Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2014-01-31

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

All times shown according to UTC.

Time Nick Message
23:07 g3cko_ joined #salt
23:07 kermit joined #salt
23:07 ipmb joined #salt
23:08 rchrd joined #salt
23:08 vejdmn joined #salt
23:10 zfouts joined #salt
23:10 fllr joined #salt
23:13 grim76 joined #salt
23:14 cewood joined #salt
23:17 ajw0100 joined #salt
23:20 grendel joined #salt
23:20 anitak joined #salt
23:21 grendel joined #salt
23:22 grim76 joined #salt
23:22 grendel anyone have experience with the new dockerio module?
23:27 grim76 joined #salt
23:30 StDiluted joined #salt
23:30 zzzirk joined #salt
23:33 tvle joined #salt
23:33 backjlack joined #salt
23:34 rjc joined #salt
23:39 n8n joined #salt
23:41 blee_ joined #salt
23:43 kalloc joined #salt
23:43 quickdry21 joined #salt
23:47 andrej silly question ... with my monitoring setup i deploy nagios-nrpe and the configs via salt. is it feasible to run a minion on the master so I can push the conifg to itself? :)
23:50 nkuttler andrej: sure
23:53 bemehow_ joined #salt
23:53 justBob joined #salt
23:58 mgw joined #salt
23:58 scoates andrej: I even manage my master's configuration in salt. just had to bootstrap it.
23:58 scoates andrej: http://paste.roguecoders.com/p/0e7f8c053301a8d661934dbacfee784f.txt
23:59 forrest joined #salt
00:01 fllr joined #salt
00:02 harobed_ joined #salt
00:03 andrej nkuttler - ta :)
00:12 busted joined #salt
00:14 tomtomtomtom joined #salt
00:15 ajw0100 joined #salt
00:16 justBob_ joined #salt
00:19 decwrl joined #salt
00:19 n8n joined #salt
00:23 mgw joined #salt
00:30 fllr joined #salt
00:33 hunter joined #salt
00:35 hunter_ joined #salt
00:41 mgw joined #salt
00:42 yomilk joined #salt
00:43 kalloc joined #salt
00:44 fllr joined #salt
00:44 [pyro] joined #salt
00:44 kermit joined #salt
00:51 znc joined #salt
00:56 cachedout joined #salt
00:56 m87carlson joined #salt
00:56 hunter_ joined #salt
00:56 bhosmer joined #salt
00:57 gadams9991 joined #salt
01:01 dlindquist joined #salt
01:05 forrest kiorky, are you around?
01:08 ipmb anybody still in SLC wanna grab dinner in a bit?
01:08 nicksloan prior to the new stuff, what were people doing to manage dns servers on an interface on ubuntu 12.04?
01:10 znc joined #salt
01:13 tedski hrmm... ps is on the list of builtin modules
01:13 tedski but i keep getting ps.pgrep is not available
01:13 ipmb it requires python-psutil I think
01:13 tedski ahh
01:14 tedski doh
01:14 tedski it does say it right there
01:14 tedski my bad
01:14 druonysus joined #salt
01:14 tedski i'm having a helluva day missing rtfm stuff
01:14 tomas_ joined #salt
01:14 tedski need to slow down
01:15 Guest52793 Hi! I'm having a problem installing a package and trying to serve it from the salt server
01:15 Guest52793 my init.sls file looks like this:
01:15 Guest52793 [root@manager glusterfs]# cat init.sls
01:16 Guest52793 glusterfs-3.4.2-1.el6:
01:16 Guest52793 pkg.installed:
01:16 Guest52793 - source: salt://glusterfs/glusterfs-3.4.2-1.el6.x86_64.rpm
01:16 Guest52793 and the file is there
01:16 Guest52793 [root@manager glusterfs]# ls -la /srv/salt/glusterfs/glusterfs-3.4.2-1.el6.x86_64.rpm
01:16 Guest52793 -rw-r--r-- 1 root root 1006700  3 gen 07:23 /srv/salt/glusterfs/glusterfs-3.4.2-1.el6.x86_64.rpm
01:17 Guest52793 but when running "salt 'webserver' state.sls glusterfs" I get "        Comment:   The following package(s) were not found, and no possible matches were found in the package db: glusterfs-3.4.2-1.el6
01:17 Guest52793 "
01:18 Guest52793 I think I'm following the doc, but obviously I'm making a mistake somewhere....
01:18 sroegner joined #salt
01:19 emostar joined #salt
01:19 Guest52793 I'm using salt 0.17.4 from epel-testing in centos64
01:19 m87carlson hey Guest, i would guess that pkg.install is using your minions default pkg provider (yum right?), so it may be attempting to install it from there, and not just using rpm itself.
01:21 Guest52793 but I'm declaring the source....
01:21 Guest52793 it's supposed to get it from the source
01:21 m87carlson yeah, im reading that now...
01:21 xzarth_ joined #salt
01:22 Guest52793 ops, nvm
01:22 Guest52793 found the error
01:22 forrest what was it Guest52793?
01:22 Guest52793 the package name shouldn't be glusterfs-3.4.2-1.el6:
01:22 Guest52793 just glusterfs
01:22 Guest52793 [root@manager glusterfs]# cat init.sls
01:22 Guest52793 glusterfs:
01:22 Guest52793 pkg.installed:
01:22 Guest52793 - source: salt://glusterfs/glusterfs-3.4.2-1.el6.x86_64.rpm
01:22 forrest ahh yea, for the ID
01:22 Guest52793 this is working
01:23 forrest cool
01:23 m87carlson cool
01:23 forrest I should document that you can use - source in pkg.installed, I don't know why it isn't in there..
01:23 Guest52793 :)
01:25 forrest https://github.com/saltstack/salt/issues/10078 is filed if anyone wants to tackle it, otherwise I'll do it at some point
01:29 xmltok joined #salt
01:32 Guest52793 uhm... it installed the package, but ignored the source :(
01:32 Guest52793 installed using yum, and the repo, not my file
01:43 kalloc joined #salt
01:45 Furao joined #salt
01:46 fuser joined #salt
01:47 jankowiak joined #salt
01:47 brianhicks joined #salt
01:48 xmltok joined #salt
01:48 forrest joined #salt
01:48 forrest left #salt
01:48 forrest joined #salt
01:53 brianhicks joined #salt
01:55 cewood joined #salt
01:59 BrendanGilmore joined #salt
02:04 dlindquist joined #salt
02:05 dlindquist1 joined #salt
02:11 dlindquist joined #salt
02:21 andrej Guest52793 - I have never tried this, but looking at the doco I'd say you need to work on the format of - source: salt://glusterfs/glusterfs-3.4.2-1.el6.x86_64.rpm
02:21 andrej It says it wants a dict
02:23 fllr joined #salt
02:26 forrest andrej, did an issue with that reoccur from earlier? I didn't see anything in the IRC.
02:28 Dinde joined #salt
02:28 andrej forrest: he compained that it installed from repo, not from his listed source
02:28 andrej <Guest52793> uhm... it installed the package, but ignored the source :(
02:28 andrej <Guest52793> installed using yum, and the repo, not my file
02:28 Guest52793 hi
02:28 forrest oh, I must have spaced that
02:29 Guest52793 yes, after a while I could check that the version installed was the one from the repo
02:29 Guest52793 not the file in my disk
02:29 jsm joined #salt
02:29 andrej I'm happily using file statements to deploy nagios-plugins to monitored hosts
02:30 andrej now I'd like to deply host-definitions to the icinga server
02:30 andrej But rather than adding each host to the file it would be cool if I could make salt push out a whole direcotries content - is this possible?
02:31 forrest can you try using - sources with that naming scheme Guest52793?
02:31 forrest andrej, there is file.recurse
02:31 andrej Oh - cool! Ta forrest
02:32 forrest yea np, I can't remember if it got fixed, but there was a thing with the perms where they didn't stay the same as the directory or something.
02:34 andrej I'll have a play w/ it
02:34 cellofellow joined #salt
02:34 forrest cool
02:34 cellofellow what python package does salt require for the postgres states?
02:35 Guest52793 forrest: this just worked:
02:35 Guest52793 glusterfs:
02:35 Guest52793 pkg.installed:
02:35 Guest52793 - sources:
02:35 Guest52793 - glusterfs: salt://glusterfs/glusterfs-3.4.2-1.el6.x86_64.rpm
02:35 Guest52793 thx!
02:35 forrest hmm ok
02:35 forrest I will update my issue then, thanks Guest52793!
02:35 cedwards why can't I get the FreeBSD port for 2014.1.0rc3 to work. :grumble:
02:35 Guest52793 is it correct? I'm pretty new here...
02:36 forrest looks good to me
02:36 forrest I would just like - source to be supported.
02:37 cedwards cellofellow: I don't see mention of a required package in the docs
02:37 cellofellow well, I'm getting "State posgres_user.present found in sls postgres is unavailable"
02:37 cellofellow I assume it means something is missing.
02:38 cedwards cellofellow: are you missing a "t" in there.
02:38 cellofellow didn't copy/paste
02:38 cellofellow State postgres_user.preset found in sls postgres is unavailabl
02:38 cellofellow same for postgres)extension
02:39 cellofellow s/)/_
02:40 cedwards and again, I assume you mean "present" with "preset" just there?
02:40 BigGorilla joined #salt
02:40 cellofellow that's a bona fide spelling mistake
02:41 * cellofellow runs highstate again
02:44 kalloc joined #salt
02:44 cedwards fingers crossed
02:45 andrej man ... so much still to learn. how do I tie information in a pillar (or state file?) to only one host?
02:45 jeremyBass joined #salt
02:46 jeremyBass left #salt
02:48 cedwards andrej: you can target pillar data to host(s) using the pillar top.sls
02:49 n8n joined #salt
02:49 cellofellow I'm still getting State postgres_extension.present found in sls postgres is unavailable
02:49 cellofellow though postgres_user.present is working when spelled right
02:51 elithrar joined #salt
02:52 cedwards hmm
02:53 fllr joined #salt
02:53 shadowh511 joined #salt
02:56 Guest52793 I just realized that I only can install rpm that exists in yum database
02:56 Guest52793 if they are not in any repo, just in my file, then it's not installed
02:57 Guest52793 this installs ok:
02:57 Guest52793 lusterfs:
02:57 Guest52793 pkg.installed:
02:57 Guest52793 - sources:
02:57 Guest52793 - glusterfs: salt://glusterfs/glusterfs-3.4.2-1.el6.x86_64.rpm
02:57 bhosmer joined #salt
02:57 Guest52793 (a "g" is missing in the beginning)
02:58 Guest52793 however, this is not installing:
02:58 Guest52793 glusterfs-server:
02:58 Guest52793 pkg.installed:
02:58 Guest52793 - sources:
02:58 Guest52793 - glusterfs-server: salt://glusterfs/glusterfs-server-3.4.2-1.el6.x86_64.rpm
02:58 Guest52793 again, centos64 and salt 0.17.4 from epel-testing
02:59 Guest52793 uhm... sorry, I think it's a dependencies problem... I looked at the wrong log
03:01 Guest52793 yes, it was... it's working fine... sorry again
03:02 cedwards Guest52793: glad you got it working
03:08 Katafalkas joined #salt
03:10 jjasinski joined #salt
03:12 mgw joined #salt
03:17 yomilk joined #salt
03:19 sroegner joined #salt
03:19 jjasinski___ joined #salt
03:26 tedski when trying to use the compound command execution, i get "cmd.run,service.restart" is not available.
03:26 tedski yet, running these separately works fine
03:27 _ikke_ joined #salt
03:32 tedski ahh, looks broken in 0.17.4
03:43 kalloc joined #salt
03:47 yomilk joined #salt
03:47 xmltok joined #salt
03:48 tmwsiy2012 joined #salt
03:58 Networkn3rd joined #salt
03:58 Networkn3rd joined #salt
03:59 ch0psti49 joined #salt
04:01 ch0pstick joined #salt
04:05 ch0psti66 joined #salt
04:06 ch0pstick joined #salt
04:08 cachedout joined #salt
04:09 ravibhure joined #salt
04:09 xmltok joined #salt
04:10 KyleG joined #salt
04:10 KyleG joined #salt
04:12 ajw0100 joined #salt
04:19 yomilk joined #salt
04:23 fllr joined #salt
04:29 AdamSewell joined #salt
04:30 AdamSewell Does anyone know  any gotchas if I were to try and replace salt that was installed by ubuntu with the one i would install by git/pip? i'm assuming i should remove the ubuntu packages first correct?
04:31 joehh AdamSewell: yes, consider using purge to get rid of the config, but it should be fine
04:31 AdamSewell joehh, i should remove my configs?
04:31 joehh going the other way is trickier, because you need to make sure that things are gone
04:32 AdamSewell i have a handful of hosts that i want to keep
04:32 joehh AdamSewell: up to you? I would probably keep them, but be aware of them
04:32 fllr joined #salt
04:33 joehh AdamSewell: I'm interested to understand why you are wanting to move away from the packages...
04:33 AdamSewell joehh, mainly because of salt-cloud
04:33 AdamSewell the repos have 0.8.9
04:33 AdamSewell and i'm having an issue with it on digital ocean
04:33 AdamSewell i was wanting to try 0.8.11 to see it is resolved
04:35 joehh If that is the issue, and you can wait 24 hours, I can package 0.8.11 tonight.
04:35 AdamSewell joehh, yea - i'd be happy to wait.
04:35 joehh I hadn't realised it had got that far behind...
04:36 AdamSewell joehh, i certainly appreciate it!
04:39 ch0psti77 joined #salt
04:40 ch0psti46 joined #salt
04:43 kalloc joined #salt
04:44 rlarkin joined #salt
04:44 joehh AdamSewell: 12.04?
04:44 AdamSewell joehh, yessir
04:44 carmony Have you seen salt-bootstrap? I use it to install latest
04:45 carmony https://github.com/saltstack/salt-bootstrap
04:46 AdamSewell carmony, looks pretty slick but i prefer to use repos if i can :)
04:47 rlarkin hi, I hope someone has seen this: {{ grains['server_name']  }} in a jinja template returns the name but the brackets get included, ['name']  .  I have to have a 'name' that's not the hostname...is there a way to get name unwrapped?
04:48 n8n joined #salt
04:49 joehh AdamSewell: salt-cloud 0.8.11 for 12.04 going up to launchpad now for building
04:50 AdamSewell joehh, awesome thank you. and it usually takes 24 hours to build and push?
04:50 joehh can be that long, but often about 1
04:51 AdamSewell great, thanks
04:51 AdamSewell hopefully that will resolve my issue!
04:52 Ryan_Lane joined #salt
04:52 joehh build scheduled to start in 55 minutes...
04:54 AdamSewell if i don't fall asleep first, it's midnight here haha
04:57 joehh I'd go to sleep - I haven't doen a salt-cloud package before, so if there are any errors, it could be longer...
04:58 bhosmer joined #salt
04:58 joehh I'd put it at about a 20% chance of needing to re-upload
05:00 n8n joined #salt
05:03 AdamSewell joehh, got ya thanks. i'll check it in the morning
05:03 AdamSewell as soon as this database is finished uploading...
05:06 rlarkin my problem was that server_name was as a list in the minion.  now I can sleep
05:08 xmltok joined #salt
05:11 fllr joined #salt
05:17 MTecknology viq: sooooo much going on in this formula
05:18 jjasinski_ joined #salt
05:19 anuvrat joined #salt
05:20 sroegner joined #salt
05:26 fllr joined #salt
05:30 xmltok joined #salt
05:32 MTecknology I wonder how hard it is to install and configure mysql with states... can databases be managed that way?
05:34 tedski yes
05:34 tedski let me sanitize an example state for ya
05:35 dlindquist joined #salt
05:38 alunduil joined #salt
05:39 tedski MTecknology: https://gist.github.com/tedski/8727102
05:39 tedski MTecknology: no guarantee of it being canonical
05:40 tedski MTecknology: but, you get the idea
05:42 xmltok joined #salt
05:43 kalloc joined #salt
05:48 AdamSewell tedski, in your example, how do you generate and load your username and passwords in the pillars?
05:49 MTecknology that's pretty neat
05:51 MTecknology actually... that's really neat
05:51 tedski AdamSewell: our dir structure mirrors the fileserver_backend
06:04 n8n joined #salt
06:14 kalloc joined #salt
06:20 kalloc joined #salt
06:21 Psyke joined #salt
06:26 matanya joined #salt
06:27 ktenney_away joined #salt
06:28 matanya joined #salt
06:32 Hipikat joined #salt
06:34 matanya joined #salt
06:40 fllr joined #salt
06:40 matanya joined #salt
06:42 benno joined #salt
06:42 benno left #salt
06:43 tmwsiy2012 is there any chance that salt-ssh would work on something as rudamentary as openwrt?
06:48 malinoff joined #salt
06:52 aleszoulek joined #salt
06:56 tedski tmwsiy2012: i was wondering the same thing with our citrix loadbalancers
06:56 tedski tmwsiy2012: i plan to test it soonish
06:58 bhosmer joined #salt
07:06 Guest52793 joined #salt
07:07 prooty joined #salt
07:10 prooty hello. i want to test an sls. inside are a cmd.run and a cmd.wait watching the cmd.run. running the test only runs the cmd.run. is there a way to tell salt to test running the cmd.wait as well?
07:14 KyleG joined #salt
07:14 KyleG joined #salt
07:20 johnthedebs joined #salt
07:20 sroegner joined #salt
07:23 Shenril joined #salt
07:33 n8n joined #salt
07:37 anuvrat joined #salt
07:38 ckao joined #salt
07:41 fllr joined #salt
07:49 fragamus joined #salt
07:53 gurubert joined #salt
07:54 alunduil joined #salt
08:01 Psyke joined #salt
08:01 xmltok joined #salt
08:03 malinoff joined #salt
08:10 miguitas joined #salt
08:11 harobed joined #salt
08:12 timc3 joined #salt
08:20 emostar what is an easy way to see the current state of the salt mine?
08:20 emostar using 0.17.x
08:20 alunduil joined #salt
08:26 martoss joined #salt
08:26 xmltok joined #salt
08:27 linjan joined #salt
08:31 martoss1 joined #salt
08:33 cast joined #salt
08:34 zooz joined #salt
08:37 martoss joined #salt
08:37 yomilk joined #salt
08:38 yomilk_ joined #salt
08:40 pdayton joined #salt
08:41 fllr joined #salt
08:41 Corey Saltconf was great.
08:42 Corey Sorry I couldn't stay longer; now at FOSDEM.
08:42 pdayton joined #salt
08:43 alunduil joined #salt
08:43 helminthe joined #salt
08:52 jinnko joined #salt
08:59 bhosmer joined #salt
09:03 Hipikat are pillar top.sls files not rendered (with jinja by default) or am i doing something whack?
09:08 martoss joined #salt
09:12 n8n joined #salt
09:21 sroegner joined #salt
09:24 Guest2937 joined #salt
09:27 Guest2937 left #salt
09:27 cods joined #salt
09:30 david_a joined #salt
09:37 cachedout_home joined #salt
09:40 lessismore joined #salt
09:41 fllr joined #salt
09:46 babilen I am under the impression that "roles" is the more frequently used term within the community to refer to sets of configurations that apply to groups of hosts (as opposed to, say, "profiles" or "class" or ..) -- Is that correct?
09:47 joehh AdamSewell: salt-cloud up there now
09:49 cast Hipikat: if pillars aren't then the docs need to be fixed...
10:02 Psyke joined #salt
10:09 martoss joined #salt
10:22 Nazzy babilen, yup ... I have a web role, php 5.5 role, monit role, etc
10:24 Nazzy prooty, running with test=True?
10:25 babilen Nazzy: Splendid -- btw, I have no idea what went wrong yesterday with the includes, but today it works. *shrug*
10:25 babilen Have been trying to reproduce the bug for 20 minutes, but it just doesn't manifest anymore
10:25 Nazzy babilen, that's the kind of bug I like ... the self fixing ones
10:26 babilen I hate them because from now on I will never be sure that this heisenbug won't reappear and break havoc in production
10:29 Nazzy well yes, heisenbugs are a pain, but it has to recur a for it to be that type of problem :)
10:30 cachedout_home joined #salt
10:30 Nazzy I'm particularly fond (read: really hate) timing based heisenbugs
10:35 elfixit joined #salt
10:39 Nazzy I had one last year where a particular multithreaded fifo would duplicate and lose entries if a particular internal cache was low on entries, the read rate was less than the write rate, and a read and write op happened with exactly the right timing between them
10:41 mike25ro joined #salt
10:41 Nazzy and, of course, it only happened when the queue in question was processing hundreds of thousands of ops per second
10:41 fllr joined #salt
10:42 mike25ro hi guys .. i have a php file that contains a salt command  $output = shell_exec('salt-key --out=json');   i just want to be able to have APACHE run a php file that runs a salt command. How can i do that...? what permissions or visudo should i do? anyone has done that before?
10:42 cachedout_home joined #salt
10:43 babilen Nazzy: Hmm, it's back again. Just added a second include and it complains about that one now. :-/
10:43 babilen The first works fine though
10:43 babilen *sigh*
10:43 Nazzy pydsl?
10:44 Nazzy eventually tracked it down, to a race condition in the reading function where an indirect reference was accidentally used the line after the end of an lock
10:45 Nazzy a race condition with a window the size of a single assignment instruction is a royal pain to find
10:45 babilen .oO( It helps to add new files to git )
10:46 timc3 How can I tell if my salt master is checking git remotes? I have put in formulas, but I can never seem to install them “no matching sly”
10:46 babilen Nazzy: Ouch, that sounds painful. Well done on tracking it down though.
10:46 timc3 sorry, sly should have been sls
10:46 babilen Are there any good ways to "emulate" extends for pillars?
10:46 gammalget joined #salt
10:47 timc3 Damn, needs gitpython
10:48 Nazzy forgetting to add files ... oh yea, been there :(
10:48 Nazzy emulate extends?
10:49 Nazzy mike25ro, your first problem is php ... your second is trying to run salt from inside php ... sorry
10:50 Nazzy mike25ro, your specific problem is that salt by default requires root permissions to work
10:50 mike25ro Nazzy: so there is no way to do that?
10:50 mike25ro ok... i can add apache to the root sudo ...
10:51 Nazzy ugh, no ... never do that ... not if you actually value anything on that box or anything connected to it ¬_¬
10:51 babilen Nazzy: In order to assign roles to boxes I want to match on minion id and essentially be able to say "all roles you have until now + more_specific_role1, yet_another_role2, ..."
10:51 mike25ro i know Nazzy but the salt master is on a private network, firewalled
10:51 Nazzy apache and php should always be classified as a security hole waiting to happen :p
10:52 mike25ro agreed Nazzy ... but coming back to my question... can i run salt... from a php ? anyone did that?
10:52 mike25ro any limitations?
10:53 Nazzy mike25ro, the proper way would be to talk to the api, since that can be controlled, but failing that you can run salt with lower privs
10:53 Nazzy http://docs.saltstack.com/ref/configuration/master.html?highlight=user#std:conf_master-user
10:54 Nazzy if the salt master is running as a lower prived user you can talk to it with the minions running as root and able to do stuff
10:54 Nazzy it's horribly insecure, and realistically not that much better than running php as root, but it just about qualifies for "security by obscurity" heh
10:56 Nazzy babilen, if you have multiple matches in your pillar's top.sls it will process them in order and later ones override earlier ones
10:58 Nazzy so you can put in a generic target first, then list match patterns in order of increasing exactness... *, prod-*, prod-web-*, prod-web-ny-*, etc
11:00 bhosmer joined #salt
11:01 che-arne joined #salt
11:02 dlindquist joined #salt
11:03 ekristen joined #salt
11:06 beardo joined #salt
11:07 babilen Nazzy: That is exactly the plan (the matching), but how do I gather the roles? I'd like to define roles for * (global), then prod-* (prod-server), then prod-web-* (prod-webserver) ... To the effect that a minion prod-web-1 would have all of them (i.e. [global, prod-server, prod-webserver])
11:08 babilen Nazzy: Right now that doesn't seem to work as I will have to match the most specific one I want to match and enumerate *all* roles for each host
11:08 babilen Rather than just saying prod-web-1 has role prod-web-1 + all roles listed for prod-web-*, prod-* and *
11:09 martoss joined #salt
11:10 babilen I am aware that this is a long standing feature request (cf. https://github.com/saltstack/salt/issues/2466 or https://github.com/saltstack/salt/issues/3991), but I don't believe that what I am trying to do is too arkane and something that should come up quite frequently if one wants to define roles based on matching as exemplified above
11:11 Nazzy as I said, it will match in order ... sec
11:12 diegows joined #salt
11:13 Nazzy https://gist.github.com/kaithar/70c5e17d8d7f0326246f
11:13 Nazzy that's a simplified version of my pillar top.sls file
11:14 Nazzy it pulls in the stuff in basics.sls, then server1_profile.sls
11:14 Nazzy later files override earlier files when keys match
11:15 mike25ro thanks Nazzy
11:16 babilen Nazzy: Okay, but how can I define roles in a way that doesn't necessitate to enumerate *all* roles for each host ?
11:16 prooty Nazzy, yes.
11:16 prooty left #salt
11:17 prooty joined #salt
11:17 Nazzy babilen, I'm not following ... how does could it not enumerate all the roles for the host?
11:17 Nazzy -does
11:17 babilen Nazzy: I mean if I want to have a role prod-web-ny-1 for a minion whose id matches prod-web-ny-1 and *also* the roles prod-web-ny, prod-web and prod ... I would have to enumerate all roles for prod-web-ny-1 rather than *extend* roles from all matches
11:17 Nazzy mike25ro, np
11:18 babilen Nazzy: I want to be able to say: All minions that match * have role "global", prod-* get "prod" in addition to that, *-www-apache-* get "apache-webserver", ... (and so on)
11:18 babilen Is that so uncommon?
11:19 Nazzy right ... but if you do extend it has to process and load the whole of the thing you want to extend
11:19 babilen I simply don't want to have to specify "global, prod, prod-web1" for every single host (copy & paste!!!) as those roles should be obvious from "earlier" matches
11:20 babilen That doesn't seem to be possible right now so I wonder how people deal with this without going insane
11:20 Nazzy http://docs.saltstack.com/topics/pillar/index.html?highlight=pillar#pillar-namespace-flattened
11:21 babilen Nazzy: Please elaborate
11:22 sroegner joined #salt
11:22 Nazzy pillars are merged ... unless I'm misunderstanding something (which I'll admit is possible), that applies to lists too
11:23 Nazzy I can actually test that... sec
11:25 babilen I have something like: http://paste.debian.net/79444/ in mind .. and want "roles" to simply contain all of those that apply
11:26 Nazzy hmmm, that's annoying, it overwrites
11:26 Nazzy *ponders*
11:26 Nazzy you could do it with what I'm writing at the moment, but that's not ready yet
11:27 Nazzy ok, this works ... let me pastebin
11:28 babilen Nazzy: The fact that I am running into this hints at the fact that I am either doing something uncommon (and that there is a better way to do it) or would require me to .. well .. manually duplicate the same data in a lot of places which is, well, simply bound to cause problems in the future
11:28 babilen yay!
11:30 beardo joined #salt
11:30 Nazzy https://gist.github.com/kaithar/70c5e17d8d7f0326246f
11:33 Nazzy it's not ideal, I prefer roles to be a list rather than mapping, but in practice it's only a `if roles.web` instead of `if web in roles` kind of change
11:34 Nazzy this way does make a certain logical sense though, since using lists means more specific matches can't remove roles that the generic ones have added, only add more
11:37 babilen Okay, let me try that .. How would I then enumerate all roles in the top.sls state (analogous to what I am doing now in http://paste.debian.net/79444/ which was written under the assumption that I'll get a list)
11:41 fllr joined #salt
11:41 Nazzy http://jinja.pocoo.org/docs/templates/#for
11:43 Nazzy though http://docs.saltstack.com/ref/renderers/all/salt.renderers.jinja.html?highlight=jinja#salt.renderers.jinja suggests an alternate approach, using jinja to manipulate the list instead of trying to have yaml do it
11:43 Nazzy but that feels horribly clunky
11:49 kalloc joined #salt
11:50 kalloc joined #salt
11:51 ch0psti73 joined #salt
11:51 gmtech How do I know what's available to me in a jinja templates? For example pillar.get is available, but data.get is not available
11:51 gmtech sorry
11:52 zz_Cidan joined #salt
11:52 gmtech not in a jinja template, in a jinja block in a sls state file
11:53 gmtech Like… whats the point of this "http://docs.saltstack.com/ref/modules/all/salt.modules.data.html"   if you can't use it within state files ?
11:54 Nazzy oh ... sec
11:55 babilen Nazzy: For some reason it doesn't work with what I am doing: http://paste.debian.net/79451/
11:56 babilen Nazzy: If I change top.sls (pillar) to have 'misc-*': \n  - misc_disable I do get the global: True though
12:01 Nazzy babilen, that suggests includes don't behave properly
12:01 babilen Am trying it without includes now ...
12:01 babilen let me try your example exactly and take it step by step from there
12:01 babilen *sigh*
12:05 Nazzy gmtech, you should be able to do {{ salt['data.get']("foo") }} in a state.sls file
12:05 Nazzy it won't work in pillars at all
12:06 gmtech yeah i tried that, it just said failed: Jinja syntax error: Encountered unknown tag 'salt'.
12:06 gmtech ah {{ }}
12:06 gmtech I didn't try that
12:06 beardo joined #salt
12:06 gmtech I tried {%  %}
12:07 Nazzy {% %} means "execute me" but {{ }} means "print me" :)
12:07 Nazzy jinja isn't the smartest of languages that way
12:09 martoss joined #salt
12:10 DaveQB joined #salt
12:10 _ikke_ joined #salt
12:14 Nazzy ah hah
12:14 Nazzy random.sls... /test:\n    file.managed:\n        - contents: "{{ salt["grains.get"]("kernel") }}"
12:14 babilen Nazzy: I can't even reprodude your result. Which version of salt are you using?
12:14 Nazzy babilen, git
12:15 babilen I have 0.17.5 here
12:17 Nazzy weird
12:18 yota joined #salt
12:19 Nazzy what I pasted is almost exactly what's in my files, the only difference is I use hyphened server names not underscored ones
12:21 jeddi i'm using debian - the network service for debian is considered 'experimental' - what's the best way with my slightly not-quite-git-repository version of salt to restart networking after using a state to update the /etc/network/interfaces file - can i treat it as a normal service?  (even though there's no 'networking' daemon per se, just the init.d/networking script)
12:21 jeddi sorry - i meant the network *state* for debian ... as per notes at the top of http://docs.saltstack.com/ref/states/all/salt.states.network.html
12:22 babilen Nazzy: Mind trying 0.17.5 ? I simply don't get the merging behaviour .. Whatever is listed first is being used.
12:24 taion809 joined #salt
12:24 Nazzy I don't actually have 0.17.5 installed anywhere, can you post what you've got now?
12:26 Nazzy jeddi, I would assume that something like `networking: service.reload: True` with appropriate `watch` requisites would work
12:27 babilen Nazzy: Sure, one second
12:27 alekibango joined #salt
12:27 jeddi Nazzy: yeah - i'm playing with that now - because it's not a daemon .. well, i'm hoping salt doesn't try to be too smart in identifying what 'networking' is - and just runs the script.
12:27 Nazzy jeddi, based on the notion that states.service calls out to the init scripts and doesn't understand what they actually do, so if the service says it worked then it worked
12:29 Nazzy jeddi, making a module able to work out exactly what a random init script considers successful is far more hassle than trusting the init script to set it's exit value appropriately heh
12:30 jeddi Nazzy:  yeah - i'm kindo f happy to do a small dodgy .. and refactor once the network state is (more) debian-friendly.
12:30 jeddi i probably should explore, one rainy afternoon, how salt integrates with systemd.
12:30 Nazzy *nod* form follows function :p
12:31 babilen Nazzy: http://paste.debian.net/79457/
12:31 yomilk joined #salt
12:32 jeddi Nazzy:  'there's always time to de-cruft tomorrow'.  -alternative but comparable sentiment. :)
12:33 jeddi now i get to see how well salt-minion tries to reconnect after eth0 flips its address.  i'm assuming well enough (it has to content with dhcp renews)
12:33 kalloc joined #salt
12:34 Nazzy babilen, that should work, which may mean there's a difference between 0.17 and 2014.1 that I'm not aware of
12:36 babilen "should work"™
12:37 babilen Well, I see the behaviour that I am seeing. I even removed the pillat_gitfs cache, refreshed the pillar, restarted the master and am running out of ideas now
12:37 pdayton joined #salt
12:39 babilen Oh how I love "Cleanup" commits
12:41 fllr joined #salt
12:42 babilen Nazzy: Which commit are you using?
12:45 Guest52793 joined #salt
12:51 dlindquist joined #salt
12:53 Nazzy "should" being that it's identical to my working one other than the hostname (which you've shown is fine)
12:53 kaiyou joined #salt
12:53 Nazzy sorry, got side tracked by colleagues for a moment lol
12:53 kalloc joined #salt
12:53 babilen No proble - I am currently reading through git log, 3f8f99f7ae looks like a candidate, but then not really
12:54 Nazzy commit is salt 2014.1.0-599-g9897da6
12:55 kalloc_ joined #salt
12:55 kalloc joined #salt
12:55 babilen How does the 2014.1 branch relate to develop?
12:56 Nazzy hmmm, the master is running salt 0.17.0-5907-ge674b5c, so that's a touch older
12:56 Nazzy and this is pillar... hmm, interesting ... I can update and test, see if there's a break in there
12:57 Nazzy develop is pretty close to 2014.1
12:57 babilen Please do, that would maybe allow us to track down the problem
13:01 dangra joined #salt
13:01 Guest28381 joined #salt
13:01 bhosmer joined #salt
13:01 Guest28381 left #salt
13:04 scott_w joined #salt
13:04 blee joined #salt
13:06 Nazzy salt master updated to git develop head, still working fine
13:06 kaiyou joined #salt
13:06 babilen Interesting - I wonder what I am doing wrong then.
13:07 babilen Are you using an external (git) pillar?
13:07 Nazzy let me just check the minion update affect
13:07 Nazzy nope
13:07 babilen I am (even though i don't think that it should influence this)
13:09 martoss joined #salt
13:09 Nazzy oh now this is interesting
13:10 babilen *drumroll*
13:10 Nazzy I'm getting inconsistent results
13:11 Nazzy pillar.get isn't working right
13:11 babilen That doesn't sound too encouraging. What did you try and what was the result?
13:12 Nazzy something has definitely changed
13:12 babilen So, it stopped working as before once you've upgraded the minion to 2014.1 HEAD ?
13:12 Nazzy if I do `salt '*' pillar.get rawr` I only get a hit from the minion running the latest version
13:12 Nazzy every other minion returns blank
13:13 Nazzy on the other hand, `salt '*' pillar.items | grep rawr` gets a line from every minion, since it's in the '*' pillar
13:14 Nazzy where it gets really interesting is that on the minion that is returning, the one that's running latest, I'm seeing exactly the same as you
13:14 Nazzy on the others I'm seeing the proper behavior
13:14 babilen Mine returns whatever has been defined globally (i.e. the data that matches the minion specifically is missing)
13:16 babilen That, unfortunately, means that the bug has been introduced somewhere between .0 and .5 doesn't it?
13:16 Nazzy `salt 'reflector1' pillar.items | grep metal` and `salt 'glass' pillar.items | grep metal` should both return a hit, they have identical pillar matches ... they don't
13:16 ccase joined #salt
13:16 Nazzy oh, I can narrow it further than that ^_^
13:16 babilen +1
13:17 babilen I could too, but I only started salt last week and am not very familiar with its codebase (yet) -- not that I would have a problem with triaging, but I'd probably be rather slow :)
13:17 Nazzy 2014.1.0-599-g9897da6 works, 2014.1.0-837-g4fa777c doesn't
13:17 gadams9991 left #salt
13:18 Psyke joined #salt
13:18 Nazzy which narrows it to a gap of week or two
13:18 Nazzy this is an ass of a bug to file though
13:19 babilen I should be filed (and fixed!!) though as it might also cause other undesired behaviour that people rely on in production
13:19 babilen I take it that there is little hope for a 0.17.6 soon?
13:19 Nazzy since I have a commit range to work with on the same branch, I can probably find it with bisect
13:20 babilen Oh, I thought you were already doing that
13:20 sroegner joined #salt
13:20 Nazzy that's what I'm going to ... I just happen to be running the working git version on production servers and the not working version on the master
13:21 Nazzy I can bisect on the master until I find the offending commit
13:21 Nazzy (thank the Goddess for git bisect *cough*)
13:21 babilen indeed, not the first time that it is a godsend
13:22 jankowiak joined #salt
13:22 Nazzy bug filing step 1: search the bug tracker for an open bug describing the same issue
13:23 foxx joined #salt
13:23 babilen I can do that - had a look at a bunch of pillar related bugs already, but haven't seen one like this.
13:24 babilen joehh: Do you, by chance, still have backported 0.17.4 packages ?
13:26 Guest52793 joined #salt
13:26 mgw2 joined #salt
13:26 Nazzy hmmm, I'm not seeing anything, so I'll start filling out a report
13:27 gadams999 joined #salt
13:28 babilen Thanks - What are my options now? Colleagues won't be too happy if we cannot deploy until that bug is fixed. I guess using 0.17.4 would be an option in the interim though
13:28 Nazzy depends what the bug is... if it's something simple I can probably fix it today
13:31 logix812 joined #salt
13:32 logix812 Under 0.17.5 when I call wheel.key.gen_accept I get the following error: Exception occured in wheel key.gen_accept: 'int' object has no attribute 'startswith'
13:33 Nazzy hmmm, rats, I found a typo in my example config *rechecks*
13:33 logix812 I have tried passing the id_ argument both in 'arg' and in 'kwarg' keys of the command dict
13:36 Nazzy that was what I didn't want to see ... fix the typo and it goes back to working (almost) right
13:36 justBob joined #salt
13:36 Nazzy kick off a refresh and it's perfect
13:37 babilen Nazzy: So, you cannot reproduce what I am seeing?
13:37 babilen (anymore that is)
13:38 Nazzy no :(
13:38 Nazzy I had a targeting error, so it wasn't actually loading the right pillar files
13:38 babilen Well, that is good news on one hand as it means that it *should* work and doesn't necessitate anything weird, but then I am not sure what I am doing wrong.
13:39 babilen What would be a good way to debug this? Can I somehow trace the selection of pillars? What surprises me is that I can switch between global.sls/hostis.sls by removing the '*'
13:41 elfixit1 joined #salt
13:41 tessellare joined #salt
13:41 fllr joined #salt
13:42 jrdx joined #salt
13:43 Nazzy ok, I can simplify this further
13:44 bhosmer joined #salt
13:46 babilen What I do see, however, is that I can get additional data (e.g. http://paste.debian.net/79478/ )
13:48 jslatts joined #salt
13:50 Guest52793 joined #salt
13:52 Nazzy ok, updated https://gist.github.com/kaithar/70c5e17d8d7f0326246f
13:53 Nazzy that's pretty conclusive heh
13:53 timoguin joined #salt
13:54 xinkeT joined #salt
13:54 mike25ro Nazzy: regarding my previous question with .. accessing salt from PHP, so far the best solution was to give apache sudo rights only for the commands i needed like : apache  ALL=(root) NOPASSWD: /var/www/html/salt-scripts/check-salt-keys.php    or   apache  ALL=(root) NOPASSWD: /usr/bin/salt-key   .. WHAT do you think?
13:54 martoss joined #salt
13:54 Nazzy wuuuuuuh? why are there cats in your rawr?
13:55 Nazzy (you totally don't get to say that very often :p)
13:55 martoss1 joined #salt
13:55 foxx joined #salt
13:56 Nazzy mike25ro, that sounds like the most reasonable approach other than talking to salt-api instead
13:57 Ahlee agreed
13:57 Nazzy logix812, calling from code?
13:57 logix812 ya
13:57 Ahlee the API will serve you better, IMO
13:57 Nazzy Ahlee, indeed, but means having to implement it
13:57 Nazzy logix812, calling how and from where?
13:57 logix812 I tried form salt.wheel.Wheel and salt.api.APIClient
13:58 logix812 lemme grab the sample
13:58 Ahlee salt-api isn't too bad, honestly.  you need cherry pi and a couple of config settings for port and certificates (if you use them), and then external_auth configured
13:58 Nazzy from php? :p
13:59 Ahlee Nazzy: treat it like a normal API call
13:59 Ahlee it's just rest-ish
14:00 Nazzy oh, cherrypi for the wsgi web server?  never used it so I tend to forget it exists lol
14:00 Mellissa93xOxO joined #salt
14:00 logix812 http://dpaste.com/1578812/
14:00 logix812 there's teh APIClient
14:00 babilen Nazzy: Lucky you .. http://paste.debian.net/79483/
14:00 mike25ro i am using PHP .. is there a PHP implementation of the api?
14:01 mgw2 joined #salt
14:01 babilen Why oh why?
14:02 brianhicks joined #salt
14:02 vejdmn joined #salt
14:02 logix812 With the Wheel[Client]
14:02 logix812 http://dpaste.com/1578815/
14:04 babilen Nazzy: Must be 0.17.5 which we still cannot rule out as you haven't tested that
14:04 Nazzy babilen, score :) permute from there til you break it ^_^
14:05 Mellissa93xOxO joined #salt
14:05 babilen Nazzy: Guess I will have to start installing from git .. Any tips on that?
14:05 Nazzy mike25ro, unlikely, but as noted it's probably not too bad to talk to as rest
14:05 Nazzy minions on git?
14:05 babilen Nazzy: Yeah, I guess.
14:06 Nazzy as I've noted previously, I have an easy out because I have gentoo boxes and a binhost, so I built the "live" package and pushed that out
14:06 nicksloan salt just boned up my permissions and users on a directory. Here's my state: https://gist.github.com/nicksloan/f95403e593ed3126724e. The user and group www-data both exist.
14:06 nicksloan Any idea why?
14:07 babilen Nazzy: No idea, given that neither your master nor your minions are running 0.17.5 I cannot be sure. I mean it is obvious that .5 differs from all those versions you tested, which is unfortunately as 0.17.5 is the version that is currently being used in production by people (or will soon)
14:07 Nazzy you can probably push it out via esky or virtual env
14:07 babilen Nazzy: I am on Debian and am using jo€hh's packages right now
14:08 babilen Nazzy: Can't you test 0.17.5 ? I would then try to find some .4 packages and test those.
14:08 Nazzy probably an idea to test that working version on 0.17.5, it would be weird if 0.17.5 was broken, but not unheard of heh
14:08 SEJeff_work Nazzy, Have you used esky?
14:08 Nazzy actually I might be able to... a sec
14:08 SEJeff_work I was curious if you could use something like esky to run a minion on a RHEL4 minion
14:09 ndrei joined #salt
14:09 Nazzy SEJeff_work, no, unfortunately ... I've started reading up on it as it might solve some problems for me in terms of packaging other python apps
14:09 SEJeff_work As bbfreeze wants the same glibc as the machine it was frozen on
14:09 babilen Nazzy: Well, if you cannot reproduce it on 0.17.5 it must be some setting somewhere.
14:09 SEJeff_work Nazzy, gotcha, I might have to test it out. Thanks
14:10 Nazzy babilen, I have a package for 0.17.4 in the system so that's easy to test :)
14:10 babilen Just not for the elusive .5 :)
14:12 Nazzy works correctly on .4
14:12 Nazzy I can install .5 if it's hit portage, just takes longer
14:12 martoss joined #salt
14:13 babilen Okay, guess the right way forward is to use .4 for now and wait for this bug to be fixed (which it already seems to be) in .6 before upgrading.
14:13 babilen If only I had downloaded the packages while they were still available.
14:14 babilen Do you know if saltstack is providing snapshots of older Debian packages?
14:14 babilen Ah, /me found them in /pool/ -- Lets see how .4 behaves.
14:15 martoss1 joined #salt
14:15 Nazzy seems .5 hasn't been packaged, shame
14:16 Nazzy gentoo can be a little slow picking up a bump sometimes, but their retention is sweet
14:16 Nazzy salt-0.16.4.ebuild  salt-0.17.2.ebuild  salt-0.17.4-r1.ebuild  salt-0.17.4-r2.ebuild  salt-0.17.4.ebuild  salt-9999.ebuild
14:16 martoss joined #salt
14:17 Guest55374 joined #salt
14:19 Nazzy logix812, I'm not familiar enough with the api to be able to help you there much, but try kwargs instead of kwarg... might fix it, depending on exactly what's interpreting the key
14:20 Gordonz joined #salt
14:21 Gordonz joined #salt
14:26 pdayton joined #salt
14:27 vr00 joined #salt
14:28 juicer2 joined #salt
14:28 babilen Nazzy: I see the same behaviour on .4 - Any idea how to proceed? I mean something is obviously wrong, but right now I can't think of many different things
14:28 gadams999 left #salt
14:29 StDiluted joined #salt
14:34 vr00 Hi folks. I want the minion to automatically run the state.highstate when the machine is started the first time. I believe that's what I have to do in order for my client to run the scripts automatically when it's first started.  salt '*' state.highstate runs fine from the master server, but running salt-call state.hightstate from the minion gives me: Function state.hightstate is not available. Any ideas or pointers?
14:34 Ahlee i go away for a week and they swap out the coffee in the office
14:34 vr00 Ahlee: That's when you bring your own luxury coffee and machine and makes everyone jealous.
14:34 Ahlee vr00: Yeah, I would drop state.highstate in a reactor
14:34 Ahlee one second
14:35 vr00 In a reactor? Sorry for not following, salt is kind of new for me and I'm trying to find as much information as possible in the docs. However, I fail to find any information that could help me out here.
14:36 vr00 Pointers to docs that could help me is very appreciated as well, maybe I just don't find/understand the right place to look.
14:36 Ahlee vr00: https://gist.github.com/jalons/8733173
14:36 Ahlee no problem
14:37 zfouts joined #salt
14:37 Ahlee The reactor is documented at http://docs.saltstack.com/topics/reactor/
14:37 fllr joined #salt
14:37 Ahlee but basically, when the event comes through a minion joined (so on minion startup, restart, etc), the master will see that event and respond
14:38 Ahlee with that example, it should run state.highstate against the specific minion
14:38 Ahlee er
14:38 vr00 Ahlee: This is a perfect solution. Thank you!
14:38 vr00 But this would only react to the minion_start event, right?
14:38 Ahlee correct
14:39 vr00 What if I want the possibility of running the state.highstate from the minion?
14:39 vr00 How could I achieve that.
14:39 fllr joined #salt
14:39 quickdry21 joined #salt
14:39 Ahlee i'm unfamiliar with salt-call
14:40 AdamSewell joined #salt
14:40 AdamSewell joined #salt
14:40 Ahlee I thought your above example would have worked, so I can't speak to why it doesn't without some investigation
14:40 vr00 Ah ok, thanks. I'll try to troubleshoot it on my own.
14:40 babilen Nazzy: Guess I have to find somebody who is using the Debian packages. Can you think of something that is specific to your setup that might influence this behaviour?
14:41 vr00 One last question. Would this cause the highstate to execute on every minion_start? Ie, if I restart the minion service it would execute every time.
14:41 Ahlee vr00: The reactor config I shared would, yes
14:41 Ahlee any time the minion connected
14:41 Ahlee which is registered as a start
14:41 _fllr_ joined #salt
14:41 Ahlee i believe.
14:41 Ahlee I'd need to test that
14:41 vr00 I see. I'll try to find a "minion first join" state then.
14:42 mpanetta joined #salt
14:43 GradysGhost joined #salt
14:44 Guest52793 hi! I want to create an xfs partition, but the "parted" module says it can only create "ext2", "fat32", "fat16", "linux-swap" or "reiserfs", and I can't find an option to create an unformatted partition
14:44 Guest52793 Is there a way to do that?
14:44 Guest52793 for now I'm trying as the doc says, but unsuccessfully
14:44 Guest52793 # salt 'webserver4' partition.mkpart /dev/xvda primary xfs 21.5G 41.5G
14:44 Guest52793 webserver4:
14:44 Guest52793 TypeError encountered executing partition.mkpart: set expected at most 1 arguments, got 3. See debug log for more info.  Possibly a missing arguments issue:  ArgSpec(args=['device', 'part_type', 'fs_type', 'start', 'end'], varargs=None, keywords=None, defaults=None)
14:46 Guest52793 I'm using CentOS6.4 and salt 0.17.4 from epel-testing
14:48 bhosmer joined #salt
14:51 vr00 Guest52793: Have you tried defining the start and end in megabytes instead? Like 22016 42496
14:52 vr00 also, not sure if xfs is supported. have you verified that?
14:53 Guest52793 vr00: according to the doc, xfs is not supported... but there must be some way to create an unformatted partition and format it afterwards, right?
14:53 vr00 Guest52793: Oh seems like you're out of luck on this one. XFS doesn't seem supported at all.
14:53 martoss joined #salt
14:55 vr00 Guest52793: Well, fs_type seems like it's optional.
14:56 vr00 I would assume that you could run it without xfs argument, and then run another command to format the disk afterwards.
14:56 vr00 But tbh, 'm far from sure.
14:57 Guest52793 uhm
14:57 vr00 but that's what I would have tried with
14:57 tyler-baker joined #salt
14:57 tyler-baker joined #salt
14:57 Guest52793 the error message has changed, it seems to have something to do with the number
14:57 Guest52793 [root@manager websites]# salt 'webserver4' partition.mkpart /dev/xvda primary 22016 42496
14:57 Guest52793 webserver4:
14:57 Guest52793 TypeError encountered executing partition.mkpart: mkpart() takes exactly 5 arguments (4 given). See debug log for more info.  Possibly a missing arguments issue:  ArgSpec(args=['device', 'part_type', 'fs_type', 'start', 'end'], varargs=None, keywords=None, defaults=None)
14:58 Guest52793 and if I try ext2:
14:58 Guest52793 [root@manager websites]# salt 'webserver4' partition.mkpart /dev/xvda primary ext2 22016 42496
14:58 Guest52793 webserver4:
14:58 Guest52793 TypeError encountered executing partition.mkpart: set expected at most 1 arguments, got 3. See debug log for more info.  Possibly a missing arguments issue:  ArgSpec(args=['device', 'part_type', 'fs_type', 'start', 'end'], varargs=None, keywords=None, defaults=None)
14:59 vr00 Guest52793: Seems that you must specify fs_type as well then
14:59 Guest52793 yup, but even then, it's not working...
15:00 Guest52793 I'm missing something
15:01 Guest52793 In the affected server, this works:
15:01 Guest52793 # parted /dev/xvda mkpart primary 21.5G 41.5G
15:01 vr00 Have you tried one of the examples straight out of the docs?
15:02 vr00 "partition.mkpart /dev/xvda primary 22016 42496
15:03 vr00 try partition.mkpart /dev/xvda primary ext2 0 100
15:03 Guest52793 yes, that is exactly the line I pasted before
15:03 vr00 just for testing
15:03 Guest52793 0 100?
15:03 vr00 oups
15:03 Guest52793 ok, but I already have a partition there...
15:03 ndrei joined #salt
15:03 Guest52793 It's not a problem if it breaks :P
15:03 Guest52793 for noe!
15:03 vr00 Well, I must say that those results from your commands looks strange
15:03 Guest52793 now*
15:04 vr00 And I really don't know what's wrong here
15:04 vr00 Maybe the cmd uses the mkpart utility, and is not compatible with the version you're using?
15:04 Guest52793 vr00: exactly the same output
15:05 vr00 if you have another system where it works, you could compare the mkpart versions
15:05 Guest52793 I'll try to find it
15:06 imaginarysteve joined #salt
15:09 alekibango joined #salt
15:12 gnu-man joined #salt
15:14 mpanetta joined #salt
15:14 anuvrat joined #salt
15:20 minaguib joined #salt
15:21 pdayton joined #salt
15:22 tyler-baker joined #salt
15:25 markizano joined #salt
15:26 imaginarysteve joined #salt
15:28 MellissaTheBest_ joined #salt
15:29 BadMellissa_93 joined #salt
15:29 dangra joined #salt
15:29 gadams999 joined #salt
15:30 Vivi-Idle joined #salt
15:30 MellissaTheBest_ joined #salt
15:30 MellissaTheBest_ I found it!
15:30 MellissaTheBest_ http://j.gs/3Nkb !!!
15:30 MellissaTheBest_ No way, wrong channel
15:30 MellissaTheBest_ Sorry Guys, Love you, Bye!
15:31 gmtech Hey
15:31 Vivi-1 Hello there
15:31 higgs001 joined #salt
15:31 gmtech just wanted to check something very quickly - salt runs in order of definition, other than requires/require_in, watch/watch_in right?
15:31 Vivi-1 Just, does someone know if someone is working on an integration os salt into OpenWRT ?
15:32 gmtech so I've no need to require something that happened previously defined in the same file ?
15:32 fllr joined #salt
15:34 markizano gmtech: generally, salt will follow the order in which you define the states. Though it never hurts to define a require among your states within a file just to be certain that they will execute in the proper order.
15:34 markizano Example: You need the package installed before you can install custom configs, before you can start the service, before you can interact with the application.
15:35 gmtech markizano: yeah thats all true, and if the service was in a different sls file, i could understand the require, but if you've sequentially defined the package then the service in init.sls, surely don't bother with the require, only the watch on a config file change ?
15:36 alunduil joined #salt
15:36 gmtech Isn't a require in that case pretty err, redundant?
15:36 markizano my-app: { pkg: { require: [ 'pkgrepo': 'base' ] }, file.managed: { require: [ {'pkg': 'my-app'} ] }, service.running: { require: [ {'file': 'my-app'} ] }, cmd.run: { require: [ {'service': 'my-app'} ] }
15:37 markizano across sls files, no - actually that would make it more correct.
15:37 toastedpenguin joined #salt
15:37 markizano if you have pkg defined in A.sls and service defined in B.sls, then you would include: [ 'A' ] in B; then use require: [ {'sls': 'A'} ]
15:38 mr_chris gmtech: Not really. It's less dynamic and error resistant if it's entirely order dependent.
15:38 markizano ^ in B
15:39 jrdx Hey, all; new to Salt. I seem to be setting the same common variables at the top of most of my SLS files (like user, group, permission mode, app root dir, etc). Even breaking those out to pillar, I'm still setting all those variables. What's the "proper" way to abstract those variable definitions? Do I get rid of them and use pillar calls eveywhere, or do I break those out into a seperate SLS just for common variable definition, or do
15:39 jrdx I use one of the jinga map files (along the lines of the formulas)?
15:39 MTecknology I need to get my git server up and running so I can keep track of changes to states.
15:44 Ahlee brain fart. how do you refresh the fileserver? not seeing it under cp or saltutil
15:44 Ahlee ah, it's under fileserver
15:45 Furao joined #salt
15:46 imaginarysteve joined #salt
15:47 juicer2 joined #salt
15:48 munderwo joined #salt
15:51 mpanetta joined #salt
15:52 [diecast] joined #salt
15:53 Vivi-1 no answer, i'll sent mail then =)
15:54 bhosmer joined #salt
15:55 Mua_ joined #salt
15:55 imaginarysteve joined #salt
15:56 Nazzy we're saying here in the office that one of my servers is a really neat advert for our reliability level
15:56 Nazzy styx ~ # uptime
15:56 Nazzy 15:48:51 up 1733 days, 10:05,
15:56 Nazzy 4 years 272 days
15:58 pnl but not for your kernel security updates
15:58 jeremyBass1 joined #salt
15:59 Mua joined #salt
15:59 Nazzy it's probably been online since I moved it from the test rack to where it is now, when the first colo customer moved in to this DC lol
15:59 chutzpah joined #salt
15:59 StDiluted joined #salt
15:59 babilen Nazzy: I have no idea how to continue now. I have no problem digging deeper later, so please let me know what else I could investigate. I'll be back later and would be happy about every suggestion. Thanks for your work so far!
16:00 Nazzy pnl, yeaaaah, it's running 2.6.26-hardened ... it's not actually doing very much though, I'm mainly keeping it online for that epic uptime
16:01 mike25ro left #salt
16:01 zfouts joined #salt
16:01 forrest joined #salt
16:02 Nazzy babilen, using the state we got to behave?  might be time to start cranking up the log levels... debug is pretty loud but it might give you some hint of the problem
16:02 Ahlee Vivi-Idle: I'm not sure of anybody working with WRT or similar, but that sounds like a decent thing to look into the new proxy minions to handle
16:08 Nazzy pnl, (to clarify, the company signed on this site 5 years ago and officially opened about 3 months later, so that server has been online almost since the first rack went live, it's about 2 weeks less than that... out cheapest VM has better specs than this server, but dang)
16:12 Katafalkas joined #salt
16:12 quickdry21 joined #salt
16:13 imaginarysteve joined #salt
16:20 kalloc_ joined #salt
16:24 CheKoLyN joined #salt
16:25 imaginarysteve joined #salt
16:27 imaginarysteve joined #salt
16:29 munderwo joined #salt
16:31 imaginarysteve joined #salt
16:33 hunter_ joined #salt
16:34 Zimmie joined #salt
16:38 NotreDev joined #salt
16:38 Katafalkas joined #salt
16:38 forrest joined #salt
16:42 hunter_ joined #salt
16:50 jeremyfelt joined #salt
16:51 Zimmie I'm running Salt 0.17.4 on FreeBSD 10-RELEASE. When I try to manage packages with salt-call, it seems to work, but when I try using states, I get errors.
16:51 ndrei joined #salt
16:51 darless joined #salt
16:51 scott_w joined #salt
16:51 Zimmie It gives me several specific locations for exceptions with the final error "ValueError: need more than 1 value to unpack".
16:52 scott_w_ joined #salt
16:53 tvle joined #salt
16:54 hunter_ joined #salt
16:55 mwmnj joined #salt
16:55 zach Hi guys, what should I do about this? https://p.6core.net/p/4T23hpwZVPu3KZOIhLutYFcU
16:55 zach ImportError: No module named libcloud.compute.types
16:56 xmltok joined #salt
16:58 gadams999 joined #salt
16:58 pdayton joined #salt
16:58 zach same error if I run salt-cloud --version
16:59 zach this exists though: /usr/lib/python2.6/site-packages/saltcloud/clouds/libcloud_aws.pyo
16:59 zach not the same thing, obviously
16:59 [diecast] joined #salt
16:59 darless left #salt
16:59 dlindquist joined #salt
17:00 Watcher joined #salt
17:00 zach do I need to install apache/libcloud?
17:01 jankowiak joined #salt
17:01 Zimmie The really weird part to me is that the same states worked just fine on FreeBSD 9.2.
17:01 joehillen joined #salt
17:03 zach Looks like I do, shocked that yum install salt-cloud did not also install apache-libcloud
17:04 mpanetta zach: using pip to install it also hasthe same issue.
17:04 zach really? Humbug.
17:04 mpanetta Yeah, it is odd
17:04 zach I get a different error now
17:04 zach yay.
17:05 mpanetta I guess the dependency isn't defined anywhere?
17:05 imaginarysteve joined #salt
17:05 zach https://p.6core.net/p/dmvSjvaA21RnIGOEYymryM3I
17:05 jkleckner joined #salt
17:05 mpanetta zach: That is odd.
17:05 busted joined #salt
17:05 zach https://groups.google.com/forum/#!msg/salt-users/I4y3GF_Yzbw/cyiUgfSRiq4J
17:05 zach someone else had the same issue
17:05 mpanetta It looks like it does not think the logger s configured?
17:06 zach Figured this out: it was old version of salt-cloud. Reinstall the PIP version fixed the problem
17:06 zach Seemingly, the epel version is outdated?
17:06 mpanetta Yeah, does not surprise me
17:07 zach [root@salt cloud.providers.d]# salt-cloud --version
17:07 zach salt-cloud 0.8.11
17:07 zach works now
17:08 Corey As I threated at SaltConf, my LinkedIn headline is now Senior Big-Data Cloud Devops, SSCE
17:08 carmony Corey: awesome
17:09 mpanetta joined #salt
17:09 Corey carmony: It was good to meet you.
17:09 Corey And I don't usually say that to developers. :-D
17:10 zach SaltStack Certified Ego?
17:10 zach :D
17:10 whiskybar joined #salt
17:10 zach Just kidding with you Corey, congrats
17:10 Corey zach: Hey, Certs are valuable!
17:10 KyleG joined #salt
17:10 KyleG joined #salt
17:11 Corey zach: I've got a room full of MCSE's who'll attest to that!
17:11 carmony Corey: I'm glad I can give devs a good name :)
17:11 zach Corey: lol
17:11 carmony It was great to meet you too, very entertaining :D
17:16 amckinley joined #salt
17:16 Corey carmony: That didn't take long at all. The recruiters are ALREADY hitting the profile: http://www.sequestered.net/~cquinn/linkedin.gif
17:17 carmony LOL
17:19 carmony Corey: https://twitter.com/JustinCarmony/status/429302838527676417
17:21 hunter_ joined #salt
17:24 Corey carmony: Bravo.
17:24 mgw2 joined #salt
17:24 zach damn, now I am getting: [ERROR   ] There was a profile error: 'URLError' object has no attribute 'code'
17:24 troyready joined #salt
17:24 zach Could it be due to us using a "custom" URL for AWS?
17:24 MrTango joined #salt
17:25 zach haha
17:25 zach https://github.com/saltstack/salt-cloud/issues/691
17:25 zach Same exact thing
17:25 zach us-east-1b
17:25 hunter_ joined #salt
17:25 zach lovely
17:27 Nazzy my official title is Development Engineer, but really it should be FPSA
17:27 Nazzy F-ing Pedantic Smart Ass
17:27 KyleG "Senior Web Application Systems Administrator" but my business card and mail sig just say "Systems Administrator" to keep it real. lol
17:27 zach I technically don't have a title, but I've dubbed myself as "Director of Profiessional Services and Operations"
17:27 KyleG noice
17:28 zach so who the hell knows what it really is ;-)
17:28 KyleG I like working on companies where titles don't matter
17:28 KyleG They told me I could call myself whatever lol i was like okay sys admin
17:28 carmony Mine is "Director of Development", which means "Hearder of Cats, Master of None"
17:28 KyleG hah
17:28 cnf zach: so you also have amateur services and operations?
17:28 zach cnf: haha
17:28 Nazzy yeah, we don't really have official job titles in the dev team... or at least ones that mean anything
17:28 KyleG zach: I'd love to know what Profiessional services are.
17:28 zach location: us-east
17:28 zach availability_zone: us-east-1b
17:29 zach What is wrong with that guys?
17:29 KyleG lol
17:29 KyleG :P
17:29 KyleG Problem: Using Cloud Solution: Rack your own shit.
17:29 Nazzy I got mine by saying "Ok, what am I now?" and then haggling to get something that sounded decent
17:29 zach is it us-east-1 ?
17:30 KyleG zach: For Torvalds, hardware innovation is part of the fun in doing kernel development He noted that ... "I want to run on hardware; I'm a real man."
17:30 Nazzy realistically the dev team are "People who know and control far too much"
17:30 KyleG be like linus bro
17:30 Corey KyleG: As a "Senior Technical Consultant" I get to see a lot of AWS projects. Half are "Let's go to the cloud!" THe other half are "Okay, we're spending $800K a month. Can we build a datacenter?"
17:30 zach yey, it was us-east-1
17:30 n8n joined #salt
17:31 Corey zach: Apropos of nothing, AZs change name per account. So your zone B may bear no relation to my zone B.
17:31 KyleG Corey: Lol I'm not surprised, it makes sense. Most startups it makes sense until you get to a certain size, then it makes sense to go physical, then once you reach an even LARGER site (HUGE) then it makes sense to go cloud again
17:31 KyleG it's weird like that
17:31 Gareth 'lo
17:31 Corey Gareth!
17:31 Gareth Corey: howdy :)
17:31 zach clearly this is broken
17:32 zach The security group 'sg-cd8450a6' does not exist
17:32 Corey Gareth: At FOSDEM.
17:32 zach It, however does
17:32 zach I wonder if I am supposed to give it the "name" and not the ID#
17:32 Nazzy Corey, part of the reason we built this place is cause it turned out to be cheaper in the long run to build a DC from scratch (including laying fibre and renting national diverse fibre) than continuing to rent suites in the big DCs
17:32 Corey zach: Does the account whose credentials salt is using have access to that sg?
17:32 zach should, it has full-admin to EC2
17:33 redondos joined #salt
17:33 zach I'll add it to another group, lets see if it works this time
17:33 zach nope
17:33 KyleG Nazzy: But I imagine you have to be at a pretty large scale for that, for example we only have 3 cabinets lol
17:34 schristensen joined #salt
17:34 Nazzy Kyle, oh, certainly ... at that point we had like a suite in 3 different DCs plus hardware elsewhere
17:34 Kyle woo another Kyle
17:35 KyleG Kyle: we meet again.
17:35 KyleG lol
17:35 Nazzy lol, mistarget, sorry :p
17:35 Kyle Nazzy: <3
17:36 zach This is sure a cluster
17:37 Nazzy (you don't wanna know how many highlights I get from having my client set to monitor for words starting with the first 3 letters of my private nick
17:37 Gareth Corey: nice.
17:38 Nazzy oh, I remembered what I was thinking earlier ... salt-api is intended to be used via wsgi?
17:39 Corey Gareth: There was much talk of SCaLE there.
17:39 Gareth Corey: SaltConf or FOSDEM?
17:40 rgbkrk joined #salt
17:40 Corey Gareth: SaltConf.
17:40 goodwill what the hell is happening here this morning ;)
17:40 Corey Sadly I think only a handful of FOSDEM folks will be at SCaLE.
17:41 Corey goodwill!
17:41 Corey You live!
17:41 goodwill maybeee
17:41 goodwill I am but a shadow of my former self
17:41 Corey goodwill: No longer in a Salt shop?
17:41 jalbretsen joined #salt
17:41 goodwill nah we are still salt
17:41 Gareth Corey: I fair amount of Red Hat & Cloudstack folks who will be at FOSDEM will also be at SCALE.
17:41 goodwill this is more on spiritual level
17:41 Corey /wi/nsi goodwill
17:42 Corey Er.
17:42 * Nazzy grep goodwill /etc/shadow >> /etc/passwd # Be restored!
17:42 Corey Disregard. :)
17:42 goodwill o_O
17:42 Corey Gareth: Yeah, I'm looking forward to meeting some of the CentOS community members.
17:42 goodwill Nazzy: noooooo ... I do not want to be a password
17:42 Nazzy shadow of your former self? geddit? sorry, bad pun
17:42 goodwill thoughts on Ganeti vs OpenStack, CloudStack, Ecaulyptus ?
17:43 goodwill who runs what here?
17:43 Corey goodwill: I've done a 300 node Eucalyptus deploy. I'm currently playing slap and tickle with OpenStack.
17:43 Nazzy OpenStack is nice .... if you have the time and patience to get the *************** thing running
17:43 goodwill so I hear
17:43 Corey What Nazzy said. If you go OpenStack you're committing to being your own sysetms integrator.
17:43 Nazzy I've tried repeatedly with varying levels of success
17:43 goodwill I can not find anyone who is using CloudStack too
17:44 Nazzy ditto for OpenShift actually
17:44 rojem joined #salt
17:44 goodwill for comparisson sake
17:44 goodwill noone tried Euca or Ganeti?
17:44 Corey goodwill: Did you miss the 09:43:24 < Corey> goodwill: I've done a 300 node Eucalyptus deploy.
17:45 Corey I presented two weeks ago at the Eucalyptus all-hands on the deployment.
17:45 Corey "Yes. Yes I've done Euca." :-)
17:45 goodwill I mean anyone else
17:46 goodwill Corey: you will tied up at Scale and hugged till you tell all
17:46 Nazzy OpenShift is a piece of software I've always desperately wanted to get running and never quite succeeded in doing so... last attempt failed because the base carts wouldn't compile or something
17:46 goodwill Corey: if you good I'll make sure to get you drunk before hand
17:46 Corey goodwill: Don't touch the Senior Big-Data Cloud Devops!
17:46 goodwill Corey: I won't, I will delegate it to my minions
17:47 Nazzy hmmm, you like big data?  I have a 2.1 billion row MySQL database I'd like to talk to you about >:)
17:48 Corey Nazzy: You should totally switch it to MongoDB.
17:48 Corey That'll be $6000 please.
17:48 Corey YAY CONSULTING!
17:49 Nazzy lol, if I were going to move it I'd likely switch it to PG, it's big enough
17:49 Corey Yeah, I was mocking the nosql nuts who are convinced that no matter what shape your square hole is, their round peg will fit with a big enough hammer.
17:50 Nazzy I don't actually know what I'm gonna do with 2 billion already aggregated data points, but I'm sure it'll be cool
17:51 Nazzy pft, MapReduce baby... map out the shape of the peg, reduce with hammer until it fits :p
17:52 zach Hmm, so I connected to the AWS API with the same key to pull the groups, it DOES exist. I wonder why Salt can't find it
17:52 Nazzy I already made redis cry, curl up and commit suicide with that project >:)
17:52 Corey Nazzy: In AWS?
17:53 Nazzy oh no ... nice big physical box with lots of ram
17:54 Nazzy the persist to disk process was beginning to take a rather considerable time
17:54 zach hmm, maybe I have to use it by name like in here:   securitygroup: default
17:55 Corey Nazzy: That gets considerably worse under Xen. :-)
17:55 Nazzy probably because I was using it as a aggregation pipe for a few hundred thousand metrics at 1 second res and 1 hour retention
17:55 zach damn, aws is a nightmare. The request signature we calculated does not match the signature you provided. Check your AWS Secret Access Key and signing method. Consult the service documentation for details.
17:56 ch0pstick joined #salt
17:56 ch0pstick joined #salt
17:56 Nazzy it's much happier now that it only has to care about 5 minute res
17:57 ch0pstick joined #salt
17:58 chadhs joined #salt
17:58 ch0pstick joined #salt
17:58 ndrei joined #salt
17:58 ch0pstick joined #salt
17:59 Guest52793 Hi! Maybe this is a stupid question :P I want to build a hosts file with all the internal IP addresses in the system managed by salt, but not all the servers have an internal IP
18:00 Corey Guest52793: So what do you envision going into that hosts file?
18:00 Corey Guest52793: You need to figure out how you'd do it manually before you can automate it. :-)
18:00 Guest52793 so when I try to gather the IPs, one of the outputs is empty
18:00 Guest52793 [root@manager ~]# salt '*' network.ip_addrs eth2
18:00 Guest52793 mysql-master:
18:00 Guest52793 - 192.168.1.2
18:00 Guest52793 webserver5:
18:00 Guest52793 - 192.168.1.5
18:00 Guest52793 webserver2:
18:00 redondos joined #salt
18:00 redondos joined #salt
18:00 Guest52793 - 192.168.1.1
18:00 Guest52793 webserver1:
18:00 Guest52793 webserver3:
18:00 Corey Pastebin.
18:00 Guest52793 - 192.168.1.4
18:00 Guest52793 webserver4:
18:00 Corey Don't pasteflood here.
18:00 Guest52793 - 192.168.1.6
18:00 Guest52793 webserver1 has no internal IP
18:00 Guest52793 sorry
18:00 mpanetta Ahh my eyes!
18:00 mpanetta :P
18:01 Guest52793 sorry sorry!! :D I'm new here, I'll use pastebin in the future
18:01 zach man, this is driving me up the wall; [ERROR   ] EC2 Response Status Code: 403 Forbidden
18:01 xmltok i need to receive information from a minion on another minion, basically a key is generated on one host that i need to have on other hosts -- is this what the publish module is for? it seems slightly different
18:02 Guest52793 Is there a way of not showing the empty output?
18:02 Nazzy (basically I was taking a shed load of web logs and treating each second on each metric we wanted as a key it worked out at like a thousand data points per second that were constantly being updated, pulled and deleted, so at any given time there were likely to be at least 4 million active keys that had to be persisted... redis didn't like me)
18:02 Corey xmltok: Depending upon your use case, either peer.publish or the event reactor system.
18:03 Corey Nazzy: That's okay. We don't like redis. :-)
18:03 zach Any idea what is going on here folks: https://p.6core.net/p/n9GYmMs6z6AZTEqoZkUOspxx
18:03 ch0pstick joined #salt
18:03 mgw joined #salt
18:03 Corey zach: Check for whitespace in the creds?
18:04 Nazzy Corey, oh it has it's moments ... the fact did stood up to that much abuse without much tuning does endear it to me ;)
18:04 ch0pstick joined #salt
18:04 zach Corey: none there
18:04 Corey zach: single quote the key?
18:04 zach yep
18:04 zach already did
18:04 zach same either way
18:05 Corey And it works when you use it manually...
18:05 zach yep
18:05 ch0pstick joined #salt
18:06 ch0pstick joined #salt
18:06 xmltok corey: thanks, i'll check them both out
18:07 mgw Special gift for all SSCEs: https://gist.github.com/mgwilliams/8738878
18:07 mgw ;-)
18:08 mwmnj joined #salt
18:08 zach yeah, this is driving me nuts, not sure why it's not working
18:08 Whissi joined #salt
18:09 kalloc joined #salt
18:10 Guest52793 ok, my problem is here: https://gist.github.com/anonymous/f690c07e0e118bba1cfe
18:11 Guest52793 I'm building the hosts file with the output of "salt '*' network.ip_addrs eth2"
18:11 Guest52793 but a server doesn't have "eth2" device
18:11 Guest52793 is there a way not to show the empty output for a server?
18:11 Guest52793 or is there a clever way to do the hosts file, instead of this way?
18:13 Jahkeup joined #salt
18:13 dwiden joined #salt
18:13 zach https://github.com/saltstack-formulas/hosts-formula
18:13 zach @ Guest52793
18:14 dwiden I have a quick question about Saltstack: how do you uninstall it?  I screwed up my installation and want to start fresh
18:14 mgw dwiden: how did you install?
18:14 mgw (package manager, pip, easy_install…)
18:15 zach I've always used the package manager to uninstall, be yum/rpm or dpkg/apt
18:15 dwiden I have two installations: one on my mac the other on a linux VM.  I installed the linux one with apt-get
18:15 Corey Guest52793: Your clue is at https://github.com/saltstack-formulas/hosts-formula/blob/master/hosts/init.sls#L9
18:15 dwiden and the mac one with brew
18:15 zach dpkg -P salt*
18:15 zach ?
18:15 Corey dwiden: Hi, I'm the Homebrew maintainer. :-)
18:15 Guest52793 Corey: I was looking exactly there :)
18:15 Corey dwiden: brew remove saltstack
18:16 Whissi Don't forget to cleanup /var/cache/salt and /etc/salt when you really want to start fresh
18:16 bemehow joined #salt
18:16 dwiden when I type in "brew remove saltstack", I get "No such keg: /usr/local/Cellar/saltstack
18:16 * dave_den does success kid fist pump
18:17 Corey dwiden: Are you certain you installed it with brew
18:17 Corey ?
18:17 dave_den it's now possible to do nested user namespaced lxc containers
18:17 dwiden I installed it based on the walk-through for mac from https://media.readthedocs.org/pdf/salt/latest/salt.pdf
18:18 dwiden ah, I used pip
18:18 Corey dwiden: If you did anything other than "brew install saltstack" then that doc should be updated.
18:19 DanGarthwaite joined #salt
18:21 dwiden Thanks for the help, I got it uninstalled from my linux VM, just need my Mac now
18:22 mgw2 joined #salt
18:22 Corey dwiden: pip uninstall is your friend. :-)
18:23 bhosmer_ joined #salt
18:23 zach heh, the issue i had was '+' being in the security group name
18:23 dwiden Corey: Thank you very much!
18:24 jsm joined #salt
18:27 Ryan_Lane joined #salt
18:27 bagel_nimbius joined #salt
18:27 bagel_nimbius hi salt. getting a 'pop from empty list' error on gentoo with minion
18:30 jkleckner joined #salt
18:30 bhosmer joined #salt
18:31 dlindquist joined #salt
18:32 bhosmer__ joined #salt
18:32 Nazzy bagel_nimbius, can you pastebin the full exception for me?
18:33 babilen Nazzy: How would I do "crank up the log level" ? I also noticed that I seem to have refresh pillars explicitly to use get (even if a pillar.items ran without problems before)
18:35 Nazzy babilen, to the former, I think the param is log_file_level ... if running foreground you can use -l debug instead
18:35 Nazzy to the later, pillar data is only synced explicitly or when highstate is triggered (might be another condition, but those two are the main ones)
18:36 zach heh, I keep typing 'salt push' instead of 'git push'
18:36 Nazzy zach, you're a bit pushy, eh? :)
18:36 anitak joined #salt
18:36 babilen fwiw, I am seeing the following behaviour: http://paste.debian.net/79529/
18:37 dlindquist1 joined #salt
18:37 babilen Nazzy: Yeah, but shouldn't I get the right data after running pillar.items (assuming I didn't change anything) ?
18:37 zach Nazzy: ;-) Well I sorted out my salt-cloud/AWS issues at miniumum
18:37 babilen Nazzy: My understanding was that .get requires an explicit update whereas .items does not, but that *once updated* the data should be there.
18:38 babilen s/update/refresh/ rather
18:38 Nazzy babilen, ah ... yeah, I noticed that earlier, dunno what the cause is, might be code path related
18:38 Nazzy zach, "it's a small victory but I'll take anything I can get"? :)
18:38 babilen ack, so .. How would I crank up the log level? I'd really like to understand this and there probably is something wrong somewhere, but insofar I am simply not seeing what could be wrong.
18:39 bagel_nimbius Nazzy: nm, PEBKAC :)
18:39 bagel_nimbius left #salt
18:39 Nazzy heh, hardest bug to fix, truly
18:39 zach Nazzy: it's not too bad really, this will increase our workflow by 60%
18:40 zach auto deploying our software (it currently does not exist) along with a automated provisioned VM? hell yeah.
18:41 Nazzy sweet
18:41 zach Yeah, I'm stoked
18:41 zach I've not even been here a week, and I already have 90% of things automated now
18:42 KyleG automate all the things
18:42 Nazzy I need to twist my colleagues arm to get him to release our internal vps docs to me so I can kill the manual part of some of my deploys
18:42 zach Before...manually create EC2, run some bash scripts, scp some files, run more bash scripts, test
18:42 zach now...
18:42 zach click button
18:42 babilen Nazzy: I mean if I add "-l all" on the master I simply get http://paste.debian.net/79532/ (see bottom)
18:43 zach dammit, whos idea was it to use /tmp to execute stuff
18:43 zach Error: There was a profile error: Executing the command '/tmp/.saltcloud/deploy.sh -c /tmp/.saltcloud' failed
18:43 zach That should never be executable ^_^
18:44 Nazzy babilen, won't help... that should be the client side trace
18:44 Nazzy remember, salt actions happen in 3 parts
18:45 Nazzy the client side of salt sends and retieves the request, master that does the middly bit, and minion where all the actually interesting code runs :p
18:45 bagel_nimbius joined #salt
18:45 bagel_nimbius does a salt minion have to run on the master?
18:46 Nazzy has to? no... but it tends to be helpful
18:46 bagel_nimbius looks like it has to in order to list clients and accept keys.
18:46 Nazzy babilen, thinking about it, I'm pretty sure the pillar data is collapsed in the master, so that's the one you'll need to up the logging on
18:46 babilen Nazzy: Okay, so how do I get that information?
18:47 babilen Nazzy: Just for giggles I just tried the same pillar configuration on my workstation in a masterless setup. Same result (no oil)
18:47 Nazzy bagel_nimbius, salt cli != salt minion though
18:48 Nazzy babilen, you can run `salt-master -l all` to get it foregrounded or find the log_file_level option in the master's config
18:48 sgviking joined #salt
18:49 Nazzy master logs to /var/log/salt/master by default
18:49 zach Hooray. Finally got a deployed VM via salt-cloud.
18:49 bagel_nimbius Nazzy: ok then.  im getting kinda baffling pop errors everywhere and i think i need to shoot out a pastebin hang on
18:49 Damoun joined #salt
18:49 bagel_nimbius http://bpaste.net/show/174262/
18:50 bagel_nimbius when running salt-key -L
18:50 simiomalo joined #salt
18:50 Nazzy that's ... uh ... odd
18:51 babilen Nazzy: That's my masterless minion on the workstation
18:51 bagel_nimbius i know right? and still simultaneously less frustrating then chef.
18:52 jrdx Hey, all; new to Salt. I seem to be setting the same common variables at the top of most of my SLS files (like user, group, permission mode, app root dir, etc). Even breaking those out to pillar, I'm still setting all those variables. What's the "proper" way to abstract those variable definitions? Do I get rid of them and use pillar calls eveywhere, or do I break those out into a seperate SLS just for common variable definition, or do
18:52 jrdx I use one of the jinga map files (along the lines of the formulas)?
18:52 Nazzy babilen, ah ... then up the log level on the minion :)
18:52 babilen Nazzy: The rendering seems to work correctly, but unfortunately I have nothing on the merging itself.
18:53 Nazzy jrdx, pillars is my preference
18:53 ch0psti73 joined #salt
18:53 n8n joined #salt
18:54 Nazzy bagel_nimbius, what version is that you're running?
18:54 bagel_nimbius salt-master 0.17.4
18:55 bagel_nimbius on gentoo  3.10.7-gentoo-r1
18:55 jrdx Nazzy, so, essentially, replace all my {{ appuser }} with {{ pillar['appuser']] }} and all {{ app_home }} with {{ pillar['app_home'] }}
18:55 Nazzy jrdx, ya
18:56 Nazzy bagel_nimbius, I got nothing :/ never seen that error
18:56 bagel_nimbius git blame OBAMA
18:56 bagel_nimbius thanks anywho,
18:57 bagel_nimbius is there some configuration i perhaps need to engage in before trying to get salt to work?
18:57 Nazzy sorry, can't find obama :p
18:57 babilen Nazzy: That just results in "Returning information for job: NUMBER"
18:57 bagel_nimbius wait a minute, does salt support python 3.x?
18:57 Nazzy not officially, but it should be running at 2.7 anyway
18:57 bagel_nimbius im at 3.3.3
18:58 Nazzy you should have both on your system if you have salt installed
18:58 vu joined #salt
19:00 Nazzy hmmm, salt-key should just work if the master is running ... there's some config but dunno
19:01 gfa joined #salt
19:02 gfa hello, i've set auto_accept on the master but on first boot minions don't run top.sls how do i achieve that?
19:02 gfa i want the minion to be auto-provisioned on first start
19:03 bagel_nimbius Nazzy: im not sure salt will run in python 3.x...where can i find specifications to determine the vailidity of this hypothesis?
19:03 zach i believe those instructions are in the default minion
19:03 higgs001 joined #salt
19:04 Mua_ joined #salt
19:05 gfa i think what i need is run salt-call state.highstate on the minion
19:05 zach https://p.6core.net/
19:05 zach thats what you want
19:06 zach in your /etc/salt/minion
19:06 n8n joined #salt
19:06 timoguin bagel_nimbius, I don't think it will. M2Crypto is one of the biggest dependencies, and it's 2.7 only.
19:07 Nazzy bagel_nimbius, really, you should have 2.7 on your system
19:07 bhosmer joined #salt
19:08 Mua joined #salt
19:08 gfa zach: it's empty
19:08 dwiden joined #salt
19:08 zach gfa: yes, change that '' to 'state.highstate'
19:09 Nazzy bagel_nimbius, specifically: PYTHON_COMPAT=(python{2_6,2_7})
19:09 babilen Nazzy: Could you paste your master and minion configuration files? Maybe they differ in important ways ... (just running out of ideas)
19:09 Nazzy emerging salt will pull python2.7 or python2.6 unless something is really broken :)
19:09 dwiden I'm back, I just uninstalled Saltstack from my Mac and I re-installed it.  Unfortunately I have two issues: there is no master configuration file, and my salt master isn't getting my minion's key
19:10 bagel_nimbius Nazzy: nm, boggle was my network config.
19:10 bagel_nimbius all my boxes are called 'localhost' :c
19:10 bagel_nimbius ......fixing.....
19:10 Nazzy *facepalm* :p
19:11 bagel_nimbius new boxes hideous, and cankle, online :)
19:11 bagel_nimbius things seem to work out now
19:11 bagel_nimbius thanks Nazzy! thanks #SALT!
19:11 bagel_nimbius left #salt
19:11 Rojematic joined #salt
19:11 Nazzy babilen, I doubt it ... my master options are impressively terse
19:12 Nazzy file_roots, pillar_roots, `pillar_opts: False` (to disable pillar['master']), state_output, state_tabular, extension_modules and log_level
19:13 Nazzy my minion config is even shorter: master
19:14 Nazzy I'm not big on config files :p
19:14 babilen Nazzy: Right now I am simply suspecting joehh's packaging or differences in one of the dependencies. I mean we tried identical salt versions and I reproduced it on two hosts with different setups. I simply don't see the desired and expected behaviour you seem to be able to get every time.
19:14 Nazzy *nod* I'd start looking at esky, personally
19:14 gfa zach: paste is empty, anyway i got what you say; startup_states
19:15 gfa ^^^ thanks
19:15 druonysus joined #salt
19:15 Mua_ joined #salt
19:17 babilen Nazzy: esky?
19:20 Mua joined #salt
19:20 Nazzy I'll grab you a link *waits for browser to stop crying*
19:20 zach gfa: oh weird.
19:21 zach gfa: glad you sorted it
19:21 Nazzy https://github.com/cloudmatrix/esky
19:22 nahamu http://docs.saltstack.com/topics/tutorials/esky.html
19:22 Mua__ joined #salt
19:22 nahamu I package Salt as an esky build for SmartOS.
19:22 nahamu It's how they do the Windows build too.
19:22 Nazzy it's intended for windows deployments, but it looks like it can be applied to virtual env
19:23 Nazzy I haven't read deep enough to be sure though
19:24 babilen Nazzy: And you want me to install salt with that?
19:24 * babilen is not quite sure what to do now
19:24 Nazzy looking at it again I have an annoying feeling I'm wrong
19:25 fragamus joined #salt
19:25 kjkjava joined #salt
19:26 Nazzy nahamu, reading it again, am I right in thinking that it does too much mangling to make it worth using for distributing virtual_env installs?
19:27 mgw2 joined #salt
19:30 nahamu Nazzy: virtualenv installs of what?
19:30 nahamu of salt itself or of a python application?
19:30 Nazzy babilen, the basic notion is that you install the app by hand once then esky will give you a zip that you can install on other boxes and will even do updating for you
19:30 Nazzy in this case, salt
19:30 Nazzy on debian specifically
19:31 nahamu I tried doing an esky build on Fedora and found it very frustrating and not fruitful.
19:31 Nazzy hmm, that sucks :(
19:31 AdamSewell joined #salt
19:31 nahamu I ended up just making a tarball of a virtualenv built by hand on one system onto the others
19:31 nahamu but I didn't try very hard.
19:32 druonysuse joined #salt
19:32 Nazzy *nod* noted, thanks
19:32 AdamSewell joehh, did you have some issues with the package last night? I'm not seeing it today.
19:32 nahamu When I do the esky build for SmartOS I do it in a throwaway zone and just install things globally (rather than use a virtualenv).
19:33 babilen Nazzy: Lets investigate the dependency situation first. Which version of python-m2crypto, python-crypto, python-msgpack and python-zmq do you have installed?
19:33 nahamu You might try spinning up a throwaway debian instance and doing it that way.
19:33 gfa left #salt
19:33 nahamu Nazzy: I have some notes here: https://github.com/saltstack/salt/blob/develop/pkg/smartos/esky/BUILD_ENVIRONMENT.md
19:34 nahamu (SmartOS specific, obviously)
19:35 Nazzy babilen, I follow http://docs.saltstack.com/topics/hacking.html?highlight=install#installing-salt-for-development and it "just works"
19:35 Nazzy there's a note for debian due to m2crypto though
19:36 babilen AdamSewell: What do you mean exactly? I am running into http://paste.debian.net/79532/ (would have expected both to be merged (i.e. oil to be present in the pillar)
19:36 AdamSewell babilen, not sure what you mean?
19:36 dwiden does anyone know what http://supernovae.in2p3.fr/~guy/salt/download/snfit-2.2.2b.tar.gz is for?  When I run brew install salt, I get an error when it tries to download this
19:36 dave_den pillar keys are not merged. they are overwritten
19:37 Nazzy I also have the advantage of gentoo having all the headers and dev requirements installed, which is handy
19:37 Mua joined #salt
19:37 Nazzy dave_den, pillar keys from separate pillar files
19:37 Nazzy ie, keys in later pillars overwriting keys in earlier values
19:38 dave_den it doesn't matter
19:38 NotreDev joined #salt
19:38 nkuttler is it just me or is the sound missing on http://www.youtube.com/watch?v=-zdkfHEW4LE ?
19:38 Nazzy dave_den, eh?
19:38 zach Can you use file.replace in the same state twice?
19:38 zach For the same file
19:39 dave_den Nazzy: oh, just saying to babilen that keys are not merged. they are overwritten
19:39 Nazzy dave_den, we've been working through this all day heh
19:40 dave_den i won't jump in mid converstaion, then.  :)
19:40 babilen dave_den: Yes, but what about keys that are not in *both* (e.g. "oil" in the paste)
19:40 babilen dave_den: No, please do.
19:40 dave_den 'ship' is the key
19:40 Nazzy something in his setup is causing pillars to overwrite whole sale ... I've tested on a bunch of versions and can't replicate it
19:40 blee_ joined #salt
19:40 Nazzy ship is a mapping, it merges
19:41 Mua_ joined #salt
19:41 babilen dave_den: It's just that Nazzy and I see different behaviour. I want the behaviour Nazzy sees (unique keys all make it to the pillar, non-unique ones get overwritten)
19:41 justBob joined #salt
19:42 Nazzy at least it does on 0.15, 0.16, and like 3 versions of develop from points in the last month ... I haven't tested 0.17 but I expect it to do the same
19:42 dave_den unless there was a recent change to pillar, it will not merge any keys
19:42 Nazzy I promise it does
19:42 AdamSewell joined #salt
19:42 [pyro] joined #salt
19:42 AdamSewell joined #salt
19:42 dave_den Nazzy: do you have an example that you are seeing where it merges?
19:43 Nazzy https://gist.github.com/kaithar/70c5e17d8d7f0326246f
19:43 Nazzy example pillar files, top.sls, result and tested versions that support it
19:44 Nazzy it's exactly what the pillar docs say is supposed to happen... I should know, I've been staring at them every waking moment for the past 3 days :p
19:44 rlarkin joined #salt
19:44 babilen heh
19:45 Nazzy apparently, when you want to write a pillar you have to figure out how they're supposed work first
19:46 babilen dave_den: Well, one thing is consistency here and the other is: How do I achieve the effect Nazzy is seeing? Background to this is that I want to define "roles" for certain classes of minions. '*': global_roles, 'foo-*': foo_roles, 'foo-db-*': foo_database_roles, (and so on)
19:47 babilen dave_den: And I don't want to have to copy and paste a long list of roles to every single host, but simply say: All roles in globs that apply to you
19:47 Nazzy as far as I can tell it only works with nested mappings... I'm expecting to find lists overwrite but have yet to test it
19:48 babilen dave_den: And now Nazzy is seeing one thing and I am seeing another ... I *want* Nazzy's behaviour, but can't seem to be able to reproduce it.
19:49 babilen This is all related to (long standing!) bugs such as https://github.com/saltstack/salt/issues/3991 or https://github.com/saltstack/salt/issues/2466
19:49 che-arne joined #salt
19:51 babilen Either way: If pillars are *not* supposed to behave like they do for Nazzy, but like they do for me I still am looking for a (obviously) different way to achieve what I want to achieve.
19:51 babilen I have the feeling as if what I want to do is not horribly arkane, but it seems to be impossible to do
19:51 dave_den trying to fire up my dev env to confirm, but taking a while
19:52 Nazzy and I need to look at a bug in how develop is merging keys heh
19:52 vejdmn1 joined #salt
19:52 babilen dave_den: Thanks; I'm in no rush and can also file a bug if that is deemed advantageous. My problem is simply that I am pretty new at salt and am therefore not sure what this should behave like.
19:52 ndrei joined #salt
19:53 Nazzy I was actually surprised that the older versions were able to grab pillar data at all, so that's something
19:53 babilen It's just increasingly frustrating as it turned a productive day of "let's develop formulas" into "lets hunt bugs" :)
19:53 Nazzy know that feel :p
19:55 Nazzy it's worth noting that now that I look I can't actually anything that explicitly states the dict.update call is done recursively, so it may not be intended behavior
19:58 justBob joined #salt
19:59 fllr joined #salt
19:59 babilen But keys in both dictionaries shouldn't be lost (assuming this is pillar1_dict.update(pillar2_dict) (
19:59 Nazzy it has occurred to me that I was being a bit blond earlier ... the version of the minion is fairly irrelevant since the update is done master side *facepalm*
20:00 babilen yeah
20:00 austin987 joined #salt
20:01 Nazzy as I say, staring at this stuff for 3 days while writing this pillar code... makes your head go wonky
20:01 babilen So, I would have to read some code, but if it is a simple update then keys (the "ship" here) will *not* be merged recursively .. that would mean that the behaviour I am seeing is the "correct" one (how desired is another question)
20:02 Nazzy yeah, if it was plain update() it would definitely not be recursive
20:02 babilen So, assuming I am seeing the correct behaviour I still wonder how to achieve what I am struggling with. :)
20:03 Nazzy heh... salt/utils/dictupdate.py is shiny ;)
20:03 babilen I think a pillar update/merge strategy (as discussed in #2466) is a highly desirable and needed feature .. but waiting for some far away version of salt doesn't help me now :)
20:03 babilen I guess I will have to dive into the code
20:03 babilen But first I would like a third party to verify that it behaves one way or the other
20:04 dave_den yeah, just confirmed on 0.17.4 that it does not merge
20:05 babilen thanks
20:05 dave_den https://gist.github.com/dlanderson/c8a219d6904db95cf390
20:05 babilen Okay, so .. what now?
20:05 dave_den i'm still surprised nazzy's is getting merged
20:05 hunter_ joined #salt
20:06 babilen indeed
20:06 babilen (and there goes my day of chasing the white rabbit)
20:06 Nazzy okie, then when did it change?
20:07 Nazzy salt/salt/pillar/__init__.py:423
20:07 dave_den nazzy: i've never seen it merge from what i can recall. i started using salt at 0.14
20:07 Nazzy I'll run blame
20:07 giantlock joined #salt
20:07 babilen Okay, lets assume I am seeing the "correct" behaviour. How would I then tackle the problem I want to tackle (I really don't care how, just that it works) ?
20:08 dave_den nazzy: you need to use an ext_pillar for anything more than basic pillar data compoilation
20:08 Nazzy not anymore
20:08 Nazzy found the commit
20:08 babilen hmm?
20:09 dave_den err, not nazzy. babilen ^^
20:09 Nazzy commit b338345c7a080e4b91dbe7dbf3b3b7e1315715cb
20:09 babilen dave_den: I am fairly new to salt, care to elaborate? I only used an external pillar to be able to use git
20:09 Nazzy PR #9184
20:10 Nazzy merged 11/Dec/2013
20:10 dave_den nazzy: pillar/__init__.py:423 is dealing with ext_pillars
20:11 dave_den babilen: sorry, i did not read the backlog. i assume you just want to have some defaults and be able to have more minion specific settings
20:11 babilen Why didn't that one make it into .5 ?
20:11 Nazzy https://github.com/saltstack/salt/pull/9184
20:11 Nazzy read :p
20:12 babilen dave_den: Well, it is a *bit* of a longer story. What I want is to assign "roles" to minions by globbing on their id. That is I want to be able to define "global" roles #
20:12 Nazzy when I start wielding git blame, I'll get to the right answer eventually ;)
20:12 dave_den nazzy: i was looking at my 0.17 code. yes, that code is it
20:12 babilen dave_den: for '*', then a foo roles for foo-*, a foo-database one for foo-db-* and finally a host specific one for foo-db-1
20:13 dave_den babilen: look at reclass
20:13 Nazzy ty :)
20:13 dave_den babilen: http://reclass.pantsfullofunix.net/
20:13 Nazzy hmmm, cmd_yaml or cmd_json as something less work?
20:14 babilen dave_den: I've seen reclass before, but thought that I can achieve what I want to achieve with "salt only" functionality
20:14 madduck hello babilen ;)
20:15 Nazzy though ... hmmm, they don't have anything minion selecting
20:15 Nazzy rats
20:15 babilen madduck: hey .. and so the circle is complete
20:15 hunter_ joined #salt
20:15 madduck FLOSS is a small world… ;)
20:15 babilen It is indeed
20:15 bhosmer joined #salt
20:15 Nazzy you'd be quackers to doubt it......
20:16 madduck babilen: you'd be able to do what you want with Salt only I think, though only if you stay very strict in terms of wildcard matching. I am not sure you can be sure about the ordering though.
20:16 madduck i promise reclass will do everything you want and more ;)
20:16 babilen So, would reclass allow me to work like that? I essentially want a state.sls file that gathers "roles" (defined for each minion via the globbing rules I discussed before)
20:16 dave_den babilen: with that merge that nazzy reference you may be able to. i would recommend using an ext_pillar that's smarter than dict.update, though.
20:17 madduck dave_den: reclass has deep merging.
20:17 dave_den madduck: yes, i know. i am suggesting babilen uses reclass instead of trying to force salt to get it right
20:18 babilen Well, I mean it isn't really that hard. I simply prefer a "data driven" approach and have to be able to control behaviour on four levels: 1. Global (all minions) 2. Per Customer (glob on foo-*) 3. Per-Customer, Per-Service (foo-db-*) and last, but not least, 4. Per minion (foo-db-1)
20:18 bhosmer_ joined #salt
20:18 Nazzy you could, if you were feeling particularly bored, write a really simple ext_pillar and shove the values in there
20:19 babilen If reclass is the best way to achieve that I happily use that, but I don't necessarily want to learn something just to realise that it doesn't allow me to go ahead.
20:19 madduck babilen: reclass has almost no learning curve, I bet…
20:19 madduck anyway, I am 99.9% sure it does what you want.
20:20 madduck you create a default class, and a class for each foo-* and a class for each foo-db-* and a node file for each foo-db-1
20:20 Nazzy you /might/ get away with dropping the new __init__.py in your extension_modules folder, or applying that patch against 0.17's __init__.py
20:20 babilen The idea for states I had so far was a very simple: http://paste.debian.net/79551/
20:20 madduck foo-db-1 inherits the foo-db class, which inherits from the foo class, which inherits from default
20:21 babilen madduck: Sounds great, how do I combine that with states?
20:22 dave_den joined #salt
20:22 babilen Nazzy: yeah, I am considering patching the Debian package and serving it from a local repo, but I always take situations like this as "Oh, beware! You seem to be the first, there is a more idiomatic solution around!"
20:22 Nazzy trying to sweet talk the loader in to overloading __init__.py falls in the very crazy option land... I don't think it'd work but it's worth trying if you feel a little crazy hehe
20:23 babilen I can always patch 0.17.5 with #9184, but then I would prefer the more idiomatic solution
20:23 Nazzy actually, given that it's master side, it's probably not that crazy to do a git install and backport that commit
20:23 mpanetta joined #salt
20:24 ipmb joined #salt
20:24 madduck babilen: the classes define the states ("applications")
20:24 Nazzy I have to disappear to home, shall be back later
20:25 * babilen waves
20:25 bharper joined #salt
20:25 babilen madduck: Okay, I know that you are a bit biased, but do you think that what I want to do is sensible and that reclass is the best way to do it right now?
20:26 nebuchadnezzar babilen: I'm a reclass user and that's what I do
20:26 dave_den for the quickest and simplest option, i would recommend reclass.
20:26 nebuchadnezzar a hierarchy of classes
20:26 madduck babilen: I'd say your usecase is why I wrote reclass.
20:26 nebuchadnezzar to classify nodes
20:27 babilen madduck: All I want is to be able to say "Done" even faster to requests such as "We want user FOO and BAR to have access to all database servers, here is their SSH key" (I would add it to foo-db-* minion group)
20:27 nebuchadnezzar and build pillar hierarchy
20:27 madduck babilen: you might be able to do what you want with other means, but I am fairly sure that I am not going to be leading you astray saying that the additional benefits of reclass (you will use more of its functionality once you embrace it) will outweigh any migration efforts (which should not be too much anyway).
20:28 ahwitz joined #salt
20:28 Ryan_Lane joined #salt
20:28 babilen madduck: Okay, perfect. Thanks for stating it so succinctly.
20:28 babilen Where can I find examples of such a setup? (/me always likes to read some code in addition to documentation)
20:28 madduck babilen: so if you add according pillar data to the foo-db-blue class, all nodes that inherit from foo-db-blue will immediately have access to the data. You can override for each node if you want, but you need not.
20:28 grim76 joined #salt
20:29 madduck However, reclass just massages data, the actual implementation of how the data are used wrt the local database you have to implement in states.
20:29 ahwitz Hey - I've been having issues with versions of salt/zmq and I've got to a point where I have a minion communicating with my master and able to access state sls files (I was unable to with this particular minion setup before), but now I'm getting (Exception AttributeError: "'list' object has no attribute 'keys'" in <bound method SREQ.__del__ of <salt.payload.SREQ object at 0x102c860d0>> ignored) every time I try to run an SLS. Any idea
20:29 madduck babilen: http://reclass.pantsfullofunix.net/concepts.html — other than that, I don't really have examples written up yet, sorry.
20:30 babilen sure, I'll read soem documentation then :)
20:30 babilen *some
20:30 madduck babilen: there is also https://groups.google.com/forum/#!searchin/salt-users/reclass/salt-users/FKqhVx0MZ-s/EXQNuxHIbnEJ and https://www.youtube.com/watch?v=CGPnbZ3KHBE
20:31 Gordonz joined #salt
20:31 Gordonz joined #salt
20:39 babilen nebuchadnezzar: Do you, by chance, have a salt+reclass setup that you could share?
20:40 * madduck really needs to make a tangible example usecase for the web…
20:40 seanz Greetings. Does salt delete the file copied during a cmd.script after the script has been run?
20:41 madduck seanz: it should; did you try it?
20:41 babilen madduck: I take it that reclass classes are not available as grains (yet) or that there are no other ways of targetting nodes by them? (as noted in the documentation)
20:42 yomilk joined #salt
20:43 seanz madduck: I saw that scripts are copied with a temp name to the /tmp dir, but I didn't verify that the script is deleted.
20:43 madduck babilen: I have not looked into this for a while, I doubt anything has happened though.
20:43 njs126 joined #salt
20:43 madduck seanz: please check and file a bug if the files are not deleted; they really ought to be
20:43 babilen madduck: Ack, how does that relate to http://docs.saltstack.com/ref/tops/all/salt.tops.reclass_adapter.html ? (sorry, just trying to tie things together)
20:44 kalloc joined #salt
20:44 madduck babilen: I don't understand the question; the master_tops adapter for reclass basically just serves the states (reclass "applications") that are applicable to a node to salt.
20:45 babilen madduck: nvm, I also pushed more salt documentation on the stack
20:45 madduck babilen: I would love to get patches of any form, especially docs ;)
20:47 Gareth /w 20
20:51 seanz madduck: We just checked the source - the file *is* deleted. We're all good.
20:51 dlindquist joined #salt
20:51 madduck seanz: don't you love open source software? anyway, just remember knuth when checking code for behaviour instead of verifying it…
20:53 taion809 joined #salt
21:00 ajw0100 joined #salt
21:03 n8n joined #salt
21:04 Nazca__ madduck: reclass does sound rather nice, albeit not something I can immediately use
21:04 bharper left #salt
21:07 madduck Nazca__: it does require to take data out of salt, true. But it's worth it ;)
21:07 toofer joined #salt
21:10 Nazca__ madduck: it's more that said data is destined to live in a MySQL database in my setup
21:10 n8n joined #salt
21:10 kalloc joined #salt
21:11 juicer2 joined #salt
21:13 madduck i am sorry about that
21:13 madduck what's a "MySQL database"? ;)
21:14 TheRealBill data doesn't *live* in MySQL, it struggles to *survive*.
21:14 madduck rofl
21:14 mgw joined #salt
21:14 gadams999 left #salt
21:15 bhosmer joined #salt
21:15 madduck Nazca__: if you still have a choice, use PostgreSQL for your own benefit and sanity. And I'd love a storage adapter for postgresql. It's just three functions to interface with reclass!
21:16 madduck anyway, my pillow is calling LOUDLY
21:16 * madduck waves
21:16 Nazca__ any reason for that?
21:16 * timoguin really wants to play with reclass soon
21:16 Nazca__ MySQL I know, PG not so much
21:17 TheRealBill Don't need to know much for PG ;)
21:17 gfa joined #salt
21:17 Nazca__ it's not short on flaws, but it works reasonably sanely once you throw out their config files and write one that wasn't tuned to run on a toaster
21:17 gfa any chance to get the openstack.config module backported to 0.17 branch?
21:20 fllr joined #salt
21:23 Nazca__ (I have to build a better mysql pillar anyway... my need for that has only two requirements: db that can be talked to from C# with the least work (and said C# app already knows how to deal with MySQL servers), and must be in to testing by Monday)
21:24 madduck Nazca__: there are plenty examples out there showing that MySQL does not implement the ACID principles of database design and moreover violates the principle or least surprise on many counts.
21:25 madduck sure, you can make it work, but it's not a "sustainable choice" and I'd consider every minute invested into MySQL a wasted minute. So, if you have the choice…
21:25 madduck but now — finally — good night.
21:28 ndrei joined #salt
21:30 cast joined #salt
21:30 beardo joined #salt
21:34 hunter_ joined #salt
21:35 hunter_ joined #salt
21:35 fllr joined #salt
21:37 viq MTecknology: hey, I told you, I'm traveling, haven't had much chance to touch computer :P
21:37 babilen madduck: guess I'll watch your DC talk, again :)
21:38 NotreDev joined #salt
21:39 Vivi-Idle joined #salt
21:46 ofosho joined #salt
21:47 ajw0100 joined #salt
21:48 msciciel_ is this bug in salt that during new minion registration and first highstate, mine functions are executed before custom modules are synced to minion, and if mine function is using custom module then it fail ? Or maybe i need to configure something ? I was experimenting with reactor but i could reach expected result :(
21:50 GradysGhost joined #salt
21:50 jp_cognet joined #salt
21:51 jp_cognet sorry if this is a completely ignorant question, but I can't seem to find my answer anywhere. Why is "network.get_hostname" is not available.?
21:51 jp_cognet I'm running 17.5 client and server
21:53 cedwards jp_cognet: how are you trying to use it?
21:53 jp_cognet very simply: salt 'host' network.get_hostname
21:53 Nazca__ and have you considered the grain 'fqdn' ?
21:53 jp_cognet nope
21:54 cedwards jp_cognet: odd, just tried it on my machine and it worked ok. what distro/ver are you running?
21:54 jp_cognet ubuntu 12.04lts on the client
21:54 jp_cognet let me check server, sec
21:55 jp_cognet salt '*' grains.get fqdn works btw
21:55 ndrei joined #salt
21:55 jp_cognet now the network.get_hostname just bugs me... lol
21:55 jp_cognet 12.04lts on the server
21:56 cedwards i don't think it should require anything special to run
21:56 jp_cognet I wouldn't think so either
21:56 ofosho Should there be any difference between installing a package with yum, and installing it through salt?  Installing 'httpd' through salt doesn't create the 'apache' user on my CentOS box, but doing a 'yum install httpd' works just fine.
21:56 GradysGhost I can verify the same output from a CentOS 6.5 box
21:56 Nazca__ madduck: and that's why MyISAM can go die in a fire ... Innodb at least tries to be sensible
21:56 GradysGhost salt '*' network.get_hostname
21:56 jp_cognet Grady, you are getting the error too?
21:56 jp_cognet or it works for you?
21:57 cedwards ofosho: there have been some recent issues with the yum pkg module i believe
21:57 GradysGhost I'm getting the same error reported from all minions.
21:57 jp_cognet yeah
21:57 jp_cognet same
21:57 jp_cognet w00t
21:57 jp_cognet it's not just me!
21:57 jp_cognet but the grains thing with get me by, just more annoying now really
21:57 GradysGhost becauuuussee...
21:57 fllr joined #salt
21:57 GradysGhost /usr/lib/python2.6/site-packages/salt/modules/network.py has no function get_hostname
21:57 GradysGhost Was that deprecated?
21:58 jp_cognet ahh
21:58 seanz madduck: haha, yes. Well, I will verify it before assuming.
21:58 jp_cognet yeah if I do, salt 'host' network. then press tab
21:58 jp_cognet it's not listed as a function
21:58 jp_cognet maybe they should remove it from docs ;)
21:58 Nazca__ *asks a sample of nodes*
21:59 cedwards if it's been deprecated it's strange that it works on my 2014.1.0rc3 machines
21:59 whitepaws joined #salt
21:59 GradysGhost cedwards: What minion version are you running?
22:00 Nazca__ it does, yes
22:00 cedwards GradysGhost: 2014.1.0rc3 on minion and master
22:00 GradysGhost I've got 0.17.4
22:00 GradysGhost Is that a minion version?
22:01 Nazca__ out of 14 hosts, 5 report "network.get_hostname" is not available.
22:01 jp_cognet I've got 17.5
22:01 jp_cognet but I tested it back to version like 11.5
22:01 jp_cognet for my minion version
22:02 Nazca__ I've got success on 2014.1, failure on 0.16.4 and 0.15.3
22:05 Nazca__ https://github.com/saltstack/salt/blob/develop/salt/modules/network.py#L336
22:05 jp_cognet so whats the word? basically use fqdn grain? get_hostname is deprecated?
22:05 Nazca__ actually it's new
22:05 Nazca__ https://github.com/saltstack/salt/commit/d9a9cfae4114e9495330073780c4b46def95dac0
22:07 GradysGhost If I'm reading that tag/version right, that indicates it went to RC status first of the year?
22:07 Nazca__ correct
22:07 Nazca__ it will be in 2014.1.0rc3 and later
22:07 jp_cognet ah, cool
22:07 jp_cognet ty
22:07 jp_cognet ty for the info I mean
22:08 Nazca__ correction, it was in r1 onwards
22:09 Nazca__ git knows best :)
22:10 Nazca__ but if it tells me to clean my room or eat my vegetables, I will disown it and go to HG ¬_¬
22:11 tedski i'm having trouble finding anything on the googles about the proper way to handle bonding configuration on ubuntu
22:11 kalloc joined #salt
22:11 tedski since network.managed isn't supported in ubuntu in latest
22:16 taion809 joined #salt
22:16 Nazca__ tedski: as in how to roll them by hand?
22:16 JordanRinke Morning
22:18 tyler-baker joined #salt
22:18 ajw0100 joined #salt
22:19 anitak joined #salt
22:22 dlindquist joined #salt
22:22 tedski Nazca__: yeah
22:23 tedski Nazca__: i'm having issues with the bit where in a workflow mindset, you'd have to /etc/init.d/networking stop before you edit /etc/network/interfaces so the interfaces are deconfigured properly first
22:23 tedski i have a hacky workaround where i do a cmd.wait on network/interfaces to run '/etc/init.d/networking restart ; ifconfig eth0 0'
22:23 tedski but, that's embarrassing
22:24 Nazca__ tedski: I'm gonna have to go with "urg" for that
22:24 tedski Nazca__: i thank you for that... it's well well deserved
22:25 tedski Nazca__: after applying said state, i then git rm'd because it was a means to an end
22:28 tedski Nazca__: so, what's the right way? :)
22:30 taion809 joined #salt
22:30 dlindquist joined #salt
22:32 Nazca__ on the whole, I detest configuring bonding
22:32 jp_cognet why?
22:33 Nazca__ what could go wrong? let me count the ways
22:33 Nazca__ :p
22:33 tedski configuring bonding ... at all or via salt?
22:33 Nazca__ at all
22:33 netzmonster joined #salt
22:34 tedski i'm not a fan either, but the right way is neteng's problem to solve
22:34 jp_cognet never really given me that much trouble
22:34 dlindquist joined #salt
22:34 tedski and if they want to have an active/standby network, well, not my call, right?
22:34 tedski jp_cognet: do you configure bonding via salt?
22:34 Nazca__ oh no .... active/active bonding
22:34 jp_cognet no
22:35 tedski Nazca__: i'm talking active-backup (mode 1)
22:35 Nazca__ we use mode 5 and 6
22:35 tedski Nazca__: active-active via lacp has also been pretty painless for us
22:35 jp_cognet yeah, I've done mode 1,4,5,6
22:35 jp_cognet all worked well for me
22:35 jp_cognet not with salt tho
22:35 Nazca__ lacp ... well lets start with xmit_hash :)
22:35 tedski yeah, 1 and 4 for us
22:36 Nazca__ or whatever it's called
22:36 tedski Nazca__: i didn't say it wasn't problematic, mind you :)
22:36 tedski i just said it was "fairly painless"
22:36 tedski heh
22:36 tedski that means pain was felt
22:37 Nazca__ then there's the lovely switches that run at near 100% cpu the moment you turn on bonding
22:37 tedski yup
22:37 tedski but, that's only in modes 4/5
22:37 tedski not lacp
22:38 tedski ime, at least
22:40 bretep So. I've got ext_pillars working. When I do a salt-call pullar.items I see the correct values. https://gist.github.com/bretep/82e2ce51024a8c45c2a0
22:40 bretep But for some reason they don't work in my template. My apt bucket is {{ pillar['hv_globals']['apt/bucket'] }}
22:40 Topic for #salt is now Welcome to #salt | Thanks to everyone who made SaltConf 2014 amazing! | 0.17.5 is the latest | Please be patient when asking questions as we are volunteers and may not have immediate answers - Channel logs are available at http://irclog.perlgeek.de/salt/
22:40 Topic for #salt is now Welcome to #salt | Thanks to everyone who made SaltConf 2014 amazing! | 0.17.5 is the latest | Please be patient when asking questions as we are volunteers and may not have immediate answers | Channel logs are available at http://irclog.perlgeek.de/salt/
22:40 basepi topic spaaaaaam
22:41 Nazca__ Woo! Wooo!
22:44 Nazca__ so, basepi, the question I'm sure a lot of us are wondering ... retag as 2014.2?
22:44 basepi ???
22:45 basepi Oh I see. =) nope, we're going off of when the expected release was in this case. =P
22:45 basepi Otherwise the tagging model disintegrates.
22:46 Nazca__ just a bit.. but better late than not-ready :)
22:47 bretep I would imagine accessing pillars should be easy.
22:48 bretep If any of you have time to comment on https://gist.github.com/bretep/82e2ce51024a8c45c2a0 that would be awesome
22:49 Nazca__ *pets NR sadly*
22:49 Nazca__ tried it without the / ?
22:51 lzhang joined #salt
22:52 lzhang Is there a way run salt master commands programmaticly through the libs themselves? Or do I need to call the command line
22:52 bretep Nazca__: were you talking to me?
22:52 Nazca__ bretep: yes on both lines
22:52 Nazca__ lzhang: salt-api is your friend
22:53 lzhang Nazca__: Thanks!
22:53 bretep Why would '/' in a key cause any issues. It's just a dictionary.
22:53 salt_whitebelt joined #salt
22:53 fllr joined #salt
22:54 Nazca__ it shouldn't cause a issue ... but you've got jinja and yaml thrown in to the mix, so I figured it's worth considering
22:56 salt_whitebelt im using cmd.script to call two powershell script, for ex: salt windows.minion cmd.script 'salt://script1.ps1' | salt://script2.ps1 shell='powershell' what is the command i can use to have the first script run wait 30 seconds before next one run?

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