Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2013-08-30

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

All times shown according to UTC.

Time Nick Message
00:01 dthom91 joined #salt
00:06 sgviking joined #salt
00:12 tyler-baker joined #salt
00:21 littleidea joined #salt
00:26 kenbolton joined #salt
00:29 Eugene I would rather spoon out my eyeballs with a rusted spork
00:29 Eugene But that's me.
00:29 quinnat joined #salt
00:34 danielbachhuber joined #salt
00:38 ccase Ah well, maybe it's not such a grand idea :o)
00:38 dthom91 joined #salt
00:39 akoumjian whiteinge Great news!! https://github.com/mitchellh/vagrant/pull/1626
00:39 kula one of the things i love about salt is that it doesn't care what your configuration files are written in; if you can get a parser to turn it into a python data structure, go with it.
00:40 quinnat so I am trying to find material on how saltstack uses zeromq and why that is a good thing out there, and in general and why that might be better than say using puppet or chef?  Are there comparison paper/articles out there?
00:40 quinnat and how it differs from job scheduling system like Google's Borg
00:41 whiteinge akoumjian: DUDE. NICE!
00:42 akoumjian whiteinge: Now to fix all the open tickets I've been putting off O.o
00:43 quinnat also would salt stack be salt stack with out zeromq?
00:45 kula i think thomas hatch has a blog post up somewhere about why he started work with zeromq and msgpack when he was creating salt stack.
00:45 whiteinge akoumjian: this is huge, man. nice freakin' work
00:46 kula also, i believe there are plans in work to allow salt stack to use different messaging layers. i believe one built on top of ssh is in the works.
00:47 kula oh, or maybe he talks about it in one of the presentations he's done. i remember seeing it somewhere from the salt stack website when i was first reading about it.
00:47 akoumjian whiteinge: thanks. :-)
00:49 whiteinge ssh as a transport is coming in 0.17
00:49 kenbolton joined #salt
00:51 whiteinge quinnat: i can field some of those questions
00:52 quinnat whiteinge: awesome!
00:52 whiteinge salt would still be salt without zeromq. that said, zeromq gives salt a ton of win
00:53 whiteinge speed and scalability are the big two. it gives near-instant communication with 10k+ servers
00:53 robertkeizer joined #salt
00:53 whiteinge another is queuing and automatic reconnections. e.g., if you restart the master, minions will wait for it to come back online before delivering responses
00:53 quinnat what about zeromq sockets are more special than using regular unix sockets and epoll?
00:54 quinnat that is how most pub/sub systems work though right
00:55 quinnat also how does it compare say with Kafka?
00:55 whiteinge (disclaimer: i'm not a zeromq expert) i believe the win there is nice interface, portable interface, and ease of use
00:56 whiteinge salt runs on OSX, the BSDs, windows, etc and that is made much easier because zeromq is on those platforms
00:56 whiteinge i'm not familiar with kafka, sorry
00:56 kula the zeromq documentation is rather extensive, and while it's a bit over the top in places, the description of "sockets the way you expect them to work" is rather apt.
00:58 kula which makes it easy to build applications on top of it. zeromq is rather lightweight, portable, and implements many communications patterns (request/reply, pub/sub) well, so that application writers can focus on their application, not diddling with standard unix sockets pain.
01:00 whiteinge in response to your question about comparison with puppet/chef (strictly from the zeromq perspective): it depends on your workflow. if you just want to push configs out to a few servers once or twice a day, you probably don't care about speed.
01:01 whiteinge if you want to run commands across servers all day, you start to care about speed. if you want real-time notifications of things happening on your servers, you definitely care about speed. :)
01:01 whiteinge there are reasons to run salt other than speed, of course. i'm just sticking to the zeromq line of questions :)
01:02 aat joined #salt
01:04 dthom91 joined #salt
01:13 quinnat yeah I am beginning to find out that zeromq is more than just speed, which adds a lot to saltstack
01:13 whit joined #salt
01:17 carmony joined #salt
01:19 gatoralli joined #salt
01:19 pdayton joined #salt
01:20 quinnat I decided to move over to #zeromq for a while until I get back into salt land
01:20 gatoralli has anyone been able to bootstrap ssl client certificates with salt's master/minion keys?
01:21 imaginarysteve joined #salt
01:25 whit joined #salt
01:33 liuyq joined #salt
01:34 santagada joined #salt
01:35 jY anything in salt to setup interface aliases.. either the old eth0:0 way or ip addr way?
01:35 liuyq joined #salt
01:37 xl1 joined #salt
01:39 danielbachhuber joined #salt
01:42 eightyeight how can i tell if a server is being actively salted at any given time
01:42 eightyeight i see the salt-minion PID is taking up ~2% of RAM on a 1024 MB RAM install. that seems high to me.
01:43 eightyeight i got that it's only 20 MB, but for a (mostly) idle PID, 20 MB seems like a lot
01:43 eightyeight maybe i'm paranoid
01:43 dcmorton joined #salt
01:43 eightyeight ... and as i type that, it drops down to .7 %. which seems more like it.
01:45 jbunting joined #salt
01:48 eightyeight ok. next question; i'm getting the following error, and i'm not sure how to approach it: http://ae7.st/p/52h
01:49 eightyeight seems to be complaining about "KeyError: 'Parallels Server Bare Metal'", but i have this setup correctly, as i have 4 other servers that are PSBM
01:50 littleidea joined #salt
01:59 Furao joined #salt
02:02 bejer joined #salt
02:03 aat joined #salt
02:05 sixninetynine joined #salt
02:09 pdayton joined #salt
02:10 malinoff joined #salt
02:10 mgw1 joined #salt
02:11 aat joined #salt
02:11 malinoff Hi all
02:15 pdayton joined #salt
02:23 vipul joined #salt
02:25 dthom91 joined #salt
02:26 gldnspud joined #salt
02:28 emocakes joined #salt
02:29 aat joined #salt
02:29 dcmorton joined #salt
02:31 aat joined #salt
02:41 Gifflen joined #salt
02:45 danielbachhuber joined #salt
02:47 renothing joined #salt
03:06 ds_shadof joined #salt
03:09 mannyt joined #salt
03:13 whit joined #salt
03:22 aat joined #salt
03:38 eightyeight got it
03:40 malinogg joined #salt
03:46 mannyt joined #salt
03:50 danielbachhuber joined #salt
03:53 malinoff joined #salt
03:56 Asphyxiated|Dk joined #salt
03:56 Asphyxiated|Dk joined #salt
04:07 liwen joined #salt
04:13 jalbretsen joined #salt
04:15 robertkeizer joined #salt
04:17 liwen_ joined #salt
04:20 berto- joined #salt
04:20 malinoff joined #salt
04:26 terminalmage eightyeight: that looks like more detection issues, you can solve that with a provider. you said that this is CentOS 6.4-based right? So that would use the rh_service module.
04:26 terminalmage by solving it with a provider, I mean in the minion config like before
04:28 troyready joined #salt
04:29 berto- joined #salt
04:33 Katafalkas joined #salt
04:35 Katafalkas joined #salt
04:43 sgviking joined #salt
04:56 danielbachhuber joined #salt
05:09 danielbachhuber joined #salt
05:19 aat joined #salt
05:24 Katafalkas joined #salt
05:27 Katafalkas joined #salt
05:28 Katafalkas joined #salt
05:30 linjan_ joined #salt
05:32 gildegoma joined #salt
05:33 troyready joined #salt
05:52 dthom91 joined #salt
06:02 linjan_ joined #salt
06:04 fibo joined #salt
06:10 geecu joined #salt
06:10 chrisgh joined #salt
06:15 SIFTU joined #salt
06:19 chrisgh Gooood morning vietnam!
06:19 chrisgh is there a way to alter debian selections with salt-stack
06:21 whiteinge chrisgh: http://docs.saltstack.com/ref/modules/all/salt.modules.debconfmod.html
06:21 chrisgh Thanks a bunchie whiteinge! :)
06:21 whiteinge np :)
06:22 Katafalkas is there supposed to be problems with salt-cloud attaching devices to a VM  on ec2 ?
06:43 NV joined #salt
06:44 TheCodeAssassin joined #salt
06:45 NV looking at migrating from puppet to salt, fair number of hosts (on the order of hundreds) trying to get the current puppet config reproduced in salt and hit a bit of a wall with use of augeas
06:46 NV afaict every augeas set would require around a dozen lines or so, which is quickly going to make config files rather hard to maintain due to the massive amounts of boiler plate to call the augeas module, etc
06:46 NV is there a way to use augeas in a more concice manner? I'm guessing if there was an augeas state module it would be easier, but from what I can tell there isn't yet
06:47 NV (or alternatively, a more salt-friendly augeas equivalent?)
06:48 intchanter joined #salt
06:52 dthom91 joined #salt
06:54 balboah_ joined #salt
06:58 druonysus joined #salt
06:59 druonysuse joined #salt
07:02 madeira joined #salt
07:05 carlos joined #salt
07:08 ml_1 joined #salt
07:14 vaxholm_ joined #salt
07:14 MrTango joined #salt
07:17 pdayton joined #salt
07:20 gildegoma_ joined #salt
07:21 madeira joined #salt
07:29 xinkeT joined #salt
07:31 xl joined #salt
07:36 pdayton joined #salt
07:38 mmilano joined #salt
07:38 TheCodeAssassin Is there a simple way of removing a user across all hosts for a mysql server?
07:39 TheCodeAssassin say root with all hosts
07:42 p3rror joined #salt
07:45 emocakes joined #salt
07:49 xl joined #salt
07:50 scott_w joined #salt
07:56 malinoff TheCodeAssassin: Is this http://docs.saltstack.com/ref/states/all/salt.states.mysql_user.html?highlight=state%20mysql#salt.states.mysql_user.absent helpful?
07:57 TheCodeAssassin malinoff: did not see it there
07:58 TheCodeAssassin also, did anyone build php from source using salt?
07:58 zooz joined #salt
07:59 Eugene You should be able to call mysql_user.absent via `salt`
07:59 malinoff TheCodeAssassin: What's the problem? Write a script e.g. "wget http://source.php.com; cd source; make; make install" and use http://docs.saltstack.com/ref/states/all/salt.states.cmd.html?highlight=cmd.script#salt.states.cmd.script
08:00 liwen joined #salt
08:01 malinoff Salt is not only configuration managment tool, it's also remote execution tool. Don't forget about it.
08:02 jcristau joined #salt
08:04 Furao joined #salt
08:08 emocakes joined #salt
08:10 TheCodeAssassin yes yes exactly, thnx malinoff
08:11 TheCodeAssassin also, i want to change the root users password in a state. But then the connection info changes so salt can no longer connect
08:12 TheCodeAssassin https://gist.github.com/thecodeassassin/14b4d64b5825aec0daec
08:13 TheCodeAssassin i also have a specify a host, i tried * but that doesn't work
08:13 ronc joined #salt
08:15 liwen_ joined #salt
08:15 qba73 joined #salt
08:15 lemsto joined #salt
08:23 dthom91 joined #salt
08:24 malinoff TheCodeAssassin: You mean, you've specified mysql user and password in the master config file, then you want to change password in a state and reconnect to the database?
08:25 liwen joined #salt
08:29 TheCodeAssassin well the next time the state runs it won't be able to connect
08:34 malinoff TheCodeAssassin: Does it connect to the database with user which password you've changed?
08:34 TheCodeAssassin well im trying something now
08:34 TheCodeAssassin something different, but then i can never change the root password
08:34 TheCodeAssassin which could also be a good thing i guess
08:35 TheCodeAssassin well the user is just root
08:35 TheCodeAssassin basically im creating users from pillar data
08:35 TheCodeAssassin with their databases
08:35 TheCodeAssassin but root should always be present
08:35 TheCodeAssassin and i want it to have a certain (preferably changeable password)
08:36 Jason-AVST joined #salt
08:37 xl joined #salt
08:41 malinoff TheCodeAssassin: Well, you could try to use http://docs.saltstack.com/ref/modules/all/salt.modules.mysql.html?highlight=mysql%20user#salt.modules.mysql.user_exists with jinja template
08:41 malinoff Let me a few minutes, i'll provide an example
08:41 mmilano left #salt
08:43 tomviner joined #salt
08:44 malinoff http://pastebin.com/4WBvm78Q
08:44 malinoff I am not sure about the syntax
08:45 malinoff password={{ pillar["dbpass"] }} - don't know, is this brackets really necessary
08:49 backjlack joined #salt
08:51 TheCodeAssassin lemme try this malinoff
08:57 kbar left #salt
09:01 lemao joined #salt
09:08 malinoff TheCodeAssassin: Any results?
09:09 liwen joined #salt
09:10 TheCodeAssassin hehe a lot of stuff going on, one moment :)
09:13 liwen_ joined #salt
09:14 TheCodeAssassin takes a while to run
09:20 felixhummel joined #salt
09:23 ml_11 joined #salt
09:26 malinoff TheCodeAssassin: http://funkyimg.com/u2/1613/240/260150vunsh.jpg this image appeared in my mind after your message
09:26 TheCodeAssassin hahahah
09:28 lemao joined #salt
09:39 ggoZ joined #salt
09:40 TheCodeAssassin but it seems like it should work, but i'm kinda trying something else now.
09:48 Psy-Q i'm installing a pkg on debian and have given the keyserver and keyid options, but it does not add the key, just the repo, and thus doesn't trust the packages and doesn't install them
09:48 Psy-Q how could i debug that?
09:48 ml_1 joined #salt
09:49 Psy-Q ah, maybe it doesn't like the keyid format
09:50 malinoff TheCodeAssassin: Why? It's readable and simple solution
09:50 TheCodeAssassin malinoff: https://gist.github.com/thecodeassassin/8d8f2df0f3870e41cb0d
09:50 TheCodeAssassin because i'm doing some refactoring ;)
09:51 TheCodeAssassin i'm setting the root password before installing the packages, and then the root password should not change for the minion
09:51 TheCodeAssassin basically
09:51 malinoff TheCodeAssassin: Yes, this state should definitely be refactored...
09:51 TheCodeAssassin ?
09:52 TheCodeAssassin what i just send you is actually my refactoring hehe
09:52 malinoff TheCodeAssassin: LOL
09:52 TheCodeAssassin https://gist.github.com/thecodeassassin/485260709787eb5719e7
09:52 TheCodeAssassin this is the pillar file
09:52 p3rror joined #salt
09:52 TheCodeAssassin tell me what you don't like (don't say everything) lol
09:53 TheCodeAssassin well i do see some syntax errors
09:55 malinoff TheCodeAssassin: Spaghetti code. But it seems to be a problem with many possible mysql options, users and databases, not a bad design
09:55 TheCodeAssassin well i dunno how to make it much cleaner, maybe more spacing and stuff
09:56 TheCodeAssassin but is should work
09:56 TheCodeAssassin "technically"
09:58 malinoff TheCodeAssassin: I LOVE python's module system. So, i'd rather separate service, mysql_user, mysql_database, mysql_grants into different state files. Probably, with including each other. So it is not a *better* solution, it's better just for me
10:00 TheCodeAssassin well malinoff, it is a seperate file for uers
10:00 TheCodeAssassin i have database, database.users etc
10:00 TheCodeAssassin but i do grants in the same file, just to keep it in the same place
10:00 hughsaunders joined #salt
10:00 TheCodeAssassin might not be the best idea, but i'm still pretty new
10:16 whiskybar joined #salt
10:21 LucasCozy joined #salt
10:22 isomorphic joined #salt
10:22 lemao joined #salt
10:23 LucasCozy joined #salt
10:26 giantlock joined #salt
10:27 Furao joined #salt
10:28 lemao_ joined #salt
10:35 felixhummel can salt-cloud set the hostname when creating an instance?
10:41 felixhummel when I run salt-cloud -p small foo, then the hostname is generated (like ip-10-0-0-23)...
10:48 liwen joined #salt
10:48 Katafalkas felixhummel: I think there is some option, but as far as I am aware, it is not recommended. something to do with ec2 itself
10:50 Katafalkas Can you import salt-cloud to a script and make a script that would combine salt and salt-cloud ?
10:52 efixit joined #salt
10:52 santagada joined #salt
10:55 eightyeight terminalmage: yeah. got it.
10:57 felixhummel Katafalkas: currently i go with echo "{{ grains['id']}}" > /etc/hostname && hostname {{ grains['id'] }}"
10:57 felixhummel i was just curious if salt-cloud can bootstrap instances with the correct hostname settings
10:58 felixhummel it can and will not, if i understand this conversation correctly: https://github.com/saltstack/salt-cloud/issues/712
11:00 felixhummel i agree, because it's not salt-cloud's responsibility. this should be done with ec2 user data and cloud-init (https://help.ubuntu.com/community/CloudInit).
11:00 ml_1 joined #salt
11:05 * felixhummel is not the first one thinking about this: https://github.com/saltstack/salt-cloud/issues/625
11:08 az87c joined #salt
11:08 intchanter joined #salt
11:08 az87c_ joined #salt
11:20 jbunting joined #salt
11:35 blee joined #salt
11:37 felskrone joined #salt
11:37 felskrone hi
11:44 lemao joined #salt
11:48 santagada joined #salt
11:49 liwen joined #salt
11:56 jcristau joined #salt
12:00 littleidea joined #salt
12:04 alunduil joined #salt
12:05 emocakes joined #salt
12:05 jcristau joined #salt
12:18 aat joined #salt
12:23 lineman60 joined #salt
12:25 aat_ joined #salt
12:32 lemao joined #salt
12:32 lineman60 joined #salt
12:34 tuxIO joined #salt
12:35 liwen joined #salt
12:37 kenbolton joined #salt
12:41 TheCodeAssassin guys how i can make a requirement for another id
12:41 lineman60 joined #salt
12:42 TheCodeAssassin say given this: https://gist.github.com/thecodeassassin/4e1166bf08009b29cb48 how can i require this? That the root user password changed
12:47 littleidea joined #salt
12:47 jbunting joined #salt
12:50 anteaya joined #salt
12:52 juicer2 joined #salt
12:55 lemsto joined #salt
12:57 whit joined #salt
12:57 drew__ joined #salt
12:59 racooper joined #salt
13:00 TheCodeAssassin is there a problem with including pillar files?
13:03 Hans_ joined #salt
13:05 Gifflen joined #salt
13:07 drew__ hello, how can i repeat the same instruction multiple times in a state file?
13:07 drew__ i need to create a directory at the start of my state, install some software into it, then chown it at the end
13:08 drew__ but the paths for both the create and the chown operations are the same so I'm getting circular dependency issues
13:08 drew__ see http://pastebin.com/Bw3gcg7i for what I've got
13:10 mwillhite joined #salt
13:10 ipmb joined #salt
13:10 tuxIO_ joined #salt
13:12 hughsaunders drew__: require the last step of the install from the chown state?
13:13 hughsaunders Can I require all states in another sls file? Or would I have to list them?
13:14 bhosmer joined #salt
13:14 brianhicks joined #salt
13:16 cron0 joined #salt
13:17 drew__ hughsaunders: I'm doing that. The problem is that the 'name' arg for file.directory seems to overwrite the label, so in fact salt can't differentiate between both commands
13:17 drew__ should that work and there's a problem in my state, or is this an issue with salt?
13:18 hughsaunders interesting problem...
13:18 hughsaunders cmd.run ;-)
13:19 imaginarysteve joined #salt
13:21 jeffasinger joined #salt
13:22 Jason-AVST joined #salt
13:33 drew__ grubby solution, but hopefully it'll work. thanks
13:35 aleszoulek joined #salt
13:35 Kholloway joined #salt
13:36 djn when can 0.17 be expected? and will halite make the cut?
13:39 bhosmer joined #salt
13:40 hughsaunders drew__: there probably is a better answer, but I am new to salt so don't know of one..
13:41 drew__ i'm now quite new to salt but haven't found a better one. it's working so far (build is still running)
13:41 terminalmage joined #salt
13:46 maxl0 joined #salt
13:50 ProT-0-TypE joined #salt
13:51 ProT-0-TypE joined #salt
13:54 Hans_ Hello, does anyone know if I can configure my minion to use the master as file server but disable all remote execution on the minion? I would like to create a minion that can pull states from the master but doesn't allow the master to push a new configuration to the minion.
13:55 aat joined #salt
13:56 Hans_ So far I've tried to use "whitelist_modules" and "disable_modules" on the minion. If I disable modules like "state" the master can't push anything to my minion, but I can't pull anything either of course.
13:59 liwen joined #salt
13:59 hughsaunders bother drew_ left but you can use extended IDs in requirements: https://gist.github.com/anonymous/6390141
14:01 gamingrobot hughsaunders: handy
14:04 kaptk2 joined #salt
14:05 [diecast] joined #salt
14:06 mannyt joined #salt
14:07 jbunting joined #salt
14:08 p3rror joined #salt
14:14 cron0 djn, yesterday I heard 0.17 should be released next week
14:14 cron0 djn, http://irclog.perlgeek.de/salt/2013-08-29#i_7517989
14:24 pdayton joined #salt
14:25 ingwaem joined #salt
14:26 ingwaem Greetings everyone. When I look at the jobs on salt, the data returned has a formatted date like 2013, Aug 30 02:06:00.759997. Is there anyway to have this pre formatted so that I don't have to do any crazy date reformatting?
14:27 whit joined #salt
14:27 lemao joined #salt
14:29 tca joined #salt
14:30 liwen joined #salt
14:31 [diecast] joined #salt
14:32 teskew joined #salt
14:35 tyler-baker joined #salt
14:35 tyler-baker joined #salt
14:37 cnelsonsic joined #salt
14:39 cwright according to the docs for mysql_grants it seems I should be able to put a *.* in the  database field.  but that does not seem to be allowed in the sls file.
14:39 cwright based on the notes about the database parameter on here http://docs.saltstack.com/ref/states/all/salt.states.mysql_grants.html#module-salt.states.mysql_grants
14:40 cwright this is my sls and the output from state.highstate: https://gist.github.com/corywright/c0825ff1dc63f17e5f60
14:40 cwright does anyone have any recommendations?
14:41 ksalman cwright: try "*.*"
14:41 ksalman in quotes
14:41 ksalman i think that should work
14:41 cwright ksalman: ok, lemme try that.
14:42 jalbretsen joined #salt
14:42 opapo joined #salt
14:45 cwright ksalman: that works.  i guess its not intuitive because you only need to quote if there is a leading asterisk?  using mydb.* doesn't require quotes
14:45 UtahDave joined #salt
14:46 ksalman cwright: i am not sure =) i only know that quotes work becuase I got hit by the same problem, but I never dug into it
14:46 cwright ksalman: well thanks for the help :)
14:46 ksalman sure!
14:49 oz_akan_ joined #salt
14:57 ricicle joined #salt
14:58 gldnspud joined #salt
14:59 m_george|away joined #salt
14:59 ckao joined #salt
15:00 aat joined #salt
15:00 m_george left #salt
15:01 carmony joined #salt
15:03 forrest joined #salt
15:03 Thiggy anyone using the gitfs backend for their states and taking advantage of the branches=environments thing?
15:03 forrest No, are you having issues with it?
15:04 Thiggy My issues currently are conceptual. Just having a hard time understanding what the workflow would look like.
15:05 Thiggy I got the state files being served this morning, and that part seems to be working ok.
15:06 UtahDave Thiggy: we have quite a few people that will have a dev branch. When that's looking good they'll push it into a "testing" branch
15:06 UtahDave When the testing/qa people pass it off, then it gets merged into "staging" branch
15:06 UtahDave Then if everything still looks good they'll merge it into master.
15:07 UtahDave That's one way it has been used.
15:07 kula Thiggy: I've just started using it for a project. I do a slightly different workflow: the master branch I use only for top.sls. Everything else lives in a specific branch, e.g. production for production stuff.
15:07 Thiggy @utahdave that makes sense
15:08 Thiggy @kula that's what I was thinking I needed to do at first. Does that work out for you? It feels weird to have a master branch with top.sls on it, but only from a "feelings" perspective, I can find no reason that it would be bad.
15:09 kula I've done very little of it, but because git 'master' has to translate to salt 'base', it's already weird, so i figure i use the weirdness to my advantage and keep it for just one thing.
15:09 ricicle so can anyone point me to an sly snippet that installs a couple of gems please? In particular I'm finding it difficult to require ruby gems to be installed first :/
15:10 forrest http://docs.saltstack.com/ref/states/all/salt.states.gem.html does that not have requires?
15:10 forrest or can you not use requires in there?
15:10 kula besides, since you have to muck with top.sls all the time to deal with things not in production (e.g. put a minion in a 'test' environment), i like that the logs in the master branch only contain notes about changing what things go where, and the logs in the production environment aren't muddled with "Put minion_foo into test_webserver environment" logs.
15:11 forrest better yet ricicle, are you trying to install gems, then something else requires them?
15:11 JoAkKiNeN joined #salt
15:11 dthom91 joined #salt
15:11 forrest Yea we did the same thing with git that Utah Dave is talking about, 3 branches, one dev, one qa, one prod
15:11 Thiggy @kula I think I'm gonna try that route and see how that works out.  I only have a staging/production environment
15:12 kula to put it shorter, having only top.sls in the master branch is slightly less weird to me then constantly having to think "production is in the master branch"
15:12 UtahDave forrest: everything has requires.
15:13 UtahDave that's a very interesting workflow, kula
15:14 kula i guess i just really like "test environment is in the test branch, qa environment is in the qa branch, production environment is in the production branch"
15:14 gatoralli joined #salt
15:14 forrest You could have a production branch
15:15 forrest UtahDave, good to know. I wasn't sure since it didn't have that information on the gems page.
15:15 inverat joined #salt
15:15 loungin_light joined #salt
15:15 kula i do, in this one project.
15:15 StDiluted joined #salt
15:15 carmony joined #salt
15:16 ksalman since "unless" , or "onlyif" doesn't exist for something like svn, how would i replicate that functionality? I don't want to "svn up" at every highstate, only when a condition is true for example
15:16 forrest ugh svn
15:17 forrest ksalman, you could use jinja2 templating
15:18 forrest it would be messy, but you could for example compare the md5sums of the local file versus the remote svn file
15:19 ksalman i can do conditionals with jinja?
15:19 ksalman i am not very familiar with jinja, i guess
15:19 forrest you can do normal python code in jinja
15:19 forrest http://docs.saltstack.com/topics/tutorials/starting_states.html#getting-to-know-the-default-yaml-jinja
15:19 inverat hi all -- I'm trying to feed a dozen or so scripts of varying run time through command-line salt cmd.script but the timeouts are giving me fits.  If I specify no timeout, then long-running scripts seem to terminate prematurely, and if I specify a long timeout then short-running scripts do not return until the end of the timeout... I feel like I'm missing a critical feature here
15:20 forrest check out that section ksalman
15:20 ksalman forrest: thanks
15:21 forrest yea np, if you want to know more about jinja2 in general, check out: http://jinja.pocoo.org/docs/templates/ It's more for HTML, but it has some great examples
15:21 forrest Just to give you ideas.
15:21 forrest inverat, can you run the scripts straight from the command line and they all complete?
15:22 ricicle forrest: I've tried but am getting The following requisites were not found: require: {'pkgs': 'rubygems'}. The rpms install fine; foopkgs: pkg.installed: - pkgs: - ruby-devel - rubygems (apologies for the formatting there)
15:22 forrest can you post your sls ricicle?
15:22 forrest stripped of any proprietary stuff?
15:23 ricicle it's basically in what I posted there but I can do a paste bin if it helps
15:23 inverat forrest: yes -- some are short backup scripts that complete in 15 seconds and others are long rsync scripts that could take 1-5 hours depending on deltas
15:23 forrest hmm
15:23 forrest is anything getting logged into the salt logs?
15:24 BrendanGilmore joined #salt
15:24 ricicle forrest: http://pastebin.com/hnEGMsG2
15:25 ricicle I must be doing something wrong otherwise none would ever have had a gem installed!
15:25 forrest heh
15:26 littleidea joined #salt
15:26 ricicle or maybe salty-ruby people don't use CentOS?
15:27 gatorall1 joined #salt
15:29 forrest can you try to expand things ricicle?
15:29 ricicle ?
15:29 ricicle sorry, new to salt. Expand things?
15:30 forrest http://pastebin.com/ruZjT99j
15:30 forrest something like that
15:30 forrest which might not be right still
15:30 forrest sometimes the yaml is weird when you use the dot notation
15:31 pdayton joined #salt
15:31 forrest If anyone can correct mistakes in my example I'd appreciate it
15:31 mgw joined #salt
15:31 UtahDave forrest gave a great example.
15:31 UtahDave You can also require against the ID declaration: http://pastebin.com/nJaKuurY
15:31 forrest Oh good idea UtahDave
15:32 forrest Try both of those ricicle
15:32 kermit joined #salt
15:35 jacksontj joined #salt
15:35 forrest UtahDave, was there anything actually wrong with ricicle's version? I don't understand why he's getting an error.
15:36 forrest is it because he specified a requires of pkgS: rubygems, instead of pkg: rubygems maybe?
15:36 forrest since technically the pks: notation only applies to mymodulepkgs
15:37 forrest *pkgs
15:37 UtahDave yes, it must be "pkg"  ,   plus, when you use ' - pkgs' I'm not sure you can then require against just one package in that list.  I think you may only be able to require against the ID dec at that point
15:37 forrest Yea since it's all wrapped up
15:37 forrest makes sense
15:38 forrest inverat, did you find anything in the logs?
15:39 liwen joined #salt
15:40 forrest or if you're running them as a cmd (not sure of your distro), can you pass through nohup cmd & to push it to a background process?
15:42 ricicle forrest: Your first paste bin gave me 'not formatted as a list', when I added dashes I got "No function declared in state "rubygems" in sls repos.mymodule". Your second paste bin worked great, thanks. UtahDave I tried all three of 'pkg', 'pkgs' and 'packages'. An example or two in the docs would be great.
15:42 TheCodeAssassin joined #salt
15:42 forrest that second one was UtahDave :P
15:42 conan_the_destro joined #salt
15:43 UtahDave ricicle: there are examples.   Just FYI 'pkg' refers to the state you're requiring against.   That's why it's spelled that way.
15:44 ricicle Thanks UtahDave, I'm sure I'll grok it fully soon
15:45 forrest mine probably didn't work because of what UtahDave brought up earlier ricicle, where since it's part of an ID declaration, if you changed my example to http://pastebin.com/YVKVSjQt
15:45 ricicle No way to require multiple gems at the moment though?
15:45 Charlie_Rangel joined #salt
15:45 forrest ?
15:45 forrest what do you mean
15:45 ricicle or rather, have one snippet of yaml to install several gems
15:45 Charlie_Rangel Hello good day to you all
15:45 ricicle like the groups of packages
15:45 Charlie_Rangel I have a question about state ordering
15:46 forrest Charlie_Rangel, what's up
15:46 Charlie_Rangel Is there a way to have a state wait and poll before it runs?
15:46 forrest you can use jinja2 templating
15:46 forrest to do some sort of command
15:46 Charlie_Rangel I'm trying to set up streaming postgres replication, and the replica needs to know the master is alive and turned off
15:46 Thiggy does gitfs do some sort of caching? It seems like updates to my state files show up "some time later" after I push to the repo.
15:47 UtahDave Thiggy: gitfs does a refresh every so often.  I think it's every minute.
15:47 Charlie_Rangel or has anyone already solved setting up postgres replication some other way?
15:47 Thiggy @UtahDave Gotcha. Thanks!
15:48 forrest ricicle, why not do something like http://pastebin.com/mcQhggTR
15:48 aat joined #salt
15:49 sgviking joined #salt
15:49 UtahDave Charlie_Rangel: You may want to use an overstate. That way you can make sure the master is in the state you need
15:49 inverat forrest: nothing particularly damning in the logs, but from a test run yesterday my wrapper script got this back: ['Execution is still running on host.on.my.domain\n',...,'host.on.my.domain: Minion did not return\n'] after quite a bit of some time running -- I assumed this was due to some timeout, so I added a -t <very long timeout, like 23 hours>
15:49 forrest but it still bombed on you
15:50 inverat well, the problem is that with the long timeout the short jobs never return, I figure they would have returned at the end of the timeout, but that isn't viable
15:50 Charlie_Rangel UtahDave: many thanks, that is what I'm looking for
15:50 lazyguru joined #salt
15:51 forrest that seems weird inverat, you think they'd just return once they finished, regardless of the timeout
15:51 UtahDave you're welcome, Charlie_Rangel
15:55 ricicle forrest: That fails even without the require - it tries to install 'myrubygems' and cries: result: false, could not install gem
15:55 forrest weird.
15:55 inverat forrest: it appears I can replicate the behavior with cmd.run on a smaller scale, just running a sleep command, like so: http://pastebin.com/Mczj0Yv8
15:56 forrest interesting
15:56 forrest that doesn't make any sense
15:57 mgw Does current develop have any known (showstopper) issues?
15:59 ricicle forrest: I'm guessing you're talking to inverat, I'm happy now, will be around if you want me to test anything
15:59 forrest oh yea sorry I was talking about his timeout issue.
15:59 forrest If you figure out how to do groups of gems, let me know
15:59 TheCodeAssassin joined #salt
15:59 forrest and then do a pull with an example for the gems page :P
16:01 aat joined #salt
16:02 forrest inverat, the actual run call is on line 501 over here: https://github.com/saltstack/salt/blob/develop/salt/states/cmd.py
16:02 forrest but I don't see how the logic works for timeout itself :\
16:03 gldnspud joined #salt
16:05 juanlittledevil joined #salt
16:05 KennethWilke joined #salt
16:09 liwen_ joined #salt
16:10 inverat forrest: you're looking at the state code, code for command module is here https://github.com/saltstack/salt/blob/develop/salt/modules/cmdmod.py#L324
16:10 forrest oh duh
16:13 felskrone joined #salt
16:14 kermit joined #salt
16:15 ccase joined #salt
16:16 joakki joined #salt
16:16 conan_the_destro joined #salt
16:18 liwen joined #salt
16:18 gldnspud joined #salt
16:21 chrisgilmerproj joined #salt
16:21 erik__ joined #salt
16:23 mwillhite is there a way to use the cmd.run state but not stop if it exits poorly? is that what 'quiet' is for?
16:25 forrest I've got nothing inverat, the only thing I would say is to try using the subprocess module to run a test script locally to see if the wait command causes the process to just hang out
16:27 inverat forrest: yea i'm tempted to just submit a "timeout doesn't work as expected" but even after reviewing the code I don't understand why it doesn't work
16:27 jdenning joined #salt
16:27 forrest same
16:28 forrest I'd submit it as an issue with that example you provided
16:28 forrest the pastebin one
16:28 dizzyd joined #salt
16:28 forrest because clearly, sleep should return after just 5 seconds, not 20
16:28 dizzyd does anyone know if the master has a cache of pillar information per minion?
16:29 UtahDave dizzyd: yes, it does
16:29 dizzyd can I force it to refresh/clear that cache?
16:31 druonysus joined #salt
16:32 forrest you might be able to just trash /srv/pillar dizzyd
16:32 saurabhs joined #salt
16:32 forrest but I'd make a backup
16:32 forrest I don't know how the rebuild process works.
16:32 UtahDave no, the cache is not in /srv/pillar
16:32 dizzyd yeah, i think the heart of my question is that when I make a change to a pillar, how do i get that to take effect on minions
16:32 dizzyd right now it appears things don't get propogated
16:32 dizzyd or maybe I'm just Doing It Wrong :)
16:32 UtahDave dizzyd: salt \* saltutil.refresh_pillar
16:33 inverat dizzyd: I have to run the saltutil.refresh_pillar thing
16:33 forrest totally missed it in the docs, dizzyd
16:33 forrest http://docs.saltstack.com/topics/pillar/index.html#refreshing-pillar-data
16:33 dizzyd haha
16:33 dizzyd beautiful
16:33 forrest it's in the main index, not the pillar walkthrough :P
16:33 dizzyd forrest: so did I, apparently
16:34 dizzyd thanks UtahDave
16:34 saurabhs Hi can somebody help me with log rotation setting for salt master and for minions? I don't find any such settings in /etc/salt/master
16:34 forrest saurabhs, are you talking about rotating the salt logs, or just logs in general?
16:35 forrest I assume you just configure the logrotation via logrotate as you normally would: http://docs.saltstack.com/ref/modules/all/salt.modules.logrotate.html
16:35 saurabhs I mean logs of salt master and minions it self
16:36 forrest yea
16:37 Jahkeup joined #salt
16:37 druonysuse joined #salt
16:37 druonysuse joined #salt
16:39 forrest saurabhs, yea just set them via salt
16:42 bitz joined #salt
16:44 whyzgeek joined #salt
16:44 whyzgeek wow I didn't expect this many people here!
16:44 whyzgeek I am new to salt
16:45 forrest UtahDave holds us hostage here :(
16:45 whyzgeek I used repo state to add a repo to a Centos machine
16:45 whyzgeek forrest: lol
16:45 Thiggy @kula - I setup my states in gitfs like you described, and it's working out really nicely.
16:45 bitz I tried to leave once, it wasn't pretty :(
16:45 whyzgeek however when I change stuff it doesn't apply the changes
16:45 forrest bitzh, hah
16:45 forrest when you try to run the state it doesn't get applied whyzgeek?
16:46 Thiggy @kula - I did move some of the more boring stuff into the master/base environment (like default editors blah blah), but the separation of the other stuff is I think exactly what I want.
16:46 whyzgeek forrest: it gets applied
16:46 whyzgeek but if for example I add a feild it adds it
16:46 whyzgeek but removal doesn't work
16:47 forrest so these are changes you're making in pkgrepo.managed or whatever?
16:47 forrest so you'd add something like a gpgkey: asdf entry
16:47 forrest and it would add
16:47 forrest but if you removed that, and reapplied the state it wouldn't remove that line?
16:48 whyzgeek forrest: correct
16:48 whyzgeek forrest: like added mirrorlist and removed it and added baseurl
16:48 forrest yea
16:48 whyzgeek baseurl gets added but mirrorlist stays there
16:49 forrest that's really weird
16:49 littleidea joined #salt
16:49 whyzgeek forrest: as I am new maybe I am missing something
16:50 dthom91 joined #salt
16:50 forrest I'm still new myself, but I don't know why it would do that, it should be cleaning it up and only including what you have.
16:50 whyzgeek forrest: you are saying it should have removed unrelated line then
16:50 forrest I thought it was supposed to
16:50 UtahDave whyzgeek: you have to use pkgrepo.absent to remove something
16:51 forrest but what if he just wants to remove a single line in a repo declaration?
16:51 UtahDave whyzgeek: otherwise it just manages the actual repos you're specifying
16:51 jdenning joined #salt
16:51 whyzgeek UtahDave: have you got an example config? I have seen some absent files in salt-states
16:52 whyzgeek I guess those are the ones you are refering to?
16:53 carmony joined #salt
16:55 whyzgeek UtahDave: is there any reason for this kind of design? in puppet it actually gives you what you define and it exclusively gives that. That's kind of most of the time the expectation. This way configs gets bloated with a lot of absent stuff.
16:55 forrest pkgrepo.absent is specifically used to delete an entire repo according to the docs
16:56 forrest but maybe you can pass in specific values via the kwargs
16:56 whyzgeek forrest: yes I guess what DaveUtah saying is that if you want to get what you define, you always have to delete whatever that is there and recreate it
16:57 ksalman How come I can a custom pillar with pillar.items but I don't with pillar.get? https://gist.github.com/ksalman/6391997
16:57 troyready joined #salt
16:57 ksalman shouldn't i get "bar"?
16:57 forrest Yea I'm not sure whyzgeek
16:58 forrest can you try to add a pkgrepo.absent to your command that removes that mirrorlist line whyzgeek?
16:58 cron0 joined #salt
16:58 forrest see what happens?
17:00 jacksontj joined #salt
17:00 felskrone joined #salt
17:02 whyzgeek forrest: I am trying that
17:02 forrest cool
17:04 ksalman ah, I need to refresh the pillar data...
17:05 gldnspud joined #salt
17:06 mwillhite I need to run specs as part of my setup…(using salty vagrant), but when the specs fail it stops building…is there anyway I can not halt the build?
17:06 whyzgeek forrest: yep he was right, it did the trick
17:07 whyzgeek however I am not sure I like it, this way
17:07 whyzgeek I expected it to apply stuff when there is any change
17:07 pmcg joined #salt
17:07 whyzgeek this way it always will be a change
17:08 forrest yea I understand
17:08 whyzgeek unless I do check if it was any change
17:08 whyzgeek with a unless condition
17:08 faust joined #salt
17:08 forrest I guess the assumption is that you won't screw up your repo maybe
17:10 Ryan_Lane joined #salt
17:11 dthom91 joined #salt
17:16 backjlack joined #salt
17:17 faust joined #salt
17:18 carmony joined #salt
17:18 KyleG joined #salt
17:18 KyleG joined #salt
17:20 ksalman is something like this possible? {{ salt['pillar.get']('foo:grains['host']', 'bar') }}
17:20 ksalman I'd like to use grains['host'] in a nested pillar
17:21 ksalman I get an error expected token ',', got 'host'; line 198 in template
17:22 UtahDave ksalman: 'foo:grains:host'
17:22 ksalman ah!
17:22 ksalman UtahDave: thanks!
17:23 juanlittledevil joined #salt
17:23 kula isn't that going to look for a literal key 'foo:grains:host' in your pillar, instead of 'foo:<the value of grains['host']>'?
17:23 UtahDave you're welcome. that's one of the nice things about using pillar.get is that you can traverse down arbitrarily deep
17:23 ksalman yea, UtahDave that doesn't work =(
17:23 ksalman what kula said is what happened
17:24 UtahDave Oh, I see.
17:24 UtahDave 'foo:'+grains['host'], 'bar'
17:24 UtahDave try that
17:25 ksalman oh that works =)
17:25 ksalman thank you
17:27 eculver joined #salt
17:27 eculver joined #salt
17:27 zach Has anyone had luck with installing a set of perl modules via salt?
17:27 berto- joined #salt
17:28 zach via cpan
17:31 mattbarto joined #salt
17:31 tomviner joined #salt
17:31 ipmb joined #salt
17:33 devinus joined #salt
17:34 carmony joined #salt
17:38 zach Is there a way to copy files from the minion(s) to the master within a state?
17:38 z0rkito does anyone have any sort of doc on how you get return data out of a #!py sls file?
17:38 zach I want to copy all of the .gz files within a direectory for archiving purposes that are located on my minion(s)
17:38 opapo joined #salt
17:39 z0rkito everything i've tried seems to make salt think im returning rendered data that needs to be run instead of return data from a state already run.
17:40 TheCodeAssassin joined #salt
17:52 carmony joined #salt
17:57 forrest zach, I can't think of anything other than just running a command to rsync them.
17:59 UtahDave zach:  you might try using module.run  to execute the cp.push execution module function
17:59 UtahDave http://docs.saltstack.com/ref/modules/all/salt.modules.cp.html#salt.modules.cp.push
17:59 UtahDave z0rkito: not sure on that.  You might ask on the mailing list
18:02 krissaxton joined #salt
18:03 jesusaurus UtahDave: i dont remember if i asked you this before, but whats the recommended way to test salt formulae? i have some formulae under code-review, but i want automated unit tests and integration tests
18:03 jesusaurus manual testing is a pain and doesnt scale :(
18:04 luminous joined #salt
18:05 dthom91 joined #salt
18:14 ipmb joined #salt
18:16 ml_1 joined #salt
18:18 UtahDave jesusaurus: we're in the process right now of setting up jenkins to run all our tests.  Then we'll be testing the salt-formulae repos
18:20 jesusaurus what process will you be using to test the formulae? deploy to a testing/staging environment and then test for errors?
18:20 dthom91 joined #salt
18:20 aleszoulek joined #salt
18:20 whiteinge jesusaurus: we'll be running integration tests by actually running the formulas on various OSes & versions and looking for success/changes & failing on errors
18:21 whiteinge we haven't settled on a way to unit test formulas yet. (if you have any ideas, we'd love to hear!)
18:22 copelco joined #salt
18:23 copelco is it possible to use salt to provision your master? maybe running a minon on the master?
18:23 whiteinge copelco: yes, and yes :)
18:23 jesusaurus what i really want (as a first step) is to be able to run something along the lines of: salt-verify --state-top /opt/salt/state/base/top.sls --pillar-top /opt/salt/pillar/base/top.sls
18:24 jesusaurus and have that return success if it can successfully compile the datastructure
18:24 copelco whiteinge: would you just set the master to localhost in the minon config?
18:24 jesusaurus but then you need to get into targeting, and testing for each target
18:25 whiteinge copelco: yes
18:25 copelco cool, thanks :)
18:26 UtahDave jesusaurus: I really like where you're going with that.
18:27 mmilano joined #salt
18:30 pdayton joined #salt
18:31 robertkeizer joined #salt
18:31 littleidea joined #salt
18:32 nliadm is there a blessed way to do service discovery via salt?
18:33 Jahkeup_ joined #salt
18:33 joehh joined #salt
18:34 dthom91 joined #salt
18:50 p3rror joined #salt
18:55 whyzgeek I am trying to install emacs via salt on centos machine. When I do yum install emacs it searches it and finds it and prompts me to agree on installation. When I do it through salt it spits this nstall failed: Cannot retrieve repository metadata (repomd.xml) for repository
18:56 littleidea joined #salt
18:57 Corey whyzgeek: Sounds like a network / repository problem?
18:58 whiteinge whyzgeek: centos 5 or 6?
19:00 krissaxton joined #salt
19:01 whyzgeek whiteinge: 6
19:01 whyzgeek its a local repo
19:02 whyzgeek it works with yum
19:04 loungin_light Anyone have an example of the password hash for salt.states.user.present? Would it basically be a copy of what is in /etc/shadow?
19:05 whiteinge whyzgeek: salt on centos6 uses the python yum library instead of shelling out which may explain the difference. does doing a pkg.refresh_db also fail?
19:06 mattbarto joined #salt
19:06 whyzgeek whiteinge: yes
19:06 whyzgeek that also fails
19:06 whyzgeek wnother hint is that I am using a proxy
19:06 krissaxton left #salt
19:06 carmony joined #salt
19:06 whyzgeek whiteinge: how can I set it to shell out?
19:07 UtahDave joined #salt
19:07 whiteinge whyzgeek: http://docs.saltstack.com/ref/configuration/minion.html#std:conf_minion-providers
19:11 mr_chris Just found an odd issue with the mysql state.
19:12 carmony joined #salt
19:12 whyzgeek whiteinge: can you be more specific? I am very new to salt
19:12 mr_chris If you grant process or super to a user, the grant gets successfully but salt returns the error "Failed to execute: "GRANT SUPER ON *.* TO user@host""
19:13 mr_chris *gets successfully added
19:14 mr_chris left #salt
19:15 whiteinge whyzgeek: yeah, sorry. if you add that "providers" section to your minion config it instructs salt to override the default behavior of which modules to load
19:16 whiteinge on centos5 the yumpkg5 module is loaded as "pkg" and that is the module that shells out to yum
19:16 whiteinge so basically you want to add those first to lines in that example to your minion config
19:16 whiteinge and then your centos6 machine will shell out to yum
19:16 whyzgeek whiteinge: thanks I did it
19:17 whyzgeek it works! kind of!
19:17 whyzgeek I can see the error now
19:17 whyzgeek it runs yum
19:17 whyzgeek however it can't resolve the host
19:17 whyzgeek as it needs the proxy
19:17 whiteinge ah
19:17 whiteinge how do you specify the proxy when running yum on the CLI?
19:17 whyzgeek http_proxy env
19:18 whyzgeek whiteinge: under what user the shell works? I assume it should run it under current one
19:20 whiteinge salt doesn't pick up tne shell of the current user
19:20 whiteinge i don't see a way in the yum module to specify environment variables. i'm sure it would be easy to add though. can you please file a ticket?
19:21 ksalman Anybody tried to install salt on Debian Lenny?
19:23 whyzgeek whiteinge: sure
19:23 whyzgeek whiteinge: I will see if I can fix it and send you a patch
19:23 dthom91 joined #salt
19:24 whiteinge awesome! let us know if you have any questions
19:33 berto- joined #salt
19:50 devinus joined #salt
19:52 juanlittledevil joined #salt
19:56 tuxIO joined #salt
19:57 chrisgilmerproj joined #salt
20:02 KennethWilke joined #salt
20:08 pdayton joined #salt
20:08 logix812 joined #salt
20:15 gatoralli joined #salt
20:26 dthom91 joined #salt
20:26 jY could salt do this.. issue a command on the master to find which server has an ip.. if it does remove the ip and put it on another server?
20:28 Kholloway joined #salt
20:29 bitz joined #salt
20:30 bitz joined #salt
20:31 robertkeizer joined #salt
20:42 zonk1024 joined #salt
20:42 lemao joined #salt
20:44 mattbarto joined #salt
20:51 zonk1024 It's possible to run commands from minions with an ACL, right? or do i have to go multi master?
20:53 whit joined #salt
20:55 gldnspud joined #salt
20:55 littleidea joined #salt
20:57 gldnspud joined #salt
21:00 zonk1024 multi master it is...
21:02 dthom91 joined #salt
21:05 zonk1024 Finally got approved to run salt today, after 2 months or so of asking for it.  One of my bosses is already slightly enamored. :D
21:06 whyzgeek whiteinge: it seems that yumpkg5 provider is using cmd.run module but calling it is a little bit odd. Its not using python imports, it uses __salt__. Checking cmd.run, it seems that it support env parameter. I am not sure how it is possible to pass that paramter to it.
21:06 berto- does the salt minion do any sort of pillar caching?
21:07 whyzgeek The idea is if in the yaml file I add env:<whatever> it should pass it along with run command
21:07 whyzgeek or adds it to provider as an argument
21:07 berto- i have a dynamically generated pillar that gets 'stuck' on boot if the web service it gets data from is not yet available.
21:07 mmilano_ joined #salt
21:08 whyzgeek alternatively, I can simplely change yumpkg5 to use subprocess
21:09 whyzgeek or even write my own provider without touching that
21:09 berto- the error i see is: 'Targeted pillar "server_cluster" not found'
21:09 whyzgeek but I guess its a useful thing for everybody
21:10 jacksontj joined #salt
21:10 berto- if i stop and start the minion, everything works.
21:11 ccrouch_ joined #salt
21:15 nkuttler joined #salt
21:15 nkuttler joined #salt
21:15 TheCodeAssassin joined #salt
21:18 juanlittledevil joined #salt
21:18 UtahDave joined #salt
21:18 mwillhite joined #salt
21:25 druonysuse is there any issue in Salt with running two of the same functions functions like this :  http://paste.opensuse.org/7100982
21:26 TheCodeAssassin Hey utahdave
21:26 druonysuse I remember I once ran into an issue with that, but I don't know if that was a bug that is fixed or it is just the way salt is designed.
21:26 TheCodeAssassin Is it correc that including pillar data does not work?
21:28 UtahDave druonysuse: yeah, you can't do that.
21:28 UtahDave you can't have multiple of the same state and function call under one ID declaration
21:28 TheCodeAssassin as in using pillar data from other pillar files
21:29 UtahDave TheCodeAssassin: That's correct. Right now you can't do that.  There was a discussion today that we're not too far off from being to be able to do that. But right now you can't, unfortunately.
21:29 isomorphic joined #salt
21:30 TheCodeAssassin Might have to a work around then, saves me from making a bug report
21:30 TheCodeAssassin Or is that still advisable?
21:30 druonysuse UtahDave: Oh, awesome. Thank you very much... glad to know I won't run into issues down the line doing this
21:34 UtahDave TheCodeAssassin: search for an existing open issue first.
21:35 UtahDave you're welcome, druonysuse!
21:43 TheCodeAssassin Hehe i always do UtahDave :)
21:43 TheCodeAssassin Thnx :)
21:44 TheCodeAssassin I was banging my head
21:49 tca joined #salt
21:50 TheCodeA_ joined #salt
21:50 jacksontj joined #salt
21:51 tca joined #salt
21:54 QauntumRiff joined #salt
21:55 QauntumRiff really quick question before I can leave for a long weekend.. I have a directory I am creating with file.directory.  set user, group, perms.. but need to set the group setUID bit..
21:55 QauntumRiff ie, I need to do "chmod -R 2775" not chmod -R 775"
21:55 QauntumRiff is that possible to set in the states file?
21:55 UtahDave QauntumRiff: Yeah, set the mode to 2775
21:56 QauntumRiff well, that is way too easy
21:56 UtahDave :)
21:56 QauntumRiff Long week :)
21:57 asimakov joined #salt
21:57 Gifflen joined #salt
21:59 Gifflen joined #salt
22:00 Thiggy Have a good weekend all!
22:02 kermit joined #salt
22:02 jacksontj joined #salt
22:03 tca btw UtahDave do you see anything wrong with this: - connection_pass: {% if salt['mysql.user_exists']('root', 'localhost', pillar["root_password"]) %} {{ pillar["root_password"] }} {% else %} '' {% endif %} ?
22:11 UtahDave tca: are you getting an error with that?
22:12 tca nope, but salt '*' mysql.user_exists with that info returns True
22:12 tca so it should go into pillar["root password"] but it doesn't
22:12 tca goes into the else
22:15 blee_ joined #salt
22:15 randall joined #salt
22:18 UtahDave maybe add == True
22:20 UtahDave s0undt3ch: ping!
22:20 s0undt3ch UtahDave: is it quick?
22:20 UtahDave yeah.
22:20 UtahDave In salt-jenkins
22:20 UtahDave python/setuptools.sls
22:20 s0undt3ch UtahDave: ;)
22:20 s0undt3ch yes
22:21 s0undt3ch UtahDave: using my fork?
22:21 UtahDave We're installing setuptools with this script:  - name: curl -L https://bitbucket.org/pypa/setuptools/raw/bootstrap/ez_setup.py
22:21 s0undt3ch UtahDave: yes
22:21 UtahDave that fails on Cent5 and Cent6
22:21 s0undt3ch UtahDave: to get latest stable, since not all distros provide the same
22:21 s0undt3ch UtahDave: it does
22:21 s0undt3ch hmm
22:21 UtahDave becauase it calls wget
22:21 s0undt3ch UtahDave: ah!
22:21 s0undt3ch ;)
22:21 UtahDave and it needs --no-check-certificate
22:21 s0undt3ch UtahDave: can I fix that in 8/9 hours?
22:21 s0undt3ch UtahDave: ah
22:22 UtahDave that's called within the ez_setup.py
22:22 s0undt3ch ah....
22:22 s0undt3ch bummer7
22:22 UtahDave maybe we call the wget ourselves?
22:22 s0undt3ch UtahDave: need to get the wife. I promisse I'll have a look when I get to work, cool?
22:22 s0undt3ch UtahDave: that could solve it
22:22 UtahDave that's cool, man.
22:22 UtahDave Thanks!
22:23 s0undt3ch UtahDave: or find out which is the env var that does the same as --no-*cert
22:23 s0undt3ch and set that
22:23 s0undt3ch k
22:23 s0undt3ch need to go
22:24 aantony joined #salt
22:25 chrisgilmerproj left #salt
22:26 g4rlic joined #salt
22:26 aantony quick question....   my master is missing some modules needed for salt.virt so it didn't load.   I got my minion set up to have the dependencies.   can I execute virt commands against a minion if the master doesn't have the module loaded?
22:27 g4rlic hello again..  :D  I'm trying to watch a file inside of a service.running: declaration.  The file exists, yet it claims that it cannot satisfy the requirement.
22:27 UtahDave aantony: most execution modules only have to exist on the minion.  I haven't used salt.virt myself yet, though
22:28 UtahDave g4rlic: are you managing that file somewhere in your sls file?  Salt doesn't look at the file system to see if the file is there
22:28 whiteinge whyzgeek: __salt__ is just a dictionary of functions from other loaded modules. you could add an env arg to the yumpkg.install function then pass it through in the cmd.run call
22:28 alekibango joined #salt
22:29 g4rlic UtahDave: I believe so.
22:29 g4rlic http://pastebin.centos.org/4135/
22:29 whiteinge whyzgeek: it looks a little weird to python programmers but it's done that way because salt only loads modules specific to the system it's running on, and also loads them under varying names (as you're doing with the providers setting)
22:29 g4rlic Inside is the error and the offending state.
22:29 kenbolton joined #salt
22:29 aantony dave: yeah thats the impression i got.  the debug log this time does not report the attempt to load virt (one the develop branch)  when I used the 0.16.2 release, it would report the attempt to laod virt, but fail becuase of a missing xml module  dependency
22:29 aantony *on the..
22:29 alunduil joined #salt
22:30 UtahDave g4rlic: ah, you can't have to file.managed entries under one ID Declaration
22:30 g4rlic Oh?
22:30 g4rlic My coworker was under the impression that was OK now. :
22:34 g4rlic I'm guessing managing xinetd daemons with salt is non-trivial.
22:34 druonysuse g4rlic & UtahDave: Sorry. I had just asked about this earlier today... sorry for the confusion, I probably wasn't clear enough with my question... however that is why I used an example pastebin (http://paste.opensuse.org/7100982)
22:34 zonk1024 Exception len([]) != len(['Host key verification failed.', '']) occurred in file server update
22:35 zonk1024 look familiar to anyone for gitfs?
22:35 zonk1024 i can ssh from root into the user that hosts the git repo
22:35 terminalmage zonk1024: this may be an issue that I fixed last week
22:35 terminalmage I recall something just like this
22:35 UtahDave druonysuse: Yeah, you can't do that.  You can't have the same  state function twice under an ID dec
22:36 zonk1024 terminalmage: which issue was it?
22:36 g4rlic Can you have that twice somewhere else? :)
22:36 terminalmage zonk1024: https://github.com/terminalmage/salt/commit/b8eb2e35f2544194c117e106169ebf86e12036dc
22:37 terminalmage zonk1024: it was not reported
22:37 terminalmage I discovered it and fixed it
22:37 zonk1024 terminalmage: sweet.  tyty
22:37 terminalmage yeah this is going to be in 0.16.4
22:37 druonysuse UtahDave: sorry about that... I just like keeping things interesting ;)
22:38 druonysuse UtahDave: thanks again for the help
22:38 terminalmage zonk1024: np
22:38 UtahDave druonysuse: you're welcome!
22:39 JoAkKiNeN joined #salt
22:39 g4rlic UtahDave: what's weird, is that it works in, well, at least one case...
22:40 santagada joined #salt
22:40 g4rlic http://pastebin.centos.org/4138/
22:40 g4rlic That actually works.
22:41 g4rlic which is why it weirded me out when the previous state didn't.
22:41 g4rlic I'm going to go back and un-do those consolidations.  Not worth having to guess about corner cases.
22:42 druonysuse g4rlic: Yep... I agree. Just one of those interesting things. But now we know for sure.
22:44 UtahDave g4rlic: no, that does not work. ever
22:44 g4rlic UtahDave: except it did!
22:44 g4rlic I'll happily paste the output from highstate.
22:44 g4rlic I'll even "recordmydesktop"
22:44 g4rlic because it's freaking me out a little too.
22:45 mwillhite joined #salt
22:45 UtahDave sure, I'd like to see the output.
22:45 UtahDave only one of them can work.
22:46 g4rlic I stand corrected..
22:46 tca UtahDave: == true also does not work
22:46 tca * True
22:46 g4rlic in the case I pasted, I'm only seeing green states for the initrd files.managed, not the vmlinux ones.
22:46 g4rlic Bleh.
22:47 * g4rlic bows to UtahDave's superior salting knowledge.
22:47 g4rlic good thing those pastes expire in 6 hours, my ignorance will hopefully be lost.
22:47 * druonysuse is an irc noob
22:47 mwillhite joined #salt
22:48 UtahDave lol, nah, that's a common idiosyncrasy that people run into
22:48 UtahDave it's because each of those become a dict key.
22:48 UtahDave dict keys have to be unique
22:48 g4rlic Oh!
22:48 g4rlic I see..
22:49 g4rlic So in shor,t there's no way that having multiple file.managed declarations will *ever* work.
22:49 g4rlic no matter how syntactically nice it would be.
22:50 lineman60 joined #salt
22:51 zonk1024 is there an ls for your fileserver?
22:52 UtahDave g4rlic: yes, exactly
22:53 UtahDave you'll have to have another ID declaration for each one
22:53 g4rlic man, that's annoying.  But at least I understand now.  I just need to group at the SLS level, instead of at the ID declaration level.
22:58 jacksontj joined #salt
22:59 ipmb joined #salt
23:00 zfouts joined #salt
23:15 dthom91 joined #salt
23:25 asimakov joined #salt
23:33 mmilano joined #salt
23:43 littleidea joined #salt
23:51 ninkotech joined #salt
23:54 santagada joined #salt
23:55 mmilano joined #salt
23:57 dthom91 joined #salt
23:58 littleidea joined #salt

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