Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2016-03-21

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

All times shown according to UTC.

Time Nick Message
00:01 Valfor joined #salt
00:02 rhodgin joined #salt
00:02 Valfor Hey, out of curiousity, is there any inbuilt support for proxying salt-minion through a SOCKS or HTTP proxy of any kind?
00:02 mavhq joined #salt
00:03 feliks joined #salt
00:08 brianfeister joined #salt
00:10 timoguin joined #salt
00:13 edrocks joined #salt
00:21 Glasswalker joined #salt
00:21 amcorreia joined #salt
00:22 Glasswalker Hey, so I'm having a bit of an issue. New server CentOS 7, using svnfs as back-end. Now when trying to apply a state from master to a test minion, it returns "Salt request timed out. The master is not responding."
00:22 Glasswalker In master log, it's throwing an un-handled exception "TypeError: string indices must be integers, not str"
00:24 Glasswalker It throws this error when doing anything (test.ping, test.versions and so on)
00:24 akhter joined #salt
00:24 Glasswalker If I disable the svn back-end things like test.ping and so on work fine
00:24 Glasswalker but of course my states (which live in SVN) are in-accessible
00:26 Nazzy joined #salt
00:26 cpowell joined #salt
00:28 mdupont joined #salt
00:31 mosu_ joined #salt
00:34 josuebrunel joined #salt
00:34 Valfor Sorry Glasswalker, I'm new to Salt myself, so have no clue!
00:36 punkoivan joined #salt
00:43 punkoivan Hi again, guys. I read theory and now go to practice.
00:44 punkoivan try to run first sls from tutor
00:44 edrocks_ joined #salt
00:44 punkoivan and have error
00:44 punkoivan No matching sls found for 'vim.sls' in env 'base'
00:44 punkoivan but test.ping is ok.
00:45 punkoivan When I run it with debug there is no error, and only one warning about insecure logging.
00:46 punkoivan in vim.sls only 2 strings - vim:
00:46 punkoivan pkg.installed
00:47 punkoivan vim.sls in /srv/salt.
00:47 keimlink_ joined #salt
00:50 edrocks joined #salt
00:51 akhter joined #salt
00:51 timoguin joined #salt
00:54 captain_magnus joined #salt
00:54 mTeK i have a minion on ubuntu 14.04 that disconnects when running highstate. This is the 3rd vm that does this, I have even synced the time with my time server.
00:56 hemebond Valfor: Do you mean the minion to master connection.
00:56 hemebond punkoivan: Do you have a top.sls?
00:57 mTeK Yes I can see do down with salt-run manage.down
00:57 hemebond Oh wait...
00:57 mTeK Yes
00:57 mTeK It works fine with 5 other minions
00:57 iceyao joined #salt
00:57 hemebond punkoivan: Would need to see your directory layout and files.
00:57 timoguin joined #salt
00:57 hemebond And the commands you're running.
00:58 punkoivan hemebond:no, I hvaen't, but read thats ok dont have thihs one and run wiith state.sls
00:58 punkoivan salt '*' state.sls vim.sls --log-level=debug
00:58 punkoivan this one.
00:58 punkoivan /srv/salt/vim.sls
00:58 hemebond Hmm, I normally use state.apply. Lemme read up on state.sls
00:59 hemebond What is your file_roots?
01:00 punkoivan file_roots in master config?
01:00 hemebond Yeah
01:00 hemebond Oh.
01:00 hemebond It shouldn't be vim.sls, it should just be vim
01:00 punkoivan It doesn't specify.
01:00 hemebond Okay, default is /srv/salt/ I think.
01:01 hemebond salt '*' state.sls vim
01:01 punkoivan yeah.
01:02 punkoivan four hours with computer. i need sleep (3 am).
01:03 punkoivan Sorry for dumb question and thanks!
01:03 hemebond np
01:04 hemebond Glasswalker: Is your issue similar to https://github.com/saltstack/salt/issues/18048 ?
01:04 Valfor hemebond: yeah, that's the connection I'm referring to :)
01:04 hemebond Valfor: The minion does support HTTP proxy for stuff that does HTTP requests.
01:04 hemebond I don't know about the zeromq connection though.
01:05 Valfor Yeah, it's the zeromq I'm specifically looking at :)
01:05 Valfor Thanks anyway!
01:06 iceyao joined #salt
01:06 Glasswalker homebond: one sec, checking :)
01:08 Glasswalker hemebond, I don't think so. I'm using default naming (trunk, branches, tags) they exist, my base sls files are in trunk. I'm not even defining that config value. Also I'm using 2015.8.x
01:11 Glasswalker Specifically this is the error and stack dump I get: http://pastebin.com/TS57UNqg
01:11 RobertChen117 joined #salt
01:12 Glasswalker This is the config I'm using to configure SVN back-end, it's pretty straightforward: http://pastebin.com/ma8F5cPj
01:17 hemebond Hmm, unfortunately I don't have the setup to test it myself..
01:17 hemebond You'll have to wait for an svnfs user to pipe up.
01:17 bhosmer_ joined #salt
01:19 hemebond Hmm, it looks like svnfs.py was updated recently.
01:21 hemebond Hmm, possibly some sort of config issue.
01:22 hemebond The examples in the docs use svn:// protocol, not http;?/
01:23 edrocks__ joined #salt
01:23 hemebond I have no idea if that is an issue or not, I'm just trying to understand the settings.
01:24 ajw0100 joined #salt
01:27 cpowell joined #salt
01:27 kliquori joined #salt
01:32 izrail joined #salt
01:43 k_sze[work] joined #salt
01:49 joe_n joined #salt
01:52 ageorgop joined #salt
01:53 iceyao_ joined #salt
01:59 racooper joined #salt
02:00 inire joined #salt
02:01 kliquori joined #salt
02:02 iceyao joined #salt
02:04 terratoma joined #salt
02:06 felskrone joined #salt
02:10 NightMonkey joined #salt
02:10 SheetiS joined #salt
02:11 brianfeister joined #salt
02:11 bhosmer_ joined #salt
02:14 kliquori joined #salt
02:17 edrocks joined #salt
02:20 iceyao_ joined #salt
02:21 _JZ_ joined #salt
02:23 quasiben joined #salt
02:26 johnkeates joined #salt
02:27 euidzero joined #salt
02:32 akhter joined #salt
02:36 onlyanegg joined #salt
02:38 catpigger joined #salt
02:53 joe_n joined #salt
03:03 quasiben joined #salt
03:06 bhosmer_ joined #salt
03:11 jfelchner joined #salt
03:11 mTeK What os do most of you use for salt master?
03:12 hemebond mTeK: I'm running it on my workstation, an Ubuntu-based OS.
03:12 jfelchner I have a crucial question. hahahaha If I include the same formula multiple times, does it actually get *executed* multiple times?
03:12 hemebond jfelchner: What do you mean by "include"?
03:12 jfelchner I have a formula that, for some reason is firing a *bunch*.
03:13 evle joined #salt
03:13 jfelchner Like if at the top of a `.sls` I put `include:\n  - 'some_formula.some_file'`
03:13 tampakrap joined #salt
03:13 jfelchner So that I can reference it in a `require` or `watch`.
03:13 hemebond Hmm.
03:14 mTeK I have a master on freebsd and it seems to be working great in my testing setup but I have a ubuntu master and ubuntu minions and they seem to be disconnecting quite alot, I don't want to be running highstates 3 or 4 times.
03:14 hemebond I would expect an error.
03:14 mTeK Sometimes it doesn't reply back and give me status but I can see that states were apllied
03:14 hemebond mTeK: Disconnected? The minion actually loses the connection?
03:14 hemebond mTeK: Lack of response doesn't mean it got disconnected.
03:14 mTeK master shows down
03:15 hemebond The master will only wait a certain amount of time for a response. If the minion is crashing, have you run it in debug mode to see the output?
03:15 mTeK then about 1min later they come back
03:16 mschiff joined #salt
03:16 mschiff joined #salt
03:16 jfelchner hemebond it definitely causes problems.  I would 100% expect that if I use an `include` it means "do this if it hasn't already been included, but just reference it if it has"
03:16 mTeK Not yet, only my second full day with salt
03:16 jfelchner Like I can absolutely see a scenario where you just need to reference another formula.
03:17 jfelchner Maybe I'm doing it wrong, but I don't know of another way to reference another formula/sls without including it first.
03:19 edrocks_ joined #salt
03:20 cliffstah joined #salt
03:20 Jarus joined #salt
03:21 iceyao joined #salt
03:21 hemebond jfelchner: Include should just include it (and apply it) but shouldn't trigger multiple times.
03:23 edrocks joined #salt
03:25 mTeK hemebond: It workes fine in debug mode. The minion and master have been stopped for awhile, That shouldn't fix anything.
03:26 antpa joined #salt
03:35 alemeno22 joined #salt
03:38 jfelchner hemebond hmmmmmm ok, so that's good I guess, but it's definitely executing multiple times.  Know any way I could debug that?
03:39 jfelchner Is there like a mode where I can see what is going to execute and in what order but not actually run it?
03:40 cliffstah joined #salt
03:40 hemebond jfelchner: You can run highstate with test=true and it _shouldn't_ make any changes.
03:40 jfelchner hemebond Boom!  Sweet I'll try that.  Gracias!
03:40 hemebond There is also state.show_sls
03:41 mohae_ joined #salt
03:41 hemebond Which will list all the states that will apply I think.
03:43 cberndt joined #salt
03:43 cliluw joined #salt
03:45 Zachary_DuBois joined #salt
03:49 cliffstah joined #salt
03:53 Jarus joined #salt
03:59 bhosmer_ joined #salt
04:00 brianfeister joined #salt
04:02 cliffstah joined #salt
04:06 eichiro joined #salt
04:08 joe_n joined #salt
04:11 Jarus joined #salt
04:15 kliquori joined #salt
04:19 aixtools joined #salt
04:19 alemeno22 joined #salt
04:20 aixtools o/, just a bit of feedback (update from last night)
04:21 edrocks joined #salt
04:21 RobertChen117 joined #salt
04:21 aixtools after getting libsodium installed, and correcting /opt/lib/python2.7/site-packages/libnacl/__init__.py (version .18, e.g.), plus some other AIX specfic stuff
04:22 aixtools the command: ]salt-call local state.highstate
04:22 aixtools ends with:
04:22 aixtools [WARNING ] The tools 'dmidecode', 'lspci' and 'dmesg' failed to execute because they do not exist on the system of the user running this instance or the user does not have the necessary permissions to execute them. Grains output might not be accurate.
04:22 aixtools 'local' is not available.
04:22 brianfeister joined #salt
04:22 aixtools and these are all "gnu/linux" commands not found on AIX.
04:23 cpowell joined #salt
04:25 kshlm joined #salt
04:25 jfelchner hemebond awesome.  That helped me track down the problem.  Thanks again!
04:27 scooby2 joined #salt
04:28 hemebond jfelchner: Ooh, what was the issue?
04:29 jfelchner hemebond something completely unrelated, but those helped me figure out it wasn't running things multiple times. :D
04:29 hemebond lol, excellent
04:29 jfelchner I accidentally had the wrong file in a `creates` statement, so it wasn't running it because the (incorrect) file already existed.
04:30 hemebond Aha
04:30 euidzero joined #salt
04:34 joe_n joined #salt
04:50 RobertChen117 joined #salt
04:50 rem5 joined #salt
04:57 RobertChen117 joined #salt
04:58 cyteen_ joined #salt
04:58 brianfeister joined #salt
05:00 kliquori joined #salt
05:12 SheetiS joined #salt
05:13 TTimo joined #salt
05:14 debian112 joined #salt
05:15 Patch joined #salt
05:19 ninjada joined #salt
05:24 cyteen_ joined #salt
05:27 i90rr joined #salt
05:48 bhosmer_ joined #salt
05:57 anmol joined #salt
06:06 fgimian joined #salt
06:18 ninjada_ joined #salt
06:25 cpowell joined #salt
06:31 ninjada joined #salt
06:33 mrmak joined #salt
06:34 mrmak hi
06:35 rdas joined #salt
06:35 hightekvagabond joined #salt
06:51 cliffstah joined #salt
06:58 colttt joined #salt
06:58 antpa joined #salt
07:02 kliquori joined #salt
07:04 slav0nic joined #salt
07:08 alemeno22 joined #salt
07:13 brianfeister joined #salt
07:16 Rumbles joined #salt
07:22 joe_n joined #salt
07:26 jhauser joined #salt
07:26 hemebond Anyone around able to help me with tests/unit/pillar_test.py? Seems to be a bit of a mess and not working the way I expect.
07:27 ninjada joined #salt
07:27 edrocks_ joined #salt
07:29 eseyman joined #salt
07:40 fooma joined #salt
07:48 kliquori joined #salt
07:49 ninjada joined #salt
07:50 fooma joined #salt
07:51 opdude joined #salt
07:51 RobertChen117 joined #salt
07:54 bhosmer_ joined #salt
07:56 KermitTheFragger joined #salt
07:58 edrocks joined #salt
08:00 elsmo joined #salt
08:06 kshlm joined #salt
08:19 dgutu joined #salt
08:20 punkoivan joined #salt
08:20 onlyanegg joined #salt
08:21 punkoivan joined #salt
08:22 yuhlw_ reading this doc https://docs.saltstack.com/en/latest/topics/targeting/pillar.html, I try using "salt -I" ( this is a i) and I'm not able to use it. For the purpose, I've created a pastebin : http://pastebin.com/KBiR3zCJ
08:23 hemebond yuhlw_: role:production is a list.
08:23 hemebond What happens if you set role:production to just a string first?
08:23 punkoivan joined #salt
08:24 jab416171 joined #salt
08:25 punkoivan joined #salt
08:26 punkoivan joined #salt
08:27 yuhlw_ hemebond: so what should it be ?
08:27 hemebond A dict.
08:28 yuhlw_ A dict ??? so how does command will work "salt -I 'foo:bar:baz*' test.ping"
08:28 hemebond role: \n production: lsbioweb1.unige.ch
08:28 hemebond foo: \n bar: bazzar
08:30 yuhlw_ hemebond: ok.... I thought that the * on "foo:bar:baz*", tells that it takes each element of the list that matches.
08:30 hemebond Nah, just a glob.
08:30 iceyao joined #salt
08:30 hemebond *it's just a glob
08:34 yuhlw_ hemebond, I think that I'm a bit stupid, as I'm not able to have a config that works
08:35 hemebond It'll click, don't worry. Let me put together a config that works.
08:36 yuhlw_ http://pastebin.com/r2GjEAzb
08:37 yuhlw_ hemebond: and I'll update the doc on saltstack
08:37 hemebond You might also need to do a salt '*' saltutil.pillar_refresh
08:37 hemebond Which forces the minion to update its pillars.
08:37 antpa joined #salt
08:40 josuebrunel joined #salt
08:41 hemebond Did that help?
08:41 ninjada joined #salt
08:41 linjan_ joined #salt
08:44 hemebond I just tested with your pillar data and it worked for me.
08:45 hemebond But you will need to refresh the pillar data to make sure the minion knows about it.
08:45 hemebond (don't ask me why the minion needs to know about its pillar data for the master to target it)
08:48 bhosmer_ joined #salt
08:51 joe_n joined #salt
08:51 keimlink joined #salt
08:53 fredvd joined #salt
08:55 bdrung_work joined #salt
09:00 edrocks_ joined #salt
09:02 brianfeister joined #salt
09:03 remyd1 joined #salt
09:04 dmaiocchi joined #salt
09:11 SpX joined #salt
09:11 JohnnyRun joined #salt
09:14 joe_n joined #salt
09:15 babilen hemebond: The master probably updated its own data about the minion at the same time.
09:17 euidzero joined #salt
09:19 Rumbles joined #salt
09:21 sfxandy joined #salt
09:24 GreatSnoopy joined #salt
09:25 brianfeister joined #salt
09:30 hemebond babilen: Yeah, still seems odd. And very annoying.
09:31 dkrae joined #salt
09:31 hemebond It's rarely mentioned in the docs that you might have to refresh the pillar data. I still forget to do it.
09:32 babilen hemebond: I typically run "salt-run fileserver.update", "salt '*' saltutil.refresh_pillar" and "salt '*' saltutil.sync_all" before I do things manually (just to make sure)
09:32 lero joined #salt
09:36 hemebond Cripes.
09:36 hemebond Doesn't sync_all do a pillar_refresh?
09:38 JohnnyRun joined #salt
09:40 felskrone joined #salt
09:42 bhosmer joined #salt
09:44 yuhlw_ hemebond: refreshing the pillars did not help. :(
09:44 iceyao_ joined #salt
09:44 babilen hemebond: My understanding is that it syncs all custom modules to the minion and reloads modules on the minion. The order is therefore ""salt-run fileserver.update", "salt '*' saltutil.sync_all" and finally "salt '*' saltutil.refresh_pillar"
09:44 hemebond Ah I see.
09:45 hemebond yuhlw_: That's strange. Try the commands babilen just pasted :-D
09:45 babilen heh
09:45 hemebond And paste your files.
09:49 kliquori joined #salt
09:51 yuhlw_ hemebond: http://pastebin.com/U8PR3PRP
09:51 yuhlw_ hemebond: Do you need to see other files ?
09:51 hemebond try
09:52 hemebond salt -I "role:production:lsb*" test.ping
09:52 yuhlw_ it does a ping on many machines
09:53 hemebond On the lsb* machines?
09:54 yuhlw_ http://pastebin.com/DZiS5ehE
09:54 hemebond Compare my target string to yours.
09:54 hemebond See how I'm targeting the value of production?
09:54 * babilen doesn't see the command at all
09:54 babilen (and recommends one of http://refheap.com, http://paste.debian.net, https://gist.github.com, http://sprunge.us, … )
09:54 yuhlw_ I see, but the result (on my production) is that it ping every machines
09:55 hemebond yuhlw_: Yes, because you've given the pillar to every minion.
09:55 hemebond That's what '*' means.
09:55 hemebond All minions.
09:57 yuhlw_ http://paste.debian.net/hidden/b5e8ca71/
09:57 hemebond Yes, every minion has that pillar value.
09:57 hemebond Your top.sls is applying the test.sls pillar file to '*' which means all minions.
09:58 yuhlw_ hemebond: I'm kind of lost. I thought that the "-I" restrict the list of machine to who we ask the ping for ?
09:58 losh joined #salt
09:59 yuhlw_ I'll see,
09:59 hemebond Okay so the top.sls applies the pillar files (dusergroup, smart, etc) to a certain set of minions; in this case every minion.
09:59 hemebond So now all your minions have all that pillar data.
09:59 yuhlw_ hemebond: I got it !
09:59 kawa2014 joined #salt
10:01 yuhlw_ hemebond: I think that I, somehow, made a big mistake as all my minions can see the configuration of any other of my minion.
10:01 hemebond Yeah? Clicked?
10:01 yuhlw_ hemebond: It, almost, hurts :)
10:01 hemebond They're not seeing the config (you mean pillar?) of other minions, they're all getting the same pillar data because of your top.sls.
10:02 yuhlw_ Yeah, that is better formulated  !
10:03 josue joined #salt
10:03 euidzero joined #salt
10:04 ninjada joined #salt
10:07 josuebrunel joined #salt
10:07 VSpike_ joined #salt
10:07 VSpike joined #salt
10:09 josuebrunel joined #salt
10:11 yuhlw_ hemebond: Somehow, I should have programatic pillar, so that the pillar adjust it's content per minion !
10:12 babilen Could you elaborate on that?
10:13 hemebond yuhlw_: Have you been through the tutorial?
10:13 hemebond Do you understand how to target a minion by its ID?
10:14 hemebond I suggest starting with targeting a minion or two by their ID in your pillar top.sls.
10:23 cpowell joined #salt
10:33 yuhlw_ babilen: my pillars are a big config file for everyone.
10:33 yuhlw_ Then in my state, I somehow do the job that should be done in pillar.
10:33 Fusepumper joined #salt
10:34 babilen yuhlw_: Yeah, that's exactly the wrong way around :)
10:34 Fusepumper When will 2016.03 be released?
10:35 babilen yuhlw_: You should aim to write rather generic states that are then tailored to your needs by providing suitable pillar data.
10:35 yuhlw_ babilen, that's the reason I previously say that "it hurts" :)
10:35 kliquori joined #salt
10:35 babilen Fusepumper: My impression was "end of June", but that depends on it being ready
10:36 dendazen joined #salt
10:36 chmod666org joined #salt
10:36 yuhlw_ babilen: for example in one of my state, I have a: hostconfig=__pillar__["dhost"].get(hostname,{}).get("hostconfig", "unix")
10:36 bhosmer joined #salt
10:38 dariusjs joined #salt
10:38 yuhlw_ As you see I choose a config for my hostname in a state. *This is a terrible mistake !!*
10:39 babilen Yeah, you should just make sure that the 'unix' pillar contains the data you want for that minion
10:40 babilen Pillar data is minion specific and you should take advantage of that fact as often as you can.
10:40 RobertChen117 joined #salt
10:41 Fusepumper Has anyone tested the development branch of salt 2016.3?  and major issues with it?
10:49 teryx510 joined #salt
10:59 Lutz__ joined #salt
11:01 amcorreia joined #salt
11:04 favadi joined #salt
11:05 _Cyclone_ joined #salt
11:12 _Cyclone_ joined #salt
11:22 kbaikov joined #salt
11:22 AndreasLutro Fusepumper: I use it with salt-ssh for personal use, seems to be ok
11:31 bhosmer joined #salt
11:31 quasiben joined #salt
11:35 edrocks joined #salt
11:36 andrew_v joined #salt
11:47 Lionel_Debroux joined #salt
11:55 punkoivan joined #salt
11:55 evle joined #salt
11:57 punkoivan joined #salt
11:57 punkoivan left #salt
11:58 punkoivan joined #salt
11:59 punkoivan joined #salt
12:01 punkoivan joined #salt
12:02 iceyao joined #salt
12:02 punkoivan joined #salt
12:05 punkoivan joined #salt
12:06 punkoivan joined #salt
12:07 LondonAppDev joined #salt
12:08 punkoivan joined #salt
12:09 punkoivan joined #salt
12:09 metalseargolid joined #salt
12:10 punkoivan joined #salt
12:11 izrail joined #salt
12:11 punkoivan joined #salt
12:13 punkoivan left #salt
12:21 onlyanegg joined #salt
12:22 bluenemo joined #salt
12:30 slav0nic joined #salt
12:34 kliquori joined #salt
12:36 TooLmaN joined #salt
12:36 dyasny joined #salt
12:36 kawa2014 joined #salt
12:37 edrocks_ joined #salt
12:45 Adi1 joined #salt
12:49 rem5 joined #salt
12:53 mavhq joined #salt
12:57 fuzzy_id joined #salt
12:57 fuzzy_id hi there
12:57 josuebrunel joined #salt
12:58 quix joined #salt
12:58 gh34 joined #salt
12:58 rhodgin joined #salt
13:00 josuebrunel joined #salt
13:01 JohnnyRun hi all. In my new network I'm going to use salt. All test with vagrant are ok. I would like to replace
13:01 anmol joined #salt
13:01 JohnnyRun zabbix with something  else. Can you suggest something?
13:02 JohnnyRun I need monitoring and graphing
13:02 dijit that's not related to salt at all.
13:03 JohnnyRun dijit: I see that someone use the salt "channel" to deliver metrics.. I think it's salt related.. that's why I ask here
13:05 subsignal joined #salt
13:07 XenophonF joined #salt
13:07 XenophonF joined #salt
13:08 edrocks joined #salt
13:10 teryx5101 joined #salt
13:14 cpowell joined #salt
13:16 favadi joined #salt
13:18 subsigna_ joined #salt
13:19 bhosmer joined #salt
13:20 Hetman Hello I've conflicting ID error in my SLS. I've got elasticsearch: <something> as a settings to my cluster, but later I've got kibana: settings: elasticsearch: host ... so elasticsearch is conflicting ID even if it's not top level . Any idea howto get aroud this ?
13:21 rhodgin joined #salt
13:23 Hetman fixed my mistake
13:27 jhauser joined #salt
13:27 VSpike Is there a way to do a file.recurse with clean: True while excluding a specific directory?
13:28 VSpike I want to deploy a directory tree and clean the target, but also keep files in one specific subdir (/etc/nginx/sites-enabled in this case)
13:29 ThomasJ joined #salt
13:31 mapu joined #salt
13:32 andrew_v joined #salt
13:37 ALLmightySPIFF joined #salt
13:40 molen3 joined #salt
13:41 marnom joined #salt
13:42 DammitJim joined #salt
13:42 kliquori joined #salt
13:42 marnom hi guys, is it correct that salt-cloud providers need to contain plaintext credentials and cannot be GPG encrypted or something like the pillardata?
13:45 _JZ_ joined #salt
13:48 racooper joined #salt
13:52 SheetiS joined #salt
13:55 roock joined #salt
13:55 hasues joined #salt
13:55 hasues left #salt
13:57 punkoivan joined #salt
13:57 punkoivan left #salt
13:58 jerredbell joined #salt
14:01 Tanta joined #salt
14:01 timoguin joined #salt
14:02 totte joined #salt
14:04 mavhq joined #salt
14:04 totte joined #salt
14:04 mavhq how do I override zones_source_dir in https://github.com/saltstack-formulas/bind-formula/blob/master/bind/map.jinja with a pillar?
14:05 totte joined #salt
14:05 Guest19218 joined #salt
14:05 Guest19218 left #salt
14:06 Guest19218 joined #salt
14:06 ht joined #salt
14:07 lero is there a way to not print colors in salt output?
14:07 ht joined #salt
14:07 ht left #salt
14:08 XenophonF lero: i think it's --no-color or something like that
14:08 XenophonF man salt should tell you
14:09 XenophonF there are other output options, too
14:09 XenophonF --out i think, like --out=json
14:10 XenophonF just in case you want the return results structured in a different manner (e.g., for scripting)
14:10 numkem joined #salt
14:12 lero yeah, the --no-color worked
14:12 lero thanks
14:13 lero is there a way to force it inside a python code that salt is running?
14:13 lero i'm just using "return" on it and then it appears green
14:13 akhter joined #salt
14:14 thejrose1984 joined #salt
14:16 CeBe1 joined #salt
14:16 akhter_1 joined #salt
14:18 bhosmer joined #salt
14:24 scoates joined #salt
14:25 zmalone joined #salt
14:25 ronnix joined #salt
14:30 akhter joined #salt
14:32 kaptk2 joined #salt
14:34 mavhq joined #salt
14:35 shiriru joined #salt
14:36 JohnnyRun joined #salt
14:40 SheetiS joined #salt
14:40 akhter joined #salt
14:43 mpanetta joined #salt
14:46 keltim joined #salt
14:47 keltim_ joined #salt
14:49 punkoivan joined #salt
14:50 punkoivan left #salt
14:51 XenophonF lero: sorry not following you - inside what python code? do you want a different compiled-in default?
14:51 punkoivan joined #salt
14:52 punkoivan joined #salt
14:53 punkoivan joined #salt
14:54 punkoivan joined #salt
14:54 Cablegunmaster joined #salt
14:55 Cablegunmaster hey im trying to run salt but its giving me a ascii error, while it needs to be utf-8
14:55 Cablegunmaster is there a environment variable im forgetting?
14:55 punkoivan joined #salt
14:56 babilen I guess that the error is actually something else. Could you paste the problematic state/pillar/... to one of http://refheap.com, http://paste.debian.net, https://gist.github.com, http://sprunge.us, … or elaborate on the actual problem you are seeing?
14:57 punkoivan joined #salt
14:57 dariusjs_ joined #salt
14:58 punkoivan joined #salt
14:59 punkoivan joined #salt
15:01 DammitJim joined #salt
15:01 rm_jorge joined #salt
15:01 racooper joined #salt
15:02 Cablegunmaster pip install works but the installation with salt doessnt work
15:02 Cablegunmaster something with ascci and utf-8
15:02 thejrose1984 joined #salt
15:03 akhter_1 joined #salt
15:03 Cablegunmaster how to force it in utf-8
15:05 anmol joined #salt
15:08 Ron1 joined #salt
15:08 babilen Cablegunmaster: Installation of what?
15:08 Ron1 HI All
15:09 Ron1 I have a question
15:09 babilen Okay
15:09 Ron1 I run schedule in the minion and the master and it does not work
15:09 Ron1 http://paste.debian.net/417710/
15:09 babilen cmd.runi != cmd.run
15:10 babilen fwiw, I also don't like to use states for schedules. You can simply define them in pillar data and they'll work
15:10 timoguin joined #salt
15:10 babilen That is documented on https://docs.saltstack.com/en/latest/topics/jobs/schedule.html which shows some examples of suitable "schedule:" pillar data
15:11 Ron1 thank you babilen
15:11 babilen If you define schedule in pillars you don't have to re-run states all the time simply to update them.
15:12 Ron1 How can I define in pillar?
15:12 babilen Have you worked with pillars before?
15:12 Ron1 yes
15:12 babilen So you know how to target data to minions with it?
15:12 Brew joined #salt
15:13 babilen Schedules simply look at entries in the "schedule:" pillar key. If you have data defined there the jobs defined therein will be automcatically executed.
15:14 babilen There is no need to "activate" or "run" them
15:14 LondonAppDev__ joined #salt
15:14 Ron1 So I just need to add a new pillar with the name schedule?
15:15 babilen Exactly
15:15 Ron1 ok.
15:15 murrdoc joined #salt
15:15 babilen Examples of suitable pillars are mentioned on https://docs.saltstack.com/en/latest/topics/jobs/schedule.html
15:15 Ron1 thank you
15:15 Cablegunmaster babilen, got it sorted :).
15:15 babilen This seems to be a bit too easy for some people :)
15:16 babilen Cablegunmaster: Ah, what was it?
15:16 Cablegunmaster testing it now, LC_ALL not set and all the others aren't yet so just need it for 1 bit so temp set it in the command.
15:17 babilen I still have no idea what you are actually doing, but if you are happy I'm happy :)
15:17 Cablegunmaster Yep !
15:18 Ron1 babilen, I can't find how  can exactly add them in the schedule.html doc
15:18 fredvd joined #salt
15:19 babilen Ron1: You create a file "/srv/pillar/foo.sls" with content such as http://paste.debian.net/417722/ and then include "- foo" in /srv/pillar/top.sls
15:19 babilen (for the minion you want to target)
15:19 babilen Adjust as needed
15:20 Ron1 oh
15:20 onlyanegg joined #salt
15:20 Ron1 Thank you very much
15:20 babilen You literally target suitable data to the minion .. there is nothing else you have to do.
15:21 cyborg-one joined #salt
15:21 Ron1 and when I change the foo.sls
15:21 Ron1 and refresh pillar
15:21 inire joined #salt
15:21 Ron1 then it will be change automatically?
15:21 babilen .. the schedule is refreshed
15:21 babilen yeah
15:21 Ron1 :)
15:21 Ron1 greate
15:21 Ron1 great
15:22 babilen Yeah, this is one of the moments in which you realise: "Hmm, why make it more complicated than that?"
15:22 Ron1 haha :)
15:23 XenophonF joined #salt
15:24 Ron1 Yes, I think this example must be added to the document
15:25 vilitux joined #salt
15:26 Ron1 Can I call to pillar without using top.sls file?
15:26 babilen no
15:26 Ron1 but in state I can , right?
15:26 babilen You can use state.sls to call states directly, yeah
15:27 babilen But pillars aren't executed .. they are simply there.
15:27 Ron1 oh I understand
15:27 Ron1 thank you
15:28 Ron1 Are you salt developer?
15:28 babilen No
15:29 Ron1 :)
15:29 Ron1 Your'e helpping a lot here
15:30 Ron1 Well done
15:30 babilen Thank you
15:31 Ron1 Your'e welcome
15:31 Ron1 :)
15:32 Ron1 How can I direct the std error of salt
15:32 Ron1 when I run salt -G ipv4:<some ip>
15:32 Ron1 If the ip isn't exist I get an error
15:32 Ron1 But if I am using a perl script to run it
15:33 Ron1 I want to redirect the error, not from perl
15:33 Ron1 I can do it, from perl but I have look for, a salt way
15:34 lero XenophonF: there are some python code inside some files and it's output by default is green.. but it's ok, i'll just use --no-color :)
15:36 SheetiS joined #salt
15:36 anmol joined #salt
15:37 Hetman guys is syntax {{ somevar|default('default_value') }} standard YAML syntax ? I'm looking on some formula that using formhelper module but wondering is this is actualy core or this module
15:39 babilen Hetman: http://jinja.pocoo.org/docs/dev/templates/#default
15:40 buhman Hetman: yaml syntax?
15:40 buhman Hetman: you mean jinja syntax
15:40 euidzero joined #salt
15:40 buhman and that's a jinja thing, not a salt thing
15:41 euidzero joined #salt
15:41 buhman (also not syntax)
15:42 numkem joined #salt
15:42 relidy joined #salt
15:43 spankalish joined #salt
15:44 Hetman Hetman: yeah jinja , ok i was just wondering can I start using it without any custom modules etc, as it's core language thing I should be fine
15:45 spankalish Hi could someone please explain this part of the haproxy.cfg.jinja file?  https://gist.github.com/Spankalish/3a15d7c5c3546ae4b0c6
15:45 mgresser joined #salt
15:46 onlyanegg joined #salt
15:46 spankalish I got this form an example online, It gets the ips from the salt mine, my webservers are called minion1, minon2, minion3.....ect
15:47 metalseargolid joined #salt
15:47 XenophonF lero: you can disable output color in the minion config iirc
15:48 spankalish Anyone?7
15:49 babilen spankalish: What do you want to know?
15:49 babilen I wouldn't implement it like that, but I'm not sure what to explain either :)
15:50 spankalish It's an example I found babilen
15:51 Rale joined #salt
15:53 spankalish I want the format to be like https://gist.github.com/Spankalish/e051682224d1f823f2ee
15:53 hightekvagabond joined #salt
15:55 spankalish I've never used salt or jinja config files before so I'm deep in learning here, any help would be appreciated
15:56 darvon joined #salt
15:56 babilen spankalish: https://www.refheap.com/116225 is config I'm using (in slightly different form) in production ..
15:56 spankalish What I would like to know babilen is how do you get the ip addresses of the minions from the salt mine
15:56 babilen You use mine.get for that -- https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.mine.html#salt.modules.mine.get
15:57 babilen I tend to set the arguments to mine.get in the pillar in which I configure the application in question. That way I don't have to edit my states just because the target expression changes (etc)
15:57 spankalish Yes I know, but do you have to specify the name of the minons? Like minon* <-----
15:57 babilen You can target them in whatever way you like
15:58 babilen https://docs.saltstack.com/en/latest/topics/targeting/
15:58 babilen Targeting by glob is default, but you can use whatever you deem most useful
16:01 spankalish Ok so can I ask what part of https://gist.github.com/Spankalish/3a15d7c5c3546ae4b0c6 is targeting the minions? As I don't know what naming structure they are using
16:01 alemeno22 joined #salt
16:01 babilen 'G@role:webserver and G@env:' ~ env, 'internal_ip'
16:02 euidzero_ joined #salt
16:02 babilen Which constructs a compound matcher (cf. https://docs.saltstack.com/en/latest/topics/targeting/compound.html) -- That bit is the "lb.backend_tgt" bit in my config.
16:03 VSpike Is there any way to make the salt-formula's salt.formulas use a specific branch for a specific formula? It looks like you could do it for all the formulas in an environment, but I don't see anything for a single formula
16:04 babilen spankalish: I like to use network.ip_addrs with a specific cidr to get IP addresses into the mine
16:04 babilen (more explicit)
16:06 bhosmer joined #salt
16:06 spankalish Yea I get what you mean babilen
16:06 hightekvagabond joined #salt
16:09 chiui joined #salt
16:12 spankalish I'm currently doing a honors degree in Computer Network Management and Cloud Infrastructure. I'm using Salt in a project I am doing. I really like all DevOps, it's really cool, but it's a large learning curve trying to teach yourself, as well as doing all the rest of the modules in college.
16:15 babilen spankalish: Sure, but then .. so is everything :)
16:15 babilen In the end you will have learned some things that you can use right away
16:16 timoguin joined #salt
16:16 whytewolf o/
16:17 spankalish Dam right. Salt is such a handy tool to learn. Can't wait to finish in college and I can play around with it properly and master it. It's just I don't have that much time at the moment
16:17 adelcast joined #salt
16:17 spankalish Thanks for the help babilen
16:19 amcorreia joined #salt
16:23 writtenoff joined #salt
16:26 antpa joined #salt
16:27 zionsofer joined #salt
16:28 zionsofer Hi, I'm looking for a baseline of how many minions can a single master with X cores handle, assuming it has to run the a single module on all minions concurrently without using --batch-size.
16:31 babilen I don't think that precise measurements like that exist. For some deployments we are simply running jobs asynchronously and look up the result in the job cache later.
16:33 babilen https://docs.saltstack.com/en/latest/topics/tutorials/intro_scale.html might be of interest
16:34 edrocks joined #salt
16:35 wise0wl joined #salt
16:35 UtahDave joined #salt
16:36 wise0wl I am trying to install the release candidate using the bootstrap shell script on centos 7, but am receiving an error: error: Developers, please remove the salt paths configuration setting from the setup's install command
16:37 UtahDave wise0wl: can you pastebin the exact command you ran and the error?
16:38 wise0wl UtahDave: http://pastebin.com/uqUrtzJZ
16:40 ageorgop joined #salt
16:41 UtahDave wise0wl: what version of the bootstrap script are you using? The latest released version?
16:42 wise0wl Yes, I just downloaded before installing
16:43 UtahDave ok, testing right now
16:44 grumm_servire joined #salt
16:44 zionsofer bailen: Thanks, I read the same tutorial and still trying to figure out the ballpark numbers for our configuration. Ie how many worker_threads we want per core on the master or managed minions.
16:46 mgresser Does anyone know if the elasticsearch returner is compatible with elasticsearch 2 or if it only works with 1.5.2 through 1.7.7
16:48 UtahDave wise0wl: I got the same result as you. Let me try the develop version of the bootstrap script
16:48 babilen zionsofer: How many minions do you have?
16:48 zionsofer about 100, managed by dual core machine right now.
16:49 babilen zionsofer: And are you seeing problems with that? I'd use at least 4 cores, but I don't have to pay for them ;)
16:50 zionsofer babilen: when I crossed the ~60 minion mark,  I noticed the some module runs simply do not return, even stuff like test.ping. Oddly enough, running the same module via the state.orch runner would be successful on all targeted minions.
16:51 UtahDave zionsofer: have you checked your job cache?  Most likely the minions are returning successfully.
16:53 UtahDave wise0wl: still there.
16:54 ajw0100 joined #salt
16:55 zionsofer UtahDave: when looking the job outout using salt-run jobs.lookup_jid this looks ok
16:55 wise0wl Yeah
16:55 wise0wl I'm still here UtahDave
16:55 zionsofer at least most of the times
16:56 UtahDave zionsofer: yeah, it's just not waiting long enough on the cli.   You might try adding    '-t 8' to your commands and see if that works.
16:56 UtahDave wise0wl: tracking down what's causing the error
16:59 ronp_usa joined #salt
16:59 punkoivan hi all.
16:59 UtahDave wise0wl: OK. I found out what happened
16:59 samed joined #salt
16:59 samed joined #salt
16:59 UtahDave wise0wl: that was a known issue in that RC and has been fixed in the 2016.3 branch
16:59 punkoivan I run htop for now and saw about 20 salt-master running, what is it?
17:00 Gareth o/
17:00 UtahDave wise0wl: so the next rc won't have that problem once we cut it.  If you install from the 2016.3 branch it should work
17:00 UtahDave hey, Gareth!
17:00 Gareth UtahDave: hey hey :) hows it going?
17:03 zionsofer UtahDave: I actually checked and the test.ping module did not run on minions that show up as ' Minion did not return. [No response] ' on the salt master...
17:04 LondonAppDev joined #salt
17:04 wise0wl OK, thanks UtahDave
17:10 kawa2014 joined #salt
17:12 dhscholb joined #salt
17:12 zionsofer punkovian: it's related to the amount of worker_threads you have in  /etc/salt/master file.
17:13 edrocks joined #salt
17:13 gcorey joined #salt
17:14 mgresser I've also seen returns fail to show up correctly when the high water marks in zeromq aren't adjusted.
17:14 mgresser I usually see that with over 1000 minions though, not with just 60.
17:14 mgresser But it's another thing that's worth looking at.
17:15 punkoivan zionsofer, thanks.
17:16 dhscholb where do i put a custom returner if i want to use it as a master_job_cache? i put it in /srv/salt/_returners, but the salt-master can't seem to find it
17:16 ajw0100 joined #salt
17:17 dhscholb it does work fine with --return
17:18 dhscholb Failed to allocate a jid. The requested returner 'rabbitmq' could not be loaded.
17:18 aljosa joined #salt
17:18 dhscholb the returner does have a prep_jid function
17:20 kutenai joined #salt
17:20 djural joined #salt
17:21 jerredbell1 joined #salt
17:21 whytewolf you need to set https://docs.saltstack.com/en/latest/ref/configuration/master.html#extension-modules to include where the returner is. or move copy the returner to a spot seen by that. _returners is for the sync_all option which does this automaticly for minions
17:23 whytewolf dhscholb: you could also use module_dirs instead of extension_modules
17:23 pfallenop joined #salt
17:23 dyasny joined #salt
17:27 dhscholb whytewolf: thanks. I didn't realize at first that that was a relative path. works now
17:27 beardedeagle joined #salt
17:28 rem5 joined #salt
17:29 baweaver joined #salt
17:31 rem5 joined #salt
17:33 LondonAppDev joined #salt
17:34 alemeno22 joined #salt
17:38 cberndt joined #salt
17:38 zionsofer UtahDave: I actually noticed something weird. When running salt with -b (x), test.ping returns fine and when not using -b, it doesn't. I then looked at salt-run jobs.list_jobs and noticed that when running salt with -b (x), there are two jobs listed: the original one and another one with a list of minions rather than the grain, compound or glob  target_type
17:38 SheetiS joined #salt
17:39 baweaver joined #salt
17:44 pfallenop joined #salt
17:45 kickerdog joined #salt
17:45 kickerdog Does anyone have an apache storm sls they could share?
17:46 kickerdog joined #salt
17:48 jmedinar joined #salt
17:48 UtahDave zionsofer: yeah, each batch gets it's own job id.
17:48 sfz- joined #salt
17:50 zionsofer UtahDave: Oddly enough batches as big as my entire entire minion fleet succeed while running without batch don't come up with results, regardless of timeouts. Am I missing something?
17:51 kickerdog1 joined #salt
17:52 deus_ex joined #salt
17:52 beardedeagle is it possible to throw a custom error inside of a .sls ? not finding anything in the docs except for salts exceptions for itself
17:52 beardedeagle custom error/exception, whichever
17:54 bhosmer joined #salt
17:57 baweaver joined #salt
17:57 jmedinar Q: How can I evaluate the output of a parameter on a state?
17:57 jmedinar Ex:  salt master ps.swap_memory
17:57 jmedinar free:
17:57 jmedinar 10735353856
17:58 jmedinar how can I access that value to run an if against it?
17:59 Ron1 joined #salt
18:01 GreatSnoopy joined #salt
18:01 beardedeagle @jmedinar: you could try this: https://docs.saltstack.com/en/latest/ref/renderers/all/salt.renderers.jinja.html#calling-salt-functions
18:03 ageorgop joined #salt
18:04 linjan_ joined #salt
18:05 insomniaSalt joined #salt
18:08 Ron1 joined #salt
18:08 jmedinar is there any example I can look at ?
18:12 mavhq joined #salt
18:13 sfxandy joined #salt
18:14 punkoivan joined #salt
18:15 punkoivan joined #salt
18:15 jmedinar_ joined #salt
18:15 baweaver joined #salt
18:17 punkoivan joined #salt
18:17 punkoivan joined #salt
18:18 punkoivan left #salt
18:20 rhodgin joined #salt
18:20 jacksontj joined #salt
18:23 Ron1 joined #salt
18:28 zer0def joined #salt
18:34 hightekvagabond joined #salt
18:35 alemeno22 joined #salt
18:44 kickerdog1 left #salt
18:45 beardedeagle i find it odd, based on the documentation at least, that pip_state only allows for the installation of multiple packages in one go with requirements files but not pkgs:
18:45 beardedeagle could be wrong though, so going to dive into the code
18:47 beardedeagle ah looks like it does accept a comma delimited list, but that is nowhere in the docs.
18:51 jfindlay also, all states should be able to work with `names`
18:53 beardedeagle @jfindlay: if using pkgs: it ignores names: Or are you saying they do the same thing?
18:54 bhosmer_ joined #salt
18:54 jfindlay names is a basic service provided for all states, basically running the state multiple times, one for each name
18:55 jfindlay https://docs.saltstack.com/en/latest/ref/states/highstate.html#names-declaration
18:55 jfindlay I'm not quite sure why pkg states also have `pkgs`.  There must be something that `names` doesn't do in that case
18:56 tpaul joined #salt
18:56 beardedeagle ah. depending on what you are doing that could be inefficient though.
18:56 jfindlay yeah, that makes sense
18:57 beardedeagle interesting on the names thing though, did not know that. thanks.
18:58 JohanR joined #salt
18:59 jfindlay sure
19:03 murrdoc joined #salt
19:07 mgresser joined #salt
19:15 west575 joined #salt
19:16 zionsofer UtahDave, babilen: I just wanted to note that I made some more analysis, and noted some nodes show "OSError: [Errno 12] Cannot allocate memory" in the minion logs. When excluding such nodes (or restarting the minion and freeing up memory) things to seem work ok... So now I'm just left with finding out why my salt minion (or neighboring processes)  eat up memory
19:16 UtahDave zionsofer: interesting
19:18 UtahDave jfindlay and beardedeagle:   The Salt State compiler creates an individual state for each item in "names".  In the pkg state "pkgs" will make one call to the package manager with all the packages
19:18 UtahDave whereas with names the package manager will get called once for each package.
19:18 beardedeagle yar, that is what I figured
19:19 zionsofer UtahDave: the weird thing is that on the memory stressed minions, I do see an entry "User root Executing command test.ping" but not followed by "Starting a new job with PID" entry. I guess there's more shown in a debug or trace level.
19:21 martoss joined #salt
19:26 chiui joined #salt
19:27 josuebrunel joined #salt
19:28 baweaver joined #salt
19:32 baweaver joined #salt
19:32 onlyanegg joined #salt
19:38 baweaver joined #salt
19:39 murrdoc joined #salt
19:40 josuebrunel joined #salt
19:41 ronnix joined #salt
19:42 josuebrunel joined #salt
19:45 josuebrunel joined #salt
19:47 fracklen joined #salt
19:48 jnilsson joined #salt
19:51 rem5 joined #salt
19:51 mfournier joined #salt
19:52 josuebrunel joined #salt
19:55 josuebrunel joined #salt
19:58 josue joined #salt
20:02 dhscholb i'm using a custom runner for master_job_cache, and whenever i run "salt-run jobs.lookup_jid <jid>", it seems to be trying to call get_jid
20:03 dhscholb but get_jid isn't a listed requirement for master_job_cache, according to https://docs.saltstack.com/en/latest/ref/returners/
20:03 dhscholb is the doc wrong?
20:03 AndreasLutro https://docs.saltstack.com/en/latest/ref/returners/#external-job-cache-support doesn't this say exactly that?
20:04 mpanetta joined #salt
20:04 cberndt joined #salt
20:04 dhscholb AndreasLutro: i'm not using it for external_job_cache
20:04 dhscholb only master_job_cache
20:06 AndreasLutro ah yeah
20:06 Azid joined #salt
20:06 UtahDave I'm pretty sure the master_job_cache does require get_jid
20:10 baweaver joined #salt
20:11 akhter joined #salt
20:12 akhter joined #salt
20:12 dhscholb so, looks like documentation for master_job_cache might need to be updated
20:14 UtahDave dhscholb: Yeah, I guess that's not clear by looking at that.
20:14 UtahDave dhscholb: returners, external job cache and master job cache all use a lot of the same code
20:14 dhscholb UtahDave: does master_job_cache need all of the external_job_cache functions?
20:14 UtahDave the returner is the inserting minion data part, and the external job cache is used by the master  job cache
20:14 UtahDave dhscholb: yes
20:15 UtahDave the master job cache needs everything from both the returner and external job cache
20:18 riftman joined #salt
20:19 mTeK I just put up a new salt master but my net eng hasn't opened up the ports yet. Can I manually copy over the keys to get this going with out waiting?
20:19 hemebond mTeK: If the ports aren't open, how will the minion connect to the master?
20:20 mTeK Well of the master goes out the firewall wont block it.
20:20 whytewolf master doesn't go out
20:20 whytewolf all minions connect to master
20:20 UtahDave mTeK: yes, you can manually copy over keys to your new master, but the minions reach out to the master, not the other way around
20:20 mTeK Thanks that what I wanted to know
20:21 mTeK I thought when I run highstate it pushed.
20:22 hemebond When the master wants the minions to do anything it just pops a message on the message bus and leaves the minions to pick it up.
20:22 mTeK oh
20:22 mTeK well i guess I have to wait then
20:22 XenophonF mTeK: at the TCP layer the minion must connect to the master
20:22 alemeno22 joined #salt
20:22 XenophonF the TCP connections are bi-directional after that
20:22 XenophonF think HTTP or SSH even though it isn't
20:23 XenophonF an SSH server pushes all kinds of data to the client, even though the client is the one who must initiate the connection
20:24 subsignal joined #salt
20:26 hemebond XenophonF: So the master actually sends the command to the minion?
20:27 hemebond I wonder why I thought it just used a message queue.
20:27 XenophonF i don't know how 0mq is implemented
20:27 XenophonF i think at that layer, the 0mq client polls the 0mq master, but don't quote me on that
20:29 XenophonF http://api.zeromq.org/2-1:zmq-tcp
20:32 XenophonF http://api.zeromq.org/2-1:zmq-socket
20:32 XenophonF well that's more than i ever wanted to know about 0mq
20:33 XenophonF which reminds me to try out raet some time
20:34 whytewolf iirc tornado is the socket layer now
20:34 XenophonF i guess publish-subscribe is only one of several possible 0mq messaging patterns
20:34 whytewolf RAET isn't being maintained and iirc was never finished
20:34 XenophonF oh really? i'm behind the times, i guess
20:34 XenophonF i was interested in it since it used udp
20:35 whytewolf it was a okay idea but ran into issues associeated with UDP
20:35 AndreasLutro isn't there a tornado transport being worked on now
20:35 whytewolf UDP is an okay concept but you can never be sure if it gets the job done
20:35 jfindlay RAET is still maintained, we're making sure tests pass on it before 2016.3.0 is tagged, for example
20:36 whytewolf ahhh okay. just after tornado dropped things got REALLY quiet about RAET
20:36 jfindlay also it may be ported to tornado/TCP at some point
20:36 mavhq joined #salt
20:36 whytewolf like it was swept under a rug
20:37 jfindlay yeah, the engineer who designed it leaving saltstack for another job didn't help :-)
20:37 zmalone Yeah, PRs stopped getting merged, etc.
20:38 whytewolf ahh yeah those pesky engineers and their pet projects :P
20:38 jfindlay but we have engineers maintaining it, and we're continuing to test it
20:38 Ahlee RIP RAET
20:38 Ahlee ;)
20:38 Ahlee I kid, I kid.
20:38 jfindlay RAET is still a cool idea
20:38 jfindlay and we're not going to drop it
20:39 Ahlee It's interesting, but complicated. Not just complex
20:39 jfindlay one of the main reasons from decoupling from 0mq is that with raet, for example, we have no non-python deps
20:39 iggy can you run one master with raet and other minions yet?
20:40 jfindlay the one missing piece being the crypto back end, which is where https://github.com/jfindlay/pure-pynacl comes in
20:41 jfindlay iggy: I'm unsure what you're asking.  You can run a master and minions with RAET the last time I checked
20:41 whytewolf i thought the issue with RAET wasn't a single master but it compleatly broke with multimaster
20:42 whytewolf [more then multimaster already does]
20:42 Ahlee That'd be impressive
20:42 mavhq joined #salt
20:43 brianfeister joined #salt
20:46 iggy Last I heard, you couldn't run zmq and raet minions on the same master
20:50 mTeK joined #salt
20:53 ZappedC64 joined #salt
20:53 beardedeagle I always thought you had to do one or the other
20:53 ZappedC64 Greetings all.
20:55 bhosmer_ joined #salt
20:55 ZappedC64 Does anyone have a link that shows how to change the IP address of a minion?
21:01 UtahDave I'm pretty sure you can run both raet and zmq minions from the same master now
21:02 rhodgin joined #salt
21:05 UtahDave ZappedC64: https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.rh_ip.html#module-salt.modules.rh_ip
21:05 UtahDave for rhel based systems.  There's a debian counterpart as well as others
21:06 notnotpeter joined #salt
21:09 hightekvagabond joined #salt
21:10 aw110f joined #salt
21:11 teatime The easiest thing might be to edit the config and reboot the minion.
21:11 Rumbles joined #salt
21:13 ZappedC64 Ah.. cool. TY.
21:14 ZappedC64 My Google search didn't come up with any useful info. :)
21:16 whytewolf wait ip address of the minion? that doesn't affect the master on it's own.
21:16 whytewolf it would if the minion id changes though
21:16 Nazca__ joined #salt
21:19 teatime I assume the issue is the state or call doesn't finish/return because the minion loses connectivity toward the end (when the IP actually changes.)
21:19 teatime or he just wanted a pointer to a formula or something.
21:21 zahna joined #salt
21:22 ZappedC64 yeah... I need to change the IPs on a bunch of non-production RHEL servers.
21:23 ZappedC64 Not quite sure I understand how to set the actual minion IP with ip.apply_network_settings.
21:23 _Cyclone_ joined #salt
21:24 teatime I assume it applies whatever is in /etc/sysconfig/network
21:24 teatime if it were me, I'd put my IP's in pillars, and make a state that manages /etc/sysconfig/network
21:26 ZappedC64 @teatime - Agreed.  I'll have to read up on pillars.  I'v only just started with Salt.
21:27 teatime see also https://docs.saltstack.com/en/latest/ref/states/all/salt.states.network.html#module-salt.states.network
21:27 teatime which I would use if I was managing different distros, but since I only have one I prefer managing the network config files manually.
21:28 akhter joined #salt
21:30 sfxandy joined #salt
21:31 oscar__ joined #salt
21:34 Ahlee is there a way to arbitrarily specify a file.managed location yet?
21:34 Ahlee better qeustion, is file.managed extendable yet?
21:34 hemebond arbitrarily?
21:35 Ahlee be able to fetch files for file.managed over say an internal API
21:35 Ahlee or really any API
21:35 hemebond You mean programatically?
21:35 Ahlee i.e. file.managed some link in bitbucket
21:35 Ahlee Yes
21:35 hemebond You could use an external pillar
21:35 Ahlee I can't put all my repos in the file system so I can't serve out /some/repo/somefile.jinja
21:36 hemebond Or you could use pydsl for your state.
21:36 hemebond Oh, you mean can you use a non-salt:// source for a file?
21:37 Ahlee I want to plug into the non-salt:// and extend it
21:37 hemebond Sorry, I'm lost. Can you please explain what it is you're trying to achieve?
21:37 Ahlee as I can't/don't want to generate md5 sums for all the files I want to pillarize for a pure source: http://our_endpoint
21:37 Ahlee sure. I have a thousand repos that I want salt to manage one file from each repo
21:38 rem5 joined #salt
21:38 Ahlee sorry, I want to reach into said repo for that one file, and have salt file.managed that one file on each server
21:38 Ahlee so, I can't do it with file.managed: and source: http:// due to A. user/password permissioning, and B. it requires md5sums that aren't available
21:38 hemebond So you want to use a file as a source without knowing its md5.
21:38 Ahlee I can't add all the repos to file_roots, as taht nukes permissions
21:39 Ahlee Pretty much.
21:39 Ahlee Yes yes, security this and that
21:39 Ahlee We're not concerned about that
21:39 Ahlee er, sorry
21:39 Ahlee not nukes permissions. It nukes performance
21:40 hemebond I think if you search for that (file source without md5) you'll find examples and suggestions.
21:40 Ahlee We tried 20 repos with gitfs. It worked, it just took an hour for highstate
21:40 jeddi joined #salt
21:40 Ahlee ah, doing it with cmd.run and curl
21:41 hemebond Yeap./
21:41 Ahlee eh, I guess that could work
21:41 hemebond Since you're not really managing the file, you're just downloading, it should work fine.
21:41 Ahlee yeah, i just lose jinja templating
21:41 hemebond Why is that?
21:41 Ahlee I don't think that's necessarily bad
21:41 oscar__ I'm trying to use file.managed with s3:// protocol but the minion is saying "Could not fetch from s3://" when triggering the state from the master. However if I do 'salt myminion s3.get mybucket myfile' it works fine, so privileges are not the issue. I printed the exception just before it prints the error in fileclient.py and the exception says ['module' object has no attribute 's3']
21:42 Ahlee if the minion curl's the file down, i'd have to pre-render the file before serving it
21:42 hemebond oscar__: I doubt the s3 protocol can be used for file.managed.
21:42 hemebond Use http if possible.
21:42 hemebond Ahlee: Oh it's a jinja file?
21:43 Ahlee from the error it sounds like whatever's providing the s3 object insn't installed on the master host for the minion
21:43 Ahlee hemebond: Yeah, it's logstash configs. We want to move them out of the saltmaster repo into individual application repos. Though that isn't going to work really as the apps will still need to know how to render the configurations out
21:44 hemebond Oh, s3 urls can be used as a source.
21:44 Ahlee It's very not DRY to go this route, but it's a POC to try to make developers' lives easier so they don't have to dip into the operations repo
21:45 hemebond What's the performance issue you ran into?
21:45 SpX joined #salt
21:45 hemebond You don't want to add the repos to gitfs_... thingy?
21:45 Ahlee The number of times the master checks the file system for updates
21:45 oscar__ since it's possible I'd like s3 over http
21:46 Ahlee last count we had 944 repositories, when I put 20 gitfs repos under file_roots highstate went from taking minutes to an hour
21:46 Ahlee that was also hosting all the gitfs repos in a tmpfs filesystem which I also don't have the memory for to hold all of the repos
21:46 oscar__ Ahlee: any clue about that can be missing in the master or minion?
21:47 Ahlee oscar__: are you positive the python environment on the master can talk to s3?
21:47 hemebond Ahlee: Can you clone the repos to the master or is it too much?
21:47 Ahlee hemebond: it'd take a fair amount of disk, but yeah it's not completely out of the question. A simple state at the start that does a git.latest: or similar
21:48 oscar__ I'm not. I'll check
21:48 baweaver joined #salt
21:48 Trauma joined #salt
21:48 hemebond Ahlee: On my master I don't use the gitfs stuff even for formulas. I just clone the repo and then symlink to them from wherever I need them.
21:48 Ahlee Guess that'll be my first task tomorrow, then verifying that if I git.latest in the file_roots repo the 'latest' files are actually available
21:49 hemebond You could use some sort of git hook to make the master pull down changes.
21:49 Ahlee hemebond: yeah, We originally were using gitfs and then moved to just having highstate manage a local checkout of the repo
21:50 Ahlee hemebond: we were hesitant to go that route as we really want to prevent the githook service being down/unavailable/misbehaving causing us to miss an update
21:50 Ahlee but, yeah. That was discussed as well.
21:50 Ahlee hemebond: appreciate your insights. It was helpful
21:50 hemebond Good luck :-)
21:50 kliquori joined #salt
21:52 quix joined #salt
21:54 bhosmer joined #salt
21:56 oscar__ Ahlee: boto is present. what do you suggest to check the s3 is working on python environment at master? also, why should the master be able to talk to s3 if the minion is the one who needs to gather the file?
21:56 ninjada joined #salt
21:57 hemebond oscar__: Doesn't the master fetch the file and give it to the minion?
21:57 ninjada joined #salt
22:00 oscar__ I think the minion fetches the file on master request to apply the managed_file configured in the state
22:00 alxr joined #salt
22:01 oscar__ hemebond: I think the minion fetches the file on master request to apply the managed_file configured in the state
22:03 mavhq joined #salt
22:05 hightekvagabond joined #salt
22:08 baweaver joined #salt
22:09 aw110f joined #salt
22:11 oscar__ it seems the s3 import statement https://github.com/saltstack/salt/blob/develop/salt/fileclient.py#L32
22:11 oscar__ is missing in my version 2015.5.3 (Lithium). Manually adding the line fixes the issue
22:16 lahwran joined #salt
22:19 zahna joined #salt
22:22 jnials_ joined #salt
22:23 kliquori joined #salt
22:23 Muchoz joined #salt
22:27 jaybocc2 joined #salt
22:29 oscar__ however it's present here: https://github.com/saltstack/salt/blob/2015.5/salt/fileclient.py#L30
22:30 ahammond any news on the 2016.3 release?
22:30 timoguin joined #salt
22:30 ajw0100 joined #salt
22:31 zmalone I've been watching, but I have not seen anything recently
22:33 kliquori joined #salt
22:36 brianfeister joined #salt
22:38 djgerm1 joined #salt
22:40 zenlot joined #salt
22:40 djgerm1 what's the best way to protect DNS zones from being over written with the initial salt state that have ddns additions that are not part of the state at all (but are part of the orchestration events that happen when new minions are added to the master)?
22:40 djgerm1 i was thinking using an unless under file.managed?
22:41 djgerm1 (context: one needs a base zone file in order for ddns clients to update it, so I want my dns masters to have that file of course, but not to overwrite it everytime the state is applied)
22:42 drawsmcgraw joined #salt
22:44 hemebond djgerm1: unless
22:44 djgerm1 something like file.managed blah blah blah and with a "unless: 'test -f /etc/bind/zones/zone_i_don't_want_overwritten"
22:44 hemebond or creates
22:45 djgerm1 creates huh?
22:45 hemebond Wait... that might be from something else.
22:46 hemebond unless should work
22:46 djgerm1 with the test -f?
22:48 hemebond Yeah
22:49 djgerm1 heh… anyway to use unless for file.recurse?
22:49 djgerm1 like… all the files in the dir...
22:49 hemebond Um, just check for one file.
22:50 hemebond If you're copying down just the initial data then you don't want to touch the directory again, correct?
22:50 djgerm1 should be protected. i guess…. true.
22:50 djgerm1 whoa.. that was an epic derailed thought train.
22:50 djgerm1 let me try that again.
22:51 djgerm1 right. all the files in that "dynamic zone" directory should be written once by the masters (upon first "highstate") and never again, as long as they still exist.
22:51 djgerm1 so test -f any of them there files should be enough
22:51 hemebond And if you create a new zone file?
22:52 djgerm1 oh yeah… that's likely.
22:52 djgerm1 so that test of one file would break deploying the new dynamic zone
22:52 djgerm1 zones
22:53 djgerm1 hrmm. heh test -f /etc/bind/zones/{ all the files that currently exist, cross referenced with the files in the file.recurse source directory}
22:53 nidr0x joined #salt
22:57 mavhq joined #salt
23:02 baweaver joined #salt
23:03 uictamale joined #salt
23:05 uictamale Hello salty friends.  I'm looking for some help with salt-cloud.  I just finished installing salt-master and salt-cloud on centos7 on a host in GCE and after following the guides here: https://docs.saltstack.com/en/latest/ref/configuration/index.html and here: https://docs.saltstack.com/en/develop/topics/cloud/qs.html#salt-cloud-qs   I tried to lau
23:05 uictamale nch an instance but it threw the error "[ERROR   ] There was a profile error: 'NoneType' object has no attribute 'startswith'"
23:05 uictamale Any help appreciated, thanks!
23:09 ninjada joined #salt
23:09 mpanetta joined #salt
23:17 baweaver joined #salt
23:20 mavhq joined #salt
23:26 om joined #salt
23:31 ahammond So, is there an rsync capable source for the saltstack repos? I'm currently syncing using wget and... it's wasteful.
23:32 jfindlay I don't think so, but that would be useful
23:36 hasues joined #salt
23:36 hasues left #salt
23:46 mavhq joined #salt
23:49 mapu joined #salt
23:51 teryx510 joined #salt
23:51 kliquori joined #salt
23:53 aphor joined #salt
23:54 aphor Do we love macs on #salt ?
23:55 bhosmer joined #salt

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