Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2013-08-28

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

All times shown according to UTC.

Time Nick Message
00:07 jaredr Anyone still around?
00:08 jacksontj i'm here-- although i don't work for salt if thats what you are looking for ;)
00:08 chrisgilmerproj left #salt
00:11 jaredr I want to be able to configure individual mogo servers with salt. What's the best way to do the individual-host configuration? Drive per-host config via salt-modules from a script? Or individual sls file per host?
00:12 jaredr substitute "mogo" with "mongo".
00:12 g4rlic I'm going to take a stab and say templates?
00:12 * g4rlic is a salt noob fwiw
00:16 littleidea joined #salt
00:26 dthom91 joined #salt
00:29 TOoSmOotH left #salt
00:32 littleidea joined #salt
00:33 NotreDev joined #salt
00:41 blee joined #salt
00:45 oz_akan_ joined #salt
00:45 jacksontj joined #salt
00:49 krissaxton joined #salt
00:49 jaredr I want to be able to configure individual mogo servers with salt. What's the best way to do the individual-host configuration? Drive per-host config via salt-modules from a script? Or individual sls file per host?
00:50 littleidea joined #salt
00:50 jaredr s/mogo/mongo/
00:58 Teknix joined #salt
00:58 wolfy joined #salt
00:59 wolfy evening
00:59 Guest24627 I have an issue with windows repo
00:59 mgw joined #salt
01:01 Guest24627 I have done a setup appropriatly and issued   salt-run winrepo.genrepo
01:01 Guest24627 and then salt 'mybox' pkg.refresh_db
01:01 Guest24627 where I got a true as response
01:02 Guest24627 when I try to salt 'mybox' pkg.available_version python
01:03 Guest24627 I got nothing, onn minion it telling me that I cannot find the package. However I can transfer the package from same location while using state
01:03 mgw joined #salt
01:04 oz_akan_ joined #salt
01:06 whit joined #salt
01:07 Furao joined #salt
01:09 whit joined #salt
01:11 mahimahi left #salt
01:12 dthom91 joined #salt
01:12 whit joined #salt
01:14 imaginarysteve joined #salt
01:16 whit joined #salt
01:17 mianos joined #salt
01:21 juicer2 joined #salt
01:21 mianos hello, if I want to use  salt.client.LocalClient() to, say run the command: client.cmd('*', 'cmd.run', ['whoami'])
01:22 mianos I want to allow a particular non root user to use all saltstack
01:23 mianos once they logged into the local host, much the same as root but key and sudo but I don't want stuff run as root
01:24 whit joined #salt
01:27 mianos OK got it, needed to find 'client_acl'
01:27 mianos it just looked up the current user id and matches it up, a bit scary but exactly what  I want
01:38 whit joined #salt
01:42 KennethWilke joined #salt
01:42 whit joined #salt
01:48 liuyq joined #salt
01:49 krissaxton joined #salt
01:49 liuyq joined #salt
01:58 malinoff joined #salt
01:59 whit joined #salt
02:05 littleidea joined #salt
02:12 mianos joined #salt
02:12 mianos so nice, when I installed salt it was just me, now training the 3rd staff member
02:13 malinoff Hi all
02:16 dthom91 joined #salt
02:24 xl1 joined #salt
02:28 faldridge joined #salt
02:32 littleidea joined #salt
02:35 vipul joined #salt
02:41 Jahkeup joined #salt
02:50 krissaxton joined #salt
02:52 NotreDev joined #salt
03:00 Guest24627 solution found master was not to latest 0.16 version
03:00 Guest24627 left #salt
03:02 mianos joined #salt
03:12 oz_akan_ joined #salt
03:23 imaginarysteve joined #salt
03:25 mwillhite joined #salt
03:31 Kili joined #salt
03:32 Thiggy joined #salt
03:34 racooper joined #salt
03:39 robertkeizer joined #salt
03:42 mr_chris joined #salt
03:46 Furao joined #salt
03:50 krissaxton joined #salt
03:58 Kilsharion joined #salt
04:35 UtahDave joined #salt
04:37 auser joined #salt
04:50 dthom91 joined #salt
04:51 krissaxton joined #salt
05:03 Furao joined #salt
05:06 mafrosis joined #salt
05:07 Ryan_Lane https://github.com/saltstack/salt/issues/6928 <-- I have a pretty major issue
05:09 UtahDave joined #salt
05:16 robertkeizer Ryan_Lane: Include your versions and particulars. https://github.com/saltstack/salt/blob/develop/Contributing.rst
05:16 Ryan_Lane 0.16.3-1precise
05:16 Ryan_Lane debian
05:16 Ryan_Lane err
05:16 Ryan_Lane ubuntu
05:17 Ryan_Lane debian package :)
05:19 malinoff salt --versions-report
05:19 malinoff And put the output in your issue
05:20 bemehow joined #salt
05:24 UtahDave Ryan_Lane: I saw your email.  When did this start?
05:25 Ryan_Lane well, I'm just noticing it now
05:25 Ryan_Lane I think it's just on the hosts that use the rsync deployment stuff I did
05:26 Ryan_Lane because I ran it on the monitoring system and it returned relatively quickly
05:27 UtahDave how many minions?
05:27 Ryan_Lane but that shouldn't cause salt to stat every file
05:27 Ryan_Lane it would cause rsync to do so
05:27 andyshinn` joined #salt
05:28 Ryan_Lane 22 minions total
05:28 Groovious joined #salt
05:29 Ryan_Lane yep, I just ran a highstate and am doing an strace
05:29 Ryan_Lane and it's stating every single file in the file roots
05:29 Ryan_Lane that's running it on a single node
05:29 Ryan_Lane it's using 100% of the cpu
05:29 UtahDave We've fixed a few bugs recently with the linkedin guys that might address this.
05:30 jchen joined #salt
05:30 UtahDave I think it's checking for changes to push down to the minions
05:30 UtahDave any way you can test against git develop branch?
05:30 Ryan_Lane against every single file in the roots?
05:31 Ryan_Lane I can't, really...
05:31 UtahDave ok
05:31 Ryan_Lane I'm not setup to run development code in production
05:31 UtahDave gotcha
05:31 UtahDave how many files are in your file_roots?
05:31 Ryan_Lane an absurd amount
05:32 Ryan_Lane I should really move most of those out of there
05:32 Ryan_Lane since rsync is actually serving them
05:32 UtahDave And this hasn't been an issue in the past?
05:32 Ryan_Lane this is new
05:32 Ryan_Lane I can't say when exactly it started, though, but I'd imagine within the 0.16 series
05:34 Ryan_Lane well, I'll move this stuff out of the path, since salt isn't really using it anyway
05:34 UtahDave Yeah. I'm trying to think of a bugfix or new feature that might cause this
05:34 UtahDave OK, I'm sure that will help.
05:35 Ryan_Lane oh, its behavior of following symlinks is really bad, btw
05:35 Ryan_Lane and pretty dangerous, at that :)
05:36 UtahDave I'm not sure why it's doing that. I'm pretty sure that has been brought up in the past.
05:39 UtahDave thanks for the bug report. I've marked it high severity
05:39 UtahDave We're hoping to get 0.16.4 released in the next week or so.
05:41 Ryan_Lane thanks :)
05:46 TheCodeAssassin joined #salt
05:49 linjan__ joined #salt
05:50 mianos joined #salt
05:51 krissaxton joined #salt
05:52 sgviking joined #salt
06:16 TheCodeA_ joined #salt
06:22 waverider joined #salt
06:24 pdayton joined #salt
06:26 joehh UtahDave: are there any "automatic" ways to update salt itself on windows?
06:26 * goodwill bellybumps UtahDave and runs off
06:26 joehh are we best to make a salt package and use that?
06:29 * UtahDave high fives goodwill
06:29 UtahDave joehh: good question
06:29 UtahDave So the esky frozen build is supposed to allow for easy upgrading of Salt over the network
06:30 UtahDave I haven't been able to get it to work.  i havent' spent much time trying to get it to work, though
06:30 goodwill UtahDave: oh you know what I want
06:30 UtahDave ok fine
06:30 * UtahDave bellbumps goodwill
06:30 UtahDave :)  how was that?
06:30 goodwill wooohooo
06:30 goodwill excellent
06:30 goodwill great job!
06:30 middleman_ joined #salt
06:31 UtahDave how've you been, goodwill?  Haven't seen you since Pycon
06:31 goodwill just went to PyConCA
06:31 goodwill it was AWEEEESOMEEEE
06:31 UtahDave joehh: also, I've found a very easy to upgrade the Windows Salt minion that could be automated
06:32 UtahDave inside the .egg there's a salt directory.  I delete the contents and replace it with the contents of the salt directory from a freshly clone repo
06:32 joehh UtahDave: that sounds interesting...
06:33 joehh so if I clone the repo down, go to the right branch/tag and drop that in there
06:33 joehh do you know if I can do that via salt or if there are any file locking issues?
06:34 joehh you could even make an installer that "automatically" tracked git like that
06:34 joehh I'll see where I get to. Thanks.
06:36 UtahDave yeah, there's no locking issues.
06:36 UtahDave Just restart the minion after dropping those files in there
06:36 UtahDave It's really easy.
06:36 UtahDave I've been doing a lot of testing and development that way
06:37 joehh nice - thanks for that
06:37 malinoff Hi, nice to see you guys. I need some help again :) I'm trying to mine my minion's data with salt mine. I have one minion with eth0 defined interface, and one with eth1. Both of them has grain 'roles:webnode'. I run salt -G 'roles:webnode' mine.send network.ip_addrs, and i have no response.
06:38 malinoff In minion's log i see such message: 013-08-28 13:30:09,416 [salt.utils.network                          ][ERROR   ] Interface False not found.
06:40 matanya joined #salt
06:42 UtahDave can you run the minion in the cli in debug mode and run that?    salt-minion -l debug
06:45 krissaxton joined #salt
06:46 krissaxton1 joined #salt
06:48 malinoff UtahDave: yes, i can. I will try it.
06:49 malinoff [DEBUG   ] grains target: roles:webnode [INFO    ] User sudo_user Executing command mine.send with jid 20130828134734852719 [DEBUG   ] Command details {'tgt_type': 'grain', 'jid': '20130828134734852719', 'tgt': 'roles:webnode', 'ret': '', 'user': 'sudo_user', 'arg': ['network.ip_addrs'], 'fun': 'mine.send'} [ERROR   ] Interface False not found.
06:50 unicoletti_ joined #salt
06:52 mianos joined #salt
06:54 UtahDave what version of Salt are you on?
06:55 malinoff 0.16.3
06:55 balboah_ joined #salt
06:56 vaxholm_ joined #salt
06:58 kleinishere joined #salt
06:58 UtahDave yeah, i'm not getting results that I would expect.
06:58 UtahDave Can you open an issue on that?
06:59 Koma joined #salt
07:00 malinoff UtahDave: yes, of course. Btw, i see a lot of range(len(collection)) in salt code, why don't you use enumerate?
07:01 UtahDave I don't know. I'd have to ask Tom
07:01 Spads joined #salt
07:01 Gwynn left #salt
07:02 malinoff Ok. If you want, i can pull request with fixes.
07:06 UtahDave malinoff: sure!
07:06 UtahDave enumerate was introduced in python 2.3 or something, right?
07:07 ml_1 joined #salt
07:10 mrpull joined #salt
07:11 malinoff UtahDave: Yes, enumerate appeared in python2.3 :)
07:11 malinoff So, you can definitely use it
07:11 qba73 joined #salt
07:13 malinoff Such lines: "for index in range(0, len(lines))" makes my eyes bleeding.
07:22 Nexpro1 joined #salt
07:23 mastrolinux joined #salt
07:26 bemehow joined #salt
07:27 ronc joined #salt
07:30 jeddi joined #salt
07:35 Psy-Q joined #salt
07:36 Psy-Q since vagrant switched to the "download from our webpage" approach, i can't install the vagrant-salt plugin anymore. it complains of missing mkmf while installing the ffi gem. but that's not quite true, on the base system, i can install ffi just fine in my ruby environment. just not when using vagrant's ruby environment
07:36 Psy-Q what could be wrong?
07:39 carlos joined #salt
07:46 malinoff Psy-Q: *offtopic* May be, Ruby? :)
07:48 Psy-Q malinoff: the salty-vagrant page says to come to freenode #salt if there's any trouble :)
07:48 Psy-Q i think it's because vagrant package their own Ruby environment nowadays
07:48 Psy-Q back in the olden days, it just used my system ruby, then things worked better
07:52 Furao joined #salt
07:53 az87c joined #salt
08:05 Xeago joined #salt
08:07 felixhummel joined #salt
08:09 syngin hmm, misread that as giantcock...
08:13 bemehow joined #salt
08:14 zooz joined #salt
08:15 tomviner joined #salt
08:27 cetex joined #salt
08:28 cetex question: how do i log the commands that's run from salt?
08:28 cetex without logging an enormous amount of info into the log-file?
08:28 cetex like "user X: salt '*' state.highstate"
08:30 malinoff cetex: Do you want to log an output of this command? Or what?
08:31 cetex failed states / minion did not return and such could be nice, but not needed. i mainly want to log who ran what command and when so we can push it to our event-flow.
08:33 malinoff cetex: Can http://docs.saltstack.com/ref/returners/all/index.html fulfill your needs?
08:33 cetex i'll check. :)
08:38 cetex i can't seem to find a way to activate it in the master or minion config.
08:38 cetex is it only commandline?
08:39 malinoff cetex: No, you can specify a returner almost anywhere
08:40 malinoff http://docs.saltstack.com/topics/jobs/schedule.html?highlight=scheduler in scheduler, for example
08:40 cetex ah, nice. so i could implement it in top.sls as well? :)
08:44 ProT-0-TypE joined #salt
08:46 malinoff cetex: Returner is simply "device, where should salt print it's output", but top.sls in mapping between nodes and states. So, no, you can't specify it in top.sls.
08:47 cetex so where would i tell salt to use a returner by default when running any command?
08:48 malinoff cetex: Trying to figure that
08:49 malinoff cetex: But anyway, you will get a standart stdout output even with a returner
08:49 cetex yeah. that's great.
08:51 cetex nothing will change from the salt-users perspective, but we will get an additional output somewhere via a returner which can be pushed to our event-system where we can do stuff with it.
08:51 malinoff cetex: Yep.
08:53 malinoff cetex: Finally i found that: http://docs.saltstack.com/ref/configuration/master.html#ext-job-cache
08:53 Psy-Q maybe vagrant is broken
08:53 Psy-Q i'll see if they have that as a bug
08:54 malinoff cetex: Don't ask me why 'default_returner' option is names 'ext_job_cache'. This is pretty hell.
08:54 cetex that's why i didn't find it. :)
08:54 cetex (after i learned about returners)
08:54 malinoff I've found it through the issue https://github.com/saltstack/salt/issues/3543
08:54 malinoff I'll comment about that
09:00 Psy-Q they fixed the problem in vagrant
09:00 Psy-Q now the plugin installs fine. on to salting my vagrants.
09:02 malinoff Psy-Q: cool :)
09:02 nielsbusch joined #salt
09:16 lemao joined #salt
09:19 aleszoulek joined #salt
09:29 Psy-Q we're trying to saltify all our development virtual machines so that it's easier to create production machines with salt that are very similar
09:36 aleszoulek joined #salt
09:42 mrpull joined #salt
09:55 Furao_ joined #salt
09:55 MrTango joined #salt
10:18 zooz joined #salt
10:18 tomeff joined #salt
10:22 mianos joined #salt
10:23 giantlock joined #salt
10:27 Furao joined #salt
10:28 xl1 joined #salt
10:34 bhosmer joined #salt
10:51 imaginarysteve joined #salt
11:05 jbunting joined #salt
11:05 whiskybar joined #salt
11:07 aleszoulek joined #salt
11:14 unicoletti_ joined #salt
11:18 cetex regarding logging of commands, solved now. copied syslog_returner.py and modified it so it only outputs the parts i want to syslog.
11:21 zooz joined #salt
11:32 ProT-0-TypE joined #salt
11:36 aat joined #salt
11:40 liwen joined #salt
11:59 blee joined #salt
12:02 tuxIO joined #salt
12:16 [diecast] joined #salt
12:24 liwen joined #salt
12:25 qba73 joined #salt
12:29 kenbolton joined #salt
12:31 NotreDev joined #salt
12:33 aat joined #salt
12:37 entil Missing arguments executing "saltutil.sync_all": ArgSpec(args=['env', 'refresh'], varargs=None, keywords=None, defaults=(None, True))
12:37 entil did you guys see that before?
12:37 mwillhite joined #salt
12:37 entil loose references on google, but nothing really clear
12:40 anteaya joined #salt
12:42 liwen joined #salt
12:50 oz_akan_ joined #salt
12:51 NotreDev joined #salt
12:52 gildegoma joined #salt
12:53 Kilsharion joined #salt
12:54 ccase joined #salt
12:54 ipmb joined #salt
13:00 sgviking joined #salt
13:02 pdayton joined #salt
13:02 littleidea joined #salt
13:05 Gifflen joined #salt
13:06 Furao http://www.reddit.com/r/PHP/comments/1l7baq/creating_a_user_from_the_web_problem/
13:07 Furao why salt when you can use php to create user?
13:11 brianhicks joined #salt
13:12 racooper joined #salt
13:13 Psy-Q hell, rvm.install is really tough to get working
13:21 fivethreeo joined #salt
13:23 Psy-Q is there a way to get some debug output from salt-call? installing a ruby on top of rvm just gets me "Could not install ruby."
13:24 alunduil joined #salt
13:24 Furao salt-call -l all
13:24 Psy-Q ah, thanks
13:25 juicer2 joined #salt
13:25 Psy-Q seems to be a user session/shell/$PATH problem
13:25 Psy-Q of course rvm when installed system-wide lives in /usr/local/rvm/bin/rvm, but the rvm state expects it in /root/.rvm/bin/rvm
13:26 tuxIO_ joined #salt
13:29 danielbachhuber joined #salt
13:30 ccase joined #salt
13:31 oz_akan_ joined #salt
13:32 teebes joined #salt
13:36 ggoZ joined #salt
13:39 entil Furao: :D
13:39 Kholloway joined #salt
13:39 entil but that missing arguments thing. I'll dcc send a beer to anyone who has a clue ;)
13:40 entil the stuff on google points to malformed yaml, but it works sometimes, like I could do a highstate regardless of that fail
13:40 entil if I could see where exactly the problem is
13:40 Furao entil: just run saltutil.sync_all
13:41 liwen joined #salt
13:41 Furao no need to specify refresh
13:41 Furao it's true by default
13:41 entil salt@ip-10-34-204-69:~/src/skyhood_salt$ salt -G 'ec2_tags:environment:duunitori-db' saltutil.sync_all --verbose
13:41 entil that was my command
13:43 Furao I don't know, you should look at the minion logs
13:43 Furao run at debug level
13:44 mattmtl joined #salt
13:44 qba73 joined #salt
13:45 entil sure I did that, it's the same, http://pastebin.com/0MT0d30C
13:45 entil the amount of google hits pointing to bad yaml makes me think I need a way to parse all the states and see if there's a glitch somewhere, and see where exactly the glitch is
13:47 Furao pots or mopts is a string
13:47 Furao and show be a dict
13:47 Furao show -> should
13:48 entil yes
13:48 entil but why
13:48 entil how do I trace the source of the problem?
13:48 entil I could run it in pdb but that seems like overkill, and pdb isn't always nice with software like this
13:49 Furao which version of salt this is?
13:49 entil 0.15.3
13:49 entil 0.15.2 even
13:49 Furao the minion will spawn an other process
13:49 Furao it's not fun with pdb
13:49 Furao I use remote debug and pycharm most of the time
13:50 entil but it broke spontaneously also so some kind of "salt render all the states" command would be useful. does one exist?
13:50 Furao https://github.com/saltstack/salt/blob/v0.15.2/salt/state.py#L1549 hard to says
13:50 Furao you should edit that code
13:51 Furao add log.debug("mopts: %s opts: %s", mopts, opts)
13:51 Furao just before 1549
13:51 entil yeah, I can do that, sure
13:51 Furao on the minion
13:51 oz_akan_ joined #salt
13:51 alunduil joined #salt
13:51 entil but you know I was just hoping for something more elegant than that :)
13:52 Furao maybe this had been fixed in 0.15.3
13:52 entil it would be a question of error reporting anyway, because the issue came out of nowhere, which probably means I made a mistake somewhere that I didn't spot
13:53 entil I was thinking about delaying upgrading until 0.17.0 is out
13:53 Furao I'm using only 0.15.3
13:53 Furao 0.16.0 was unusable
13:53 Furao and they replaced plenty of grains key name in 0.16.x
13:53 entil oh
13:53 Furao everything is broken
13:54 Furao so far my 0.16.x branch of my states is already 300 lines of difference
13:54 entil thanks for telling me, I really don't need a broken update anymmore than a hole in my head
13:57 ronc joined #salt
13:58 NotreDev joined #salt
13:59 aat joined #salt
14:02 Psy-Q it fluctuates quite a bit until 1.0.0, i'm guessing?
14:06 cnelsonsic joined #salt
14:07 kaptk2 joined #salt
14:08 bhosmer_ joined #salt
14:09 ccase joined #salt
14:16 faldridge joined #salt
14:20 pdayton joined #salt
14:21 jbunting joined #salt
14:23 [diecast] joined #salt
14:28 jalbretsen joined #salt
14:30 isomorphic joined #salt
14:31 JoAkKiNeN joined #salt
14:32 NotreDev can pillars consist of data outside of the pillar sls files? E.g. can I move .pub keys from my states to the pillar?
14:32 Jahkeup joined #salt
14:33 NotreDev i want to separate private information from implementation data (pillar, state respectively)
14:36 tyler-baker joined #salt
14:36 tyler-baker joined #salt
14:37 mannyt joined #salt
14:43 teskew joined #salt
14:43 jpeach joined #salt
14:46 [diecast] NotreDev use grains for that
14:46 liwen joined #salt
14:46 [diecast] grains are used for static data typically
14:46 jacob_s joined #salt
14:47 abe_music joined #salt
14:48 NotreDev [diecast]: i don't think i understand. i want to ensure that the ssl package is installed, and that certain keys are set in ~/.ssh/; how would i use grains for that?
14:49 [diecast] store the key in the grain
14:53 faldridge joined #salt
14:54 samsalt joined #salt
14:56 SunSparc joined #salt
14:56 toastedpenguin joined #salt
14:56 aleszoulek joined #salt
14:57 ckao joined #salt
15:00 hazzadous joined #salt
15:02 chrisgilmerproj joined #salt
15:05 littleidea joined #salt
15:06 jacob_s I'm using the salt local client to run some commands on a windows salt minion.
15:06 jacob_s From the command line  sudo salt '*' cmd.run 'tasklist' this works fine
15:07 jacob_s but if I run client.cmd('*','cmd.run','tasklist') I get an error
15:07 jacob_s {'vm1qa2': 'ERROR: Attempted to render file paths with unavailable engine s'}
15:09 conan_the_destro joined #salt
15:10 NotreDev joined #salt
15:10 yota joined #salt
15:17 forrest joined #salt
15:25 mrpull joined #salt
15:42 mattmtl joined #salt
15:42 UtahDave joined #salt
15:43 jacksontj joined #salt
15:47 pdayton joined #salt
15:49 devinus joined #salt
15:51 hvn joined #salt
15:52 zetsuboudev joined #salt
16:00 Gifflen_ joined #salt
16:03 lazyguru joined #salt
16:04 Gifflen joined #salt
16:04 Gifflen joined #salt
16:05 deedubs joined #salt
16:07 jacob_s figured it out should be client.cmd('*','cmd.run',['tasklist'])
16:09 juanlittledevil joined #salt
16:11 unicoletti_ left #salt
16:16 kermit joined #salt
16:16 vmdsch joined #salt
16:18 Ahlee whiteinge: Around? new data points on my continued API issue
16:19 Xeago joined #salt
16:26 Ahlee it's cherrypy, and it's refusal to not-print to stdout, so when the controlling terminal goes away it can't print the output.  Leave the user that started cherrypy logged in, and the api functions, as soon as you log out the output can't print, and hte intput/output errors begin
16:28 devinus joined #salt
16:28 KyleG joined #salt
16:28 KyleG joined #salt
16:30 SunSparc joined #salt
16:33 dthom91 joined #salt
16:36 dthom911 joined #salt
16:36 juanlittledevil joined #salt
16:36 SunSparc joined #salt
16:36 UtahDave Ahlee: whiteinge should be here soon
16:38 bhosmer_ joined #salt
16:38 CheKoLyN joined #salt
16:38 CheKoLyN joined #salt
16:39 Ahlee https://github.com/saltstack/salt-api/issues/83
16:41 Ahlee However, I can't get hte cherrypy object to honor log.access_file = None
16:41 hvn left #salt
16:42 mgw joined #salt
16:46 ml_1 joined #salt
16:48 juanlittledevil joined #salt
16:54 brianhicks joined #salt
16:56 wyrd1 joined #salt
16:57 bhosmer_ joined #salt
16:59 zooz joined #salt
16:59 mortis_ how do i do THIS: http://pastie.org/8277596 in away that actually works? :)
16:59 kenbolton joined #salt
16:59 mortis_ away=a way
17:04 jacksontj joined #salt
17:04 Xeago joined #salt
17:04 mmilano joined #salt
17:05 KennethWilke joined #salt
17:05 beardo__ mortis_, I'd use a for loop like:  http://pastie.org/8277615
17:05 mortis_ oooooooooooooh ofc!!!
17:05 mortis_ beardo__: tnx :D
17:06 beardo__ np
17:06 UtahDave There's another easier way using - names.     just a sec
17:07 gildegoma joined #salt
17:08 mortis_ oh - names
17:08 mortis_ ill try that too
17:08 faust joined #salt
17:08 UtahDave mortis_: http://pastie.org/8277632
17:09 UtahDave I also added the group option in there because you had it in the - recurse   list
17:09 mortis_ perfect!
17:10 mortis_ tnx both of you
17:12 ccase How do can I use salt-cloud to select the public_ip? I see -S but I'm not sure how to make it work and I didn't find any examples in the docs.
17:13 alunduil joined #salt
17:14 jaddison joined #salt
17:16 UtahDave ccase: Do you mean to have salt-cloud use the public_ip?
17:17 tdilly joined #salt
17:18 tdilly Are there any docs describing how salt masters and minions communicate? For instance, do the minions poll the master or just wait for commands? Does the master ever poll the minions?
17:20 ccase I'm looking for a way to programatically extract the public ip from salt-cloud so I can then pass that off to ssh.
17:20 KennethWilke tdilly: http://docs.saltstack.com/ref/topology.html kind of describes it a little bit
17:21 Ryan_Lane joined #salt
17:23 tdilly KennethWilke: thanks, I'm trying to determine if salt is going to noticeably increase bandwidth usage if we roll it out to some servers where we are charged for bandwidth.
17:23 NotreDev is there a way to validate my states reference pillar data that actually exists?
17:24 UtahDave ccase: salt-cloud -Q should contain that info
17:24 KennethWilke tdilly: you're welcome! I've never tried to measure it's usage so I'm not sure what to expect there
17:25 KennethWilke my expectation is that since it's using 0mq the overhead on the transmission is pretty low, and that it'll probably come down to things like how files are being distributed and things of that nature
17:26 UtahDave NotreDev: salt \* state.highstate test=True     might help
17:26 devinus joined #salt
17:26 NotreDev UtahDave: hey, that looks worth exploring. thanks!
17:27 tdilly KennethWilke: That's what I am thinking, I think I'm going to roll out one minion and measure usage
17:27 troyready joined #salt
17:30 ronc joined #salt
17:30 KennethWilke has anyone here used the mysql_user state module? mine seems to be setting the password incorrectly somehow
17:32 eculver joined #salt
17:32 eculver joined #salt
17:36 lineman60 joined #salt
17:36 KennethWilke oddly enough the mysql.create_user function from the mysql salt module works fine, but when i try to set the password via the mysql_user.present state module I am not getting the same password hash
17:36 m_george|away joined #salt
17:36 jaredr joined #salt
17:38 renoirb Hey guys
17:38 renoirb is there a way to use a -name with multiple values in a state file?
17:39 KennethWilke that may depend on the specific state module, is there a particular one you're using like pkg.installed?
17:39 mortis_ renoirb: i just asked a similar question, check this http://pastie.org/8277632#2
17:39 mortis_ renoirb: prolly workes for files too
17:39 renoirb thanks!!
17:39 mortis_ np
17:39 renoirb mortis_: this works?
17:39 renoirb I thought, but wasn't sure
17:39 KennethWilke mortis_: do you know if names is ubiquitous?
17:40 m_george left #salt
17:40 mortis_ renoirb: yeah workes for dirs
17:40 mortis_ KennethWilke: i dont know what ubiquitous means lol
17:40 UtahDave yeah, names is ubiquitous
17:40 UtahDave it's provided by the Salt state compiler
17:40 mortis_ :)
17:40 KennethWilke alrighty, mortis_ i mean consistent among state modules
17:40 KennethWilke but the law has spoken
17:41 lazyguru joined #salt
17:41 KennethWilke UtahDave's law
17:41 UtahDave It basically creates an identical low state for each item
17:41 KennethWilke that makes sense
17:41 KennethWilke so for pkg.installed it'd be interchangeable with pkgs?
17:41 UtahDave Yes, but for one important difference
17:41 KennethWilke though if handled as you mean that'd be multiple runs
17:41 KennethWilke vs just a single run with pkgs?
17:41 UtahDave - names will make an apt or yum call for each package
17:42 KennethWilke yeah
17:42 KennethWilke or emerge!
17:42 KennethWilke buddy!
17:42 UtahDave - pkgs makes one apt or yum call
17:42 KennethWilke lol
17:42 KennethWilke but it would be apt on these boxes i'm working on lol
17:44 * KennethWilke facepalm
17:44 KennethWilke figured out my mysql_user state issue
17:45 KennethWilke def a problem with A user
17:45 ipmb joined #salt
17:45 KennethWilke and that user is me
17:47 jbunting joined #salt
17:50 ccase UtahDave: It does, I was hoping I could use -S to select just the ip. :o} I guess I'll just dump it through some yaml parsing. What is the intended use of -S?
17:51 UtahDave I'm not sure.  If you're using Salt Cloud, aren't your minions already connected to a Salt Master?
17:51 NotreDev joined #salt
17:52 UtahDave nice, KennethWilke
17:53 mortis_ what causes salt to say a state is unavailable?
17:54 mortis_ wrong configparams?
17:54 abe_music joined #salt
17:54 brutasse joined #salt
17:55 ccase They are, but they don't have the public ip in their grains. So while I can run salt commands I can't ssh into the boxes without some hoops. fwiw I'm using this to setup a dev environment where the intended use is to then ssh into the box.
17:56 KennethWilke mortis_: i think that often has to do with dependencies
17:56 KennethWilke mortis_, ie, i had that message with the mysql state modules when i didn't have the mysql python lib installed on my minion
17:56 mortis_ ooooh
17:56 mortis_ right
17:57 KennethWilke and remember to bounce your minion after installing the package ;)
17:57 KennethWilke i've forgotten it a couple times and been like, y u no werk!
17:57 NotreDev the return data from running a salt command (e.g. salt-call --local state.highstate) doesn't appear to be yaml. how can i parse this and relate it to the individual state files?
17:57 mortis_ running masterless atm, just trying to set up entire gitlab with salt
17:59 UtahDave ccase: try salt \* network.interfaces
18:00 UtahDave NotreDev: salt-call --local state.highstate --out yaml
18:01 ccase Unfortunately no dice. These are on aws so they only know about their private networks.
18:01 devinus joined #salt
18:02 racooper joined #salt
18:03 UtahDave mortis_: also, make sure your names and paths of your states are correct.
18:04 mortis_ UtahDave: imo this should work http://pastie.org/8277751
18:04 mortis_ OH WAIT
18:04 mortis_ derp
18:05 mortis_ nah nm
18:05 mortis_ it should work shouldnt it
18:05 mattbarto joined #salt
18:06 mattbarto joined #salt
18:07 ipmb joined #salt
18:08 druonysus joined #salt
18:08 UtahDave mortis, there's no git.clone state
18:09 UtahDave http://docs.saltstack.com/ref/states/all/salt.states.git.html#module-salt.states.git
18:15 tuxIO joined #salt
18:16 mortis_ ah its a module
18:17 mortis_ i was on this http://salt.readthedocs.org/en/v0.16/ref/modules/all/salt.modules.git.html
18:17 mortis_ :)
18:23 ml_1 joined #salt
18:24 auser joined #salt
18:26 TheCodeAssassin joined #salt
18:28 liwen joined #salt
18:29 kvt joined #salt
18:30 mwillhite joined #salt
18:33 jaredr joined #salt
18:36 NotreDev joined #salt
18:39 StDiluted joined #salt
18:40 johntron joined #salt
18:40 johntron hey guys, my minion isn't finding a topfile from my master: https://gist.github.com/johntron/6369651 any ideas?
18:42 johntron i just added the output from running salt-call on the minion: https://gist.github.com/johntron/6369651
18:45 UtahDave johntron: where's your top.sls?
18:46 johntron /etc/salt/roots/salt/top.sls (on master)
18:46 johntron i can add the output from `salt-master -l debug` to the gist if you'd like
18:46 UtahDave johntron: there's your problem.  your file_roots is defined as /etc/salt/roots
18:47 johntron UtahDave: oh, should it be /etc/salt/roots/salt?
18:47 UtahDave Salt looks for the top.sls in the root of your file_roots
18:47 johntron oh ok! I was trying to find that information
18:47 johntron thanks
18:48 johntron I was also trying to figure out what i'd grep from `salt-master -l debug` to tell if it's finding a top file
18:49 johntron hmm, still not finding a topfile though
18:51 Xeago joined #salt
18:52 UtahDave did you restart your master after changing the config?
18:53 johntron yeah
18:54 UtahDave johntron: /etc/salt/roots/top.sls    ?
18:54 johntron i just realized i'm running 0.16.0 on the minion - running the bootstrap
18:55 bauser joined #salt
18:56 johntron UtahDave: top.sls is at /etc/salt/roots/salt/top.sls, and master now reads file_roots: base: /etc/salt/roots/salt
18:58 gildegoma_ joined #salt
18:59 UtahDave johntron: That should work, I think.
18:59 UtahDave restarted master and minion?
19:03 TheCodeAssassin joined #salt
19:04 juanlittledevil joined #salt
19:04 robertkeizer joined #salt
19:07 johntron UtahDave: yeah, but now the minion doesn't seem to be able to reach the master
19:07 whiteinge Ahlee: hi. did you try setting ``log.screen = False`` ?
19:09 whiteinge https://github.com/saltstack/salt-api/compare/cpy-logging
19:09 aat joined #salt
19:10 faldridge joined #salt
19:12 StDiluted can i list multiple gems in a gem.installed: state with 'names:'
19:13 UtahDave you can use - names in any state if you want an identical configuration for each one
19:16 aat joined #salt
19:16 Ahlee whiteinge: I thought so, but apparently not.  Trying now
19:18 StDiluted so like gems:\n gem.installed:\n - names:\n - bundler\n - whenever\n
19:18 StDiluted or can i just list them under gem.installed:
19:18 UtahDave like your first example, StDiluted
19:19 StDiluted ok cool
19:19 whit joined #salt
19:19 StDiluted and would i watch that with watch:\n - module: gems
19:20 Ahlee whiteinge: issue remains
19:21 jacksontj joined #salt
19:21 UtahDave StDiluted: no, you watch the state and the ID declaration or name
19:21 StDiluted id delcaration is gems:
19:21 UtahDave watch:\n  - gem: bundler
19:21 UtahDave ok
19:21 StDiluted but if i want to watch the whole list of gems I'm installing?
19:21 whiteinge Ahlee: are you seeing that it's not outputting to the screen but you still get the IOError?
19:21 UtahDave watch\n  - gem: gems
19:22 Ahlee whiteinge: putting together pastebin now
19:22 StDiluted ah ok
19:22 StDiluted cool
19:22 UtahDave remember, the "watch" or the "require" always starts with the state being used, then the ID dec or name
19:22 StDiluted ok. That makes sense
19:23 UtahDave and actually, now that I think about it, if you're using - names   I'm pretty sure you can only watch or require against the ID dec
19:23 StDiluted yeah that's what i would think
19:25 Ahlee was started with --log-level debug, trying again without to make sure that didn't somehow propagate in
19:29 qba73_ joined #salt
19:30 johntron man, i can run salt-call on the master and it starts provisioning, but calling it on the minion just ends shortly after "Fetching file ** attempting ** 'salt://top.sls'"
19:31 whit joined #salt
19:31 dthom91 joined #salt
19:32 jaddison left #salt
19:32 johntron i can see the master responding to requests from the minion, but it still cannot seem to find the top.sls
19:33 Ahlee I messaged you whiteinge, I hope that's acceptable
19:35 UtahDave johntron: I'd recommend sticking with all the path defaults until you know you can get it working correctly.
19:35 UtahDave johntron: then if you really must have paths be different start slowly
19:36 johntron UtahDave: I did, that's why I'm trying other paths - the default paths didn't seem to work
19:36 johntron I believe my files are still located at the default installation (at least on Ubuntu - not sure about OS X)
19:37 johntron s/installation/location/
19:37 StDiluted so UtahDave: if I want to make sure that a particular script ran before I do another state, I can include the previous state with it's cmd.script state and then require:\n - cmd: declaration ID from that previous state?
19:37 UtahDave Well, the defaults should work. Something else must have been causing you some issues
19:38 UtahDave StDiluted: exactly
19:39 johntron UtahDave: Yeah, that's what I was thinking, but there are no obvious errors in the logs, and I don't know enough about salt to know if it's running correctly
19:39 johntron by logs, i mean when using `-l debug`
19:39 StDiluted and if I include the state in that state, that will cause whatever was in the previous state to happen as well, yes, so if I have state 1, which installs X, and State 2, which installs Y, and I include state A in state B, and then include state B in top.sls, both X and Y will get installed?
19:40 UtahDave johntron: sure.  I'd recommend going back to the default configs.  The only change you should need to make is in the minion config   /etc/salt/minion   just change  the "master" option to the ip or hostname of your master
19:40 UtahDave yep! StDiluted
19:41 StDiluted ok cool
19:41 StDiluted thanks
19:41 StDiluted I've had some states that get run multiple times in a highstate and am trying to weed them out.
19:41 johntron ok, i'll try it, but the only other thing I have besides that is my grains
19:42 StDiluted have there been requests for compiling the list of actions first and then culling duplicate actions?
19:44 UtahDave StDiluted: duplicate actions shouldn't happen.
19:44 StDiluted UtahDave, even if a state is included in multiple other states that get run in the highstate?
19:45 StDiluted because it seems to me that my 'users' state which is included in a few states as I require certain users to be there in more than one state, gets run multiple times
19:45 UtahDave correct.  Everything gets compiled into one big dict.  So even if they're split up into different sls files, in the end Salt just sees one big dict
19:45 StDiluted ok
19:52 Xeago joined #salt
19:53 mwillhite joined #salt
19:56 NotreDev joined #salt
19:59 jacksontj joined #salt
19:59 mattbarto joined #salt
20:01 m_george|away joined #salt
20:05 ninkotech__ joined #salt
20:06 johntron arrrrg
20:06 johntron UtahDave: I figured it out
20:06 UtahDave what was it?
20:06 johntron in my master's file_roots for base, i had: roots/salt and it needed to be: - roots/salt (with a hyphen)
20:07 johntron there was no error or anything, so i figured it was reading the option correctly
20:08 johntron UtahDave:
20:10 johntron https://gist.github.com/gists#file-master-L7 needed a hyphen
20:10 nbari joined #salt
20:10 nbari joined #salt
20:11 johntron * https://gist.github.com/johntron/6369651#file-master-L7
20:11 nbari hi all, I just instaled salt and giving a try, I have already setup a master and minion, but for now I just want to make the minion, do a git clone of a project
20:11 druonysus joined #salt
20:13 nbari i creatd a file 'dalmp.sls' containing this: https://github.com/nbari/DALMP: git.latest: - rev: master  - target: /usr/local/share/dalmp
20:13 nbari later just run salt '*' state.higstate but get a message saiing State - no
20:13 nbari any ideas ?
20:14 johntron nbari: if you haven't setup the master yet, you need to use salt-call --local
20:14 johntron salt-call --local state.highstate
20:14 UtahDave nbari: if you want to use a highstate you have to set up your top.sls
20:15 UtahDave nbari: otherwise right now you can try    salt '*' state.sls dalmp
20:15 nbari mmm I still getsometingin red saying State: -no
20:15 johntron nbari: here's some info on running masterless: http://docs.saltstack.com/topics/tutorials/quickstart.html as dave says, add a top.sls
20:15 nbari ok
20:16 johntron UtahDave: so salt '*' state.sls dalmp would just require the minion process to be running?
20:16 johntron whereas salt-call --local would not?
20:16 johntron sweet! system is up and running :) thanks UtahDave!
20:17 nbari dalmp.sls" is not available.
20:17 nbari mmm
20:17 nbari dalmp.sls is in /user/local/etc/salt/states/
20:18 johntron nbari: where's your minion file?
20:18 jbunting left #salt
20:18 UtahDave nbari: why is dalmp.sls there?  It needs to be in your file_roots
20:19 nbari mmm I don't have a minion file
20:19 cedwards UtahDave: that is the file roots on FreeBSD
20:19 nbari my file_roots: base: /usr/local/etc/salt/states/
20:19 nbari right it is a FreeBSD
20:20 cedwards nbari: you should have a minion.sample in /usr/local/etc/salt/
20:20 nbari both server/minion FreeBSD, salt '*' test.ping works :
20:20 nbari right I have one but on the server only using master
20:20 nbari and on the minion /user/local/etc/salt/minion
20:21 juanlittledevil joined #salt
20:21 nbari but I had an idea that I coud just configure the states.sls on master and that send the order to the minons
20:22 cedwards and you say you have /usr/local/etc/salt/states/dalmp.sls?
20:22 cedwards you shouldn't really need to configure anything on the minion-side
20:23 nbari i will put the top.sls also in /user/local/etc/salt/satates/
20:24 vipul joined #salt
20:26 nbari salt-call --local state.highstate working :)
20:26 nbari now how to make the same on the minion ?
20:26 cedwards you should be able to salt-call state.highstate on your minion as well
20:27 cedwards or state.sls dalmp..
20:27 nbari but from the master how can I send the order without need to enter to minion?
20:28 cedwards nbari: if things are working now, you should be able to 'salt $minion state.highstate'
20:28 NotreDev what exactly would i use salt's event system for? i might have a better clue if i could get it to work with a standalone :) http://docs.saltstack.com/topics/event/index.html
20:28 nbari wow this just worked :)
20:29 nbari salt '*' state.highstate
20:29 nbari :D
20:29 cedwards nbari: as long as you've got the files in the right places :)
20:29 cedwards nbari: if you run into any FreeBSD-related salt questions feel free to hit me up.
20:29 nbari many thanks
20:29 cedwards nbari: I maintain dozens of jails using Salt on my FreeBSD servers.
20:30 nbari great, I just start using salt to send from a master to a jail
20:31 nbari one question related to the git with salt, if later I want to do a pull not a clone, salt handles that automatically ?
20:31 forrest with gitfs nbari?
20:31 cedwards forrest: i think he's using git.latest in a state
20:32 nbari mmmm don't know if iis gifs, my dalmp.sl is this: http://pastebin.com/qxtYf1AH
20:32 forrest the states class supports pulling as opposed to cloning
20:32 nbari right I am using git.latest
20:33 cedwards I would assume the next time you make a commit salt will detect you have an existing checkout and do a pull
20:33 forrest line 108 here nbari: https://github.com/saltstack/salt/blob/develop/salt/states/git.py
20:34 nbari perfect :D
20:34 nbari cedwards: related to FreeBSD how do you deploy multiple jails / hosts ? you compile on a single server and share your packages or build on each jail ?
20:35 cedwards nbari: I use poudriere to build a package repository and subscribe my jails to it.
20:36 cedwards nbari: It really is pretty simple to get setup. We even have salt packages (pkgng) available at http://freebsd.saltstack.com
20:36 nbari oh ok
20:36 cedwards nbari: I used to use portmaster and build on each host, but that really doesn't scale very well.
20:36 devinus joined #salt
20:37 auser be back soon
20:37 nbari right, currently I am using portmater on each jail/host but is tired to repeat a lot the task
20:37 cedwards nbari: these steps are pretty easy: https://wiki.freebsd.org/PkgPrimer#Building_custom_repositories
20:38 ipmb joined #salt
20:38 nbari and what about the thunderxxxx I forgot the name, but I think is similar to poudriere
20:39 cedwards nbari: those instructions assume ZFS, but if you don't have that you can toggle it off in the config.
20:39 cedwards nbari: tinderbox?
20:39 nbari that one
20:39 nbari right
20:39 cedwards nbari: i haven't used tinderbox, but i've heard it's much more complicated
20:39 alunduil joined #salt
20:40 nbari I see, I will give a try, since I have a micro (ec2) instance and is a pain to complie things there
20:40 cedwards nbari: as something to think about for later.. if you do much work with ec2 you might look into salt-cloud as well.
20:41 nbari but if I am right that is for creating instances right ?
20:41 cedwards yes
20:41 cedwards creating and automagically 'salt'ing
20:42 nbari well I still "micro" for now jails are ok
20:43 cedwards nbari: do you use ezjail?
20:43 nbari no, I use jails2 and zfs
20:43 nbari I create one base jail and clone it N times
20:44 nbari ( I have less than 10 jails ) but anyway wanted to give a try to salt
20:44 cedwards well good luck :)
20:45 cedwards let us know if you have more questions
20:45 nbari ok many thanks
20:45 nbari one last question related to frebsd, do you have a specific deployment schema ?
20:46 nbari or just install packages in their default destination ?
20:46 cedwards i just install packages to defaults
20:46 nbari ok
20:47 nbari well thanks again, bye for now
20:47 forrest Hey UtahDave, do you know if there are ever plans to get travis CI integrated into saltstack-formulas?
20:47 zach Can someone let me know if I'm on the right track here please: https://p.6core.net/p/Rh1g7PKu20c8CkJR5KXUe8N3
20:47 johntron left #salt
20:48 UtahDave forrest: we're standing up our own jenkins server that uses Salt cloud to run our tests on more than just Ubuntu (which is all Travis runs on)
20:48 forrest oh nice
20:48 forrest that will be sick
20:48 UtahDave We have plans to get the saltstack-formulas on there, but we're still getting the formulas sorted out and organized
20:49 zach jenkins++
20:50 UtahDave Yeah, we'll be able to say, "Look! All of our SaltStack Formulas are tested EVERY NIGHT against 3 versions of Ubuntu server, RHEL 5 & 6, CentOS 5 & 6, latest Fedora"  and whatever other OSes we have demand for
20:50 forrest Yea that's a big plus
20:50 talso joined #salt
20:50 forrest especially when you have people contributing for an OS they aren't familiar with (or trying not to break some other OS)
20:53 Ryan_Lane joined #salt
20:53 forrest So will salt cloud only support providers, or will it eventually support stuff like vmware?
20:54 cedwards forrest: i think salt-cloud is for public and salt-virt is for private
20:54 forrest oh duh you're right
20:56 UtahDave Salt Cloud will support vmware very soon.
20:56 aleszoulek joined #salt
20:56 UtahDave Salt Cloud already supports openstack, parallels and eucalyptus
20:56 UtahDave we're just working on access to vmware so we can get it working
20:56 forrest awesome
20:57 forrest yea we're using vmware so if I could hook that up, another win for salt against the encroaching red hat puppet support :P
20:58 druonysus joined #salt
21:01 jacksontj joined #salt
21:01 UtahDave forrest, what vmware products are you all using?
21:02 forrest our entire rhel infrastructure runs on vmware
21:02 forrest hosted in house
21:02 forrest so I don't know what product that qualifies as to be honest
21:02 forrest that's another team
21:03 forrest Pretty sure it's vsphere.
21:03 UtahDave gotcha
21:04 forrest vmware has so many products now, I don't know what add-ons we're using :\
21:04 forrest so maybe vsphere plus some other stuff, who knows.
21:04 faldridge joined #salt
21:04 UtahDave yeah
21:05 forrest Did you get a chance to look at that white paper on satellite 6?
21:09 zach UtahDave: that would be great to get VMware support. That's what my entire infrastructure (800 VMs) is built ontop of.
21:09 zach It would have been very useful for when we cloned everything to do our RHEL5 -> RHEL6 migration (which is still ongoing)
21:10 Ahlee possible to specify a grain in a state file? like file.managed: match grain foo?
21:10 zach {% if grain['stuff'] ' blah %} ?
21:10 Ahlee that'll play
21:12 Kilsharion joined #salt
21:12 zach http://docs.saltstack.com/ref/states/all/salt.states.grains.html
21:12 zach Not much info but that's all I can find
21:12 zach and the modules one
21:12 zach http://docs.saltstack.com/ref/modules/all/salt.modules.grains.html
21:12 forrest yea I agree zach, it would be awesome
21:13 p3rror joined #salt
21:13 forrest we still haven't settled on a configuration management tool, so the more push I can get behind something like ansible or salt, the less ruby I have to deal with.
21:15 zach I told my boss "We're using salt."
21:15 zach "Okay I guess we're using salt then." - bossman
21:15 cedwards ..if it were only always that easy
21:16 zach If only I could convince my boss to move our infrastructure to something in-house...it would be even easier ;)
21:16 forrest ahh yea it doesn't go like that for me at all.
21:16 zach budget problems, mo money, mo problems
21:16 forrest more like 'we should use config management' and then "red hat satellite is config management!"
21:16 SEJeff_work I'm kind of lucky in that regard
21:16 SEJeff_work budget? What budget
21:16 zach can't afford to double our budget for two months for an infrastructure move
21:16 SEJeff_work if it helps you do your job better... do it
21:17 zach $160,000/mo hosting bill already
21:17 forrest I'm pretty sure SEJeff works for the government
21:17 forrest 160k/month???
21:17 SEJeff_work We spent over 150k for 3 graphite caches
21:17 zach yessir
21:17 forrest screw that man
21:17 SEJeff_work forrest, I work for a finance company
21:17 forrest ahh ok SEJeff_work
21:17 zach I know, I am trying to get that lowered
21:17 zach I work for a financial company as well ;-)
21:17 SEJeff_work And we pay millions per month in hosting
21:17 zach We add $10-20k every other month onto our bill
21:17 forrest how can you work for a finance company and they don't understand the benefits of paying two months now, to save over the next year or more?
21:18 zach forrest: it's a different type of financial company in that sense
21:18 SEJeff_work My company is high frequency trading
21:18 SEJeff_work We are a tech company to the core and make money in finance
21:18 zach financial services is the genre of company i guess you could say
21:18 zach we deal with id theft mainly
21:18 SEJeff_work We have 0 clients, but billionaire owners :D
21:18 forrest every company at the core is a tech company
21:18 SEJeff_work no no no
21:18 zach politics
21:18 zach is the core fo EVERY company
21:19 forrest lol
21:19 SEJeff_work yes
21:19 SEJeff_work Some companies understand tech
21:19 SEJeff_work most do not
21:19 SEJeff_work very few companies are tech companies at their core
21:19 forrest oh I am well aware of that
21:19 forrest I mean whether they want to be or not, tech is at the core
21:19 SEJeff_work Apple is a design company that has very good tech for instnace
21:19 forrest even if they don't get it
21:19 SEJeff_work but I'm talking about companies that *do* get it :D
21:20 forrest *looks around*
21:20 zach so yeah...my hosting bill sucks. especially for such a small ifrastructure
21:20 forrest how many VMs? 800?
21:20 zach we have 3 racks in 2 locations
21:20 forrest that's it?
21:20 zach 6 racks total, yes
21:20 forrest 3 racks per location?
21:20 g4rlic y u no aws? :D
21:21 zach it's hardly anything
21:21 zach g4rlic: compliance
21:21 forrest ahh PCI
21:21 forrest so much fun, we have that here too
21:21 zach ISO 27001, HIPPA, PCI-DSS, amongst other things
21:21 g4rlic zach: I figured, I'm just funnin' around.
21:21 forrest makes me want to shoot myself.
21:21 forrest are they in a cage?
21:21 forrest made of chicken wire
21:21 zach they're with Rackspace - that's the real cost
21:21 g4rlic oooh.. :\
21:21 forrest ahh ok
21:21 forrest there's usually some rackspace guys in here
21:22 zach I'm trying to get them colo'd locally and use RS purely for DR
21:22 forrest is rackspace charging you for taxes?
21:22 zach would cut our bill down quite a bit
21:22 forrest you know they are rolling that out over the next few months right?
21:22 zach yeah, like $20k/mo
21:22 forrest for sales tax or whatever based on your state
21:22 zach 12*
21:22 forrest yea they just did that for me
21:22 forrest so I moved
21:22 zach we're inc'd in TX
21:23 forrest I don't know, with RS you're paying for support
21:23 zach So yeah, we have to pay
21:23 forrest I don't need support
21:23 zach neither do we
21:23 g4rlic Then why RS?
21:23 zach predates me
21:23 zach ;)
21:23 forrest Because it was the best option at the time when they were looking probably
21:23 * g4rlic lols
21:23 zach pre-dates my entire IT operations group
21:23 g4rlic my money is that someone at Rackspace made a great sales pitch to someone who didn't know better. ;)
21:23 zach most likely, yes
21:24 zach 100% sure of that
21:24 zach Considering we have 30 dom0's with 4 dedicated netapp filers with like 8 shelves each
21:24 g4rlic damn.
21:24 forrest lol
21:24 forrest but you only have 6 racks?
21:24 zach yep, 2 racks in each location is storage
21:25 zach 1 rack of servers
21:25 zach they're in a DR configuration, so mirrored between locations
21:25 forrest yea my stuff at rackspace was just a personal VM
21:25 g4rlic 2 racks of NetApp?  Man.  That must've cost a pretty penny on its own.
21:25 zach yeah, that's 60% of our bill
21:25 forrest are you pushing those things??
21:26 forrest I'm just having a hard time seeing that happening, I used to work for a company with thousands of physical machines, and we used netapp filers
21:26 Eugene Still 7-Mode?
21:26 g4rlic As do I currently.
21:26 g4rlic We have one filer head and one shelf in each DC
21:26 g4rlic with ~1000 nodes talking to each one
21:26 g4rlic as long as things are mounted with noatime option, the CPU's are barely registering any load.
21:28 zach forrest: yeah we max those netapps out...we are trying to get away from them
21:28 * Eugene puts on his NetApp hat
21:28 zach g4rlic: we do mainly data processing / mysql - we kill the IO on those things
21:28 Eugene What's not scaling?
21:29 zach I actually had to build a new dedicated machine to handle our production DB load
21:29 zach 24 cores, 256gb ram, connected to a VNX with 48x600GB SAS 15k
21:29 zach had to use so many disks to get our IOPS requirements for the DB
21:29 g4rlic Damn.
21:30 g4rlic iSCSI or NFS?
21:30 zach the VNX is connected with FC, the netapp uses NFS (which is where we are pointing the finger for performance)
21:30 Eugene FlashCache works great for cheap IOPS
21:30 Eugene Unless you have old heads, anyway
21:30 g4rlic Which is odd, because NEtApp was the first time I've ever seen NFS + MySQL not fall over in seconds..
21:31 forrest Yea I don't like netapp
21:31 zach g4rlic: keep in mind...it's inside of a vdk (vmware)
21:31 g4rlic I do. ;)
21:31 forrest way too expensive
21:31 zach container ontop of container ontop of container
21:31 g4rlic Ugh.
21:31 zach yes, it's gross
21:31 g4rlic No wonder you're having small block I/O problems.
21:31 zach if we NFS'd the netapp directly to the VM for MySQL it would be great
21:31 g4rlic (which I'm imagining your MySQL path looks like)
21:31 Eugene Oh, ew. No wonder.
21:31 dthom91 joined #salt
21:31 oz_akan_ joined #salt
21:31 zach indeed
21:32 zach not my design
21:32 g4rlic Wait a sec..  you're running mysql in a VM under ESXi?
21:32 forrest netapp is just so expensive
21:32 zach g4rlic: lol. yep.
21:32 forrest :\
21:32 g4rlic I'm sorry man, I Feel that pain.
21:32 zach yep.
21:33 g4rlic I've had to do some interesting tuning experiments on old Ubuntu+KVM dom0's in my last job, made me decide never to virtualize databases.  (or anything I/O heavy, if I can avoid it.)
21:33 forrest yea
21:33 forrest that's not a good plan
21:33 g4rlic I've seen it work, but you have to have a good performance baseline on the original system, to know if it will virtualize well.
21:34 g4rlic Capacity planning for VI isn't as easy as I'd hoped 5 years ag.
21:34 forrest a friend at my old job moved their entire DB infrastructure over to SSDs at least
21:34 g4rlic ago.
21:34 forrest ended up being cheaper per gig than normal drives.
21:34 g4rlic Eh, that's just buying time..
21:34 forrest this was on physical machines, not virtual
21:35 g4rlic Oh I know..  We've got a, well, massive hadoop grid that runs on all physicals with a boatload of SSD's per chassis.
21:35 g4rlic But my opinion still stands that moving to SSD for "performance" is a cop-out.
21:35 forrest I agree
21:36 forrest but my old shop was hosting
21:36 forrest so there's only so much you can do to tune the DB of a customer
21:36 g4rlic Fair point.
21:36 g4rlic We're the only customer of our internal services, thankfully.
21:36 g4rlic So when someone does somethign stupid, I have someone to berate and shame.
21:37 g4rlic (someone who's not paying us.)
21:37 forrest Yea in that case the SSDs are a cop out
21:38 g4rlic We do have some external SLA's to hold up to though..
21:38 g4rlic I believe one of our services has a 160ms SLA...
21:38 forrest that's pretty tight, jeez
21:38 g4rlic Most of those calls run through fairly beefy systems that are SSD backed.
21:42 giantlock joined #salt
21:45 TheCodeAssassin joined #salt
21:49 NotreDev is `extend` lazy?
21:50 NotreDev -> is `extend` a lazy way of updating a state?
21:52 forrest not really
21:53 forrest it's so you can make a change to something that's already included in another sls
21:53 forrest which is friggin amazing if you've used puppet
21:53 forrest http://docs.saltstack.com/topics/tutorials/states_pt3.html#extend-declaration
21:53 berto- joined #salt
21:54 forrest is an awesome example, where you have your base where all it does is install apache, so you could include apache on every machine, but then on your web server or wherever where you make a lot of changes, you could have it watch your vhost file
21:54 berto- is it possible to access __salt__ or __grains__ within a pillar's run() function?
21:55 mafro joined #salt
22:02 UtahDave berto-: look at the bottom of this page: http://docs.saltstack.com/topics/development/external_pillars.html
22:02 kula berto-: yes.
22:04 Ahlee complex matching of grains, isn't it -C "G@grain1:foo G@grain2:bar" ?
22:04 Ahlee er, compound, not complex
22:04 UtahDave I think you need an "AND" in there
22:05 kermit joined #salt
22:05 berto- UtahDave: so, rather than calling the method `run()` call it `ext_pillar()` ?
22:05 berto- kula: thanks.
22:05 jaredr joined #salt
22:06 kula berto-: yes. the ext_pillar function is what's actually called to generate the pillar data.
22:06 berto- hmm, ok, i need to update.  i've been using run() for a while.  when was ext_pillar() introduced.
22:06 bhosmer_ joined #salt
22:07 kula i don't know. ext_pillar has been around since at least 0.14, since that's about the time i wrote that documentation.
22:09 ronc joined #salt
22:09 jaredr I want to be able to configure individual mongo servers with salt -- port numbers, peers, replication set number, etc. What's the best way to do the individualized-host configuration? Do it with "salt cmd.run" from a script? Or something else?
22:10 berto- kula: great, thanks.
22:11 jaredr I ask, because I was trying to use pillars, and including more than one pillar with the same key names, but pillar data isn't additive -- it clobbers what was already there if another pillar file defines the same data.
22:11 robertkeizer left #salt
22:12 forrest what do you mean by the same key names jaredr? Doesn't every host have a unique name?
22:12 forrest and then you pull the grain data from each unique machine?
22:14 kula jaredr: you can either create different pillar files, and use pillar's top.sls to say "this machine gets that file, that machine gets that file", etc. or you can generate a pillar that creates a datastructure where the options are keyed off of something like the minion id
22:15 kula it might be helpful to see an example of your pillar data so we can try to come up with an example that works better.
22:16 chrisgilmerproj left #salt
22:17 isomorphic joined #salt
22:20 antsygeek joined #salt
22:30 littleidea joined #salt
22:33 txmoose joined #salt
22:33 aptiko joined #salt
22:37 forrest Do any of you debian guys know off hand what the command is to update a repository?
22:37 forrest similar to repo-update on rhel
22:38 racooper never heard of repo-update
22:38 StDiluted When I run a state.highstate, and it's taking a while (ruby compilation, etc)
22:39 StDiluted it returns and there's no data
22:39 StDiluted and jobs.active shows no jobs
22:39 StDiluted but if I go onto the highstated machine, it's running everything
22:39 StDiluted and the compiles are running
22:39 StDiluted etc
22:39 g4rlic forrest: apt-get update ?
22:39 forrest sorry racooper, it would be createrepo
22:39 g4rlic oh
22:39 forrest apt-get update is to update packages
22:40 g4rlic I misread.
22:40 StDiluted is there a way to get state.highstate to block and return the highstate data, or to obtain the highstate data after the fact?
22:40 forrest I'm talking about actually rebuilding your repository and the associated xml file or whatever the equivalent is on ubuntu
22:40 Ryan_Lane forrest: it's the same thing
22:40 Ryan_Lane apt-get update
22:41 Ryan_Lane is yours corrupted or something?
22:41 forrest I'm talking about on the location you're pulling the actual package from
22:41 forrest no
22:41 druonysus joined #salt
22:41 druonysus joined #salt
22:41 Ryan_Lane yeah, apt-get update updates the repo locations and such
22:41 forrest no not the repo location
22:41 forrest I mean if you have a local repository of packages
22:41 Ryan_Lane then you need to add it to the config file, then apt-get update
22:42 forrest you have to update the xml data so that the systems that are pulling from you know those updated packages exist
22:42 UtahDave StDiluted: you can get the full results of the job after the fact
22:42 oz_akan_ joined #salt
22:42 Ryan_Lane the config file is /etc/apt/sources.list and /etc/apt/sources.list.d/*.list
22:42 racooper forrest,  sounds like you are wanting to create a debian repository?  'fraid I'm no help then
22:43 forrest yea
22:43 forrest a debian repo
22:43 Ryan_Lane oohhh.
22:43 Ryan_Lane look at reprepro
22:43 forrest ahh thank you
22:43 Ryan_Lane yw
22:43 forrest I couldn't remember what it was for the life of me
22:43 forrest googling key terms was completely failing
22:43 StDiluted UtahDave: I found the jid, and a jobs.lookup_jid <jid> doesn't return any data
22:45 UtahDave what os are you on StDiluted?
22:45 StDiluted ah, now it works, I think it finally finished.
22:45 StDiluted it had not returned the data to the master yet
22:45 g4rlic Ryan_Lane: got a moment to answer an off-topic convo in PM?
22:45 StDiluted is there a way besides poking through the job list to get a jid when you do a highstate?
22:46 Ryan_Lane g4rlic: sure.
22:46 UtahDave StDiluted: add -v
22:46 StDiluted ok cool
22:46 StDiluted thanks
22:46 UtahDave you're welcome.
22:46 UtahDave also, you can query the minion to see what jobs it has running
22:49 hazzadous joined #salt
22:57 KennethWilke joined #salt
23:01 ronc joined #salt
23:03 mgw does anyone have experience using pgrep/pkill with cmd.run?
23:06 mianos joined #salt
23:06 dthom91 joined #salt
23:06 mgw in particular, I'm using the -f flag, and it is causing it to match the salt cmd itself it appears
23:07 matanya joined #salt
23:07 isomorphic joined #salt
23:11 mianos left #salt
23:15 bhosmer_ joined #salt
23:16 lineman60 joined #salt
23:17 mgw It looks like psutils has some kill support, but it is not wrapped by the salt ps module. UtahDave: Would it make sense to a) add this and b) create a state that can ensure a process (by pattern) is not running?
23:17 whit joined #salt
23:22 UtahDave mgw: sounds reasonable to me
23:22 jacksontj joined #salt
23:26 mgw UtahDave: looking over the doc strings, it seems the examples usually use '*' as the target. I'm adding a kill_pid to ps module, which really only makes sense on a specific target.
23:26 mgw What should I use in the doc string?
23:27 UtahDave salt '<minion id>'  bla.bla
23:27 UtahDave that's what I usually do
23:27 UtahDave but you can put whatever you want.  I'd just make it obvious it's a minion name
23:27 DredTiger joined #salt
23:30 druonysus joined #salt
23:42 teebes joined #salt
23:45 cewood joined #salt
23:58 whit joined #salt

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