Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2015-10-13

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

All times shown according to UTC.

Time Nick Message
00:00 kivihtin joined #salt
00:00 mattiasr joined #salt
00:01 otter768 joined #salt
00:06 sjorge joined #salt
00:06 sjorge joined #salt
00:12 otter768 joined #salt
00:14 pravka joined #salt
00:15 perfectsine joined #salt
00:16 dendazen joined #salt
00:18 subsignal joined #salt
00:22 amcorreia joined #salt
00:23 kinetic joined #salt
00:24 MadsRC joined #salt
00:25 kinetic joined #salt
00:26 kinetic joined #salt
00:27 kinetic joined #salt
00:28 kinetic joined #salt
00:30 kineticy joined #salt
00:31 kinetic joined #salt
00:32 otter768 joined #salt
00:32 kinetic joined #salt
00:33 hasues joined #salt
00:38 ageorgop joined #salt
00:38 baweaver joined #salt
00:39 tmclaugh[work] joined #salt
00:43 falenn joined #salt
00:53 jeffpatton1971 joined #salt
00:55 otter768 joined #salt
01:01 cberndt joined #salt
01:23 breakingmatter joined #salt
01:25 penguinpowernz joined #salt
01:26 cberndt joined #salt
01:34 tmclaugh[work] joined #salt
01:35 Chafouin joined #salt
01:37 amcorreia joined #salt
01:45 catpiggest joined #salt
01:45 Chafouin Hi everyone! I have a question: using dockerng, is it possible to pull an image from a public repo without configure the authentification in the pillar data?
01:47 ilbot3 joined #salt
01:47 Topic for #salt is now Welcome to #salt | 2015.8.0 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/
01:49 mattiasr joined #salt
01:53 ^C <iggy> ^C: because the default config file is expected to be empty
01:54 ^C iggy: not if its salt-cloud provisioned
01:57 TyrfingMjolnir joined #salt
02:01 ajw0100 joined #salt
02:10 quasiben joined #salt
02:11 malinoff joined #salt
02:28 tmclaugh[work] joined #salt
02:35 iggy ^C: what does salt-cloud set?
02:48 evle joined #salt
02:55 sunkist joined #salt
03:06 ^C when you tell it to provision salt-minion, i believe it uses the package config and adds master: <ip> and preseeds the minion keys
03:07 favadi joined #salt
03:07 ^C so if you setup using salt-cloud, then switch to managing salt-minion using the salt formula, you get 2 full configs
03:10 bougie joined #salt
03:10 iggy the formula should probably say to rm /etc/salt/minion or put it back to the defaults
03:12 ^C you still need the defaults: minion.d item in the /etc/salt/minion file tho?
03:12 ^C otherwise it would load the config folder items?
03:12 ^C ^wont
03:13 iggy rephrase the question
03:15 otter768 joined #salt
03:17 whytewolf ^C: as per https://docs.saltstack.com/en/latest/ref/configuration/minion.html#default-include default_include defaults to minion.d/*.conf so no, that setting doens't need to be set
03:18 ^C whytewolf: ah, sweet. is it worth adding this as an issue to the salt formula, to get the package config removed?
03:22 tmclaugh[work] joined #salt
03:22 whytewolf only value that the docs say needs to be in the minion config file is master:. so I would say the salt-formula if it takes over in minion.d should reduce the minion config to only that setting.
03:24 whytewolf but then again i am sure that will nuke perfectly good setups that people are using
03:24 hasues left #salt
03:30 amcorreia joined #salt
03:34 ^C yeah, dangerous. perhaps it should just be a toggle
03:36 cberndt joined #salt
03:38 mattiasr joined #salt
03:41 ramteid joined #salt
03:42 guardianL joined #salt
03:44 CaptainMagnus left #salt
03:44 CaptainMagnus joined #salt
03:51 opal joined #salt
03:53 capricorn_1 joined #salt
04:05 opal left #salt
04:05 iggy there have been tickets about this subject before
04:05 iggy no need to open a new one
04:06 iggy if you can find a better way to handle the problem than what was mentioned in the existing tickets, feel free to open a PR
04:07 iggy it's a mess, and adding some half thought out hack isn't going to improve the situation ($0.02)
04:08 ^C yeah, true. i'll go read those tickets. i doubt my current salt abilities will allow me to come up with something better than already discussed :)
04:11 iggy fortunately, this doesn't really take salt skills, just the ability to see multiple points of view
04:13 ashirogl1 joined #salt
04:13 Trivium joined #salt
04:14 Trivium Hey folks. I'm running Void Linux, and want to bring xbps (package manager) support to Salt Stack. What do you recommend I do?
04:14 dingo joined #salt
04:14 dingo YAY COMPUTRES YAY YAY
04:15 * dingo throws a fistfull of car bolts at JoeJulian
04:15 iggy Trivium: look at the existing pkg modules and reimplement the interfaces they already specify
04:16 tmclaugh[work] joined #salt
04:33 mehakkahlon joined #salt
04:34 jodv joined #salt
04:35 jodv joined #salt
04:37 djinni` joined #salt
04:41 TOoSmOotH joined #salt
04:50 viq joined #salt
05:10 tmclaugh[work] joined #salt
05:15 rdas joined #salt
05:19 cberndt joined #salt
05:28 favadi joined #salt
05:36 Aidin joined #salt
05:39 MaZ- joined #salt
05:46 ashmckenzie joined #salt
05:46 impi joined #salt
05:48 ashmckenzie joined #salt
05:49 anmol joined #salt
05:58 tedbot joined #salt
05:58 tedbot left #salt
06:04 slav0nic joined #salt
06:04 tmclaugh[work] joined #salt
06:07 Guest25336 joined #salt
06:15 sirex joined #salt
06:16 lb joined #salt
06:18 katyucha_ joined #salt
06:22 dynamicudpate joined #salt
06:23 dh__ joined #salt
06:24 rawzone^ joined #salt
06:25 av___ joined #salt
06:25 av___ joined #salt
06:25 Garo_ joined #salt
06:26 Buzer joined #salt
06:26 armguy joined #salt
06:26 vlcn joined #salt
06:26 Qwazerty joined #salt
06:27 murkey joined #salt
06:27 TOoSmOotH joined #salt
06:27 ropes joined #salt
06:27 twiedenbein joined #salt
06:28 bernieke joined #salt
06:29 larsfronius joined #salt
06:34 dh joined #salt
06:36 Aidin left #salt
06:40 edulix joined #salt
06:41 mattiasr joined #salt
06:47 ashirogl joined #salt
06:49 mortis_ can you in any way do something like (in file.managed for instance), if source exists do this or that? by source i mean - source: salt://blattiblah
06:53 _JZ_ joined #salt
06:54 babilen mortis_: What are you really trying to do?
06:54 mortis_ babilen: haha :D
06:54 mortis_ import skynet
06:54 mortis_ sort of
06:55 mortis_ im trying to see if the source-file exists, and if it does, do something with it
06:55 babilen https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.cp.html#salt.modules.cp.list_master
06:55 mortis_ its generic, so its not always there on all hosts
06:55 mortis_ oh
06:55 mortis_ oooooh
06:55 mortis_ thats perfect!
06:55 mortis_ havent seen that before
06:55 mortis_ thanks )
06:55 mortis_ :)*
06:56 babilen You are welcome. I do, however, have the feeling as if there *might* be a better way to achieve what you are trying to achieve :)
06:57 mortis_ always is, but you gotta start somewhere
06:57 mortis_ refactor all the things :x
06:58 tmclaugh[work] joined #salt
06:58 mortis_ earlier we defined what files was being copied to a minion
06:59 elsmo joined #salt
06:59 mortis_ i wanna change that to letting the included service do it itself, without any definitions
06:59 mortis_ if it exists on the master, it should be on the minion too
06:59 tedbot joined #salt
06:59 mortis_ typically for nginx and varnish-config
07:01 babilen Can't you get away with a file.recurse and a .d directory there?
07:04 mortis_ .d ?
07:04 mortis_ file.recurse gives an error if files dont exist
07:04 mortis_ but thats what im doing, just thought of adding some jinja logic
07:05 CeBe1 joined #salt
07:05 babilen A lot of services allow for automatic inclusion of all files in a foo.d directory. So you could, for example, recursively sync a directory that is then being used in toto. In the case of nginx you would, probably, make that sites-available (and introduce symlinks)
07:05 babilen Okay, I'm happy if you are happy :)
07:06 Ztyx joined #salt
07:07 mortis_ in nginx i sync sites_avail, yeah, and just symlinks all that exists to _enabled
07:07 mortis_ maybe i could just set include_empty and it would be ok :o
07:08 babilen That was the idea more or less
07:08 mortis_ :)
07:09 mortis_ any salt-folks going over to amsterdam in a couple of weeks btw? oscon and velocity
07:09 tedbot interesting problem: https://gist.github.com/tedops/131097f7e0a5f419f86d, but easy for one of the "Salterati" to solve
07:09 hvn joined #salt
07:09 hvn joined #salt
07:10 tedbot tryna make a /backup/mount directory whose name is the same as the DNS name of the server
07:10 babilen tedbot: Line 13 is broken
07:10 tedbot yes indeed...how?
07:11 tedbot babilen: I'm having trouble with that syntax
07:11 babilen That would be: pillar['admin.yar:%s'|format(....)
07:12 tedbot gah, the darn quote again
07:12 babilen ''|format(...) corresponds to ''.format(...) in Python.
07:13 tedbot babilen: thanks..going to test again now and will let you know
07:13 babilen That being said: I'd rather use salt['pillar.get']('admin.yar:%s'|format(fqdn), SOME_SENSIBLE_DEFAULT)
07:13 babilen Or salt.pillar.get('.....', ....) (whatever syntax you prefer)
07:14 babilen Oh, and it is makedirs not makesdirs
07:14 tedbot could I even do format(grains['fqdn'])? or is that little too Perl-y ?
07:14 tedbot yeah, just threw in makedirs for context
07:14 tedbot that's just a typo
07:14 babilen I'd use salt['grains.get']('fqdn', DEFAULT) again, but yeah
07:14 tedbot ok, I'll go with that then
07:18 tedbot babilen: worked like a charm…thank you very much
07:18 sirex joined #salt
07:19 tedbot I went with the salt['pillar.get'] and salt['grains.get'] as you suggested in both cases
07:21 babilen https://docs.saltstack.com/en/latest/ref/renderers/all/salt.renderers.jinja.html#calling-salt-functions
07:21 Shirkdog_ joined #salt
07:21 guardianL joined #salt
07:22 babilen I might use salt['...'](...) simply because I learned salt before 2014.7.0 was released and it also feels natural in terms of Python, but salt.MODULE.FUNCTION(arg1, arg2, ...) might be better to read.
07:22 babilen I actually haven't quite made up my mind on that :)
07:24 katyucha joined #salt
07:24 saltman_ joined #salt
07:24 tedbot from what i was reading, it sounded like people were using the straight pillar['value
07:24 tedbot pillar['value'] for strong defaults
07:24 tedbot and the original syntax if an actual default will suffice
07:24 saltman_ Hello, I try to  upgrade from salt-2014.1.7 to salt 2015.5.2 with this state:
07:24 saltman_ salt-master:   pkg.latest:     - name: salt-master   service.running:     - name: salt-master     - require:       - pkg: salt-master   cmd.wait:     - name: echo service salt-master restart | at now + 1 minute     - watch:       - pkg: salt-master
07:25 saltman_ IT exit with the following error: [INFO    ] Loading fresh modules for state activity [WARNING ] Failed to import module test, this is due most likely to a syntax error. Traceback raised: Traceback (most recent call last):   File "/usr/lib/python2.6/site-packages/salt/loader.py", line 707, in gen_functions     tag='module',   File "/usr/lib/python2.6/site-packages/salt/modules/test.py", line 25, in <module>     @depends('non_existant
07:25 saltman_ What can be the reason for that?
07:25 saltman_ File "/usr/lib/python2.6/site-packages/salt/modules/test.py", line 25, in <module>     @depends('non_existantmodulename')   File "/usr/lib/python2.6/site-packages/salt/utils/decorators/__init__.py", line 60, in __call__     class wide depandancy_dict   File "/usr/lib64/python2.6/inspect.py", line 953, in stack     return getouterframes(sys._getframe(1), context)   File "/usr/lib64/python2.6/inspect.py", line 931, in getouterframes
07:25 saltman_ tframeinfo(frame, context))   File "/usr/lib64/python2.6/inspect.py", line 906, in getframeinfo     lines, lnum = findsource(frame)   File "/usr/lib64/python2.6/inspect.py", line 568, in findsource     if pat.match(lines[lnum]): break IndexError: list index out of range
07:25 babilen tedbot: No, that's the difference between pillar['key'] and salt['pillar.get']('key', default). salt.pillar.get(..) is simply a different way of writing the latter
07:26 babilen saltman_: Please don't paste into the channel, but use a pastebin such as http://refheap.com, http://paste.debian.net, https://gist.github.com, http://sprunge.us, http://dpaste.de, …
07:26 breakingmatter joined #salt
07:26 Depado joined #salt
07:27 tedbot oh I See
07:27 saltman_ babilen: sorry
07:27 tedbot babilen: I think I prefer the monkey patching…looks cleaner and more logical
07:27 saltman_ this is the issue: https://www.refheap.com/110570
07:28 s_kunk joined #salt
07:28 s_kunk joined #salt
07:29 babilen saltman_: Looks if your upgrade might have gone wrong. How did you do it?
07:30 Grokzen joined #salt
07:30 saltman_ babilen: I put the salt 2015.5.2 (Lithium) in my yum repository, and then run this state
07:31 saltman_ It success to upgrade to salt 2015.5.2 (Lithium)
07:31 CeBe joined #salt
07:31 saltman_ If I run the state again, it will not failed. But in the first time it gives the error I attached
07:32 babilen But not the second time?
07:32 saltman_ Not..
07:32 babilen (or any time thereafter?)
07:33 babilen depandancy_dict is a wonderful name, isn't it btw? ;)
07:34 saltman_ In the first time I run the state, it upgrade from 2014.1.7-3.el6 to 2015.5.2-3.el6 but failed in the error I attached
07:34 saltman_ In the second time, It will not failed, and the state will success
07:34 babilen And your master is working fine?
07:34 saltman_ Yes..
07:35 babilen Do you have other boxes on which you could test this?
07:35 saltman_ The problem is that it failed in the first time with this error stack
07:36 saltman_ I did it in other boxes, and it happen akso on these boxes..
07:36 illern_ joined #salt
07:36 babilen Are all of them on using the ancient Python 2.6 ?
07:36 saltman_ Yes
07:36 babilen -on
07:37 saltman_ we have to use python 2.6
07:37 babilen That would be my initial guess that there's something amiss due to the Python version. Feel free to open an issue on this.
07:37 saltman_ I will open an issue, but there is something to so now? as temporary solution?
07:37 saltman_ It urgent
07:38 felskrone joined #salt
07:38 eseyman joined #salt
07:40 babilen saltman_: Can you upgrade with pkg.upgrade followed by a manual service restart? How many masters do you have?
07:41 saltman_ I have one master
07:41 |Trend| joined #salt
07:41 babilen Which is now upgraded, isn't it?
07:41 saltman_ exactly
07:42 babilen So, why do you want to run that state again?
07:42 babilen That should be a no-op at the moment
07:42 saltman_ I don't want to run the state again
07:42 babilen Are you *also* getting exactly the same error when you try to upgrade minions with a slightly adapted state?
07:42 saltman_ The state failed in the first time
07:42 babilen Or why is this an urgent issue?
07:43 favadi joined #salt
07:43 saltman_ This is urrgent issue because I can't upgrade the salt-master withput getting this error
07:43 babilen You already upgraded your master and it is, according to you, working fine
07:43 saltman_ and it failed in the first attempt
07:44 saltman_ But I write script that to it in all other environments
07:44 saltman_ and in these environments it will fail
07:44 saltman_ *to=do
07:44 nickermire joined #salt
07:44 babilen You just said that you only have a single master. Which is now upgraded. I don't understand your issue nor the pressing need to address this immediately if this is only about the master.
07:44 saltman_ I need smooth upgrade process from 2014 to 2015
07:45 saltman_ In my environment there is only one master
07:45 saltman_ But we have other environments
07:45 saltman_ with one master and 8 minions
07:45 babilen So you have more than one master ?
07:45 saltman_ Not in the same environment
07:46 saltman_ I have other environments with one master
07:46 babilen Well, how many masters do you have that are affected by this?
07:46 saltman_ we have about 20 environments
07:46 saltman_ so 20 masters
07:47 babilen Okay, fair enough. Could you try upgrading one of those masters with pkg.upgrade ?
07:48 saltman_ Yes, I will try it now
07:49 mr-op5 joined #salt
07:52 chiui joined #salt
07:53 tmclaugh[work] joined #salt
07:57 tedbot babilen: thanks again…really appreciate the super quick help
08:01 saltman_ babilen: State pkg.upgrade found in sls opt.salt.update_salt is unavailable
08:02 thefish1 joined #salt
08:02 toanju joined #salt
08:02 mattiasr joined #salt
08:06 traph joined #salt
08:11 babilen saltman_: That isn't a state, but an execution module function that you would run manually
08:11 babilen The test is "What happens if I upgrade that without a salt state?"
08:13 toanju joined #salt
08:14 saltman_ haaa.. ok
08:16 babilen Essentially "upgrade through yum and see how that behaves"
08:16 ssc joined #salt
08:16 ssc left #salt
08:17 thalleralexander joined #salt
08:22 ramteid joined #salt
08:23 Guest__ joined #salt
08:23 stefan_ joined #salt
08:23 Guest__ hi everyone...
08:28 impi joined #salt
08:28 GreatSnoopy joined #salt
08:32 elsmo joined #salt
08:43 keimlink joined #salt
08:47 tmclaugh[work] joined #salt
08:59 larsfronius joined #salt
09:01 mehakkah_ joined #salt
09:06 saltman_ babilen: I upgrade it with the pkg.upgrade
09:06 saltman_ It success, but after I run in the same script another state
09:06 saltman_ It failed with the same error
09:06 saltman_ It seems that the new version of salt is not loaded to the python package
09:08 Ztyx joined #salt
09:10 ws2k3 joined #salt
09:15 patchedmonkey joined #salt
09:25 MadHatter42 joined #salt
09:27 babilen What have you done exactly and what happened?
09:28 patchedmonkey joined #salt
09:29 saltman_ I have a script, called upgrade.py in python: In the first, I run the module: 'salt-call pkg.upgrade salt-master'
09:29 saltman_ and it upgrade the salt-master and salt successfully
09:30 saltman_ In the samr script, I run some states after I upgrade the salt-master and salt. So I run the salt "salt.config" which is my custom state
09:30 saltman_ and this state failed on the same error
09:30 saltman_ of: salt.exceptions.LoaderError: The renderer yaml_jinja is unavailable, this error is often because the needed software is unavailable
09:31 saltman_ If I will run the script again, then the okg.upgrade will pass successfully (Because no upgrade is nedded)
09:32 saltman_ and also the state: salt.config will pass successfully
09:35 VSpike Some has spectacularly broken the https://github.com/saltstack/salt-winrepo
09:35 VSpike Someone^
09:37 VSpike It also exposes a major flaw in the way Salt identifies installed Windows packages - namely that if you want a 32-bit and 64-bit version, it won't work if they have the same "FullName" attribute
09:38 VSpike That problem probably goes away in 2015.8. To be replaced by other, more fundamental problems :)
09:41 trph joined #salt
09:41 tmclaugh[work] joined #salt
09:42 cberndt joined #salt
09:47 Ztyx left #salt
09:48 ericof joined #salt
09:49 CeBe joined #salt
09:49 oravirt joined #salt
09:49 Ztyx joined #salt
09:52 denys joined #salt
09:59 traph joined #salt
09:59 traph joined #salt
10:13 Ztyx1 joined #salt
10:18 favadi joined #salt
10:20 larsfron_ joined #salt
10:21 patchedmonkey joined #salt
10:21 larsfro__ joined #salt
10:21 thefish joined #salt
10:21 permalac joined #salt
10:25 aqua^c joined #salt
10:25 darkpaladin joined #salt
10:25 mehakkahlon joined #salt
10:27 mehakka__ joined #salt
10:29 darkpaladin joined #salt
10:31 larsfronius joined #salt
10:32 giantlock joined #salt
10:32 chrismckinnel joined #salt
10:33 CeBe1 joined #salt
10:34 chrismckinnel hi all, was wondering if anybody could give me some insight as to the best way to structure my users config
10:35 chrismckinnel i have a couple of hundred servers and 4 teams of devs, each team works on a different group of servers
10:35 chrismckinnel i've inherited the current salt config which isn't repeatable at all and it seems users are added by individually calling salt "some-server" state.sls users.username
10:35 tmclaugh[work] joined #salt
10:36 chrismckinnel i want to group the users into their logical teams, and group the severs into their logical groups, and then assign the users access to each group of servers
10:37 chrismckinnel ideally i want to be able to point salt at the config and have it talk to every machine to make sure each server has the correct users on it based on config files
10:37 chrismckinnel does anybody know of a good article, or the section of the docs, which has a good example of what i'm after?
10:39 larsfron_ joined #salt
10:42 CryptoMer joined #salt
10:43 larsfro__ joined #salt
10:44 CeBe joined #salt
10:44 illern_ joined #salt
10:46 lule01 joined #salt
10:46 guardianL joined #salt
10:46 guardianL left #salt
10:46 lule01 hi all, i'm new to salt an playing around a bit...
10:47 lule01 my question is, how often salt minion calls forr highstate?
10:47 lule01 is it done automaticly or does it have to be scheduled?
10:47 lule01 by automaticly i mean out of the box
10:48 lule01 or do i have to schedule runs from the salt master
10:48 lule01 i can't find that in docs
10:49 dendazen joined #salt
10:54 chrismckinnel lule01: salt won't run highstate automatically
10:55 lule01 ok, so it needs to bu ran by cron or tasksched (win minion)
10:56 chrismckinnel lule01: it seems like you can set up a startup_state: highstate in your minion config which will run when the minion starts (like reboot of the machine), but otherwise you'll need to schedule it
10:57 lule01 ok, thanks :)
11:02 babilen lule01: How often do you want your highstate to run? Should it be triggered by certain events or simply run on a set schedule?
11:02 breakingmatter joined #salt
11:02 babilen I run my highstates explicitly btw
11:03 CeBe1 joined #salt
11:04 lule01 babilen: i'm just curious  and learnming... i have it all in my lab only
11:05 babilen I'd play with "salt 'some_minion' state.highstate" on the master and "salt-call -ldebug state.highstate" on the minion while you are learning.
11:07 lule01 i am :)
11:08 babilen Welcome to salt and your route into the depths of infinity btw
11:08 lule01 thank you :)
11:09 lule01 i think it's awsome
11:09 lule01 i'm managing win server with powershell DSC
11:09 lule01 but salt is...
11:10 lule01 i don't have a word fir that..
11:10 lule01 for
11:10 lule01 maybe in a year.... when i real all of the docs
11:10 lule01 read*
11:10 trph joined #salt
11:11 lule01 does it play well with all cloud providers? aws, azure, g cloud?
11:12 Ztyx joined #salt
11:12 Ztyx joined #salt
11:12 chiui joined #salt
11:15 qx joined #salt
11:17 babilen lule01: There is https://docs.saltstack.com/en/develop/topics/cloud/index.html but I don't use it much
11:17 Ztyx joined #salt
11:22 jespada joined #salt
11:22 numkem joined #salt
11:23 illern_ joined #salt
11:24 lule01 babilen: i use vagrant to create ec2 instances, but seems that salt can do it also :)
11:24 lule01 + provisioning
11:27 bluenemo joined #salt
11:32 maveas joined #salt
11:36 ubikite1 joined #salt
11:36 VSpike sanity check please. https://github.com/saltstack/salt-winrepo/blob/master/xming.sls and https://github.com/saltstack/salt-winrepo/blob/master/xming_x86.sls contain the same version and full_name, and that's a major problem. Correct?
11:37 ubikite joined #salt
11:37 VSpike Well, for some values of major, obviously :)
11:37 Reiner030 joined #salt
11:38 linjan joined #salt
11:39 VSpike What i mean is, if you try to install "xming", the minion will scan the installed packages and quite possibly detect the installed package as "xming_x86" since it looks for the first match on name and version. It will then show the installation as failed.
11:39 VSpike It's not just xming either.. there are dozens of packages with this problem
11:40 VSpike Before I got to considerable trouble of de-duplicating them and submitting a PR I just wanted to make sure I'm not being completely stupid
11:40 ubikite1 joined #salt
11:41 shiriru joined #salt
11:42 ubikite joined #salt
11:44 Ztyx joined #salt
11:44 scarcry joined #salt
11:44 CeBe joined #salt
11:46 denys joined #salt
11:48 scarcry joined #salt
11:51 anmol joined #salt
11:53 otter768 joined #salt
11:53 Ztyx left #salt
12:03 giantlock joined #salt
12:04 otter768 joined #salt
12:10 PI-Lloyd afternoon, anyone have any idea why salt is throwing "KeyError: 'fingerprint'" when calling ssh_known_hosts.present on a state run?
12:11 PI-Lloyd that's on 2015.5.2 btw
12:13 tmclaugh[work] joined #salt
12:17 mapu joined #salt
12:20 mapu joined #salt
12:23 bluenemo I run a Startup in Berlin and am looking for Admins to hire. You should have Experience with Debian and Saltstack. We are interested in both Experts and Junior Admins atm. Please send me a PM if you are interested. Thank you basepi for letting me post this here :)
12:24 trapha joined #salt
12:24 mapu joined #salt
12:26 DanyC joined #salt
12:27 TyrfingMjolnir joined #salt
12:29 impi joined #salt
12:29 netcho i'm new to salt but i think smh is fishy here...can someone explain me this?
12:30 netcho http://pastebin.com/yRv2DxDB
12:31 netcho why is centos returning error whe grains targeting win os?
12:33 netcho actualy is NOT returning and imho it SHOULD NOT
12:33 netcho confused
12:37 babilen Can you list packages on dc.home if you target os:CentOS ?
12:37 SheetiS joined #salt
12:37 babilen And is it included if you run test.ping ?
12:39 dendazen joined #salt
12:40 netcho babilen: when i list pkg targeting os:centos i lists centos pkks BUT also win7:
12:40 netcho Minion did not return. [Not connected]
12:41 debian112 joined #salt
12:42 netcho http://pastebin.com/keFJKdh5
12:42 babilen Could you paste the output of "salt 'dc.home' grains.item os" and "salt 'win*' grains.item os" to one of http://refheap.com, http://paste.debian.net, https://gist.github.com, http://sprunge.us, http://dpaste.de, … ?
12:42 netcho sure
12:42 netcho sec
12:42 pravka joined #salt
12:44 fusionx86 joined #salt
12:46 netcho babilen: https://paste.debian.net/315611/
12:47 netcho babilen: https://paste.debian.net/315612/
12:47 babilen netcho: Okay, now: "salt -G 'os:Windows' test.ping" and "salt -G 'os:CentOS' test.ping" ?
12:48 netcho babilen: https://paste.debian.net/315613/
12:50 babilen "centos" != "CentOS"
12:50 netcho the same
12:50 netcho need to paste it?
12:51 dyasny joined #salt
12:51 babilen no
12:51 netcho babilen: https://paste.debian.net/315614/
12:52 babilen How about "salt -C 'P@os:Windows' test.ping" ?
12:52 babilen and "salt -C 'G@os:Windows' test.ping"
12:54 netcho babilen: https://paste.debian.net/315615/
12:55 babilen That is weird
12:55 * babilen blames Bill Gates
12:55 babilen Which version of salt is that? (salt --versions-report)
12:55 subsignal joined #salt
12:56 netcho 2015.5.3
12:56 Guest__ joined #salt
12:57 sfxandy hi everybody
12:57 sfxandy time for another question about that age old topic of roles.  what is the consensus on the best pillar structure if you want to be able to assign more than one role to a server?  a lot of the examples I've seen assign and deal with only a single role being assigned to a given host
12:58 anotherZero joined #salt
12:59 edulix joined #salt
13:00 impi joined #salt
13:04 racooper joined #salt
13:05 numkem joined #salt
13:05 netcho babilen: it's rly strange, i agree
13:07 JDiPierro joined #salt
13:07 sfxandy here's a simplified setup of how we currently implement roles on our infrastructure.... http://pastebin.com/BqNPML9u
13:07 patchedmonkey joined #salt
13:07 sfxandy it just looks "messy"
13:07 aqua^c joined #salt
13:07 sfxandy is there a better way to implement this in Pillar?
13:07 sfxandy any input/comments gratefully received.
13:08 jdesilet joined #salt
13:08 sfxandy i'm trying to take advantage of Pillar namespace flattening so it merges data under ROLES but using that in practice is functional but messy
13:11 MadHatter42 joined #salt
13:16 tpaul left #salt
13:21 tpaul joined #salt
13:24 VSpike Has anyone seen "latest" used as a version number here before? https://github.com/saltstack/salt-winrepo/blob/master/chrome.sls
13:24 VSpike Is that actually valid?
13:26 eliasp VSpike: it is valid, as this version number is just, what the installer/msiexec writes to the registry…
13:27 eliasp VSpike: whether Salt is able to properly non-SEMVER versions is a different topic…but at least for Win/MSI that's valid
13:27 catpig joined #salt
13:29 quasiben1 joined #salt
13:32 VSpike eliasp: ah, ok. thanks
13:33 VSpike did you see my question earlier about the windows repo, by any chance?
13:35 perfectsine joined #salt
13:36 Ztyx joined #salt
13:36 Ztyx left #salt
13:38 Dan_ joined #salt
13:40 kant_ joined #salt
13:41 Guest62253 Can I make a question?
13:42 Guest62253 Where I can find a python exaple that use Salt API for learning?
13:43 _JZ_ joined #salt
13:44 felskrone have a look up https://github.com/saltstack/pepper , that should get you started
13:44 felskrone at
13:46 Guest62253 Thank you! I'll take a look!
13:47 cpowell joined #salt
13:50 zwi joined #salt
13:51 debian112 joined #salt
13:51 quasiben joined #salt
13:54 mpanetta joined #salt
13:54 jeffpatton1971 joined #salt
13:54 zmalone joined #salt
13:55 TranquilityBase joined #salt
13:56 Brew joined #salt
13:56 mpanetta joined #salt
13:57 JDiPierro joined #salt
13:58 amcorreia joined #salt
13:58 kaptk2 joined #salt
14:02 andrew_v joined #salt
14:03 braqoon joined #salt
14:06 tpaul Should I prefix/namespace custom grains to prevent collisions with future versions of salt?
14:07 ferbla joined #salt
14:11 mrtesty joined #salt
14:11 braqoon Hi, Is there a way of feeding output of eg. status.meminfo into eg. variables and not just text output ?
14:11 mrtesty left #salt
14:14 favadi joined #salt
14:20 dkrae joined #salt
14:21 ferbla joined #salt
14:23 clintberry joined #salt
14:27 Guest82546 joined #salt
14:27 sunkist joined #salt
14:28 pm90_ joined #salt
14:29 larsfronius joined #salt
14:32 pm90__ joined #salt
14:33 perfectsine_ joined #salt
14:35 teebes joined #salt
14:42 Gareth o/
14:42 polishdub joined #salt
14:42 indispeq Is it possible on the salt-master, with the salt command to override the pillar and file roots in the configs?
14:43 indispeq i know salt-call has these options
14:43 indispeq but the salt command does not seem to have the --file-root and --pillar-root options
14:43 indispeq if i want to temporarily test with a set of states in another directory
14:44 whidbeywalker joined #salt
14:44 jcockhren This PR fixes the salt provisioner in vagrant. I need people to review and make comments to show interest and get it merged! -> https://github.com/mitchellh/vagrant/pull/6382
14:44 mapu joined #salt
14:47 thefish i have a state that installs some software from a tarball, and i'd like to clean up the extracted files after its completed - can anyone help with making sure its not deleted before everything is done? if i add requires: into the file.absent state, it complains that this is not available
14:47 Guest____ joined #salt
14:48 Guest____ hi everyone.
14:48 babilen jcockhren: Finally! I'm so tired of that/those issue(s)
14:48 sfxandy am after a bit of advice really on a Pillar structure for roles and applying mutiple roles to a host.
14:49 sfxandy this is what we've got currently.... http://pastebin.com/BqNPML9u
14:49 sfxandy but its messy when referencing or searching for specific roles
14:50 sfxandy just wondered what might be a more suitable Pillar structure for this scenario
14:51 babilen jcockhren: How does that play with https://github.com/mitchellh/vagrant/pull/6073 ?
14:51 Kelsar joined #salt
14:54 Kelsar joined #salt
14:56 TooLmaN joined #salt
14:59 giantlock joined #salt
14:59 dfinn joined #salt
15:02 jcockhren babilen: TBH, I think the 6073 is addressing a false issue. The minion config isn't failing to upload due to the privilege, it's failing to upload due to improper config. I can say this b/c my PR eliminates the error posted in #5973.
15:02 Akhter joined #salt
15:03 jcockhren not saying #6073 is wrong per se, saying that the error messages surrounding the salt provisioner is misleading.
15:04 RedundancyD joined #salt
15:05 jcockhren babilen: we should merge both b/c one does address privilege on some level while my PR reverts the bugs introduced in the config
15:05 jcockhren I"m placing this and more on the PR comment
15:06 jcockhren probably a bit more edited and polished than this blurb. lol
15:07 sirex joined #salt
15:07 anotherZero joined #salt
15:10 btorch hi, I have a couple of services (new init.d files managed by salt) that have the service checked with service.running. it always seem to return saying it has been changed
15:11 btorch is that likely because update-rc.d hasn't created the start/stop files ?
15:11 btorch update-rc.d: System start/stop links for /etc/init.d/swift-proxy-sos do not exist .. I thought salt would take care of that
15:15 favadi joined #salt
15:19 jcockhren babilen: see PR comment. A bit more correct. heh
15:20 jalbretsen joined #salt
15:21 favadi joined #salt
15:22 perfectsine joined #salt
15:23 favadi joined #salt
15:23 bw joined #salt
15:31 ageorgop joined #salt
15:31 kinetic joined #salt
15:31 iggy btorch: salt doesn't... it just calls out to the distro tools
15:32 btorch yeah, I guess I need to add a cmd.run state for running "update-rc.d -f  <service name> defaults"
15:33 scoates joined #salt
15:33 iggy that sounds hacky... you can't figure out why the service isn't starting/being set to start automatically?
15:33 btorch the strange thing though is that it somehow enabled it on 2 boxes but failed on 4 others
15:34 btorch the service started up on all 6 though
15:36 pm90_ joined #salt
15:37 aboe joined #salt
15:38 ageorgop joined #salt
15:47 clintberry joined #salt
15:52 RandyT Ryan_Lane: noob here with question about bootstrapping masterless..
15:52 RandyT Ryan_Lane:was reading through your post: http://ryandlane.com/blog/2014/08/26/saltstack-masterless-bootstrapping/
15:53 RandyT Ryan_Lane: can you help me understand what the command or series of commands are when you refer to "when the orchestration is run"?
15:54 RandyT Ryan_Lane: thanks for taking time to share what you have learned on this. I'm intrigued about possibility to run masterless...
15:55 tyler joined #salt
15:55 scoates joined #salt
15:55 Guest99054 hello
15:56 Guest99054
15:58 sfxandy hi
16:00 melog1 joined #salt
16:01 ashirogl joined #salt
16:03 bw hi - i read that i can set test: True in master config file, to globally apply this to states i run, but it is not behaving that way. any one else have this issue or solution?
16:04 bw running 2015.5.5-1
16:08 bhosmer joined #salt
16:08 malinoff joined #salt
16:09 favadi joined #salt
16:09 pcn joined #salt
16:11 sfxandy bw, needs to be set in the minion config file....
16:11 katyucha joined #salt
16:15 aparsons joined #salt
16:16 Akhter joined #salt
16:19 thebwt Is there a way to provide  a custom CA cert so we can use internal certificates with private cloud setups? the insecure: true bit is leaving a sour taste in my mouth.
16:20 sporkd2 joined #salt
16:20 thebwt ^re: salt-cloud
16:22 bw https://docs.saltstack.com/en/latest/ref/configuration/master.html
16:22 bw per this, it can be set in master config, which is simpler than in all the minions for our environment
16:22 tedbot joined #salt
16:25 genomer joined #salt
16:26 sfxandy not really sure bw in that case...
16:26 sfxandy so youre saying that even though the test parameter is set to True, the states still run in non-test mode .... i.e. they apply changes?
16:28 JaredR joined #salt
16:28 bw yes
16:28 tedbot left #salt
16:29 sfxandy not really sure where to go with it - if it were me I might try increasing the log level to see if that configuration directive ie being picked up.  you said setting it in the minion config file was not an option?
16:30 JaredR Question: I generate my /etc/nginx/conf.d/ssl_proxy.conf based on pillar data. When I update the pillar data, I want it to be regenerated. How do I do that?
16:30 Fiber^ joined #salt
16:30 aparsons joined #salt
16:30 bw it's just that those minions are already all out there, and it just seems safer to me to have it set at master level -- ie we don't want someone to accidentally push a state during production hours
16:32 sfxandy ok...
16:32 ViciousLove joined #salt
16:32 tpaul fa
16:33 beardedeagle joined #salt
16:33 pcn JaredR: have you looked at the reactor? https://docs.saltstack.com/en/latest/topics/reactor/
16:33 sfxandy guess if it were me i'd go for increased logging and see if that turned anything up
16:34 genomer Question: I'm having trouble restarting salt-master.  The status states that it entered failed state, but I can still run things via salt and I get responses from each minion.  I ran 'salt-call -l debug state.highstate' and I get all (24) successes.
16:34 pcn genomer: you've probably got an instance that never shut down listening and running.
16:34 pcn Consult your supervisor :(
16:35 teryx510 joined #salt
16:35 genomer that's another weird thing, I'm not seeing any of the daemons under the process status list
16:36 genomer are there times where it's just necessary to restart the master node?
16:36 netcho babilen: i've rebooted the master now grains os targeting works with centos but when targeting os:windows still returnss centos machine :S
16:37 writtenoff joined #salt
16:37 VSpike I'm especially enjoying today how the win_path.exists state is broken in the Windows minion :/
16:37 genomer Question: I guess my main issue is being able to check for instances that 'never shut down listening and running'.  Shouldn't this be easy to kill?
16:38 JaredR pcn: I haven't tried the reactor. I don't need immediate change, but I do want a hightstate or a state.sls to know that it needs to rewrite the file because the pillar data has changed, somehow.
16:38 pcn genomer: netstat -pan | egrep '450[56]' and see what the process name is?
16:39 genomer omg, thank you so much!
16:40 pcn JaredR: Interesting, are you running into this: https://github.com/saltstack/salt/issues/24050 ?
16:40 Norbell joined #salt
16:42 jodv joined #salt
16:43 pm90__ joined #salt
16:44 aqua^c joined #salt
16:45 iggy so... yeah, salt.states.schedule docs contradict each other
16:47 JaredR pcn: Yes.
16:52 netcho where can i find a repo for old centos minions? i want to roll back
16:53 iggy make your own?
16:53 netcho :)
16:54 netcho ok, i'm gonna ask like this.. how do i roll back from 2015.8 ti 2105.5 :)
16:54 netcho to
16:54 jcockhren netcho: fwiw, it's easiest to roll forward and back if you install from git.
16:54 hasues joined #salt
16:55 jcockhren system packaging is always all over the place.
16:55 iggy I kind of doubt there is any place that just keeps infinite copies of old packages around
16:55 netcho i sintalled sfrom saltstack repo
16:55 iggy and afaik, saltstack only plans on maintaining packages for branches starting from 2015.8
16:55 netcho geez this keyboard...
16:56 netcho i've got this new vesrion and it's acting strange
16:56 ageorgop joined #salt
16:56 netcho with the old master in place
16:56 keimlink joined #salt
16:56 iggy repo.saltstack.com never had 2015.5 packages, so it most certainly won't have older versions
16:56 kalnajar joined #salt
16:56 iggy you should always upgrade your master first
16:57 netcho i was addding a new minion
16:57 zmalone joined #salt
16:58 hasues left #salt
16:58 iggy welcome to the exciting world of SaltStack packaging... you never know what you're gonna get
16:58 netcho :)
16:58 netcho iggy: http://pastebin.com/yRv2DxDB
16:59 netcho this is what i get D
16:59 keltim_ joined #salt
16:59 iggy that's with a 2015.5 master and 2015.8 minion?
16:59 netcho yes
16:59 iggy does a test.ping work?
16:59 netcho yes
16:59 iggy or salt-run manage.status?
17:00 netcho https://paste.debian.net/315612/
17:00 netcho https://paste.debian.net/315613/
17:00 iggy windows minion? /me runs
17:01 iggy wait, wtf is going on there? (and for the love of god, use gist and just add more files to the paste)
17:02 iggy you have minions replying to -G os:windows and -G os:centos
17:02 netcho yes
17:03 impi joined #salt
17:03 kinetic joined #salt
17:03 pcn I thought I was told that newer minions with an older master was a no-no
17:04 iggy it's frowned upon
17:04 netcho this is my home lab
17:04 iggy then just upgrade the master and see what happens
17:04 Bryson joined #salt
17:05 kinetic joined #salt
17:06 geekatcmu for pretty much every piece of software, ever, "when using mixed versions, the server-side has to be at the highest in-use version".
17:10 KyleG joined #salt
17:10 KyleG joined #salt
17:10 CeBe joined #salt
17:11 Jahkeup joined #salt
17:12 pcn geekatcmu: sometimes we can do better than that.
17:12 geekatcmu Sure.
17:12 geekatcmu But you don't *bet* on it.
17:14 cyborg-one joined #salt
17:17 scoates joined #salt
17:21 seanie joined #salt
17:22 seanie Running highstates via bamboo. But status returned by bamboo is always going to be successful. Anyone using bamboo and salt?
17:22 quasiben joined #salt
17:24 iggy salt has issues returning meaningful return codes
17:24 amcorreia joined #salt
17:25 geekatcmu Is there a way with jinja+yaml to do logging?
17:26 iggy geekatcmu: be more specific?
17:26 geekatcmu As in, I've got a state that runs in one situation, but outside of that situation, I just want to log a line saying that I didn't run it, and why.
17:26 woodtablet joined #salt
17:27 geekatcmu With any of the py* renderers I can basically just call out to the logging module.
17:27 iggy {% if some_var %}.... states ...{% else %}no-run-id:\n  test.succeed_without_changes{% endif %}
17:27 iggy or something
17:27 geekatcmu OIC.
17:28 geekatcmu So, a "fake" state that just logs a message.
17:28 breakingmatter joined #salt
17:28 geekatcmu And so my log message would just need to be the state name.
17:28 breakingmatter joined #salt
17:28 iggy that's one way anyways
17:28 iggy https://docs.saltstack.com/en/latest/ref/states/all/salt.states.test.html#module-salt.states.test
17:29 denys joined #salt
17:29 iggy has some other interesting state functions
17:29 iggy (i.e. 2015.8 has test.show_notificiation)
17:29 geekatcmu yeah, I'm just reading that page now
17:29 geekatcmu 8)
17:29 geekatcmu It wouldn't have occurred to me to use the test state
17:30 ashirogl joined #salt
17:30 pravka joined #salt
17:31 woodtablet hello i am someone thinking about switching to salt. quick question, does salt have like a prepackaged dev kit or i should say environment kit that wraps around vagrants (similar to chef's test kitchen) ?
17:31 iggy kitchen sink?
17:31 iggy something like that
17:31 woodtablet the chef version is called test-kitchen
17:32 stanchan joined #salt
17:32 woodtablet the salt one is called kitchen-sink ?
17:32 iggy https://github.com/simonmcc/kitchen-salt
17:32 woodtablet ahhh
17:32 woodtablet you are awesome
17:32 woodtablet thank you!
17:33 seanie Anyone got some nice dynamic monitoring ideas for salt states etc?
17:34 KyleG How do you mean?
17:36 geekatcmu seanie: that depends *entirely* on, well, everything.
17:36 geekatcmu You could easily have every state include, say, an appropriate HTTP GET request.
17:37 geekatcmu Or you could have your job cache hooked into your monitoring system by $TECHNIQUE
17:37 seanie YOu mean monitor my job cache?
17:37 winsalt joined #salt
17:37 geekatcmu yes
17:39 iggy I wrote a custom module that sent messages to graphite's webui events page, so I could see lines on all my graphs for every highstate/etc
17:40 seanie do you have a picture of that?
17:40 seanie I can't picture it
17:40 iggy could probably be done with the http module at this point (it didn't exist when I wrote that)
17:40 iggy that was at $last_job, so don't really have anything to show
17:41 CeBe joined #salt
17:42 iggy saltpad I think sets up the redis returner and can display that output
17:44 seanie <3
17:46 iggy looks like it uses mongo actually (/my bad
17:49 beardedeagle left #salt
17:51 jmreicha joined #salt
17:52 dingo 17:28 < geekatcmu> So, a "fake" state that just logs a message.
17:52 dingo geekatcmu: i have your solution
17:53 dingo https://github.com/joejulian/saltstack-logging there you are
17:53 geekatcmu heh
17:53 dingo thisis my job, to wait in #salt for people to ask for a logging module in salt
17:53 dingo and i link them this url
17:53 dingo i do it every few days, it makes me look really important
17:53 * iggy replaces dingo with a bot
17:53 dingo geekatcmu: is CMU == central michigan?
17:53 geekatcmu nice
17:53 geekatcmu Carnegie Mellon
17:53 dingo far more impressive :)
17:54 JaredR pcn: It turns out I wasn't running into that bug. I had failures in a prerequisite that I missed seeing.
17:54 Bryanstein left #salt
17:54 geekatcmu I actually left them 5 years ago, but am unwilling to give up/change the nick.
17:54 saltstackbot joined #salt
17:54 dingo i almost worked at CMU a few years ago, i should have, i regret working for jstor.org instead
17:54 geekatcmu After 21 years there, it's built in to my fingers, and exists in a lot of places.
17:54 dingo it was some python project of using imaging to transfer old mechanical draft drawings into autocad electronic
17:54 dingo it sounded fascinating
17:55 geekatcmu Sounds like something Red would do.
17:55 geekatcmu (I didn't know him, but it still does)
17:55 dingo are you still in PN, geekatcmu ?
17:56 geekatcmu ',PN
17:56 dingo i may fly to pittsburgh next week to purchase a car :)
17:56 geekatcmu Oh, PGH.
17:56 geekatcmu No.
17:56 geekatcmu Reno, NV
17:56 dingo oh yuck
17:56 dingo i met my girlefriend in reno, but its still a yuck town :)
17:56 geekatcmu Eh, it's actually kind of nice.
17:57 geekatcmu I don't often go into Reno proper, and suburbia is suburbia
17:57 dingo yeah good house costs there i'm sure
17:57 dingo i'm out in the santa cruz mountains of california
17:57 iggy better than LA
17:57 geekatcmu Not so great, really.
17:57 geekatcmu Too close to CA and Tahoe
17:57 dingo python user group in santa cruz tonight if anyone in the area
17:57 dingo http://www.meetup.com/Santa-Cruz-Python/events/225774780/
17:58 zmalone joined #salt
17:58 quasiben joined #salt
17:58 geekatcmu Really, the only thing I don't like about Reno is that it feels like the casino's have stamped out all of the non-casino entertainment (no top golf, only one public swimming pool left, all the board gaming groups seems to have shut down, etc)
17:59 saltstackbot [reddit-saltstack] jira service issues with salt https://www.reddit.com/r/saltstack/comments/3o4ro9/jira_service_issues_with_salt/ - 2015-10-09 - 19:45:09
17:59 dingo mmm, yeah, that's how i see it too -- vegas has MORE than just gambling, right
17:59 dingo theres a lot of shows, special events, convention centers, etc.
17:59 dingo plenty of non-gambling things to do in vegas, not so sure about reno
17:59 dingo i don't care for tahoe anyway, i prefer yosemite national park
18:00 woodtablet there are those great reno 911 pd peeps
18:00 dingo lol i forget about that reno 911 show, that's a halarity isn't it
18:00 woodtablet ya, till you go to reno lol
18:01 * dingo resumes asyncio/telnetlib3/pexpect python project work
18:02 woodtablet another chef comparison.. so chef we have cookbooks, and occasionally the cookbook would change functionality or make grand assumptions, breaking my production stuff. i "fixed" that by forking every cookbook I ever used and its dependencies.. which was not fun. Can this happen to me in salt and how do i avoid it ?
18:02 woodtablet lucky dingo
18:02 perfectsine_ joined #salt
18:03 ViciousLove joined #salt
18:03 geekatcmu woodtablet: of course it cant
18:03 geekatcmu s/cant/can/
18:03 geekatcmu And will
18:03 geekatcmu That happens for every CM, ever.
18:03 geekatcmu Because "people"
18:03 woodtablet ya..
18:03 woodtablet i know
18:03 woodtablet how do you avoid it ?
18:03 iggy salt is like a swiss army knife... with a gun
18:03 geekatcmu Even if all your formulae are written internally, it will happen.
18:04 woodtablet internally, i can find that person
18:04 dingo woodtablet: you should manage your deployment of the "salt package", build it yourself
18:04 woodtablet and do fun things to
18:04 dingo internally
18:04 geekatcmu You do your best to learn from everything that has happened before and write "appropriate states" to avoid it.
18:04 dingo so if you need to manage downstream forks while awaiting upstream pull requests, etc., you can manage all of this in your custom "salt" package independent of upstream changes
18:04 iggy it is standard practice to fork whatever formulas you use into your own repo anyway
18:04 fusionx8_ joined #salt
18:04 dingo then you can *plan* upgrades, say every quarter, where you rebase all of your work upstream, etc.
18:05 dingo my biggest issue was tracking salt itself, sometimes we weren't sure when our pull requests would get merged
18:05 dingo might be 1 hour, might be 3 days, but we couldn't stall development work meanwhile
18:05 zmalone woodtablet: In chef, you'd have managed that by wrapping cookbooks and using the version limits in berkshelf to pin yourself to a version after you found that it worked.
18:06 openfly i think yaml was a mistake for salt
18:06 zmalone Then the maintainer can go extend the cookbook if they wanted, and you could continue to follow along until it broke for you or no longer did what you want
18:06 woodtablet iggy & dingo - thanks, i ll do that
18:06 openfly it probably needed an actual declarative language.
18:06 woodtablet zmalone - yep, that is what we switched to
18:06 openfly =/
18:06 dingo openfly: i disagree
18:06 zmalone (I realize you are no longer using chef, but that is a problem that has a work flow to accommodate for it)
18:06 JoeJulian openfly: I agree. It should have been properly formatted xml.
18:06 openfly ew
18:06 openfly dingo i can't see it
18:06 Arbusto joined #salt
18:06 openfly as it is, it lacks necessary linguistic features
18:07 openfly and there's no fix
18:07 dingo openfly: by being *independent* of python, a yaml+jinja -- salt could conceptually be authored in an entirely different language.
18:07 openfly just a big fat... that's expected functionality, when there is in fact no functionality
18:07 openfly dingo and i am down with that
18:07 openfly but yaml isn't a declarative language
18:07 openfly and jinja isn't enough
18:07 dingo I don't think it lacks the necessary linguistic features. I am also frustrated by limited jinja (or mako, if you prefer), but I think you might have to think you may require programatically generating yaml as part of your chain
18:07 openfly jinja is a templating language
18:08 openfly dingo if i do that salt is no longer a cfm
18:08 openfly and then i should just kick it out of the environment
18:08 dingo right, but that's good, because you're not tied into salt
18:08 openfly as it's not a very good datastore.
18:08 dingo when i used chef with the ruby-based DSL, i was absolutely pigeonholed into chef and ruby
18:08 dingo there was no translation
18:08 openfly rvm makes chef useless.
18:08 openfly tbh
18:09 dingo with a declaritive yaml language, we could actually translate salt states into any other language, it can be fully comprehended without being interpreted at run-time
18:09 openfly puppet at least has a real declarative language and you can survive without getting drawn into ruby hell
18:09 dingo much of the problem with these DSL's is they are also dynamic languages
18:09 openfly and proven by enterprise puppet the puppet declarative language is portable to other underlying languages.
18:09 dingo so the only way to know the effect is to execute it
18:10 openfly you need to unit test in salt as much as in any of them
18:10 dingo with a yaml state, you can predict its outcome without fancy intepretator logic of objects and magic methods and their effects
18:10 openfly only unit testing salt is basically impossible
18:10 openfly and you need to dynamically state verify in most cfm workflows
18:10 openfly again... can't do that in salt.
18:11 openfly i don't believe salt was successful until i run unit and functional tests against the target it acted on
18:11 openfly this isn't because i expect salt to fuck up
18:11 zmalone That's a fair assumption
18:11 openfly but because deployment can be complexe
18:11 sgargan joined #salt
18:11 zmalone I've had enough weird salt bugs to expect it to fuck up in some way
18:11 openfly it's not even just salt
18:11 openfly deploy tomcat
18:11 openfly it has spurious delays in loading
18:11 dingo openfly: i think you might investigate including *verifying* salt states, so that salt states are in ofthemselves your unit tests
18:11 openfly wierd little bugs pop up
18:12 Gue______ joined #salt
18:12 openfly if you have dependencies in services you need to check off your unit tests and functional tests before proceeding
18:12 openfly dingo i did
18:12 openfly and it couldn't be done
18:12 dingo i would write salt states such as, "yes i know the daemon is started, required, but ensure the port is binded, require that, and when it is, ensure this curl http api command works, also"
18:12 openfly at least not without programmatically creating dynamic unreadable salt
18:12 dingo i thought it was trivial
18:13 openfly look at watch, require, and include
18:13 dingo cmd.run /opt/my/deployed/program-checker.py {host} {port}
18:13 openfly and how they get scheduled as runs
18:13 openfly if you have an iterative
18:13 dingo thats all that was sufficient for a unit-test-as-state
18:13 dingo functional rather
18:13 openfly you need to change the state name for each individual pass
18:13 dingo wow you have a lot of complaints :) I'm not sure I'm up for such discussion
18:13 dingo i don't even work with salt currently
18:13 openfly =P
18:13 woodtablet lol
18:13 openfly i do and it's slowly driving me mad
18:13 dingo i've used most of the CM's also, and I hate them all, honestly
18:13 dingo frnakly, I hate linux!
18:14 openfly i really liked parts of opsware
18:14 openfly the global shell mainly
18:14 dingo i'm a grumpy old bastard so this is a conversation i should defer :)
18:14 openfly but it was too broken to work with otherwise
18:14 toanju joined #salt
18:14 openfly puppet is survivable tbh
18:14 woodtablet openfly - why wouldnt a functional test like dingo work ?
18:14 woodtablet dingo's i mean
18:14 openfly i'd like to see something more in the vein of global shell meets mesos
18:14 openfly woodtablet cmd.runs nonstop
18:14 dingo i found it more than satisfactory, i recommend such design woodtablet
18:15 openfly i'd like to do a cmd.script instead
18:15 openfly to avoid having to install a battery of test scripts on every single node
18:15 openfly but i can't
18:15 openfly i'd have to use cmd.run or cmd.mod or something
18:15 iggy I have that person /ignore'd for a reason
18:15 openfly and even then
18:16 geekatcmu Repeat after me: all hardware sucks, all software sucks, but the PDP-11 was kind of neat.
18:16 geekatcmu The monastery is over ---->
18:16 whytewolf lol geek
18:17 woodtablet i am too young for such things
18:17 openfly you know... a friend of mine has a pdp-11
18:17 openfly discovered something cool about them
18:17 Riz They'll heat your house?
18:17 openfly dec bus has same number of pins as GPIO on beaglebone blacks
18:17 openfly he made an expansion board for it
18:18 openfly the beaglebone.
18:18 openfly to plug into the dec bus on the pdp 11
18:18 openfly kinda mad.
18:18 geekatcmu That bus was used to talk to pretty much everything, so not really that surprising.
18:18 openfly actually pretty useful as it turns out
18:18 rickflare quick questions guys if I am deploying a file via source and using jinja. How do I prevent salt from ammending the file on each run. Its as if salt is not seeing that the file has been deploy and has not changed.
18:18 openfly linux has software for emulating most of the hardware
18:19 geekatcmu Yes, DEC had a lot of great tech
18:19 openfly so you can dynamically emulate hardware components on the PDP-11 from the beaglebone
18:19 openfly rickflare I don't know.
18:20 openfly are your pillar / jinja variables changing?
18:20 iggy rickflare: can you gist the state?
18:20 whytewolf rickflare: gist it
18:20 Riz rickflare: are you using file.appaned rather than file.managed (or some such) by chance?
18:22 rickflare let me check
18:22 rickflare yea
18:22 rickflare I am
18:22 Riz That'll do it :)
18:22 rickflare ok
18:23 Riz file.append will append the file on each run (from my understanding)
18:23 rickflare is it bad to use the append vs manage
18:23 Riz Depends on what you are trying to do I suppose
18:23 rickflare like from a programming standpoint
18:23 Riz Personally, I don't think I've ever used the append function.
18:23 rickflare I prob should not use the append
18:23 Riz Though, still somewhat newish.
18:23 rickflare no riz
18:24 rickflare i think you are right sir
18:24 rickflare appreciate the help
18:25 zmalone Riz/rickflare: I use file.append to work around the bugs in pkgrepo.<foo> on debian
18:25 rickflare im just really being picky
18:25 Riz zmalone that would be a good use case for append
18:25 iggy heaven forbid anybody actually fix the complaints they have about pkgrepo
18:25 * iggy runs
18:25 Riz lol
18:25 rickflare and trying to ensure my salt runs are fast as possible
18:25 rickflare the append
18:26 rickflare seems to add to the run time
18:26 rickflare nothing major but still slow none the less
18:26 nofxrok joined #salt
18:26 Riz I stick to managed, as that way I know, without question (in theory), the file that is in my git repo is 100% the same as what ends up on my hosts
18:26 rickflare yea
18:26 Norbell joined #salt
18:26 dingo ^ double plus good riz statement
18:26 zmalone I think .append also makes sense for system files where you REALLY don't want to manage the entire thing, because your package manager needs to be able to change the content
18:26 dingo determinism is what this is all about
18:26 Riz but again, depending on what you are doing/trying to do, that might not be the right choice :)
18:26 iggy I didn't even know file.append had a source arg
18:27 dingo if you work against deterministic deployments, you work against salt, you struggle
18:27 rickflare in my case now that im looking at its kind of lazy
18:27 rickflare im basically using append on modprobe blacklists
18:27 dingo can't blacklists be split in multiple files?
18:27 rickflare I did the append because I was nervous about the system adding stuff to that file depending on hardware etc
18:27 dingo just make one file per system-choice of inclusion
18:28 zmalone Like /etc/default/grub or /etc/sysctl.conf or maybe your network config files
18:28 edrocks joined #salt
18:28 rickflare dingo really?
18:28 * openfly goes back to writing scripts
18:29 whytewolf rickflare: you could just file.manage a seperate blacklist such as /etc/modprode.d/blacklist-salted.conf
18:30 whytewolf that way you don't interfear with automated systems And you have your own seperate blacklist
18:30 Riz +1
18:30 Riz that's what I'd be doing
18:32 ashirogl left #salt
18:32 rickflare outstanding
18:33 rickflare thanks fellas
18:34 losh joined #salt
18:36 pm90_ joined #salt
18:36 wryfi joined #salt
18:36 wryfi is it possible to use pyobjects renderer for top files? if so are there any examples anywhere?
18:40 nate_c joined #salt
18:40 pm90__ joined #salt
18:42 iggy wryfi: you could always try it
18:43 wryfi iggy: i haven't been able to figure out a syntax that works, thus i am asking here. ;)
18:43 whytewolf sometimes I sit and wonder what happened to the K.I.S.S. principles when it comes to IT engineers.
18:44 iggy they grew up and figured out it's not always best/possible/etc to keep it simple (or stupid... or you're stupid... or whatever)
18:50 ajw0100 joined #salt
18:52 breakingmatter joined #salt
18:55 fusionx86 joined #salt
18:57 Joke joined #salt
19:00 giantlock joined #salt
19:00 fusionx86 joined #salt
19:03 whytewolf iggy: true. but there is a happy med. between K.I.S.S. and OMG I need six layers of reduntent failovers cause this might break.
19:03 edrocks joined #salt
19:05 kinetic joined #salt
19:08 RandyT Greetings
19:09 RandyT Hoping someone can give me some guidance regarding the following failure to deploy windows ec2 instances.
19:09 RandyT https://gist.github.com/rterbush/97efd9356ac498677fc8
19:09 RandyT I am successfully deploying Linux instances.
19:10 RandyT This server is running and port 445 is open for incoming.
19:12 nzero joined #salt
19:12 perfectsine joined #salt
19:15 ziro` joined #salt
19:15 clintberry joined #salt
19:15 ziro` What strategies exist for hands off deployment of instances I can control via salt
19:15 zmalone anecdotally, I'd say "avoid 2015.8.0 for now"
19:16 zmalone I'm rolling back an initial 2015.8.0 test for now
19:16 bVector ziro`: salt-cloud ?
19:16 ziro` For example, I’m trying to come up with a strategy for bringing up jenkins instances, but I don’t want to bake in my ssh keys… Nor ssh in and set it up manually
19:17 ziro` bVector: I have my own images I want to manage
19:17 bVector on what platform
19:18 ziro` Whatever
19:18 ziro` I’ll be using docker images if one must know
19:19 kinetic joined #salt
19:19 Ztyx joined #salt
19:19 dingo typically ziro you bake an image in a form that is very easy to rapidly reassemble
19:19 dingo docker provides something like that with snapshots right, so theres that
19:20 ziro` I mean I can have my own salty-jenkins image that phones home to the master on a domain name
19:20 ziro` but… feels icky
19:20 dingo in the idea of baking an image, sprinkle a few toppings in a low-cost operation to prepare it for a CI
19:20 bVector ziro`: if baking a hostname makes you feel icky what you're looking for is a service discovery framework
19:21 bVector people typically use etcd or consul for something like that
19:21 dingo ^ +1 agree w/b
19:21 Ztyx left #salt
19:22 ziro` Okay, I can solve that
19:22 ziro` What about masters?
19:23 ziro` Do people nominate and maintain a master?
19:23 ziro` Make master their workstation?
19:23 ziro` stand alone master?
19:23 bVector salt master? depends on your needs
19:23 seanie Our setup is master per env and then master of masters
19:24 seanie Could be over kill
19:24 ziro` env?
19:24 bVector dev/test/production etc
19:24 seanie environment dev qa qa1 stg
19:24 ziro` kk, never considered that approach before
19:24 ziro` solves some other problems I’ve been thinking about though
19:25 brd_ left #salt
19:25 ziro` I guess part of my problem is I’m over engineering my personal stuff
19:26 ziro` I don’t have a real problem to throw this at
19:26 whytewolf personally I typicall have a system that I make the over all master that handles all aspects of deployments. from baremetal deployments to salt deployments. it is also the only system i actually log into.
19:29 kinetic joined #salt
19:29 ziro` tbh, I just want to manage my jenkins configuration
19:29 gekitsuu joined #salt
19:29 ajw0100 joined #salt
19:30 ziro` and I’m throwing salt at this problem
19:31 gekitsuu Has anyone run into config files being so close to Jinja syntax that the parser tries to read elements of the config as if they were jinja and fails? The problem I'm having is with an Exim config file and I was wondering if there's an easy way to change the Jinja {% %} tags to stop the file from causing the state to fail
19:31 whytewolf single system? hell go masterless then
19:31 ziro` whytewolf: actually, it’s going to be containerised & cloud distributed
19:32 bVector ziro`: you can even go masterless
19:32 ziro` I’m planning to run jenkins ina docker container, with salt, so I can push the configuration out to it
19:33 gekitsuu ziro` the pipelines plugin is good for visualizing that
19:33 opensource_ninja joined #salt
19:33 ziro` gekitsuu: visualising what?
19:34 gekitsuu You can set up dependencies in jenkins and the pipeline plugin can show you how deployments are rolling out. It sounded like you were going to use Jenkins to push out config changes to envrionments
19:34 kinetic joined #salt
19:34 ziro` gekitsuu: no, salt to push jenkins configuration
19:35 gekitsuu Ahhh ok nevermind then :)
19:37 kinetic joined #salt
19:37 whytewolf ziro`: spinning  up a master to control a single instence does seem like a bit of overkill. as everything almost everything you would use from the master [except salt-cloud] can be done directly in the minion with salt-call --local
19:38 ziro` tbh, if i’m gonna go masterless i might as well just curl a tar down
19:39 gazarsgo why is pkg.installed so slow (yum/rpms) and is there an easy way to speed it up ?
19:41 ziro` does anyone else use dns based minion home calling?
19:41 ziro` i’m thinking of going saltmaster.mydomain.com
19:41 ziro` any better suggestions for managing this?
19:41 bVector if you use salt.yourdomain.com you wont have to set the master id
19:42 bVector salt minion looks for 'salt' hostname automatically
19:42 whytewolf ziro`: use salt.<domain> and have search domains setup.
19:42 bVector but no, for regular system, a dns name is fine
19:43 ziro` hmm
19:44 whytewolf ziro`: also you might want to actualy read Ryan Lanes blog. http://ryandlane.com/blog/ gives some good insites into how salt is being used at a large company. in a masterless configuration.
19:44 ziro` thank you
19:45 ziro` I really *don’t* want to start baking salt into all my docker images
19:45 ziro` but jenkins seems like a legit reason to do so
19:45 ziro` (for that image, anyway)
19:47 wryfi i'm trying to add an external (python) pillar to my master
19:47 wryfi i am able to query the pillar data successfully from the command line
19:47 wryfi but with the pillar enabled, i cant seem to call any states
19:47 wryfi if i try state.show_top, i get 'Malformed topfile (state declarations not formed as a list)
19:47 wryfi but the same topfile is fine if i disable the external pillar
19:47 wryfi any ideas?
19:48 chiui joined #salt
19:49 danlsgiga joined #salt
19:50 Ztyx joined #salt
19:50 danlsgiga Any way to call an orchestrate from inside another orchestrate?
19:51 whytewolf danlsgiga: no
19:52 Ztyx left #salt
19:54 Akhter joined #salt
19:54 slav0nic joined #salt
19:55 danlsgiga whytewolf: :'(
19:55 nofxrok joined #salt
19:55 danlsgiga whytewolf: So I can't have a moduler orchestrate...
19:55 danlsgiga *modular
19:55 nate_c left #salt
19:56 danlsgiga whytewolf: But can I send an event from an orchestrate?
19:56 danlsgiga whytewolf: That would work fine for me and the reactor would call another orchestrate
19:57 whytewolf danlsgiga: didn't say you couldn't have a modular orch. just said you coudn't call and orch from orch. and yes. module.event does exist so you can call an event
19:57 whytewolf danlsgiga: https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.event.html
19:58 danlsgiga whytewolf: But the orchestrator runs on the master, the modules event seems to be available only in the minion
19:58 whytewolf danlsgiga: ... is your master not a minion also?
19:58 danlsgiga whytewolf: no
19:59 ashirogl joined #salt
19:59 danlsgiga whytewolf: I'm avoiding the approach of having master managing itself
20:00 RandyT Still hunting this problem with windows instance deployment: https://gist.github.com/rterbush/97efd9356ac498677fc8
20:00 whytewolf danlsgiga: why? why avoid something that makes your life easier?
20:00 RandyT If I execute the winexe command without the password, I am prompted, but the generated password does not work.
20:01 danlsgiga whytewolf: easier doesn't always mean better or desirable
20:01 danlsgiga whytewolf: would be easier to call an orchestrate from inside another one :)
20:02 danlsgiga whytewolf: not complaining, I have my own reasons to not go over that path
20:03 ashirogl joined #salt
20:03 aparsons joined #salt
20:03 scoates joined #salt
20:04 mike2512 joined #salt
20:04 chiui joined #salt
20:05 whytewolf danlsgiga: theoreticlly you could use module.runner to run something. but again. this would require your master to be a minion. pretty much everything that you want orch to do on the master requires the master also be a minion
20:06 mike2512 hi all - is it possible to have a state that runs a mysql query, BUT the credentials have to be in the state itself (the state will be a reactor state - so the pillar data is not yet generated) - thanks
20:07 whytewolf anyway. lunch time
20:07 elsmo joined #salt
20:07 geomyidae_ joined #salt
20:09 danlsgiga whytewolf: Much appreciated anyways... That's the thing... A runner is something that runs in the master by definition, why do I need to be a minion to actually have that?
20:10 toanju joined #salt
20:10 Ryan_Lane basepi: https://github.com/saltstack/salt/issues/27911 <-- very critical bug that I'll have a PR incoming for soon
20:10 saltstackbot [#27911]title: rules_egress in boto_secgroup should not manage egress rules, if set to None | This is a critical issue in 2015.8+. rules_egress support was added to boto_secgroup, and it changes the behavior in a way that will cause outages for anyone using it....
20:11 Ryan_Lane it affects salt stable and it'll cause outages for anyone using it
20:11 basepi Ryan_Lane: Thanks! Very good catch!
20:16 Ryan_Lane basepi: I don't know if thay feature made its way into a point release yet, but if it hasn't the point release should wait until the fix
20:16 teebes joined #salt
20:17 DammitJim joined #salt
20:17 perfectsine joined #salt
20:17 iggy hooray for adding features in point releases
20:18 basepi Ryan_Lane: =\ do you have an originating pull request? 2015.8.1 is announcing to packagers like today. Or maybe yesterday.
20:18 basepi Hoping that breakage is not in there.
20:18 Azulez joined #salt
20:19 Ryan_Lane this may be one that cause the backwards incompat: https://github.com/saltstack/salt/pull/23482
20:19 saltstackbot [#23482]title: Boto secgroup improvements | Minor changes - removing requirement to specify egress rules (even if empty). Fixing change detection when a rule is for all ports - boto expects input of -1 to -1 as port range, but seems to return None, which breaks the comparison in _check_rule.
20:20 Ryan_Lane more likely this one, though: https://github.com/saltstack/salt/pull/23309
20:20 saltstackbot [#23309]title: Egress rule support for boto_secgroup | I've written a patch to add a new feature and a minor fix to modules.boto_secgroup and states.boto_secgroup. The changes are:...
20:20 zmalone https://docs.saltstack.com/en/latest/topics/releases/2015.8.1.html looks like it's been announced
20:24 whytewolf iggy: adding features in point releases isn't as bad as breaking backwords compate in point releases. [both are pretty bad though]
20:26 keimlink joined #salt
20:27 Ryan_Lane iggy: I'm pretty sure this one in the PR was "should this be against 2015.8, or should we go into develop"
20:27 breakingmatter joined #salt
20:27 Akhter joined #salt
20:27 Ryan_Lane the answer always seems to be it should be in develop
20:27 Ryan_Lane of course, I pull boto modules from develop, so I'm just asking for problems
20:28 pravka joined #salt
20:34 woodtablet zmalone you are both the #salt and #chef channels. do you work for either company or do just use both ?
20:34 zmalone I use both.
20:36 woodtablet zmalone - you are the perfect person then to as this next question. if i was writing a new formula, what do you use to test the formula to make sure it worked ? (similar to the "unit-tests" in test-kitchen)
20:36 kinetic joined #salt
20:37 moogyver danlsgiga: you could always just use any minion to send an event back to the reactor to fire off another orchestrate
20:37 moogyver woodtablet: serverspec
20:37 zmalone I love test kitchen / chefdk.  https://github.com/simonmcc/kitchen-salt exists, but I've been told that you'll run into issues with it right now by a coworker who tried it out this week.
20:38 zmalone serverspec is platform agnostic though.
20:38 woodtablet moogyer - ohh it works for salt too ?
20:38 woodtablet ohhh awesome
20:38 woodtablet thanks
20:38 moogyver serverspec doesn't care about the cfg mgt platform
20:38 CeBe joined #salt
20:38 Arbusto joined #salt
20:39 |Fiber^| joined #salt
20:39 danlsgiga moogyver: Ugly workaround in my opinion
20:39 Akhter joined #salt
20:39 whytewolf danlsgiga: the logic is not so cut and dry about runners. yes runners run on the master. but module.runners is an exacution module that is run through a minion. all of orchs built in functions focus on running either states. or exacution modules. [IE not runners] it is meant to orchestrate minions.
20:40 moogyver danlsgiga: I suppose?  you can also use  something like: event = salt.utils.event.get_master_event(__opts__, __opts__["sock_dir"], listen=False)
20:40 moogyver and just use that to fire the event
20:41 moogyver in the runner, anyways
20:41 moogyver we do our orchestration w/out orchestrate and in a runner.
20:41 moogyver the orchestrate runner is too limited
20:41 whytewolf orchestrate is a runner.
20:41 moogyver ( for our purposes )
20:41 moogyver whytewolf: i meant our own custom runner.
20:42 whytewolf yeah figured that out right as I hit enter :P
20:42 moogyver :)
20:43 whytewolf I think the food is helping me be less grouchy. I do appologize for the grufness of my answers danlsgiga
20:44 moogyver danlsgiga: custom runner functions run modules on minions, then the functions send 'success' events back to reactor with a payload.  reactor parses payload to get next step and executes the next runner function.
20:44 moogyver helps keep it modular as well, we can restart the orchestrate at any point by just sending the event we need.
20:44 gekitsuu joined #salt
20:45 aidalgol` joined #salt
20:46 Arbusto joined #salt
20:47 whytewolf danlsgiga: I think moogyver solution might be a better choice for you. as orchestrate does not seem to fit what you are actually trying to do. orchestrates goal is 100% focused on minion control. what you might want to look at building is a new runner that is a master-level orchestration.
20:47 danlsgiga moogyver: I understand that, my first attempt to do what I want was exactly using events and reactors back and forth, but reactors doesn't use the requisites functions that I need to make sure the command executed successfully
20:48 aidalgol` joined #salt
20:48 moogyver danlsgiga: you do that logic in the runner.
20:48 moogyver the reactor only ever fires off a runner event.  the runner starts the module on the minion, gets results and makes sure things went ok
20:49 moogyver then it sends the 'success' event back to the reactor
20:49 subsignal joined #salt
20:50 danlsgiga moogyver: Well, for that I don't actually need a runner, executing a state should work and worked in my previous implementation... what I want is something more "elegant" and orchestrate provided me the requisite level of control
20:50 lisa2lisa joined #salt
20:51 kinetic joined #salt
20:52 moogyver danlsgiga: except that orchestrate is a runner.. so making your own runner is really just making a better orchestrate, no? :)
20:53 danlsgiga moogyver: Also I'm thinking on performance... Using the event / reactor ping pong I was basically exchanging almost 15 events between the master and the minion... using the orchestrate runner I reduced that to 5 + I got the requisites working
20:54 mapu joined #salt
20:55 danlsgiga moogyver: That makes sense, the thing is, I don't really need to have my own runner if orchestrate allowed to run another orchestrate from inside itself, which in my opinion makes sense for a more modular approach
20:55 moogyver danlsgiga: sure, there's alot of messaging to handle it.  we couldn't use orchestrate because requisites don't work on execution modules that well.
20:55 keimlink joined #salt
20:56 danlsgiga moogyver: It depends on the execution module though... I found a bug in the cmd.run module that does not return failure if the command fails, but using cmd.run_all it works as expected
20:56 scoates joined #salt
20:56 danlsgiga moogyver: So far, I've not found any requisites issues from the execution modules I'm using, except the cmd.run one
20:57 zmalone The return codes are all messed up in salt
20:57 zmalone there are some long running tickets for that
20:57 zmalone https://github.com/saltstack/salt/issues/18510 for example
20:57 saltstackbot [#18510]title: salt exit codes | is there a reason why test.* returns False and exits with bash exit code 0 instead of 1 ? ...
20:57 sunkist joined #salt
20:58 zmalone Unfortunately, the best thing to do is to just read the full output
20:58 danlsgiga zmalone: test.ping works just fine for me :)
20:58 danlsgiga zmalone: inside an orchestrator
20:58 moogyver really? cuz test.false doesn't
20:58 moogyver I would expect that to fail for the next step, but it won't
20:59 danlsgiga zmalone: I target multiple servers to validate my cluster is responsive before doing any actions... if any of the nodes fails then I abort the orchestrate
20:59 perfectsine joined #salt
21:00 Akhter joined #salt
21:00 kinetic joined #salt
21:00 ajw0100 joined #salt
21:01 danlsgiga moogyver: I have a cluster of 8 servers and I don't want to apply any changes if any of them are not online, or I could get a split brain situation
21:01 moogyver danlsgiga: no, I get that.  I'm just saying not all the functions return correct ( for example, test.false )
21:01 moogyver danlsgiga: I'd expect that to return a false result and cause a 'require' to fail, but it doesn't.
21:03 danlsgiga moogyver: But afaik the test execution module is not maintaned for a long time... I saw somewhere that this was one of the first execution modules written and it was used for early debbuging purposes
21:05 moogyver danlsgiga: sure, that was just an example, but one I had tested orch's with.  there were some other orch modules that we tried as well that did the same thing.  as zmalone said, it's just not consistent
21:07 moogyver danlsgiga: and it's hard to do good branching logic in orchestrate or other control flows.  it's less feature rich than doing a pure runner.
21:07 moogyver danlsgiga: so, which is why you either have to enhance the orchestrate runner or write your own :)
21:08 danlsgiga moogyver: My "solution" for now is call a state.apply from the orchestrate and the state will then call another orchestrate
21:09 danlsgiga moogyver: I'll take a deeper look in the pure runner approach
21:10 danlsgiga But having the ability to call another orchestrate from inside an orchestrate would be awesome
21:14 kinetic joined #salt
21:17 wryfi should `salt-run cache.clear_all` normally return False?
21:20 kinetic joined #salt
21:23 breakingmatter joined #salt
21:23 jaybocc2 joined #salt
21:23 jaybocc2 ehlo
21:23 Aleks3Y joined #salt
21:23 jaybocc2 can anyone explain to me why when i run salt-run jobs.lookup_jid it only returns the output for one minion?
21:24 jaybocc2 instead of all minions for that job?
21:25 Akhter joined #salt
21:27 RandyT I've done considerable hair tearing over this today... https://gist.github.com/rterbush/97efd9356ac498677fc8
21:27 RandyT One last try here to see if anyone has any thoughts as to how to further debug this before reporting it as an issue.
21:28 polishdub left #salt
21:29 kinetic joined #salt
21:30 stanchan joined #salt
21:34 pcn jaybocc2: have you searched the event cache to see if all minions have the same jid?
21:37 teryx510 joined #salt
21:39 moogyver wryfi: 'salt-run cache.clear_all \*'
21:47 kinetic joined #salt
21:54 kinetic joined #salt
21:54 KyleG joined #salt
21:54 KyleG joined #salt
21:55 pm90_ joined #salt
21:57 pm90__ joined #salt
21:58 wryfi moogyver: ty
22:10 RandyT FWIW: reverted back to deploying a Windows_Server-2008 image and the problem goes away. I was trying to deploy a Windows_Server-2012 which appears to not work as I have described.
22:12 shawnbutts anyone else seeing lots of "Bad load from minion" after upgrading to 2015.8.0 from the new repo?
22:13 sunkist joined #salt
22:25 cberndt joined #salt
22:33 aparsons_ joined #salt
22:35 irctc041 joined #salt
22:36 pm90_ joined #salt
22:38 pravka joined #salt
22:40 pcn shawnbutts: https://github.com/saltstack/salt/issues/27856 ?
22:40 saltstackbot [#27856]title: Master stop responding when I've specifics symlinks | I've created a symlink pointing to itself in order to [bypass this hardcoded `salt://` restriction](https://github.com/saltstack-formulas/bind-formula/issues/37):...
22:43 moogyver what's the difference between extension_modules and module_dirs?  one takes a string, one takes a list - why not just have module_dirs and be done with it?
22:43 moogyver not sure i understand the point of both
22:44 mike2512 guys in a STATE ... to use mysql credentials should i use alternative.mysql.host etc ?
22:44 jeffspeff joined #salt
23:01 otter768 joined #salt
23:05 nzero joined #salt
23:06 sunkist joined #salt
23:19 danlsgiga hey guys... if I send an orchestrate run starts and try to run an execution module in a minion and this minion is already running another job (like a highstate), will it abort the execution or queue the job?
23:24 fusionx86 joined #salt
23:29 mike2512 is anyone around ? :)
23:33 wryfi since upgrading to 2015.8.0, my external pillar data is suddenly showing up twice
23:33 wryfi duplicates
23:33 wryfi why would that have happened?
23:43 breakingmatter joined #salt
23:44 otter768 joined #salt
23:45 fusionx86 joined #salt
23:46 mike2512 wryfi:  not sure.. i just installed .8
23:46 mike2512 is anyone around who can help me with return mysql? i have an issue... might be me being an idiot
23:47 iggy moogyver: extension_modules predates module_dirs
23:48 moogyver so should use module_dirs?
23:48 iggy probably more future proof, yeah
23:48 thalleralexander joined #salt
23:48 iggy although there's been no talk (afaik) of deprecating extension_modules
23:49 moogyver iggy: and if i want modules sync'd to minions, I still have to use /srv/salt/_modules?
23:49 iggy yeah
23:49 iggy the only place I've ever used extension_modules was on a master
23:50 wryfi oh, i wonder if this is relevant to my problem
23:50 iggy because that's the only way custom modules/states/etc work in master context
23:50 iggy (i.e. runners, reactor, etc)
23:52 Ssquidly joined #salt
23:53 mike2512 can anyone take a short look at https://dpaste.de/akR8
23:55 moogyver iggy: seems like too many ways to specify modules :) module_dirs, extension_modules, runner_dirs, etc.

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