Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2013-09-14

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

All times shown according to UTC.

Time Nick Message
00:00 berto- joined #salt
00:07 halfss joined #salt
00:09 pipps1 joined #salt
00:14 aat joined #salt
00:15 LethalTeamX joined #salt
00:15 LethalTeamX Hey
00:22 cedwards joined #salt
00:26 pipps joined #salt
00:29 troyready joined #salt
00:29 ipmb joined #salt
00:30 StDiluted joined #salt
00:31 m_george|away joined #salt
00:50 halfss joined #salt
00:57 redondos joined #salt
00:59 Gwayne joined #salt
00:59 Gwayne joined #salt
01:01 mgw I just read this post by Tor Hveem: http://www.bidne.org/salt-icinga-nrpe-replacement
01:01 mgw How are others integrating salt with monitoring?
01:02 m_george left #salt
01:02 KyleG mgw: I'm waiting for the native monitoring to come out. Having had conference calls w/ the salt stack CEO and talking to Dave and the others, as well as Tom, it's coming on it's own.
01:02 KyleG at least that's what it seemed like the last I heard
01:03 mgw KyleG: I've discussed it with Tom too, but not in detail… certainly the native monitoring is where it's headed, but I'm speaking of the present :-)
01:03 KyleG very well then. :P
01:04 KyleG I personally have found that check_mk's features and the way it does checks means that if I tried to add salt to the equation I would be slowing things down considerably.
01:04 KyleG 900+ service checks every minute on a low powered VM, and it's fast
01:05 StDiluted joined #salt
01:05 KyleG anyways i"m out, have a good weekend everyone
01:07 oz_akan_ joined #salt
01:09 oz_akan_ joined #salt
01:29 bhosmer joined #salt
01:37 imaginarysteve joined #salt
01:41 StDiluted joined #salt
01:42 alunduil joined #salt
01:46 fxhp joined #salt
02:02 mmilano joined #salt
02:05 sixninetynine joined #salt
02:10 mmilano left #salt
02:14 jalbretsen joined #salt
02:19 oz_akan_ joined #salt
02:27 baniir joined #salt
02:32 oz_akan_ joined #salt
02:42 juicer2 joined #salt
02:42 Ryan_Lane joined #salt
02:43 m_george|away joined #salt
02:53 mmilano joined #salt
02:55 Ryan_Lane joined #salt
03:00 berto- joined #salt
03:11 vipul joined #salt
03:23 zooz joined #salt
03:34 sssslang joined #salt
03:35 forrest joined #salt
03:40 mannyt joined #salt
03:43 mannyt joined #salt
03:49 robertkeizer joined #salt
04:00 PolyDOS joined #salt
04:02 sssslang is there a way to let minion reload it's config?
04:02 forrest did you restart the service?
04:03 sssslang I don't use service to manage minion. just start them from command line.
04:03 forrest ?
04:03 forrest service salt-minion start right?
04:04 forrest or do you actually run the bash script
04:04 sssslang does minion accept some singnals like sighup?
04:04 sssslang I just run '/path/to/salt/bin/salt-minion -d' to start minions
04:05 forrest why not use the service
04:06 sssslang cause salt is not installed by system package management tools. I compiled it myself.
04:06 forrest with the bootstrap?
04:06 sssslang so there isn't some control script in /etc/init.d
04:06 sssslang nope.
04:06 forrest ugh
04:06 forrest that sounds horrible
04:06 forrest lol
04:07 sssslang I use pretty old system. I need to compile zeromq, openssl myself.
04:07 forrest check the minion script and see if there's an option to restart it
04:07 forrest I don't know offhand if there is
04:07 sssslang there's a big environ with salt.
04:07 sssslang ok, I'll read the minion's source code :-(
04:08 forrest or just restart salt using salt
04:09 PolyDOS left #salt
04:09 sssslang using 'pkill salt-minion; salt-minion'?
04:09 forrest no
04:10 forrest just salt-call --local service.restart salt-minion
04:10 forrest but that might require the init script :\
04:10 sssslang I've no idea what's the content in /etc/init.d/salt :-/
04:15 sixninetynine joined #salt
04:16 bkonkle left #salt
04:18 xl joined #salt
04:21 UtahDave joined #salt
04:22 forrest UtahDave, you ready for that Friday night IRC party? Wooooooooo
04:25 whit joined #salt
04:26 sssslang forrest: I want to give different minions some tags, how should I do this in salt?
04:26 forrest do you mean you want to group minions, or you want to apply specific states to certain ones?
04:28 sssslang the latter. e.g. minion1:[loc1, role2, dep1], minion2:[loc2, role2, dep3]
04:28 forrest I'd suggest looking at overstate: http://docs.saltstack.com/ref/states/overstate.html
04:28 forrest if you need to do them in order.
04:29 forrest if you just want specific roles applied to each one, you can define that value in your top.sls
04:29 forrest just take a look at http://docs.saltstack.com/ref/states/top.html there is a good example in there of items getting applied
04:29 sssslang yes, just like roles.
04:30 forrest if you wanted get fancy, you could base what gets applied off of grains, or pillar data.
04:31 sssslang It seems to need manage a minions list on master. But I've no idea where to save this list.
04:32 sssslang if the minions are tagged, I could using these tags in top.sls.
04:32 forrest https://gist.github.com/gravyboat/6558845
04:33 sssslang Or the top.sls would be confused when number of minions are large..
04:33 forrest right which is why you incorporate logic within the top.sls, along with logic within the states themselves
04:34 forrest so you know that by looking up the grain data, the role2 file either sets the hostname to minion1, or minion2.
04:34 forrest because it references the data off that box you're configuring
04:34 forrest there's http://docs.saltstack.com/ref/configuration/minion.html#std:conf_minion-id
04:34 forrest but that's basically the same as hostname.
04:35 forrest which it defaults to
04:35 forrest so I guess you could modify the tag for groups of machines
04:36 sssslang The tags could be too much to fill in the id field.
04:38 forrest hmm
04:38 sssslang Sometimes need logic operation using tags: 'tag: dep1 and tag: role2', I don't think this could be done in top.sls.
04:38 forrest yea I don't know of a direct equivalent to tagging a system in salt :\
04:38 UtahDave sssslang: a good place to set roles is either in grains or in pillar
04:38 forrest well, you should be writing the logic to do that right? So you know that in your top .sls *.webserver get a bunch of states
04:39 UtahDave then you can match on those roles
04:39 forrest We discussed grains already UtahDave :\
04:40 sssslang UtahDave: thanks. I'm a little confused about grains and pillar. I thought grains are like hard-class in cfengine, but I found it can be changed during runtime.
04:41 UtahDave sssslang: Grains are generally static items of information about each minion.  But you can programmatically create new grains and set your own grains as well
04:42 forrest sssslang, as much as terminalmage hates it because it summons him, check out: https://github.com/terminalmage/djangocon2013-sls He's got a good example in there of using pillar for app configuration
04:43 sssslang here is my top.sls in pillar: http://pastebin.ca/2450287. The problem is, however, I don't know which is a proper way to manage a minion list.
04:44 pygmael joined #salt
04:45 UtahDave sssslang: that's your top file?
04:46 sssslang sorry, roles.sls mentioned in pillar/top.sls
04:47 sssslang I think there should be a minion list, then I can import this list, setting tags to each minion one by one. Is it a right way?
04:55 halfss joined #salt
04:56 forrest I don't think it is sssslang, but I don't know the official saltstack stance
05:01 forrest Are all your machines so unique that you're unable to write logic to handle it within the states/files that are templatized?
05:01 ldlework is there a way to remotely control a salt master?
05:03 sssslang forrest: Sometimes. different departments have different requirements.
05:04 forrest so why can't you create file_roots per department or something
05:04 forrest http://docs.saltstack.com/ref/file_server/file_roots.html
05:07 sssslang forrest: thank you. I'll try to change it.
05:07 aat joined #salt
05:07 forrest yea np, sorry if I'm confusing you, just trying to understand what you need, and what the easiest/most future proof solution is.
05:09 sssslang no. I'm not a system administrator. Maybe I'm heading a wrong direction in configuration management.
05:09 forrest Yea, just gotta explore and figure out what works for your environment.
05:11 oz_akan_ joined #salt
05:12 sssslang forrest: :-)
05:14 terminalmage forrest: maybe I can configure weechat not to notify me when you mention me :D
05:14 terminalmage j/k :P
05:14 forrest lol
05:14 forrest I'll start linking my rip off example :P
05:14 terminalmage haha
05:14 terminalmage it's not a problem, really
05:14 forrest 'screw that terminalmage guy and his example with more details!'
05:14 forrest Yea I know, I was just joking
05:14 terminalmage :)
05:14 terminalmage had a bit of an issue with my laptop
05:15 forrest yea? You didn't want to work so you broke it?
05:15 terminalmage :)
05:15 terminalmage that's one of the few things that will bring me away :)
05:15 forrest hah
05:15 forrest Oh were you able to check who was going to devops atlanta?
05:15 terminalmage oh yeah
05:15 terminalmage it's the sales guys
05:15 forrest oh ok cool
05:15 terminalmage and our marketing guy, rhett, I believe
05:16 forrest Ahh I don't think I've met him. I'm sorry to say I can't remember the name of the sales guy who was at pycon :\
05:16 terminalmage I think that one of the guys that is going to be there is a very recent hire
05:16 terminalmage like, <2 mos
05:17 terminalmage so it's doubtful you'd have met him
05:17 forrest ahh ok
05:17 terminalmage anyway, so there is a bug in the intel video drivers in recent kernels
05:17 forrest for which distro?
05:18 terminalmage Arch Linux, but they release vanilla kernels
05:18 forrest why am I even asking which distro? I should have said 'which distro does not constantly have video driver issues'
05:18 terminalmage :)
05:18 terminalmage actually this is the first time I've had a video issue
05:18 forrest really??
05:18 terminalmage and I've used arch for almost 5 years
05:18 forrest Maybe that's just a nice feature of arch
05:19 forrest last time I installed Fedora it took me 4 hours to get the drivers working
05:19 terminalmage Arch is the best.
05:19 terminalmage https://wiki.archlinux.org/index.php/Arch_is_the_best
05:19 forrest lol
05:19 terminalmage see, it's right there in the wiki
05:19 forrest :P
05:20 terminalmage :D
05:20 terminalmage so yeah, I had to use my phone to find a thread with some dude that has a patched kernel, download the PKGBUILD, and compile it
05:20 forrest http://www.youtube.com/watch?feature=player_detailpage&amp;v=yphLKSjnSU8#t=2346
05:20 forrest That's for you terminalmage
05:21 terminalmage well, I downloaded and compiled on my laptop, while chrooted off of a boot USB
05:21 forrest heh
05:21 lynxman joined #salt
05:21 lynxman joined #salt
05:21 terminalmage forrest: ahh I haven't seen this
05:21 forrest Yea it was not fun when fedora was acting up, had to install the drivers in run level 3, then reboot into run level 1, remove some stuff that couldn't be remove in 3/5, then configure the drivers, was not fun
05:22 terminalmage forrest: meh, I live on the command line
05:22 forrest Yea this was just messing around with the release after they redid the GUI
05:22 forrest wanted to see what it looked like
05:22 terminalmage ah
05:23 terminalmage the one thing that isn't working right now on this laptop is screen backlight control
05:23 terminalmage it's all the way turned up
05:23 forrest ouch
05:23 terminalmage yeah... but the intel bug basically makes the backlight not work at all
05:24 terminalmage so, there's a tradeoff
05:24 forrest You could TRY to use http://justgetflux.com/linux.html
05:24 terminalmage brightness up to 11, or nothing
05:24 forrest so it at least adjusts the screen to a better hue
05:24 terminalmage eeeew, a GUI?
05:24 forrest I assume that someone already filed a bug report for those drivers? Or did you
05:24 terminalmage :P
05:24 forrest *shrug*
05:25 forrest hey man what do you want, it's free
05:25 forrest if you're booting to run level 3, worthless
05:25 terminalmage someone else has, if they're not fixed soon then Arch might roll the changes into their core kernel
05:25 forrest ahh ok
05:25 terminalmage I use systemd, the notion of runlevels is different
05:25 terminalmage but I boot to a tty and run startx to launch my WM
05:26 terminalmage no display manager, don't like 'em
05:27 forrest do you do all local development or do you ssh into a VM?
05:29 terminalmage I write the code on Arch, but I have VMs that I use for testing
05:29 terminalmage mostly virtualbox
05:29 forrest gotcha
05:29 terminalmage but a few on rackspace
05:31 terminalmage yeah, I set up a 2nd interface on the VMs, and use host-only networking so that I can SSH directly into it
05:32 terminalmage for the local ones that is
05:32 forrest Yea standard setup
05:33 terminalmage yeah, you'd be surprised how few people think of that
05:33 terminalmage I run a LUG and I've showed at least half a dozen people that
05:33 forrest at work I had to set mine up through our DHCP server because of some shennanigans
05:34 forrest but at home *shrug*
05:34 forrest granted I mostly do my work on remote VMs
05:49 forrest hey s0undt3ch, did your merge for https://github.com/saltstack/salt/blob/develop/salt/states/cmd.py test properly? Travis CI is just showing errors, and I get errors on building the docs.
05:51 forrest oh yea I see the error now
05:52 akio joined #salt
05:53 hotbox_ joined #salt
05:53 pmcg joined #salt
05:53 heckman joined #salt
05:53 kamal_ joined #salt
05:56 djinni` joined #salt
05:58 terminalmage woot fixed screen backlighting
05:58 terminalmage band-aid fix but whatev
05:58 forrest That's kind of a bummer, I was going to make a witty mark about wearing sunglasses at night
06:01 terminalmage hehe
06:02 terminalmage the backlight stuff wasn't broken, per se, it's just that it was broken via ACPI events
06:02 forrest oh nice, what did you end up doing to fix it
06:04 terminalmage the Arch User Repository has a little shell script that basically does echo $value >/sys/class/backlight/intel_backlight/brightness
06:04 forrest lol
06:05 terminalmage yeah. mapped that script to the normal brightness keys
06:05 terminalmage voila
06:06 forrest did you make it auto-increment the value based on the button you hit? So it just pulls it out and adds 1?
06:06 forrest or subtracts
06:06 terminalmage nah, the script takes care of all that
06:06 terminalmage there's a down or up option
06:06 forrest oh ok
06:06 terminalmage https://wiki.archlinux.org/index.php/ASUS_Zenbook_Prime_UX31A#Method_2
06:07 forrest ahh
06:09 xl left #salt
06:28 az87c joined #salt
06:28 az87c_ joined #salt
06:29 bhosmer joined #salt
06:39 terminalmage ok, way past my bedtime. night folks.
06:40 forrest later
06:51 balboah_ joined #salt
07:01 Ryan_Lane joined #salt
07:11 nu7hatch joined #salt
07:35 durnik joined #salt
07:36 pmcg joined #salt
07:37 sgviking joined #salt
07:37 cowmix joined #salt
07:46 Lue_4911 joined #salt
07:48 p3rror joined #salt
08:01 xl joined #salt
08:09 mgw joined #salt
08:22 Ryan_Lane joined #salt
08:25 sgviking joined #salt
08:38 xinkeT joined #salt
09:03 txmoose So, on my CentOS 6.4 server, I just noticed that salt-minion had 113 instances running, and salt-master had more than 10 instances running.
09:03 txmoose Has anyone noticed that before?
09:09 sfello joined #salt
09:12 TheCodeAssassin joined #salt
09:23 redondos joined #salt
09:29 bhosmer joined #salt
09:35 jrgifford joined #salt
09:38 mgw txmoose: multiple salt-master is normal
09:38 mgw I think the salt-minion procs is a bug
09:53 foxx[cleeming] joined #salt
10:17 redbeard2 joined #salt
10:20 zooz joined #salt
10:37 lemao joined #salt
10:49 eliasp joined #salt
10:49 eliasp hi
10:49 eliasp coming from a Puppet background, I'm just taking a look at Salt and it looks really really nice…
10:50 eliasp I'm wondering whether there's something like puppet-librarian or r10k for Salt to deploy the environments/modules/etc. on the Master…
10:52 sgviking joined #salt
11:15 MrTango joined #salt
11:16 nbari joined #salt
11:16 nbari hi all, any idea of how to call command from the mysql pluging for example mysql.version  within a sate ?
11:17 nbari I want to send a dump.sql via file.managed and wait until is on the server and later call a series of mysql.* commands
11:17 nbari any idea ?
11:20 djinni` joined #salt
11:20 unicoletti_ joined #salt
11:20 xl A cmd.wait that watch: the sql file?
11:21 nbari /root/dump.sql.gz:  file.managed:- source: salt://mysql/dump.sql.gz after that I want to do something like mysql.user_list: - watch: - file: /root/dump.sql.gz
11:22 nbari but wondering how to call the plugin 'mysql' within the salt.sls
11:23 ghanima joined #salt
11:23 xl You mean salt.modules.mysql? You can use module.wait/module.run state to call its functions
11:23 nbari right
11:24 nbari ok let me search an example :)
11:25 xl http://docs.saltstack.com/ref/states/all/salt.states.module.html
11:26 nbari thanks
11:26 nbari https://github.com/saltstack/salt/issues/5936 I am folloing that one
11:26 nbari I think that name mut match in my case 'mysq' (hame of the module) right ?
11:28 xl name is the module function name
11:29 nbari right I am trying now with -m_name: mysql
11:29 xl Something like "name: mysql.user_list" should work
11:31 nbari many thankx xl :)
11:32 JoAkKiNeN joined #salt
11:46 nbari mmmm how to do something like -name: "mysql.query table 'Select 1 from table where a=""'" <-- how to properly quote?
11:46 nbari if I remove the "" the dot is treated as a dir
11:46 nbari but I need it so that I can use the mysql.query module but now how to properly put the "" on the query ?
11:46 mapu joined #salt
11:50 xl - name: mysql.query - database: somedb - query: select 1 from table where a = ""
11:51 xl might work
11:51 nbari many thanks, let me try
11:51 ronc joined #salt
11:53 nbari worked :D
11:54 xl :)
11:55 xl so it is generally - name: module.function - para: value - para: value
11:55 nbari i see
11:59 xl left #salt
12:15 yidhra joined #salt
12:18 samuel joined #salt
12:24 samuel Hello, I am trying to install package using salt and I want to preset some settings using debconf. What is the correct way to do it? I can not run debconf-set-selections before package is installed, however, after the package is installed, it is pretty useless.
12:28 sgviking joined #salt
12:31 ccase joined #salt
12:34 sibsibsib joined #salt
12:40 aat joined #salt
12:40 kvt joined #salt
12:50 superflit joined #salt
12:54 baniir joined #salt
13:10 tuxIO joined #salt
13:25 racooper joined #salt
13:31 mwillhite joined #salt
13:35 txmoose mgw: Is it normal to have 113 of them? When I only run 3 minions?
13:35 kchr joined #salt
13:40 gildegoma joined #salt
13:41 joehh samuel: This is probably useful: https://salt.readthedocs.org/en/latest/ref/states/all/salt.states.debconfmod.html#module-salt.states.debconfmod
13:42 joehh txmoose: have you been restarting the minion regularly?
13:43 joehh txmoose: Not sure with centos, but I suspect that the init.d scripts on debian (as they currently are) risk leaving
13:44 sfello joined #salt
13:44 joehh minion processes hanging around sometimes
13:45 ronc joined #salt
13:46 eliasp joehh: sounds like it's about time for you to switch to systemd :)
13:48 joehh eliasp: :) currently the packages support sysv-init, systemd and upstart. The difficulties of getting agreement in something the size of debian...
13:49 eliasp :)
13:58 felixhummel joined #salt
14:09 sfello joined #salt
14:11 redondos joined #salt
14:19 adepasquale1 joined #salt
14:36 bhosmer joined #salt
14:45 MrTango joined #salt
14:57 sofiacampus joined #salt
14:58 HumanCell joined #salt
14:58 premera_w_j joined #salt
14:58 premera joined #salt
15:05 felixhummel joined #salt
15:05 GoKage_ joined #salt
15:06 GoKage_ i am coming from puppet, i was wondering how to do the equivalent exec and file function in salt
15:07 GoKage joined #salt
15:12 GoKage hi
15:12 GoKage how do you deal with files and execution of commands like what you would normally do with puppet's file and exec calls? is that all done with grains?
15:15 joehh I'm not that familiar with puppet, I think this is what you want: http://pastebin.com/FihxF1dc
15:16 joehh ie myscript is downloaded and executed (cmd.wait) whenever it changes
15:16 joehh download is file.managed
15:16 GoKage so this would have the salt master manage this file right?
15:16 avienu joined #salt
15:17 GoKage sorry for the ignorance... i'm just getting started with salt. very impressed with the many salt presentations on youtube
15:17 joehh I think so - (not too sure on the terminology)
15:18 GoKage how would i run something like say.... apt get update before having the system go and install packages
15:18 joehh This fragment would be in an sls file (state file) on the master, as would the myscript script
15:19 joehh This will be "run" on the minion whenever it is called from the master
15:19 joehh either by: salt 'MINIONNAME' state.highstate
15:19 joehh if the state file is allocated to the minion in the top.sls file
15:19 joehh or: salt 'MINIONNAME' state.sls SLSNAME (without the .sls)
15:19 joehh if it is done as a one off
15:20 joehh no worries about the ignorance, we all start somewhere :)
15:22 joehh If you wanted puppet functionality of being updated every n minutes, then you would put the salt .. state.highstate in a cron job
15:22 joehh or use the scheduling functionality
15:22 joehh hope that helps
15:24 GoKage that makes sense... so there is no daemon
15:28 joehh there is a daemon, but in general the flow is from master to minion rather than minion (whatever it is called in puppet) asking the master what to do
15:29 joehh As far as I am aware, the master and minion maintain a fairly permanent zero mq link between them
15:32 ronc joined #salt
15:36 GoKage ic
15:36 GoKage in puppet there is the puppetmaster and puppet
15:36 GoKage so the minions don't call home?
15:36 joehh that makes sense
15:37 joehh not regularly - when they start up, they make a connection to the master, then wait/listen for instructions
15:37 GoKage but rather master will open multi threaded connections to the many minions right?
15:38 GoKage the puppet syntax and many files is starting to piss me off and hence why i'm looking into salt
15:38 joehh yes - getting a little deeper than I fully understand here, but I'll describe my perspective and hope the technicalities are correct
15:39 joehh many people seem to find salt much simpler - I haven't done much more than basics with puppet
15:39 joehh master will then either ask minion directly to do something or
15:40 joehh send out to all to do something (possibly with criteria) ie salt -G 'os:Debian' would make all debian machines respond
15:40 joehh similarly with -G 'os:Windows'
15:40 GoKage so more of a push scenario
15:40 joehh all minions with the "os" grain equal to "windows" would response
15:40 joehh yes
15:41 joehh I think that is a good way of differentiating, puppet is pull, salt is push
15:41 StDiluted joined #salt
15:42 UtahDave joined #salt
15:47 bhosmer joined #salt
15:49 felixhummel joined #salt
15:58 GoKage thanks
15:58 GoKage i'll give it a try
15:59 nbari joined #salt
16:00 nbari how could i copy files from minion to master ?
16:00 nbari using  salt
16:01 horus_plex joined #salt
16:03 Lue_4911 joined #salt
16:06 horus_plex joined #salt
16:07 superflit joined #salt
16:07 UtahDave nbari: http://docs.saltstack.com/ref/modules/all/salt.modules.cp.html#salt.modules.cp.push
16:07 nbari thanks
16:07 UtahDave you're welcome
16:08 nbari this salt is giving a very good flavor to my suff :)
16:09 UtahDave :)  I'm glad!
16:13 sibsibsib joined #salt
16:14 fragamus joined #salt
16:18 Katafalkas joined #salt
16:19 Katafalkas joined #salt
16:19 nbari is there a way to specify a diferent destination dir on the fly when using cp.push ?
16:20 nbari I want to get a mysqldump file , storit on the master and send it to the minions
16:20 nbari or any better aproach ?
16:21 UtahDave That's all you can do directly through Salt right now
16:22 UtahDave You could use cmd.run  to execute scp ./mysqldump root@salt-master:/path/to/proper/directory/
16:23 nbari ok
16:27 ccase Should I see salt/states/augeas.py in the v0.16.4 release tarball? I was expecting to see it, but its not there. It's also not in the git tag at that version, nor the ubuntu packages.
16:33 UtahDave ccase: if it's a new feature, then it will only be in the next full release. Which will be 0.17.0
16:36 redondos joined #salt
16:37 drags UtahDave: speaking of .17, are the .17 docs supposed to be showing up on docs.saltstack.com? (ex: http://docs.saltstack.com/ref/modules/all/salt.modules.grains.html#salt.modules.grains.append)
16:37 bhosmer joined #salt
16:37 ccase UtahDave: Thanks for the clarification. Is the augeas state a new feature then? I isn't listed as one on the docs.
16:37 UtahDave Yeah, some people are testing against the develop branch
16:37 drags ahh
16:38 UtahDave ccase: I'm not sure, exactly. There has been some work done lately on augeas stuff.
16:38 UtahDave ccase: My vague memory of it is that we had an augeus execution module, but not a state.
16:38 UtahDave I could definitely be wrong though
16:41 ccase Ah yeah, I see it's pretty recent (2013-08-07, last update 09-05).
16:46 Thiggy joined #salt
16:48 ronc joined #salt
16:54 jcockhren joined #salt
16:56 jcockhren joined #salt
17:06 jcockhren joined #salt
17:09 unicoletti_ joined #salt
17:10 mmilano joined #salt
17:10 ccase Anyone else having trouble with sphinx generating the man pages on ubuntu raring? NotImplemented errors at the end: http://pastebin.com/jVjZQex8
17:11 ccase Googled around, but I could only find this http://www.mail-archive.com/sphinx-dev@googlegroups.com/msg01399.html
17:11 ccase Which basically says this should never happen.
17:13 jcockhren joined #salt
17:13 UtahDave ccase: not sure what's going on there. I wonder if you're missing a dependency or something.
17:22 jcockhren finally remembered my freenode PW. now I have finally have some presence here. \o/
17:23 ccase UtahDave: It appears there was something unhappy with the docutils 0.10 + sphinx 1.1.3. pip -U and its all happy again.
17:23 UtahDave ah, good
17:23 UtahDave yay, jcockhren!
17:23 jcockhren akoumjian: salty-vagrant's issue #100, I went ahead and did a PL directly to vangrant
17:24 jcockhren s/PL/PR
17:24 jcockhren vagrant
17:24 jcockhren UtahDave: yeah! I haven
17:26 aat joined #salt
17:26 jcockhren UtahDave: I haven't gotten around to doing any saltcloudmod just yet.
17:27 UtahDave cool
17:27 jcockhren been working through how automated website deployments would happen
17:27 jcockhren (for the dayjob)
17:27 jcockhren I've put them on salt.
17:28 UtahDave nice!
17:42 unicoletti_ joined #salt
17:42 unicoletti_ left #salt
18:02 copelco joined #salt
18:05 m_george left #salt
18:13 mech422 joined #salt
18:13 mezgani joined #salt
18:15 mech422 Hi all - anyone know a good way to manage config files with salt ?  (eg adding a new section to a config when you install addon, or removing it when you remove an addon )
18:15 mech422 all I've managed to find is file.patch
18:21 jcockhren mech422: maybe leverage jinja?
18:22 jcockhren mech422: you could create a config file "template" then use file.manage setting :template:jinja"
18:23 jcockhren "template: jinja"
18:23 jcockhren then have sections that appear only when pkg is cataloged as installed
18:24 mech422 jcockhren: hmm...so re-create the entire config file every time based on the current 'includes' ?  that sounds...complex ?
18:25 mech422 I'd want to be able to call out from the 'main' sls to each 'optional' component to get its config block
18:25 mech422 not sure my saltFu is up to that yet :-P
18:25 redbeard2 joined #salt
18:27 jcockhren mech422: I may be misunderstanding your request. a little more info?
18:28 jcockhren mech422: when you say "config file" you mean system specific configuration files right?
18:29 mech422 jcockhren: yeah - my current one is ceph.conf (/etc/ceph.conf) .. it has an '.ini' style config file
18:30 mech422 with 'sections' for different functions... for instance, I need to add a 'section' for a web gateway
18:32 jcockhren mech422: ok. cool. then yes.
18:32 Katafalkas joined #salt
18:32 mech422 or maybe I want to add/remove an entire interface from /etc/network/interface ?  pretty much anything with multiple options and 'sections' ?
18:33 jcockhren mech422: you can then have a service restart when changes are detected in that file
18:34 mech422 jcockhren: yep
18:35 mech422 guess I can just use a multi-line append for now... not sure how I'd take it out again later though
18:38 jcockhren mech422: my approach would use jinja's b/c I don't know of any salt based way to selectively edit lines of text in a file
18:39 mech422 jcockhren: oh? I'll have to read up more on jijna ... I never played with it before I tried salt
18:39 jcockhren mech422: there's 2 cons to that appraoch though
18:40 jcockhren 1. there's has to be some salt based way to detect addon installation/removal
18:40 jcockhren 2. You have to know all the possible sections before hand
18:41 mech422 eww..#2 could be a PITA
18:41 jcockhren yeah
18:41 jcockhren hmmm
18:41 mech422 jcockhren: well - let me RTFM on jinja a bit. maybe something will jump out at me
18:41 jcockhren cool
18:42 mech422 jcockhren: thanks for the help!
18:42 jcockhren no prob
18:45 sfello joined #salt
18:46 drags mech422: I'm not familiar with ceph, but the way a lot of applications make this easier to do is have an includedir type directive, usually called <appname>.d that config snippets get dropped intop
18:46 drags if there's anyway to include arbitrary files in ceph.conf try including /some/directory/*.conf
18:47 drags then have the /some/directory/*.conf snippets managed by salt
18:47 mech422 drags: yeah - I'm not sure thats supported yet...at least, I have noticed it in the docs
18:47 mech422 drags: I'll add that to my google list
18:47 drags (still run em through jinja though, it's a very flexible template tool and you can access salt grains in there)
18:48 drags mech422: if the program doesn't allow including a whole directory (/yadda/*) you could also still use the snippets dir, then use salt to manage the include lines (one for each snippet) in ceph.conf
18:48 JesseC joined #salt
18:49 drags also lastly: python's ConfigParser (included in standard dist I think) can read and write .ini files rather well
18:49 drags could make a custom module for ceph that used ConfigParser
18:49 auser joined #salt
18:49 mech422 drags: yeah - I was thinking about that... that would be a better long term solution I think ( failing config.d availability )
18:50 drags yeah if you have to go that route, good luck :)
18:52 mech422 drags: blah - as of 3/4/13, no #include support
18:52 drags mech422: if you're on github you might want to click 'Watch Thred' on https://github.com/saltstack/salt/issues/4043
18:53 drags mech422: ouch
18:54 mech422 drags: cool.. the ceph-deploy tool would be a start - but tbh just having config.d support would prolly be more flexible
18:54 drags yeah, very
18:54 mech422 ceph-deploy drives me batshit anyway :-P
18:59 tuxIO joined #salt
19:00 MKFG joined #salt
19:03 dyim42 joined #salt
19:05 dyim42 joined #salt
19:43 kiorky uhm, is there a way to have a meaningful return code for salt-call if at least one state failed or if the total exeuction fails (like state.sls on an inexistent state)
19:43 kiorky getting 2 or 0 is so useless
19:46 terminalmage kiorky: the return code is decided by analyzing the state results
19:47 kiorky terminalmage: well even with no errors, i have 2
19:47 kiorky terminalmage: and for an inexistent state, i get 0
19:47 terminalmage kiorky: I think that there's a bug in that bit of code
19:47 terminalmage to be honest
19:48 gcbirzan left #salt
19:48 terminalmage the core devs have a meeting on Monday, I'll bring it up then
19:50 kiorky terminalmage: pastening somethign
19:50 kiorky terminalmage: http://pastebin.com/BiBkMwGa
19:51 kiorky terminalmage: so parsing output is a bit crazy
19:51 kiorky first check logfile if there are ERROR entries
19:51 kiorky then if retcode != 2 or != 0
19:51 kiorky :D
19:51 terminalmage we don't parse output
19:52 terminalmage the state results are in a data structure
19:52 kiorky terminalmage: i mean, in the script we run some salt cmd, and we need to check if everything had been done fine
19:52 kiorky terminalmage: but, then to have the data structure, you need it to run
19:53 terminalmage yes
19:53 kiorky terminalmage: if the state is non existent for exemple you have an empty struct
19:53 kiorky terminalmage: for shell integration, is there any strong objection for having retcode != 0 with at least one failing state ?
19:54 berto- joined #salt
19:56 terminalmage yes, I know the struct is empty when there are no states being run
19:56 redondos joined #salt
19:56 terminalmage this is part of the aforementioned bug
19:59 kiorky terminalmage: well, i think the most easy thing for me to do for now is a double step shell function to check (if empty struct; check log)
20:02 Ryan_Lane joined #salt
20:04 terminalmage kiorky: https://github.com/saltstack/salt/issues/6973#issuecomment-24348712
20:09 baniir joined #salt
20:18 kiorky terminalmage: yep, but the problem is "2" for retcode when there are no errors
20:18 kiorky terminalmage: and also in fatal errors
20:19 kiorky terminalmage: http://pastebin.com/YpwsuDH6 that's my workaround for now
20:21 forrest joined #salt
20:22 cedwards s0undt3ch: I just tried to bootstrap a centos6 machine and it's failing..
20:22 forrest on what part cedwards?
20:23 cedwards forrest: package dependencies actually.. can't find python-jinja2.
20:23 redondos joined #salt
20:23 forrest did it pull in the epel repo?
20:23 cedwards yes
20:23 forrest can you do a yum list python-jinja2*
20:24 cedwards Available Packages
20:24 cedwards python-jinja2-26.noarch
20:24 mech422 heh - it would be nice if file.append could pass stuff thru jinja.. I just had to hack it via a file.managed /tmp/foo and cmd.wait 'cat /tmp/foo >> /etc/conf.file && rm -rf /tmp/foo'
20:24 diegows joined #salt
20:24 forrest yea that's the problem cedwards
20:24 forrest python-jinja2 should be coming from the centos repo, hmm
20:25 forrest can you remove the /etc/yum.repos.d/epel repo file, yum clean all, then relist yum list python-jinja2* ?
20:25 cedwards none found
20:25 forrest *sigh*
20:25 cedwards did epel rename the package?
20:25 forrest no
20:26 forrest it's had that terrible name for a while
20:26 forrest I had to commit a patch a while back because python-jinja2 is only in the optional packages for RHEL
20:26 forrest centos used to be ok
20:26 forrest let's see...
20:28 quinnat joined #salt
20:28 forrest I'm spinning up a centos 6 instance now cedwards, give me a minute
20:29 forrest Looks like digital ocean is under maintenance right now. Which repos do you have in /etc/yum.repos.d cedwards?
20:30 cedwards the base repos, and epel
20:31 forrest ok, so here I'm seeing it's available: http://mirror.centos.org/centos/6/os/x86_64/Packages/
20:31 forrest python-jinja2-2.2.1-1.el6.x86_64.rpm
20:32 cedwards i'm using a local repo mirror. i wonder if it's not up to date
20:32 forrest This package hasn't been updated since july 3rd 2011
20:32 forrest so maybe you're really out of date, or someone pulled in select RPMs?
20:33 sfello joined #salt
20:33 forrest but if you resolve that you'll be good to go. I was worried there for a second centos was screwed up too, was gonna start on another crusade to hunt down the jinja2 packager.
20:34 cedwards i'll tinker with my repo. thanks
20:34 forrest yea np
20:34 forrest if it still fails after you fix the repo let me know
20:37 SunSparc joined #salt
20:48 cedwards forrest: realized i didn't have the second DVD imported into my local mirror
20:49 cedwards forrest: i see the jinja package now
20:49 forrest ahh that would make sense :D
20:50 redondos joined #salt
20:50 kiorky terminalmage: https://github.com/saltstack/salt/pull/7224
20:51 kiorky terminalmage: do i need to make this comment a new related to this issue ?
20:51 cedwards it's working now
20:52 kiorky terminalmage: uhm, i think i just polluted something closed, will try to find somewhere more appropriate
20:52 forrest cedwards, awesome
20:52 robertkeizer joined #salt
20:54 kiorky terminalmage: https://github.com/saltstack/salt/issues/7247
21:01 mmilano joined #salt
21:05 ben_ joined #salt
21:07 redondos joined #salt
21:11 sibsibsib joined #salt
21:13 Jahkeup joined #salt
21:20 Gifflen joined #salt
21:20 logix812 joined #salt
21:27 redondos joined #salt
21:46 redbeard2 joined #salt
21:50 rspectre joined #salt
21:50 SunSparc_ joined #salt
21:50 rspectre Just upgraded to 0.16.4 - one of my pillars will not render now due to an asterisk?
21:50 rspectre Quoting the string doesn't solve the issue?
21:54 redondos joined #salt
21:55 SunSparc joined #salt
21:57 forrest can you put the pillar up on pastebin or gist rspectre?
21:59 rspectre @forrest: https://gist.github.com/RobSpectre/6566060
21:59 forrest and you wrapped it in single quotes and got the same error?
22:00 rspectre @forrest: Affirm.  Also this pillar compiled without issue on Salt 0.16.0.
22:00 forrest did you go straight from 0.16.0 -> 0.16.4?
22:00 forrest and I assume you restarted the master right?
22:00 rspectre @forrest: affirm.
22:00 aat joined #salt
22:00 forrest can you paste in the error it spits back when you wrap it in single quotes?
22:00 rspectre @forrest: Restart - unsure if apt does this automatically or not.  Let me give that a try.
22:01 forrest apt won't automatically restart it
22:02 rspectre @forrest: Check.  Restarted - same error.  Edited the gist for the result with single quotes: https://gist.github.com/RobSpectre/6566060
22:02 rspectre @forrest: Produces same error.
22:03 forrest Yea I was curious to see whether it was caching the data or something
22:03 forrest hmm, but if you leave the asterisk off there, no issue?
22:03 rspectre @forrest: Giving that a try...
22:04 rspectre @forrest: Hrm.  Issue persists - agree there must be a cache in play.  Do you know how I can clear it?
22:04 forrest /var/salt/cache
22:04 forrest does that exist?
22:05 forrest if so remove it and restart the service
22:05 forrest err sorry /var/cache/salt :P
22:06 rspectre @forrest: Cache dir removed.  Service restarted.  String in single quotes.  Same error.
22:06 forrest ok, what does your sls look like?
22:07 forrest maybe mysql was changed between 0.16.0 and 0.16.4
22:07 rspectre sls for that pillar is in the gist.
22:07 rspectre Just a dict within a dict.
22:09 rspectre hrm.  removing cache, removing asterisk, restarting salt-master, problem persists.
22:10 forrest yea I'm trying to look through the docs now, I just don't see where root_password is at in here
22:10 forrest haven't messed with the mysql module much
22:10 mmilano joined #salt
22:11 rspectre @forrest: root_password is just my key.  It gets rendered in the mysql.sls salt state for the configuration of a root mysql_user.present state.
22:12 forrest ahh ok, so you're not storing the creds inside of pillar.
22:12 rspectre @forrest: I am storing the password hash in the pillar, then rendering them in the mysql salt state.
22:13 rspectre @forrest: Interestingly, I just changed the key from "root_password" to "password_hash" and am getting the same error...
22:13 rspectre @forrest: Which suggests to me there is definitely some other cache here.
22:14 forrest after chaing to password hash, trash /var/cache/salt and try restarting the service again
22:14 forrest see what happens then
22:15 rspectre @forrest: Grah.  Same error. :/
22:15 forrest even saying root_password is wrong still?
22:15 rspectre @forrest: Affirm. :/  Very frustrating.  That key no longer exists in that file.
22:15 forrest weird
22:15 forrest can you go onto the minion you're trying to apply this to
22:16 forrest trash the cache, and restart the salt-minion service there?
22:18 Thiggy joined #salt
22:18 rspectre @forrest: kk - looks like that did it.  Seems like the pillar caching behavior changed.  Bet it's not invalidating anymore.
22:18 forrest cool
22:19 rspectre @forrest: Thanks for the help - frustrating hitch.
22:20 forrest yea no problem man, it's pretty common that you need to at the very least restart both the minion and master after upgrades, sometimes clearing the cache too. Did you change it back to root_password, or leave it as password_ash?
22:20 forrest *hash
22:20 redondos joined #salt
22:20 rspectre @forrest: Changed it back.
22:20 forrest awesome
22:22 superflit left #salt
22:22 rspectre @forrest: and now my mysql states are working as intended. w00t.
22:23 SunSparc_ joined #salt
22:26 TheCodeAssassin joined #salt
22:27 jalbretsen joined #salt
22:37 redondos joined #salt
22:47 forrest nice rspectre
22:51 fragamus joined #salt
22:52 emocakes joined #salt
22:52 fragamus joined #salt
22:53 mmarch left #salt
22:59 Tekni joined #salt
23:02 bhosmer joined #salt
23:16 SunSparc joined #salt
23:21 SunSparc joined #salt
23:26 SunSparc_ joined #salt
23:26 Thiggy joined #salt
23:27 cowmix joined #salt
23:27 aat joined #salt
23:35 rspectre joined #salt
23:45 mech422 Does anyone know how you make a file.managed ONLY update when a 'parent' file.managed actually changes ?
23:45 mech422 (specifically, I'm creating a 'flag' file (.this_step_done), and I only want the 2nd file.managed to fire off the first time the state is executed (when I create .this_step_done) )
23:50 cowmix mech422: just to a required, right?
23:51 mech422 cowmix: sorry?
23:52 mech422 I've tried: file.managed - watch: file: .this_step_done ...
23:52 robertkeizer joined #salt
23:53 Jahkeup joined #salt
23:53 mech422 but it still runs the file.managed templated even if .this_step_done hasn't changed
23:58 redondos joined #salt

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