Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2014-02-12

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

All times shown according to UTC.

Time Nick Message
00:00 Ryan_Lane1 faef: add an issue in github for pillars to use gitfs
00:00 faef Ryan_Lane1: seems like they have a philosophical argument against it? https://github.com/saltstack/salt/issues/3745
00:00 vbabiy_ joined #salt
00:01 dpac|away joined #salt
00:01 Ryan_Lane1 faef: oh: https://github.com/saltstack/salt/issues/5732
00:02 Ryan_Lane1 there's an external pillar for it
00:02 Ryan_Lane1 why not just use the external pillar?
00:02 fllr joined #salt
00:02 faef Ryan_Lane1: because that doesnt autodiscover new pillar envs
00:03 Ryan_Lane1 :D
00:03 faef I'd like to have 2 repos, one for file_roots and one for pillar_roots, where people can create branches and salt creates the corresponding envs
00:03 faef gitfs works for file_roots, but pillar_roots is still manual
00:03 * Ryan_Lane1 nods
00:03 schimmy hmm- looks like there is one: https://github.com/saltstack/salt/issues/8959
00:03 schimmy I'll put that to fixed, pending verification
00:05 faef guess i'll make an issue
00:08 faef https://github.com/saltstack/salt/issues/10373
00:09 andrej I"m still trying to wrap my head around how to maintain users on our servers (don't ask me why, but we don't have/want a directory - not my call). What I'd like to be able to do is have a group ops who are meant to be on all machines, plus the ability to add some individuals to certain machines
00:10 andrej I found http://dev.mlsdigital.net/posts/SaltStackBeyondJinjaStates/ and thought it was cool, but can't apply that to my requirements ...
00:13 andrej How would I do the mapping of individuals to certain servers?
00:13 vbabiy__ joined #salt
00:15 ClausA joined #salt
00:19 juniorzoid andrej:/clear
00:24 m87carlson andrej, I typically use a pillar called allowed_users, based on that pillar, a state will create those users (if they do not exist), and ensure their home directory and ssh keys exist. I only do this for a few users so its maintainable (otherwise we use ldap or AD)
00:25 m87carlson then I have messy if/elif condition blocks in the allowed_users pillar file if a server needs different users
00:28 joehoyle joined #salt
00:29 faef andrej: salt-api
00:29 faef oh nvm, I misread
00:33 kermit joined #salt
00:47 hunter_ reading back I see master_tops which I think I understand - but that led me to https://github.com/saltstack/salt/commit/6032b2a2db6381513e3db7622e65367cec85b9db
00:47 hunter_ What is master_roots?
00:49 hunter Quick search on docs.saltstack didn't yield anything
00:51 Gifflen joined #salt
00:51 andrej juniorzoid: what did you mean by /clear? :)
00:52 kalloc joined #salt
00:52 juniorzoid andrej: I was about to type something then typed the /clear command by accident. User error on my part.
00:53 andrej m87carlson ... do you need to have two classes of people? and how do you handle deletions?
00:53 andrej Oh juniorzoid - ok :)
00:53 hunter I've confused myself - I have no idea why this minion is getting pillar data from the dev environment.
00:54 hunter ALL the minions are getting the dev pillar data.
00:54 m87carlson That is a good question. I haven't had to delete a user, our list is very small, basically the core it staff and non of us have left yet :)
00:55 hunter <--- clearly does not know what he is doing.
00:55 m87carlson I have two different users types, local only, and LDAP backed
00:56 m87carlson only the local accounts are completely created by salt, ldap accounts just require the correct PAM stack and then directory create + ssh keys
00:57 andrej Uh-huh.  We're a small team, too, but one of us left - which raised the question on how to maintain this
00:57 robawt andrej: the manual!
00:57 frasergr_ joined #salt
00:57 robawt :D
00:58 andrej And LDAP (it was the first thing I suggested when I started here ;)) is unfortunately not welcomed.
00:58 * andrej ignores robawt
00:58 robawt sorry andrej
01:00 hunter joined #salt
01:03 m87carlson I understand the position you are probably in, sometimes its not worth fighting a political battle. You can manage users and groups with salt, but if someone leaves, you'll have to have an absent state, or disable their account (set shell to /sbin/nologin and remove keys?)
01:05 m87carlson odd someone would be against a directory service, they must have been bitten by an ldap server not being available during some emergency :)
01:06 jnials Or they just don't like the complexity.  I've heard that argument many times.
01:07 m87carlson yeah, that too. Or its part of some security plan that does not allow centralized systems
01:07 danielbachhuber joined #salt
01:07 m87carlson except salt would then be the centralized syste,
01:07 jnials Yeah, well, security plans that don't allow centralized systems are not security plans, they are hacker playgrounds.
01:09 zzzirk joined #salt
01:09 m87carlson I dont disagree. When that is the case, it usually stems from a lack of understanding of the technology
01:10 zzzirk_ joined #salt
01:11 twinshadow joined #salt
01:12 jnials A multitude of reasons in my experience, from that to empire building to plain old NIMBY.\
01:12 jnials But I'm an old bitter grey haired sysadmin :)
01:16 andrej Heh
01:16 Eugene Just For Men: Touch of Xeon
01:16 bemehow joined #salt
01:17 andrej As I said ... I won't lose sleep over it, and just because I think a a few replicationg LDAP VMs would be affordable doesn't mean that I can't see some valid reasons for maintaing state of users via salt
01:18 andrej Now I am in the unfortunate position of having to make it happen ;}
01:18 andrej I've dabbled w/ salt for a few months now, and I dabbled in programming for many years ... getting these two together is what I'm struggling with
01:21 xzarth_ joined #salt
01:22 jnials go look at http://reclass.pantsfullofunix.net/index.html and see if it will help.  I've used it to replace my master_tops file and so far it is working well.
01:22 jnials It should allow you to do composition of user lists.
01:23 jnials I'm still learning salt, so it took me a little while to get my head around and I'm still not convinced I'm doing it right though.
01:33 thayne joined #salt
01:37 favadi joined #salt
01:46 joehoyle joined #salt
01:52 kalloc joined #salt
02:12 fllr joined #salt
02:15 TJ1980 joined #salt
02:18 faldridge joined #salt
02:31 mgw1 joined #salt
02:34 ravibhure joined #salt
02:37 ravibhure1 joined #salt
02:40 xl1 joined #salt
02:42 brianhicks joined #salt
02:45 joehoyle_ joined #salt
02:50 ajw0100 joined #salt
02:52 kalloc joined #salt
02:57 vbabiy_ joined #salt
03:00 pass_by_value joined #salt
03:00 seanz joined #salt
03:09 seanz joined #salt
03:11 jacksontj_ joined #salt
03:11 jacksontj_ anyone from saltstack online?
03:11 jacksontj_ weirdish zmq question
03:11 jacksontj_ ;)
03:11 mbarrien joined #salt
03:19 dave_den jacksontj: might as well ask the question just in case a lurker knows the answer :)
03:19 jacksontj_ will do then ;)
03:19 * dave_den goes back to lurking
03:19 jacksontj_ basically i have a large salt cluster. I just started sending fairly large amounts of data (10k-- so not all that big) in the request
03:19 jacksontj_ after doing that i noticed i was pegging the master's outbound network interface
03:20 jacksontj_ it seems (after some digging) that we aren't using the filtering capabilities of zmq
03:20 jacksontj_ so i was wondering if anyone was working on that-- if not i'm messing with it now
03:20 dave_den right now the salt master snds commands via PUBSUB to all minions
03:20 jacksontj_ zmq support topics and filters in pub/sub
03:20 dave_den it's up to the minion to decide whether the command applies to them
03:20 jacksontj_ which are filtered publisher side
03:21 dave_den right, but salt sends commands to all minions, at least last time i checked.
03:21 jacksontj_ yea, and it is currently ;)
03:21 jacksontj_ what i'm wondering is why we don't add some sort of topic filtering support
03:21 jacksontj_ super easy-- although we'd only be able to support lists or something
03:22 jacksontj_ just seems like a bad idea sending all jobs to all minions as we get big
03:22 rostam joined #salt
03:22 dave_den i agree. personally i think the master should make the decision about which minions receive the commands. i can't recall exactly, but it is a deliberate decisions to behave this way
03:23 jacksontj_ i believe it was intended like this for compound matchers etc which also used to execute minion side-- now with the mine that targeting is done on the master and converted to a list
03:23 dave_den in certain situations (like grains values), it may still make sense to send to *minion, but when you target from data that the master knows without a doubt, i would love selective targeting in 0mq
03:24 jacksontj_ yea, i was thinking of adding the support for list targeting only
03:24 jacksontj_ at least for now
03:24 jacksontj_ then we can do some magic on the master to convert the other types into lists if we can, if not we'll still send out to all
03:25 dave_den check out utils.minion.ckminion for targeting
03:26 jacksontj_ yea, thats the stuff i saw before :)
03:26 dave_den how many minions do you have?
03:26 jacksontj_ ~10k in this cluster
03:26 dave_den how many syndics?
03:27 jacksontj_ 0
03:27 dave_den nice
03:27 jacksontj_ :)
03:27 dave_den well, i for one would prefer precision targeting. i'd suggest hitting the mailing list or filing a github issue for it before you start any work
03:28 jacksontj_ i think i'm 50% done actually while chatting here ;)
03:28 jacksontj_ although, as im messing with it im pretty sure this will not be backwards compatible :(
03:28 jacksontj_ but someone smarter than me can come up with a better way
03:29 manfred terminalmage: ping.
03:30 techdragon joined #salt
03:32 krow joined #salt
03:32 joehoyle joined #salt
03:49 krow joined #salt
03:51 harkx joined #salt
03:52 kalloc joined #salt
03:53 frasergraham joined #salt
03:53 terminalmage sup?
03:54 manfred where in the code can I tell that the master is completely started?
03:56 manfred I want to put in a systemd notify to send a message to systemd that the service has completed starting, so that the systemd unit can be type=notify instead of simple
03:56 jacksontj_ dave_den: https://github.com/jacksontj/salt/tree/pubfilter if you are interested :)
03:56 manfred because the current one lists it as started when it is still starting.
03:57 manfred and I am looking through this, and I don't see a good place
03:57 jacksontj_ crud my local git checkout has some.. issues ;)
03:57 manfred terminalmage: techhat said you might know
03:58 KyleG joined #salt
03:58 KyleG joined #salt
03:58 joehoyle joined #salt
04:07 mbarrien joined #salt
04:07 krow joined #salt
04:12 middleman_ joined #salt
04:17 mgw1 joined #salt
04:19 pass_by_value joined #salt
04:25 Corey jacksontj_: What did you break? :-)
04:25 jacksontj_ nooothing ;)
04:25 jacksontj_ lol
04:25 Corey jacksontj_: just yolo it.
04:25 jacksontj_ Corey: https://github.com/saltstack/salt/pull/10374
04:26 jacksontj_ today i tried sending a fairly large salt job (~2k of data i believe) and it sent to all minions
04:26 Corey yolo: aliased to git commit -am "DEAL WITH IT" && git push -f origin master
04:26 jacksontj_ not the 1 i intended ;)
04:26 jacksontj_ Corey: rofl
04:26 jacksontj_ *typing up alias*
04:26 msciciel joined #salt
04:27 Corey jacksontj_: alias such=git; alias very=git; alias wow='git status'
04:28 Corey wow; such git; very push
04:28 manfred +1
04:29 jacksontj_ i think i can get around my need for large arguments by doing some grains passing magic to a custom ext_pillar
04:29 jacksontj_ but, figure thats a useful feature anyways
04:29 jacksontj_ since large clusters == lots of commands flying around
04:30 hunter joined #salt
04:30 jacksontj_ but it does seem that i end up breaking a lot ;)
04:30 jacksontj_ lol
04:31 jacksontj_ the channels thing of mine apparently broke everything
04:31 jacksontj_ but i can't get it to break ANYTHING on my box :(
04:31 TJ1980 joined #salt
04:32 jacksontj_ i'm hoping to have another feature pull in the next week or so, but its just a code refactor and those *never* break anything ;)
04:32 terminalmage manfred: sorry! had to step away
04:32 terminalmage reading scrollback now
04:33 manfred np :)
04:33 vbabiy_ joined #salt
04:33 terminalmage manfred: so, I know that the *minion* fires an event when it starts
04:33 terminalmage but I am not sure about the master
04:34 terminalmage if not, this should be relatively simple to add
04:34 terminalmage the events are published to a socket
04:34 terminalmage which can be listened to
04:34 manfred hrm... ok
04:34 terminalmage how does this sort of thing normally work with systemd
04:35 terminalmage ?
04:35 terminalmage I'm not as familiar with it as I should be
04:35 manfred So, the systemd.daemon.notify('READY=1') Has to be fired off from the main pid
04:35 manfred so
04:35 manfred you have a process that doesn't start instantly, so you don't want to just use Type=simple.  BUt there is a systemd python library that comes with the systemd package that has a python wrapper for sd_notify
04:36 terminalmage ok
04:36 jacksontj_ terminalmage: doesn't look like the master fires an event on startup (since im in there anyways)
04:36 terminalmage jacksontj_: ok
04:36 manfred when the service is done startnig up, the main pid just needs to fire systemd.daemon.notify('READY=1')
04:36 terminalmage manfred: we have a function that checks to see if the system was booted with systemd
04:37 terminalmage so it'd be easy to add this
04:37 terminalmage I'll file an issue
04:37 terminalmage to remind me
04:37 manfred http://www.freedesktop.org/software/systemd/python-systemd/index.html
04:37 manfred I was trying to add it, but I couldn't figure out where to add it.
04:38 terminalmage manfred: that python module would need to be added to the PKGBUILD as well
04:39 manfred We would need to make a python2-systemd package, since the systemd in arch comes with the python3 version
04:39 terminalmage or, this could just be optional, it's easy enough to try/except an import
04:39 manfred that is what I was going to do
04:39 terminalmage ok
04:39 manfred fedora already has a systemd-python package that i the python2 version
04:39 terminalmage cedwards manages the PKGBUILD, so you can coordinate with him to get it updated once python2-systemd is packaged
04:39 manfred it is in community now
04:40 terminalmage it is?
04:40 manfred https://www.archlinux.org/packages/community/any/salt/
04:40 manfred seblu put it there this afternoon
04:40 terminalmage ha!
04:40 TJ1980 joined #salt
04:40 terminalmage I heard at SaltConf that a Racker was going to package it
04:40 terminalmage cool stuff
04:41 manfred well... I was going to
04:41 yomilk joined #salt
04:41 manfred and i overwrote seblus work on it...
04:41 manfred and there was a little drama on arch-dev-public
04:41 terminalmage ha
04:47 jacksontj_ jenkins just came back on my pullreq https://github.com/saltstack/salt/pull/10374
04:47 kalloc joined #salt
04:47 jacksontj_ i can't really seem to interpret the output :/
04:47 Fin1te joined #salt
04:50 AdamSewell joined #salt
04:51 anuvrat joined #salt
04:53 mgw1 joined #salt
05:01 yomilk joined #salt
05:02 Furao joined #salt
05:03 Mua joined #salt
05:04 Mua joined #salt
05:05 alunduil joined #salt
05:06 TheSojourner joined #salt
05:06 TheSojourner joined #salt
05:09 Mua joined #salt
05:10 terminalmage manfred: https://github.com/saltstack/salt/pull/10375
05:10 manfred nice
05:11 vbabiy_ joined #salt
05:11 manfred just so you know
05:11 manfred systemd.daemon.booted() will tell you if it is booted with systemd
05:12 terminalmage ok
05:12 terminalmage I can add a check for that as well
05:12 terminalmage if the module is present
05:13 manfred and instead of checking for /run/systemd/system, i would just check if /proc/1/comm is systemd or init
05:14 xDamox joined #salt
05:14 terminalmage hmm, /proc/1/comm doesn't seem to be universal
05:14 xDamox joined #salt
05:16 terminalmage manfred: also, http://dpaste.com/1607294/
05:17 terminalmage ran that on my laptop running arch w/systemd
05:17 manfred it will return an exception isn't hooked into a service tries to return it
05:17 terminalmage ahh ok
05:17 manfred put it in a file and put it in ExecStart
05:17 terminalmage duh... makes sese
05:17 terminalmage sense
05:17 manfred sleep 10, and then notify and it will work :)
05:17 manfred yeah
05:18 manfred and it does this if you accidentally return it with one of the workers
05:18 manfred Feb 12 03:37:13 salt.gtmanfred.com systemd[1]: salt-master.service: Got notification message from PID 5316, but reception only permitted for PID 5294
05:18 terminalmage yay cgroups!
05:18 terminalmage so much easier than initscript hacks
05:19 manfred yar
05:19 manfred http://cgit.freedesktop.org/systemd/systemd/tree/src/python-systemd/_daemon.c#n49
05:20 manfred http://cgit.freedesktop.org/systemd/systemd/tree/src/libsystemd-daemon/sd-daemon.c#n506
05:20 kermit joined #salt
05:20 manfred that is what it is using.
05:20 manfred so you were right :0
05:21 manfred :D
05:21 terminalmage :)
05:21 terminalmage I'm going to refactor a bit
05:22 manfred kk
05:22 terminalmage I don't need to touch all that other code if I can just use systemd.daemon.booted()
05:22 manfred yar, cause you can't really use notify if you don't also have booted :P
05:24 terminalmage manfred: what do you think: https://github.com/saltstack/salt/pull/10375/files
05:24 terminalmage much simpler that way
05:25 manfred well...
05:25 manfred except ImportError:
05:25 manfred HAS_PYTHON_SYSTEMD = True
05:25 thayne joined #salt
05:25 manfred something seems wrong with that :P
05:25 terminalmage whoops
05:25 terminalmage d'oh
05:26 terminalmage fixed
05:26 terminalmage that's what you get when you yank and paste
05:27 manfred heh
05:28 sroegner joined #salt
05:28 manfred terminalmage: hrm... it isn't working for me...
05:28 manfred it binds to the socket, but doesn't appear to finish startnig
05:29 terminalmage hmmm
05:29 manfred one secnod
05:30 terminalmage ok, in the meantime I updated the pull request comment
05:31 manfred kk
05:31 terminalmage so it isn't merged until we make sure it works
05:31 manfred oh
05:32 manfred it works, i am just a dummy :) I copied where you put your code, but didn't change it to match the variable I was using
05:32 terminalmage oh, it does?
05:32 terminalmage cool
05:32 terminalmage can you comment?
05:32 manfred yup, works
05:32 manfred yar
05:33 drags joined #salt
05:33 favadi joined #salt
05:33 terminalmage I updated the commit msg again
05:33 terminalmage :)
05:34 donatello joined #salt
05:37 manfred it matches up at the same time as if we used
05:37 manfred type=forking
05:37 manfred and PIDFile
05:37 manfred and salt-master -d
05:37 manfred so it is teh equivalent, but you get to let systemd do all the forking into the background and tracking :)
05:37 terminalmage yeah, I need to learn more about systemd
05:38 terminalmage right now I know enough to be dangerous
05:38 terminalmage I did teach myself enough to get a tty working within lxc under systemd
05:38 manfred yar
05:38 terminalmage so I consider that to be a triumph
05:38 manfred indeed
05:39 terminalmage had to make my own getty@tty1.service, and point it at a getty@.service with this fix applied: https://wiki.archlinux.org/index.php/Linux_Containers#lxc-console_does_not_provide_a_login_prompt
05:39 manfred I spent a lot of time digging into systemd back before I had a job :) well, when i was still working at the georgia tech math department :)
05:39 terminalmage which reminds me, I need to update the wiki since that was not mentioned
05:40 manfred nice
05:40 terminalmage I know there are tons of people a SATX, but have you ever bumped into Marty Buchaus?
05:40 terminalmage I used to work with him in Chicago
05:41 manfred I have not
05:41 __number5__ wondering Ubuntu's reaction to Debian switching to systemd
05:41 manfred __number5__: they are being pissy, 3 of the 8 tech council people were ubuntu devs
05:41 terminalmage yeah... only one racker with whom I've spoken is familiar with the name, even :)
05:41 terminalmage lots of people there
05:41 manfred and ian lost his mind
05:41 manfred terminalmage: indeed :)
05:42 manfred __number5__: https://lists.debian.org/debian-ctte/2014/02/msg00359.html
05:42 manfred right after it was passed
05:42 manfred he went freaking insane https://lists.debian.org/debian-ctte/2014/02/msg00344.html
05:42 manfred https://lists.debian.org/debian-ctte/2014/02/threads.html
05:43 terminalmage oh wow...
05:43 manfred he even wrote a test email to the list, to test his filter
05:43 __number5__ hmm, doesn't look good
05:43 manfred https://lists.debian.org/debian-ctte/2014/02/msg00360.html
05:43 manfred https://lists.debian.org/debian-ctte/2014/02/msg00362.html
05:43 manfred he is a lunatic, and an canonical employee
05:44 xDamox Hello, I just had a quick question. Salt can be used for configuration management, how often does it enforce configuration files? Say I changed a managed file locally on a system, how long would it take for that file to be restored to the original managed file?
05:44 __number5__ I might consider abandon ubuntu/debian if this goes messy
05:44 manfred __number5__: if it went to general resolution, it was going to be systemd anyway
05:44 terminalmage __number5__: I abandoned it back in 2006 :x
05:45 manfred most of the packages in debian already support systemd unit files
05:45 terminalmage er, actually 2007
05:45 __number5__ terminalmage: which one is your favorite now? ;)
05:46 terminalmage __number5__: Arch Linux, of course. It is the best, of course: https://wiki.archlinux.org/index.php/Arch_is_the_best
05:46 manfred I started using linux in January of 2011... i did ubuntu for about 2 hours, had to have help installing it.  got mad at it, read the archwiki beginners guide and installed it on my own, and then installed dwm and mplayer to watch tvshows
05:46 terminalmage that was one too many "of course"s
05:46 manfred lol
05:46 terminalmage dwm... nice
05:46 manfred terminalmage is a horse of course of course
05:47 terminalmage I've tried it, but I prefer awesome
05:47 terminalmage manfred: har har
05:47 __number5__ I abandoned Arch Linux few years ago because the dev ask me to go away when I asked if they can support both python2.6 and python2.7 in same system
05:47 manfred heh
05:47 vbabiy_ joined #salt
05:48 manfred i went round and round with tiling window managers, my first was wmii, but i wanted dwm because i like the tags, where i just tag a window with a display number, and i can display multiple tags at the sametime
05:48 terminalmage hahaha
05:48 manfred I also tried i3 for a while, then xmonad for about 6 months, and then have been back on dwm for the last year
05:49 terminalmage __number5__: FreeBSD and Gentoo are more your speed then, it seems
05:49 manfred or rhel/centos6
05:49 manfred ius repos ftw
05:49 __number5__ yep, if not ubuntu might go to Funtoo/gentoo
05:49 manfred i like gentoo.
05:50 manfred If I ever have to reinstall my work laptop, i might put gentoo on it purely for all the patches I have to do for arch packages to get dumb old stuff at work to work
05:50 terminalmage I never got a working gentoo system, after about 2 days of fumbling
05:50 manfred first thing i do is strip out openrc in favor of systemd
05:50 terminalmage then I started with Arch and have been with it since
05:51 __number5__ I love openrc ;)
05:51 terminalmage for about 4 1/2 years now
05:51 terminalmage __number5__: SystemRescueCd is gentoo-based and (at least used to) use OpenRC
05:51 manfred __number5__: it is fine, but when they start saying it is even remotely on par with systemd, they are just wrong
05:51 terminalmage my fav rescue distro
05:51 manfred my biggest contribution to systemd is the zshell completion, and researching for that is probably how i learned the most about it
05:52 manfred i should go back through it and see if any flags are missing
05:52 __number5__ funtoo is all openrc and yep I always keep a copy of systemrescuecd somewhere
05:53 manfred i think the best part of this dumb chart is the timer-based activation, as if systemd can't do cron or atd... and saying that .timer units are proprietary https://wiki.gentoo.org/wiki/Talk:Comparison_of_init_systems
05:53 manfred oh, and the idea that systemd is un documented
05:54 manfred every possible command has a man page, every configuration file too, and half the source is comments
05:54 manfred anyway :)
05:54 manfred I am off to bed!
05:54 manfred nite
05:55 terminalmage night
05:55 __number5__ night
05:56 __number5__ "The tabular comparison appears to be biased negatively against sysvinit and upstart. It is copied from the systemd website." <- lol
06:03 ravibhure joined #salt
06:08 vbabiy_ joined #salt
06:11 ravibhure1 joined #salt
06:14 salt_stac joined #salt
06:15 salt_stac if want to install a specific version of mongodb or anyother pkg.#salt#salt how do i do this?
06:15 frasergraham joined #salt
06:17 frasergr_ joined #salt
06:18 frasergraham joined #salt
06:19 jalbretsen joined #salt
06:20 frasergraham joined #salt
06:22 frasergr_ joined #salt
06:24 frasergraham joined #salt
06:24 tzero joined #salt
06:26 frasergraham joined #salt
06:28 frasergr_ joined #salt
06:29 frasergraham joined #salt
06:30 jY what do i need to do to move my grain dir out of /srv/salt/_grains
06:30 Psyke joined #salt
06:30 jY i have a file_root set to /etc/salt/srv/salt and a /etc/salt/srv/salt/_grains
06:30 jY but it only seems to work in /srv/salt/_grains
06:30 frasergraham joined #salt
06:32 frasergraham joined #salt
06:33 kalloc joined #salt
06:34 frasergraham joined #salt
06:35 kalloc joined #salt
06:35 frasergraham joined #salt
06:36 frasergraham joined #salt
06:37 yano joined #salt
06:39 frasergraham joined #salt
06:40 frasergraham joined #salt
06:41 frasergr_ joined #salt
06:43 frasergraham joined #salt
06:44 frasergraham joined #salt
06:45 dirjax joined #salt
06:46 frasergraham joined #salt
06:47 frasergr_ joined #salt
06:48 frasergraham joined #salt
06:48 mbarrien joined #salt
06:48 n8n joined #salt
06:49 frasergraham joined #salt
06:49 chru joined #salt
06:50 frasergraham joined #salt
06:51 frasergraham joined #salt
06:53 frasergr_ joined #salt
06:53 L0j1k joined #salt
06:54 jY issue was my custom grain was throwing an exeception
06:55 micko joined #salt
06:55 frasergraham joined #salt
06:56 frasergraham joined #salt
06:58 frasergraham joined #salt
06:58 mbarrien joined #salt
07:00 frasergraham joined #salt
07:01 frasergraham joined #salt
07:03 alunduil joined #salt
07:03 frasergr_ joined #salt
07:05 frasergraham joined #salt
07:05 mbarrien joined #salt
07:05 frasergraham joined #salt
07:07 frasergraham joined #salt
07:09 frasergraham joined #salt
07:10 kalloc joined #salt
07:10 frasergraham joined #salt
07:12 frasergr_ joined #salt
07:14 frasergraham joined #salt
07:15 frasergr_ joined #salt
07:16 gw joined #salt
07:16 gw joined #salt
07:17 sroegner joined #salt
07:17 frasergraham joined #salt
07:19 frasergraham joined #salt
07:20 frasergr_ joined #salt
07:21 frasergraham joined #salt
07:22 Shenril joined #salt
07:23 frasergraham joined #salt
07:23 dirjax joined #salt
07:25 frasergraham joined #salt
07:27 frasergr_ joined #salt
07:28 frasergraham joined #salt
07:29 frasergr_ joined #salt
07:31 yomilk joined #salt
07:32 frasergr_ joined #salt
07:32 alunduil joined #salt
07:47 anuvrat joined #salt
07:48 toastedpenguin joined #salt
07:49 Katafalkas joined #salt
07:49 fllr joined #salt
07:51 _fllr_ joined #salt
07:55 krow joined #salt
07:55 Furao joined #salt
08:00 balboah joined #salt
08:03 harobed_ joined #salt
08:03 sgviking joined #salt
08:03 alunduil joined #salt
08:04 superlazy joined #salt
08:07 kalloc joined #salt
08:09 kalloc joined #salt
08:10 L0j1k joined #salt
08:10 superlazy Is that a known issue that a minion is not "pingable" by using "salt 'minion_id' test.ping" if the minion is running in a vmware instance using NAT? When I set it to bridged the minion is reachable. Just for Your information: The minion is running in a vmware instance running on a host which is behind a firewall. This firewall has dedicated rules which makes it possible to open a connection to my master on port 4505 and 4506. Any i
08:12 fllr joined #salt
08:15 middleman_ joined #salt
08:15 linjan joined #salt
08:16 zooz joined #salt
08:17 frasergraham joined #salt
08:22 miguitas joined #salt
08:26 mike25ro guys ... in a jinja variable i want to get the xxx from the xxx.server.com ... is that possible .. like a split or smth?
08:26 Kenzor joined #salt
08:29 anuvrat joined #salt
08:30 Furao mike25ro: {{ domain_name.split('.')[0] }}
08:30 kadel joined #salt
08:30 mike25ro Furao: the split function is from jinja? ... cause i couldn't find it in their docs
08:31 mike25ro Furao: thanks btw :)
08:31 Furao it's from python
08:31 Furao in fact it's a method to any string
08:31 Furao >>> "xxx.example.com".split('.')[0]
08:31 Furao 'xxx'
08:32 mike25ro yeah i get it ... thanks buddy ... i didn't know i can use python methods inside jinja... i have to learn py ...
08:32 mike25ro again thanks a lot
08:32 yomilk_ joined #salt
08:39 masm joined #salt
08:42 MrTango joined #salt
08:44 giantlock joined #salt
08:44 krow joined #salt
08:47 ml_1 joined #salt
08:54 abique joined #salt
08:54 mjulian joined #salt
08:54 mjulian joined #salt
08:54 d10n joined #salt
08:57 yomilk joined #salt
09:00 pwistrand joined #salt
09:00 hvn joined #salt
09:02 frasergraham joined #salt
09:02 slav0nic joined #salt
09:02 slav0nic joined #salt
09:03 frasergraham joined #salt
09:04 frasergraham joined #salt
09:05 sroegner joined #salt
09:06 frasergraham joined #salt
09:07 yomilk_ joined #salt
09:08 frasergraham joined #salt
09:08 babilen Can you import modules in jinja? (e.g. urlparse would be just what mike25ro needs)
09:09 mike25ro babilen: the answer from Furao is great ... just what i needed...
09:09 Furao babilen: no
09:09 frasergraham joined #salt
09:09 mike25ro i just needed to target in the pillar top.sls :   - minions.{{ grains['id'].split('.')[0] }}
09:09 babilen Well, wouldn't be an option anyway
09:10 NV joined #salt
09:11 mike25ro in my case i have all my minions .. in a folder using a name like xxx.sls where xxx is taken from grains[id]  xxx.site.com
09:11 frasergraham joined #salt
09:11 mike25ro i am just curious.... grains['nodename'] ... is based on what sys info?
09:11 mike25ro on the hostname?
09:12 frasergraham joined #salt
09:12 Furao one state folder per minion?
09:12 Furao or pillars?
09:12 mike25ro i have a folder for all minions
09:12 mike25ro pillars
09:12 mike25ro minions.{{ grains['id'].split('.')[0] }}
09:13 frasergraham joined #salt
09:13 Furao I was afraid you had minion-specific states :P
09:14 mike25ro Furao: when a ... minion is discovered... i have a web dashboard.. where i click .. add minion .. and it adds a pillar file for each minion
09:14 mike25ro the states are global states...
09:15 mike25ro i have some values in the pillar files like environmet: office ... or location: Frankfurt
09:15 mike25ro and i target based on that
09:15 frasergraham joined #salt
09:16 mike25ro Q
09:16 mike25ro how do you guys manage... for example content of /etc/fstab?
09:17 Poooogles joined #salt
09:17 frasergraham joined #salt
09:17 mike25ro taking in consideration that you can not manage the entire file... just append data maybe? I use append data... but .. i saw in the documentation that in an upcoming release ther ewill be a feature like block ... so you can add a block of info into a file... and only that is managed by salt.. the rest of the file can differ from minio to minion
09:17 Poooogles joined #salt
09:17 ninkotech__ joined #salt
09:17 mike25ro that is a GREAT feature... i do that ... now using some bash and SED
09:18 frasergraham joined #salt
09:20 frasergraham joined #salt
09:22 kalloc joined #salt
09:22 frasergr_ joined #salt
09:23 frasergraham joined #salt
09:24 frasergraham joined #salt
09:26 frasergraham joined #salt
09:28 frasergr_ joined #salt
09:28 rojem joined #salt
09:29 ninkotech__ joined #salt
09:29 frasergraham joined #salt
09:31 frasergraham joined #salt
09:33 frasergraham joined #salt
09:33 debaser joined #salt
09:33 frasergraham joined #salt
09:35 frasergraham joined #salt
09:37 Katafalkas joined #salt
09:37 frasergraham joined #salt
09:38 frasergraham joined #salt
09:40 frasergraham joined #salt
09:41 frasergraham joined #salt
09:42 frasergraham joined #salt
09:42 baoboa joined #salt
09:43 frasergr_ joined #salt
09:44 frasergraham joined #salt
09:45 yomilk joined #salt
09:46 frasergraham joined #salt
09:47 sgviking joined #salt
09:48 frasergraham joined #salt
09:49 qba73 joined #salt
09:49 frasergr_ joined #salt
09:49 kalloc joined #salt
09:51 frasergraham joined #salt
09:53 frasergraham joined #salt
09:55 carlos joined #salt
09:55 frasergraham joined #salt
09:57 frasergr_ joined #salt
09:58 Aleksander joined #salt
09:58 frasergraham joined #salt
09:59 joehoyle joined #salt
09:59 frasergraham joined #salt
10:00 Aleksander I'd like Pillar to import the same public key for all users, so I'm modifying /pillar/target/common.sls - what is the proper option for importing the key? system.user.sshpub? id_rsa? id_rsa.pub? Where could I find documentation on that?
10:01 frasergr_ joined #salt
10:03 frasergraham joined #salt
10:05 frasergr_ joined #salt
10:06 frasergraham joined #salt
10:08 frasergraham joined #salt
10:08 sroegner joined #salt
10:09 frasergraham joined #salt
10:10 frasergraham joined #salt
10:12 fllr joined #salt
10:12 karolpogonowski joined #salt
10:12 frasergraham joined #salt
10:13 frasergr_ joined #salt
10:15 frasergraham joined #salt
10:16 frasergraham joined #salt
10:17 frasergraham joined #salt
10:19 frasergr_ joined #salt
10:22 frasergraham joined #salt
10:24 MrTango joined #salt
10:25 millz0r hi there fellow salters
10:29 ggoZ joined #salt
10:33 diegows joined #salt
10:43 vbabiy_ joined #salt
10:45 ndrei joined #salt
10:49 austin_laptop joined #salt
10:52 mgw1 joined #salt
10:55 BogdanR Hi. I have a config file in the that I am deploying for all machines but for some I want to deploy a different config. Apparently all the time I run the highstate the different configs are overwritten by the general config
10:56 BogdanR My top state looks like this: http://pastie.org/pastes/8725623/text?key=ecwrprz2jifrx7cj0rkada
10:56 BogdanR And collectd.lg-us-manager gets overwitten by collectd in '* and not nboffice'
10:58 vbabiy_ joined #salt
11:00 Katafalkas joined #salt
11:01 vbabiy__ joined #salt
11:05 n8n_ joined #salt
11:07 n8n__ joined #salt
11:07 ndrei joined #salt
11:07 vbabiy_ joined #salt
11:12 fllr joined #salt
11:14 fllr joined #salt
11:15 Damoun joined #salt
11:16 backjlack joined #salt
11:18 beardo_ joined #salt
11:22 frasergraham joined #salt
11:23 backjlack joined #salt
11:25 yomilk joined #salt
11:27 yomilk joined #salt
11:32 johnny_bravo joined #salt
11:32 johnny_bravo hi
11:33 johnny_bravo anyway in salt to put in a list of packages in the sls
11:33 johnny_bravo like:
11:33 johnny_bravo wget, mc, nmap:
11:33 johnny_bravo pkg:
11:33 johnny_bravo installed
11:34 delkins joined #salt
11:45 techdragon johnny_bravo you can just create multiple pkg. installed statements if you want multiple packages.
11:46 johnny_bravo ok so no list of packages
11:46 johnny_bravo separate row for each
11:47 viq johnny_bravo: well, kinda. You could do that with jinja
11:47 johnny_bravo good point
11:48 johnny_bravo i might give that a try
11:54 ml_1 joined #salt
11:54 halfss joined #salt
11:54 frasergraham joined #salt
11:56 kadel joined #salt
11:59 RizKhan joined #salt
12:01 david_a joined #salt
12:03 zzzirk joined #salt
12:05 ml_1 joined #salt
12:06 n8n joined #salt
12:09 RizKhan Hi Guys, Any one know if a setup where Master is on ZMQ3.x/CentOS6.x and minions on ZMQ2.x/CentOS5.x will work fine in terms of master/minion communication ?
12:12 fllr joined #salt
12:12 millz0r RizKhan: I'm pretty sure if the packages can be installed then it doesn't matter
12:14 cb joined #salt
12:17 apergos joined #salt
12:22 viq RizKhan: which salt versions?
12:23 elfixit joined #salt
12:24 maveas joined #salt
12:27 AdamSewell joined #salt
12:28 RizKhan Thanks millzor . Basically I had performance issues and wanted to upgrade to ZMQ 3, but current servers are all centos 5.x and had no luck upgrading zmq to v3 on them, but able to setup a new master on CentOS 6.x to ZMQ 3.x ,
12:28 RizKhan so thinking if this will work and secondly if that will still give any performance improvement with only master being on ZMQ 3 and minions still on ZMQ 2
12:29 apergos joined #salt
12:30 Nazzy *sigh* life on the cutting edge, ain't it fun
12:31 Nazzy I think we need to redefine "cutting edge" and "leading edge" to "front line deployment" ... in the military sense ¬_¬
12:32 Nazzy so s/cutting edge project/target for enemy fire/g
12:33 RizKhan @viq  the minions are currently on salt-minion 0.17.1 and salt version on current master is 0.17.4  , the new master with ZMQ 3 would have the version available by default with latest repo,
12:35 Nazzy RizKhan, I would be concerned about ZMQ 2 vs 3 comms, but salt wise that should be fine
12:40 RizKhan Thanks Nazzy , I guess I will keep the current master active , bring a new one up in parallel , update DNS for master URL and then see how things go, if they go north, I will revert back :)
12:47 krow joined #salt
12:54 ndrei joined #salt
12:55 AdamSewell is it possible to manage an entire directory with salt as you would with file.manage?
12:55 AdamSewell er file.managed
12:55 frasergraham joined #salt
12:55 Eugene AdamSewell - there's recurse
12:56 Eugene http://docs.saltstack.com/ref/states/all/salt.states.file.html#salt.states.file.recurse
12:57 babilen madduck: Thanks for your mail, much appreciated. Do you plan to backport reclass to wheezy or would I have to do that internally?
12:57 madduck babilen: planning… ;)
12:57 AdamSewell Eugene, thanks, thats what i was looking for
12:57 madduck babilen: you could backport it and I upload it ;)
12:58 babilen madduck: haha, alright. Do you expect a backport to be problematic? I've seen that all B-Ds are easily satisfied, but haven't backported it yet.
12:58 madduck should be a simple matter of rebuilding in the wheezy environment
12:59 babilen "debcheckout reclass"
12:59 madduck babilen: I can look at it, but next week…
12:59 MrTango joined #salt
13:00 babilen madduck: No worries, I don't mind backporting it. Maintaining it in bpo would, however be a bit more work. But then I have my build environment already configured for wheezy-backports, so ...
13:01 jeddi joined #salt
13:01 madduck as you wish
13:11 gadams999 joined #salt
13:12 fllr joined #salt
13:16 blee joined #salt
13:24 techdragon ok I'm trying to get a master minion pair running in a virtualenv, issue is the minion isn't generating a key pair for the master to accept.... not sure what I can do to force it.
13:27 halfss_ joined #salt
13:27 brianhicks joined #salt
13:29 pille joined #salt
13:36 vbabiy_ joined #salt
13:38 blee_ joined #salt
13:38 sroegner joined #salt
13:39 zooz joined #salt
13:39 Luijk joined #salt
13:40 thayne joined #salt
13:41 babilen madduck: yeah, that was painless
13:43 jrdx joined #salt
13:44 techdragon anyone got an idea what i can do about this pki issue
13:44 schristensen joined #salt
13:46 ipmb joined #salt
13:47 elfixit joined #salt
13:51 Luijk Any reason `salt-ssh '*' saltutil.refresh_pillar` would not update pillar? Even after the minion is terminated and a fresh node created!!
13:53 Luijk This solved the problem: `sudo rm -rf /var/cache/salt/master/` Is there a better solution?
13:55 jeremyBass2 joined #salt
13:56 frasergraham joined #salt
13:56 dwiden joined #salt
13:58 vbabiy_ joined #salt
13:59 dwiden do minions fire off events after they have completed a task?
14:04 juicer2 joined #salt
14:10 blee joined #salt
14:10 jaimed joined #salt
14:11 linjan_ joined #salt
14:11 techdragon is it worth bothering to debug this virtualenv issue or should i just use some VM's to run it for real using my own git repo?
14:11 L0j1k joined #salt
14:12 dwiden I'm trying to debug reactor events by looking at the events themselves, using eventlisten.py.  I'm manually firing events from my minions to my master, but they aren't appearing in my terminal running eventlisten.py.  Can anybody give me a hand?
14:12 techdragon I kinda want to run tests & ensure my pull requests actually work before bothering to send them to the project.
14:12 techdragon :-/
14:12 fllr joined #salt
14:14 techdragon guess ill dig into this tomorrow
14:14 techdragon night
14:16 Poooogle_ joined #salt
14:17 Poooogle_ joined #salt
14:18 TJ1980 joined #salt
14:20 racooper joined #salt
14:23 tessellare joined #salt
14:24 AdamSewell joined #salt
14:26 vbabiy_ joined #salt
14:34 faldridge joined #salt
14:37 mpanetta joined #salt
14:37 fllr joined #salt
14:38 mpanetta joined #salt
14:39 fllr joined #salt
14:53 faldridge joined #salt
14:56 frasergraham joined #salt
15:00 joehoyle joined #salt
15:01 kalloc joined #salt
15:08 frasergraham joined #salt
15:08 gw_ joined #salt
15:09 Gifflen joined #salt
15:12 GradysGhost joined #salt
15:12 brianhicks joined #salt
15:13 Poooogles joined #salt
15:17 dwiden joined #salt
15:18 Gordonz joined #salt
15:18 techhat joined #salt
15:18 Gordonz joined #salt
15:19 rojem joined #salt
15:21 nobody_314 joined #salt
15:21 ccase joined #salt
15:22 abe_music joined #salt
15:23 rostam joined #salt
15:25 Katafalkas joined #salt
15:25 abe_music has anyone attempted to include salt as a dependency in their own project's setup.py? if so, did you have issues with M2Crypto not building?
15:29 Ahlee let me guess, you're on a red hat based distro?
15:29 kermit joined #salt
15:30 Ahlee if so, you'll need to build it with the fedora_setup.py or whatever, i opted to build a bdist_egg and install with easy_install
15:30 toastedpenguin joined #salt
15:31 ndrei joined #salt
15:31 debaser has anyone tried running halite via apache/mod_wsgi?
15:33 abe_music Ahlee, yeah, on centos 6.4, but if I include M2Crypto as a requirement in my own setup.py it builds fine
15:33 abe_music no need for the fedora_install workaround
15:33 abe_music very odd
15:37 kolaman joined #salt
15:37 sgflt joined #salt
15:38 Mua joined #salt
15:39 fllr joined #salt
15:40 sroegner abe_music: aren't you doing hadoop/hbase with salt?
15:40 cnelsonsic joined #salt
15:41 Mua_ joined #salt
15:41 abe_music sroegner: yup, cdh4 to be specific
15:41 sroegner any of that code out on github?
15:42 abe_music sroegner: yeah, we have a lot of our formulas up...keep in mind they're for a platform we're currently building so may need a bit of work: https://github.com/stackdio-formulas/
15:45 Ahlee abe_music: very odd, i built on 6.2 about 6 months ago and needed to
15:46 johnny_bravo_ joined #salt
15:47 chuffpdx_ joined #salt
15:47 forrest joined #salt
15:47 debaser anyone using halite at all? :p
15:48 alunduil joined #salt
15:48 faldridg_ joined #salt
15:49 pass_by_value joined #salt
15:49 _david_a joined #salt
15:49 sroegner abe_music: are you still using accumulo?
15:50 pass_by_value Hi debaser
15:50 abe_music Ahlee: yeah, we've seen this issue and had to use that workaround...it's in our README, but for some reason it appears to work....i should probably go test to make sure things are working as expected :-p
15:50 abe_music sroegner: yeah, we support several distributed stores for our software
15:51 dirjaway_ joined #salt
15:51 hvn_ joined #salt
15:51 puppet_ joined #salt
15:51 rjc_ joined #salt
15:51 utahcon_ joined #salt
15:51 vbabiy__ joined #salt
15:51 debaser hi pass_by_value :)
15:51 pass_by_value I can help you with Halite related issues
15:51 Dinde496 joined #salt
15:52 vejdmn joined #salt
15:52 tedski- joined #salt
15:52 debaser cool, do you know how to make it run via mod_wsgi?
15:52 debaser im banging my head against the wall and docs are sparse on the subject
15:52 sroegner abe_music: accumulo-formula on saltstack-formulas should have some updates
15:54 abe_music sroegner: thanks, i'll take a look :)
15:54 pass_by_value We use bottle.py which is a wsgi framework
15:55 d10n_ joined #salt
15:55 pass_by_value Let me take a quick peek in there and see if I am able to find something interesting for you debaser
15:55 debaser thanks
15:58 lionel joined #salt
16:00 diegows joined #salt
16:01 debaser i also tried pointing a proxy balancer to halite app but that didnt work out either
16:01 pass_by_value It looks like we could definitely add Apache support to Halite
16:02 gmoro joined #salt
16:02 pass_by_value We will require some code additions to Halite
16:02 LucasCozy joined #salt
16:02 baoboa joined #salt
16:02 qba73 joined #salt
16:02 ktenney joined #salt
16:02 linjan_ joined #salt
16:02 debaser that would be awesome
16:03 blee joined #salt
16:03 pass_by_value http://bottlepy.org/docs/dev/tutorial_app.html
16:03 pass_by_value "Running Bottle on Apache with mod_wsgi"
16:03 mattikus` joined #salt
16:03 AdamSewell joined #salt
16:03 AdamSewell joined #salt
16:03 cellofellow joined #salt
16:03 rhand joined #salt
16:04 drags joined #salt
16:04 mordred joined #salt
16:04 tyler-baker joined #salt
16:04 tyler-baker joined #salt
16:04 debaser yeah, we actually have our internal salt bottle app deployed via mod_wsgi
16:04 pass_by_value So that config might be a good thing to add Halite's source. That way people can re use it.
16:04 mgw1 joined #salt
16:04 Luijk joined #salt
16:04 debaser so we'd like halite and our app to coexist nicely
16:04 ckao joined #salt
16:06 davet joined #salt
16:07 cachedout joined #salt
16:08 abe_music Ahlee: here's the results of M2Crypto build when I include it vs when I let salt do it: https://gist.github.com/abemusic/8958392
16:09 pass_by_value https://github.com/saltstack/halite/blob/master/halite/test/functional/halite_functional_tests.py#L22-#L25
16:09 pass_by_value the above code should be sufficient to get reference to app as shown in that adapter.wsgi file
16:11 Ahlee abe_music: yeah, salt's include is the error i always got when I tried to build it on centos
16:11 FarrisG joined #salt
16:12 salt_whitebelt joined #salt
16:12 ndrei joined #salt
16:12 salt_whitebelt is there a minion yum package i can use to deploy it on rhel 6 vm?
16:13 salt_whitebelt our rhel 6 vm doesn't have internet access
16:13 [diecast] joined #salt
16:13 nahamu salt_whitebelt: I think it's in EPEL.
16:13 nahamu salt_whitebelt: 32 or 64 bit?
16:14 salt_whitebelt 64
16:14 xmltok joined #salt
16:14 chrisjones joined #salt
16:14 nahamu http://dl.fedoraproject.org/pub/epel/6/x86_64/repoview/salt-minion.html
16:14 nahamu might need to get dependencies, though.
16:15 salt_whitebelt isn't the yum package comes with all the dependencies?
16:15 swa_work joined #salt
16:15 nahamu no, that's not how RPMs work.
16:16 nahamu *not how they generally work
16:16 nahamu the machine will need python, zeromq, etc.
16:16 nahamu Are you looking for an Esky build?
16:16 salt_whitebelt yes we need to deploy minon on rhel 6 vm with all dependencies
16:17 nahamu salt_whitebelt: how do you install packages onto your RHEL VMs that don't have internet access in general?
16:17 salt_whitebelt we set up local repo with all the rpms and dependencies
16:17 thayne joined #salt
16:17 blee_ joined #salt
16:18 nahamu so you're going to want to figure out what all the dependencies are and get them into your local mirror.
16:18 alekibango joined #salt
16:18 nahamu If you have a test VM that *does* have internet access you should be able to figure out what packages the other VM will need.
16:18 nahamu And you'll want to take a look at this: https://fedoraproject.org/wiki/EPEL
16:20 salt_whitebelt i know what those dependencies are, i understand
16:20 BRYANT__ joined #salt
16:20 salt_whitebelt i meant is there a way to package the dependencies with the rpm in one package?
16:20 nahamu Esky builds.
16:21 nahamu http://docs.saltstack.com/topics/tutorials/esky.html
16:21 salt_whitebelt windows msi does that i don't understand why linux doesn't a self contain package
16:21 salt_whitebelt have
16:21 nahamu I've done it on SmartOS, but in theory it should be possible on Linux too.
16:22 nahamu I don't know if anyone hosts pre-built Esky builds of Salt for RHEL 6.
16:22 abe_music Ahlee: m_george from #pynash gave me a heads up that setting the USE_SETUPTOOLS environment variable forces salt to use setuptools, which installs M2Crypto correctly
16:22 abe_music seems like something salt would need to figure out or warn the user
16:23 jalbretsen joined #salt
16:24 pass_by_value Ahlee: Hi, I remember you mentioning having issues getting mongo returner to work from the config file? Did that work out for you?
16:25 RizKhan joined #salt
16:25 ipmb joined #salt
16:27 Ahlee pass_by_value: the cli won't honor a job cache, you have to specify --return=mongo
16:27 Gifflen joined #salt
16:28 Ahlee and even then, if you run salt \* test.ping --return=mongo it'll return both to the file system based cache as well as mongo
16:28 vlad___ joined #salt
16:28 vlad_paciu joined #salt
16:29 mgw1 joined #salt
16:29 blee joined #salt
16:29 vejdmn joined #salt
16:30 bastion1704_ joined #salt
16:32 Gifflen joined #salt
16:32 delkins joined #salt
16:32 ml_11 joined #salt
16:39 pass_by_value Ah I thought you were using ext_job_cache. I was wondering what your preferred method would be to have jobs go to the job cache?
16:41 schimmy joined #salt
16:43 Ahlee I am using ext_job_cache
16:43 Ahlee per http://docs.saltstack.com/ref/configuration/master.html#ext-job-cache, it says it sets up the default returner
16:44 Ahlee I would expect the salt command line to honor that
16:44 Ahlee it also says it will disable the local job cache, however that's not the case
16:44 pass_by_value Yes, that was where I was looking as well. I remember there was an issue with mongo returner (similar to this) 2 - 3 months ago
16:44 Ahlee Interesting
16:44 vlad_paciu Hello i need help with onlyif: directive , someone can help me ?
16:45 Ahlee just mongo?  i'll fire up a test instance with redis and check
16:45 miguitas joined #salt
16:46 pass_by_value Let me take a look and see what info I can find Ahlee.
16:46 jeremyfelt joined #salt
16:47 HeadAIX joined #salt
16:55 Ahlee pass_by_value: looks like redis return goes to both by default
16:55 Rory joined #salt
16:55 Ahlee with redis as ext_job_cache, just test.ping without specifying the returner works
16:55 Ahlee still goes back to both on disk and the ext cache
16:56 Rory In my sls files, how do I restart a service if a highstate causes a file's contents to be changed (ie restarting apache after httpd.conf is updated)
16:56 Ahlee checking now if jobs.list_jobs and jobs.lookup_jid work with the ext_job_cache now
16:56 Ahlee Rory: You'll want to set a watch on the service, watching the config
16:56 pass_by_value Thanks, I will check and see what is up with the mongo returner and why it still goes to both.
16:57 Ahlee yeah, jobs.list_jobs only looks locally
16:57 Rory Ah right Ahlee I think I see the idea
16:57 Ahlee taking a peak to see what would be involved in making the job cache queryable
16:58 quickdry21 joined #salt
16:58 Rory Ahlee: http://rory.sh/init.sls is correct?
16:58 kermit joined #salt
17:00 thayne joined #salt
17:00 Ahlee Rory: looks right to me
17:01 Ahlee pass_by_value: hrm. looks like it should be honoring: https://github.com/saltstack/salt/blob/develop/salt/runners/jobs.py#L67
17:01 Ahlee digging in
17:05 carmony non-salt but system question: I have a Vagrant/VirtualBox setup, and it's eth0 interface freezes up on me when I run our test suite on it
17:05 vlad_paciu I trying to impliment the capistrano deploy logic with salt, but i have problems with onlyif:
17:05 vlad_paciu http://pastebin.com/tSGnRALG
17:05 vlad_paciu can you help me ?
17:06 pass_by_value I am setting up a VM to debug/test this as well.
17:06 carmony turns out I think it is a problem with our StatsD/Graphite code sending out thousands of UDP packets that dont make it out
17:07 carmony like the interface is having problems trying to deliver those UDP packets. Any suggestions on how to confirm this diagnosis? Right now if disable our StatsD packet sending fixes the problem, but I want to inspect the eth0 interface itself with them on to make sure I'm on the right track
17:07 carmony if that makese sense
17:08 smkelly joined #salt
17:10 smkelly joined #salt
17:11 smkelly joined #salt
17:12 jslatts joined #salt
17:13 teratoma joined #salt
17:14 blee_ joined #salt
17:14 troyready joined #salt
17:16 amckinley joined #salt
17:16 jY is there anyway to tell what service start/stop command to use in a sls file?
17:20 felixhummel joined #salt
17:24 dave_den jY: not a simple custom command, but you can change the provider http://docs.saltstack.com/ref/states/providers.html
17:25 bemehow joined #salt
17:25 dave_den jY: http://docs.saltstack.com/ref/states/providers.html#arbitrary-module-redirects
17:25 jY thanks
17:25 terminalmage joined #salt
17:26 Ahlee carmony: are the packets hitting the phyical machine's nic?
17:27 Ahlee carmony: or is it just vbox being stupid
17:27 carmony I think vbox is being stupid
17:30 Ahlee carmony: nat? tried bridge? also, what os is vbox running on? have you tried changing the emulated nic type?
17:30 carmony host is os x mavericks
17:30 carmony latest VBox
17:30 carmony Nat
17:30 carmony I use NFS for the mounts
17:31 yurei joined #salt
17:32 carmony its the intel pro/1000
17:32 carmony because I use NFS for mounts its a huge pain to try and use bridged
17:32 Ahlee i'v not heard of those specific issues, but i also try to avoid vbox for all but the most basic of items
17:32 Ahlee yeah.
17:33 Ahlee gotta love virtualized networking
17:33 Ahlee i guess roll through the differnet intel pro emulated models
17:34 vejdmn joined #salt
17:42 [diecast] joined #salt
17:45 salt_whitebelt dave_den: is the recent salt-conf slide deck avail?
17:46 xDamox joined #salt
17:46 salt_whitebelt dave_den: i m looking for integration with vmware vcac and front end web interface
17:48 rojem joined #salt
17:48 jacksontj_ joined #salt
17:53 salt_whitebelt where is the minion rpm package?
17:56 dave_den salt_whitebelt: i don't think the saltconf speaker decks are available yet
17:56 racooper salt_whitebelt,  rpms are available in EPEL, currently 0.17.4 available.
17:57 dave_den salt_whitebelt: the salt packages for RHEL are in EPEL: http://dl.fedoraproject.org/pub/epel/6/x86_64/repoview/salt-minion.html
17:57 erchn joined #salt
17:58 racooper looks like 0.17.5 is ready to be pushed out from testing now, whenever terminalmage has the opportunity :)
17:58 mgw1 joined #salt
17:59 kolaman hi all,
17:59 Mua joined #salt
18:01 lyddonb joined #salt
18:02 melinath joined #salt
18:02 kolaman I am trying to install nagios nrpe package on a minion but getting error while runing highstate, is my this file OK or any synteax problems http://pastebin.com/8MnKCnX4
18:04 lyddonb joined #salt
18:04 erchn kolaman, what kind of errors?
18:06 erchn I don't see any errors in syntax
18:08 dpn` joined #salt
18:08 dancat joined #salt
18:09 Kyle- joined #salt
18:09 Nazca joined #salt
18:10 Nazca joined #salt
18:11 forrest is the name of the nagios nrpe seriously nagios-nrpe-server?
18:11 forrest wow it really is, jeez
18:14 kolaman erchn: got it, the issu was that there was an extra "." in top.sls
18:14 erchn ok, good
18:15 jacksontj_ joined #salt
18:16 joehillen joined #salt
18:16 [diecast] joined #salt
18:17 n0arch joined #salt
18:17 joehillen joined #salt
18:17 n8n joined #salt
18:20 frasergraham joined #salt
18:22 Ryan_Lane1 joined #salt
18:23 jfzhu_us joined #salt
18:24 jfzhu_us left #salt
18:24 jfzhu_us joined #salt
18:26 terminalmage racooper: pushed it last night
18:27 frasergraham joined #salt
18:28 jfzhu_us joined #salt
18:28 terminalmage racooper: actually, I guess I didn't
18:28 terminalmage the heck
18:28 blee_ joined #salt
18:28 terminalmage I know I clicked the "Mark as stable" link
18:29 terminalmage anyway, just pushed to stable
18:29 BenCoinanke joined #salt
18:29 frasergr_ joined #salt
18:31 frasergraham joined #salt
18:32 Poooogle_ joined #salt
18:32 n8n joined #salt
18:34 psyl0n joined #salt
18:34 Ryan_Lane1 … old versions of the debian packages have disappeared from the ppa
18:34 Ryan_Lane1 it seems only 0.17.5-1precise1 is now available
18:35 krow joined #salt
18:35 Ryan_Lane1 any reason the old packages were removed?
18:36 kolaman erchn: is there any example for modify some text in a file with grains or some simple formula to do that ?
18:37 vbabiy_ joined #salt
18:37 Ryan_Lane1 it's a problem if versions of salt disappear from the ppa, since upgrading requires upgrading the master before the minions, so if a version disappears you can't install any new minions without constantly updating
18:37 * Ryan_Lane1 opens an issue
18:38 thayne joined #salt
18:38 forrest Ryan_Lane, pretty sure this has been brought up before. joehh, any input?
18:38 forrest I can't remember what th reason is though
18:38 Ryan_Lane1 it's a pretty major issue
18:38 forrest yea
18:38 Ryan_Lane1 I'm going to be forced to have a ppa mirror
18:38 Ryan_Lane1 that includes old versions
18:39 Ryan_Lane1 this will cause autoscaling groups to fail
18:39 vlad_paciu I am trying to impliment the capistrano deploy logic with salt, but i have problems with onlyif:
18:39 vlad_paciu http://pastebin.com/tSGnRALG
18:39 vlad_paciu HELP ME PLZ
18:42 mwmnj joined #salt
18:43 Ryan_Lane1 yep: https://github.com/saltstack/salt/issues/4276
18:44 ipmb joined #salt
18:44 forrest I don't understand that about ppas
18:44 forrest who designed it that way in the first place
18:45 forrest Ryan_Lane1, you upgrade your master before the minions?
18:45 Ryan_Lane1 I could have sworn that PPAs allow old version of packages
18:46 Ryan_Lane1 for sure hosting a custom debian repo would work
18:46 * kolaman still finding a way to update specific text string with salt stack in a file
18:47 danielbachhuber joined #salt
18:48 racooper kolaman,  as a state or as a remote command?
18:48 kolaman racooper: as a state
18:49 krow joined #salt
18:49 racooper http://docs.saltstack.com/ref/states/all/salt.states.file.html#module-salt.states.file has file.append, file.replace
18:50 dave_den Ryan_Lane1: i would encourage having your own repo anyway. but if you continue to use the PPA you can easily set up apt-cacher-ng as a transparent proxy.
18:51 joehoyle joined #salt
18:51 kolaman racooper: :)
18:52 Ryan_Lane1 dave_den: I will eventually have my own repo, but in the interim it would be nice if this worked right :)
18:53 Ryan_Lane1 lots of people won't have a custom repo
18:53 dave_den Ryan_Lane1: i personally use apt-cacher-ng specifically to keep all versions of everything that is ever proxied through it.
18:54 dave_den and you can also do repo rewriting/mapping which is also handy
18:54 Ahlee pass_by_value: so i'm not unable to replicate the issue with mongo return not being used by default, but file system return is not disabled.  Is it a bug, or a case of docs needing updated?
18:54 Ahlee er, now unable
18:55 forrest is anyone going to Pycon this year?
18:56 Ryan_Lane1 dave_den: yep, but then it's a point of failure and another instance that needs to be running
18:57 Ryan_Lane1 I'm going to solve this problem for myself, but I'd also like to see it solved for the community as a whole
18:58 Gordonz joined #salt
18:58 [diecast] joined #salt
18:59 pass_by_value I will have to test and find out Ahlee. I will get back to you on that.
18:59 Gordonz joined #salt
19:00 Ahlee pass_by_value: i think the issue with mongo not working with default returner was my ext_job_cache was set to use mongodb not mongo
19:00 Ahlee so, yay me there.
19:00 Ahlee Also, it does not appear that the jobs runner looks to the job cache, still weighing implications there
19:01 HeadAIX joined #salt
19:01 snuffeluffegus joined #salt
19:01 pass_by_value Ah, cool.
19:05 L0j1k joined #salt
19:05 mgw1 joined #salt
19:11 AviMarcus joined #salt
19:12 apergos joined #salt
19:13 salt_whitebelt is there a central place where i can get all of the dependencies?
19:14 salt_whitebelt minion dependencies that is
19:17 zooz joined #salt
19:17 Ryan_Lane1 well, I guess since I'm using masterless salt for now, I'll just always ensure latest :D
19:17 Ryan_Lane1 I guess that's a dangerous game to play
19:18 Ahlee what's the salt dev channel?
19:18 Ryan_Lane1 there's a salt dev channel?
19:18 Ryan_Lane1 I thought this is it
19:19 Ahlee I thought there was a channel for discussing actual development, not just this one aimed at general support
19:19 Ahlee mayhaps i'm misremembering.
19:21 Poooogles joined #salt
19:23 mbarrien joined #salt
19:24 druonysus joined #salt
19:25 druonysus joined #salt
19:30 d1rk joined #salt
19:32 hunter joined #salt
19:32 lksjdlfjoiwef joined #salt
19:33 lksjdlfjoiwef can anyone please tell me how to use file.replace or file.sed to remove a line from a file?
19:33 lksjdlfjoiwef from what i can tell i can only replace the existing text with some other text
19:33 lksjdlfjoiwef using '' creates a new line
19:37 pass_by_value I have seen some discussions taking place on #salt-devel Ahlee. But this might be a good channel also.
19:37 Ahlee ah, -devel. i tried -dev
19:38 pass_by_value Like Ryan_Lane1 said this is probably the best place though.
19:39 Ahlee yeah, no worries
19:39 KyleG joined #salt
19:39 KyleG joined #salt
19:39 Ryan_Lane1 huh. a dev channel exists. thankfully salt's main channel is pretty responsive despite that
19:40 Ryan_Lane1 for most projects the main channel becomes a ghost town when a dev channel is created
19:40 krow joined #salt
19:41 ml_1 joined #salt
19:41 faef joined #salt
19:42 faef anyone know if runners_dir works with gitfs?
19:44 dren joined #salt
19:44 pass_by_value Your keynote talk at Salt Conf was awesome btw Ryan_Lane1. I loved seeing the custom UI stuff that you implemented.
19:45 pass_by_value Sorry I did not get a chance to say hi there.
19:47 frasergraham joined #salt
19:48 Ryan_Lane1 pass_by_value: thanks :)
19:49 Ryan_Lane1 I'm sure we'll get a chance to meet in person at some point
19:49 Ryan_Lane1 still lots of work to do on trebuchet :)
19:49 ajw0100 joined #salt
19:49 Ryan_Lane1 salt gives for a lot of potential things it can do, though
19:49 Ryan_Lane1 <3
19:52 seanz left #salt
19:52 pass_by_value I am sure we will. Yes, customizability with GUIs is more or less infinite.
19:52 Ryan_Lane1 especially since I can just make calls to salt via salt-api
19:52 Ryan_Lane1 and the auth at that layer is extensible
19:53 jeremyBass1 joined #salt
19:54 pass_by_value Yeah, it was great to see how y'all used salt's functionality with your code and had that info integrated on the GUI. I am a big fan of salt-api!
19:54 faef are there any projects that are a GUI to salt-api?
19:54 pass_by_value faef there is Halite and we also have a custom enterprise GUI based off Halite by the name of Crystal
19:55 faef ohhh that's what halite is, saw references to it but never put two and two together
19:56 kasanop joined #salt
20:01 n8n joined #salt
20:03 zach I never realized how much hate I had for AWS until I Started using it full time
20:07 Ryan_Lane1 joined #salt
20:08 forrest zach, lol
20:08 forrest still struggling today?
20:11 lksjdlfjoiwef How can I remove a line from a file using file.sed or file.replace?  The only way I can figure out how to do it is using a cmd.run using sed which is not what I want to do.
20:13 zooz joined #salt
20:13 manfred file.replace will do it. you could try file.comment too probably?
20:14 lksjdlfjoiwef do i need to use a flag?  file.replace 'before' '' /tmp/file doesn't work
20:14 lksjdlfjoiwef it replaces the text with a new line
20:15 manfred That is what it should do
20:15 moped joined #salt
20:17 hunter joined #salt
20:22 faldridge joined #salt
20:22 tmmt_ joined #salt
20:24 DaveQB joined #salt
20:26 Ahlee Those are all very fragile and prone to breaking.  It's recommended to take control of the file with file.managed and change the items you need modified via templating
20:27 Ryan_Lane1 joined #salt
20:29 lksjdlfjoiwef manfred: :) I want to completely remove the line, not replace the line with a newline.
20:30 lksjdlfjoiwef I guess the best way to do it is just to cmd.run 'sed'
20:32 KyleG joined #salt
20:32 KyleG joined #salt
20:33 lksjdlfjoiwef ahlee: understood, i needed to make a quick one-off change.  no need for a state or template.
20:33 hunter joined #salt
20:33 madduck would you think google actually bothered fixing their DKIM signatures on googlegroups outgoing mail? They've been signing with invalid signatures for almost a week now.
20:34 manfred can you just comment out the line? That is what I am doing for redis.conf
20:34 lksjdlfjoiwef manfred: yes, adding a comment would also have been possible but not as clean as I wanted it to be
20:36 jtang joined #salt
20:36 manfred shouldn't I be able to put
20:36 manfred minion:
20:36 manfred startup_states: highstate
20:36 manfred in the salt-cloud mapfile and have it do what I expect... it doesn't work for some reason :(
20:38 AdamSewell joined #salt
20:40 manfred hrm... it looks like it worked this time...
20:40 manfred no idea!
20:40 hunter I'm working on pillars and am confusing myself.
20:41 hunter I have 3 environments set up (for both states and pillars) (base,dev,prod)
20:41 hunter As a test I just set an "info" variable with different values in each env and assigned the same host to _all_ of them.
20:42 hunter The minion gets the pillar data from "dev" along with a flag variable I set in all three.
20:42 hunter I know I'm making this harder/more complicated than it has to be (esp for salt).
20:43 hunter For those who _dont_ use grains to assign roles/properties, how do you layout your pillar top files?
20:45 joehoyle joined #salt
20:46 ajw0100 what is the extension_modules config option for? I see it mentioned in the external pillars docs (http://docs.saltstack.com/topics/development/external_pillars.html) but I can't find it documented anywhere
20:48 mgw1 joined #salt
20:53 hunter For pillar data, do people match host->environment in the base/top.sls? Or what?
20:53 AdamSewell Could someone look this over and tell me what I'm missing? I'm getting an error every time I run this state. http://pastie.org/8727393
20:53 forrest AdamSewell, what error?
20:53 AdamSewell forrest, http://pastie.org/8727396
20:54 forrest AdamSewell, looks like you are using recurse wrong
20:54 xmltok any jinja masters? i have a list that i need to join with "," but i need to quote each entry in the list first
20:54 forrest recurse doesn't support sources, only user/mode/group
20:54 AdamSewell forrest, i had thought i could use that to sync the directory
20:55 forrest AdamSewell, you're thinking file.recurse.
20:56 AdamSewell so i'd have to use them separately then?
20:56 forrest xmltok, are you using join?
20:56 xmltok yeah
20:57 xmltok list | join(', ') does half of what i want, [1,2,3] => 1, 2, 3, but what i need is "1", "2", "3"
20:58 xmltok http://stackoverflow.com/questions/12794148/build-list-of-links-using-jinja2 it looks like i need to do something like this
20:59 jaimed_ joined #salt
20:59 Katafalkas joined #salt
21:01 forrest xmltok, yea it's kinda weird in jinja since you can't really iterate like you normall would in python. Where is this data coming from?
21:01 hunter http://docs.saltstack.com/topics/tutorials/states_pt4.html  -- based on that, do people use multiple environments of pillar data? IE, do you have a prod/base/dev of pillars? Or just one pillar and multiple environments only applies to states?
21:01 abe_music has anyone encountered this before? TerminalException: Failed to spawn the VT. Error: out of pty devices
21:02 forrest abe_music, weird, is that occurring when trying to run a command off the commandline?
21:03 forrest hunter, it's possible to do it different ways depending on what you need.
21:03 abe_music forrest: nope, that's coming from salt-cloud during a create using a map file
21:03 forrest there isn't a 'standard'
21:03 forrest abe_music, odd
21:03 abe_music forrest: s/odd/bad/ :(
21:03 hunter forrest: I know - I'm trying to figure out what I need/want.
21:03 hunter forrest: right now I really think I need a best practices guide until I know salt (and my needs) better.
21:04 xmltok forrest: its in a pillar, it will be hostnames that will get shoved into a rabbitmq config. i dont want to make people have to quote the hostname inside the pillar
21:04 forrest abe_music, I'm pretty sure that error is usually from the VM itself crapping out in some fashion, but I'm not 100% sure
21:04 rallytime joined #salt
21:04 forrest xmltok, yea fair enough
21:04 xmltok anyway, this seems to work fine, it just looks like i dont konw what im doing in the config template. im ok with that :)
21:04 abe_music forrest: anyone you can recommend i bring this issue to?
21:04 cewood joined #salt
21:04 cewood joined #salt
21:05 forrest abe_music, nope, I'd say ensure you can duplicate it elsewhere to confirm it isn't just wherever you are trying to provision, then create an issue.
21:05 rojem joined #salt
21:05 hunter forrest: what do YOU recommend - you've given good advice before?
21:06 frasergraham joined #salt
21:06 forrest hunter, the reason it supports that stuff is for flexibility, if you're just getting started it's probably better to stick with one env pillar
21:06 hunter forrest: and multiple state envs?
21:07 mpoole joined #salt
21:07 forrest To begin with yea
21:07 forrest it wouldn't be too bad to transition later, then you can kind of get a grasp on the pillars to begin with
21:08 hunter What I'd really like ot see is one based on reclass
21:08 hunter But thats also probably more advanced than what I need at the moment.
21:08 forrest I'd probably have a pillar for each app that has multiple environments in it, so within the apache pillar, a prod, dev, and test section or whatever.
21:09 hunter Hmmm.
21:09 harobed_ joined #salt
21:09 forrest Or you could split it into 3 *shrug*
21:09 hunter What I'd like is a single place to assign a host to a given environment, rather than counting on having each host only specified in one pillars/top.sls
21:10 hunter file
21:10 hunter I know the base/top.sls overrides them....
21:11 hunter Is environment encoded within salt? In the base/top.sls can you say "minon.foo.com": - env: production and then have all the other minions info coming from only the production pillar?
21:14 chadhs joined #salt
21:22 mephx joined #salt
21:25 madduck hunter: what do you want to see based on reclass?
21:26 hunter a working example
21:26 madduck ouch ;)
21:26 madduck yeah, i have to work on that…
21:26 yomilk joined #salt
21:27 hunter something that (preferably doesn't use hostnames) and classifies each machine into categories (web server, analysis server, mongodb server) and whatnot
21:27 hunter madduck: even a SIMPLE setup would get me started.
21:30 madduck hunter: what do you mean with "doesn't use hostnames"?
21:30 madduck hunter: have you watched the LCA2014 video?
21:33 jaimed joined #salt
21:34 hunter I started it but haven't finished.
21:34 hunter Our hostnames are based on bloodiest battles in history.
21:34 madduck you know that secretly reclass is for all those people who actually still use proper hostnames ;)
21:35 hunter The war tells you the functional group and the naming scheme tells you how happy I was to take over IT for these people. :)
21:35 mgw1 madduck: can reclass be used in a masterless config?
21:35 hunter Thats a good question.
21:35 hunter madduck: given my host count (60+) I can live with specifiying them manually.
21:36 madduck mgw1: never tried it, it *is* a pillar plugin for the master currently. I don't know how masterless would use it, but I'd imagine it would be possible to write an adapter
21:36 madduck but then you need the whole database on every client, no?
21:37 mgw1 madduck: yeah, I don't want to do it… was proxying a question for someone else
21:37 mgw1 I didnt' think it would work
21:38 faef can you specify environments in an external pillar?
21:38 zzzirk joined #salt
21:41 nebuchadnezzar joined #salt
21:41 mgw1 faef: you mean have access to the env?
21:42 mgw1 top specifies the environment
21:45 faef mgw1: sorry, I meant if you have top files in each external pillar, are they merged or overwritten?
21:46 bemehow joined #salt
21:48 mgw1 faef: you have access to the existing pillar, and you can make updates to it. Or you can return pillar data which will get merged, but conflicts will be updated with the later pillar
21:50 [diecast] joined #salt
21:53 mr_chris What is the proper way to create a a new ordered dict from within a salt state using the py renderer? I've tried OrderedDict(), collections.OrderedDict and import collections collections.OrderedDict. Nothing works.
21:54 mgw1 mr_chris: did you import collections?
21:54 mr_chris See my previous statement.
21:54 mgw1 sorry, missed that part
21:54 mr_chris With just OrderedDict I get NameError: global name 'OrderedDict' is not defined
21:55 alunduil joined #salt
21:55 mr_chris With collections.OrderedDict I get "NameError: global name 'collections' is not defined"
21:55 mgw1 you definitely need to import collections
21:55 mr_chris With import collections, collections.OrderedDict I get "AttributeError: 'module' object has no attribute 'OrderedDict'"
21:56 mgw1 what version of python?
21:56 mr_chris 2.6.6
21:56 jaimed joined #salt
21:57 mr_chris But I already have plenty of OrderedDict objects in my states from the pillar data.
21:57 mr_chris So I know that they do work.
21:57 mgw1 http://docs.python.org/2.6/library/collections.html
21:57 mgw1 I'm not sure how salt is handling ordered dicts on 2.6
21:58 dave_den mr_chris: check out utils.odict.py
21:58 thayne joined #salt
21:58 dave_den 2.6 doesn't have native support for ordereddict
21:58 mr_chris dave_den: Ah.
21:59 mgw1 there you go, so import salt.utils.odict.OrderedDict
21:59 dave_den try pip install ordereddict
22:01 jfzhu_us Does anyone know how to do the equivalent of a yum group install in salt?
22:02 mr_chris dave_den: My test environment does not currently have pip nor do I want to have to pip install something on all of my servers for what I'm trying to do. Considered that I've got plenty of OrderedDict objects already floating around in the running state, and considering that I just want to manually add one more, how would I utilize the OrderedDict class that's already there?
22:03 mr_chris jfzhu_us: http://docs.saltstack.com/ref/modules/all/salt.modules.yumpkg.html#salt.modules.yumpkg.group_install
22:03 dave_den from salt.utils.odict import OrderedDict
22:03 manfred jfzhu_us: there isn't a state for it, but there is a module
22:04 n8n joined #salt
22:04 forrest mr_chris, looks like that is only supported in yumpkg, not yumpkg5 as well
22:04 mr_chris dave_den: That was it. Thanks. Learned a bit more about python today.
22:04 jphall joined #salt
22:04 mr_chris Thanks mgw1 too.
22:04 jfzhu_us Rats. Ah well I guess i'll just define a pillar with the packages
22:04 dpn` joined #salt
22:05 _jslatts joined #salt
22:05 forrest terminalmage, do you know why group_install is in yumpkg, but not yumpkg5? And did you guys ever make headway on this comment: https://github.com/saltstack/salt/pull/3524 ?
22:05 mr_chris forrest: http://docs.saltstack.com/ref/modules/all/salt.modules.yumpkg5.html#salt.modules.yumpkg5.group_install
22:06 forrest mr_chris, https://github.com/saltstack/salt/blob/v0.17.5/salt/modules/yumpkg5.py
22:06 mr_chris forrest: I stand corrected.
22:06 forrest mr_chris, it's in there because the docs are built from develop
22:07 forrest and yumpkg5 isn't in develop
22:07 lyddonb_ joined #salt
22:07 darless joined #salt
22:07 forrest so it must be pulling the docs from yumpkg?
22:07 forrest I don't know
22:07 dpn` joined #salt
22:07 forrest mr_chris, not trying to 'prove you wrong' or anything, just pointing out that consistency issue :P
22:07 darless Is there a way to have git do a checkout (force_checkout: True, and force: True) doesn't do it.
22:07 mr_chris forrest: No. I understand. If I'm wrong I want to know it.
22:07 forrest cool!
22:07 smkelly joined #salt
22:08 forrest I really have no idea where those docs are coming from in develop though :|
22:08 manfred jfzhu_us: you could use the module.run state
22:08 forrest darless, https://github.com/gravyboat/hungryadmin-sls/blob/master/salt/hungryadmin/app.sls#L36
22:08 forrest I'm using that, works like a charm to force the checkout every time.
22:08 jfzhu_us manfred: Oh wow. Didn't know that existed. Thanks!
22:08 terminalmage yumpkg5 no longer exists
22:09 forrest Right, the docs are still built for it on develop it seems though
22:09 terminalmage ah
22:09 forrest or they are coming from yumpkg
22:09 forrest since they have group_install which never existed in yumpkg5
22:09 terminalmage ok
22:09 terminalmage it's probably files left over
22:09 forrest Did you guys ever have any luck getting that into the state?
22:09 forrest or figuring out how to do so
22:09 [diecast] joined #salt
22:09 terminalmage the file.recurse we use to deploy the docs probably needs a "clean: True"
22:10 terminalmage no I didn't even start to look into it
22:10 forrest gotcha
22:10 terminalmage we'd need a way to resolve the groups and turn them into package names
22:10 terminalmage I have a rough idea of how to do it
22:10 jaimed joined #salt
22:10 forrest why would you have to resolve the group names?
22:10 forrest instead of simply passing them to yum
22:11 terminalmage so the state knows what packages need to be installed
22:11 terminalmage the check at the end that checks to see if the state was fulfilled
22:11 terminalmage needs to know what packages constitute a successful result
22:13 forrest wouldn't a binary 'yes/no' be ok in that case? Because either all the packages in the group succeed, or you would consider it a failure.
22:13 nextdoorwarren joined #salt
22:14 terminalmage I don't know
22:14 forrest you're the yum expert though! :P
22:14 jslatts joined #salt
22:14 terminalmage I don't know if there's a way to check if a group is fully installed
22:14 terminalmage without checking if the packages that make it up are installed
22:15 ipmb joined #salt
22:15 nebuchadnezzar joined #salt
22:15 forrest as far as I know you'd have to query using 'groupinfo' and ensure those are installed, which is a real pain
22:16 giantlock joined #salt
22:18 yomilk joined #salt
22:18 quickdry21 joined #salt
22:18 elfixit joined #salt
22:19 VladK joined #salt
22:19 zzzirk_ joined #salt
22:20 terminalmage ok, then I think you answered your question :)
22:20 forrest yea, it doesn't make it any less lame
22:23 vbabiy_ joined #salt
22:24 jslatts joined #salt
22:26 andrej If I install a new minion ... is there a way to make salt push out config as soon as it connects, rather than having to run a highstate?
22:28 Ahlee andrej: yes, you can do it with reactor
22:28 dave_den andrej: there is startup_states in the minion, or you can react to a minion being accepted event on the master
22:28 Ahlee andrej: https://gist.github.com/jalons/8733173
22:28 Ahlee ooo, startup_states
22:28 Ahlee good call.
22:32 Ryan_Lane1 joined #salt
22:32 darrend hi.. where does the _modules directory live on the master?  I'm struggling to sync a simple custom module.  Docs suggest "inside file_roots on the master" but this isn't working for me
22:33 krow joined #salt
22:40 fauxweasel joined #salt
22:40 david_a joined #salt
22:42 lurker joined #salt
22:42 dave_den darrend: how are you trying to sync it?
22:43 forrest darrend, if it's in /srv/salt/_modules, you still have to sync it over with salt '*' saltutil.sync_all
22:43 joehh Ryan_Lane1, forrest: Unfortunately this is a side effect of the way launchpad works.
22:43 forrest joehh, gotcha, I know you've explained it a few times before
22:43 forrest but I couldn't remember.
22:43 forrest I will try to do so this time :P
22:43 joehh The 0.16 releases are still around in the https://launchpad.net/~saltstack/+archive/salt16 ppa
22:44 forrest and most likely fail
22:44 joehh no worries :)
22:44 psyl0n joined #salt
22:44 darless forrest: So that's the way I have my git.latest, but the problem is with the untracked files which might be in use after a git pull, thus a [git checkout] is not enough. It would be nice to have an option that would do a [git reset HEAD --hard] before doing the pull
22:45 forrest why would there be untracked files?
22:45 faef is there a way to set root directory on external pillars from git, like you can with gitfs and file_roots?
22:45 VladK Hi I'm installing salt-cloud on AWS insatnce with Ubuntu 12.04 (with repository ppa:saltstack/salt). when I apt-get install salt the version is 0.17.5, but it does not have salt-cloud command. If I install salt-cloud it's version is 0.8.11. I know that salt-cloud is merged into salt package. How can I use salt-cloud 0.17.5 ? Any links to the installation/usage would be nice (I could not find 0.17.5 specific info for the cloud in docs.salt
22:46 forrest VladK, the salt-cloud version is not identical to the version release of salt itself as they were not started at the same time. 0.8.11 should work with 0.17.5
22:46 darrend dave_den: forrest: i think it's sync'd ok.. can see the file in /var/cache/salt/minion/extmods/modules/
22:47 VladK ok. thanks, forrest
22:47 forrest VladK, yea np
22:47 forrest faef, I'm not sure.
22:47 dave_den faef: no, there is no pillar git equivalent of gitfs_root at the moment
22:47 darrend but I get "mymod.foo is not available" when trying to call it
22:48 dave_den darrend: set your minion log level to debug and see if there is an exception when loading the module
22:48 darrend dave_den: will try, thx
22:48 faef dave_den: are top.sls merged across different external pillars from git ?
22:49 VladK and what is the latest version of salt-cloud? I see that people are talking about starting instances in VPC, but I can not reference for the syntax to specify vpc_id, subnet_id or sg_id and also in which congig file to put them. providers, profiles, maps?
22:49 dave_den darrend: also, are you trying to run the module as a state?
22:49 _jslatts joined #salt
22:50 VladK :-) correction: "can not find" reference to syntax
22:50 darrend dave_den: no, just doing "salt '*' mymod.foo"
22:50 chadhs joined #salt
22:51 dave_den faef: i don't know the answer to that for sure. that is the way tops work with gitfs state top.sls files though
22:52 darrend dave_den: yes, error when loading it.  syntax error - typo in the module file :(
22:54 dave_den darrend: doh!
22:55 darrend dave_den: yup.
22:55 darrend dave_den: works now: )
22:56 dave_den cool
22:57 vbabiy_ joined #salt
22:57 andrej Thanks Ahlee and dave_den
22:58 faef dave_den: seems like a small fix to add root directory to git_pillar ?
22:58 dave_den andrej: you're welcome. which method did you choose?
22:58 sroegner joined #salt
22:59 dave_den faef: yes, it should be a rather small fix. if you can open an issue or submit a PR if you know how to fix it it would be appreciated
22:59 andrej I haven't decided yet
22:59 andrej trying to understand the startup_states
23:00 darless forrest: reason for untracked file - I'm special. This server I rsync changes to it test it out, then push my changes. When I do a salt run it can't pull the changes because of the untracked files (my changes)
23:00 andrej is that something you configure in /etc/salt/minion?
23:00 darless forrest: I can get around this pretty simply by having a cmd.wait which does a [git reset HEAD --hard] before doing the git pull
23:01 andrej yup ... just found the commented section
23:01 dave_den andrej: http://docs.saltstack.com/ref/states/startup.html
23:01 andrej thanks dave_den - that doesn't really mention WHERE it happens.
23:02 andrej Only says what it does, and what the options are
23:02 andrej But that's what I'm gonna run with
23:02 forrest darless, yea but it would be nice if the command supported it. You should create an issue for that.
23:02 dave_den yeahs it's not super clear - it has a tiny mention of "As of Salt 0.10.3 the minion config reads options"
23:03 danielbachhuber joined #salt
23:03 darless forrest: k will do
23:04 Ryan_Lane1 joined #salt
23:04 faef dave_den: will do
23:08 forrest darless, cool
23:11 hunter forrest: thank you for you advice earlier
23:11 hunter madduck: if you get even a gist of stuff, let me know.
23:14 AdamSewell joined #salt
23:19 mgw1 joined #salt
23:20 kermit joined #salt
23:21 forrest hunter, no worries, it's my job to give out bad advice for free.
23:21 __number5__ Is there a way to get stdout/stderr when running cmdmod.script_retcode?
23:21 sroegner joined #salt
23:22 sgviking joined #salt
23:22 bwq joined #salt
23:24 __number5__ oh, cmdmod.script actually return retcode,stdout,stderr :) thanks
23:26 gnugnu1 joined #salt
23:27 krow joined #salt
23:28 kalloc joined #salt
23:28 jeremyfelt joined #salt
23:29 tonthon joined #salt
23:32 krow joined #salt
23:32 fllr joined #salt
23:33 jcristau joined #salt
23:37 justBob joined #salt
23:40 faldridge joined #salt
23:40 techhat joined #salt
23:41 ajw0100 joined #salt
23:42 krow joined #salt
23:47 gw joined #salt
23:47 gw joined #salt
23:48 gw joined #salt
23:48 gw joined #salt
23:48 faldridg_ joined #salt
23:52 halfss joined #salt
23:53 gnugnu1 joined #salt
23:57 falican joined #salt
23:58 bemehow joined #salt

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