joined #salt
13:48 DEger joined #salt
13:49 zoski joined #salt
13:51 mapu joined #salt
13:56 jerredbell joined #salt
13:56 ronp_usa joined #salt
13:56 Craig_ joined #salt
13:57 yetAnotherZero joined #salt
13:58 debian112 joined #salt
13:58 Craig_ Hi, wondering if anyone can help me build an SLS file using grains to support multiple OS types. Having trouble with the syntax. Thanks
13:58 Rumbles pastebin what you have and I'm sure soeone will give their input
13:58 west575 joined #salt
13:59 tapoxi joined #salt
13:59 edrocks joined #salt
14:01 ravenx start with:  {% if grains['os'] == yada yada %}
14:01 ravenx then one hyphen, then whatever you want
14:01 ravenx then elif
14:01 ravenx then endif
14:01 ravenx :P
14:01 tapoxi joined #salt
14:03 Craig_ ah, thanks ravenx - that is different to how I used it in the top.sls file, so probably explains why it is failing elsewhere
14:03 hasues joined #salt
14:04 ravenx let me show you a snippet of my syntax
14:04 ravenx instead of checking for OS i'm checking for sdaltversions
14:04 ravenx but the concept is the same.
14:06 ravenx Craig_: https://paste.debian.net/702777/
14:06 Craig_ this is what I have that is broken: http://pastebin.com/EHJaUcHR - it even looks wrong!
14:06 ravenx i set a variable on line 7.  but you dont have to.  you could, or you could not
14:06 ravenx oh..hm....that SHOULD work
14:06 ravenx i remember seeing that in the docs
14:08 VSpike Craig_: yeah, you can only use that syntax in the top.sls file
14:08 VSpike You'll need to use Jinja conditions like ravenx suggested
14:08 Craig_ when run you get 'state is not formed as a list'
14:09 Craig_ right, thanks VSpike. I'll go and have an experiment. What's the reasoning behind the different methods in the top.sls and other files?
14:10 ravenx they want us to learn jinja (or at least that's how it feels like to me)
14:10 VSpike Or alternatively a jinja map file or grains filter by ... see https://docs.saltstack.com/en/latest/topics/best_practices.html
14:11 VSpike Craig_: not sure.. it confused me at first too, calling both "SLS" files when they have quite different structures
14:12 zmalone joined #salt
14:13 VSpike when you call salt-call, does the minion process execute it or does salt-call execute directly?
14:13 hasues left #salt
14:14 Craig_ sorry vSpike - still learning the ropes here - how do you mean?
14:16 Tanta joined #salt
14:16 VSpike Sorry Craig_ , last question was directed to the channel generally :)
14:21 morissette joined #salt
14:22 remyd1 joined #salt
14:23 ssplatt VSpike: you can run salt-call without the minion running
14:24 ssplatt i.e. masterless, the minion daemon is not supposed to be running
14:24 remyd1 Hi. I would like to use service.running to check if service is running in a salt state. Although I thought my states works as expected, it is not. When I copy (for example) the binary from a service and check it back again, salt tells me this daemon is running and is in desired state, while it is not. Any idea ?
14:24 ssplatt and you’d use salt-call [—local] to configure things
14:24 Craig_ VSpike, ravenx - Got it working, thanks. Here is my working config: http://pastebin.com/Bea85sjz
14:24 DEger joined #salt
14:24 remyd1 s/copy/moved/
14:25 thekabal joined #salt
14:25 hasues joined #salt
14:26 hasues left #salt
14:26 hellosalt joined #salt
14:29 remyd1 However, the module service.status works fine
14:29 thekabal How can I query to see the contents of the Salt Mine for a given minion? Am I correct in thinking that mine simply puts a value into either pillar or grains, and I just query those?
14:30 komputes joined #salt
14:30 jnials I'm trying to get an external pillar on git working.  I have 2 paths I need to import from the same repo.  It's only ever picking up the path "base".  the config file is here: https://gist.github.com/jnials/3557a508d38e2d6a30f6ff57372793de
14:31 jnials basically, I'm trying to do in pillar_git what I did with the pillar_roots that is commented out.
14:34 tmclaugh[work] joined #salt
14:36 Sylvain31 can auto magically add an header: "DON'T EDIT THIS FILE: salt managed file" in a file.managed?
14:36 ronp_usa1 joined #salt
14:36 Tanta Sylvain: define a global pillar value for "salt-banner" and tag the files with {{ salt-banner }} at the top
14:37 hellosalt Does anyone has a recommended workflow for multiple people working on the same states repository?
14:37 oida joined #salt
14:37 Sylvain31 hellosalt: talking, with voice? ;)
14:38 Sylvain31 Tanta: it has to be a Jinja so?
14:38 ssplatt hellosalt: are you lookign for alternatives to git?
14:38 hellosalt @Sylvain31, all engineers are operating on master?
14:38 drawsmcgraw hellosalt: I treat it like a software project. Branch, occasional rebase on master, review, then merge branch into master.
14:39 Sylvain31 hellosalt: git copy on their workstation?
14:40 teryx510 joined #salt
14:42 hellosalt thanks drawsmcgraw/Sylvain31,  maybe i need to clarify a bit more. I'm looking for a nice workflow to test saltstate locally, pre-test states remotely on AWS/GCP and deployment to production.  This means 1) project structure in git need to be able to work locally agiante docker or vagrant (clusters sometimes), the same states can be applied remotely for testing.
14:43 hellosalt for example, how to handle branches, maybe let minion run only against certain branch.
14:44 drawsmcgraw hellosalt: Right. Like, say, via 'saltenv' ?
14:45 Brew joined #salt
14:45 N-Mi joined #salt
14:46 hellosalt something like that. Maybe environment would be match to the branch name. For example, "Bob" branched off to Feature_1, he can label all test minions run Feature_1 branch.
14:46 babilen hellosalt: I typically develop in a vagrant box, that references a local checkout in /srv/{salt/file_roots/foo,pillars/foo} for states and pillars respectively. /srv/salt and /srv/pillars is being mounted as shared folder into the box and I just checkout whatever branch I'm working on there.
14:46 hellosalt same applies when you work on a cluster of Vagrant?
14:47 babilen hellosalt: Once I'm happy with that we have two masters, one for production and one for qa that track different branches. I decided against environments as they make everything complicated and I hate the way git branches are equated to environments anyway
14:47 babilen hellosalt: A "cluster" ?
14:47 drawsmcgraw babilen: "more than one machine"
14:47 hellosalt yah
14:47 hellosalt :)
14:48 drawsmcgraw I've used salt cloud and map files to make VMs, then run tests on those new VMs (vs using Vagrant)
14:48 hellosalt i meant that more than one machine, then orch help to balance the state on the number of servers.
14:48 drawsmcgraw But that means you have a cloud in front of you...
14:50 babilen hellosalt: https://www.refheap.com/119587 -- I always have a master (installed normally) and multiple minions. I obviously only need the /srv/{salt,pillar} on the master
14:50 tapoxi hi everyone, so I'm still diving into salt. what's the advantage of the orchestrate runner over writing my own runner? seems like I'd have more control with the latter.
14:50 babilen hellosalt: That is what I use for *local* development .. I then push it into a qa branch that I can run against a bunch of qa boxes
14:50 Mowens Quick question on beacons for services, "onchangeonly" only checks for the service and alerts if it starts OR stops correct?
14:51 hellosalt i see, that could be a good way for working.
14:51 aw110f joined #salt
14:51 hellosalt for a larger number of server inventories, let's say 1k servers, mixed with different functional set of severs. Does it make sense to use the same salt repo?
14:52 babilen hellosalt: So I essentially write scripts such as the "update-salt-formula" that clone a number of repositories. This will get me the current branch of that repo and it is being treated as normal salt state or pillar data by the vagrant box (it knows nothing of branches or git)
14:52 babilen hellosalt: Are they all different?
14:53 babilen I mean we have some larger number of minions, but they are mostly identical.
14:53 remyd1 Is it possible to overwrite a module function to retrieve more informations than the basic module function; e.g. adding service name in the comments of service.status module ?
14:54 hellosalt babilen: i have that similar script just to manage/trackdown dependencies of formulars.
14:54 hellosalt babilen: they are all different, from message queue, databases, webservers, data process to anything...
14:55 babilen I'm very happy that I adopted ~/src/$HOST/$USER/$REPO for my $HOME directory .. it allows for easy and stable references to repositories
14:55 babilen hellosalt: Well, if they are *all* different and they *all* belong to different projects then you can do what you want. It will be a nightmare to maintain 1k completely different boxes though.
14:56 hellosalt lol. 1k completely different boxes is insane - i would quick if that's the case.
14:56 hellosalt quit*
14:56 babilen I mean, normally, you have webserver 1-500, database servers 1-50, compute boxes 1-500, ... and so on
14:57 babilen hellosalt: Well, you said "they are all different" :D
14:59 babilen I'd split the projects semantically .. keep all boxes in the same repo that have to be present in order for whatever system they belong to "works"
14:59 hellosalt does that mean you have multiple states git repository for each functional group?
15:00 babilen Really depends on your project and team structure. If you want 20 DBAs to work on your database states, then it might make sense to give them their own repository
15:00 babilen (for example)
15:01 babilen I use formulas a lot, so I typically use a large number of repositories already (cf. https://github.com/saltstack-formulas)
15:01 babilen But *projects* have their own repo and simply define the boxes and the states that go in there. The states themselves typically come from specialised project SLS files combines with very generic "formulas"
15:02 babilen (that are all configured and tailored to my needs by pillar data)
15:03 babilen So .. say I have a project "foo.com" that needs a MySQL database. I'd use a foo-project repository in which all the minions are configured, but would use the generic "mysql-formula" to install the db on foo-db1
15:03 Rkp joined #salt
15:03 Ayo joined #salt
15:04 babilen Does that make sense?
15:04 hellosalt what if we need to improve mysql-formula, what would be a sensible way to ensure that foo.com bar.com database are not broken by the formula changs?
15:04 babilen Well, you would obviously use your own fork and merge changes as needed
15:05 Sylvain31 state for recuse remove a directory? rm -rf like
15:05 XenophonF that's what i do
15:05 XenophonF i've forked all the formulas i use
15:05 babilen Do *not* reference the github.com formulas directly .. that way you grant root to all members of that organisation! ;)
15:05 jenastar joined #salt
15:05 babilen (I mean .. thank you kindly, but ..)
15:05 XenophonF and i periodically update the forks, review the changes, and run highstates in my dev environment to make sure nothing broke
15:05 babilen exactly
15:06 XenophonF and sometimes things break, like how salt-formula made a bunch of changes to how the salt.cloud SLS works
15:06 babilen And if something breaks I fix it and get it upstream
15:06 XenophonF really annoying, that one
15:06 XenophonF had to rework my aws configs
15:06 hellosalt that's really helpful!
15:06 Sylvain31 babilen: if they use windows 10 they can safely grant root acces to everyone… I guess :-P
15:07 XenophonF i'm considering moving to a different model
15:07 XenophonF in which i create branches within each of my formula forks
15:07 snergster joined #salt
15:07 hellosalt windows is far from me... I used to see that something requires zookeeper to be run on windows... nightmare.
15:08 brianfeister joined #salt
15:08 XenophonF that way i can keep formulas continuously in sync with upstream but only deploy into dev/test/stage/prod under my control
15:08 XenophonF it'd make reviewing changes easier, too
15:08 Sylvain31 hellosalt: I'm speaking about the privacy policy (=None)
15:08 babilen hellosalt: The other side is pillar organisation .. I tend to keep "project" pillars also, that might combine settings from multiple formulas so that they are all in one place.
15:09 babilen XenophonF: You have dev/test/stage/prod ?
15:09 XenophonF yup
15:09 XenophonF the staging environment isn't really used
15:09 babilen Nice, I have dev (on my laptop), qa and prod ..
15:09 XenophonF it's just a convenient way to host changes to production
15:10 babilen But we are a small shop anyway
15:10 XenophonF i can apply prod changes using "salt ... state.sls ... saltenv=staging"
15:10 hellosalt does minion suppose to be run in a pulled model all the time in cron (in combination of the master push)?
15:10 XenophonF and i can backout with "salt ... state.highstate"
15:10 nmadhok joined #salt
15:10 XenophonF that way the prod environment always reflects what's actually in production
15:10 debian112 joined #salt
15:10 babilen hellosalt: You could schedule a highstate run every k minutes (with either cron or salt scheduler)
15:10 XenophonF kind of a github flow-style approach
15:11 babilen hellosalt: I tend to run highstates manually though
15:12 hellosalt i come from a puppet world... without cron it makes me feel insecure. lol.
15:12 sjmh joined #salt
15:12 XenophonF hellosalt: i'm aiming at a point when the CI system drives highstates
15:13 babilen Yeah, that is a nice goal
15:13 hellosalt maybe getting logs from each local state run, make sure everything is still green after rollout.
15:13 XenophonF e.g., one merges changes from dev to test, the CI system sees that, runs highstate in test plus follow-on functional/system/integration tests
15:13 mohae joined #salt
15:14 XenophonF i want humans in the loop for production deployments though
15:14 hellosalt does the CI ssh into the salt-master node and run salt apply?
15:14 XenophonF well my goal is to be reactor-driven
15:14 hellosalt reactor is life!
15:14 XenophonF so the Ci would use a webhook or something to fire off a reactor that triggered the highstates on the specified nodes
15:15 XenophonF i'm still learning the CI stuff
15:15 XenophonF it's like, a whole nuther toolset on top of salt
15:15 hellosalt i need to read a bit more about how reactor works.
15:15 XenophonF and then the testing stuff, which is a whole nuther toolset on top of the CI tools
15:16 dfinn joined #salt
15:16 hellosalt are there anyway to trigger apply on master without being on the salt-master host?
15:16 XenophonF doesn't help that i'm like john the baptist here, crying "make unit tests the way of all deployments"
15:16 XenophonF yeah
15:16 XenophonF webhooks are my weapons of choice
15:17 hellosalt is it built-in?
15:17 XenophonF yeah but you have to set up salt-api (or rest_tornado or whatever's the recommended WSGI app these days)
15:17 tapoxi joined #salt
15:17 hellosalt ah, that makes sense.
15:18 XenophonF i went the route of using an unauthenticated webhook with a secret in the URL
15:18 hellosalt cool, that's good to know.  sorry for throwing those questions - I'm really new to salt.
15:18 hellosalt probably with ssl enabled :P
15:18 nmadhok1 joined #salt
15:18 XenophonF yes
15:18 XenophonF of course
15:19 XenophonF i wrote the reactor in python, too, so that i could add some authentication there
15:20 hellosalt i need to go, but thanks for all the help here!
15:20 XenophonF https://github.com/irtnog/salt-states/blob/development/salt/files/reactors/github.sls
15:21 XenophonF i had it working but then it broke and i haven't had a chance to fix it
15:22 XenophonF that one was a github webhook that triggered updates of either the salt-master's file server (gitfs) or of checked out formulas (salt.formulas SLS from salt-formula)
15:23 N-Mi joined #salt
15:23 N-Mi joined #salt
15:24 XenophonF writing that reactor was also the point at which i realized that thatch45 tricked us all
15:25 XenophonF salt is obviously a lisp implementation ;)
15:25 pmcg joined #salt
15:25 VSpike Can't help but think that salt formulas need a test branch and stable branch, or similar
15:26 XenophonF well, i kind of understand why they don't
15:26 XenophonF it'd make referencing a formula using gitfs much trickier
15:27 VSpike Would it? I've created my own forks of each with a develop branch that tracks upstream master and a master branch that gets updated when I'm happy with what's in develop
15:27 DEger joined #salt
15:29 VSpike Pillar for the salt formula for master looks like https://bpaste.net/show/c95ec7e61605
15:29 VSpike So at the moment, the dev environment gets its own version of the formulas (from the develop branch) while beta and prd get the stable one
15:30 VSpike I only just set that up recently so I might change it yet :)
15:30 west575_ joined #salt
15:30 zmalone joined #salt
15:31 VSpike I suppose that's the idea of forking the formulas. Stuff gets accepted into them with minimal testing, so you have to treat them as development software and implement your own test/release policy on top of it
15:31 viq hah, out of obdi, saltpad, molten on a test vagrant box only molten seems to actually do anything
15:33 barmaley joined #salt
15:33 autofsckk joined #salt
15:34 rojem joined #salt
15:34 CeBe1 joined #salt
15:35 barmaley joined #salt
15:36 babilen viq: Really?
15:37 viq babilen: could be PEBKAC, but I didn't manage to make obdi nor saltpad display/do ANYTHING
15:39 babilen wow
15:39 debian112 joined #salt
15:40 DEger joined #salt
15:43 amcorreia joined #salt
15:44 Number6 joined #salt
15:45 viq molten has quite a bit of a feel of halite
15:45 Ch3LL joined #salt
15:46 Number6 Is there a way to NOT have the ability to do a cmd.run on a single minion? I still want to be able to do a pkg.upgrade
15:46 Trauma joined #salt
15:46 komputes joined #salt
15:47 watersoul joined #salt
15:48 edrocks joined #salt
15:49 yetAnotherZero Number6: what's the scenario?  could you clarify?
15:49 DEger joined #salt
15:50 bshelton229 joined #salt
15:52 Number6 yetAnotherZero: I have a machine that I don't want a (essentially) root prompt from my master on. It looks like this might work, though, https://github.com/saltstack/salt/issues/25854
15:52 saltstackbot [#25854]title: Disabling CMD module on minion | I was trying to disable the CMD module on minion using 'disable_modules' configuration, so that it should not be available even from master ....
15:52 babilen Number6: https://docs.saltstack.com/en/latest/topics/eauth/access_control.html ?
15:53 aw110f joined #salt
15:53 dgutu_ joined #salt
15:58 N-Mi joined #salt
16:00 mierst joined #salt
16:00 Number6 cmdmod is the correct module to disable :-)
16:01 dgutu joined #salt
16:03 haaj joined #salt
16:03 rgrinberg joined #salt
16:05 Number6 Actually, I take that back... the minion crashes
16:05 viq What's a good way to assign in states 'roles' grains to minions? https://pbot.rmdir.de/FCWF6SOhNoyxdDjPwrs5EQ doesn't work, as apparently applying one overwrites the other
16:05 viq (also I just realised I'll have state ID conflicts, but I'll deal with that in a bit)
16:08 spuder joined #salt
16:10 hasues joined #salt
16:10 haaj joined #salt
16:12 hasues left #salt
16:17 haaj_ joined #salt
16:19 bltmiller joined #salt
16:25 ferbla joined #salt
16:26 Qlawy Do I remember correctly that nova-compute should not connect to database directly?
16:26 Qlawy and nova-conductor should do this instead?
16:34 babilen viq: grains.append?
16:35 lero joined #salt
16:35 pcn Qlawy: do you mean to ask that here, or in an openstack channel?
16:36 writtenoff joined #salt
16:40 zenlot joined #salt
16:41 onlyanegg joined #salt
16:42 Fiber^ joined #salt
16:47 jonher joined #salt
16:48 jonher Hello, I'm having issues with a jinja syntax, could anyone have a look if the syntax is correct? It's hard to find in the jinja docs: https://gist.github.com/Jonher937/3ba33089411a686ec6063cb82ecccd92
16:49 jfindlay jonher: what is the error you are getting?
16:49 debian112 joined #salt
16:50 yetAnotherZero it looks fine to me
16:50 jonher jfindlay: I attached the error that I get when pushing the state (see github gists)
16:52 viq babilen: oh, right, for some reason I was understanding the docs that the grain needs to exist first, now I see I was mistaken. Thank you.
16:52 yetAnotherZero so.. you don't have access to 'logstash' make sure the pillar is assigned in top.sls.  You may want to use salt['pillar.get']('logstash:server', 'fallback value') jonher
16:53 yetAnotherZero i *think* that's what's happening
16:53 jfindlay yetAnotherZero's suggestions are all good
16:54 jonher yetAnotherZero: I would like to not push a default pillar if possible. So the logic I would like is: If the pillar is not defined or the value is not x.
16:54 jonher But I can't get that to work :(
16:54 viq jonher: I much prefer https://docs.saltstack.com/en/latest/topics/pillar/index.html#pillar-get-function
16:54 ageorgop joined #salt
16:54 DEger joined #salt
16:55 viq Ah, yeah, was suggested, sorry
16:55 jonher viq: One less click :) Thanks
16:55 viq But you don't have to define the default value, and I still much prefer that syntax
16:56 mosen joined #salt
16:58 jonher viq: I'll give it a try
16:58 yetAnotherZero jonher: you should be able to structure the logic the same way, perhaps: {% if salt['pillar.get']('logstash:server', False) or (salt['pillar.get']('logstash:server', False) != "log-listener.mydomain.com") )
16:59 yetAnotherZero and jinja won't puke
16:59 jonher yetAnotherZero: Thanks, I'll give it a try :)
16:59 jfindlay jonher: I also commented on your gist
17:00 jonher jfindlay: Awesome, thanks
17:00 yetAnotherZero orrrrr in this specific case you could be all tricky and do this:  {% if salt['pillar.get']('logstash:server', 'log-listener.mydomain.com') != 'log-listener.mydomain.com' %}
17:02 viq yetAnotherZero: you don't need the ', False', you can have just salt['pillar.get']('logstash:server')
17:02 yetAnotherZero viq: right-o  see above
17:03 yetAnotherZero only one compare == kiss
17:03 spuder joined #salt
17:04 jfindlay yeah, I think that is much nicer
17:04 poiuasdf joined #salt
17:05 Qlawy damn ... still having problems with state is not available
17:06 jfindlay what state?
17:07 Brew joined #salt
17:08 yetAnotherZero ok guys, I have one: I want to have a single list of all my servers IP addresses (and some other info) that each minion can access (for things like config file templates) question is: how can I have access to all servers for some states?  like on 'uber-minion' it needs to build a comprehensive list of all of the servers so it can do some operations on each of them
17:08 viq yetAnotherZero: mine?
17:09 Qlawy jfindlay: I wrote my own state and 3 of 4 substates are aviable for minions, but this one? no :/
17:09 yetAnotherZero viq: are you asking about 'one compare' ?
17:09 yetAnotherZero viq: I was talking about mine:  {% if salt['pillar.get']('logstash:server', 'log-listener.mydomain.com') != 'log-listener.mydomain.com' %}
17:09 jfindlay yetAnotherZero: it seems that you could use the salt mine for that
17:10 viq yetAnotherZero: your list of IPs, sounds like a job for the mine system
17:10 yetAnotherZero ahhhh
17:10 yetAnotherZero I'll look at that and see if I can figure it out
17:10 yetAnotherZero I was like "that was a pretty delayed response"
17:10 viq "as if coming from a mine" ;)
17:11 DammitJim joined #salt
17:11 jfindlay Qlawy: what do you mean by substate?
17:12 Brew left #salt
17:13 yetAnotherZero viq: mine reaches out to the minions for information?
17:13 yetAnotherZero all the information I need is in pillars... is there a way I could glob all the pillar files for some states?
17:14 tristianc joined #salt
17:15 sjmh joined #salt
17:15 Qlawy jfindlay: hmm, im kinda new in salt world so ;)
17:16 jfindlay Qlawy: can you paste anything specific?
17:17 Qlawy jfindlay: http://wklej.org/hash/08d3742d757/
17:17 Qlawy this is tree of my formula(?)
17:17 Qlawy I can easly run intrastrucutre.openstack.compute same with neutron and keystone (same level) but no way to run controller :/
17:19 aw110f joined #salt
17:19 jfindlay do you get an error message or can you paste the output when you run those commands?
17:19 Qlawy wait, I just run state.show_sls and now it seems to work oO
17:20 brianfeister joined #salt
17:20 Qlawy anyway I had minion name: intrastrucutre.openstack.controller is not available
17:21 DammitJim do you have as many issues as I do when doing pillar for a template?
17:21 jfindlay Qlawy: could be a typo, `infrastrucutre.openstack.controller`?
17:22 DammitJim my pillar data is like 5 levels deep!
17:22 bltmiller joined #salt
17:23 lero joined #salt
17:23 josuebrunel joined #salt
17:23 teryx510 joined #salt
17:25 ssplatt requisites should work in reactors…right?  like watch, watch_in, requires...
17:26 noraatepernos joined #salt
17:27 nmadhok joined #salt
17:27 Brew joined #salt
17:28 noraatepernos Anyone know of a HA way to get salt to download a zipped up git repo and “merge” it?  We want to put the zip on s3 so we don’t have to rely on github.  The problem is that we have auto-scaling groups fairly frequently scaling in/out so if github is down it would kill us.  Over 100 instances in all aws regions at any given time.
17:32 ssplatt archive.extracted?
17:32 ssplatt then cmd.run rsync -va?
17:32 Qlawy btw is archive.extracted fixed? I mean is it possible to downlad really big archives? Or still ram is an issue?
17:33 jfindlay Qlawy: should be fixed
17:33 DammitJim wow... I didn't know S3 was so populare
17:37 josue joined #salt
17:40 aberdine joined #salt
17:42 aberdine Hi - I'm trying to understand the use case for modules - do people use them in scripts? Or can they be called directly from Python? Or some other way?
17:42 poiuasdf joined #salt
17:43 jfindlay aberdine: do you mean execution modules?
17:43 aberdine yes
17:44 jfindlay in a production environment normally you don't interact directly with execution modules, but they are used by state modules to apply whatever state is desired
17:45 onlyanegg joined #salt
17:45 aberdine from the __salt__ dict?
17:45 jfindlay but there's really no wrong answer :)
17:45 jfindlay right
17:45 aberdine So they're building blocks for states, and other modules?
17:45 aberdine makes sense
17:45 jfindlay mainly for state modules
17:46 haaj joined #salt
17:46 aberdine so where there are execution modules with no state module, that may be just because someone hasn't found the time/need to write it?
17:46 poiuasdf will orchestration allow me to execute commands on minions then execute a local command on master for each minion executed (e.g. to update a local database)?
17:47 fracklen joined #salt
17:47 jfindlay you can also run modules from the command line with the `salt` command or similar, and there are the python and REST APIs as well
17:47 haaj joined #salt
17:48 aberdine I'll have a look at those. Thanks jfindlay
17:48 jfindlay poiuasdf: the easiest way to do that may be to put a minion on the master
17:48 drawsmcgraw Anyone use Vagrant for getting their team mates started with writing Salt states?
17:48 jfindlay aberdine: correct
17:48 drawsmcgraw Not asking for implementation. Just a "yeah and it's great" or "no not really"
17:49 drawsmcgraw I'm getting tired of keeping up with scripts and telling people to run them on a VM in Openstack....
17:49 aberdine drawsmcgraw  yes, it works ok, from my own experience
17:49 cilkay drawsmcgraw: Salt "Just Works" as a Vagrant provisioner. I use it.
17:49 aberdine I have two VMs, a master with a minion and a minion only
17:49 drawsmcgraw thanks aberdine. Good to have a data point
17:50 cilkay We only use masterless minions.
17:50 drawsmcgraw cilkay: You mean to say you use Salt to deploy Vagrant things?
17:50 aberdine I didn't use the vagrant salt integration, just had shell provisioner install salt and NFS mount my salt folders
17:51 poiuasdf jfindlay: i suspected that might be what i have to do
17:52 west575 joined #salt
17:52 poiuasdf not sure how best to pass data between them tho
17:52 aberdine @drawsmcgraw http://pastebin.com/vvSwEcbW is my vagrant file
17:52 cilkay drawsmcgraw: I have a shell script that takes a Debian netinstall CD, creates a custom ISO, and creates a Vagrant box. When we do "vagrant up", we use Salt as a provisioner to put that Vagrant box in the desired state.
17:52 drawsmcgraw cilkay: I see
17:52 aberdine actual a master and two minions, but I only tend to use one as two is more work
17:52 jfindlay if you're doing things on the master system that aren't directly related to salt administration, it's best to run a minion on the same system
17:53 ajw0100 joined #salt
17:53 drawsmcgraw aberdine: *wow*. That makes a master and two minions, all on a private network on (say) a laptop?
17:53 aberdine yes
17:54 drawsmcgraw I assume Virtualbox is the default driver?
17:54 aberdine and sets up /etc/hosts on the local machine and VMs
17:54 aberdine needs a vagrant plugin
17:54 drawsmcgraw plugin....
17:54 aberdine https://github.com/devopsgroup-io/vagrant-hostmanager
17:55 cilkay drawsmcgraw: Additionally, we deploy VirtualBox appliances to our customers that self-configure. The latest iteration has a Django application to provide the YAML file that Salt uses to put the VM in the desired state. It's essentially a self-configuring, disposable appliance. We don't store anything in the appliance that has to be backed up.
17:55 drawsmcgraw cilkay: Fascinating... This sounds much easier than, say, putting the appliance in a Docker container.,
17:55 aberdine in the vagrantfile dir there are folders for /etc which is a full /etc/salt dir, pillars and states. You can edit locally and test in the VMs
17:56 cilkay I've studiously avoided Docker. I might have to take a closer look at it at some point.
17:56 drawsmcgraw aberdine: I see. Effectively mounted into the VMs?
17:56 aberdine yes, via NFS so pretty quick
17:56 aberdine another option is to run the master on the laptop
17:56 drawsmcgraw cilkay: I did too for a good, long while (for similar reasons as yourself, to take a guess). But it's brought a fair bit of sanity to our ecosystem.
17:57 drawsmcgraw Still a challenge and it's certainly not as easy as the brochure says. But it can be useful.
17:57 aberdine I avoided Docker, now I'm starting to very much like it. Combined with salt it's pretty sweet.
17:57 DEger joined #salt
17:57 cilkay One of the reasons we built the Django web app to configure the VM is to eliminate any dependencies on shared folders between the host and the guest.
17:57 drawsmcgraw aberdine: Makes sense. In my case, the more self-contained, the better. I have a few developers that are... less-than-eager to make some Salt contributions.
17:57 aberdine get the whip out :)
17:58 drawsmcgraw aberdine: Definitely. Our deploy workflows have Salt setting everything up, then launching the containers
17:58 drawsmcgraw heh. I've come close :)
17:58 cilkay That will make it easier for us to build images for different hypervisors. We started with VirtualBox, next up is Hyper-V and then VMWare.
17:59 aberdine cilkay does salt also provision the VM itself?
17:59 aberdine via salt-cloud, say?
17:59 cilkay no
17:59 aberdine packer?
18:00 ssplatt are requisites supposed to work with reactors?  for example, https://gist.github.com/ssplatt/21fc7f0bdfeb70ab4a462fe655739ccb idea is only accept a minion if it’s in a dynamic autoaccept file?
18:01 tvinson ssplatt: no requisites in reactors but reactors can call other things like orchestration states
18:01 ssplatt hmm
18:01 ssplatt so not even “onlyif and unless?
18:01 tapoxi joined #salt
18:02 ssplatt i guess this is an orchestration...
18:04 XenophonF ssplatt: it depends on how you code your reactor
18:04 XenophonF if you write it in python, then the reactor can, er, react using whatever logic you want
18:05 ssplatt also true.
18:05 impi joined #salt
18:05 XenophonF that said, my reactor scripts are broken right now so maybe my advice isn't to be relied upon ;)
18:05 ssplatt sounds like i need that thorium thing or whatever...
18:06 tapoxi is it possible to execute salt runners through the client_acl system? seems I can't run manage.up, but I can run test.ping...
18:06 rakan joined #salt
18:07 rakan Hello everyone
18:08 rakan Seems like "require_in" in saltstack doesn't work as described in the docs: https://docs.saltstack.com/en/latest/ref/states/requisites.html#the-in-versions-of-requisites
18:09 s_kunk joined #salt
18:11 rakan using Salt: 2015.8.8.2, here's my issue: https://gist.github.com/rakanalh/f938ec9170df9717f7ae5cd91adfc4b8
18:11 rakan git.latest mainly tries to clone the repo before id_rsa is even executed... both id_rsa & git.latest states are in different files, different modules
18:13 rakan Anyone got any ideas?
18:15 ageorgop joined #salt
18:19 GreatSnoopy joined #salt
18:19 Aleks3Y joined #salt
18:23 nmadhok joined #salt
18:25 lero joined #salt
18:36 dendazen joined #salt
18:41 punkoivan joined #salt
18:45 ajw0100 joined #salt
18:45 Qlawy I do not understand why salt is using my old files (file.managed)
18:46 ElasticApps joined #salt
18:47 tapoxi joined #salt
18:48 Qlawy might it be due to some cache or sth?
18:48 fracklen joined #salt
18:49 dh__ joined #salt
18:49 Qlawy PEBKAC again :/
18:50 dh__ I am using cmd.run in a state file, getting a "failed: mapping values ar enot allowed here;" on the following line:       - name: docker run --net=host gcr.io/google_containers//etcd:2.2.1 etcdctl set /coreos.com/network/config '{ "Network": "10.254.0.0/16" }'
18:50 dh__ I think it's the json object that is breaking things, can anyone help me understand how I can work around this?
18:50 josuebrunel joined #salt
18:51 metalseargolid joined #salt
18:53 bltmiller joined #salt
18:59 smcquay joined #salt
19:00 noraatepernos joined #salt
19:04 manji joined #salt
19:04 aberdine super impressed by salt virt
19:04 dmaiocchi joined #salt
19:05 manji joined #salt
19:06 teryx510 joined #salt
19:08 quix joined #salt
19:08 quasiben joined #salt
19:09 brianfeister joined #salt
19:11 teryx5101 joined #salt
19:13 haaj joined #salt
19:14 deniszh joined #salt
19:16 ajw0100 joined #salt
19:23 edrocks joined #salt
19:24 rgrinberg joined #salt
19:26 djgerm joined #salt
19:27 DEger joined #salt
19:29 onlyanegg joined #salt
19:32 brianfeister joined #salt
19:39 racooper joined #salt
19:42 dh__ is there a way to escape data in cmd.run?
19:42 dh__ this json argument is breaking salt and im not sure how to fix
19:42 DEger joined #salt
19:42 nmadhok yeah
19:42 nmadhok there's a way to escape
19:42 nmadhok are you using cmd.run as a part of your state?
19:42 dh__ yeah
19:43 nmadhok Do the following:
19:43 dh__ - name: docker -H unix:///var/run/docker-bootstrap.sock run --net=host gcr.io/google_containers/etcd:2.2.1 etcdctl set /coreos.com/network/config '{ "Network": "10.254.0.0/16" }'
19:43 dh__ Rendering SLS 'base:kube.master.install' failed: mapping values are not allowed here;
19:43 nmadhok -name: |
19:43 nmadhok and then put your command on the next line
19:45 dh__ does the second line need to be yaml or no white space?
19:45 nmadhok hold on giving you an example
19:47 noraatepernos joined #salt
19:47 nmadhok Run sample curl command:
19:47 nmadhok cmd.run:
19:47 nmadhok - name: |
19:47 nmadhok curl -i -X POST -d 'payload={"key1": "value1", "key2": "value2"}' https://example.com
19:47 brew joined #salt
19:47 nmadhok Does that make sense to you?
19:47 Tanta look up multi-line YAML strings
19:47 dh__ perfect!
19:47 dh__ thank you so much :)
19:48 nmadhok Np
19:48 nmadhok Try that and let me know how that goes
19:48 dh__ worked perfectly
19:48 nmadhok Awesome!
19:50 brianfeister joined #salt
19:51 komputes joined #salt
19:51 ssplatt am i missing how i’d use wheel fucntions on the cli? like if i wanted to tell syndic to accept a key, could i run: salt ‘syndic’ wheel.key.accept ‘minion’
19:52 ssplatt trying to translate this into a runner, or orchestration ...
19:53 yetAnotherZero anyone have tips on why I get 'Minion did not return. [No response]' ?  what kinds of server changes trigger that behavior?
19:53 dh__ do you have connectivity to the minion on the salt port?
19:53 iggy ssplatt: wheel's normally get called from python code (i.e. inside the manage runner, etc)
19:54 ajw0100 joined #salt
19:54 dmaiocchi joined #salt
19:54 ssplatt ok i guess i need to make a proper runner then.  then i can have the reactor kick off teh runner module.
19:54 dh__ It can also be a long response, so if I have connectivity, I generally start the minion in debug mode and try again 'service salt-minion stop && salt-minion -l debug'
19:55 ssplatt yetAnotherZero: it can also be if the minion daemon restarted as a part of the state run
19:56 ssplatt pretty sure i’ve seen that before, when having salt salt itself
19:56 yetAnotherZero ah.  is there a way to have the master retry communication?
19:57 yetAnotherZero (I'm not restarting salt, but I may be having trouble due to a network change, or running sysctl -p, or similar)
19:57 yetAnotherZero It only does it the first time I run the high state
19:59 yetAnotherZero it's kinda lame to have to find which task is doing it, the pull it out into a separate sls so I can execute it before the highstate...
19:59 yetAnotherZero *then pull it out
20:00 punkoivan joined #salt
20:00 aberdine is there a way to lower the timeout for offline minions? I tried -t/--timeout= but it didn't seem to make a difference
20:00 tapoxi joined #salt
20:02 gh34 joined #salt
20:02 punkoivan joined #salt
20:03 freelock joined #salt
20:04 punkoivan joined #salt
20:05 punkoivan joined #salt
20:10 freelock joined #salt
20:21 impi joined #salt
20:23 macheck left #salt
20:24 macheck joined #salt
20:27 AlexF_ joined #salt
20:28 AlexF_ Hello, does anyone have experience with setting up a masterless minion on a raspbian OS?
20:29 twork_ oooh, no but i'd love to
20:29 AlexF_ I'm running into an error that might have to do with the sources.list.d file
20:30 AlexF_ W: Failed to fetch http://repo.saltstack.com/apt/debian/8/amd64/latest/dists/jessie/InRelease  Unable to find expected entry 'main/binary-armhf/Packages' in Release file (Wrong sources.list entry or malformed file)
20:31 noraatepernos joined #salt
20:34 AlexF_ I'm following the installation instructions from http://repo.saltstack.com/#debian
20:35 josuebrunel joined #salt
20:37 bltmiller joined #salt
20:37 babilen There are no armhf packages
20:38 babilen (from repo.saltstack.com, you can naturally use the packages in Debian proper, stretch has recent ones)
20:38 babilen Probably best to perform a pip install from the 2015.8 branch with the bootstrap script though if you are on jessie
20:38 tapoxi how stable is salt-pepper?
20:39 tristianc joined #salt
20:41 DEger joined #salt
20:41 AlexF_ I tried to run the bootstrap script and I ran into a problem. I'll try again
20:42 tapoxi AlexF_ looks like the bootstrap script isn't aware of ARM
20:43 tapoxi AlexF_ yeah, saltstack only has x86 packages, you'd need to find others or compile yourself
20:45 rojem joined #salt
20:45 MTecknology HAHA!! Guys.. this is awesome...
20:46 AlexF_ babilen I tried to run the bootstrap and ran into the same issue
20:46 AlexF_ tapoxi Where could I find others?
20:46 MTecknology I'm deploying a server. When I'm done, a requirement is having a chef cookbook. Instead of writing chef as I go, I'm taking notes using the salt state structure because it's quick and easy to build states as you build the system in the first place.
20:47 tapoxi AlexF_ https://docs.saltstack.com/en/latest/topics/installation/debian.html use the ones from the debian salt team
20:48 AlexF_ Alright, I'll try that route next
20:50 babilen AlexF_: Yeah, hence the pip install (pass -P and the branch name)
20:50 babilen We no longer maintian debian.saltstack.com and packages there are severely outdated
20:51 babilen AlexF_: "bootstrap_salt.sh git 2015.8" would be one approach.. or use a specific tag
20:52 babilen No idea about dependencies in that case
20:52 AlexF_ babilen I see, thanks
20:52 babilen https://docs.saltstack.com/en/latest/topics/tutorials/salt_bootstrap.html
20:52 tapoxi joined #salt
20:53 manji joined #salt
20:53 babilen I'd like to get a backports in jessie-backports soon, but I am waiting for a sphinx backport first (the version in jessie-backports is buggy and prevents salt from building)
21:02 danemacmillan left #salt
21:02 brianfeister joined #salt
21:04 djgerm left #salt
21:08 brew joined #salt
21:16 brew joined #salt
21:24 aw110f joined #salt
21:33 Rumbles joined #salt
21:34 DEger joined #salt
21:39 liskl joined #salt
21:39 rojem joined #salt
21:41 murrdoc joined #salt
21:42 eseyman joined #salt
21:43 DEger joined #salt
21:53 ajw0100 joined #salt
21:53 lero joined #salt
21:56 bltmiller joined #salt
21:56 wendall911 joined #salt
21:56 yetAnotherZero Is there a way to get information from another minion from within a state?  I need two servers to know information about eachother, and I'd like it if it could be via grains access so I don't have to maintain the information
21:59 viq yetAnotherZero: publish or mine
22:01 yetAnotherZero oh!  now, here's a use for mine ;)
22:01 yetAnotherZero viq: do you have an opinion on publish vs mine?
22:02 viq do you want information "as of now" or cached?
22:02 viq If you need reference - https://docs.saltstack.com/en/latest/ref/peer.html   https://docs.saltstack.com/en/latest/topics/mine/index.html
22:02 yetAnotherZero How cached are we talking?  manual updates?
22:03 viq no, that's configurable, one of examples they have is 5 seconds
22:04 viq But cached information is available for minions that are down, and may take some effort to remove if the minion is not coming back. It may or may not be something you want - that's a question of your use case
22:04 tristianc joined #salt
22:06 west575 joined #salt
22:07 yetAnotherZero viq: is this some overhead that might not be desireable, or is this type of info cheap/already happening at some interval?
22:08 sagerdearia joined #salt
22:09 viq yetAnotherZero: there probably would be some overhead. How many minions do you have in mind?
22:11 viq I'm guessing the main overhead would be memory on master, which shouldn't be much, but that accumulate if you have thousands. As for load on minions, I don't believe it would be significant, though that depends on data you want to gather and how often.
22:12 yetAnotherZero joined #salt
22:12 TheoSLC joined #salt
22:13 TheoSLC Greetings.  Question.  Are salt-minion automatically updated now after the salt-master is updated?
22:13 viq TheoSLC: no
22:14 TheoSLC Ok . I ran into this problem https://github.com/saltstack/salt/issues/33429
22:14 saltstackbot [#33429]title: [2015.8.10] KeyError: 'file.source_list' | ### Description of Issue/Question...
22:14 TheoSLC But I guess I don't understand it now.
22:15 TheoSLC oh.  Perhaps they mean "the salt-master was updated while a minion state was running"?
22:15 viq TheoSLC: it sounds like pacakge was updated, but the daemon was not restarted
22:16 TheoSLC viq: the minion daemon or the master daemon.  But we only updated the master.
22:17 viq TheoSLC: the bug you listed, sounds like on machine minion1 salt-minion package was updated, but the salt-minion daemon was not properly restarted
22:18 TheoSLC perhaps somebody else in the org updated the minion.  It was on the latest version.
22:19 TheoSLC (that's why I thought I might of autoupdated)
22:19 TheoSLC viq: thanks
22:19 viq TheoSLC: what distro?
22:20 casey-bowman joined #salt
22:25 * viq calls it a day
22:29 DEger joined #salt
22:32 lero__ joined #salt
22:36 tercenya_ joined #salt
22:37 ry joined #salt
22:37 lero joined #salt
22:41 ajw0100 joined #salt
22:45 nmadhok joined #salt
22:48 jeddi joined #salt
22:48 TheoSLC joined #salt
22:49 TheoSLC viq: Centos7
22:52 DEger joined #salt
22:53 DEger joined #salt
22:56 schemanic joined #salt
22:57 schemanic Hi Everyone. Does anyone have info on whether running the salt minion on Windows 10 works?
23:01 Edgan 2016.3.0 in the process of being released, and the servers behind http://repo.saltstack.com/ are flipping back and forth between 2015.8.10 and 2016.3.0 if you refresh the page. Sigh
23:04 schemanic Anyone?
23:05 rem5 joined #salt
23:07 jfindlay schemanic: I know one of our windows devs runs salt regularly on windows 10
23:09 jfindlay Edgan: if you select the an install repo from the archive, you shouldn't have issues with version changes: http://repo.saltstack.com/apt/debian/8/amd64/archive/2015.8.10/, for example
23:10 jfindlay Edgan: also, we're tracking down some weirdness with the load balancer on repo.saltstack.com
23:10 nmadhok joined #salt
23:10 Edgan jfindlay: http://repo.saltstack.com/apt/ubuntu/14.04/amd64/latest/ is the one I see flipping
23:10 Edgan jfindlay: I am guessing there are multiple servers and at least one of them is out of sync, and has the older version
23:11 jfindlay my guess is that not all of the servers behind the balancer are updated yet
23:11 jfindlay yeah
23:11 drawsmcgraw joined #salt
23:12 ahammond jfindlay 2016.3.0 has finally clawed it's way free? Yay!
23:12 murrdoc is 2016.3 1604 only ?
23:12 Edgan jfindlay: I have a reverse proxy I used for caching apt repositories. The site flipping is breaking the cache and installs.
23:12 murrdoc http://repo.saltstack.com/index.html#ubuntu
23:13 jfindlay ahammond: not just yet, so don't tell anybody until we announce for reals :)
23:13 jfindlay or tell, whatever
23:13 ahammond jfindlay :thumbsup:
23:13 Edgan jfindlay: It is already on github's release page
23:13 jfindlay murrdoc: no, there should be 14.04 packages too
23:14 Edgan jfindlay: Thank you and the team for getting it out the door. I already needed it for salt-ssh.
23:14 ahammond Edgan github's release page is dangerous. I can't remember the details, but I once blew things up by trying to use code pulled from there. :)
23:14 jfindlay and I think 12.04?
23:14 murrdoc all repo links have 1604
23:14 jfindlay murrdoc: example?
23:14 Edgan murrdoc: it is there, but servers are out of sync
23:14 murrdoc http://i.imgur.com/fNDGTMc.png
23:15 murrdoc the links on that page all have 1604 on it
23:15 Edgan murrdoc: It is good news for me. If everyone else would update their repositories for 16.04.
23:15 murrdoc wow
23:15 murrdoc <— is an idiot
23:15 murrdoc missed the tabs to select 1404
23:15 Edgan murrdoc: Refresh http://repo.saltstack.com/apt/ubuntu/16.04/amd64/2016.3/ a few times
23:16 murrdoc https://docs.saltstack.com/en/develop/topics/releases/2016.3.0.html
23:16 murrdoc yeah
23:17 Edgan murrdoc: I see it for 12, 14 and 16
23:18 murrdoc you are corret
23:24 nmadhok1 joined #salt
23:26 eightyeight joined #salt
23:27 smcquay left #salt
23:29 ahammond W: Failed to fetch http://repo.saltstack.com/apt/ubuntu/14.04/amd64/latest/dists/trusty/main/binary-i386/Packages  Hash Sum mismatch
23:29 ahammond @jfindlay from a salt-cloud spin up about 10 min ago.
23:31 jfindlay ahammond: yeah, I don't think the repo update is done yet
23:31 repl1cant so it appears there is something wrong w/the rhel repos too
23:31 repl1cant https://repo.saltstack.com/yum/redhat/7/x86_64/latest/python-tornado-4.2.1-1.el7.x86_64.rpm: [Errno -1] Package does not match intended download. Suggestion: run yum --enablerepo=saltstack clean metadata
23:31 jfindlay in my opinion it's always better to use the 'archive' version of the repo so you know exactly what you're getting :)
23:31 Edgan jfindlay: But then you are version locked :\
23:32 repl1cant and salt-cloud deploys the latest by default :-)
23:32 repl1cant which is where I'm getting my issue from
23:32 jfindlay Edgan: true, but I would imagine that upgrading should happen on your own terms
23:32 Edgan jfindlay: Also seeing flipping here too, http://repo.saltstack.com/apt/ubuntu/16.04/amd64/archive/2016.3.0/
23:33 Edgan jfindlay: same with 14.04 archive. So if I want the latest release now, I still can't get it right via archive.
23:34 jfindlay Edgan: what about an older version of salt?
23:35 Edgan jfindlay: should work, but I was just remaking AMIs, so I wanted 2016.3.0 now. After 3-4 attempted rebuilds it finally worked well enough to build them.
23:35 ahammond So much winning. :)
23:36 ahammond @jfindlay what's the special sauce to pass to bootstrap to target 2015.8 ?
23:36 jfindlay Edgan: from what I can tell, the repo update is the last thing before we officially announce, so it should be soon
23:36 Edgan jfindlay: In my book inconsistent views of the apt repo should not happen at any time. It is a fixable problem.
23:36 amcorreia joined #salt
23:37 salty_solution joined #salt
23:37 salty_solution anyone ever have this error when using vagrant and saltstack?
23:37 salty_solution Failed to fetch https://repo.saltstack.com/apt/ubuntu/14.04/amd64/latest/dists/trusty/main/binary-amd64/Packages  Hash Sum mismatch
23:37 jfindlay ahammond: `stable <version>` where <version> is a branch or a specific release
23:37 Edgan salty_solution: Probably has nothing to do with vagrant
23:38 Edgan salty_solution: They are releasing a new version right this minute and things are inconsistent.
23:38 jfindlay Edgan: yeah, I'm not familiar with the details, but there are some issues with repo.saltstack.com that we're trying to resolve
23:38 ahammond thanks!
23:38 jfindlay ahammond: that will install packages from repo.saltstack.com
23:38 salty_solution Sweet thanks guys. I love freenode
23:39 salty_solution I am just grabbing the stable version vagrant style and it just started failing all of the sudden. :P
23:39 ahammond @jfindlay perfect. I have the code for installing from pip and elsewhere, wanted to add this. Relevant since it's borking a salt-cloud spin up in DO right now. :)
23:40 mosen_ joined #salt
23:41 jfindlay salty_solution: yeah, we're trying to resolve some issues with repo.saltstack.com
23:41 repl1cant i'm attemping --script-args="stable 2015.8.9" with salt-cloud atm myself to see if that fixes it
23:42 mosen__ joined #salt
23:42 salty_solution saul goodman. That must mean that its time to go home. :)
23:43 repl1cant ya, forcing salt-cloud to use a specific version fix it
23:44 repl1cant s/fix/fixed
23:56 nethershaw joined #salt
23:57 rojem joined #salt

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