Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2013-11-08

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

All times shown according to UTC.

Time Nick Message
00:00 oz_akan_ joined #salt
00:01 oz_akan_ joined #salt
00:01 jhulten joined #salt
00:02 ctdawe joined #salt
00:03 KyleG joined #salt
00:03 KyleG joined #salt
00:07 jacksontj joined #salt
00:11 NotreDev joined #salt
00:17 KyleG joined #salt
00:17 KyleG joined #salt
00:18 cdcalef anybody use salt-cloud much?
00:19 xmltok cdcalef: i found that looking at the source helped me a little there, you can see the connection between states/modules
00:20 xmltok and the pydoc for salt.states.* helps a lot when youre writing a state file
00:20 cdcalef xmltok: yeah its not immediately apparent when you're looking at yaml's with whats a state, whats a module, whats an arg..
00:20 xmltok the more i am using pydsl the more i prefer it over yaml
00:21 aleszoulek joined #salt
00:24 ajw0100 joined #salt
00:25 tty-1 joined #salt
00:26 cbloss Anyone have any idea why the service is failing to reload? http://pastebin.com/f2B96i2E
00:28 NotreDev how do i proof of concept at the cli that filesystem.gitfs works?
00:28 KyleG left #salt
00:29 jhulten joined #salt
00:31 jslatts NotreDev: i just just run the master in debug mode on the command line and make sure it doesn't throw any errors
00:31 jslatts salt-master -l debug
00:31 NotreDev cool
00:31 Jahkeup cbloss: try changing - file: /etc/btsync/* to be an id or use watch_in in your files
00:31 trickyher joined #salt
00:32 Jahkeup haha jk
00:32 Jahkeup cbloss: my bad, since when could you use globs on ids?
00:32 __number5__ cbloss: can you manually start/stop btsync service?
00:32 jslatts NotreDev: and you can run a state.highstate test=TRUE to verify its working without changing anything
00:33 __number5__ I never get that working with their default init.d script :P
00:33 cbloss __number5__: yeah, i put that in the bottom of the paste :)
00:35 NotreDev any way to take an action if a file does not exist?
00:36 cdcalef whats the 'manual reload' part mean?  you had 64 that reloaded fine and 1 that didn't?
00:37 jgarr_afk left #salt
00:38 tty-1_ joined #salt
00:39 __number5__ cbloss: wait..., why line 37 file.absent required btsync pkg?
00:39 cbloss for debconf-default.conf?
00:39 cbloss becuase the default install puts that sample config there
00:39 __number5__ yep
00:40 __number5__ try to remove require: after it
00:41 cbloss same problem
00:42 mwillhite joined #salt
00:42 __number5__ anything in logs with -l debug?
00:43 cbloss jesus now i broke everything what did i do
00:43 cbloss hah
00:45 jhulten joined #salt
00:46 cbloss ahhh
00:46 cbloss here is the problem
00:46 cbloss [INFO    ] Executing command 'service btsync reload' in directory '/home/chris'
00:46 cbloss [ERROR   ] {'btsync': False}
00:46 cbloss the command is restart not reload
00:46 cbloss is it possible to customize that?
00:49 cbloss changed "- reload" to "- restart"
00:49 cbloss good to know, thanks guys
00:49 Ryan_Lane joined #salt
00:49 trickyher left #salt
00:59 pentabular joined #salt
00:59 bhosmer joined #salt
01:05 cbloss how do you guys modify state/pillar files on a master remote server? rsync? btsync? something else?
01:08 __number5__ rsync or gitfs
01:10 jkleckner joined #salt
01:13 redondos joined #salt
01:13 redondos joined #salt
01:16 gw joined #salt
01:18 redondos joined #salt
01:21 gw Hi, I have a quick question about good practice setting up ssh keys between users  during provisioning ....
01:21 gw e.g. the provision step creates user1 and user2, and user1 needs to be able to ssh to localhost (and possibly to remote hosts that are provisioned by salt as well) as user2
01:23 Cidan ssh to it self?
01:23 gw could think of 2 ways doing this .... 1st would  be to put the ssh keys (incl. private keys) into pillars or accessible through salt:// url and deploy them via normal state handling
01:23 gw 2nd would be to use overstate system
01:23 gw no ssh as different user ... but on same machine
01:23 Cidan but you said, "user1 needs to be able to ssh to localhost"
01:23 Cidan so I'm lost right there
01:23 gw as user2
01:24 Cidan do you mean sudo as user2?
01:24 Cidan on the localhost?
01:24 Cidan (forget remote right now, we'll get to that in a sec)
01:24 gw yes would one way, but I thought I'd like to use ssh as user1 needs to do the same for remote machines as well
01:25 Cidan yeah don't do that, that's... bad practice
01:25 Cidan setup sudo so user1 can become user2
01:25 Cidan then install the keys on user2 all around
01:25 gw yes ... that's easy :)
01:26 Cidan then user1 can do sudo -u user2 "ssh remotehost 'some command'"
01:26 xl1 joined #salt
01:26 gw ok ... would it be possible to generate the ssh keys during provisioning or is the only reasonable way to have the user1 priv and pub key in the salt states/pillar tree
01:26 ctdawe joined #salt
01:27 Cidan you can generate it during provisioning, no problem, but that SAME key needs to exist everywhere, well, the pub does
01:27 gw ok forget user1 distinction ..... with your sudo ssh combination this doesn't matter anymore
01:27 Cidan your best bet is to pre-generate the key tbh
01:28 Cidan this way if you want to change it, update the pillar, state.highstate, and you're done
01:28 Cidan key updates everywhere
01:28 gw yeah thought so ....
01:28 Cidan just make sure your key lives in a pillar that only those hosts can access
01:28 Cidan and you're fine, :)
01:28 gw one more question here .... I didn't quite understand data sharing of pillars and states between minions
01:28 Cidan sure, so
01:29 gw so pillars are visible to host only? and state tree to all hosts?
01:29 Cidan so
01:29 Cidan here's a few stuff
01:29 Cidan http://pastie.org/8464251
01:29 Cidan that's in my master config
01:30 gw ok ... so far mine looks similar
01:31 Cidan http://pastie.org/8464255
01:31 Cidan that would be in /srv/pillar/base/top.sls
01:31 gw cool .... still following you
01:32 Cidan machines in prod, with the salt id of 'id-of-machine-with-key-*' (glob match)
01:32 Cidan can read the variables stored in pillar-sls-with-key
01:32 Cidan which would be in /srv/pillar/prod/pillar-sls-with-key.sls
01:33 gw as well as the was and salt pillars from base
01:33 Cidan yes
01:33 Cidan exactly
01:33 gw ok pillars are a safe way to keep private data for matching hosts
01:34 Cidan anything that does not match 'id-of-machine-with-key-*' can not read the pillar's under it
01:34 Cidan yes, they are
01:34 Cidan so long as you don't turn on auto accept of keys in salt-master
01:34 Cidan :)
01:34 gw wouldn't ever do that ;)
01:34 gw does this work for the state tree as well?
01:34 Cidan yep!
01:34 Cidan same basic concept
01:35 gw so states are only transferred for matching id's
01:35 Cidan right
01:35 gw just thought, within state tree I can keep the keys as files there instead of squeezing them in to yaml
01:35 Cidan yep, whatever works for you
01:36 Cidan though I think the point of pillar is to store data like keys, etc
01:36 Cidan you can always write a small script to generate the yaml for you using python
01:36 gw yes, pillars would be the much more obvious place... I agree
01:37 Cidan ./make-my-pillar.py /path/to/key
01:37 gw mm... cool idea .... using a #!py pillar
01:37 flebel joined #salt
01:37 Cidan or that
01:38 gw thanks for your help... I'll try the #!py pillar .... maybe I am able to generate a key there one the fly in case it doesn't exist yet
01:39 alunduil joined #salt
01:40 Cidan totally, great idea
01:40 jslatts joined #salt
01:42 gw reading up on self renderers and have another question :) ..... it says a renderer get's all the __salt__, __grains__, __pillar__, __opts__, __env__ and __sls__ dictionaries ....
01:42 gw is there some documentation what's actually in them?
01:42 gw __grains, pillar, opts is obvious
01:42 gw but I am not sure about the other ones
01:43 gw or is this one of the cases .... would be great if someone could write a bit about them ;)
01:49 Ryan_Lane joined #salt
01:52 Cidan sorry I was in the shower
01:53 gw np.  just started my work day .. so I have time ;)
01:53 Cidan ha
01:53 Cidan well, let's see
01:53 gw and your help was great so far .... :)
01:53 nmws_ joined #salt
01:53 Cidan I don't actually know what all those are
01:53 Cidan you could yaml dump them to a file and find out! lol
01:53 gw so my last statement was pretty accurate :)
01:53 gw yes ... that's what I try do now ;)
01:54 UtahDave joined #salt
01:54 nmws_ how does salt handle user permissions, like user foo is only allowed to run modules x, y and z
01:55 Cidan nmws_: not... following.  Sorry
01:56 gw nmws_: client_acl in the salt-master config  e.g. http://docs.saltstack.com/ref/clientacl.html if you meant this
01:56 nmws_ gw: perfect, thanks :) :)
01:56 gw nmws_: there is also a separate config for user identification in the master config
01:56 Cidan oh neat, I didn't even know that existed
01:56 gkze joined #salt
01:57 gw tried it recently .... being a lazy person I didn't want to use sudo all the time ;)
01:57 Cidan pft
01:57 Cidan I just use root everywhere
01:57 josephholsten joined #salt
01:57 * Cidan is a bad person
01:58 gw IMO it's usually better to  explicitly change user before doing any potentially harmful stuff ;)
01:58 Cidan you're absolutely right
01:58 Cidan I just, don't do that because I'm a dork
01:58 nmws_ gw: and is there some sort of audit log where I can check user X ran module Y?
01:58 gw nmws_: sorry don't know
01:59 jkleckner joined #salt
02:00 gw but for development or testing I rather choose the lazy way :)
02:01 backjlack joined #salt
02:01 bemehow_ joined #salt
02:01 ajw0100 joined #salt
02:06 gw ok __salt__ has all the modules. e.g. 'pillar.items' ... just the python function references
02:08 gw __pillar__ is empty during in a #!py pillar sls. (no surprise here, as the pillars are still built)
02:09 gw __env__ is a string that has the current environment name
02:09 gw __sls__ is the name for the current sls as you would reference it from the top.sls file
02:10 redondos joined #salt
02:10 redondos joined #salt
02:10 gw __opts__ seems to have all the master config values, and things like command line parameters etc....
02:10 ajw0100 joined #salt
02:11 joehh joined #salt
02:17 ipmb joined #salt
02:17 Tekni joined #salt
02:25 jhulten joined #salt
02:25 ipmb joined #salt
02:29 bemehow joined #salt
02:37 morty_ joined #salt
02:37 pentabular left #salt
02:38 bemehow joined #salt
02:44 NotreDev joined #salt
02:44 Thiggy joined #salt
02:46 ipmb joined #salt
02:49 bemehow joined #salt
02:53 Ryan_Lane joined #salt
02:56 bemehow_ joined #salt
02:59 oz_akan_ joined #salt
03:06 bemehow joined #salt
03:14 Thiggy joined #salt
03:14 redondos joined #salt
03:14 redondos joined #salt
03:25 oz_akan_ joined #salt
03:29 NotreDev how should i be storing/distributing id_rsa and id_rsa.pub files?
03:29 NotreDev public keys ...
03:29 gw NotreDev: Ideally within the yaml in pillars, but within the state tree works as well
03:30 gw NotreDev: within the state tree you can access them via salt:/// urls
03:30 NotreDev gw: how should i be protecting the pillar data?
03:30 gw NotreDev: always
03:30 NotreDev :) how to do it though?
03:31 NotreDev git submodule?
03:31 gw restrict them to the user the salt master runs as
03:31 gw normal file system permissions
03:31 NotreDev we have a company bitbucket account, and i don't want to make these files super public
03:32 gw public keys is normally no big issue. but yes it's always a bit of a bad feeling to put key data into some place you don't have full control over
03:32 gw gifts might be a solution for you?
03:32 Ryan_Lane NotreDev: use multiple pillar or state roots
03:32 Ryan_Lane then have two repos
03:32 gw not sure ... never tried gifts
03:32 Ryan_Lane one public and one private
03:34 NotreDev Ryan_Lane: where do i find that under the docs?
03:34 gw or write a #!py pillar which fetches the data from wherever you want?
03:34 NotreDev https://salt.readthedocs.org/en/latest/ref/file_server/backends.html?highlight=multiple%20roots
03:34 Ryan_Lane it's in the configuration
03:34 Ryan_Lane sorry I don't have direct links :)
03:34 NotreDev gw: also not a bad idea
03:35 NotreDev Ryan_Lane: yeah, i think that's it
03:35 gw Cidan gave me the idea about 2hrs ago in this channel ;)
03:35 jhulten joined #salt
03:36 ctdawe joined #salt
03:38 Cidan yay I did something!
03:38 pentabular1 joined #salt
03:39 NotreDev should i put the key data into a submodule or subtree?
03:39 bfeister_ joined #salt
03:39 pipps joined #salt
03:41 gw NotreDev: sorry not exactly sure what you mean
03:44 NotreDev gw: thinking about putting the pub/priv key data into a git submodule. dealing with git is always an adventure
03:45 whitepaws what's up my righteous salt brothers?
03:45 whitepaws when you are running salt minions on an old version of zmq (<3) .. is it normal to get a lot of "Minion did not return" messages?
03:45 whitepaws that is, zmq < version 3
03:46 gw NotreDev: mm... I am afraid I can't help much there ..... I never use git submodule, ... I find it to complicated ;)
03:46 whitepaws and should i expect those messages to stop happening after i upgrade them to 3.2.2 or so?
03:46 NotreDev haha
03:47 redondos joined #salt
03:47 redondos joined #salt
03:53 redondos joined #salt
03:53 redondos joined #salt
03:56 racooper joined #salt
04:02 anuvrat joined #salt
04:04 whitepaws it's weird because you can see the master logs saying "Got return from blahblahblah for job 20131108040020108176" .. and the minion's debug logs similarly say "[INFO    ] Returning information for job: 20131108040020108176"
04:06 redondos joined #salt
04:06 redondos joined #salt
04:08 NotreDev i'm looking at the filter_by docs, and don't understand what's happening here - merge=salt['pillar.get']('apache:lookup')) %}
04:09 NotreDev what is apache:lookup?
04:10 NotreDev e.g. is 'apache:lookup' actually a key, or is lookup a key of apache and this is some special dsl
04:11 pears that's a key
04:11 jesusaurus its not really a dsl, but its an alternate syntax since you are passing a string to a function call
04:11 pears I'm assuming you're looking at this? http://docs.saltstack.com/ref/modules/all/salt.modules.grains.html
04:11 jesusaurus its equivalent to pillar['apache']['lookup']
04:11 NotreDev jesusaurus: exactly what i was wondering. thanks
04:12 pears not quite equivalent, it's safe in the way that python's dict.get() method is
04:12 jesusaurus no prob
04:12 pears you can specify a default if nothing is there
04:12 pears nitpickin'
04:12 jesusaurus as far as the data structure of your pillar is concerned, its equivalent
04:13 jesusaurus :P
04:16 ctdawe joined #salt
04:19 pentabular joined #salt
04:20 sciyoshi1 joined #salt
04:26 oz_akan_ joined #salt
04:26 lemao joined #salt
04:30 pears joined #salt
04:33 pipps_ joined #salt
04:43 pentabular joined #salt
05:01 Ryan_Lane joined #salt
05:04 [1]VertigoRay joined #salt
05:12 linjan_ joined #salt
05:17 faldridge joined #salt
05:26 oz_akan_ joined #salt
05:28 [2]VertigoRay joined #salt
05:33 gkze joined #salt
06:00 jacksontj joined #salt
06:02 oz_akan_ joined #salt
06:10 ravibhure joined #salt
06:10 oz_akan_ joined #salt
06:11 sgviking joined #salt
06:11 APLU joined #salt
06:13 jacksontj joined #salt
06:18 malinoff joined #salt
06:24 bhosmer joined #salt
06:27 lemao joined #salt
06:33 oz_akan__ joined #salt
06:52 IJNX joined #salt
06:55 Nexpro joined #salt
07:12 Ryan_Lane joined #salt
07:14 tty-1 joined #salt
07:15 amahon joined #salt
07:18 scott_w joined #salt
07:24 ravibhure hi
07:24 ravibhure any one alive :)
07:24 ravibhure need help on pillar programming
07:25 ravibhure condition based variables in pillar
07:25 faldridge joined #salt
07:30 scott_w joined #salt
07:31 NotreDev hmm
07:31 ravibhure anyone?
07:31 NotreDev i'm just wondering why salt refuses to install mongo-10gen-server: http://pastebin.com/1XkJTeRD
07:31 scott_w_ joined #salt
07:40 mbrgm joined #salt
07:41 mbrgm hey! does anyone know how to combine a pkg.installed and pkg.removed within one superstate?
07:42 malinoff ravibhure, What's the question?
07:42 ravibhure Condition based variables in pillar
07:42 malinoff mbrgm, What do you mean?
07:42 ravibhure have large network in different IPs
07:42 malinoff ravibhure, {% if condition %} {% endif %}, if you use jinja as template engine
07:43 ravibhure it apply into pillar ?
07:43 malinoff ravibhure, Let me pastebin an example
07:43 ravibhure mailnoff: ack
07:44 mbrgm malinoff: right now my state tree looks like this: http://pastebin.com/raw.php?i=5W9FKV3w
07:45 mbrgm but it won't work as pkg state is included twice. so what's the right way?
07:45 malinoff ravibhure, http://pastebin.com/suazhEQR here is project.sls pillar file
07:45 ravibhure ack
07:46 petersabaini joined #salt
07:47 malinoff mbrgm, you should use this: http://pastebin.com/iwZirJAL
07:47 malinoff oops, there is a typo
07:47 malinoff sorry
07:48 mbrgm nm
07:48 mbrgm malinoff: ok, i tried this but forgot to prefix installed and removed function with dashes
07:48 malinoff mbrgm, now it works?
07:49 mbrgm i didn't figure out what the difference between prefixing with dashes and not doing it is?
07:49 mbrgm malinoff: booting vm right now...
07:49 malinoff mbrgm, well, salt is verifying the syntax
07:50 malinoff dash is saying that this is a list element
07:50 malinoff If you are not familiar with yaml syntax, I'd rather commend to read about that
07:50 ravibhure mailinoff it works
07:51 malinoff ravibhure, of course :)
07:51 mbrgm malinoff: ok, got it. i'll read about yaml
07:52 giantlock_ joined #salt
07:53 malinoff mbrgm, Btw, you are not limited to yaml: http://docs.saltstack.com/ref/renderers/all/index.html
07:54 mbrgm oh, didn't know about that. i just jump-started into salt ;-) thx!
07:54 mbrgm btw. doesn't work like you proposed.
07:55 mbrgm it says "no function declared in state "pkg" in sls gitlab"
07:56 mbrgm i just remember that i got that one before :-D
07:57 mbrgm malinoff: you got another idea?
07:59 malinoff mbrgm, ah, right. Wait a second
07:59 scott_w joined #salt
08:00 mbrgm leaving out the dashes from installed and removed results in ordered dict has no attribute append error
08:02 malinoff mbrgm, hmm, looks like you can't do that that way
08:02 malinoff mbrgm, check this example: http://docs.saltstack.com/topics/tutorials/starting_states.html#default-data-yaml
08:02 malinoff mbrgm, but there are two different modules - pkg and service
08:02 gw joined #salt
08:04 mbrgm malinoff: what does it have to do with my issue? sry, can't see the point.
08:06 malinoff mbrgm, http://pastebin.com/CEHcBzwk
08:07 mbrgm ok, so i have to split it up into two states
08:07 malinoff mbrgm, yep
08:07 mbrgm tried to avoid this but i think i'll be ok with it
08:07 malinoff well, not actually states
08:07 malinoff just two actions under one state
08:07 mbrgm ah ok
08:07 mbrgm ty
08:08 pentabular joined #salt
08:13 slav0nic joined #salt
08:13 slav0nic joined #salt
08:13 gildegoma joined #salt
08:16 oz_akan_ joined #salt
08:17 harobed_ joined #salt
08:18 mbrgm can i chain requires?
08:19 networkpadawan joined #salt
08:20 gw mbrgm: not necessarily chain, but you can have a list of requirements for a state
08:21 gw mgbrm: and required states can have their own requirements
08:21 mbrgm gw: ok, and can a requirement be an action?
08:21 mbrgm like http://pastebin.com/2k7bYric
08:22 mbrgm last line
08:22 gw mbrgm: usually the elements in a requires list refer to other states
08:22 mbrgm ok, so what i want to do cannot be done?
08:22 creich joined #salt
08:22 gw your require in someotherthing should be : -require: pkg.rbenv_deps
08:23 gw oh derry
08:23 gw sorry
08:24 gw it should read
08:24 gw -require:
08:24 gw - pkg: rbenv_deps
08:25 malinoff mbrgm, http://docs.saltstack.com/ref/states/requisites.html
08:28 lemao joined #salt
08:29 mbrgm gw: well, but rbenv_deps is not a package?
08:29 mbrgm malinoff: i already read this, but did not find an answer
08:30 mbrgm rbenv_deps shall combine several states, e.g. pkg.installed, cmd.run etc.
08:30 mbrgm and in another action i want to ensure that rbenv_deps was run before that action
08:31 malinoff mbrgm, gw said correctly, use
08:31 malinoff - require:
08:31 malinoff - pkg: rbenv_deps
08:31 xl1 Hi, is it possible to set minion options in pillar? I added "backup_mode: minion" but config.option doesn't show it. "config.option backup_mode omit_opts=True" does, but I guess the option still isn't in effect?
08:31 NotreDev after setting grains in /etc/salt/grains, how do i get to see those changes?
08:32 pears grains.items should show them
08:32 malinoff NotreDev, http://docs.saltstack.com/ref/modules/all/salt.modules.saltutil.html#salt.modules.saltutil.sync_grains
08:33 NotreDev my bad, i was doing pillar.get rather than grains.get
08:37 hvn joined #salt
08:38 hvn left #salt
08:38 hvn joined #salt
08:45 tty-1 joined #salt
08:48 jjasinski__ joined #salt
08:49 jjasinski__ Hey all, I've been trying to install salt 0.17.1 on ubuntu 13.10 using the instructions from the main website.  http://docs.saltstack.com/topics/installation/ubuntu.html
08:49 jjasinski__ However, after doing an aptitude update
08:50 jjasinski__ and apt-git install salt-minion, I only get salt 0.16.0-1
08:50 jjasinski__ but more interestingly,
08:51 jjasinski__ whenever I try to uninstall, aptitude actually hangs
08:51 jjasinski__ this only happens when installing salt-master (not other apt packages)
08:51 jjasinski__ and when aptitude hangs, I have to kill -9 it and delete locks
08:52 jjasinski__ This has happened to me on two separate machines, both running Ubuntu 13.10
08:52 jjasinski__ and only when installing salt
08:52 jjasinski__ has anyone experienced anything similar?
08:53 jjasinski__ Is there a salt 0.17.1 package for Ubuntu 13.10 yet?
08:53 jjasinski__ Note:
08:53 bui 4
08:54 jjasinski__ I've tried installing both through the add-apt-repository and then when that didn't work, I removed the repo, and installed via add-key
08:56 mbrgm ok, another question: i have an include: - git, so if i want to make this a requirement, i have to write it as require: - sls: git, right?
08:56 mbrgm include:
08:56 mbrgm - git
08:56 mbrgm ...
08:56 mbrgm - require:
08:56 mbrgm - sls: git
08:57 mbrgm now it says "the following requisites were not found: require: sls: git"
08:59 pentabular joined #salt
08:59 mbrgm but the include did work
08:59 mbrgm i.e. git was installed
09:05 jjasinski__ Hmm, I also tried the bootstrap method just now… it still installed 0.16.0
09:06 hvn @mbrgm: maybe you want require pkg: git
09:06 hvn you can require states, not SLS file
09:09 mbrgm ok, the sls: ... thing in the docs
09:09 mbrgm found the*
09:09 mbrgm but pkg works, thx
09:09 balboah joined #salt
09:10 redondos joined #salt
09:10 redondos joined #salt
09:10 jjasinski__ Ok, I just tried uninstalling salt-master again and aptitude is hung.  This is the pstree output of the uninstall process.  It looks like it's stuck on salt-master.prerm
09:10 jjasinski__ http://hastebin.com/wogenujavu
09:12 ollins joined #salt
09:13 jjasinski__ Anyway, I'm just making note of this in case other people are having aptitude installation/removal issues hanging issues when installing salt-master 0.17.1 on ubuntu 13.10.  Should I open up a github ticket to track this?
09:14 ollins Hello, I have a state which installs and enables a module in apache. How can I force apache to restart from this state?
09:14 giantlock_ joined #salt
09:23 ollins Found it: "extend" works
09:24 jjasinski__ hmm, I keep hitting a loop 1) I try to uninstall salt-master 2) it hangs 3) I kill the aptitude process 4) I run sudo dpkg --configure -a to fix apt, 5) I aptitude install something else 6) it tries to configure salt-master again
09:28 giantlock_ joined #salt
09:33 joehh jjasinski__: not sure why aptitude is hanging, but Corey uploaded a build for 13.10 about 24 hours ago
09:33 joehh the 16.0 is the version that is in ubuntu 13.10 "proper"
09:34 jjasinski__ joehh - thanks for the info.
09:34 sebgoa joined #salt
09:34 jjasinski__ Is the suggested way to install the latest version to use the ppa?
09:35 jjasinski__ (latest stable)
09:35 joehh yes
09:35 jjasinski__ ok
09:35 joehh I'll check on the status of the build
09:35 jjasinski__ Thanks
09:35 ravibhure hi
09:36 ravibhure how can I get interface eth0 ip using grain when I have multiple nic
09:36 gildegoma left #salt
09:37 ggoZ joined #salt
09:39 IJNX is there a package manager for salt modules?
09:41 IJNX Here is list of modules, but how do I correctly install and manage the updates? (https://github.com/saltstack/salt/tree/develop/salt/modules)
09:42 giantlock_ joined #salt
09:43 joehh jjasinski__: looks like the build failed, I'll see if I can upload one
09:44 lemao joined #salt
09:45 jjasinski__ Thanks joehh… I assume the upload will be available via the ppa… (vs the ubuntu repo)?
09:49 jjasinski__ (sorry for my confusion… wasn't sure you were indicating that Corey is updating the ppa archive with a saucy 0.17 release or if he is updating the ubuntu repository with a 0.16 release update…not sure how that all works)
09:50 joehh sorry - I wasn't celar
09:51 joehh clear
09:51 joehh saucy has 0.16.0 in it "officially". This came from the debian packages when saucy was "split off" from debian (or whatever the technical term is)...
09:52 joehh Corey uploaded a 0.17.1 to the ppa roughly 24 hours ago, but a minor error prevented the build
09:52 networkpadawan joined #salt
09:52 joehh https://launchpadlibrarian.net/156026984/buildlog_ubuntu-saucy-i386.salt_0.17.1-2saucy_FAILEDTOBUILD.txt.gz
09:53 joehh looks like a version mismatch in building the man page
09:54 jjasinski__ Cool - thanks for that clarification… I'll give that new version a try when it's available.
09:56 Minioncito joined #salt
09:56 Minioncito Hi there
09:59 Minioncito jcockhren: hi, just in case you want to know; yesterday I was talking to you about the troubles when matching minions from masterofmasters using syndic. I opened an issue (https://github.com/saltstack/salt/issues/8322)
09:59 Minioncito jcockhren: and although it was duplicated, it has been fixed hours ago! https://github.com/saltstack/salt/pull/8335
10:07 joehh jjasinski__: just copied the raring package to saucy, let me know if you have any issues
10:08 IJNX how do I get the npm module in use?
10:11 tty-1 joined #salt
10:11 prooty joined #salt
10:12 prooty hello. how can i be sure that the opts parameter to the hg.latest state is being passed? they don't seem to be included in the hg command when looking at the minion in debug mode. using salt 0.17.0.
10:13 nocturn_ joined #salt
10:13 bhosmer joined #salt
10:15 IJNX I keep getting: '    "npm.list" is not available.'
10:18 oz_akan_ joined #salt
10:20 hvn @lJNX: have you installed nodejs? http://docs.saltstack.com/ref/states/all/salt.states.npm.html#module-salt.states.npm
10:24 matanya joined #salt
10:26 APLU joined #salt
10:29 IJNX hvm, should be
10:32 IJNX I can run npm in the minion machine fine from cmdline
10:33 IJNX e.g. salt '*' cmd.run "npm"
10:37 IJNX it doesn't make sense to use salt for running just my .sh script with salt prefix
10:37 IJNX "        Comment:   Package npm is already installed"
10:37 matanya joined #salt
10:40 hvn @ljNX sorry, I cannot help on this because I have never used that state/module
10:47 IJNX hvn: I'm just starting to use salt and I already ran into these simple setup issues. How stable project salt is?
10:50 hvn @ljNX: since I still here, not in chef/puppet or other, it mean it good enough for me for my use case :P
10:50 IJNX hvn: ok
10:51 IJNX I'll try to keep reading the docs again...
10:58 carlos joined #salt
10:58 IJNX odd thing is that salt '*' npm -d gives me the documentation of the npm module
10:59 IJNX but when I run any command like npm.list, it reports that it's not available
10:59 IJNX I have run sync_all
11:00 joehh IJNX: what version of npm do you have?
11:00 joehh npm --version
11:00 IJNX 1.1.4
11:00 joehh that will be it
11:00 joehh according to https://github.com/saltstack/salt/blob/develop/salt/modules/npm.py#L33
11:00 joehh it needs to be atleast 1.2
11:01 IJNX hmm, that's the latest that ubuntu server has
11:01 joehh and checking npm.list, https://github.com/saltstack/salt/blob/develop/salt/modules/npm.py#L153
11:01 joehh the first thing it does is check for a valid version
11:02 hvn it should check that in __virtual__
11:03 joehh hvn: that would make more sense, each function checks independantly
11:03 IJNX great, It seems I need to install npm from sources...
11:04 joehh https://launchpad.net/ubuntu/+source/npm
11:04 joehh saucy and newer have newer versions
11:04 hvn I mean if the version does not suffice to use that module, that module should return false than behave weird
11:04 joehh maybe you could try backporting...
11:05 IJNX https://gist.github.com/dwayne/2983873
11:06 joehh looks simple enough
11:08 bhosmer joined #salt
11:10 IJNX Now I need to figure out how replace the pkg. installed with this one
11:12 joehh dpkg --purge ?
11:12 josephholsten joined #salt
11:13 creich hi, is there some convenient method to get the result of the last finished job
11:13 IJNX joehh: yes, that I know. But I was meaning how to write a sls file that does the compile for me.
11:16 ollins joined #salt
11:16 joehh IJNX: Ahhhhhh - much more interesting :)
11:16 joehh I'm guessing a cmd.wait
11:17 IJNX maybe, I'm not that far in docs yet :)
11:18 IJNX some reference to file stored in salt fileserver and then the execution or it from npm.sls
11:18 oz_akan_ joined #salt
11:21 viq IJNX: don't even have to be on salt server, you can give http url and checksum
11:25 viq Something like https://gist.github.com/viq/7369690
11:28 gildegoma joined #salt
11:28 IJNX viq: great, thanks
11:29 prooty hello. is there a way to configure nginx to watch the /etc/nginx/conf.d directory for changes and reload nginx if there are?
11:30 ricicle joined #salt
11:31 prooty i know i could do something like a watch on a file inside the directory, but this requires that i know all files to watch ahead of time.
11:31 viq prooty: I guess combination of file.recurse and service with watch
11:32 ricicle Is there any way to specify multiple requirements.txt file to the same virtualenv state please?
11:32 viq prooty: no, watch state
11:32 viq prooty: http://docs.saltstack.com/ref/states/all/salt.states.service.html#module-salt.states.service
11:35 viq prooty: and https://salt.readthedocs.org/en/latest/ref/states/all/salt.states.file.html#salt.states.file.recurse
11:38 networkpadawan joined #salt
11:44 gw prooty: would a watch_in work for you? e.g. every file that needs to trigger a nginx reload would need to have this option.
11:46 gw prooty: or to be more precise ... every state that should trigger a nginx reload on change
11:48 prooty viq, file.recurse would probably work but what if i template some files? i understand that it just copies the directory contents.
11:49 viq prooty: wrong ;)
11:49 viq https://salt.readthedocs.org/en/latest/ref/states/all/salt.states.file.html#salt.states.file.recurse
11:50 prooty gw, that is what i want actually. to push nginx to reload rather than being watched.
11:50 prooty viq, oh i see. should have read better.
11:51 viq prooty: either way it will be salt that will tell nginx to reload, it's just where will you define the watch part
11:53 JasonG_TA joined #salt
11:56 IJNX some like this: https://gist.github.com/ilkkanisula/7369980
11:57 joehh IJNX: looks like a good start
11:57 joehh I think your cmd.run lines should be cmd.wait
11:58 joehh and they should watch something relevant
11:58 joehh node-compile/node-config are the ones
11:58 IJNX yep
11:59 IJNX is it the watch that cmd.wait waits for?
11:59 prooty thanks, file.recurse would probably work well in my case.
12:00 joehh yes
12:01 joehh so the cwd of node-config should be in the /tmp/node-v0.10.21
12:01 joehh directory
12:01 joehh you will not be able to watch /tmp/node-v0.10.21, you need to watch one of the labels (ie the top level bits that end with :)
12:02 IJNX updated gist
12:02 joehh so watch /tmp/node.tar.gz
12:02 rofl____ is the schedule module supposed to work with state.highstate from the salt-master?
12:03 joehh looks like it should work (not knowing anything about node or the paths in particular), but the structure looks fine
12:03 IJNX hmm, now I need to figure out how to run it?
12:03 joehh maybe your cwd s need checking
12:03 IJNX highstate does nothing
12:03 joehh to get high state to do it, it will need ot be in top.sls
12:04 joehh otherwise salt 'minion' state.sls npm
12:04 joehh no need for .sls extension at end (number one mistake for our admins)
12:05 IJNX my top has: base: '*': - npm in it
12:05 joehh and your minion's key has been added?
12:06 joehh you can test.ping it?
12:06 joehh silly question - you were doing more before
12:06 joehh :)
12:06 IJNX (updated to gist)
12:06 IJNX ping works fine :)
12:06 cron0 joined #salt
12:07 IJNX this is all I do now with salt
12:07 IJNX all I have is one top.sls and npm.sls (as in gist)
12:07 joehh trying it at my end now
12:08 joehh in /srv/salt/
12:08 joehh ie /srv/salt/top.sls
12:08 IJNX yep
12:08 IJNX and _modules/npm.py
12:08 joehh and /srv/salt/npm.sls
12:08 joehh not sure why you need the _modules one, but maybe you have a good reason
12:09 IJNX See gist
12:09 IJNX propably I don't need it. Its remaining there after my tries to get the npm to work in first place.
12:12 IJNX some problem with yaml syntax… (AttributeError: 'OrderedDict' object has no attribute 'append'
12:12 IJNX )
12:13 IJNX hmm, it's probably the pkg: twice in npm.
12:13 joehh beat me too it
12:13 logix812 joined #salt
12:13 IJNX pkgs: -?
12:13 joehh as you are using http to get the file, you will need to provide a hash
12:14 joehh see http://docs.saltstack.com/ref/states/all/salt.states.pkg.html#salt.states.pkg.installed
12:17 joehh you'll need - source_hash: md5=774bc17875c042ac42ad2d96b9c96734 for the file.managed
12:18 IJNX updated
12:19 IJNX still the same orderDict error… something wrong with some keywords
12:19 joehh I'm not seeing that one
12:20 joehh I have http://pastebin.com/TRkZpmJS
12:21 joehh pkg.installed rather than require
12:21 xl1 joined #salt
12:25 aleszoulek joined #salt
12:30 ollins joined #salt
12:32 erSitzt joined #salt
12:33 erSitzt hi, is it possible to apply a state only if a specific service is not running ? if yes, how can i achieve that ? thx
12:33 matanya joined #salt
12:37 faust joined #salt
12:39 claudep joined #salt
12:40 Minioncito erSitzt: this one?  salt.states.service.dead(name, enable=None, sig=None, **kwargs)
12:40 Minioncito 'Ensure that the named service is dead by stopping the service if it is running'
12:40 rofl____ anyone else experienced minions eating alot of memory when the master is gone?
12:41 sgviking joined #salt
12:42 matanya joined #salt
12:45 erSitzt Minioncito: but that only ensures the service is not running, does it ? salt would then stop the service and reinstall the service. i'm trying to manually install a service in linux, but only if that service does not exist... and i dont want to reinstall it when the state is applied again.
12:46 erSitzt is there a way to check for services/processes using grains ?
12:47 matanya joined #salt
12:49 claudep left #salt
12:49 viq Having groups defined like so: http://pbot.rmdir.de/ch22uvc6SnII12uhdHBCSQ what would be a good approach to make sure they all are created?
12:50 viq erSitzt: no, grains are considered to be static information
12:52 erSitzt ok
12:53 viq erSitzt: you could write a custom grain... but it would have information from the time point when minion was started, they would not get updated unless specifically told to
12:53 erSitzt can i manually set a grain with a state ?
12:54 viq probably
12:54 erSitzt eg. when i successfully installed the service set a grain
12:54 viq What do you mean, "installed a service" ?
12:54 gkze joined #salt
12:55 erSitzt i have to copy a tgz to a linux box, extract it, and manually install it
12:55 viq That's not service, that's a program
12:55 erSitzt its a service
12:55 erSitzt a daemon
12:56 viq erSitzt: maybe this will give you ideas https://gist.github.com/viq/7369690
12:57 joehh IJNX: I would move your building down a directory is a good idea
12:57 joehh and maybe doing it somewhere more permanent than tmp
12:57 erSitzt the unless part is just a command  ?
12:57 joehh so it doesn't get rebuilt each time after  a reboot
12:59 viq erSitzt: yes, there's also onlyif, they kind of work in reverse, one is "do this only if command returns true", other is "don't do this if command returns true"
13:00 erSitzt nice
13:00 erSitzt i'll try that
13:00 erSitzt thx
13:00 viq Any idea how to create all groups defined like this http://pbot.rmdir.de/ch22uvc6SnII12uhdHBCSQ ?
13:02 xl1 left #salt
13:04 matanya joined #salt
13:05 mwillhite joined #salt
13:05 IJNX I get the OrderDict now
13:05 IJNX …error I mean
13:06 IJNX It's something new in python 2.7
13:06 slav0nic joined #salt
13:07 ninkotech joined #salt
13:07 mwmnj joined #salt
13:07 IJNX http://pastebin.com/gDW1DfBb
13:08 zandy joined #salt
13:10 joehh IJNX: I think it is the require at line 2
13:11 joehh I've got pkg.installed there
13:11 IJNX joehh: I have the pkg.installed there already. didn't change behavior
13:12 joehh can you update your gist?
13:12 IJNX https://gist.github.com/ilkkanisula/7369980
13:12 srage joined #salt
13:13 elfixit joined #salt
13:13 joehh - needed in front of pkgs
13:14 joehh also, I have a different md5 sum?
13:14 ipmb joined #salt
13:19 IJNX great, it seems to do something
13:20 oz_akan_ joined #salt
13:20 joehh I think the challenge remaining is to either get it to install on the path ie into / rather than /usr/local or adjust the path of your minion to include usr/local/bin
13:21 brianhicks joined #salt
13:22 oz_akan_ joined #salt
13:23 mdasilva joined #salt
13:24 IJNX that's just --prefix=/
13:25 IJNX is there a way to minitor minion from master?
13:25 IJNX *monitor
13:27 tty-1 joined #salt
13:29 joehh not that I am aware of
13:29 NV define monitor?
13:29 joehh I ran it like salt -t 6000 'miniion' state.sls npm
13:29 IJNX seeing the progress of state execution
13:30 joehh that meant it would wait long enough to see status
13:31 NV not until the state finishes i don't think
13:31 IJNX would it be possible to put that -t 60000 inside the sls file somehow?
13:31 NV ssh + salt-call may work though
13:32 IJNX SInce this is expected to take longer due to compilation process and downloads
13:33 EWDurbin joined #salt
13:35 mbrgm joined #salt
13:35 EWDurbin terminalimage: looks like you've been maintaining the fedora/EPEL builds for salt since 0.16.4
13:35 mbrgm hi! can someone tell me how to use the connection_args with mysql_
13:35 mbrgm hi! can someone tell me how to use the connection_args with mysql_* states?
13:35 mbrgm i.e. specify a user/password?
13:36 EWDurbin mbrgm: have you followed http://docs.saltstack.com/ref/modules/all/salt.modules.mysql.html#module-salt.modules.mysql
13:36 mbrgm EWDurbin: oh! sry, just looked at the state docs
13:37 mbrgm thx!
13:37 EWDurbin no problem
13:37 EWDurbin terminalimage: i'd like to discuss the possibility of a yum repo just for salt releases when you have time
13:38 EWDurbin i've been burned by a few versions out of EPEL, and got stuck self-hosting older versions
13:38 EWDurbin EPEL doesn't keep even one version back, so yum downgrades aren't an option
13:39 scott_w joined #salt
13:39 EWDurbin so it generally means headaches until a bugfix release to EPEL, or self-hosting a yum repo with the previous version (0.16.4 in this case)
13:41 gw EWDurbin: I had the same problem. Worked around it by using salt-bootstrap and specifying specific version numbers. However there is 0.17.1 now in peel which worked fine so far
13:41 gw peel -> epel
13:42 EWDurbin gw: yeah there's a nasty rendering bug in 0.17.1 for my use case :( https://gist.github.com/ewdurbin/7235624
13:42 EWDurbin it's fixed on develop
13:42 EWDurbin just sittin on my hands waiting for a release
13:42 anitak joined #salt
13:43 scott_w EWDurbin: be careful, that method leads to pins and needles ;)
13:43 EWDurbin scott_w: elaborate?
13:43 scott_w sitting on your hands
13:43 EWDurbin haha
13:43 EWDurbin i thought you were talking about my gist
13:43 scott_w ah, haha
13:44 scott_w parasthesia is the term, apparently
13:44 EWDurbin for now i think i'm just going to track all packages on koji and set up a mirror of sorts
13:45 bhosmer joined #salt
13:47 bhosmer joined #salt
13:48 bhosmer joined #salt
13:49 bhosmer joined #salt
13:53 srage joined #salt
13:55 khodgson joined #salt
13:55 hvn joined #salt
13:56 gw EWDurbin: I might be mistaken, but shouldn't {{ pillar['sudoer_groups'] }} rather be something like {{ pillar['sudoer_groups'].keys()|join(',') }}
13:56 gw and in your salt-sudoers.jinja you would split(',') the sounders list again?
13:57 gw sounders -> sudoers
13:57 EWDurbin gw: that certainly sounds like it would be safer
13:58 EWDurbin i could just pass the comma delimited list in via the context and split it up again
13:58 gw should work as well .. yes :)
13:58 aptiko I'd like to execute a cmd.run (/etc/init.d/iptables-persistent save) whenever ANY iptables.* state changes. Is there any way to watch all states of a specific type?
14:00 EWDurbin gw: quick check says yes "groupadd: 'foo,bar' is not a valid group name"
14:00 khodgson joined #salt
14:00 EWDurbin so as long as commas aren't allowed in usernames and groups :)
14:01 * EWDurbin digs up some RFCs
14:01 _no2be joined #salt
14:01 gw :) ... good check, I just assumed they are not allowed :)
14:02 gw so you can get off your hands again? :)
14:02 _no2be Hi is anybody using salt-ssh ?
14:02 scott_w joined #salt
14:03 faldridge joined #salt
14:03 jefferai joined #salt
14:04 _no2be I have set up roster files and when I pass salt-ssh -r "uname -a" it works but when I pass salt.ssh cmd.run "uname -a" I get error messages
14:04 godog joined #salt
14:04 godog joined #salt
14:05 khodgson joined #salt
14:05 jslatts joined #salt
14:08 mzito joined #salt
14:08 _no2be It does not matter if salt-minion is installed or not .. Systems in question are all Linux a mix of Centos 6 and ubuntu 12.04
14:09 scott_w joined #salt
14:09 gw EWDurbin: I am also not sure whether this is a renderer bug. I kinda would expect this behaviour as it is. The Jinja template is rendered to a string which is then parsed as yaml.
14:10 gw EWDurbin: wit jinja template I meant your sudoers.sls
14:11 mdasilva1 joined #salt
14:11 Brew joined #salt
14:12 IJNX_ joined #salt
14:13 VSpike joined #salt
14:13 khodgson joined #salt
14:15 juicer2 joined #salt
14:18 EWDurbin gw: when i first brought the bug up, i was asked to test on develop and it worked...
14:19 EWDurbin am i wrong to consider it a regression if 0.16.4 handled it correctly and 0.17.1 didnt?
14:20 oz_akan_ joined #salt
14:20 mapu joined #salt
14:22 mwillhite joined #salt
14:23 khodgson joined #salt
14:23 oz_akan_ joined #salt
14:23 IJNX joehh: nice, it seems to work fine. only thing now is that npm.list doesn't work. let's see if --prefix=/ helped… (compilation takes quite long time)
14:24 aptiko Version 0.17.1 took an array from the pillar and made it a string, like this: "[ ... ]". Is this a bug or a feature?
14:24 VSpike joined #salt
14:24 Khollowa_ joined #salt
14:25 aptiko It was working before.
14:28 linuturk joined #salt
14:28 scott_w joined #salt
14:29 viq Any idea how to create all groups defined like this http://pbot.rmdir.de/ch22uvc6SnII12uhdHBCSQ ?
14:30 creich viq,  whats your problem with that?
14:30 georgj05 joined #salt
14:31 viq creich: that I don't know how to do two level iteration
14:32 mbrgm hey! what's the problem with this part of code? http://pastebin.com/ft2uTEAh
14:32 mbrgm it says that there is an undefined jinja variable on the - host: line
14:32 creich viq, maybe you should considre to put the group1/group3 outside of the if-statement
14:34 viq creich: but that's the whole point of the exercise - users get different groups depending what machine their on
14:34 creich viq, maybe like that: http://pbot.rmdir.de/FxA1KfjSdo3BRoo3AYEotg
14:35 viq mbrgm: try http://pastebin.com/zNdpEt6z
14:35 linjan_ joined #salt
14:35 creich viq, yeah i think i get what you want, just could not explain my idea ... so see pastebin pls ;)
14:35 viq creich: looking
14:35 Teknix joined #salt
14:35 creich viq, maybe like that: http://pbot.rmdir.de/FxA1KfjSdo3BRoo3AYEotg
14:36 mbrgm viq: thx
14:39 khodgson joined #salt
14:39 mgw joined #salt
14:40 danielbachhuber joined #salt
14:40 bhosmer_ joined #salt
14:41 bhosmer joined #salt
14:42 cnelsonsic joined #salt
14:42 austin987 joined #salt
14:43 IJNX i give up. 'npm.list' is not available
14:43 viq creich: sure, but how do I _access_ that data
14:44 viq because https://github.com/viq/cm-lab-salt/blob/users/salt/roots/salt/users/groups.sls does not quite work
14:46 creich viq, hm i'll have a look in  a few minutes...
14:46 elfixit joined #salt
14:48 petersabaini left #salt
14:49 aptiko Is salt.states.iptables working at all? I get very strange errors.
14:50 dccc joined #salt
14:50 aptiko Executing command 'iptables -t filter -A INPUT --proto icmp --jump ACCEPT --__env__ ba
14:50 aptiko se --__sls__ enhydris --order 10045 '
14:51 hazzadous joined #salt
14:51 aptiko --__env__, --__sls__, and --order shouldn't be there.
14:53 viq aptiko: which version, what's your state?
14:55 aptiko viq: 0.17.1
14:55 aptiko viq: I see that this has been subsequently fixed.
14:56 viq there you go then ;)
14:57 ipmb joined #salt
14:57 tyler-baker joined #salt
14:57 tyler-baker joined #salt
15:01 bhosmer joined #salt
15:03 creich viq, so i managed to get it work
15:03 creich w8 i paste-bin it
15:03 viq creich: sweet!
15:05 creich viq, http://pastebin.com/rPa1vRG5
15:05 creich so i used two files for that
15:05 viq looking
15:05 creich maybe it is not really necessarry
15:05 carmony Good morning #salt :)
15:05 creich but for now it works
15:06 creich carmony, hi :)
15:06 pdayton joined #salt
15:07 NotreDev joined #salt
15:17 austin987 joined #salt
15:18 viq creich: I tested it, it works, but it's not quite what I wanted to do, or I can't quite figure out how to get from here to what I want to do
15:18 creich hmm but at least you can access the data .. so one step after another
15:18 creich maybe i din't get what you want
15:18 creich could you explain your idea again
15:19 creich as far as i understood, you wanted to insert different content into some file regarding to the os us are using
15:20 viq I am trying to do quite a lot with pillars, which may be sub-optimal and troublesome. I'll explain my way of thinking, you can follow what I currently have here https://github.com/viq/cm-lab-salt/tree/users/salt/roots/salt/users
15:22 viq I want to define a couple of pillars with groups, erm, "crowds" to avoid confusion, of people. I want to define what system groups each "crowd" is assigned in a different pillar, based on machine, OS, whatever. And I want to write states, that will make sure all the relevant groups are created, and create users from appropriate crown belonging to appropriate group(s)
15:25 tyler-baker joined #salt
15:25 tyler-baker joined #salt
15:27 JasonG_TA joined #salt
15:27 viq creich: ^
15:28 creich viq, sry..still working but i've allready seen your repository...
15:28 creich i'll try to help you as soon as possible
15:29 viq creich: sure, no problem. I realize it's quite a complex thing to look into, so I certainly don't expect immediate answer, you looking at it at all is appreciated :)
15:29 creich :)
15:31 JasonG_TA joined #salt
15:31 Kareem joined #salt
15:31 higgs001 joined #salt
15:32 bhosmer joined #salt
15:33 forrest joined #salt
15:35 mwmnj joined #salt
15:35 mwmnj joined #salt
15:36 mwmnj joined #salt
15:36 mwmnj hello
15:36 EWDurbin gw: thanks for the workaround suggestion earlier... it works well. now i have to fix all the other places i was using the same methodology :-/
15:38 mwmnj I am just getting started with SaltStack and I am running into this error when trying to use the python client api:
15:38 * mwmnj Failed to authenticate, is this user permitted to execute commands?
15:38 mwmnj client = salt.client.LocalClient()
15:38 mwmnj ret = client.cmd('*', 'cmd.run', ['whoami'])
15:39 mwmnj I am the salt master and minion set up fine
15:39 mwmnj and can run the commands from the cli
15:39 mwmnj Is there some permissions I need to set on the python script?
15:39 ipmb joined #salt
15:40 viq mwmnj: as what user are you executing the python script?
15:40 scott_w joined #salt
15:42 tty-1_ joined #salt
15:42 mwmnj viq: pretty sure the default 'ubuntu' user amazon ec2 creates
15:42 viq mwmnj: try as root
15:43 mwmnj its actually a django app I am using the client in
15:43 mwmnj Would i just sudo python manage.py runserver to run as root?
15:43 viq mwmnj: yes
15:43 viq mwmnj: mind you, not a very good idea to run in production, but something to start with
15:43 forrest mwmnj, if you want to try and run salt as the non root user, you can review: http://docs.saltstack.com/topics/nonroot.html
15:46 mwmnj forrest: ok, so if I chown those directories to my nonroot user I am runing the django runserver with, it should work?
15:47 viq mwmnj: you need to run salt as that user as well
15:47 forrest yea viq has it right
15:47 mwmnj ah
15:48 mwmnj ok, let me try that, since running $ sudo python manage.py doesn't seem to include my virtual environment and the apps depenencies
15:50 mzito joined #salt
15:51 ricicle joined #salt
15:53 cachedout joined #salt
15:58 kermit joined #salt
15:59 khodgson joined #salt
16:03 bhosmer joined #salt
16:03 terminalmage joined #salt
16:03 jalbretsen joined #salt
16:05 rawzone joined #salt
16:05 Kareem joined #salt
16:06 pipps_ joined #salt
16:08 khodgson joined #salt
16:10 redondos joined #salt
16:10 redondos joined #salt
16:13 tty-1_ joined #salt
16:14 rjc joined #salt
16:15 ipmb joined #salt
16:15 mgw joined #salt
16:17 bretep[hirevue] joined #salt
16:19 nineteeneightd joined #salt
16:20 nineteeneightd joined #salt
16:23 mgw joined #salt
16:24 druonysus joined #salt
16:24 druonysus joined #salt
16:29 faldridge joined #salt
16:29 seanz Good morning, humans. I'm getting an error: "recursive requisite found". I'm not entirely sure what that means.
16:30 seanz Any advice?
16:30 bretep[hirevue] @all anyone know of a good way to store RSA keys in pillars? If you have any other suggestions, those are welcomed as well.
16:30 lineman60 joined #salt
16:30 hvn @seanz : it means A require/watach B, then B requrie/watch A
16:31 ggoZ joined #salt
16:31 forrest bretep, https://gist.github.com/UtahDave/3785738
16:31 forrest That stores passes, so you might be able to do something similar
16:31 foxx joined #salt
16:32 seanz bretep: We store keys similar to UtahDave's example. The public key is a one-liner assigned to a variable, and the private key uses the "|" pipe feature to make it a multiline string.
16:33 seanz hvn: Ah. Hm. I wonder how that's happening. Apache is watching a file that is watching a step once-removed from Apache.
16:33 bitz joined #salt
16:33 forrest seanz, if it's just one state toss it in a gist and we can take a look
16:34 seanz forrest: https://gist.github.com/szicari-cars/16e0d436c0babed9a174
16:34 bretep seanz Do you have an example you can post. That sounds like what I want to do
16:34 seanz bretep: Sure. Hold on.
16:34 forrest what's in apache seanz?
16:35 hvn can you guys show me where in the doc that have "sls: something" ?
16:35 seanz forrest: https://gist.github.com/szicari-cars/6c3601d169adf7b264a6
16:35 jkleckner joined #salt
16:35 pipps joined #salt
16:35 seanz bretep: Working on the example.
16:36 seanz hvn: I don't know where in the docs, but I know it tells salt to wait until the entire sls has been executed.
16:38 bemehow joined #salt
16:38 seanz bretep: https://gist.github.com/szicari-cars/86803e3da525b314b2d9
16:38 seanz bretep: The keys have been changed, obviously. Those keys are too short and are generally "whack".
16:38 bemehow joined #salt
16:39 bitz joined #salt
16:40 hvn @seanz: if there is something like that, I think it cause recurse require. Order run:
16:40 hvn SLS (apache service) -> config file
16:40 hvn but in your extend: SLS watch config file mean Config file -> apache service
16:41 forrest damn you hvn, typed it up faster than me, I was thinking the same thing.
16:41 jkleckner1 joined #salt
16:41 seanz hvn, forrest: Ah, I see what you're saying.
16:41 seanz Let me rearrange that a bit.
16:42 hvn I only read salt doc since 0.11, I don't know there is something like that. Hope that you guys can show me, because I searched but not found a result
16:43 forrest hvn, can you expand on what you mean?
16:43 bretep thanks seanz!
16:43 forrest I don't get 'sls: something'
16:44 pentabular joined #salt
16:44 seanz In this case, I changed it from sls: apache to pkg: apache-server, and now it works.
16:44 seanz You guys are good.
16:45 hvn @forrest: from what @seanz wrote "seanz: hvn: I don't know where in the docs, but I know it tells salt to wait until the entire sls has been executed."
16:45 forrest seanz, you should also take a look at: https://github.com/saltstack-formulas/apache-formula/tree/master/apache
16:45 forrest hvn, oh ok I thought you were asking a question
16:46 hvn @forrest: yes, I hope to know where in the doc have that example - which using "require: -  sls: something"
16:47 forrest http://docs.saltstack.com/ref/states/requisites.html#require-an-entire-sls-file
16:47 pentabular left #salt
16:47 seanz forrest: This is genius!
16:48 forrest seanz, yea formulas are amazing
16:48 seanz I love the mapping much better than the inline conditionals.
16:48 seanz forrest: Though what is this "formula" term you're using?
16:48 forrest Yeah if you need a simpler example: https://github.com/saltstack-formulas/mysql-formula/tree/master/mysql
16:48 seanz Is this a new thing in salt?
16:48 forrest it's a few months old
16:48 hvn @forrest: thanks
16:48 forrest so formulas are basically pre-built 'chunks' so you can check them out, modify the pillar example to incorporate it into your own, and then it should be good to go regardless of distro
16:49 forrest hvn, np
16:49 scott_w joined #salt
16:49 forrest seanz, check out: http://docs.saltstack.com/topics/conventions/formulas.html
16:49 seanz forrest: Ah, so formulas make heavy use of pillar, I'm guessing.
16:49 forrest sometimes
16:49 seanz Modifiable through pillar ?
16:49 bretep I'm in the process of migrating our chef to salt... We use encrytped databags because we store them in our git repo. Any suggestions on how I can achive this with pillars?
16:50 hvn btw, that kind of require "sls" just make a SLS hard to extend (as above example show)
16:50 bretep We have requirements around this from our customers and go through periodic, monthly, audits.
16:50 forrest hvn, yes it can make things a bit annoying, but it's also nice sometimes.
16:50 forrest seanz, so the pillar data a lot of the time is just 'base' information
16:50 seanz hvn: There are some cases where it's useful. We also use it to ensure that our deployment keys are all deployed prior to running certain other states.
16:50 forrest for the formulas that is.
16:50 seanz hvn: Though you're right that it can get messy.
16:51 teskew joined #salt
16:51 racooper joined #salt
16:51 seanz forrest: Hm, a lot to learn here. The use of filter_by.
16:51 seanz I'm new to all that better stuff.
16:51 forrest hvn, It depends on how you use them as well, for example https://github.com/gravyboat/hungryadmin-sls/blob/master/salt/hungryadmin/app.sls
16:51 forrest they're really easy there.
16:52 forrest seanz, yea the formulas are still pretty 'new' in the grand scheme of things, I just saw your apache state and figured you'd be interested
16:52 forrest because it looks friggin sexy when you use the map file
16:52 seanz forrest: Definitely.
16:52 pipps_ joined #salt
16:52 bhosmer joined #salt
16:53 hvn @seanz: I cannot file "sls: " in that file :p
16:54 xmltok joined #salt
16:54 seanz forrest: What is "watch_in"?
16:54 mzito joined #salt
16:56 seanz Also, how does salt differentiate a function/method call as opposed to the arguments? That's one thing I don't necessarily like, that method names are mixed in with the arguments in salt states.
16:56 forrest seanz, so watch_in is equivalent to saying 'in item x (apache), make sure this item is watched'
16:57 seanz Ah, so the reverse of watch.
16:57 forrest it's awesome for when you have a ton of files, instead of filling apache with watch: a-z, you just say in each of those items watch_in: apache
16:57 forrest pretty much yea
16:57 forrest What do you mean with differentiating between a function/method call as opposed to the arguments?
16:58 forrest do you have an example?
16:58 seanz Well, service.running: as opposed to service:
16:58 seanz - running
16:58 forrest ok
16:58 seanz - other: arg
16:58 seanz Etc.
16:58 forrest yea so I never use service.running, or virtualenv.managed, etc.
16:58 seanz Why, though? It keeps the functions separate from the arguments.
16:58 forrest it's just my personal preference.
16:58 seanz Ah.
16:59 seanz Do you know how salt handles that during parsing?
16:59 seanz Does it check every arg to see if it's a method/function call?
16:59 forrest sometimes YAML is weird, and if you mix service.running, but then do virtualenv: \n - managed it can have weird issues.
16:59 forrest seanz, I don't know if it does every check, I always figured it had to do with the usage of the ':'
17:00 forrest hey basepi are you around?
17:00 NotreDev trying to figure out how pip.installed and virtualenv.managed go together. do i need to provide the pip state a bin_env?
17:00 forrest NotreDev you don't have to, it should work with the system pip if it's installed. I've alwaysu sed one though
17:01 redondos joined #salt
17:01 redondos joined #salt
17:01 forrest seanz, but to be totally honest with you I don't know how exactly it figures that out during parsing.
17:01 NotreDev forrest: i'm trying to put my different python projects in their own virtualenvs.
17:01 forrest NotreDev, check out https://github.com/gravyboat/hungryadmin-sls/blob/master/salt/hungryadmin/app.sls#L66
17:01 jefferai joined #salt
17:02 forrest I use a virtualenv there for my environments
17:02 forrest well environment
17:02 forrest but same concept
17:02 seanz forrest: I just figured it's got to check all args because I *think* you can have the method name anywhere in the arg list. I could be wrong.
17:02 seanz I'm pretty new to salt.
17:02 seanz So I'm going to retract that statement until I know for sure.
17:02 NotreDev k
17:02 NotreDev forrest: thanks
17:03 forrest seanz, Yea, I'm pretty sure it just has to do with how salt treats specific items like the - and :, but I haven't dropped into that code enough to check
17:03 forrest NotreDev, np
17:03 forrest NotreDev, https://github.com/terminalmage/djangocon2013-sls also uses some virtualenv stuff.
17:03 NotreDev forrest: and {{ hungryadmin_venv }} is just the path to the root of the venv?
17:04 NotreDev i don't have access to the pillar data
17:04 forrest I have the pillar data a few dirs up: https://github.com/gravyboat/hungryadmin-sls/blob/master/pillar/hungryadmin.sls
17:04 forrest but yes, that's just the path.
17:04 troyready joined #salt
17:08 NotreDev forrest: thanks agian
17:08 forrest NotreDev, yea np
17:09 KyleG joined #salt
17:09 KyleG joined #salt
17:12 londo joined #salt
17:14 bemehow_ joined #salt
17:17 anuvrat joined #salt
17:18 avienu joined #salt
17:22 shinylasers joined #salt
17:24 basepi forrest: yep, sorry, was afk for a minute
17:24 lineman60 joined #salt
17:25 ctdawe joined #salt
17:25 forrest basepi, no worries. Can you explain how salt handles the parsing of the YAML files? As in how it checks to see whether something is a method/function call? Does it check every single item? I thought it had to do with identifying the usage of '-' and ':', but I'm not sure.
17:26 forrest I can dive into the code if you don't have time, I just don't want to if there's a quick yes/no answer :P
17:27 basepi forrest: i guess i don't understand the question.  YAML has a defined syntax and just parses the files into a data structure
17:27 basepi and then we interpret that data structure in the state compiler
17:27 forrest basepi, ok that makes sense.
17:27 forrest seanz, you see the response regarding the question from earlier?
17:28 seanz forrest: Thanks for getting my attention.
17:28 seanz basepi: I was just curious:
17:28 basepi which is where our input flexibility comes from.  any input can be used, as long as we end up with those data structures.  thus we can use PyDSL and other input methods
17:28 forrest Gotcha
17:28 seanz How does salt figure out which argument is the method to be called vs the arguments for that method.
17:28 forrest From the YAML data structure I imagine seanz
17:29 basepi seanz: it's formed as a list.  the `- ` in front of each argument is YAML's list syntax
17:29 seanz basepi: Right...but then is each item inspected and tested to see if it's a method call? I guess I just need to dive in and look.
17:29 seanz It *is* open source, after all.
17:29 basepi so we end up with a dictionary (`pkg` might be the key, and then we have a list of args `['installed', {'name': 'nginx'}]`
17:30 basepi as the value
17:30 mapu joined #salt
17:32 bhosmer_ joined #salt
17:35 pentabular joined #salt
17:35 pentabular left #salt
17:36 tty-1 joined #salt
17:38 Thiggy joined #salt
17:39 tempspace Hello all
17:39 forrest hey
17:43 ebodine joined #salt
17:44 quickdry21 joined #salt
17:45 cdcalef joined #salt
17:55 NotreDev is my workflow f'ed? my only way to debug an error is to manually go line by line and delete stanzas until the project compiles. This is about as good as I get in terms of errors: SaltRenderError: Undefined jinja variable; line 350 in template
17:55 NotreDev which means absolutely nothing
17:57 andersb I just had the same problem today... It was a pillar I had not defined. So kinda hard to find with that note.
17:58 andersb Not saying it is, but that was what happend to me, and I had to paste in stanzas util it would _not_ compile :)
17:59 khodgson joined #salt
17:59 tvs_ joined #salt
17:59 andersb The line it pointed to originally in the error was nothing wrong with. Made even less sense when I commented the stanza out and it was still throwing the error :)
18:00 * Gareth waves
18:01 tvs_ I have defined some settings in /srv/pillar/appserver/app_setting.sls but they don't appear when I execute 'salt '*' pillar.items' ... what could be the problem?
18:02 xmltok is it included in the top.sls?
18:02 xmltok notredev: i have begun to use pydsl and i don't think i can go back to jinja+yaml
18:03 xmltok perhaps for people that dont know python the jinja+yaml is easier, but the pydsl is much more intuitive, easier to read, and powerful imo
18:03 KyleG left #salt
18:03 NotreDev xmltok: pydsl is vanilla python?
18:04 xmltok its close, you can use vanilla python but then you use state() to create your states, see http://docs.saltstack.com/ref/renderers/all/salt.renderers.pydsl.html
18:04 cdcalef i've got some fabric scripts i need to port into salt i wonder if the pydsl would be a good avenue for them or if i should just make them modules..
18:04 NotreDev nice
18:05 xmltok notredev: loops read a lot easier to me too, and you can do stuff like this to pull in parameters from your pillars http://pastebin.com/PAqvBXzT
18:05 higgs001 joined #salt
18:06 tvs_ \xmltok that was the problem, thanks!
18:07 mr_chris Installing salt-minon from epel CentOS 5 results in salt with zeromq2. Obviously this, cause problems.I attempted install zeromq 3 from the repo at http://zeromq.org/distro:centos. Problem is that salt-minion has zeromq2 from epel as a dependency and will not install even if you install zeromq 3 first.
18:08 NotreDev mr_chris: would it if you used yum-plugins-priority?
18:08 whitepaws mr_chris: yeah it is really annoying i have been dealing with that the last couple days
18:08 mr_chris If you do install zeromq3 first then try to install salt-minion it will complain about dependency issues which you can only resolve by first uninstalling zeromq3.
18:08 whitepaws there is a specific library bind issue too
18:08 pdayton joined #salt
18:08 whitepaws i have rebuilt salt and salt-minion from source RPMS, on a system with libzmq3 and related devel packages installed
18:09 mr_chris NotreDev, I haven't tried that yet, but i think it's version specific.
18:09 scott_w joined #salt
18:09 whitepaws and that actually works .. the only thing is that you'll still get warnings about the ZMQ version being too low. but i think the warnings are wrong
18:09 mr_chris The dependency in question is libzmq. It's libzmq3 in the zmq repo but libzmq in the normal repo.
18:10 mr_chris So the salt-minion package will never look for libzmq3 as a dependency but libzmq and libzmq3 conflict and both can't be installed.
18:10 mr_chris whitepaws, In my case we're losing the connection between the two.
18:11 bhosmer_ joined #salt
18:11 whitepaws mr_chris: yes. as you can probably tell, there is no making it work with ZMQ3 + without rebuilding at least one RPM somewhere.
18:11 whitepaws so i have been getting much more familiar with rpm spec files recently
18:12 mr_chris whitepaws, Good to know. Thanks.
18:12 whitepaws sure thing. yumdownloader with source option is a major time saver too btw
18:13 forrest Did you guys file that as an issue yet, or open a discussion on it?
18:13 whitepaws forrest: ah no actually
18:13 scott_w joined #salt
18:13 whitepaws it would be my first time doing such a thing but i would be happy to
18:14 whitepaws opening a discussion .. as in on the mailing list?
18:14 forrest https://groups.google.com/forum/#!forum/salt-users
18:14 forrest yea
18:14 forrest I haven't seen one in there about this, granted I didn't check my updates from yesterday
18:14 whitepaws cool
18:15 bhosmer joined #salt
18:16 whitepaws i will gather my thoughts and notes and see what i can write up
18:16 ajw0100 joined #salt
18:17 andersb This is cheesy, but it's the weekend :) Made this yesterday to show the devs what I have been up to with salt lately: https://dl.dropboxusercontent.com/u/619726/work/svn_salt_repo_the_who_baba.mp4
18:17 whitepaws it's a bit weird since some of the "problem" is in EPEL ZMQ availability ; the recommended fix IIRC was to use fengzhuo's updated ZMQ repo
18:17 josephholsten joined #salt
18:17 whitepaws do you guys have a recommended latest version from github to use?
18:17 forrest well, develop isn't really that stable whitepaws
18:18 forrest so depending on stability, and what you need, 0.16.4 if all you care about is the stability, 0.17.1 if you need proper formula support, or salt-ssh
18:18 whitepaws ok, cool
18:19 forrest obviously some functionality that is in the docs will be missing from 0.16.4 since they are built off the develop branch
18:19 whitepaws yeah some issues i had seen were noted to be fixed in develop but i don't know that i've tried 16.4 yet
18:19 whitepaws that should be fine though. thanks forrest
18:19 forrest np
18:20 troyready joined #salt
18:21 bemehow joined #salt
18:22 seanz basepi: Thanks. That helps a bit.
18:22 pniederw joined #salt
18:22 pniederw can someone let me know what the output of `salt --version` is for HEAD?
18:23 heewa joined #salt
18:24 pniederw I installed HEAD via the homebrew formula, but it's showing 0.17.0, and hence I'm not really sure I got HEAD
18:26 forrest andersb, do your devs use SVN as well?
18:26 heewa I'm having trouble running "salt-call state.highstate" on a new minion. Complains about master's public key. Traced it down to a key-error in mine.py:960, happening on the master: if not self.__verify_minion(load['id'], load['tok']):
18:27 forrest heewa, is your minion joined to the master?
18:27 heewa @forrest What does "joined to the master" mean?
18:27 cachedout pniederw: The most recent HEAD is 0.17.0-1956-gf076895
18:27 forrest you accepted the minions key on the master
18:28 forrest heewa, http://docs.saltstack.com/topics/tutorials/walkthrough.html#using-salt-key
18:28 heewa Yea, I run in fully open mode (whatever the terminology for that is). I know it's insecure, just getting going for now, will tighten it up later.
18:28 _ikke_ Is it possible to get in a managed file information about all the hosts?
18:29 viraptor_ joined #salt
18:29 forrest heewa, oh ok, I'm not sure then, can you try to ping the minion from the master?
18:29 pniederw cachedout: is that also the output of `salt --version`? in other words, if `salt-version` gives 0.17.0 for me, that's really the 0.17.0 release?
18:29 pniederw s/salt-version/salt --version
18:30 heewa @forrest ICMP traffic is fire walled, so can't ping, but minion can reach master on the relevant ports. I've been running salt since like March.
18:31 tempspace heewa: Are the master and minion different versions?
18:31 bhosmer joined #salt
18:31 heewa Both 0.17.1
18:31 tempspace can you target the minion from the master and run test.ping successfully?
18:32 heewa What's that __verify_minion think take? The code isn't super obvious about what _mine() is as a function, or what the "load" dict is supposed to hold, or how it got it.
18:32 andersb forrest: Nah, but they need to be updated on stuff, so they understand some of the concepts and I can explain why they have to do a thing a certain way.
18:32 cachedout pniederw: It's the 0.17 branch with a hash appended to it representing the most recent updates.
18:32 heewa @tempspace As I understand it, it's not a requirement for master to be able to reach the minions, just that the minions can reach the master. Right?
18:33 forrest andersb, fair enough
18:33 andersb forrest: Also I am checking out heavily developed git trees and having scipt pulling a bare-metal machine up to a production state.
18:34 forrest andersb, yea I was just curious, it's always interesting when teams are still using SVN, since it drives me nuts
18:34 Brew joined #salt
18:34 andersb forrest: Yeah, we pretty recently switched SVN->git. Git is nicer to work with in salt when you are doing it from scratch at least.
18:35 forrest yea, svn is tough for me to use since I use git so much at home
18:35 andersb But my salt_repo is in SVn actually.
18:36 andersb That's fine though. As you can see I am pretty much alone in there. Hehe.
18:36 pniederw cachedout: ok. then I suppose something's wrong with the saltstack homebrew formula. (to be fair, it's not released yet.)
18:36 tempspace heewa: That is true, but if the master can't run test.ping, it could mean there's a communication issue, did you try running both the minion and master with -l debug
18:37 cachedout pniederw: I recommend you just grab the latest copy via git.
18:38 andersb forrest: In my dev suite I never have the need to branch. And I use git/SVN inside sublime-text with the same shortcuts anyway.
18:39 forrest ahh nice
18:39 forrest sublime text, aww yea
18:39 forrest I need to get back onto the salt sublime text syntax
18:39 tempspace I saw some conversation about that
18:39 tempspace a while back
18:39 forrest oh yea?
18:40 bhosmer joined #salt
18:40 tempspace I think I saw an empty repo
18:40 tempspace for someone who intended to write it
18:40 pniederw cachedout: I'd prefer if I wouldn't have to manually install stuff. I just want an easy & managed way to install daily salt via a package manager, so that I can work around bugs and test out fixes.
18:40 forrest are you sure that wasn't me :P
18:40 tempspace but never did
18:40 tempspace forrest: It could have been :)
18:40 redbeard2 joined #salt
18:40 heewa tempspace: output from minion & master in -l debug: http://pastebin.com/SWFa5znR
18:41 cachedout pniederw: For development, I work off a git branch and then set PYTHONPATH accordingly. You shouldn't even need to install anything on your system beyond the dependencies for Salt. (And you can install those in a virtualenv if you wish.)
18:41 cdcalef fucking rackspace. sorry off topic but.. damn them.
18:41 jcockhren cdcalef++
18:41 jcockhren carry on
18:42 rgbkrk joined #salt
18:42 tempspace cdcalef: What happened?
18:42 bemehow joined #salt
18:43 jcockhren cdcalef: don't ever say that on facebook. They'll call you. true story
18:43 cdcalef i'm just annoyed by a pervasive attitude, not specific to rackspace mind you, of 'something broke, i restarted it and now it works, so there is no problem' rather than fixing an actual problem.
18:43 tempspace heewa: If you're sure the master is currently running 0.17.1, I'd purge salt completely and re-install
18:43 tempspace on the minion
18:44 tempspace get rid of /etc/salt and /var/cache/salt
18:44 cdcalef rackconnect fails, i get e-mailed about a ticket with zero info, a 'racker' goes in and reruns the thing, everything is good!  meanwhile i get 10 e-mails every time it happens, that forces me to login to their website rather than *actually putting the ticket info in the e-mail*.  just a huge waste.
18:44 quist left #salt
18:44 pniederw cachedout: I don't develop saltstack, I don't know much about python, and I don't want to manually manage the deps. I'll only do it as a last resort.
18:44 cdcalef the delicate balance between managed hosting and hosting your own hardware.
18:45 bemehow joined #salt
18:45 cdcalef sorry to get off topic. i have a feeling plenty of people in this channel have felt that pain. :)
18:46 pniederw cachedout: for my purposes, it would be better to use a tested daily-salt, but I don't know if homebrew can deliver on that. on ubuntu there's the daily-salt ppa.
18:46 redbeard2 holmb: epic
18:46 linjan joined #salt
18:47 diegows joined #salt
18:48 tty-1_ joined #salt
18:49 tempspace cdcalef: I'm a huge fan of SoftLayer myself
18:49 bemehow_ joined #salt
18:50 tempspace it's an ideal option for us, since we have ops and don't need maanaged services
18:51 cdcalef yeah i've observed that even with the thinnest of management layers then the network becomes a problem. obvious hosting my own boxes also has that problem, but at least i can get into my own switches and routers and see whats up without having to file a ticket. it all depends. i haven't tried softlayer i should give them a shot.
18:51 cdcalef i've been through aws, rackspace, linode, heroku, more than i can remember. all sucked in different ways heh.
18:51 tempspace yeah
18:52 ipmb joined #salt
18:53 whitepaws how do you download source for a specific version, like say 0.16.4 ?
18:53 tempspace whitepaws: the releases are tagged in github
18:54 whitepaws ahh just found it tags. amazing how asking a question out loud makes you realize the answer
18:54 whitepaws tempspace — thanks
18:54 tempspace whitepaws: :-D
18:54 cdcalef there are branches as well https://github.com/saltstack/salt/tree/0.16
18:55 forrest Yea that's why at home I have a duck sitting under my monitor
18:55 forrest I just talk to him when I have a problem.
18:56 heewa When I install from pip instead of that apt repo, what setup stuff do I need to do? Because it doesn't put in an /etc/salt/minion conf file, for ex.
18:56 jcockhren ha
18:56 forrest http://en.wikipedia.org/wiki/Rubber_duck_debugging
18:56 cdcalef has anybody made a tutorial about bringing a medium sizes (double digit) typical system like a LAMP stack into salt with all the bits and pieces?
18:56 forrest cdcalef, not as far as I know, are you porting it from another config management system?
18:57 bemehow joined #salt
18:57 cdcalef no, from your workaday shell scripts and fabric file proto-management
18:57 forrest how do you have it setup, a bunch of identical web servers, connecting to some sort of database cluster?
18:57 cdcalef i haven't seen any, and i kind of realized with all my notes and stuff i'm pretty much making exactly that.
18:58 forrest yea if you're doing it, then by all means, share
18:58 forrest especially if you can upload the project somewhere (removing security concerns of course)
18:58 cdcalef forrest, essentially. but a bit more complex than that.  throw in some redis caches, rabbitmq message queues, async workers, some clustered nosql stuff, etc.
18:58 forrest github repo or whatever.
18:58 cdcalef forrest for sure
18:59 forrest yea that would e awesome to see
18:59 forrest *be
18:59 cdcalef i've been finding getting a clear explanation of sls trees, your salt state, modules, all that, and how that maps to an actual production system has been lacking as well
19:00 cachedout pniederw: If you have a git checkout another option would be to use setup.py install --force and it will install over the top of what brew has installed.
19:00 cdcalef just going through http://docs.saltstack.com/contents.html
19:00 cdcalef has anybody made a salt book?
19:00 forrest cdcalef, yea that is a problem for sure, and no one has made a salt book yet
19:01 josephholsten joined #salt
19:02 pniederw cachedout: I just verified that `brew install saltstack --HEAD` did clone the saltstack repo, and it is at the latest commit. hard to believe that I'd still get 0.17.0. I wonder if the output of `salt --version` is to be trusted when running from HEAD.
19:02 faldridge joined #salt
19:03 NotreDev joined #salt
19:04 cachedout What does --version give you?
19:04 pniederw 0.17.0
19:05 pniederw whereas `brew install saltstack` (w/o --HEAD) gives 0.17.1
19:05 pniederw the latter is based on: url 'https://pypi.python.org/packages/source/s/salt/salt-0.17.1.tar.gz'
19:06 pniederw the former is based on: head 'https://github.com/saltstack/salt.git', :branch => :develop
19:06 pniederw (I pasted this from the homebrew formula)
19:06 cachedout Just 0.17? Nothing after that in the version string?
19:06 pniederw exactly
19:06 cachedout Interesting.
19:07 pniederw 0.17.0, to be precise
19:08 pniederw but since I'm still running into a supposedly fixed issue, I'm not sure who's right.
19:09 mr_chris whitepaws, Checking out yumdownloader. This will be my first experience with spec files.
19:09 heewa joined #salt
19:11 mwillhite joined #salt
19:12 dh joined #salt
19:13 IJNX joined #salt
19:15 _ikke_ I want to create a hosts file with items for each minion. Is there a way to iterate over each minion and retrieve items from grains?
19:15 ajw0100 joined #salt
19:19 tempspace _ikke_: That's a great question, if you get a good answer let me know
19:20 tempspace _ikke_: I ended up doing a very ugly solution where I use Python to iterate over all the minions to grab grains to grab the IP address of each minion
19:20 tempspace run it on a cron
19:20 tempspace and use it for firewall rules
19:20 Psi-Jack heh.
19:21 Psi-Jack I need to start using salt for my shorewall setup.
19:21 Baribal__ joined #salt
19:21 Baribal__ Hi. I've stumbled upon SaltStack yesterday and now I'm trying to figure out what people pay for when licensing SaltStack Enterprise. Or rather: What do I not get without paying?
19:21 Psi-Jack defunctzombie: That's really annoying. Would you mind disabling away nicks?
19:21 IJNX salt-keys -L ?
19:22 tempspace _ikke_: I'd be happy to share what I wrote if it'll help
19:22 defunctzombie Psi-Jack: no, sorry, just hide nick change message if it bothers you
19:22 tempspace Baribal__: Salt Enterprise gets you a guaranteed stable build and support I believe
19:22 _ikke_ tempspace: I just found this: http://kuttler.eu/post/list-minions-template/
19:22 Psi-Jack defunctzombie: You demand the impossible, while I politely ask the simple.
19:22 defunctzombie I do away nicks for a reason, I am not changing it, get over it sorry
19:23 Psi-Jack defunctzombie: MOST IRC clients don't allow you to hide NICK changes, however, you can ignore notices, messages, people, quits. parts, joins.. But nick changes, the user is still part of the channel.
19:23 Baribal__ tempspace, so all I'm missing out on is people doing work for me, and liability, not any actual functionality?
19:23 Psi-Jack Fine, I will bug you EVERY fracking time till you do. USE /AWAY
19:23 cdcalef your client should be okay with something like /ignore *!*@*defunctzombie NICK
19:23 Psi-Jack defunctzombie: Away nick/messages: http://goo.gl/n55tH -- /away was put into IRC since 1992~1993, 1 nick is all you need, for "away", just use the native IRC away status mode. Everything else is VERY annoying to everyone.
19:23 tempspace Baribal__: That's my understanding, but basepi could do a better job selling it :)
19:23 Psi-Jack cdcalef: Very few clients can ignore nick changes.
19:24 Baribal__ ...and just like that, Puppet is dead to me! :D
19:24 defunctzombie Psi-Jack: no one looks at away, but this way when they tab complete my name they can know if I am there or not
19:24 basepi what am i selling?
19:24 cdcalef i wouldn't know, i've never used one that didn't..
19:24 tempspace Baribal__: I believe they're building out more advanced functionality specific to Enterprise, but they're not quite there yet
19:24 defunctzombie Psi-Jack: I am not changing it.. you can stop trying to make any sort of argument about it :)
19:24 Baribal__ Awwww... :(
19:24 tempspace Baribal__: Like more web based stuff
19:24 Psi-Jack defunctzombie: That's not your problem. but it's my problem that you flood every chanel YOU are in because you auto away/back your fracking nick.
19:24 cdcalef not shorting the open source community on features is good, and i think at the current state they'll make a grip of money on 'support'
19:25 tempspace Baribal__: it won't be a big deal, welcome to the puppet is dead to me club!
19:25 defunctzombie Psi-Jack: ok, then it is your problem... learn to live with it and move on.. seriously
19:25 cdcalef man i haven't been on freenode in a while. hah
19:26 tempspace Baribal__: We have membership cards and everything
19:26 cdcalef for real? give me one of those
19:27 tempspace ha
19:27 Psi-Jack Bah.
19:27 Baribal__ tempspace, that's worrying. Membership cards are the first sign of walling off the rest of the world. I'd rather have a ripped-up Puppet fanclub membership card. :)
19:27 cdcalef i used puppet on six digits worth of nodes.  it was not fun.
19:27 cdcalef i really like salt from a high level but tying it all together has been more difficult than i anticipated
19:27 tempspace Baribal__: We don't actually have cards, in fact there's no we at all
19:28 NotreDev joined #salt
19:28 Baribal__ cdcalef, I know the feeling... ...which is why I'll now dive through the tutorials to find out about heterogenous server environments.
19:28 whitepaws cdcalef: thanks (from earlier)
19:28 Baribal__ That's one of the things that *really* bugged me about Puppet.
19:28 cdcalef if you find something good let me know. i haven't.
19:29 tempspace There's a few of us talking about writing a book
19:29 tempspace There's a real need for it
19:29 cdcalef i like how the structure is built
19:29 cdcalef but i wish things were a little bit more self evident
19:30 _ikke_ tempspace: aparently, salt mines can be used for this too
19:30 cdcalef i look at a random snippet of sls from a tutorial and i have a hard time identifying what is a module, what is a state, what is a function, what is an argument..
19:30 Gareth basepi: are you selling swamp land in Florida again?
19:30 cdcalef its clearer in the pydsl
19:31 tempspace _ikke_: definitely
19:31 cnelsonsic joined #salt
19:31 _ikke_ Still trying to find out how
19:31 jcockhren cdcalef: yeah. I think a few of us are basically dealing with salt's growing pains as we build out or convert our infrastructures
19:31 cdcalef jcockhren exactly
19:32 jcockhren cdcalef: as a sys admin, that's not fun. but it's necessary since that's really the only way to make sure salt gets the coverage we need
19:32 jcockhren s/gets/gives/
19:32 cdcalef i've been going at http://docs.saltstack.com/contents.html and going through everything i can heh
19:33 tempspace cdcalef: Yeah, that's rough
19:33 tempspace tons of great info
19:33 cdcalef coming from big giant puppet i see the benefits of salt and i like the approaches its taking.
19:33 tempspace but not great for learning about fundamentals like how to use pillars, grains and jinja in a helpful way
19:33 tempspace best practices on when to use each, etc...
19:34 cdcalef now i've got the opportunity to take this new place from `for loop ssh` and fabric files to a good basis
19:34 josephholsten joined #salt
19:34 basepi Gareth: so much swamp.
19:34 xmltok tempspace: i find that that is really missing and where a book could really help
19:34 cdcalef yeah the biggest struggle i have is trying to grok what some line in a YAML file is actually doing under the hood.
19:35 xmltok along with some use case info would be good
19:35 tempspace yeah
19:35 tempspace I think one way or another it's going to happen
19:35 tempspace I think it's going to be key for higher salt adoption
19:35 xmltok i will probably write up a blog post on ours when we have it figured out, as it is its still pretty early in the game
19:36 Ahlee difference between /var/cache/salt/minion/extmods/modules/ and /var/cache/salt/minion/files/base/_modules/
19:36 cdcalef i see a lot of stuff in the tutorials that sort of uses salt in a way very similar to fabric
19:36 Ahlee ?
19:36 pipps joined #salt
19:36 tempspace cdcalef: Yeah, focusing on modules isn't helpful when getting started
19:36 tempspace at least it wasn't for me
19:36 xmltok yeah, salt started off as something like fabric and expanded into something like puppet
19:36 xmltok basically modules are fabric-ish, states are puppet-ish
19:36 cdcalef xmltok thats a good way to put it i'm stealing that line.
19:37 tempspace all I really needed to know to get started is behind the scene, the states are the rules that supply the runtime paramters to the modules
19:37 xmltok tempspace: yeah, browsing the salt source (modules and states dirs) made a lot of it clear
19:38 xmltok and using pydsl rather than jinja/yaml also helped out a ton
19:38 NotreDev i'm including "context" on a managed file, and this is something that used to work. my code is no longer working on salt 0.17. i can't find much documentation on using "context". anyone have a real code sample?
19:38 xmltok i got one, one sec
19:38 cdcalef yeah
19:38 tempspace xmltok: yeah, we need to make it easier for people who aren't super proficient in Python as well
19:38 Ahlee pydsl in place of jinja/yaml you say?
19:38 cdcalef and there is that uncomfortable puppet/fabric middle space in a lot of the examples
19:38 xmltok notredev: http://pastebin.com/jCbGVyvY
19:38 cdcalef where they're defining a state, but then applying that state to hosts via globs
19:39 tempspace such as myself who's more Ops than Dev :)
19:39 cdcalef rather than a puppet-ish node manifest
19:39 NotreDev xmltok: thanks
19:39 xmltok ahlee: http://pastebin.com/jCbGVyvY :)
19:40 Ahlee xmltok: thanks
19:40 Ahlee Is there a way to list what modules a server things it has available? I know i've run this custom module against this server before, I can see the custom module in both /var/cache/salt/minion/extmods/modules/ and /var/cache/salt/minion/files/base/_modules/, but when I call it, it's just saying not available
19:41 cdcalef if your installations are you guys mostly relying on 'globs' to target hosts or are you using some setup of node manifest (or map or whatever) and node "roles?"
19:41 xmltok the other thing i like about pydsl is that i can print debugging information in there, you can literally run print and see it in the salt-call output. that is huge IMO
19:41 cdcalef s/if your/in your/
19:42 Ahlee cdcalef: I target exclusively via grains i define on the hosts, soon to move that all to external pillars
19:42 xmltok cdcalef: i was originally going to use globs but i am now planning on using external scripts to generate my pillar data and environment/state lists. we have an inventory system with information on the node. i'll use that to make decisions and select the pillar data and states
19:42 cdcalef i'm a little weary about salt-call.  good for dev and debugging but worries me about getting hosts divergent.
19:43 xmltok yeah i just use salt-call for development/testing, everything is getting applied through scheduled runs on my minions
19:43 cdcalef yeah host inventory is a big thing i've been thinking about.
19:43 cdcalef along with automatic provisioning and registration. but i might be getting ahead of myself.
19:43 xmltok have you guys run into issues managing dependencies between states/formulas? I want to have formulas for our internal app deployments, but there is no way to define a dependency on another formula, or a specific version
19:44 higgs001 joined #salt
19:45 xmltok for example my app formula depends on my apache formula, but someone releases a change to the apache formula that breaks app somehow. i think this is one thing chef has right with its cookbook versioning
19:45 _ikke_ Someone knows a good resource explaining how to use the mine module? The documentation is a bit lacking it seems
19:45 khodgson joined #salt
19:47 blee joined #salt
19:49 cdcalef "A tool for visualising the dependency graph of a Salt highstate.
19:49 cdcalef " why hello there.
19:49 ckao joined #salt
19:52 khodgson joined #salt
19:54 tempspace cdcalef: oh that's neat
19:55 khodgson_ joined #salt
19:55 dave_den _ikke_: http://docs.saltstack.com/topics/mine/index.html     and then there's the mine execution module:  http://docs.saltstack.com/ref/modules/all/salt.modules.mine.html
19:56 cdcalef yeah it is https://github.com/DanielBryan/salt-state-graph  i'm a sucker for anything graphviz.  and it shows the module as well!
19:56 dave_den _ikke_: the thing about the mine is that it just stores data for module.function.
19:56 _ikke_ dave_den: Yeah, I've read those, and I see what they do, but what's not exactly clear is how to use them
19:57 dave_den _ikke_: salt '*' mine.send grains.get os
19:58 dave_den the problem with that is that you can't store just one or two grains in the mine. you can only store one value per module.function
19:59 _ikke_ The problem I want to solve is to get a list of ip's and hostnames for each minion and put that in a hosts file
19:59 dave_den so you can't store both "grains.get ipv4" and "grains.get os".
19:59 dave_den i know
19:59 _ikke_ okay
19:59 dave_den i'm just explaining the mine before i suggest you should create either a runner module or an ext_pillar module  :)
20:00 dave_den you could also use the 'peer' interface.
20:01 dave_den but a runner or ext_pillar would be faster, more efficient, and flexible.
20:01 _ikke_ yeah, I read an article that used the peer feature, but then later suggested the mine module
20:01 _ikke_ dave_den: okay
20:01 dave_den if you are only ever going to use the mine to store ONE grain with grain.get, then it would be fine.
20:02 dave_den you could store *all* grains in the mine, but that seems bad to be
20:02 _ikke_ If I run that command, the minion returns false btw
20:03 redondos joined #salt
20:03 redondos joined #salt
20:03 dave_den it should return true
20:04 dave_den https://gist.github.com/dlanderson/78e10c5f473dfcf80d8d
20:06 jslatts joined #salt
20:07 pipps_ joined #salt
20:07 _ikke_ my bad
20:07 Psi-Jack defunctzombie: That's still very annoying. :p
20:10 cdcalef you complaining about it is more annoying than his nick changes.
20:10 _ikke_ I don't even see the nick changes
20:10 cdcalef me neither. they're there, but they don't hit my brain, like a banner ad.
20:10 Psi-Jack _ikke_: Well, he's doing it every 2~10 minutes.
20:10 cdcalef actual words i have to look at.
20:11 _ikke_ My clients smart filter filters them out
20:12 cdcalef i should make a plugin that filters joins/parts except for people who are/have talked.
20:13 _ikke_ weechat's smart filter actually does that
20:13 cdcalef oh thats cool
20:16 andersb cdcalef: Did not read the whole backlog... Here is a start I used for building salt stuff: https://github.com/brutasse/states
20:16 faldridge joined #salt
20:16 andersb cdcalef: Some okay key managment stuff. Also see redis there.
20:17 tempspace _ikke_: not sure if this helps, http://stackoverflow.com/questions/17158665/how-to-get-a-list-of-all-salt-minions-in-a-template
20:17 cdcalef andersb thats cool i'll have to take a look at that deeper
20:17 faldridge joined #salt
20:18 andersb cdcalef: I also like the site structure he did.
20:18 Kareem joined #salt
20:19 _ikke_ tempspace: I'll look into it
20:20 tempspace _ikke_: It should help, gives syntax for a state and everything
20:20 andersb cdcalef: You have to guess his pillar stuff (he docs some of it), and that lead me to look into pillars and "get" it.
20:20 Cidan need...
20:20 Cidan 0.17.2...
20:20 Cidan please... someone...
20:20 * Cidan gasps for air
20:20 * Cidan dies
20:20 tempspace well if you're dead I guess you don't need 0.17.2 any longer
20:21 * Cidan revives
20:21 Cidan Happy?
20:21 Cidan HAPPY?!
20:21 tempspace I've never been happy :(
20:21 Cidan D:
20:21 jcockhren tempspace: even if Cidan dies, his infrastructure will live on. preserved with salt
20:21 tempspace jcockhren: ugh
20:21 andersb cdcalef: And also something to take a look at maybe: https://github.com/uggedal/states
20:22 cdcalef tempspace spoken like a true sys admin
20:22 higgs001 joined #salt
20:22 cdcalef yeah i've found a fuckton of 'states' for doing simple configuration of a simple package
20:22 cdcalef not so much in coordinating that around a whole environment
20:25 troyready joined #salt
20:25 andersb cdcalef: Start small, that's good. But soon you are gonna need some structure for the environment. take a look at the links then. The last one is better than I remembered I see.
20:25 andersb cdcalef: With good examples of pilalrs etc.
20:26 cdcalef andersb structure for the environment is what i'm worried about most
20:26 cdcalef right now at least.
20:28 tempspace These are pretty good, uggedal's
20:28 tempspace I tend to never use defaults: in a file.managed jinja template though
20:29 cdcalef hmm
20:29 tempspace use pillar.get with a default value in the template to make it easy to change on a host by host basis
20:29 dfinn joined #salt
20:29 cdcalef the best i've seen is 'top files' with base
20:29 cdcalef and then glob or explicit hostnamed states
20:29 pipps_ joined #salt
20:29 cdcalef rules whatever.  need a glossary heh.
20:29 dfinn I'm not sure what changed but all of a sudden I'm not able to run any salt commands from my master
20:29 dfinn http://pastebin.com/GNGVDA6G
20:30 dfinn I did upgrade to 0.17 yesterday but it was working after the upgrade
20:30 mdasilva joined #salt
20:30 tempspace I ended up doing something way out there with my states and pillars
20:30 cdcalef i've also seen having different 'environments' like prod, qa, stage with different salt fileroots
20:31 tempspace Let's say I want to add redis to a server, I was adding the redis pillar that specified memory limits and whatnot I wanted
20:31 tempspace then I had to add the redis state
20:31 tempspace so I was double double duty, adding it in multiple places
20:31 nmws_ joined #salt
20:32 tempspace now my state top.sls just checks pillar values and loads the state based on the pillars
20:32 tempspace so I only change my state top.sls when I add new state
20:32 jcockhren cdcalef: my approach to use pillar as my authoritative source
20:32 jcockhren is* to use
20:33 nmws_ can i define a meta-grain that represents the aggregate of multiple other grains? like if >50% of minions say TRUE for a grain, then this meta-grain becomes TRUE
20:33 jcockhren that means, me helping to flesh out the reactor system, grains and pillar
20:34 dave_den dfinn: are your minions also 0.17.1?
20:34 dfinn yeah
20:34 dfinn updated everything
20:34 cdcalef this seems to come up a lot. is there any protocol compatibility policy?
20:35 cdcalef having newer minions always able to talk to older masters so you can upgrade your minions then your master sounds like it'd be nice
20:35 dave_den dfinn:  is it intermittent? https://github.com/saltstack/salt/issues/8295
20:35 dfinn it's not, seems to be happening every time
20:36 dfinn i haven't been able to run a salt command yet
20:36 dfinn now that I think of it, I don't think I tested this the the other day after doing the upgrades
20:36 dfinn no, I did
20:36 dfinn I take that back
20:36 dfinn and it was definitely working
20:36 forrest lol
20:36 cdcalef_ joined #salt
20:37 dfinn the salt-key stuff is still working, I just had some new clients ask for keys
20:37 dave_den dfinn: have you enabled debug logging on your master and a minion to see if you get anything?
20:37 cdcalef_ lets give this badboy a whirl.
20:37 dfinn i enabled it on the master and pasted it above
20:37 dfinn i'll try and client and see if I get anything helpful
20:38 dfinn nothing gets logged by the client, it's like the connection isn't making it there
20:38 dave_den dfinn: what does salt —versions-report show?
20:38 dave_den did you upgrade zmq?
20:38 dfinn i did
20:39 dave_den to what version
20:39 dave_den that's probably the issue
20:39 dfinn zeromq3-3.2.4-1.el6.x86_64
20:39 dfinn on the master
20:39 dave_den hrm, that should be ok
20:39 dfinn on a minion
20:39 dfinn zeromq3-3.2.4-1.el6.x86_64
20:39 dave_den show the versions-report
20:39 dfinn but not all minions have the same version because some are cent5
20:40 dfinn it does not think that's a valid command
20:40 dfinn it gives me the help stuff
20:40 NotreDev "could not find a pip binary" on my virtualenv.managed state...
20:40 dfinn oh, double dash
20:41 NotreDev strange... i thought that pip installed itself when creating virtualenvs
20:41 dfinn http://pastebin.com/TCGZqM1k
20:41 forrest NotreDev, did you install pip?
20:41 forrest for the system?
20:41 NotreDev forrest: yeah
20:41 NotreDev the virtualenv is empty though
20:42 NotreDev i.e. the directory is baren
20:42 cdcalef joined #salt
20:42 forrest Was that virtualenv configured via salt?
20:42 cdcalef i like a gui native client rather than a terminal one :)
20:43 NotreDev forrest: no, it's failing to configure because it can't install the requirements
20:43 mdasilva1 joined #salt
20:43 forrest Hmm, ok so I just do this: https://github.com/gravyboat/hungryadmin-sls/blob/master/salt/python/pip.sls
20:43 NotreDev http://pastebin.com/vaGPdzgN
20:44 dave_den dfinn: not sure.. i have el6 minions that are running python 2.6.6, pyzmq 2.2.0.1 and zmq 3.2.4 and they are working fine. my master is ubuntu precise though
20:45 forrest that seems odd NotreDev
20:45 forrest https://github.com/gravyboat/hungryadmin-sls/blob/master/salt/hungryadmin/app.sls#L65
20:45 forrest I'm doing that
20:45 dfinn well, that's a bummer.  gonna make it hard to move forward with this.
20:45 forrest my requirements.txt only has pelican inside of it though
20:46 forrest oh you have no-site-packages
20:46 forrest that's why
20:46 khodgson joined #salt
20:46 forrest NotreDev, any particular reason for that? Because you won't get pip if you do that
20:47 forrest even if you make a new virtualenv by hand with no-site-packages you won't get pip.
20:47 dave_den dfinn: have you made sure to stop all salt-master processes and restart it? like actually look at 'ps' output to make sure it's being stopped if you're doing 'ervice salt-master stop'
20:47 dfinn i went as far as to reboot
20:47 dfinn so yeah ;)
20:48 dave_den dfinn: that traceback looks like it can't connect to the ZMQ socket
20:48 dfinn is there something that needs to be started for zmq?
20:48 dave_den no
20:48 dfinn i never had to mess with it previously
20:49 dave_den if you enable 'trace' logging and paste more context from your log we may be able to see more helpful info
20:49 dfinn how do I do that?
20:49 dave_den edit your master config and set log_file_level (or whatever it's called) to trace. retart master.
20:50 NotreDev forrest: not working regardless, even if i delete the .virtualenv
20:51 forrest NotreDev, what do you mean?
20:51 forrest what happens if you create a new virtualenv straight off the command line?
20:51 forrest versus creating one with the --system-site-packages
20:51 forrest does pip get installed there?
20:52 dave_den dfinn: also, have you tried just doing a test.ping against a known up minion?
20:52 NotreDev forrest: works fine.
20:52 NotreDev i'm pretty sure that the virtualenv is just not getting created
20:52 forrest ahh ok
20:52 NotreDev there is no bin/lib/include
20:52 forrest oh
20:52 dfinn yes, I did try that.  I narrowed it down to several single minions and still the same result.  timeout
20:52 forrest yea that's a problem
20:52 forrest can you activate the virtualenv that salt creates?
20:53 dfinn also strange, salt-master is failing to stop.  i run /etc/init.d/salt-master stop and it says FAILED.  it's leaving processes still running
20:53 forrest NotreDev, what's with this: - requirements: salt://sentry/files/requirements.pip
20:53 NotreDev forrest: just a pip freeze of what sentry requires
20:53 forrest ok
20:54 dave_den dfinn: killall −9 salt-master
20:54 forrest can you try to apply that state, with the -l debug option?
20:54 dave_den then start salt-master -l trace
20:54 dfinn i'm getting the trace logs now
20:56 nmws_ So how do Salt and Cobbler compare/contrast? Anyone have any specific examples of them working together?
20:57 forrest well, a cobbler makes shoes, salt makes magic
20:58 dfinn nothing seems to get logged to the salt master when I issue the salt command, it's like they aren't talking
20:58 dfinn will paste the log once it times out
20:58 forrest in reality nmws_, I don't know enough about cobbler to say
20:58 nmws_ forrest :) actually im trying to figure out how to do external node classification, which is how i ran into salt+cobbler
20:59 forrest external node classification, that sounds like puppet...
20:59 scott_w joined #salt
20:59 nmws_ forrest yes im a puppet refugee. :)
20:59 forrest did you already see this:
20:59 forrest http://docs.saltstack.com/ref/tops/all/salt.tops.ext_nodes.html
20:59 forrest not much in there though :\
21:00 nmws_ forrest yeah thats what i read, but i dont understand how this would work in practice..
21:00 nmws_ forrest basically what i want to do is that i want to externally control which nodes get which states
21:00 forrest outside of the top file?
21:00 nmws_ forrest yeah
21:00 ajw0100 joined #salt
21:00 bemehow_ joined #salt
21:01 dave_den nmws: you write an ext_nodes module which accepts a minion name. the module returns the names of the state files that apply to that minion.
21:01 dfinn1 joined #salt
21:02 nmws_ dave_den: ah so that doesnt even need cobbler etc.?
21:02 dave_den no, but there is a cobbler module already if you want to use it.
21:02 dave_den https://github.com/saltstack/salt/tree/develop/salt/tops
21:03 nmws_ dave_den, forrest: thanks guys, will read up on this
21:04 dave_den t a master_tops module that executes a system shell and uses the stdout from that as the list of states
21:04 dave_den whoops
21:05 forrest jeez dave_den
21:05 dave_den nmws: http://docs.saltstack.com/topics/master_tops/index.html
21:05 dave_den what i meant to say was:
21:05 dave_den i made a mistake. you want to write a master_tops module. ext_nodes is just a master_tops module that executes a system shell and uses the stdout from that as the list of states
21:06 redondos joined #salt
21:06 redondos joined #salt
21:06 pdayton joined #salt
21:08 aco_ joined #salt
21:08 aco_ hello, i have a problem setting mongo returner and i wish someone could help me in that
21:08 aco_ i got this error
21:08 aco_ The return failed for job 20131108232611160367  'mongo.returner'
21:09 aco_ and added the suggested data in my minion config also
21:09 aco_ http://docs.saltstack.com/ref/returners/all/salt.returners.mongo_return.html
21:09 dave_den aco_: check your minion logs/
21:10 aco_ dave_den: here is the error in minion log, The return failed for job 20131108232611160367  'mongo.returner'
21:11 dave_den turn on debug logging in your minion config and restart your minion. then check the log for errors.
21:12 aco_ dave_den: nothing more that this also after activating debug log level: 2013-11-08 23:12:06,520 [salt.minion                                 ][ERROR   ] The return failed for job 20131109001206366836 'mongo.returner'
21:13 dave_den turn logging up to 'trace' and try again
21:14 aco_ dave_den: turned to trace and still the same: 2013-11-08 23:14:30,776 [salt.minion                                 ][ERROR   ] The return failed for job 20131109001430606457 'mongo.returner'
21:15 dfinn1 pastebin has a limit of 500k and my trace logging is larger than that.  somewhere else I can share these logs?
21:15 micah_chatt joined #salt
21:15 forrest did you try gist?
21:16 forrest I don't know what the limit looks like over there
21:16 forrest you could always just push it to a file on your github account
21:16 dfinn guess it's time to sign up
21:16 forrest lol
21:17 dave_den aco_: sorry, without any more log information i can't help other than to suggest you make sure you can manually connect to the mongodb using the returner settings.
21:17 VertigoRay when setting up a git_pillar, and I want to setup a subfolder as root (like the gitfs_root option), how do I target a sub folder in my repo?
21:17 micah_chatt Has anyone else had an error running high state on Ubuntu 13.04?
21:18 forrest joehh, I know i've asked this before, but your Ubuntu 13 (raring) patch was pushed in right?
21:19 micah_chatt http://pastebin.com/FzR05Ft8
21:20 pentabular joined #salt
21:20 micah_chatt I've even tried running develop, and still no dice
21:20 forrest oh, that's odd...
21:20 redondos joined #salt
21:20 redondos joined #salt
21:20 faldridge joined #salt
21:23 micah_chatt Does that look familiar to anyone?
21:24 jcockhren micah_chatt: yeah. there's an issue for that
21:24 micah_chatt jcockhren: thanks, I searched and didn't see it
21:24 micah_chatt what issue #?
21:24 troyready joined #salt
21:25 higgs001 joined #salt
21:25 scott_w joined #salt
21:27 jcockhren micah_chatt: I can't find it either. I thought someone listed it here a few days ago though
21:27 ctdawe joined #salt
21:27 jcockhren maybe I dreamt that
21:28 dfinn ok, here are my trace errors
21:28 dfinn https://github.com/danfinn/errors/blob/master/salt_errors.txt
21:31 jslatts /join #sockjs
21:31 jslatts whups
21:33 jslatts joined #salt
21:35 dave_den dfinn: how is salt-master started? what distro are you running?
21:35 dfinn /etc/init.d/salt-master start
21:35 dfinn cent 6.4
21:36 dfinn which appears to run /usr/bin/python /usr/bin/salt-master -d
21:36 dave_den dfinn: cat /proc/$(</var/run/salt-master.pid)/limits
21:38 dfinn http://pastebin.com/Pn6Zm7Dv
21:38 dfinn i do remember seeing a warning last week about open file limits
21:38 dfinn max open files needs to be bumped?
21:38 dave_den also do cat /proc/$(</var/run/salt-master.pid)/status
21:38 dave_den yes, very likely.
21:38 dave_den you have 209 minions and only 4096 limit on nofile
21:39 dfinn http://pastebin.com/bhUtbnWw
21:39 dave_den yep
21:40 dave_den bump your nofile limit to something much higher.
21:40 dfinn is nofile the same as max open files?
21:40 dave_den yes
21:40 dfinn what's a reasonable amount?
21:40 heewa joined #salt
21:41 schristensen joined #salt
21:41 dave_den cat /proc/sys/fs/file-max  is how many the entire system can have open.
21:41 danielfrg joined #salt
21:41 dfinn 98821
21:42 dave_den go for 40k
21:42 khodgson joined #salt
21:43 jefferai joined #salt
21:43 dfinn ok, trying that now
21:43 dfinn do the per user one's need to be upped too?
21:44 jacksontj joined #salt
21:44 dave_den the only one that matters to salt-master is the one that's set during the initscript start
21:44 josephholsten joined #salt
21:44 dave_den but if you start salt-master from a shell, you will need your root user's limit set to 40k as well.
21:44 diegows joined #salt
21:45 dfinn ok, will do
21:45 aco_ dave_den: look http://i.imgur.com/fnYZ5F6.jpg
21:46 racooper what a day. had to rebuild my salt master from scratch; RAID on the vmware host corrupted.
21:46 aco_ dave_den: as you see i connected, could you help telling me what is wrong now?
21:46 heewa joined #salt
21:46 racooper now a question..what am I going to have to do with the minions to re-recognize the new master?
21:46 cdcalef_ joined #salt
21:47 khodgson joined #salt
21:47 forrest racooper, just point the systems at the new master, and accept the key
21:47 forrest and you should be good to go
21:47 forrest so you'd modify your minion config file
21:47 forrest and restart the service of course
21:48 dave_den aco_: without more logs from the salt minion about why the mongo return failed, i wouldn't know what else to suggest. you can post your config for the salt returner and we can look at it, but it may not show the problem.
21:48 racooper shouldn't need to modify...IP and hostname are the same
21:48 forrest oh
21:48 forrest you just rebuilt the master then?
21:48 forrest what do you get on the master when you do salt-keys -L ?
21:49 racooper now nothing, I had just flushed them.....
21:49 racooper but I'm starting to see them reconnect and new keys to accept
21:49 TheRealBill hrmmm so 0.17 from git is now breaking when trying to start a salt-minion. It is failing to import salt.cloud.cli. Looks to be a very recent change by Hatch. Just thought I'd pass a heads-up along.
21:50 forrest racooper ok cool
21:50 forrest you mean on the develop branch TheRealBill?
21:50 redondos joined #salt
21:50 redondos joined #salt
21:50 cjh973 i'm having an odd problem where packages fail to install with a salt-call run but i can apt-get install them just fine
21:50 dave_den TheRealBill: thanks for the heads up. Thomas reserve the right to break develop at any time ;)
21:50 forrest I just always assume develop is broken
21:50 forrest works out pretty well
21:52 forrest cjh973, are you getting an error or what?
21:52 TheRealBill No, On 0.17. Looks like there was amerge conflict and basepi just fixed it in the last few minutes.
21:53 TheRealBill so all is well again. ;)
21:53 cachedout develop is going to be bumpy for the next few hours while we finish the salt-cloud merge
21:53 cjh973 forrest: yeah it's saying failed to install
21:53 forrest cjh973, the package fails to install via salt? Can you paste your error, as well as the state? Also what OS (Debian/Ubuntu it looks like?) and what version of salt
21:53 cjh973 i can paste my state file
21:54 cjh973 ok hang on lemme dig up this info for ya
21:54 forrest ok
21:54 forrest also, if you can run the call with -l debug
21:54 forrest we'll get debug level logging
21:54 forrest take a look at that first
21:54 forrest to see what it looks like
21:54 cjh973 ok lemme do that
21:55 dfinn looks like salt commands are still timing out even with the 40k limits in place
21:55 dave_den TheRealBill: woops, didn't see you say 0.17. sorry about that
21:55 samCones joined #salt
21:55 TheRealBill dave_den: no worries.
21:55 forrest yea I always assume when people say 0.17 they mean the latest release.
21:55 dave_den dfinn:  check the /proc limits and status again
21:55 samCones Does salt work on win/.net environments? (appears not, checking)
21:55 dfinn [root@salt-master01 ~]# cat /proc/sys/fs/file-max
21:55 dfinn 40000
21:56 dave_den oh, no no
21:56 grep_awesome Has anyone else run into `"monit.summary" is not available.` when monit is definitely running?
21:56 dfinn no?
21:56 dave_den did you set sysctl sys.fs.file-max?
21:56 dfinn yeah
21:56 dfinn and rebooted
21:56 dave_den rip that out
21:56 Cidan dude
21:56 Cidan you know what would be awesome?
21:57 dfinn ok, removed that
21:57 Cidan salt 0.17.2, so I can use salt-cloud and not have to use dev branch :(
21:57 TheRealBill Cidan: living in a Martian colony? ;)
21:57 dave_den dfinn: you need to set ulimit -n 40000 in /etc/salt/init.d/salt-master
21:57 cachedout samCones: Salt certainly works in Windows environments :]
21:57 Cidan or that
21:57 dfinn ok
21:57 dave_den dfinn: just set that near the top
21:58 samCones cachedout --- thanks! I like the python basis of salt, but am working on win2008 and up servers
21:59 Cidan bzz
21:59 dfinn did that, restarted, still timing out
21:59 Cidan if you're using the deb packages
21:59 Cidan you need to edit /etc/init/salt-master.conf
21:59 dfinn centos
21:59 cachedout samCones: Here's the place to start. http://docs.saltstack.com/topics/installation/windows.html
21:59 Cidan oh nm
21:59 aco_ dave_den: here is the error log since the job starts http://pastebin.com/s5yjXRDg
21:59 Cidan :(
22:00 jcockhren looks nice
22:00 jcockhren https://salt.readthedocs.org/en/latest/
22:00 cjh973 forrest: http://fpaste.org/52822/38394798/ ubuntu 13.04, salt 0.17.1-2raring
22:00 forrest oh yea the new style jcockhren?
22:00 forrest I saw the update on the readthedocs blog thing last night, it's pretty slick
22:00 ggoZ joined #salt
22:00 jcockhren yes and better organization
22:00 bhosmer joined #salt
22:00 * jcockhren bookmarks page
22:01 aco_ dave_den: and here is the config added to minion http://pastebin.com/tN6m9R39
22:03 dfinn looks like python may be crashing, just found this in the messages log: http://pastebin.com/ri29N9ng
22:03 aco_ dave_den:  and ofcourse pymongo is installed on minion
22:04 forrest so cjh973, try this for the pkg install section: https://gist.github.com/gravyboat/7378419
22:05 dave_den aco_: you are specifying return 'mongo'. it should be 'mongo_return'
22:05 forrest cjh973, also, for your original example, can you try enclosing the version in single quotes before making that change?
22:05 forrest the leading 0 might be causing the version to be interpreted as an octal :P
22:06 forrest if you can test that I can make an issue report then update the docs when I get home tonight if the leading 0 is indeed the culprit (which I think it will be)
22:07 aco_ dave_den: still the same for sorry look: 2013-11-09 00:06:23,843 [salt.minion                                 ][ERROR   ] The return failed for job 20131109010623682676 'mongo_return.returner'
22:09 dave_den aco_: sorry, i'm not sure what's going on with it. you can try emailing the salt-users group - maybe someone there knows
22:09 Brew joined #salt
22:10 aco_ dave_den: yea, thanks for your time
22:11 cdcalef_ https://salt.readthedocs.org/en/latest/topics/pillar/index.html#pillar-namespace-flattened why is bind=named rather than bind=bind9 when packages comes first?
22:11 cjh973 forrest: single quotes first?
22:11 forrest wrap the version in single quotes
22:11 forrest '0.6.6'
22:11 forrest I was just saying for your original code you pasted
22:11 forrest to try wrapping it first
22:12 cjh973 right
22:12 cjh973 ok i'll try that first
22:12 cjh973 forrest: single quotes failed :(
22:13 cjh973 lemme try your other change
22:13 forrest ok, try it without the version wrapped in single quotes, then with it wrapped in single quotes
22:13 forrest it should require the single quotes
22:13 ndivad joined #salt
22:15 dfinn dave_den, any thoughts on why salt commands are still timing out?  i even tried bumping up the limit to 80k
22:16 cjh973 forrest: that failed also :(
22:16 cjh973 lemme try wrapping that in quotes
22:16 forrest ok
22:16 cjh973 picky picky haha
22:17 dave_den dfinn: did you check /proc/<PID>/limits and status again?
22:17 cjh973 same thing.  it says it can't find the package
22:17 forrest when you go on the box, are you specifying that exact release?
22:18 forrest so when you just do a search for lxc-docker what do you get?
22:18 cjh973 lemme check
22:18 dfinn dave_den, looks good to me http://pastebin.com/YgJcdbC5
22:18 cjh973 i see this: lxc-docker 0.6.3  amd64    lxc-docker is a Linux container runtime
22:18 cjh973 for the current version
22:19 forrest is there an update available?
22:19 forrest to 0.6.6?
22:19 m0hit joined #salt
22:19 forrest from the repo that is
22:19 dfinn although I'm not actually sure what you are looking for in the status
22:19 forrest so if you were to do the system update yourself, what version would it try to install?
22:19 cjh973 yeah there's an update available
22:19 cjh973 so i wanted to make sure everyone has the same version
22:19 forrest hmm
22:20 dave_den dfinn: the FDSize
22:20 Brew joined #salt
22:20 forrest are you still getting the same error saying that it can't find the package 'lxc-docker'?
22:20 dave_den dfinn: how much ram on the salt-master?
22:20 dfinn that's still 2048.  not sure why
22:20 cjh973 yeah same error
22:20 Brew joined #salt
22:20 dfinn 1 gig
22:20 forrest and are we also sure there aren't any additional decimals onthe version, such as 0.6.6.2?
22:20 dfinn and 2 gig of swap
22:20 dave_den oh boy
22:20 sgviking joined #salt
22:20 dave_den and >200 minions?
22:21 logix812 joined #salt
22:21 dfinn sorry, if there's a hardware sizing doc somewhere I may have missed it
22:21 dfinn it's a VM so bumping it up is easy
22:21 dfinn i'm open to suggestions
22:21 cdcalef- joined #salt
22:21 cjh973 forrest: it looks like it's looking for the right thing: http://fpaste.org/52830/94926813/
22:21 forrest dfinn, might be worth bumping it up, or reducing the number of calls that can run in tandum
22:21 forrest *tandem
22:22 forrest Also I don't think there is a system resources suggested doc...
22:22 forrest that would be good to have
22:22 cjh973 it says the following new package will be installed: lxc-docker-0.6.6
22:22 dfinn bumping up to 2 CPU and 4 gig ram
22:22 cjh973 oh... i think -0.6.6 is in the name...
22:22 forrest ahhh
22:22 cjh973 those jerks haha
22:22 dave_den dfinn: http://docs.saltstack.com/ref/configuration/master.html
22:22 dfinn although according to free the system was only using 500 meg so i'm skeptical
22:23 forrest I hate when they do that
22:23 dave_den dfinn: also, increase the worker_threads
22:23 dfinn is that set in the master config?
22:23 forrest yes
22:23 cjh973 running salt-call again to verify
22:23 dfinn ok
22:23 dfinn is 200 a lot of clients?  i figured I'd be on the smaller side for a normal salt user
22:23 forrest it's a pretty small amount
22:24 dfinn hmm..seems odd that I'm running into so many issues
22:24 forrest yea
22:24 dave_den dfinn: 200 minions on a 1GB RAM master is lopsided
22:24 forrest dave_den is usually pretty good about that stuff though
22:24 dave_den for example
22:24 dave_den every thread spawned by salt-master will take 8MB of RAM by default on most linux distros
22:24 dfinn 1 thread per client?
22:25 forrest isn't the worker_threads by default limited to 25 at one time though dave_den?
22:25 cjh973 forrest: i'm thinking it worked.  it's been paused a long time on that install line :)
22:25 dfinn isn't most of that shared though?
22:25 dave_den dfinn: or more.
22:25 forrest cjh973, awesome
22:26 dfinn looks like default worker threads default is 5.  what's a reasonable amount for 250 ish clients?
22:26 dfinn minions.  sorry, gotta get that in my head.  ;)
22:28 Whissi joined #salt
22:29 ricicle joined #salt
22:30 dave_den dfinn: try 25
22:30 dfinn finally getting a response
22:30 dfinn yeah, tried 25
22:30 dfinn and 4 gig ram
22:30 dfinn although I'm getting mixed reports from test.ping
22:30 dfinn ran it one time, had 10 clients report back.  ran it again, had double or triple that
22:30 dfinn seems unstable
22:31 dave_den your minions need to reconnect
22:31 dfinn ok
22:31 dave_den let it go for a minute
22:31 dave_den then try again
22:31 xmltok joined #salt
22:31 dfinn is there a way to have test.ping report back the failed minions?
22:31 micah_chatt basepi: while #8295 is still an issue, can you all host the previous ubuntu packages in your repo so people can downgrade?
22:33 basepi micah_chatt: no idea.  =P  Corey managed the ppa right now, I think.
22:33 basepi micah_chatt: that said, it's quite easy to downgrade to a specific version using the bootstrap script
22:34 micah_chatt basepi: where might the docs be on that?
22:35 micah_chatt I cannot pip install because that is broken for 12.04
22:35 forrest micah_chatt, https://github.com/saltstack/salt-bootstrap
22:35 micah_chatt gratzi
22:36 forrest yep, I'm pretty sure there isn't an official doc page on that yet...
22:36 khodgson joined #salt
22:36 Whissi Hi. Is git develop broken? Install will fail with "error: can't copy 'doc/man/salt-cloud.1': doesn't exist or not a regular file". I see that thatch45 was working on salt-cloud ~15min ago... not sure if he finished or if should wait :)
22:36 Whissi +i
22:37 khodgson joined #salt
22:37 forrest Whissi, yes it appears to be broken
22:38 forrest someone else was saying the 0.17 tag was broken a bit ago, but basepi merged in a change that fixed it. Not sure if that is going to happen on develop.
22:39 basepi Whissi: yes, we're in the process of merging salt-cloud into the salt repo
22:39 forrest basepi, oh nice that is gonna be awesome
22:39 basepi so develop may be broken for a little while during the process.
22:39 basepi forrest: yep.  =)
22:39 Whissi OK. Thx for the information.
22:39 forrest basepi, techhat getting a little overzealous over here: https://github.com/saltstack/salt-cloud :P
22:40 nineteeneightd Any word on when 0.17.2 will drop?
22:40 forrest basepi, are you guys gonna merge the docs over as well?
22:40 basepi indeed.
22:41 forrest sweet
22:41 forrest probably still a couple weeks nineteeneightd
22:41 nineteeneightd Roger
22:41 nineteeneightd Thanks
22:41 forrest a lot of the guys are out of the office
22:41 forrest yup, someone from the team might slap my wrists for that statement, but that's what it has sounded like
22:42 forrest regarding the timeframe
22:42 xmltok joined #salt
22:42 heewa I'm getting auth errors from my minions. Master throws an exception during auth in salt/master.py : 1926: with salt.utils.fopen(pubfn, 'w+') as fp_:   Saying that the thing it's trying to open for writing is a directory. Looks like it's trying to open '/etc/salt/pki/master/minions/[minion id]/'
22:42 khodgson joined #salt
22:42 heewa Any ideas if it's a config thing, or a code error?
22:42 Brew joined #salt
22:43 heewa I've installed salt minion & master both from pip, off the v0.17.1 tag
22:43 hhtp joined #salt
22:43 hhtp left #salt
22:43 dave_den hrm, never seen that before
22:44 cachedout joined #salt
22:45 basepi nineteeneightd: actually, we hope by Monday
22:46 basepi we were hoping today but we're running into issue getting the tests passing
22:46 dave_den dfinn: how's it working now?
22:46 basepi forrest: 0.17.2 asap.  now that we fixed the syndic issue we're just working on making the tests pass and then we'll cut it
22:46 basepi hopefully today or Monday
22:47 forrest basepi, oh damn
22:47 forrest sweet
22:47 forrest squashing tons of bugs too, woo!
22:47 * cro takes a bow
22:47 * cro takes a bow for mp as well
22:48 sgviking joined #salt
22:49 dfinn dave_den, the clients aren't reconnecting
22:49 dfinn i restarted a couple and that did the trick
22:50 dfinn but i'm hoping to not have to go to each one and restart
22:51 basepi forrest: the release notes currently have 64 bug fixes.  =P
22:51 forrest basepi, pssh slackers!
22:51 basepi ;)
22:51 forrest what bug is the syndic issue?
22:51 joehh forrest: re the patch, no, good point - I thought it was pretty isolated, but thinking about it, it is needed for sid, saucy and raring
22:51 joehh I'll do a pull request
22:52 basepi #7671 is the issue that was fixed, forrest
22:52 basepi basically minions under syndic were almost untargetable, because of a single line of code.  =P
22:52 m0hit joined #salt
22:53 forrest basepi, yea we were talking about a syndic issue earlier this week where you had to have all 3 services installed to get the syndic working
22:53 pipps_ joined #salt
22:53 forrest joehh, I don't think the issue was related. I thought micah_chatt was getting the issue specifically from the package on raring (http://pastebin.com/FzR05Ft8), but it seems like they tried develop as well without luck
22:53 forrest sorry to bug you
22:54 joehh no worries - worth getting in, otherwise corey will be in for a world of pain
22:54 forrest yea fair enough
22:54 forrest good time to get it in too
22:54 joehh yeah that pastebin is unrelated
22:55 forrest Yep I realized that after I looked at it, I hit you up before that was tossed out
22:57 bhosmer joined #salt
22:58 jacksontj joined #salt
23:03 jacksontj joined #salt
23:03 packeteer guys, i noticed this is ps while installing: /bin/sh /var/lib/dpkg/info/salt-master.postinst configure 0.16.0-1
23:03 packeteer is that right ?
23:05 packeteer also, the install failed because I have an override in place /etc/init/salt-master.conf.override
23:05 packeteer edge case i know, but anyway
23:06 packeteer Setting up salt-master (0.17.1-2raring) <- relevant info
23:08 a1j when using http as file source - why not use ETag for source_hash? It is there for you already.
23:08 joehh packeteer: are you using saucy?
23:09 packeteer yes
23:10 joehh the package was copied from raring, after the build failed
23:10 joehh contents should be identical
23:11 joehh 0.16.0 is in ubuntu saucy "proper" - no need for ppa
23:11 joehh so it may have been installed first
23:11 joehh saucy build failed due to minor man page building problem - nothing significant
23:12 joehh packeteer: does it work now it is installed?
23:13 packeteer install is hung. gonna kill it and try again
23:14 packeteer k, works now
23:14 packeteer thx for the assist
23:14 joehh no worries - any idea why it may have been hung?
23:14 packeteer override file
23:14 packeteer /etc/init/salt-master.conf.override  <-
23:15 tty-1 joined #salt
23:15 packeteer this is a test box, so i had the daemons stopped
23:19 bretep So I'm reading the minon config file and trying to understand the include:
23:19 bretep Is it looking for a pillar or salt sls?
23:19 bretep s/salt/state/
23:22 pears bretep: are you talking about this? http://docs.saltstack.com/ref/configuration/minion.html#include
23:22 ajw0100 joined #salt
23:25 khodgson joined #salt
23:26 dfinn this is slightly strange, I'm trying to script going out to each of my minions to restart salt but the salt startup script seems to be hanging onto something, it never disconnects
23:27 dfinn it does this, and then just hangs
23:27 dfinn ssh root@aklogs.vwdl.bcinfra.net "/etc/init.d/salt-minion start"
23:27 dfinn Starting salt-minion daemon: [  OK  ]
23:27 m0hit_ joined #salt
23:27 forrest Is restarting salt with salt supported? I thought there were still issues with that because of how the jobs worked or something
23:28 dfinn i'm not using salt, just ssh
23:28 dfinn i'm having to restart because the minions aren't connecting to the master
23:28 forrest oh ok
23:28 dfinn but this loop won't work if the first client never disconnects
23:28 heewa Ok, I'm trying to debug what's going on with that auth error. I've found that salt master tries to save the minion's public key in the 'pki' did under a sir called 'minions', which for me resolves to '/etc/salt/pki/master/minions/'. It tries to append the minion's id to that path and open that file for writing (or reading in other places in the master.py file). But when it runs, that path is a dir on disk, which means at some point some c
23:28 heewa made that minion id a dir in the pki/master/minions/ dir. But I can't find who's making that a dir.
23:29 dfinn tough day for salt over here today ;(
23:29 forrest dfinn, can you try to use service salt-minion start instead?
23:29 forrest yea I can tell, lol
23:29 forrest lots of issues today
23:29 dfinn i tried that, unfortunately same result
23:29 forrest I miss UtahDave
23:29 dfinn it must be hanging onto the tty or something
23:29 forrest yea that seems very strange
23:30 forrest let me bring up my 0.16.4 minion real quick hang on
23:30 dfinn stop doesn't do it
23:30 dfinn seems to be only start and restart
23:30 forrest the master is up right?
23:30 pears dfinn: does it happen from the shell if you log in normally and don't pass it directly to ssh?
23:30 dfinn this also doesn't work
23:30 dfinn /etc/init.d/salt-minion start && exit 1
23:30 dfinn it does not happen via a normal shell
23:31 higgs001 joined #salt
23:31 pears if you add -t to ssh, does it go away?
23:31 dfinn let me try
23:31 dfinn that works
23:31 * dfinn checks the ssh man page
23:31 pears strange
23:31 pears something terminal-related then
23:31 dfinn forces a tty
23:32 dfinn interesting, possible bug with the startup script
23:32 pears but looking at the init script I can't guess what
23:32 forrest heewa, I guess I don't understand your question, when the minion key is accepted by salt, it creates the required directories for the keys
23:33 dfinn is UtahDave on vaca?
23:33 cro dfinn: No, he's at the OpenStack Summit
23:33 dfinn ah
23:33 heewa The code is trying to open the path for writing, but that path is a dir. Well, the path as the string of the variable has a trailing slash in it, that string doesn't actually exist as a file or a dir on the filesystem.
23:34 forrest dfinn
23:35 dfinn ya?
23:35 forrest try ssh -q server_name "service salt-minion restart"
23:35 forrest that worked for me
23:35 forrest didn't hang
23:35 dfinn you saw that someone else suggested -t above and that fixed it?  i can also test -q if you need
23:35 pears forrest: does it hang for you without -q?
23:36 forrest nope
23:36 pears then your test is invalid :)
23:36 forrest this is on rhel though
23:36 forrest well, I was just suggesting the -q option as a maybe :P
23:36 dfinn i'm on 6.4 running salt 0.17.1
23:37 kermit joined #salt
23:37 dfinn actually, that may have been a 5.x minion
23:37 forrest yea this minion is my old test minion 0.16.3
23:38 forrest dfinn, can you run the process again, wait for it to hang, open a new terminal, then strace/ltrace the process to see where it's hanging up?
23:38 dfinn sure, i just need this for loop to finish now that it's actually working
23:38 forrest oh, what got it working?
23:38 dfinn -t
23:38 forrest weird
23:39 dfinn it forces it to use a tty
23:39 forrest yea
23:39 forrest I remember someone before was having a similar issue
23:39 dfinn sometimes, the salt-minions are very, very slow to start
23:39 forrest where they had to change some configuration in their bashrc because of the tty
23:39 forrest I can't remember what it was now...
23:39 forrest different circumstance, but the tty was screwing them
23:40 forrest heewa, the /etc/salt/pki/master/minions location should be a dir.
23:40 dfinn how often to the minions try to check in with the master?
23:40 forrest it should be created when your minion is join
23:40 forrest *is joined
23:40 forrest jeez, time to go home, can't type any more
23:40 pears dfinn: are other init scripts affected?
23:40 heewa forrest: I just downgraded to 0.17.0 on master & it looks better? I'll keep looking and if  can trace it down to the code I'll try to submit a pull reuqest.
23:41 pipps joined #salt
23:41 heewa But right now I'm putting out a completely separate fire that this is preventing me from battling.
23:41 forrest heewa, sounds good! I haven't seen anyone else with that issue in 0.17.1, so I'm not quite sure
23:41 forrest yea I understand, I've actually gotta head out, I'll be back later tonight though
23:43 heewa OH OH OH! Looks like the minion is sending its id with a trailing slash. Sometimes. Maybe some weird version of the minion code? I dunno, I'll strip it when master gets it.
23:45 troyready joined #salt
23:48 mdasilva joined #salt
23:49 mdasilva1 joined #salt
23:49 Gifflen joined #salt
23:50 mdasilva joined #salt
23:51 Gifflen joined #salt
23:52 mdasilva1 joined #salt
23:53 logix812 joined #salt
23:53 mdasilva joined #salt
23:54 pipps_ joined #salt
23:58 ipmb joined #salt

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