Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2015-05-27

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

All times shown according to UTC.

Time Nick Message
00:00 iggy and sadly the fileserver update schedule is tied to the main timer (that runs scheduler and everything else internal to salt)
00:00 RabidCicada ahh....Interesting
00:00 iggy so if you change that, you risk other bits of salt not running as expected
00:01 bfoxwell joined #salt
00:02 RabidCicada yeah...I only had 3 gitfs remotes
00:05 RabidCicada ahhh...I was thinking about gitfs as the salt states provider anyway...not the pillar provider
00:05 RabidCicada *or whatever terminology is correct*
00:08 baweaver joined #salt
00:08 RabidCicada I see...loop_interval to update multiple things.  Decoupling those would seem to be good.
00:11 glassx- joined #salt
00:11 glassx- left #salt
00:12 iggy if they did that, I'd probaby stick with it longer
00:12 iggy but at a certain point, if you can do everything event driven, might as well
00:12 RabidCicada Absolutely agreed iggy
00:13 RabidCicada Is anyone actually working on bringing ext_pillar git support to parity with gitfs?  Where might I read up about current work on that task if it exists?
00:13 iggy no
00:13 iggy it gets mentioned in bugs a lot, but it's a pipe dream right now afaik
00:13 supersheep joined #salt
00:13 RabidCicada I like how the docs say "and is not yet at feature parity with it" which gives the impression it is currently being worked on
00:14 RabidCicada That might be why....If it's not planned to be worked...how about that saying " and it wont be" :)
00:14 iggy *someday
00:14 RabidCicada haha:)
00:14 RabidCicada I certainly was under the impression that it was likely being worked
00:14 kg_ joined #salt
00:14 iggy they want to, it's a matter of manpower (specifically as it relates to the 2 people in the world that could probably actually pull it off)
00:15 RabidCicada yikes...it's that complicated?
00:15 iggy plus all the other features that people have requested in that area
00:15 iggy (i.e. pillar:// uris for file.managed/et.al.
00:16 RabidCicada hehe
00:18 RabidCicada oh...looking into the webhooks for fileserver.update...love the change for passing backends in 2015.5
00:18 RabidCicada now if we can further restrict that for the actual repo/tag
00:18 RabidCicada super specific updates based on events
00:19 Singularo joined #salt
00:23 iggy push to dev, master pulls, runs unit tests in dev env, sends results
00:23 iggy something like that anyway
00:31 cpowell joined #salt
00:31 troyready joined #salt
00:37 MatthewsFace joined #salt
00:39 otter768 joined #salt
00:40 mgw joined #salt
00:43 KyleG <3 clean: True
00:43 KyleG life changing
00:44 murrdoc does it work with listen_in yet ?
00:44 murrdoc if not fuck it i still hate it
00:44 murrdoc and am happy it works for u tho :)
00:47 mosen ashamed to admit i dont know all the requisites
00:50 subsignal joined #salt
00:51 murrdoc KyleG:  were u working with file.directory and how did u get it to work
00:51 murrdoc mosen:  yeah they keep adding to it
00:52 KyleG murrdoc: I have an nginx/includes dir with all of our sub configs, and my state entry jsut looks like this: https://gist.github.com/kylegato/d7d4444290277dc8330b
00:52 murrdoc ahyes
00:52 murrdoc nice
00:52 mosen murrdoc: in your template i noticed you always load defaults using load_yaml i think.. is that the standard modern style?
00:52 BgBee joined #salt
00:52 iggy standardish
00:52 murrdoc the template formula mosen is my understanding of the standard
00:53 iggy guess i should look at some point
00:53 mosen murrdoc: all good.. jsut recognised i was doing some things the old way
00:54 mosen Also i have some pillars accessed through templates.. but iggy you were mentioning that limiting pillar usage to states was a better/more manageable way to operate
00:57 iggy i'm not sure of what you speak
00:58 mosen iggy: is it bad practice to use pillars directly from templates?
00:58 iggy i prefer it
00:58 murrdoc i hate it
00:59 murrdoc i look at salt code as mvc pattern tho
00:59 murrdoc so thats the diff
00:59 iggy well, i usually have map.jinja that combines pillar,defaults,grain lookups, etc
00:59 iggy then impprt that into the template
01:00 murrdoc yeah that i approve
01:00 mosen i couldnt figure out how to supply a whole pillar dict as a context in the sls.. i saw some examples using var|yaml filter
01:01 iggy jinja caches stuff, so its not a performance hit to import the map in multiple files
01:04 druonysuse joined #salt
01:04 druonysuse joined #salt
01:12 johngrasty joined #salt
01:13 xMopxShell joined #salt
01:14 julez joined #salt
01:14 Rob joined #salt
01:19 iggy looking back at some states I wrote a little over a year ago
01:19 iggy should probably just scorched earth them and start over...
01:20 DammitJim joined #salt
01:20 DammitJim does one normally create a "file" repository with configuration files for the different servers?
01:20 DammitJim * cpowell has quit (Remote host closed the connection)
01:21 DammitJim say I am configuring a samba server... do I create a formula to change the smb.conf file or do I copy a new smb.conf file to the minion?
01:22 mosen typically people write states to control each individual variable within the smb.conf, but you can also just manage the whole file
01:23 david_an11 joined #salt
01:23 iggy yeah, it depends how much of pita it is to do individual edits
01:23 iggy or if the software in question supports .d dirs
01:23 mosen yeah
01:23 cpowell joined #salt
01:26 DammitJim interesting
01:26 DammitJim the smb.conf was just an example
01:27 DammitJim I'm viewing salt as a tool to #1) do configuration management and #2) be able to rapidly deploy new servers
01:27 DammitJim are any of my assumptions wrong?
01:29 aurynn well, depends on how rapid "rapid" is
01:29 aurynn but yes, overall, that's what it's for
01:30 DammitJim so, for example, I'd have to at least set the hostname and IP address of the new server and then install salt-minion
01:30 aurynn and by "rapid" I mean, sufficiently fast values of rapid get into having frozen AMIs that you pre-provision, or warm spares, or what-have-you
01:30 DammitJim accept the minion on the master
01:30 aurynn you can use pre-generated keys for that stage, if you need additional rapidity
01:31 DammitJim then have a group of things that need to be installed on the server, so like if I need to configure tomcat 6, I'll have to install java first
01:31 aurynn (with a security cost, naturally)
01:31 DammitJim sure... I think I can deal with the first piece being manual... the rest of the steps are the ones that make me nervous that we aren't doing consistently as a team
01:31 aurynn right. So yes, you'd declare that tomcat needs java via the state requirements
01:32 aurynn and that your app needs tomcat
01:33 DammitJim I have done a tutorial where vim gets installed
01:33 DammitJim how hard is it to add a new repo to install sun java for example?
01:33 aurynn depends on your OS
01:34 aurynn pkgrepo handles debian and rhel-like: http://docs.saltstack.com/en/latest/ref/states/all/salt.states.pkgrepo.html#module-salt.states.pkgrepo
01:34 DammitJim ubuntu in this case
01:34 aurynn others are a bit harder
01:34 DammitJim oh ok, so debian is very similar
01:35 aurynn ubuntu is a debian derivative, so yes :)
01:35 DammitJim so, what do I do with the link you posted?
01:35 aurynn okay, so you know how you wrote that bit for vim which was like
01:35 aurynn vim: pkg.installed
01:35 aurynn ?
01:35 DammitJim yeah
01:35 DammitJim very short... like 3 lines
01:36 aurynn right, so that used http://docs.saltstack.com/en/latest/ref/states/all/salt.states.pkg.html#module-salt.states.pkg to do the work
01:36 aurynn pkgrepo will work the same way
01:36 DammitJim so, I need to copy and paste the code?
01:37 aurynn so it'll be something like `mycustomrepo: pkgrepo.managed: - humanname: etc`
01:37 aurynn you can copy and modify the code, yes
01:38 DammitJim gotcha... .I just hear so much about git, I thought you checked out code or something
01:38 aurynn there's a bunch of contributed formulas, and there's some advanced magic for using gitfs to back the salt master
01:38 voxxit joined #salt
01:39 hasues joined #salt
01:39 beauby joined #salt
01:39 hasues left #salt
01:39 aurynn but it sounds like you're still in the learning stages, so while those will be interesting, it doesn't solve your immediate problem :)
01:39 DammitJim gotcha
01:40 DammitJim yeah, my first step is to do the java install
01:40 DammitJim so, I guess I'll mess around with the link you sent me
01:40 DammitJim thanks aurynn
01:40 DammitJim what distro are you on?
01:40 DammitJim it was weird that I had to create the /srv/salt directory to create the sls files
01:40 aurynn I use Ubuntu 14.04
01:41 aurynn mostly for testing I have a vagrant config with a master & minion
01:41 DammitJim vagrant? isn't that the competition?
01:41 djpain joined #salt
01:45 c10 joined #salt
01:45 aurynn vagrant's a dev tool
01:45 aurynn It lets you manage on-machine VMs easily, and provision them :)
01:45 DammitJim oh, this is cool! I can check for deployed war files in tomcat!!!
01:45 DammitJim on-machine vms?
01:45 DammitJim not sure what you mean
01:47 Singularo joined #salt
01:48 danjj iggy: still around?
01:49 aurynn DammitJim, your workstation,
01:49 aurynn vs. production servers
01:49 DammitJim hhhmmm
01:50 iggy danjj: sort of
01:50 aurynn DammitJim, my workflow is: Get the salt states I want to test, set up 2 vms on my workstation (one master, one minion) and run the states on that minion VM
01:50 danjj is this functionality in the users formula is expected to work: ssh_auth_sources:
01:50 danjj - salt://users/files/danjj.id_rsa.pub
01:50 aurynn then I can run tests on it
01:50 aurynn like "is there a running tomcat?"
01:50 danjj I *think* I have it set up right, but I just get some cryptic tracebacks
01:51 aurynn "Does it serve on port 80, and not stacktrace messily?"
01:51 DammitJim oh
01:52 iggy danjj: no clue, never used the users-formula, try gist'ing the tracebacks
01:52 mgw joined #salt
01:52 aurynn DammitJim, IMO it's essential to have VMs for this, since they can be destroyed and re-created trivially
01:52 danjj heh, I pinged you since I thought I saw you were a committer on it
01:52 iggy I merged someone else's commit
01:52 danjj gotcha
01:52 DammitJim aurynn, I agree... that's what I am working with (but I have snapshots)
01:53 aurynn DammitJim, snapshots are also good
01:53 iggy but go ahead and gist the errors, we might still have some ideas
01:53 aurynn I get nervous about not having a dedicated test env
01:53 DammitJim but I am thinking more long term as we switch from RHEL to Ubuntu
01:53 DammitJim I'm going to have to set up a lot of servers with different requirements
01:53 danjj iggy: I see a lot comments around getting pillar to pull salt://, most of it with issues
01:53 danjj I'll check the tracebacks and see what I get
01:54 DammitJim I mean, how would you go about editing your ntp.conf to disable all ubuntu time servers and add your own?
01:54 DammitJim do the edit on the fly or replace the ntp.conf file?
01:54 aurynn I'd replace the ntp.conf and reload it
01:54 murrdoc is there a pillar ?
01:54 murrdoc what formula are u using
01:54 aurynn if you *need* to edit the ntp.conf inline, I'd use the `augeas` state.
01:55 DammitJim aurynn, what would I use to replace the file?
01:55 danjj murrdoc: was that to me?
01:55 aurynn DammitJim, http://docs.saltstack.com/en/latest/ref/states/all/salt.states.file.html#module-salt.states.file
01:55 murrdoc danjj:  no i thought the ntp stuff was a formula
01:55 aurynn and you'd use the source: to point to something in your Salt state tree, the base fine
01:55 murrdoc whats your issue ?
01:55 aurynn er, base file
01:56 DammitJim holy cow, so this is what I read where one even sets the permissions on the file and everything!
01:56 danjj murrdoc: I'm looking at the users-formula
01:56 murrdoc k wats the issue
01:56 aurynn DammitJim, http://docs.saltstack.com/en/latest/ref/states/all/ is the best page to keep open :)
01:56 danjj ssh_auth_sources:
01:56 danjj - salt://users/files/danjj.id_rsa.pub
01:56 danjj results in a traceback
01:56 aurynn it's all the stuff Salt ships with
01:57 danjj I think I've got the pillar fs set up properly
01:57 DammitJim these are built in, right? I don't have to download anything?
01:57 tiadobatima murrdoc, iggy... I might have found the problem with docker-formula:
01:57 tiadobatima https://github.com/saltstack/salt/blob/develop/salt/modules/upstart.py#L413
01:57 aurynn DammitJim, this is all built-in
01:57 danjj is there an easy way to check if the file is avaliable to the pillar/state?
01:58 DammitJim the first link didn't tell me how to just copy a file, right? it seems to be talking about templates instead
01:59 tiadobatima docker installs an upstart script as well... but the upstart module seems to handle only sysv?
01:59 druonysus joined #salt
02:00 aurynn DammitJim, You don't need the template stuff, just the source file
02:00 aurynn the template stuff is if you're modifying the file on a per-machine basis
02:00 DammitJim where do I put the source file?
02:01 murrdoc where ever u want
02:01 murrdoc under the file_roots
02:01 aurynn what @murrdoc said
02:01 murrdoc then use salt://(ls file path)
02:01 murrdoc basically salt:// means cd $file_root
02:02 DammitJim oh really?
02:02 DammitJim does one ever have to set up a database to hold all this stuff?
02:02 tiadobatima when I look at start() and stop(), restart() in upstart.py it looks like it's all sysv... am I crazy?
02:02 DammitJim or check in the file_roots to a local git or svn?
02:02 danjj so, does pillar_roots not work?
02:02 aurynn DammitJim, yes, that's where gitfs becomes a thing
02:02 danjj or is it not needed?
02:02 DammitJim ah, gotcha
02:03 DammitJim man, but git, not svn, huh?
02:03 DammitJim (I requested the repo guy to set me up an svn project
02:03 aurynn Fortunately, git-svn is a thing
02:03 DammitJim oh wow
02:03 DammitJim awesome
02:03 aurynn or it was a thing, long ago
02:03 aurynn hopefully it remains a thing
02:05 danjj murrdoc: I went with pillar_roots since this file is getting called from a pillar.sls, otherwise what is the point of the pillar_roots directives in the master?
02:05 murrdoc point is to tell the salt master where the file based pillars are
02:06 murrdoc similar to how file_roots tell salt mster where the state config files are sitting
02:06 DammitJim can one state file call another state file?
02:06 danjj okay, but salt:// only points to file_roots and there is no pillar://
02:07 robawt DammitJim: you can include them yes
02:07 DammitJim I mean, I want to set up the sls for installing sun java
02:07 DammitJim then another sls to tell it to install tomcat with the required sun java jdk
02:07 robawt if tomcat's sls requires java's key to install java pacakge, and it included java in tomcat you'd be done
02:08 danjj and I can call salt:// from a pillar
02:08 beauby joined #salt
02:09 DammitJim robawt, even if I need to add a ppa?
02:10 DammitJim hhhmmmm how do I know the keyid?
02:10 robawt the identifier for the salt state == key, sorry
02:10 murrdoc dont worry about
02:10 robawt it all rolls into a dictionary so i think of them as keys
02:10 murrdoc the ppa state uses apt-add-repository
02:10 murrdoc that adds teh key
02:11 robawt DammitJim: include the PPA in the java state, require the ppa to install java, require java to install tomcat, you're done
02:11 DammitJim no, I mean, this page: http://docs.saltstack.com/en/latest/ref/states/all/salt.states.pkg.html has a keyid line
02:11 DammitJim robawt, so, 2 sls files?
02:11 robawt DammitJim: it's up to you, that's what I would do for reusability, you can do what you wish
02:11 DammitJim oh ok
02:12 DammitJim how do I make the minion run apt-get update after I get it to add the ppa?
02:12 robawt i believe it's default behavior of pkg.installed but i would double check
02:13 robawt you really only need to reindex after the PPA is installed, right?
02:13 DammitJim yes
02:14 danjj murrdoc: I think I'll inline the keys instead so we're not addiing files for each new user under the salt/ side. Thanks for the clarification.
02:14 murrdoc eugh
02:14 murrdoc but aight
02:15 DammitJim does this look good? http://pastie.org/10209165
02:15 danjj murrdoc: you'd go the other way?
02:15 DammitJim this is to install java jdk from sun
02:16 murrdoc danjj:  i have now hit 11 hours
02:16 murrdoc i am going to stop working
02:16 DammitJim actually, it looks like it is the jre
02:16 danjj heh, good plan
02:16 DammitJim oh yeah, go close your eyes
02:16 murrdoc if the users-formula documnetaion stinks, open an issue saying , you guys stink, i need better docs, heres waht i was trying to do
02:16 murrdoc and we ll go get someone to fix it
02:16 murrdoc g night
02:17 DammitJim can someone look at my pastie for installing java from ppa?
02:17 ALLmight_ joined #salt
02:17 robawt not sure about line 1 DammitJim
02:17 DammitJim oh?
02:17 DammitJim base: ?
02:17 danjj 'night
02:17 robawt and I would add a require: - pkgrepo: blehblerh at the end
02:18 robawt yeah usually base is reserved for the top.sls file
02:19 DammitJim what are my other options besides base?
02:19 robawt none
02:19 robawt get rid of it
02:19 DammitJim oh
02:19 robawt you sound kinda new to salt, i would look at some of the example salt states from the tutorial
02:20 robawt http://docs.saltstack.com/en/latest/topics/tutorials/starting_states.html
02:20 DammitJim I have the require_in
02:20 aurynn require_in injects the requirement into another state
02:20 robawt aurynn++
02:21 DammitJim oh, so I'm doing it backwards?
02:21 DammitJim I had read that starting_states tutorial
02:21 DammitJim I have gotten a little confused by trying to add the ppa
02:21 DammitJim by following their example
02:22 DammitJim http://docs.saltstack.com/en/latest/ref/states/all/salt.states.pkg.html
02:22 DammitJim which says that there is a require_in and cannot be replaced with a require at the end
02:22 aurynn `x: require: y\ny: etc` and 'x: etc\n y: require_in: x` are equivalent WRT the resolution graph
02:23 DammitJim WHAT???
02:24 robawt you pick one
02:24 robawt you either require or require_in, that's it
02:25 DammitJim oh ok, I left it with require_in
02:25 robawt then you're good
02:25 DammitJim I got confused with your comment <robawt> and I would add a require: - pkgrepo: blehblerh at the end
02:25 DammitJim or was that for something else?
02:27 brianfeister joined #salt
02:28 robawt you're ok
02:28 robawt you should always use the requisite system, and you're doing it
02:28 DammitJim si, it worked!
02:29 beauby joined #salt
02:29 robawt awesome, now you gotta pay forward the good info to the next user
02:29 DammitJim :D
02:30 DammitJim so much to learn
02:30 DammitJim *sigh*
02:33 aurynn very much :) but it's worth it
02:35 DammitJim what could be making my system slow?
02:36 DammitJim like running this sls is taking at least 40 seconds to just check on one server
02:36 evle joined #salt
02:36 DammitJim oh wow... the server that doesn't have the package returned False for the installed piece
02:37 DammitJim http://pastie.org/10209181
02:37 DammitJim is that because I didn't update before the install?
02:40 otter768 joined #salt
02:40 DammitJim ugh, one has to accept the oracle binary thing
02:41 DammitJim I wonder if there is a way to automatically accept it
02:41 amcorreia http://pastebin.com/1geAQi77
02:41 amcorreia anyone can help why i this error?
02:46 favadi joined #salt
02:51 iggy amcorreia: is that the entirety of your master.conf?
02:54 dalexander joined #salt
03:00 amcorreia iggy, no, just what i've changed from default
03:01 amcorreia I almost get working if I specify in command line 'saltenv=dev'
03:02 amcorreia http://pastebin.com/qQucpRTh
03:03 refnode_ joined #salt
03:05 devops_chick joined #salt
03:06 devops_chick Hey there!
03:07 aurynn hi
03:08 devops_chick Did anyone see the hacker news post about compute engine's preemptive VM's last week? https://news.ycombinator.com/item?id=9564122
03:09 devops_chick Wondering if this is able to be done by salt-cloud, or if it would take a PR. Happy to help if the latter
03:10 aurynn as with the commenters: I would rather not put things on Google, because they will probably go away
03:10 beauby joined #salt
03:12 devops_chick yeah I did see some people aren't too optimistic (and rightly so) about something like this being deprecated without notice.
03:13 devops_chick Though if this DOES intend to be a viable option for a bit, it could be a huge cost saver and be much more practical to use with salt.
03:14 devops_chick I think they're intending to compete with EC2's spot instances with preemptive offerings
03:15 aCodinMan joined #salt
03:21 amcorreia any idea what is the problem? i'm trying to set up 2 environments 'dev' & 'prod'
03:22 amcorreia I try to put a 'top.sls' inside envs but didn't work too
03:25 writtenoff joined #salt
03:26 iggy it's feature parity with aws, I wouldn't figure they'd get rid of it
03:27 dimeshake joined #salt
03:31 c10 joined #salt
03:41 devops_chick @amcorreia, did you update your master config with updated values for dev and prod (usually /etc/salt/master); and remember to restart the service after
03:44 devops_chick what error are you getting when running a highstate?
03:44 amcorreia devops_chick, yes, i restarted master, but i don't know why, daemon seems to be "stuck",
03:44 amcorreia i meant, restart didn't worked...so I kill process, and now seems to be working..
03:45 stanchan joined #salt
03:58 power8ce joined #salt
04:01 hal58th joined #salt
04:01 sunkist joined #salt
04:01 hal58th_ joined #salt
04:01 hal58th__ joined #salt
04:05 swa_work joined #salt
04:05 garthk joined #salt
04:12 aCodinMan joined #salt
04:13 Aidin joined #salt
04:18 garthk joined #salt
04:18 garthk joined #salt
04:31 tkharju joined #salt
04:36 hybridpollo left #salt
04:41 otter768 joined #salt
04:42 mgw joined #salt
04:42 ctrlrsf joined #salt
04:46 aCodinMan joined #salt
04:49 ramteid joined #salt
04:51 noway_ joined #salt
04:53 TyrfingMjolnir joined #salt
05:01 vexati0n joined #salt
05:08 ndrei joined #salt
05:09 ajw0100 joined #salt
05:13 ndrei joined #salt
05:13 dimeshake joined #salt
05:17 joeto joined #salt
05:20 c10 joined #salt
05:23 writtenoff joined #salt
05:23 aravind joined #salt
05:24 catpiggest joined #salt
05:27 Aidin joined #salt
05:30 ndrei joined #salt
05:31 scoates joined #salt
05:36 ndrei joined #salt
05:39 subsignal joined #salt
05:48 losh joined #salt
05:56 impi joined #salt
05:58 dopesong joined #salt
06:01 colttt joined #salt
06:06 stoogenmeyer__ joined #salt
06:06 mosen joined #salt
06:13 refnode_ joined #salt
06:13 Nazzy joined #salt
06:21 flyboy joined #salt
06:22 ndrei joined #salt
06:25 wendall911 left #salt
06:29 oldblade joined #salt
06:30 oldblade help!!
06:30 Aidin ?
06:31 KermitTheFragger joined #salt
06:31 oldblade can you create private chat rooms on here?
06:32 __number5__ nope
06:35 honestly joined #salt
06:37 rdas joined #salt
06:38 dimeshake joined #salt
06:39 dopesong joined #salt
06:39 ndrei joined #salt
06:40 subsignal joined #salt
06:41 mrbigglesworth joined #salt
06:42 otter768 joined #salt
06:44 soren_ joined #salt
06:46 ndrei joined #salt
06:46 c10 joined #salt
06:50 kawa2014 joined #salt
06:53 malinoff joined #salt
07:00 c10 joined #salt
07:01 stephanbuys joined #salt
07:03 tmclaugh[work]_ joined #salt
07:05 markm joined #salt
07:06 CeBe joined #salt
07:12 bfoxwell joined #salt
07:12 bastiaan joined #salt
07:12 grrrrr on irc?
07:13 grrrrr just join whatever channel you like #myownprivatechat
07:13 grrrrr blah gone already
07:16 fredvd joined #salt
07:23 eseyman joined #salt
07:34 julez joined #salt
07:43 c10 joined #salt
07:44 Auroch joined #salt
07:46 ninkotech joined #salt
07:51 MatthewsFace joined #salt
07:51 jhauser joined #salt
07:57 al joined #salt
08:02 s_kunk joined #salt
08:02 s_kunk joined #salt
08:02 julez joined #salt
08:02 refnode_ joined #salt
08:03 losh joined #salt
08:05 Grokzen joined #salt
08:06 impi joined #salt
08:16 viq joined #salt
08:17 julez joined #salt
08:18 julez joined #salt
08:19 bastiaan_ joined #salt
08:19 linjan joined #salt
08:31 slav0nic joined #salt
08:35 c10_ joined #salt
08:43 otter768 joined #salt
08:43 supersheep joined #salt
08:44 refnode joined #salt
08:47 mariusv joined #salt
08:47 mariusv joined #salt
08:49 mariusv joined #salt
08:49 __mel joined #salt
08:53 CeBe1 joined #salt
09:08 denys joined #salt
09:11 Auroch joined #salt
09:23 denys joined #salt
09:23 keimlink joined #salt
09:26 N-Mi joined #salt
09:31 huddy joined #salt
09:36 supersheep joined #salt
09:39 leev_ joined #salt
09:43 subsignal joined #salt
09:43 chiui joined #salt
09:45 ksj is there a way to set a default parameter for all my state modules - i.e. I have lots of 'pkg.installed' statements spread over many sls files. I'd like all of them to default to '- install_recommends: False', without having to write that explicitly
09:46 laax joined #salt
09:50 froztbyte sigh why does http://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.gpg.html#module-salt.modules.gpg exist but a state not
09:50 Edgan joined #salt
09:51 froztbyte guess I have to write that
09:51 mage_ how can I use python modules in salt jinja templates ?
09:51 mage_ for ex I need to use os.path.join
09:54 cberndt joined #salt
09:54 cberndt joined #salt
10:12 PI-Lloyd joined #salt
10:13 supershe_ joined #salt
10:15 VSpike ksj: you might be able to use a macro https://salt.readthedocs.org/en/v2014.1.0rc1/ref/renderers/all/salt.renderers.jinja.html#macros
10:15 VSpike just looking into those myself at the moment and though it might be relevant to you too
10:17 VSpike ksj: or for a more up-to-date link http://docs.saltstack.com/en/develop/ref/renderers/all/salt.renderers.jinja.html#macros
10:18 _mel_ Hi. i have two virtual machines on XEN. one is PVM and the other is HVM. both have grains virtual: xen and virtual_subtype: Xen PV DomU. should'nt be it "Xen HV DomU" for the one system?
10:20 supersheep joined #salt
10:22 _mel_ i found a difference, that hvm domu has a grain called productname, which pvm domU hasn't. but that is not very good imho.
10:23 ksj VSpike: thanks, but they seem kind of messy. If I can't explicitly set a default then I'll just go with the extra verbosity. Much easier for others to read - which is very important for me
10:24 impi joined #salt
10:31 ventris joined #salt
10:34 wnkz joined #salt
10:35 Berty_ joined #salt
10:36 zer0def hum, is there a way within Salt to manage a postgres user's priviledges to a particular db?
10:36 zer0def other than owner
10:36 julez joined #salt
10:37 zer0def there was a github ticket #6558 that was supposed to target 2014.7, but the docu is sparse about that
10:44 otter768 joined #salt
10:46 ksj joined #salt
10:48 giantlock joined #salt
10:49 markm joined #salt
10:52 ksj is there a way to append values to a list in jinja? I want to do something like this http://dpaste.com/1W8X18P.txt
10:52 babilen zer0def: https://docs.saltstack.com/en/latest/ref/states/all/salt.states.postgres_user.html is not working for you?
10:53 babilen ksj: Prefix it with "do" to read: {% do mylist.append(grain) %}
10:53 babilen And I wouldn't necessarily use grains for that as there isn't much of a need to store that information on the minion itself
10:54 ksj babilen: ....man, that's the second time I've been missing the magic "do" keyword
10:54 babilen (as opposed to on the master or in a database you access via external pillars)
10:54 ksj why wouldn't you store a role in grains?
10:54 babilen ksj: jinja's Python support is shite
10:54 ksj yes
10:55 ksj yes it is
10:55 babilen ksj: Because there is absolutely no need to distribute the information as to which host gets which state to the managed hosts themselves and grains are also inherently insecure. The latter might, however, not be a problem in your setup.
10:56 babilen I'd keep it in pillars or a database that is being accessed via external pillars
10:58 ksj database is too painful. grains seemed to be the most obvious way. I don't see any reason why it's insecure....it's "public" information that those servers have certain functions
10:58 ksj I wanted to be able to "tag" servers with different roles
10:58 greyrhin_ joined #salt
10:58 zer0def babilen: not sure how i'd use that to set permissions/priveledges, i could be dumb about it
10:58 babilen Sure, but I still don't see the advantage of storing that information *on* the minions
10:59 babilen (and, as I said, the security implications might not be relevant in your setup)
10:59 ksj babilen: me neither. I'd rather store it centrally and push it, which is what I was hoping to do. you can push grains
10:59 babilen But why?
10:59 lloesche joined #salt
11:00 babilen That simply means that you have to maintain two data sources and ensure that they are being kept in sync
11:00 zer0def babilen: am I missing something in the docu which should scream "SQL GRANT" in my face?
11:00 babilen (the one that decides which grains are pushed and the grains that are being set)
11:00 ksj well, because matching in the top file didn't feel right
11:00 supersheep joined #salt
11:00 ksj and everyone kept saying "roles should go in grains"
11:00 babilen ksj: How so? And why don't you match on pillars?
11:00 ksj so that's the way I did it
11:01 babilen I doubt that everyone said that
11:01 ksj but I'm open to changing it
11:01 ksj especially since pushing grains doesn't really work
11:01 c10 joined #salt
11:01 ThomasJ grains are inherently insecure too, as they can be manipulated on the minion end
11:02 babilen ksj: How are you targeting the grains at this moment? That is: How do you decide which minion gets which grains?
11:02 ksj the problem is we have a lot of servers - ~60. so if I match in the top file it could become very verbose
11:03 babilen You can generate the top file based on pillar data
11:03 ksj ThomasJ: TBH, I'd never thought about someone maliciously manipulating the grains file....but it's a good point
11:03 babilen (just loop over a minion/grain map or a grain/minion map)
11:03 bluenemo joined #salt
11:03 ksj babilen: I'm not. it's all in VMs and I'm doing it manually.
11:03 bluenemo joined #salt
11:03 babilen ksj: That is what I referred to with "inherently insecure" but you said that that isn't an issue
11:04 ksj when I set up the vm, I just copy a skeleton grains file across, then uncomment lines in it depending on what functinoalisty I want the maching to have
11:04 ksj babilen: sorry, I didn't understand what you meant
11:05 babilen ksj: The security aspect isn't really that important when you don't care that minions can change them to whatever value they want ... it is, however, something you have to keep in mind
11:05 ksj it does make sense now. if someone breaks into a machine, theoretically they could manipulate the grains....but the only thing they could do is create or delete another webnnode (at the minute)
11:05 ksj yeah, agreed
11:05 babilen I would suggest maintaining a mapping from minion id to roles (or roles to minion id) in pillars and to dynamically generate your top.sls from that data
11:05 ksj and I'd rather not use grains for lots of other reasons, but...I couldn't figure out a versatile way to "tag" machines in pillar
11:07 ThomasJ babilen: Do you have any reference resources relating to dynamically generating top.sls from pillar data?
11:07 linjan joined #salt
11:08 ksj babilen: actually, you reminded me, I actually did do that already, with the intention of dynamically pushing the grains files. maybe I should just skip the grains altogether and read info direct from that pillar file I created...but I'm pretty sure there were reasons I couldn't do that at the time
11:08 ThomasJ ksj: Keep in mind, you do not need to use a single pillar file either. Pillar merging is awesome
11:09 laax joined #salt
11:09 ksj I'd be interested in seeing any specific examples of how people organise this stuff in practice. it's something I've been tearing my hair out over
11:10 babilen ksj: https://www.refheap.com/101731
11:10 babilen (very very basic and untested example)
11:11 ksj thanks. one thing I really wanted to do was avoid any jinja (or logic at all) in pillar files. I want other people to be able to come and quickly edit them if necessar
11:12 babilen Or you match on pillars directly in your top.sls -- http://docs.saltstack.com/en/latest/topics/targeting/compound.html → I@
11:12 babilen ksj: That is *not* a pillar file, but an example the the states top.sls
11:12 evle1 joined #salt
11:12 ksj e.g., theoretically, if we need a new webnode for one particular site, they should be able to just change one line and have it all work
11:12 ksj babilen: excuse me, I was being dub
11:12 ksj dumb...
11:13 babilen ksj: You would let them edit the pillar/roles.sls file and add a new minion in there
11:13 babilen Let me find an actual example of such a top.sls file
11:13 ksj thanks
11:13 ksj ...I knew I was going to end up changing all this stuff around
11:21 babilen It's just that grains for this are great if they are being managed by something external already (e.g. your cloud platform supporting tagging), but if you simply want to tag them yourself you are facing the "how to assign grains to minions" problem. And, if you do that, you can just as well target states by that logic directly.
11:21 babilen Sorry, can't find a good example. There are some pretty decent examples of full feature dynamic top.sls files around ... you might want to write to salt-users and ask for advice
11:23 babilen I use grains primarily for information *about* the minion that needs to be kept up-to-date and changes "on its own" or to mark minions as having completed certain provisioning steps.
11:23 ksj right, so, that's what I was planning to do - i.e. push the grains from the master
11:25 babilen Which means that: 1. You use the same logic for pushing grains as you could have used for states and 2. You require multiple iterations of highstates for your minions to converge on their final state as you have to push grains first 3. Your minions might be susceptible to attacks
11:25 babilen Let me paste an example of a suitable pillar file that is easy to maintain for all minions
11:25 ksj yeah...I know it can be simplified....but there was a good reason for doing it this way. I'll have to check irc logs
11:26 ThomasJ ksj: Tbh, unless it's information that has to be generated on the minion, it really shouldnt be done using grains
11:27 ThomasJ I use grains for obtaining os information, and setting provisioning steps
11:27 c10_ joined #salt
11:27 ThomasJ Everything else is kept in pillars as there is no reason to place them in grains in the first place, especially if you push from the master
11:27 amcorreia joined #salt
11:28 ThomasJ pushing from the master means 1) The information is already on the master in the first place, so it might just as well be in a pillar instead of grain 2) It cannot be manipulated 3) It's available on the first highstate
11:28 michelangelo joined #salt
11:29 babilen exactly
11:29 ThomasJ But I have an environment where I simply cannot trust grains either
11:29 SaveTheRb0tz joined #salt
11:29 ThomasJ Which also means I cannot use them for determining anything but the abovementioned
11:30 babilen We can't trust our grains so they are a big "no no" for anything confidential
11:30 dingo_ joined #salt
11:30 CryptoMe1 joined #salt
11:30 ThomasJ I maintain a single top.sls as well (almost) using the pillar driven design pattern
11:30 _vimalloc joined #salt
11:31 dh__ joined #salt
11:31 ThomasJ And use nodegroups combined with grains matching to handle site and os specific pillars
11:32 nod joined #salt
11:32 tmmt_ joined #salt
11:32 litwol_ joined #salt
11:33 ksj ThomasJ: ok, so you target via hostnames in the top file then?
11:33 tyler-ba- joined #salt
11:34 ThomasJ ksj: I target primarily on hostnames yes. Either in the pillar top.sls or using lists in the actual pillars themselves
11:34 ThomasJ Let me throw up a quick example
11:36 tyler-ba- joined #salt
11:36 codehotter joined #salt
11:36 cptsupermrkt joined #salt
11:37 eseyman joined #salt
11:37 cberndt joined #salt
11:37 zer0def joined #salt
11:38 colonD joined #salt
11:38 tzero joined #salt
11:38 ThomasJ ksj: Not sure if it's of any use, but http://pastebin.com/xphYCgd2
11:39 ThomasJ Could probably dynamically generate the formula/top.sls if I took the time
11:41 saifi joined #salt
11:41 ThomasJ Highly slimmed down examples of course
11:42 mage_ https://dpaste.de/ji5p/raw any idea ?
11:43 fredvd joined #salt
11:43 babilen mage_: You can't import Python modules in jinja (it's python support is shite and it was a massive mistake to use it as default renderer as opposed to mako)
11:44 MatthewsFace joined #salt
11:44 mage_ so it's impossible to use os.path.join in my sls templates ?
11:45 mage_ other than: "switch to Mako" :)
11:46 babilen I wouldn't quite go that far, but you cannot do that in jinja. Other approaches would be to: 1. Implement a salt execution module to do that and call it via salt['yourmodule.pathjoiner'](....) in jinja 2. Use a different renderer (e.g. mako or py or ...) 3. Curse jinja and everything it stands for, travel back in time and ensure it has proper Python block support
11:47 ThomasJ babilen: Am I to take it that you do not use jinja as a default renderer? :P
11:47 babilen Salt would have been *so much more* powerful if mako had been chosen as default renderer, but it is way too late for that
11:48 babilen ThomasJ: You can't switch the default renderer (well, you can, but ...) as most formulas and SLS files in the wild do not declare the renderer and simply rely on jinja+yaml being the default
11:49 babilen I have accepted my fate and cry a little whenever I have to write SLS files in pure Python just because I want to use some standard Python module
11:49 tyler-baker joined #salt
11:49 ThomasJ True enough
11:49 babilen I don't use mako as I don't want to introduce yet another renderer into the setup
11:50 genediazjr joined #salt
11:50 * ThomasJ is fighting hard to get by with jinja as well, for the same reason
11:51 mage_ I used Chameleon and I especially learned Jinja for salt :(
11:51 babilen Jinja was a horrible choice as it relies to have all the logic and data munging in the backend (that is, after all what it was designed for). Salt does, however, require a templating engine with more power than jinja and people either have to write execution functions for trivial things or write horrible, horrible jinja to work around its limitations
11:52 julez joined #salt
11:53 babilen The situation would be a bit better if every SLS file had to have a shebang
11:53 mage_ yeah, the fact that you can't even import std python modules is very limitating
11:53 babilen http://docs.makotemplates.org/en/latest/syntax.html#python-blocks is all I want
11:53 mage_ maybe the next major version of salt could break things in a good way ?
11:53 babilen it'll never happen
11:53 babilen there is way too much jinja code out in the wild
11:54 babilen (+ documentation)
11:54 babilen It's too late and will be a wart in salt's face forever
11:54 mage_ :s
11:55 babilen still don't understand why they didn't choose mako
11:55 mage_ I don't use Formulas so maybe I could switch to Mako ..
11:55 mage_ but it's, again, learning a new template engine which is not very fun
11:55 babilen indeed
11:55 mage_ anyway;
11:57 litwol is there a 'vs' list i can read to help me decide beteween mako and jinja?
11:57 litwol or betyer yet, just to learn the highlits?
11:57 litwol highlights*
11:58 mage_ I think Jinja was designed for the web
11:59 supersheep joined #salt
12:01 julez joined #salt
12:03 babilen jinja was designed for web designers to dynamically generate websites from data that is already in the state in which you want it.
12:06 litwol forces separation of responsibilities eh?
12:07 litwol wouldnt call that a bad thing.
12:07 lloesche In my formula should I use {{ grains['fqdn'] }} or {{ grains.fqdn }}? What's the difference?
12:08 litwol when PMs storm in your workspace telling you 'just do the quick hack' it is good to not having to like that 'literally cannot do that in the template'
12:08 jhauser joined #salt
12:09 babilen litwol: No, it is not a bad thing, but inappropriate for salt
12:15 TheHelmsMan joined #salt
12:22 kawa2014 joined #salt
12:24 greyrhino joined #salt
12:25 linjan joined #salt
12:28 Nazzy joined #salt
12:28 agend joined #salt
12:33 TooLmaN joined #salt
12:38 DjDeaf joined #salt
12:38 DjDeaf How can I combine salt.modules.archive.cmd_zip into my sls file? it seems like i can only use it with ad hoc commands
12:41 aawerner joined #salt
12:41 babilen DjDeaf: There is no corresponding state
12:43 babilen DjDeaf: You can, however, use http://docs.saltstack.com/en/latest/ref/states/all/salt.states.module.html#salt.states.module.run
12:44 otter768 joined #salt
12:44 Aidin joined #salt
12:46 permalac joined #salt
12:48 Tecnico1931 joined #salt
12:54 Auroch joined #salt
12:56 DjDeaf @babilen: and run the salt command from it?
12:56 DjDeaf babilen: and run the salt command from it?
12:59 subsignal joined #salt
13:00 jeremyr joined #salt
13:01 subsignal joined #salt
13:02 lloesche What is the difference between {%- %} and {% %} ?
13:02 DammitJim joined #salt
13:03 lloesche I'm looking at other peoples formulas and see both versions sometimes... what's the dash doing?
13:04 julez joined #salt
13:04 FeatherKing joined #salt
13:06 subsigna_ joined #salt
13:07 babilen lloesche: http://jinja.pocoo.org/docs/dev/templates/#whitespace-control
13:07 babilen DjDeaf: That's the idea, yeah
13:08 _JZ_ joined #salt
13:10 lloesche babilen: ah jinja syntax :) that makes sense. I'm only just beginning to understand that a lot of the syntax comes from the templating engine and has nothing to do with salt. Thank you!
13:10 bhosmer joined #salt
13:10 babilen Don't worry, you'll learn to hate jinja soon enough ;)
13:11 lloesche aha :) I already had to use a filter to join a list, takes some getting used to
13:14 mapu joined #salt
13:16 bluenemo hi guys. How could I write this (has to be in one line) so that the | is not appended after the last item in the list 'redirects'?   {%- for url in redirects %}{{ url }}|{% endfor -%}    => gives me:    foo.com|bar.com|test.com|      I dont want the last |
13:16 dopesong_ joined #salt
13:16 refnode joined #salt
13:19 babilen bluenemo: {% redirects|join('|') %}
13:19 bluenemo think this is ok:   RewriteCond %{HTTP_HOST} ^({%- for url in redirect[:-1] %}{{ url }}|{% endfor %}{{ redirect[-1] }}) [NC]
13:19 bluenemo ah thats even better. join. of course. thx babilen ;)
13:19 babilen http://jinja.pocoo.org/docs/dev/templates/#join
13:19 babilen But use {{ .... }}
13:20 hasues joined #salt
13:21 hasues left #salt
13:21 dyasny joined #salt
13:25 jdesilet joined #salt
13:27 evilrob joined #salt
13:28 cpowell joined #salt
13:35 TheHelmsMan joined #salt
13:35 c10 joined #salt
13:38 mpanetta joined #salt
13:38 c10 joined #salt
13:45 wnkz_ joined #salt
13:47 Tyrm joined #salt
13:48 jhauser joined #salt
13:49 supersheep joined #salt
13:50 allanparsons joined #salt
13:51 jdesilet joined #salt
13:52 andrew_v joined #salt
13:57 losh joined #salt
13:59 litwol babilen: sry to resurrect old topic. i wasn't able to listen-in too closely at the time. about jinja "limitations". i'd like to confirm:
14:00 litwol babilen: need for custom/heavy logic doesn't magicaly disappear, it gets redirected. Wouldn't it be advisable to shift such logic into a custom salt plugin and call it as a state or a module in sls?
14:00 TheHelmsMan joined #salt
14:00 litwol which then generates needed context for the template?
14:02 jdesilet joined #salt
14:03 babilen litwol: Yes, it would be. It is simply a *huge* step to write a custom execution module/state if all you need is a single function in the standard library
14:03 vincent_vdk left #salt
14:04 emaninpa joined #salt
14:04 babilen mako would have allowed to go this extra step by simply including literal Python blocks in the "default" SLS files, but jinja forces you to do a lot of work just because you want to use the, say, re module
14:06 babilen jinja was designed with the idea that the backend handles all that heavy lifting in mind. That is fair enough for a web design template language, but salt states and pillars *do* require more logic and writing custom states and execution modules is often overkill
14:06 morsik left #salt
14:06 sunkist joined #salt
14:07 rm_jorge joined #salt
14:07 nkuttler hm? state files can be mako, python, etc, can't they?
14:09 nkuttler litwol: https://docs.saltstack.com/en/latest/ref/renderers/all/
14:09 andrew_v_ joined #salt
14:10 babilen nkuttler: Sure, but you can't switch the default renderer and maintaining states and pillars in more than two languages causes a lot of wasted efforts and requires a lot of staff training
14:11 babilen I have enough problems getting ops to use jinja and python effectively .. I will not throw in another ten template languages
14:11 babilen (and if you take a look at formulas and some of the jinja code in the wild you will simply cry)
14:12 elfixit joined #salt
14:12 allanparsons joined #salt
14:12 nkuttler yeah..
14:13 nkuttler i'm currently on v2 of my base state tree, but i think i'll switch that to python entirely
14:13 nkuttler too much gymnastics involved with jinja
14:14 debian112 joined #salt
14:14 jhauser joined #salt
14:14 scoates joined #salt
14:16 litwol babilen: it is a huge step indeed.. but doesn't it lead to several important benefits? for example (1) reusable modules, (2) interchangeability of renderers if all they do is "print"
14:17 litwol i realize this step is a pain in the butt.. i also dealt with small systems grow unusable. and no one bothers learning design patterns anyway unless you've just graduated from comp sci course. having an impassable wall to separate your logic and presentation sounds like an annoying but good thing.
14:18 litwol fyi i'm "arguing" this for personal gain of understanding better, not for "what's good for salt"
14:18 murrdoc joined #salt
14:18 babilen It simply doesn't apply that way in salt.
14:18 litwol in other words, just curious.
14:18 Brew joined #salt
14:19 murrdoc morning
14:19 babilen o/
14:23 kawa2014 joined #salt
14:24 ajw0100 joined #salt
14:25 DammitJim how do I go about troubleshooting this problem:
14:26 evilrob joined #salt
14:26 DammitJim http://pastie.org/10210210
14:26 DammitJim This is my sls: http://pastie.org/10210212
14:27 iggy protip: gist supports multiple files per paste
14:27 DammitJim why is it complaining about an AttributeError: 'NoneType' object has no attribute
14:27 DammitJim where should I look?
14:28 murrdoc its ok to use the name attribute
14:28 murrdoc if anything its more readable
14:28 DammitJim murrdoc, where, though?
14:30 DammitJim I don't even know what line I'm looking at to fix this
14:30 andrew_v joined #salt
14:30 iggy your dependencies are all wrong (I think)
14:30 bluenemo what was the nifty way to do this again?   - contents_pillar: {% print 'foo:bar:' + name + ':' + file %}      where name and file are variables containing strings
14:30 DammitJim iggy, are you talking to me?
14:30 iggy DammitJim: yes
14:30 julez joined #salt
14:31 iggy -pkg: oracle-java8-installer
14:31 DammitJim all I'm trying to do is install oracle-java8-installer
14:31 iggy missing a space after - and the pkg should be java
14:31 DammitJim the ppa has already been added to apt
14:32 iggy - pkgrepo: oracle-java8-installer  ---> - pkgrepo: java
14:32 DammitJim oh, you mean, on line 17 shouldn't be oracle... but java?
14:32 DammitJim thanks for pointing out the space between - and pkg
14:32 iggy dependencies are of the form: <state module>: <sls id>
14:32 ksj where has the documentation gone for salt mine? I can only find old versions of it
14:33 DammitJim ok
14:33 iggy ksj: it's been deprecated
14:33 DammitJim after making those changes, I still have the same error :(
14:33 iggy ksj: nah, jk
14:33 iggy file a bug
14:34 ksj iggy: ok, I've found it, it's just google is not indexing it
14:34 murrdoc DammitJim:  does this work http://pastie.org/10210222 ?
14:34 ksj and the docs search uses google...
14:35 ksj maybe mine's not the correct way to do this, but I want all minions to send back their public server ssh key, so I can manage known_hosts files
14:35 murrdoc mine has been deprecated /
14:35 ksj murrdoc: has it? really? or is this a joke?
14:35 DammitJim let me see
14:36 froztbyte sigh
14:36 froztbyte spend ~3h beating some splunk stuff into submission
14:36 froztbyte deploy it onto everything in like 8s (yay local repo)
14:36 froztbyte one machine fails
14:36 murrdoc ksj:  i dont know, iggy said
14:36 froztbyte "oh, crap, that's the centos one"
14:36 DammitJim murrdoc, thanks... Rendering SLS "base:java" failed: Unknown yaml render error; line 3
14:36 froztbyte why must centos be a thing? :(
14:37 iggy I was kidding, mine isn't going anywhere
14:37 iggy ksj: you'd probably have to mix a custom module in with the mine to get that data
14:38 murrdoc DammitJim:  oh ?
14:38 ksj the fact that we were worried does kind of show how quickly development's moving though
14:38 DammitJim murrdoc, do I need to have an sls created just for debconf-util-pkg?
14:39 murrdoc DammitJim:  i missed a :
14:39 ksj iggy: I kind of thought that would be a really common thing to do. Managing authorized keys files manually is laborious, but very very important
14:39 ksj I guess most people just type "yes" though...
14:39 murrdoc hows https://gist.github.com/anonymous/44ebc741156322d14933
14:39 DammitJim ok, so I should just know that a : is needed just like in regular python when I see indentation...
14:40 giantlock joined #salt
14:40 murrdoc nah it was my bad
14:40 DammitJim murrdoc, you don't have to give me the answer... I need to learn how to troubleshoot something like this :D
14:41 DammitJim murrdoc, I should be calling this as: sudo salt '*' state.sls java-pkgs right?
14:41 DammitJim or the name of the sls?
14:41 iggy ksj: I imagine at a certain size, a lot of companies figure out how to integrate it into ldap or something
14:41 froztbyte ksj: Route53 still doesn't have formal SSHFP support, it's nowhere /near/ popular~
14:42 froztbyte (I say that with a lot of mourning)
14:42 murrdoc DammitJim:  name of the sls
14:42 DammitJim yeah, I just figured that out
14:42 murrdoc so like say sls is in {file_root}/hi/mom/test.sls
14:42 murrdoc you ll call state.sls hi.mom.test
14:42 DammitJim line 16, I should do -java-pkgs right?
14:42 DammitJim I mean - java-pkgs and not - java right?
14:43 murrdoc what is the name of the pkg u want to install ?
14:43 murrdoc java ? or java-pkgs ?
14:44 DammitJim if I was to do this on the server, I would do: sudo apt-get install oracle-java8-installer
14:44 murrdoc oh
14:44 murrdoc then remove - java
14:44 DammitJim ok, I did that
14:44 DammitJim and the result was this:
14:44 quist joined #salt
14:45 DammitJim https://gist.github.com/anonymous/94e5d41eaeb9112247f1
14:45 froztbyte ksj: so I got this conversation mid-way, but I just found https://github.com/saltstack-formulas/openssh-formula
14:45 froztbyte ksj: ignore if that's what you're talking about, though
14:45 illern joined #salt
14:45 otter768 joined #salt
14:45 DammitJim murrdoc, I guess it isn't understanding that I want it to accept the oracle license when using debconf.set
14:46 apergos when someone with a largish installation is awake (> 1k minions), I'd like to talk to them about salt tuning
14:46 murrdoc hold up
14:46 krisdavey joined #salt
14:47 froztbyte many of the docs have been broken after this URL change happened. it seems there's now an "en/" in the URL (to prepare for translation?). should've at least maybe tried some 302s in the mix
14:47 froztbyte quite jarring to find a 404 everywhere the whole time
14:47 froztbyte apergos: I'm not there yet, but I'm keen to know what problems you're seeing
14:47 murrdoc https://gist.github.com/anonymous/bf648d7866269beab42e try that DammitJim
14:47 illern joined #salt
14:48 murrdoc you are running into an indent issue
14:48 murrdoc see http://docs.saltstack.com/en/latest/ref/states/all/salt.states.debconfmod.html
14:48 murrdoc the indent for the 'question' is 4 spaces
14:48 quist joined #salt
14:48 murrdoc instead of 2
14:48 DammitJim oh really?
14:48 murrdoc ya rlr
14:48 murrdoc left #salt
14:48 murrdoc joined #salt
14:49 apergos the usual warning message "Salt request timed out. If this error persists, worker_threads may need to be increased." we have 25 workers so I doubt that's it, load is not too bad on the host
14:49 ksj froztbyte: I have so far managed to avoid using any formulas. I think they over complicated everything. but that formula, as far as I can see, requires pushing keys out to the minions, rather than getting keys from them.
14:49 DammitJim or it could all be on the same line, right?
14:49 aparsons joined #salt
14:49 ksj maybe that's the correct way to do it though
14:49 apergos test.pings to all 1k minions take about 40-50 seconds to be recorded by the master
14:49 apergos that's pretty slow
14:50 froztbyte ksj: you could work from it, though, to write whatever you want to do
14:50 froztbyte apergos: oh, youch
14:50 froztbyte all my stuff in still in the milliseconds region
14:50 apergos I haven't set any of the recon and auth stuff on the minion side but I wonder how applicable it is to this
14:50 DammitJim man, this 'package' takes forever to run
14:50 apergos all running 2014.7.5
14:50 ksj at the minute I'm trying to figure out what other functions mine actually has. literally the only one I can see in use anywhere is network.ip_addrs
14:50 froztbyte I guess I'll be up to 300~400 in the next month or so
14:51 apergos ah with 400 we didn't see any problems
14:51 jeremyr joined #salt
14:51 murrdoc ksj:  it has ALL functions
14:51 DammitJim murrdoc, you meant to remove - java right?
14:51 DammitJim since there is no such package?
14:51 murrdoc yeah
14:52 DammitJim man, 25 seconds just to check (java is already installed)
14:52 DammitJim am I having a network issue?
14:52 DammitJim thanks for clarifying about the indentation... I wonder if I need to add this to my vim plugins
14:52 cedwards joined #salt
14:53 murrdoc are u using salt-vim
14:53 DammitJim no
14:54 babilen you should
14:54 gthank joined #salt
14:54 gthank joined #salt
14:56 DammitJim I'll get it
14:56 DammitJim hhmmmm... I"ll have to google what debian package that is
14:57 peters-tx joined #salt
14:58 numkem is docker authentication broken in 2015.0? I'm getting a weird error message:  500 Server Error: Internal Server Error ("Registration: "Wrong username format (it has to match \"^[a-z0-9]{4,30}$\")
15:01 ksj murrdoc: ....shit. ok, that's powerful
15:02 murrdoc ksj:  configure minion mine functions as pillars, it makes them much easier to use and add/remove
15:02 babilen ksj: We use it for a variety of things. it started with mine function aliases for all private networks we have, but has grown to include all sorts of information
15:02 babilen +1 for mine functions in pillars
15:03 julez_ joined #salt
15:04 ksj babilen: any real world examples would be great. the only ones I can find just use network.ip_addrs
15:04 meeteOrite joined #salt
15:06 VSpike Is there an equivalent for states of the include "defaults" syntax for pillar? Or are jinja macros the right way to do that?
15:06 julez joined #salt
15:07 VSpike I've got a small state file which ideally I need to run once for each user I'm setting up, ideally with the username as a parameter I can use in the state file
15:07 murrdoc joined #salt
15:08 babilen ksj: http://garthwaite.org/saltmine_check_mk_agent.html might be interesting
15:08 bhosmer joined #salt
15:09 meeteOrite Is it possible to capture the details of each job run on a minion and post/put it to a web uri?
15:09 meeteOrite I imagine this would be the job of a custom returner no?
15:09 schristensen left #salt
15:09 murrdoc u imagine correc !
15:09 murrdoc +1 to meeteOrite
15:10 murrdoc but recently i have found out about salt-eventsd and i think i am in love
15:10 julez joined #salt
15:11 meeteOrite murrdoc: Ok, I was attempting to build a custom returner however I am not sure where to place it on the minion?
15:11 ksj babilen: thanks. reading now
15:12 babilen ksj: it is a bit tricky, but an infrastructure in which states register checks with nagios in the mine and the nagios server is using that information to adapt its config is nice
15:12 babilen There are other examples of it, but that was the first blog post I could find
15:13 c10 is there any way to create the directory structure with file.managed?
15:13 jimklo joined #salt
15:15 murrdoc meeteOrite:  http://docs.saltstack.com/en/latest/ref/returners/#writing-a-returner
15:15 ek6 c10: file.managed with makedirs True  ?  course then you would also have to create some type of file
15:15 markm joined #salt
15:16 meeteOrite murrdoc: salt-eventsd looks pretty good too
15:16 numkem ksj: I understand your pain regarding learnign about the mine system, I've just ran into this a couple of months ago
15:16 murrdoc its secure-er
15:16 bhosmer joined #salt
15:17 dopesong joined #salt
15:18 jalbretsen joined #salt
15:20 murrdoc meeteOrite:  i am surprised there isnt a http returner
15:21 murrdoc meeteOrite:  read the code for the slack returner and reuse it
15:21 bhosmer joined #salt
15:21 jhauser joined #salt
15:22 iggy froztbyte: file a bug
15:23 theologian joined #salt
15:24 cedwards murrdoc: thanks for the mention of salt-eventsd. looks interesting.
15:24 meeteOrite murrdoc: That really put me off too, I was looking at the built in returners thinking "where is the http returner???"  I am going to try salt-eventsd first
15:25 murrdoc i get why no http returner
15:25 murrdoc because the returner does more than send data back, it has the option to return info on jobs, minions so on
15:25 imanc joined #salt
15:27 murrdoc interesting, they dont use salt.utils.http in their retueners
15:27 iggy it's new
15:28 meeteOrite murrdoc: I simply want to get the data back from scheduled jobs so I can create a dashboard showing pass/changed/fail states.  Maybe I am going about this the wrong way?  Simple and clean is what I would like, and so far putting code on every minion seems like a huge PITA
15:28 Gareth murrdoc: Most of the returners (at least the ones I wrote) are using the requests module because salt.utils.http didn't exist at the time :)
15:28 murrdoc Gareth:  ah makes sense
15:28 murrdoc meeteOrite:  install salt-pad, do less work
15:29 imanc I have a req that is installed along with other items with pkg.latest: pkgs: - pure-ftpd  (obviously there are spaces in here and the syntax is correct)   a later state in the same file does  - require: -pkg: pure-ftpd    but I get this error: The following requisites were not found: require: pkg: pure-ftpd... despite the fact that it is installed on the OS
15:29 imanc and that the previous pkg.latest state ran without a problem
15:29 meeteOrite murrdoc: We didn't like it that much, and we want to integrate this data into our own application that tracks our network
15:30 genediazjr joined #salt
15:30 TaiSHi Did they fix locale generation in debian based systems?
15:32 iggy meeteOrite: something like what saltpad does isn't terrible though (dumping to mongo and just integrating your system into that mongo db)
15:32 murrdoc Gareth:  fix yo shizzle
15:32 iggy or... what am I thinking about
15:32 murrdoc :P
15:32 meeteOrite iggy: That's the opposite of what we want to do, no more database please.  We have so many damn databases it's insane.
15:33 iggy that's the only way to keep something with historic data
15:33 iggy the master job cache evicts data >24 hours old
15:33 iggy saltobserver
15:33 meeteOrite Then what's the point of the customer returner?  I can't get the data from that and do a history on my backend db?
15:34 froztbyte iggy: not a salt problem, unfortunately
15:34 FeatherKing anyone seen this error using docker.pulled AttributeError: 'RecentlyUsedContainer' object has no attribute 'lock'
15:35 losh joined #salt
15:35 iggy froztbyte: was referring to docs 404s
15:35 FeatherKing minion version 2015.5, there is a bug report 21896 but it was closed in 2014.7.2
15:39 TaiSHi murrdoc: empty ping
15:39 murrdoc TaiSHi:  Riposte!
15:40 TaiSHi Query !
15:40 TheHelmsMan joined #salt
15:40 murrdoc 404!
15:41 TaiSHi ¬¬!
15:41 Gareth murrdoc: my kung fu is the best.
15:45 froztbyte iggy: sorry, that was unclear. it's not a salt problem, just random splunk bullshit and I was only targetting debian first
15:46 froztbyte oh, right
15:46 froztbyte that
15:46 froztbyte (I just scrolled up)
15:46 froztbyte yeah I should
15:47 iggy froztbyte: you know halite is dead right?
15:48 Gareth long live halite.
15:48 * murrdoc puts away crash cart
15:48 vexati0n halite was awful anyway :(
15:48 vexati0n it would have been cool if my laptop had 4 quad-corde Xeon processors and 2TB RAM
15:49 vexati0n and a 10gb NIC
15:49 iggy I never got it running
15:49 mpanetta I think saltpad is supposed to be a replacement?
15:50 vexati0n i just wrote my own in php
15:50 TaiSHi How often are ubuntu packages built ?
15:50 mpanetta php :(
15:50 vexati0n more intiuitive for what we actually use it for.
15:50 vexati0n yeah yeah php is the devil blah blah blah. i hear it all day.
15:50 mpanetta haha
15:50 iggy SSE has one (with subscription)
15:51 vexati0n yeah we can't use the SSE one because it only supports PAM auth so far, not LDAP/AD
15:51 vexati0n but it does seem to be nicer than Halite
15:51 iggy one would hope
15:51 iggy I mean they are effectively charging for it
15:51 murrdoc its 50 bucks a server
15:51 vexati0n mainly because it doesn't do all of its processing on the client side
15:52 andrew_v Is there a way to pass arguments to the orchestrate runner? That is, within the orchestration .sls, I want to be able to do something like: `salt.state: - tgt: {{ arg }}-rest-of-hostname`.
15:52 mpanetta https://github.com/tinyclues/saltpad
15:54 nobrak joined #salt
15:54 RedundancyD joined #salt
15:54 vexati0n eh. i'm still trying to get reactor to work. it keeps telling me 'too many functions declared in state' even though the state referenced in the error isn't the one referenced in the reactor settings.
15:57 bluenemo joined #salt
15:57 dimeshake joined #salt
15:58 iggy andrew_v: orchestrate will accept saltenv and pillar args... should be able to do something with that
16:02 schristensen joined #salt
16:03 DammitJim joined #salt
16:04 desposo joined #salt
16:04 vexati0n why would top.sls work perfectly fine when commanding a minion to do state.highstate, but when telling reactor to do the same thing on minion_start, it errors with "too many functions declared in state..." ?
16:05 julez joined #salt
16:05 iggy gist your code
16:06 froztbyte iggy: I didn't
16:07 ndrei joined #salt
16:07 froztbyte But since I saw foreman has salt capability I'll just do that
16:08 iggy froztbyte: saltpad is (so far) the suggested alternative (or going with SSE's web gui)
16:08 vexati0n https://gist.github.com/8c23fcdec08b6e9918db.git
16:08 amcorreia joined #salt
16:09 iggy take off the .git if you want to actually see it
16:09 vexati0n https://gist.github.com/anonymous/8c23fcdec08b6e9918db
16:09 vexati0n oops
16:09 andrew_v iggy: You mean {{ pillar['some:pillar:value'] }} works inside an orchestration sls? Yeah, that's not ideal but it might be enough.
16:10 iggy vexati0n: looks similar to what I have (it works)
16:11 iggy andrew_v: I mean you can pass pillar data to an orch run
16:11 vexati0n but here is what I get from the master when a minion comes online
16:11 vexati0n 2015-05-27 08:56:09,892 [salt.utils.reactor][ERROR   ][11983] Unable to render reactions for event salt/minion/edg-c-salt01-az/start due to errors (["Too many functions declared in state 'nginx_role:frontend' in SLS '/srv/salt/top.sls'", "Too many functions declared in state 'nginx_role:public' in SLS '/srv/salt/top.sls'", "Too many functions declared in state 'nginx_role:backend' in SLS '/srv/salt/top.sls'"]) in one or more of the sls files ([
16:11 kawa2014 joined #salt
16:12 froztbyte iggy: SSE?
16:12 iggy SaltStack Enterprise
16:13 froztbyte Oh
16:13 froztbyte Is that a toy everyone gets to use?
16:13 iggy no, it's for paying customers
16:13 andrew_v iggy: Do you have a link to example usage. What I'm trying to do: I have an orchestration job that sets up and initializes a database cluster. I have multiple database clusters. I want to be able to specify which cluster to run it against, without having to maintain either separate environments or separate .sls files for each.
16:14 andrew_v (usage?, not usage. Pardon.)
16:15 froztbyte iggy: ah
16:15 froztbyte Well, that's probably not a killer in the long run. I don't think we're there just yet but may be soon
16:15 iggy andrew_v: sudo salt-run state.orch dborch pillar='{"cluster": "dbcluster001"}'
16:16 mapu joined #salt
16:16 jcj_moz joined #salt
16:16 iggy andrew_v: and then in your sls, salt.state: - tgt: {{ pillar['cluster'] }}-rest-of-hostname
16:16 froztbyte I'm still getting to know much of the options available in Salt; however, been managing a generally better pace with it than I ever did with puppet
16:17 andrew_v iggy: Okay. Yeah, that will do what I want, I think. Thanks.
16:17 froztbyte Building up a tiny bit of debt though, will have to revisit early states sometime.
16:19 andrew_v iggy: Is there a way to bail out of the job with an appropriate error if it's not specified? (not a big deal if not, I assume it will just break noisily if pillar['cluster'] doesn't exist, but it would be nice)
16:20 andrew_v Hrm. Actually, never mind, I don't need that after all.
16:20 iggy it'll fail to render and should point to what line failed which should hopefully make it clear
16:21 andrew_v Okay. That'll work.
16:21 andrew_v Thanks.
16:23 lothiraldan joined #salt
16:23 ndrei joined #salt
16:24 murrdoc joined #salt
16:24 wt joined #salt
16:25 wt Is the documentation styling supposed to be different between http://docs.saltstack.com/en/latest/ and http://docs.saltstack.com/en/2014.7/?
16:25 wt The latest styling seems messed up.
16:25 dopesong joined #salt
16:25 wt the 2015.5.0 one, that is
16:26 iggy wt: yeah, I think they are tinkering
16:26 wt The right sidebar doesn't scroll, and the state modules are hard to navigate.
16:26 murrdoc also one is develop
16:26 wt ah
16:26 ctolsen joined #salt
16:26 murrdoc the other is 2014.7
16:26 iggy latest is actually 2015.5
16:26 wt The one in develop looks like 2014.7.6, which is okay.
16:26 iggy develop is /en/develop
16:27 wt I copy/pasted the URLs
16:27 wt it's "latest" in the URL.
16:27 murrdoc das new
16:28 genediazjr joined #salt
16:28 wt Does file.exists create the file if it doesn't exist?
16:29 wt It doesn't appear to do so.
16:29 ndrei joined #salt
16:29 wt I don't want to touch the file it it exists.
16:30 iggy "have been previously satisfied before deployment"
16:30 iggy I'd gues no
16:30 murrdoc basically the right side of teh docs is now useless
16:31 murrdoc wt:  use file.managed with no source or contents
16:32 wendall911 joined #salt
16:33 murrdoc Note that if you just want to create a file and don't care about atime or mtime, you should use file.managed instead, as it is more feature-complete. (Just leave out the source/template/contents arguments, and it will just create the file and/or check its permissions, without messing with contents)
16:35 writtenoff joined #salt
16:37 spookah joined #salt
16:38 wnkz_ joined #salt
16:39 cruatta joined #salt
16:40 cruatta joined #salt
16:41 cruatta_ joined #salt
16:43 stephanbuys joined #salt
16:44 ndrei joined #salt
16:44 Grokzen joined #salt
16:46 MatthewsFace joined #salt
16:46 otter768 joined #salt
16:46 felixhummel joined #salt
16:47 zerthimon joined #salt
16:48 murrdoc anyone managing kernel versions with salt
16:49 DammitJim oh goodness... that's another can of worms!
16:49 mgw joined #salt
16:50 aparsons joined #salt
16:50 mgw can the same pillar_root be used for multiple environments, distinguished by the entries in top.sls?
16:50 mgw the format of top.sls implies that, but it doesn't seem to be working
16:51 stoogenmeyer__ joined #salt
16:53 gladiatr joined #salt
16:56 saltstacknoobie joined #salt
16:57 bhosmer_ joined #salt
16:57 felixhummel joined #salt
16:58 felixhummel joined #salt
16:58 cruatta joined #salt
17:00 lothiraldan joined #salt
17:01 Ryan_Lane joined #salt
17:02 Ryan_Lane joined #salt
17:02 Ryan_Lane joined #salt
17:02 ctolsen joined #salt
17:02 ndrei joined #salt
17:03 aparsons joined #salt
17:03 DjDeaf joined #salt
17:03 tiadobatima1 joined #salt
17:04 bhosmer__ joined #salt
17:06 TyrfingMjolnir joined #salt
17:06 s_kunk_ joined #salt
17:08 s_kunk joined #salt
17:08 amcorreia root@saltmaster:/var/cache# salt '*' salt.highstate
17:08 amcorreia saltmaster.zarathustra.com.br:
17:08 amcorreia 'salt' __virtual__ returned False
17:08 amcorreia what this error means?
17:11 ndrei joined #salt
17:11 davisj I question for those who agree that mako > jinja: If you were starting your salt rollout from square one and training/existing code basee was a non-issue, would you still do jinja? Another way to put it... is there a technical downside other than fighting inertia?
17:12 gthank joined #salt
17:12 davisj s/downside/downside using mako/
17:14 bhosmer_ joined #salt
17:14 forrest joined #salt
17:16 lz-dylan amcorreia: I don't really know what to do with that, but try running with `-ldebug` for a bit more output
17:18 racooper joined #salt
17:20 amcorreia lz-dylan, http://pastebin.com/CstauqtS
17:21 amcorreia lz-dylan, this started to happen after i try to use this formula https://github.com/yandex-sysmon/nginx-blocks-formula
17:21 amcorreia then i removed and restarted salt-master, but this error still persist
17:22 ndrei joined #salt
17:22 ctolsen joined #salt
17:23 iggy davisj: docs
17:25 numkem Anyone been playing with the dockerio states regarding links?
17:25 sukh joined #salt
17:26 denys joined #salt
17:26 conan_the_destro joined #salt
17:27 julez joined #salt
17:28 andrew_v iggy: Tried the orchestration/pillar business and it worked perfectly the first time. Awesome.
17:29 TaiSHi murrdoc: You know you love me
17:29 swa_work joined #salt
17:29 cruatta joined #salt
17:31 impi joined #salt
17:33 davisj iggy: you're referring to the salt docs use of jinja, or to a shortcomming of mako docs?
17:33 iggy both
17:33 iggy I mean mako is documented, but not so much in the context of Salt
17:34 murrdoc TaiSHi:  i done do
17:34 murrdoc davisj:  if u want to l33t it up
17:34 murrdoc use pyobjects
17:34 murrdoc but for me jinja has worked fine because as a dsl is supremely limiting
17:35 MrsButterWorps joined #salt
17:35 davisj murrdoc: oh, I intend to :) But I'm the only one on the team who speaks python ;)
17:35 murrdoc and all tricky stuff ends up being either custom modules or states
17:35 murrdoc so yeah mako will let u do more things in templates and states
17:35 wnkz joined #salt
17:35 theologian joined #salt
17:35 murrdoc but i actually dont want my users to do 'smart' things in states or templates
17:35 murrdoc all the smarts are in pillars, formulas and custom python modules
17:36 davisj murrdoc: a fair point
17:36 murrdoc its what i meak
17:36 murrdoc make*
17:37 murrdoc ALL DAY!
17:37 * murrdoc toots his own horn
17:37 * murrdoc puts down horn and gtfbtws
17:37 sukh joined #salt
17:37 * davisj can't see "toots" without giggling a litte.
17:37 * davisj has a two year old.
17:37 aboe joined #salt
17:38 murrdoc :)
17:38 murrdoc nice
17:39 dopesong_ joined #salt
17:40 DammitJim do you guys use salt to deploy new instances of servers?
17:40 DammitJim or do you use it in conjunction with another tool?
17:40 DammitJim I'm not understanding how I can use salt with a vmware vm template
17:41 iggy salt-cloud, yes
17:41 stanmarsh_ joined #salt
17:41 DammitJim what's that? I have vmware vcenters
17:42 jpaetzel joined #salt
17:43 viq joined #salt
17:44 whytewolf salt-cloud is a part of salt that can request/deploy/build/destroy instences in many cloud architectures.
17:45 stanmarsh_ left #salt
17:46 ndrei joined #salt
17:48 shellhacks joined #salt
17:49 shellhacks left #salt
17:49 ren0 joined #salt
17:55 tomh- joined #salt
17:55 ndrei joined #salt
17:57 aparsons joined #salt
17:57 giantlock joined #salt
17:57 ren0 Hello. Does anybody has expirience in using salt-cloud with vCloud Director? In particular i am trying to add custom cloud provider (https://github.com/ministryofjustice/salt-cloud-provider-vcloud) but seems salt-cloud doesn't load it at all (https://github.com/ministryofjustice/salt-cloud-provider-vcloud/issues/5#issuecomment-105945534)
17:59 baweaver joined #salt
18:00 DammitJim oh, this is sooo cool! I can prep a server with the proper user!
18:01 sgargan joined #salt
18:01 bhosmer_ joined #salt
18:01 DammitJim is it normal that I am creating all my sls files as root?
18:03 numkem how can I feed a dict instead of an Ordered dict to a function? Reading through the code of docker_py regarding links I see its trying to check if the variable is a dict
18:03 tiadobatima joined #salt
18:03 numkem this is using the yaml/jinja rendered
18:03 numkem *renderer
18:04 schristensen joined #salt
18:05 LtLefse DammitJim: I don't think it matters too much what user owns the .sls files, but ideally you want them version-controlled
18:06 LtLefse failing that, at least be sure to back them up!
18:06 benw joined #salt
18:06 spookah joined #salt
18:06 DammitJim yeah, I need to start checking this stuff into svn
18:06 murrdoc svn ?
18:06 murrdoc what is that
18:06 LtLefse I keep mine in subversion and use the svn backends for state and pillar
18:07 LtLefse murrdoc: it's a version control system :P
18:07 DammitJim LOL
18:07 murrdoc is it like dropbox ?
18:07 DammitJim murrdoc, is git
18:07 whytewolf murrdoc: it is an old outdated version control system that uses cvs style logic
18:07 wt LtLefse, if you call that version control...:-P
18:07 DammitJim I take that back, it is NOT git, but murrdoc only knows of git
18:07 DammitJim :D
18:07 wt murrdoc, it's not CVS style at all
18:08 DammitJim it's good enough to get started and not loose anything
18:08 * murrdoc knows svn, was weak troll
18:08 wt svn is Apache Subversion
18:08 wt ah
18:08 LtLefse it works great for my team of 2 sysadmins
18:09 DammitJim we are 3 sysadmins
18:09 dayid joined #salt
18:09 murrdoc I AM ONE
18:09 DammitJim and I'm the only one that knows how to use it (the little bit I know)
18:09 wt LtLefse, I was kidding.
18:09 wt LtLefse, there are actually some things that I liked about svn....like partial repo ACLs.
18:10 LtLefse I suppose I should get on the git train someday
18:10 LtLefse can't use GitHub where I work, though
18:10 wt I also like the fact that svn doesn't require downloading the full project history for a checkout
18:11 ajw0100 joined #salt
18:12 murrdoc wt:  truncating history in svn is paaaaain
18:12 murrdoc why cant u use git LtLefse
18:12 DammitJim this might be a security concern, but can I set a user password?
18:13 LtLefse murrdoc: not git, GitHub. can't use any cloud service without vetting from the legal dept, corp. security, etc, etc
18:13 DammitJim or do you guys just set the password when you go to the server?
18:13 murrdoc DammitJim:  google saltstack gpg renderer
18:14 murrdoc LtLefse:  fun
18:14 DammitJim in saltstack everything is a renderer, huh?
18:14 LtLefse and I know git doesn't require GitHub. but I already knew and loved svn anyways
18:14 bbradley joined #salt
18:15 iggy sick, sick, person
18:15 amcorreia then i removed and restarted salt-master, but this error still persist
18:15 amcorreia ops
18:15 murrdoc the best part about svn
18:16 murrdoc when u commit, its for real
18:16 murrdoc its like real life
18:17 DammitJim :-P
18:17 DammitJim LtLefse, do you just check in your whole /srv/salt tree?
18:18 LtLefse DammitJim: yes
18:18 DammitJim LtLefse, do you keep all these files owned by root?
18:19 tmclaugh[work] joined #salt
18:19 LtLefse well, what I do is use Salt's 'svn' fileserver_backend
18:19 sergio_ joined #salt
18:20 LtLefse so Salt pulls the state files directly from SVN
18:20 DammitJim woot woot???
18:20 DammitJim I need that
18:20 LtLefse and me and the other admin each have a working copy in our homedirs
18:20 murrdoc dont do it
18:20 murrdoc dont do it
18:20 DammitJim does that mean I have to ditch my svn server and create a new one?
18:20 murrdoc use debs
18:20 DammitJim debs?
18:21 aawerner joined #salt
18:21 LtLefse but as iggy says, I'm a sick, sick person
18:23 LtLefse DammitJim: you shouldn't have to ditch your svn server
18:24 LtLefse read up on the svn fileserver_backend if you want to do it
18:26 LtLefse http://docs.saltstack.com/en/latest/ref/file_server/backends.html
18:26 DammitJim to be clear... if I am using a require statement in an sls and I want to call another formula, I just say: - <formulax>
18:26 DammitJim right?
18:26 cruatta_ joined #salt
18:27 shellhacks joined #salt
18:27 Ahlee yes
18:27 aawerner joined #salt
18:29 smcquay joined #salt
18:30 ndrei joined #salt
18:31 DammitJim LtLefse, is this something where you just check in stuff from another workstation
18:31 DammitJim then let the master check out the latest and run it?
18:31 murrdoc joined #salt
18:31 LtLefse right
18:32 DammitJim nice
18:32 DammitJim of course, you don't do this while you are developing new formulas, right?
18:33 LtLefse right, I apply new formulas to a test system first
18:34 DammitJim gosh, I need to do that
18:34 baweaver joined #salt
18:34 laax joined #salt
18:34 DammitJim right now I am just applying my formulas to new servers I created
18:35 LtLefse I can't say I really have a great formal process or anything
18:35 DammitJim but this is good info... great advice
18:37 laax joined #salt
18:38 DammitJim LtLefse, did you define your svnfs_remotes in /etc/salt/master?
18:39 LtLefse I used /etc/salt/master.d/fileserver_backend.conf
18:39 LtLefse actually all my master config is broken up into files in master.d/
18:39 cruatta joined #salt
18:40 LtLefse same for minion.d
18:40 DammitJim oh nice
18:40 DammitJim how do you reference your conf files?
18:40 LtLefse that makes it easier to Salt manage the Salt config :)
18:40 DammitJim Sounds like it's a clean setup the way you have it
18:40 LtLefse the cleanest one I've seen
18:40 LtLefse s/cleanest/only/
18:41 LtLefse by default Salt master picks up everything in /etc/salt/master.d/*.conf
18:41 VSpike Hmm. I just hit the issue that running commands as a specific user is not supported in Windows. Is there any reason I can't use https://msdn.microsoft.com/en-us/library/windows/desktop/aa376977%28v=vs.85%29.aspx (assuming I don't need the command to run right now?)
18:42 DammitJim so, what, you just have to call the file in master.d, fileserver_backend.conf ?
18:42 VSpike The intention is to install some powershell bits for the user so they get a nicer PS profile
18:42 LtLefse DammitJim: right. as long as it ends in .conf it will get read
18:43 cruatta joined #salt
18:43 DammitJim so, for my minion, I could have created a master.conf instead of changing master = in the minion file?
18:43 tiadobatima murrdoc, iggy, guys... What are your thoughts on how upstart.py handles status()? For context, take a look at this comment:
18:43 tiadobatima https://github.com/saltstack-formulas/docker-formula/pull/29#issuecomment-105995502
18:44 LtLefse right. I have /etc/salt/minion.d/master.conf for the minions
18:44 DammitJim and all you got in that file is: master = <master_hostname>
18:44 DammitJim ?
18:44 LtLefse yes, that's it
18:45 LtLefse well, master: <hostname>
18:45 ALLmightySPIFF joined #salt
18:45 DammitJim oh yeah, sorry no =, but ;
18:47 otter768 joined #salt
18:47 FeatherKing is there a way to tell a salt-minion about pip module it loaded during a state run? i am using reload_modules: true but i still get failures
18:48 ajw0100 joined #salt
18:48 FeatherKing i load docker-py and urllib3 during a state run and pulling a docker image at the end of the run fails. if i restart the minion it succeeds
18:48 laax joined #salt
18:52 VSpike Which would be a hell of a lot easier if the bug with setting windows registry values that's been open for a year or so had been fixed :/
18:54 awerner_ joined #salt
18:55 dfinn joined #salt
18:55 bhosmer_ joined #salt
18:55 huddy joined #salt
18:58 tonyhope joined #salt
18:59 chiui joined #salt
18:59 hybridpollo joined #salt
19:00 murrdoc1 joined #salt
19:03 murrdoc joined #salt
19:05 bhosmer_ joined #salt
19:06 linjan joined #salt
19:09 rvankleeck joined #salt
19:10 rvankleeck i'm wanting to make sure docker is installed on a minion, but I want it to install from source (e.g. wget -qO- https://get.docker.com/ | sh)
19:10 rvankleeck could someone point me to the right state to use for this?
19:11 apergos I guess people should be awake now so here's my question: we've got over 1k minions now, getting the dreaded "Salt request timed out. If this error persists, worker_threads may need to be increased."
19:11 apergos we've got 25 workers though, 16mb ram on the box, and while puppet runs on it it doesn't seem all tht loaded
19:12 apergos so can someone who has experience with a large installation let me know what's most likely to help with that?
19:12 apergos the symptom is e.g. test.ping across all minions takes 40-50 seconds for them all to check in
19:12 apergos or at least to be recorded by the master.
19:16 apergos I'm aware of the recon and auth tweaks, but I would like to know how much those affect running commands, if at all
19:19 tiadobatima rvankleeck: get.docker.com doesn't really "install from source". it still uses the underlying package management (ym, apt-get...)
19:21 tiadobatima try using the docker-formula
19:21 rvankleeck tiadobatima, thanks
19:22 tiadobatima np
19:23 murrdoc tiadobatima:  fix it right
19:23 murrdoc w.r.t your pull request
19:23 murrdoc do all the configs in the defaults.yml
19:23 murrdoc and then setup your state file
19:25 julez joined #salt
19:26 tiadobatima joined #salt
19:26 tiadobatima defaults.yml murrdoc?
19:27 murrdoc yeah
19:27 murrdoc the docker-formula doesnt hvae one ?
19:27 tiadobatima nope
19:27 murrdoc is it all in map.jina
19:27 tiadobatima yeah
19:28 murrdoc hate it
19:28 murrdoc well anyways configure it all in the map.jinja
19:28 murrdoc keep your 'sig' fix
19:28 murrdoc i like it
19:29 tiadobatima I'll send the PR in a few minutes... addressing just the pkg.latest and pkg.installed problem
19:29 murrdoc u can rebase that into your current PR
19:29 murrdoc push up to the same brnach
19:29 tiadobatima ok
19:31 tiadobatima did you see my comment on PR #29?
19:32 cruatta_ joined #salt
19:33 c10 joined #salt
19:34 noobadmin_ left #salt
19:34 Tyrm joined #salt
19:37 Tecnico1931 joined #salt
19:40 hal58th_1 joined #salt
19:40 hal58th_2 joined #salt
19:40 hal58th_3 joined #salt
19:43 bhosmer_ joined #salt
19:44 baweaver joined #salt
19:46 giantlock joined #salt
19:49 bhosmer_ joined #salt
19:50 bhosmer_ joined #salt
19:51 aphor joined #salt
19:53 rocky____ joined #salt
19:53 rocky____ howdy
19:53 aphor hai
19:54 rocky____ I have a question, when will version Beryllium  be out?
19:54 BgBee joined #salt
19:55 rocky____ I was looking into the boto_rds module but it says that it will be out with Beryllium version
19:56 aphor Question.. pillar['somekey']['subkey'] or salt.pillar.get('somekey:subkey') or salt['pillar.get']('somekey:subkey') ??
19:57 murrdoc my vote is the last one
19:57 murrdoc saltstack leans towards the one in the middle
19:57 murrdoc but it was broke for a while
19:58 aphor wouldn't it be nicer to just pillar('somekey:subkey') ? pillar could be a subclass of dict() with a custom __call__()
19:58 murrdoc i like using salt and __salt__ everywhere
19:58 murrdoc so i cant say
19:58 tomspur_ joined #salt
19:58 aphor I like to make things more pythonic...
19:59 murrdoc sure
19:59 aphor I'm grinding through all my map.jinja changing the second option to the third, and I had an epiphany: this is ugly.
19:59 bmac2 joined #salt
19:59 murrdoc ?
19:59 aphor So now I'm contemplating filing a bug and a patch.
20:00 aphor ... pull request
20:00 kaptk2 joined #salt
20:00 murrdoc the third option isnt the ugliest
20:00 murrdoc if u think about how many modules are available
20:00 murrdoc and it would be cumbersome managing the same
20:01 murrdoc they used to have __pillar__ and __grains__
20:01 murrdoc and now they have moved away from it for the same reason
20:02 felixhummel joined #salt
20:02 aphor .. could be better.
20:03 Pulp joined #salt
20:05 tiadobatima murrdoc...  https://github.com/saltstack-formulas/docker-formula/pull/33
20:05 RabidCicada joined #salt
20:05 aphor I feel like maybe this isn't really important for anyone else, so maybe I will shelve the __call__() module getter idea.
20:06 tiadobatima let me know if something is not up to standard... I wanna make sure I do it as the pros here ;)
20:06 iggy rocky____: not for a while, 2015.5 _just_ came out
20:06 rocky____ ok thanks
20:06 iggy rocky____: I'd expect 2015.11/12
20:06 rocky____ will do a work around custom module
20:06 rocky____ thanks again
20:07 iggy rocky____: you can probably just copy the boto* modules into /srv/salt/_modules (and similar with the state)
20:07 iggy /srv/salt/_modules is an example... $file_roots/_modules
20:08 rocky____ yea that is what am setting up now, ;)
20:08 rocky____ state should go under _states, correct?
20:09 iggy si
20:10 rocky____ awesomeness
20:11 jhauser joined #salt
20:12 cberndt joined #salt
20:13 ajw0100 joined #salt
20:14 murrdoc joined #salt
20:14 VSpike I'm sure this is simple, but how can I kill an executing job on a minioin from the master?
20:15 murrdoc figure out job id using salt runner for jobs
20:15 murrdoc then use saltutil.kill_jid ? to kill the job
20:17 abng88 joined #salt
20:20 bhosmer_ joined #salt
20:20 abng88 running 2015.5.0, able to get gitfs working with salt-call (masterless) for my states, was wondering if ext_pillar is also supported in the minion config?
20:22 forrest abng88: Looks like it according to this issue: https://github.com/saltstack/salt/issues/16478
20:23 abng88 hmm ok i must just have something configured wrong
20:23 abng88 is this something also worth putting in the masterless tutorial docs?
20:24 CeBe1 joined #salt
20:25 forrest abng88: Without a doubt
20:25 abng88 ok, ill create another issue
20:25 abng88 thx
20:26 rocky____ So I tried placing the custom state under _states
20:26 rocky____ and having the minion sync, which worked correctly as expected
20:26 rocky____ the minion picked the new custom state
20:26 rocky____ but when I run the state.sls it says that there is no such state
20:26 rocky____ please check link
20:27 wt joined #salt
20:28 rocky____ http://paste.debian.net/184730
20:28 nahamu Is the Windows build still using Esky or is it something else these days?
20:28 rocky____ any ideas
20:28 rocky____ ?
20:30 murrdoc check your minion salt log
20:33 rocky____ thanks
20:33 linjan joined #salt
20:33 rocky____ yup found some good info
20:33 rocky____ http://paste.debian.net/184732/
20:35 VSpike murrdoc: thanks!
20:35 murrdoc ok
20:40 rocky____ found that the boto module I have does not not come with rds2
20:40 rocky____ import boto.rds2
20:40 rocky____ sorry.. for accidental paste "import..."
20:42 baweaver joined #salt
20:46 schristensen joined #salt
20:46 markm joined #salt
20:47 tmclaugh[work] joined #salt
20:47 nodens hi there
20:48 nodens murrdoc, thanks, I can work from here
20:48 schuckles joined #salt
20:48 nodens though I'll probably end up patching cmdmod to add a cmd.exec_code_json or something like that
20:48 murrdoc send a pull upstream
20:48 nodens along with a PR and a BR :)
20:49 murrdoc and a beer ?
20:49 yawniek joined #salt
20:50 nodens a bug report, though in the unlikely event that you live near grenoble in france or geneva / lausanne in switzerland, a beer can easily be arranged.
20:50 murrdoc heh
21:01 soren_ joined #salt
21:03 Tyrm joined #salt
21:13 tmclaugh[work] joined #salt
21:15 cruatta joined #salt
21:16 pviktori joined #salt
21:17 cozby_ left #salt
21:22 badon joined #salt
21:26 baweaver joined #salt
21:27 TyrfingMjolnir joined #salt
21:31 dopesong joined #salt
21:31 dimeshake joined #salt
21:31 dimeshake joined #salt
21:34 badon_ joined #salt
21:36 dopesong_ joined #salt
21:37 pppingme joined #salt
21:41 mrbiggle_ joined #salt
21:42 _JZ_ joined #salt
21:46 andrej At the moment I'm generating a icinga hosts.cfg out of salt on the master by using a jinja template, and this worked well for us so far ...
21:46 cberndt joined #salt
21:47 andrej we have a raft of machines that are our out-of-band entry points in our POPs, and they typically have 3 interfaces, but the monitoring server can only reach one of those directly.
21:48 andrej I have written a python script that will harvest the other available interfaces, and create a dict of machines that holds the other interfaces addresses.
21:49 andrej Now I'm wondering about the best way to tie that in with  the template; is there a way to have jinija interact w/ python, or should I make my script populate custom grains for those minions, and tackle it via running the script from the same state that generates the other config items, just before the actual generation?
21:49 andrej Is there a "best practice" for this kind of thing?
21:50 VSpike grrr windows
21:51 pviktori joined #salt
21:51 VSpike I just spent about an hour of my own time trying to work out how to run some powershell in the context of each user on a windows box to give them a nice PS environment. Not even that important, but it really bugged me I couldnt do it
21:51 murrdoc if its a module, call it in a state and hand it down to template
21:51 VSpike Going to have to admit defeat though.
21:51 murrdoc if its a grain, call it in the state, set context vars, hand down to template
21:51 murrdoc no VSpike no
21:52 murrdoc fight on
21:52 VSpike I know I could do it with scheduled jobs but that's just ugly, and trying to spawn powershell is just horrible due to its wierd handling of pipes and quoting, especially when wrapped in cmd
21:53 VSpike murrdoc: i would, but balancing the fact that I really hate leaving a problem unsolved is the fact that I really ultimately dont want to know more about powershell, windows process credentials, cmd shell quoting, etc etc
21:54 murrdoc valid
21:55 VSpike I managed to make it work with psexec by using psexec -u blah -p foo cmd /c "echo . | powershell -NonInteractive -File C:\utils\script.ps1" ...
21:55 VSpike except that doing that fails when called from the context of the SYSTEM user, which is what Salt runs as
21:55 andrej Thanks murrdoc ... now I need to find out what I have written
21:55 andrej LOL
21:55 murrdoc doesnt command run have a run_as
21:55 VSpike murrdoc: it always requires the password entered interactively
21:56 murrdoc oh dear lod
21:56 TyrfingMjolnir_ joined #salt
21:56 VSpike And no, you can't pipe it in
21:57 VSpike I was thinking the registry RUNONCE mechanism would sort of work, but normally you only create that in HKCU. To create it in another user's registry hive is probably not a good idea
21:58 VSpike The only way I've seen that would probably work is this https://github.com/markuskramerIgitt/LearnSalt/blob/master/learn-run-as.sls
21:58 VSpike But I can't bring myself to do it
21:58 pviktori_ joined #salt
21:59 VSpike Besides I'd probably have to create a batch file to call the powershell file
21:59 VSpike Just.. no
22:03 Berty_ joined #salt
22:05 pviktori joined #salt
22:06 anotherZero joined #salt
22:07 tmclaugh[work] joined #salt
22:08 spookah joined #salt
22:11 bhosmer_ joined #salt
22:13 FeatherKing joined #salt
22:15 ahammond on the command line, you can use --batch-size 25%. What is the correct syntax for using this in an orchestration?
22:15 ahammond I included it as part of the salt.state and got the following error: 'batch-size' is an invalid keyword argument for 'salt.state'. If you were trying to pass additional data to be used in a template context, please populate 'context' with 'key: value' pairs. Your approach will work until Salt Carbon is out. Please update your state files.
22:18 iggy ahammond: doesn't look like you can
22:18 ahammond iggy well... that's a bummer
22:19 vexati0n so --- re: states... is there a way that i can have dynamic paths specified in a file source? e.g. i have some folders in the file_root tree named after specific minions. how can i, in a state file, say "get /some/file from salt://root/<YOUR_ID>/some/file" ?
22:19 baweaver joined #salt
22:19 iggy vexati0n: - source: salt://path/to/{{ grains['nodename'] }}/somefile.conf
22:20 vexati0n oh. ok. thx. simple enough
22:21 iggy ahammond: https://github.com/saltstack/salt/issues/12556
22:26 sunkist joined #salt
22:27 murrdoc joined #salt
22:28 ahammond can the reactor at least call the orchestration with batching?
22:28 ahammond I'm looking for ways to not have every system pound on my rpm repository at once...
22:31 Tyrm joined #salt
22:34 bfoxwell joined #salt
22:45 evidence so i noticed last week 2015.5.2 was tagged, but the docs still say release TBD, and the packages don't seem to have been generated.  is more work being down, or is the ubuntu ppa just not getting love yet?
22:45 evidence the artifact was put up, the FreeBSD port is actually already on 2015.5.1
22:45 evidence sorry 2015.5.1 was tagged **
22:46 iggy evidence: no, that's just how Salt does it (and nobody else that I've ever seen)
22:46 forrest_ joined #salt
22:46 evidence does what?  waits a week to generated the packages for the ppa?
22:47 evidence s/generated/generate/
22:47 spookah they create a tag and then keep developing before cutting the release?
22:47 evidence oh.. it's been idle for a week though?
22:48 evidence guess they might just merge forward before cutting it though.  makes sense.  odd that fbsd already bumped to it
22:48 Tyrm joined #salt
22:49 otter768 joined #salt
22:50 Singularo joined #salt
22:54 jhauser joined #salt
22:56 cruatta_ joined #salt
22:58 subsignal joined #salt
23:02 lloesche joined #salt
23:03 julez joined #salt
23:05 lloesche is there a state that allows me to recursively delete a directory, other than calling rm -rf
23:07 ahammond lloesche file.absent
23:07 lloesche will that delete entire directory trees?
23:08 c10 joined #salt
23:08 murrdoc yeah
23:08 ahammond lloesche that's what the manual says
23:08 lloesche nice, thank you!
23:09 ajw0100 joined #salt
23:10 ahammond I'm having trouble with event propegation. When I run salt-call event.send on a minion, the event does not appear on the master, however events around the running of the salt-call command do appear. When I run the same command on the salt master, the event appears and reactor does it's thing.
23:10 ahammond what do I have to do to get events to arrive from my minion?
23:15 cruatta joined #salt
23:18 schuckles joined #salt
23:18 iggy evidence: they tag and wait for distros to have the releases ready, then release tarballs, packages, pip, etc. all on the same day
23:19 mosen joined #salt
23:19 iggy ahammond: event.fire_master?
23:19 subsignal joined #salt
23:20 iggy I have no earthly idea what the difference between event.send and event.fire_master are
23:20 ahammond yeah, I'm confused, too. I'll trying the fire_master one.
23:20 andrej So I am now trying to run my (modified by following a template) bit of python as a module.  When I invoke it via salt-call I get "'oobling' __virtual__ returned False" ... what did I overlook?
23:20 Opti98 joined #salt
23:20 FeatherKing joined #salt
23:21 iggy andrej: can you gist the code?
23:21 andrej sure can iggy ... gimme a sec
23:22 Opti98 When I upgrade from Salt 2014.7.1 to 2015.5, It cleans out my /etc/salt/roster file so then I cannot use salt-ssh without repopulating it. Am I missing something or is this a bug that needs to be reported?
23:23 iggy Opti98: I'd say report it
23:23 iggy it could be a side effect of the new rosters (cache or something)
23:23 iggy but it shouldn't do that
23:23 andrej https://gist.github.com/tinkster/6311cb2f12567f62efef
23:24 iggy you don't have a __virtual__ function
23:24 andrej Fruit donations (rotten tomatos) welcome
23:24 Opti98 Will do. I snapped shoted my VM and rolled it back a few times to just verify. Does it every time.
23:24 iggy that probably means you got a traceback and salt ditched it
23:25 andrej Hmmm ... is that required? ... maybe my "template" I found online is too old :}
23:25 iggy Opti98: it'd be interesting to try backing up /etc/salt, upgrading, waiting till roster gets wiped and then copying the old files back into place and seeing what happens
23:25 andrej I'll search for a more recent code sample for modules
23:25 iggy andrej: not strictly, but it's a good idea
23:26 andrej thanks iggy
23:26 iggy andrej: did you trying running that standalone?
23:28 KyleG joined #salt
23:28 KyleG joined #salt
23:28 Opti98 iggy: I just tried that, I backed up the roster file, upgraded and put the roster file back in. test.ping works to all my hosts which means all of the keys are still in place.
23:29 iggy so it's the upgrade that wipes it somehow (not a general Salt 2015.5 blowing it away)
23:29 Opti98 I have upgrade in the past (Nothing that I can prove now) but I have been using salt-ssh for a while and haven't had the issue.
23:29 iggy package manager should prevent it being overwritten, so it's probably not a packaging thing
23:30 iggy weird
23:30 Opti98 Ill open a an issue
23:30 iggy see if there's one already
23:30 ek6 well package manager should prompt you before overwriting unless you have set a variable so that it wont
23:31 Opti98 It does create an /etc/salt/master.rpmnew since masters was already there, but there are no notifications about roster already being there or that its going to be overwritten
23:32 murrdoc u dont have a state to recreate the roster ?
23:32 andrej iggy: what do you mean by standalone?
23:34 Opti98 murrdoc: Why would I need a state for the roster that is already static? I guess I don't understand the question.
23:34 andrej duh
23:34 andrej thanks iggy
23:34 andrej :}
23:34 murrdoc Opti98:  sorry, dumb question, i have master/minion setup, and i use a state on the master to setup the roster
23:35 murrdoc so that salt-ssh is a fallback
23:35 iggy that's what the cache roster is for
23:35 Opti98 Oh. Many of my targets cannot talk back to my salt-master, so we use salt-ssh directly for them as the master can get to those minions.
23:37 dalexand_ joined #salt
23:39 aparsons joined #salt
23:40 ahammond so, event.fire_master on master works, but on minion does not actually create the event.
23:40 murrdoc yeah
23:40 ahammond there are job related events that show up, but not the custom event I'm trying to fire.
23:40 ahammond event.send has exactly the same issue
23:41 ahammond I'm running 2015.5.0
23:41 ahammond murrdoc any ideas?
23:41 ahammond I feel like I must be missing something really obvious here
23:42 aparsons_ joined #salt
23:43 murrdoc salt even.fire_master doesnt work
23:43 murrdoc also event.send ?
23:43 murrdoc cos it should
23:44 ahammond murrdoc yeah, that's what I expected, but it is not working.
23:44 mosen hiya salties
23:46 Aidin joined #salt
23:52 multiball joined #salt
23:57 mrbigglesworth joined #salt

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