Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2013-12-03

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

All times shown according to UTC.

Time Nick Message
00:06 ajw0100 joined #salt
00:09 oz_akan_ joined #salt
00:12 jpeach joined #salt
00:13 bemehow_ joined #salt
00:19 JulianGindi joined #salt
00:21 renoirb Hey guys, what are the limitations of Macros?
00:21 renoirb Can I have  a set of states definition, e.g. file X has to exist, do this, run that script, etc?
00:28 cachedout joined #salt
00:28 nmistry joined #salt
00:31 Ertac joined #salt
00:34 oz_akan_ joined #salt
00:41 redondos joined #salt
00:42 amahon joined #salt
00:43 fllr joined #salt
00:44 sroegner___ joined #salt
00:48 frosty996 left #salt
00:49 hhatfield joined #salt
00:50 zandy joined #salt
00:53 mgw1 is there a module that parses the output of 'w'?
00:57 jalbretsen joined #salt
01:00 pipps__ joined #salt
01:02 dpippenger joined #salt
01:05 AdamSewell joined #salt
01:09 zandy joined #salt
01:11 oz_akan_ joined #salt
01:11 tdillio joined #salt
01:13 hhatfield joined #salt
01:14 N-Mi joined #salt
01:16 tdillio I'm trying to connect a minion to a master behind a router, the minion seems to be able to connect to the master, but the master never sees the minion. I only have this issue with minions behind a router, is there something special I need to do? http://pastebin.com/jknKp8qE
01:16 zach tdillio: the ports are forwarded?
01:16 tdillio If I do a salt-key -L I never see the minion, if I go setup a minion somewhere in the cloud it connects no prob
01:17 jpeach left #salt
01:17 bhosmer_ joined #salt
01:17 tdillio zach: What ports would I need to forward to the minion? I thought that the minion connected directly to the master?
01:18 zach it does, is the master ports open
01:18 zach ?
01:18 zach are the*
01:18 tdillio Yeah master ports are open
01:18 svnh joined #salt
01:18 tdillio other minions connect no prob
01:18 zach minion does not have a weird firewall restricting outbound connections?
01:18 tdillio I disabled the firewall on the minion machine
01:18 zandy joined #salt
01:18 zach Still no luck?
01:19 tdillio Nah :/
01:19 zach What do you see with tcpdump/wireshark?
01:19 zach any packets getting through?
01:19 NV tdillio: by router, do you mean NAT router?
01:19 zach can you telnet/netcat to the master port?
01:19 tdillio Haven't tried tcpdump yet
01:20 zach I run my master behind a NAT with out an issue
01:20 a1j http://www.rightscale.com/solutions/managing-the-cloud/   need to put saltstack on the list there
01:20 tdillio NV: Yeah, err I think. Its a basic consumer grade router o_0
01:24 zach Is it bad I saw this and thought of salt? https://www.tanga.com/deals/941a4a90ab/what-i-really-need-is-minions-t-shirt
01:25 tdillio NV: Yeah confirmed. I'm behind a NAT router. Is that an issue w. salt?
01:25 NV shouldn't be
01:26 NV but it is a different story compared to a regular router
01:27 zach ^^
01:27 zach consumer router == "NAT Box"
01:28 malinoff zach, Everything else works well? You can ping a service, and it sends ICMP echo back?
01:29 fllr joined #salt
01:29 NV anything in the minion log? (up the logging level to trace if need be)
01:31 tdillio logs just say that it is waiting for the key to be accepted, I can access the master from this machine via SSH
01:32 kermit joined #salt
01:33 zandy joined #salt
01:33 smccarthy joined #salt
01:33 tdillio ah I think I might have found a clue...
01:37 tdillio yeah it was operator error
01:38 tdillio thanks for the help though!
01:39 Gifflen joined #salt
01:41 redondos joined #salt
01:42 zandy joined #salt
01:52 davidfischer joined #salt
01:56 jimallman joined #salt
01:57 MZAWeb joined #salt
02:01 [diecast] joined #salt
02:03 davet joined #salt
02:08 willist joined #salt
02:08 rojem joined #salt
02:14 anitak1 joined #salt
02:15 willist joined #salt
02:34 Doji joined #salt
02:41 xl1 joined #salt
02:54 oz_akan_ joined #salt
02:54 hhatfield joined #salt
03:09 oz_akan_ joined #salt
03:20 willist joined #salt
03:26 dstanek_afk joined #salt
03:28 matanya joined #salt
03:29 xl1 joined #salt
03:33 Furao joined #salt
03:35 Psi-Jack Uhh.. What's this new salt-syndic I just started seeing randomly appear? heh
03:36 pass_by_value joined #salt
03:37 zach Psi-Jack: I saw it today and thought the same thing
03:37 Psi-Jack heh
03:37 andrej Has anyone used salt to force a recompile of vmware-tools on a Ubuntu VMware guest?
03:37 zach "That's pERFECT for DR!" - first thing that came to mind
03:37 Psi-Jack It's auto-starting after an upgrade of salt-master, on CentOS, completely defying the CentOS standard.
03:38 zach Psi-Jack: oh really? I only read about it
03:38 Psi-Jack yeah. Not cool. :)
03:38 zach I will have to VPN into my master and look, sigh
03:39 zach What is the process called?
03:39 Psi-Jack heh, the main reason I noticed it is, my salt runs on a kvm guest, and the console was flooding with "Master hostname:  not found" messages, where hostname was as you can see, blank.
03:39 Psi-Jack salt-syndic
03:40 zach $ ps auxf|grep salt-sy[n] -c
03:40 zach 0
03:40 zach said it happened after an upgrade?
03:40 Psi-Jack yes
03:40 zach $ salt-master --version
03:40 zach salt-master 0.17.2
03:40 zach What is yours?
03:40 Psi-Jack Since, salt-syndic is part of salt-master.
03:41 Psi-Jack 0.17.1
03:41 zach interesting
03:41 zach I don't have that issue on RHEL 6
03:41 Psi-Jack Well, how do you have 0.17.2? heh
03:41 Psi-Jack Eh, I guess EPEL doesn't have salt 0.17.2 yet.
03:41 zach salt-master-0.17.2-2.el6.noarch
03:41 zach It does
03:41 zach It's in epel-testing
03:42 Psi-Jack yeah, as i said. :)
03:42 zach I would suggest upgrading to it
03:42 Psi-Jack I don't use *-testint.
03:42 zach There is a nasty bug in 0.17.1
03:42 zach that I ran into
03:42 malinoff Psi-Jack, Uhm, salt-syndic and salt-master are separated into two different packages
03:42 Psi-Jack yeah, I know.. There's a few.
03:42 NV from, anyone familiar with writing salt states?
03:42 zach I've hit them all, so far 0.17.2 is the most stable one I've ran
03:42 Psi-Jack zach: me and my co-worked helped identify some major bugs, inf act.
03:43 Psi-Jack malinoff: not in epel it's not.
03:43 NV more specifically, is there any way to run code at the start and end of a highstate or sls execution?
03:43 zach cmd.run:\n\s\s- name: cmd\n\s\scwd: path
03:44 zach I don't think you can specifiy the order of execution
03:45 CaptTofu joined #salt
03:45 NV im writing a module and state for vyatta router configs
03:45 CaptTofu hey all
03:45 NV on embedded routers though it takes around 13 seconds to parse the config file
03:46 zach very nice
03:46 malinoff Psi-Jack, Interesting.
03:46 CaptTofu I see on http://docs.saltstack.com/ref/modules/all/salt.modules.linux_lvm.html the various commands, but don't see lvrename
03:46 CaptTofu does anyone know if it's there?
03:46 Psi-Jack Cool, I just installed salt 0.17.2 from epel-testing as the only packages from there. Not bad. :)
03:46 NV so ideally one would only parse the running config once, figure out what changes are to be made (by comparing the state vs the running config, and making changes as appropriate including updating salts internal representation of the state of the config) then committing the changes at the end
03:47 NV commits take like 30 seconds as well (as it re-parses)
03:47 NV so basically staging all the changes and committing it as one big chunk to save time
03:47 zach yum list all |grep salt:  https://pastee.org/b36ga
03:48 NV else if you have even 10 states that change the vyatta config it'd take 5-10 minutes to apply (and even several minutes doing a highstate even if no changes are to be made just from loading the config over and over!)
03:48 malinoff NV, you can write a state with order: 1 - it will be executed at first; and a state with order: last - it will be executed at last :)
03:48 NV seems dirty, would prefer it if the state module was able to handle it internally
03:49 AdamSewell joined #salt
03:49 zach Psi-Jack: FYI...you'll need to update your minions too
03:49 zach Psi-Jack: they do NOT play well together - 17.2 and 17.1
03:49 Psi-Jack zach: yeah, already in progr
03:49 zach cool, I forgot to do that...
03:49 zach :)
03:49 Psi-Jack heh
03:49 malinoff How do you see this? You run salt 'minion-id' state.sls slsname ... and what?
03:49 NV zach: i thought that was only 17.0 and 17.1?
03:50 zach was not a fun task updating them manually across 90 machines
03:50 Psi-Jack zach: Try 200+
03:50 shinylasers joined #salt
03:50 zach oh man, no thanks
03:50 Psi-Jack :)
03:50 Psi-Jack And all CentOS boxes. :)
03:51 zach I just just ran a for loop doing yum install <epel mirror url>.rpm with some expect magic
03:52 malinoff http://sourceforge.net/projects/pacmanager/ - can join hosts into a cluster; you run a command on a single host - it runs on every node in the cluster
03:52 zach neat, sounds like a nightmare for a PCI/HIPAA environment
03:53 * zach misunderstood
03:53 zach I did not realize it was just a putty-like client
03:53 malinoff Yep
03:53 Psi-Jack Interesting.
03:53 zach now my 700 line perl script has been obsoleted
03:53 zach but it pulls passwords from a keepass store
03:54 malinoff PAC can do this :)
03:54 zach and runs them in parallel
03:54 zach Can it do that too?
03:55 Psi-Jack And even provide rpms, how sweet of them.
03:55 zach Mine can handle (what my laptop can handle) 150 forks at once
03:55 zach If I had a machine with more ram and faster CPU, I could do 200-500 at a time
03:56 malinoff It's not consecutive - if you're about that
03:57 malinoff It's like you type in many shells at one
03:57 malinoff once*
04:01 Psi-Jack It's like, basically, terminator's key repeater.
04:02 Psi-Jack Except... In this case. Better.
04:02 malinoff Psi-Jack, yes :) I use it for 1 month, and I like it. Quite simple and useful
04:02 Sheco joined #salt
04:03 Psi-Jack Eh, so far, seems like a PITA to initially setup.
04:03 malinoff What? dpkg -i, thats all :)
04:04 Psi-Jack Ahem.. Not install. Set up.
04:04 malinoff Ah, got it
04:05 malinoff But it can be tuned once, and it'll work for years
04:05 Psi-Jack heh, they make no easy means to replicate prior connections, either. even cloning is an overly stupifying tasj.
04:08 jfzhu_us joined #salt
04:10 oz_akan_ joined #salt
04:11 JoeDev joined #salt
04:11 Psi-Jack And you can't organize it well, either. :/
04:12 Psi-Jack You can't setup a singleton group with subgroups, without issues.
04:13 sandGorgon joined #salt
04:13 rojem joined #salt
04:13 JoeDev Hi. Anyone had luck using a salt state to setup a rvm wrapper for a gemset? Keep getting State rvm.wrapper found in sls is unavailable.
04:16 nhanpt-rad joined #salt
04:16 nhanpt-rad hallo
04:18 NV hrm, mod_init appears useful
04:18 dpippenger joined #salt
04:24 ckao joined #salt
04:25 bemehow joined #salt
04:25 mgw joined #salt
04:36 Guest64131 joined #salt
04:43 forrest joined #salt
04:48 willist joined #salt
04:51 Gifflen joined #salt
04:55 wilywonka joined #salt
05:04 jalbretsen left #salt
05:08 Gifflen joined #salt
05:11 oz_akan_ joined #salt
05:11 nmistry joined #salt
05:12 AdamSewell is it possible to create a state that resets all permissions in a directory and changes the owner/group as well?
05:13 forrest AdamSewell, how about a http://docs.saltstack.com/ref/states/all/salt.states.file.html#salt.states.file.directory
05:13 forrest there are some examples up above that
05:13 forrest you could also just run a recursive chown/chmod with cmd.run
05:14 AdamSewell true
05:14 AdamSewell thanks forrest
05:15 forrest np
05:23 AdamSewell forrest, have you ever setup salt to auto create a mysql database, user and such?
05:25 forrest AdamSewell, I have not, however terminalmage has a great example over here: https://github.com/terminalmage/djangocon2013-sls He's using the mysql formula, and then within foo/db.sls he's creating a database, a user, and setting up grants.
05:26 forrest you might have to modify some of the pillar data, but overall it should be very similar
05:26 AdamSewell interesting, thanks
05:26 forrest np
05:27 matanya joined #salt
05:29 xmltok joined #salt
05:31 pass_by_value joined #salt
05:31 pass_by_value left #salt
05:31 philipsd6 joined #salt
05:46 philipsd6 joined #salt
05:50 davet joined #salt
05:51 zandy joined #salt
05:52 sandGorgon joined #salt
06:00 Gifflen joined #salt
06:01 supplicant joined #salt
06:02 matanya joined #salt
06:08 mesmer joined #salt
06:13 oz_akan_ joined #salt
06:18 philipsd6 joined #salt
06:20 terminalmage joined #salt
06:22 pcarrier joined #salt
06:30 AdamSewell joined #salt
06:36 jeddi joined #salt
06:37 mr_chris joined #salt
06:40 sunand joined #salt
06:43 MTecknology *grumble*
06:43 MTecknology http://dpaste.com/1491015/ <-- All my other servers work except for this one
06:46 MTecknology salt seems to have developed sentience and for some reason wants to cause me to harm myself.
06:47 MTecknology heheheh..... I see why
06:47 packeteer dont you hate that
06:48 packeteer spend ages tryong to figure it out, as soon as you ask for help you see the problem!
06:48 MTecknology indeed
06:48 redondos joined #salt
06:49 MTecknology this one was because the fs filled up and things broke, salt had enough space (the default 5%) to create the proc file for a process, but went boom when trying to actually execute it.
06:49 MTecknology It didn't clean up dead processes
06:49 MTecknology so.... kept going boom
06:50 packeteer heh
06:52 MTecknology I feel like that could be a bug, but not sure
06:52 MTecknology it would be one heck of a thing to change.
06:53 malinoff It's like when I tried to unzip a 12 GiB archive (around 1 GiB zipped) in a system with 2 GiB free space available. unzip wrote messages like 'Copied 1075 bytes of 4096' and I thought 'what the hell'
06:54 MTecknology I'm used to that kind of thing enough that I know exactly what it's whining about. I don't screw with zip, though.
06:54 MTecknology I wish I could figure out this issue with rhel boxes.
06:55 jimallman joined #salt
06:57 packeteer ?
06:59 Nexpro1 joined #salt
07:00 MTecknology loooooooong story; but pretty much - generating pillar data breaks the master process and so the state.highstate execution breaks because the pillar data isn't available.
07:01 packeteer put up a gist, maybe somebody will know
07:04 MTecknology dave_den was working pretty hard to help me figure it out earlier today and we came up with nothing. I think it's a bug in 0.17.2. A gist won't come close to explaining all the weird crap.
07:04 packeteer k
07:06 MTecknology I don't even have any box here that I could test with
07:09 terminalmage joined #salt
07:11 MTecknology packeteer: I'll be at work in 7hr if you want to help me try to debug it then.
07:18 packeteer unfortunately i'll be sleeping then, something about being on the otherside of the globe  :)
07:18 packeteer else I'd love to help
07:20 favadi joined #salt
07:24 Furao joined #salt
07:33 vipul joined #salt
07:35 MTecknology g'night
07:38 juasiepo joined #salt
07:43 Kizano joined #salt
07:57 redondos joined #salt
08:06 giantlock joined #salt
08:09 NV hrm, is it kosher to modify the lowstate in mod_init?
08:12 anitak joined #salt
08:15 oz_akan_ joined #salt
08:18 NV oh damnit, the entire lowstate list isn't passed in, just the one for that specific state call, bugger
08:18 malinoff NV, I still think adding order: first and order: last is the best choice :)
08:20 NV well i can get around requiring order: first using mod_init
08:20 NV so i can parse the config there and save it somewhere
08:21 NV but being able to defer committing until the last state has executed needs some way of triggering afterwards
08:21 NV (plus cleanup so the config isnt sitting in memory all the time for no reason)
08:22 malinoff Damn we have overstate, have you looked at it? http://docs.saltstack.com/ref/states/overstate.html
08:24 slav0nic joined #salt
08:28 NV do you have to use overstate to use any other state module?
08:28 NV :P
08:30 malinoff I may use it, yes :)
08:32 NV you misunderstand, i mean is there a single state that is functionally broken and impossible to use unless you also use overstate?
08:33 terminalmage joined #salt
08:35 malinoff Hm, I think, if a state module is broken, you can't use it wherever you try to. Overstate is just a sequence of actions
08:36 NV exactly, your suggestion of using overstate to work around the issue if there being no mod_finalise in a similar manner as mod_init would result in a broken state ;)
08:36 NV (see: i want to be able to defer committing until the end rolling in the changes from multiple invocations of the state so it doesn't take 5-10 minutes to execute)
08:37 NV doing a commit has around 30 seconds of overhead no matter how trivial the change - but committing many changes at once only takes a little longer than committing a single change
08:38 NV but forcing everything to be munged into a single state execution sort of defeats the purpose of configuration management, might as well just copy-paste a static file and load that
08:38 NV so multiple state invocations to build up all the changes in memory, then commit them all at the end, saving 30 seconds * <number of changes> in time
08:38 malinoff Alright, got it. I don't think I know how to do that :) I know, there is failhard option that prevents highstate execution if any error occurs, but how to make atomic actions...
08:39 gasbakid joined #salt
08:39 NV eh, the idea is to break atomicy
08:39 NV if i wanted it to be atomic it'd be easy, commit for each state invocation
08:39 NV it'd just take a really long time
08:39 malinoff I mean, not actions, all the highstate. Right?
08:40 malinoff Or I'm missing something again? :)
08:40 NV basically you will have <N> invocations of a vyatta.config state in a single highstate (or even sls file)
08:40 NV all those states need to be executed, then at the end of it the changes have to be committed
08:41 NV so yes, adding a bogus state at the end of the highstate (by doing order: last) would work, it's rather dirty and doesn't follow how every other state works
08:42 NV meanwhile committing for every <N> invocation of the vyatta.config state would make it unacceptably slow (although a flag would be provided in case dependancy ordering requires it for a specific section)
08:42 NV like your config state that sets up your system package repos would have the immediate_commit flag set to true, so that pkg states can require it to ensure the pkg repos are setup before trying to install packages
08:43 malinoff Are you familiar with puppet?
08:43 NV but 90% of the time the changes wouldn't be depended on by other states, so the order of when the states changes are actually committed is irrelevant
08:43 NV yes, currently converting from puppet (although what I'm doing now isn't something i've had previously setup in puppet)
08:44 Sypher left #salt
08:45 Sypher joined #salt
08:45 malinoff Looks like you try to follow puppet's way - compiling everything on the master. I really don't know how to tell salt to group all these changes into just a single change - since every action compiles on a minion and is quite independent
08:51 malinoff Btw, I can't find vyatta state module, is it your own module?
09:03 NV I'm writing one, yeah :)
09:05 malinoff Can you pastebin vyatta.config function?
09:06 NV it's not written yet :P
09:06 NV have the config parsing stuff done, still yet to do the state side of things to reconcile between the running config and the config in the state
09:07 NV http://vyattawiki.net/wiki/Cli-shell-api is what I'm wrapping
09:08 NV note there's a 'showCfg' command which outputs the current config (which I have written a lexer and parser to read), and commands to set specific configuration nodes
09:08 Ryan_Lane joined #salt
09:09 NV the idea is read the current config, lex and parse it (~15 seconds for my current config, longer for larger ones), for each state compare against the in-memory parsed config, if changes have to be made update the in-memory config and run the vyatta set command, if not do nothing, wash wrinse repeat for all other invocations of the state within the highstate, then at the end run commit and cleanup the in-memory representation of the config (~30 seconds)
09:09 NV so most of the time it should take less than a minute, which is acceptable
09:10 juasiepo joined #salt
09:10 NV but if each invocation was done seperately (consider the entire config of vyatta is realised via this interface, so 10+ invocations would be easily hit) you'd be talking about ~45 seconds each which quickly adds up
09:11 tomspur joined #salt
09:12 balboah joined #salt
09:13 malinoff Let me think...
09:15 oz_akan_ joined #salt
09:17 donatello joined #salt
09:21 malinoff It's too complicated to write it here. I'll try to arrange my thoughts and pastebin it. Not sure when I'll do that, probably, tomorrow
09:23 Damoun joined #salt
09:35 mike251 joined #salt
09:35 * mike251 morning from Frankfurt
09:40 mike251 hey guys ... is it possible to ... append a line of text to a file ONLY if it is not already there... so when i run highstate, i do not want to have the same line like 50 times in a conf file
09:40 NV yes
09:40 NV file.append
09:40 mike251 i was just looking at that :) thanks NV
09:41 carlos joined #salt
09:43 lynxman joined #salt
09:43 zooz joined #salt
10:09 micko joined #salt
10:16 oz_akan_ joined #salt
10:16 sunand [CRITICAL] Unable to import msgpack or msgpack_pure python modules is this an error?? i am using salt-ssh to install a package of lower version where higher version is already insatlled?
10:19 rjc joined #salt
10:20 zooz joined #salt
10:23 NV sunand: that's a bug with salt-ssh
10:24 mike251 hey guys - i have an  error with file.append :   http://pastebin.com/6ZWEHQ0y
10:24 NV msgpack is a dependancy of the salt minion, but salt-ssh doesn't need it
10:25 NV mike251: state.show_sls on the sls file plox
10:25 NV wait hang on
10:25 mike251 ..?
10:25 NV sec
10:26 mike251 i want to add 2 lines of text :) to the /etc/aliases ... from a jinja
10:26 NV no nevermind, as you were, salt-call state.show_sls <slsname>
10:26 NV kpaste
10:26 mike251 on the minion or master?
10:27 NV on the minion, or salt 'minion' state.show_sls <slsname> on the master
10:27 NV either works
10:27 NV (in theory you could run salt-call on the master, but that would rely on the jinjamap having the same contents, etc - which it may not if it depends on pillar or grain data, etc)
10:27 NV show_sls shows the rendered sls, ie post-jinja/etc
10:27 mike251 ah right
10:28 NV so quite useful to figure out what's actually happening when you get weird errors with the state execution
10:28 NV sometimes what you think is happening in jinja, actually isn't
10:28 mike251 http://pastebin.com/yXi4xJEZ
10:28 NV like my suspicion is that rootemail or platformemail is a dict
10:28 NV and it is! :D
10:28 NV both are in fact
10:29 NV the former is a dict with the key 'root', and the value 'sysadmin@3605.com'
10:29 mike251 'rootemail' : 'root:           sysadmin@360t.com',
10:29 NV i suspect you actually want a string of "root: sysadmin@360t.com"
10:29 mike251 yes!!
10:29 mike251 aaa i got it now... it transformed my string into a dict
10:30 mike251 bastard :))
10:30 NV yup :)
10:30 NV show_sls is your friend
10:30 mike251 NV: another GREAT day .. to learn something
10:30 NV (also show_highstate, but often show_sls is quicker as you generally only care about a specific sls file)
10:30 mike251 if i use double quotes .. should work?
10:30 NV haha no idea
10:30 mike251 damn :)
10:30 NV I'm not a jinja exepert :P
10:30 scott_w joined #salt
10:30 NV expert**
10:30 NV can't even spell it hah
10:30 mike251 i will let you know
10:30 mike251 haha
10:31 NV actually
10:31 NV jinja isn't doing it
10:31 NV it's actually the yaml parser!
10:31 mike251 aha
10:31 NV jinjamap.postfix.aliases.rootemail
10:31 NV is a string
10:31 mike251 yep
10:31 NV but jinja then makes it
10:31 NV - root: blah
10:31 mike251 yep
10:31 NV throw quotes around the {}'s
10:32 NV - '{{ ... }}'
10:32 MrTango joined #salt
10:32 mike251 i wanted to do that when i created the sls... but i said .. wtf no need for it :)
10:32 NV haha
10:33 mike251 state.show_sls works now...
10:33 NV the thing to remember is jinja executed to make a plain ole yaml file, then the yaml is parsed
10:33 NV on one hand powerful, on the other hand, easy to bite yourself like that :P
10:33 mike251 aha... good to know
10:33 mike251 again THANKS a lot
10:34 mike251 salt is great... in couple of days i have managed to do a lot of things... on the test platform
10:34 mike251 especially files/conf management
10:34 NV hehe, a fun one is file.managed it has a contents_pillar option that can populate a file from a pillar value directly to work around issues like this with the yaml formatter
10:34 mike251 i really like it... and my boss as well.. so we are putting it on a live environment soon-ish
10:35 NV but unfortunately there isn't the same thing in file.append, and it only works if you're sourcing your data from pillar of course
10:35 mike251 yeah... i souorce data from a jinja file (one file per environment) and the pillars are used to .. add minion to env, groups etc.
10:35 mike251 at least this is my setup
10:36 NV mhmm
10:36 NV I love how easy it is to break out into python when the going gets rough to do really funky stuff in pillar and states
10:36 ro_ikou joined #salt
10:36 NV and the ease of distributing custom modules in _states and _modules
10:37 mike251 pff .. i keep away from that... first i have to learn a bit of py ... but there are so many things to do.. I also started a web app that integrates with salt... a web dashboard...
10:37 NV saved my butt a few times when I've needed a fix in a newer version of a state or module but not willing to update all my minions to git, etc :P
10:37 NV isn't there already a web dashboard jigger? halite?
10:37 mike251 ah
10:37 NV (Although I'm also in the process of writing a module and state of my own atm :P)
10:38 mike251 i think there is... but ... i want more.... i want to RUN salt commands from the web dashboard... to add minions to groups... with drag and drop etc
10:39 mike251 is more.. a web dashboard for me... but.. i might put it on github... for others to play around with.... i am still @ beginning...  but things are going in the right direction
10:39 NV ftr, adding minions to groups - how are you planning on that? ext_pillar? or using a python rendered pillar?
10:39 mike251 too bad i am the only one in my team with web skills ... i used to be a web dev... then switched to sysadmin
10:39 NV I hate webdev personally :P
10:39 mike251 good question
10:40 NV it's not that I'm bad at it, I just really don't like it :P
10:40 NV (well, and the things I make end up looking like arse. Functional, but look like arse)
10:40 mike251 via normal pillars.. php will generate the entire pillar
10:40 NV eww
10:40 mike251 i know what you mean... my boss tells me... that ... i spend too much time on making things perfect...
10:40 NV why not write a pillar sls in python and have that query your mysql db directly?
10:41 mike251 because i do not know py :P
10:41 NV python is easy :P
10:41 mike251 is pretty similar to php ... with some major differences :)
10:41 NV I learnt python by hacking on code, all things considered I'm fairly new to python
10:41 NV From a C, C++, Java, Perl, etc background
10:41 mike251 is about... not knowing how to start doing things.. that's what i hate
10:42 NV just dive in! it's easy :P
10:42 NV arguably easier than php
10:42 NV (less gotchas and sadness)
10:42 NV obligatory http://www.phpsadness.com/
10:43 NV the interactive mode (just type "python" at a shell) makes things really easy to quickly test things too
10:47 viq Does salt cache somewhere the ext_pillar git contents? I changed the git repo config, yet it still seems to try to use the old location...
10:49 pengunix joined #salt
10:53 Ryan_Lane joined #salt
10:54 mike251 NV:  thanks!! a lot
10:55 NV viq: yes, there is a command to refresh it i believe but I can't think of it off the top of my head (I don't use ext_pillar with git, I used to but ran into issues which I believe are now fixed, but I haven't gotten around to changing back yet)
10:57 mike251 salt '*' saltutil.refresh_pillar
10:57 NV mike251: that tells a minion to refresh its pillar data, but doesn't signal to the master to pull it down from git
10:57 mike251 ah sorry .. i didn't pay attention... it is ext_pillar... sorry
10:58 NV like salt-run fileserver.update for example forces a git pull of the gitfs for states
11:01 NV hrm, looks like there isn't a way to do it manually, but it should happen whenever pillar data is retrieved anyway
11:01 NV https://github.com/saltstack/salt/issues/8943 according to this bug report that was opened yesterday anyway
11:02 viq NV: thanks
11:02 NV git pillar should probably be updated to be closer to gitfs for states
11:03 viq also, for reference, /var/cache/salt/master/
11:05 NV protip, don't play with that directly yourself :)
11:06 NV if you do want to manage it manually, use the git state to manage a checked out git repo and use a regular ole file-based pillar
11:06 NV (which is what I currently do due to since-fixed issues i encountered in 0.16)
11:06 viq Since it's "cache" I just did rm -r on the stuff I wanted to refresh :P
11:06 NV ah yeah, that should be fine
11:07 NV just don't try git operations on it :P
11:07 viq Yeah, it made master reach into the new place for data, instead of old invalid one
11:07 viq Kind of got bitten, just starting with it, and just learned that states and pillar need to be in separate repos
11:09 Tiriel joined #salt
11:09 Tiriel hi!
11:10 Tiriel Is it possible to run salt-call and get the pillar data from the minion's file system?
11:10 NV are you talking about a masterless setup? or?
11:11 Tiriel yes
11:11 Tiriel actually no
11:11 viq Tiriel: how about on master 'salt minion pillar.get <something>' ?
11:12 Tiriel I want the minion to get the sls from the master but the pillar from itself
11:13 Tiriel as in, to get the minion on a state that the master knows about but with pillar data that the master does not know about
11:14 NV does it have to be pillar data?
11:14 NV http://docs.saltstack.com/topics/targeting/grains.html#grains-in-etc-salt-grains
11:14 NV etc
11:15 Tiriel I think it has to be pillar, but I'm not sure
11:16 Tiriel let's see by example. Take the apache formula (from saltstack-formulas)
11:17 Tiriel If my master has this git repo configured as a backend but has no pillar data configured for the vhosts.standard state, can I place that pillar on a minion so as to call salt-call and configure itself with the local pillar config?
11:17 oz_akan_ joined #salt
11:18 viq Tiriel: just out of curiousity, could you explain why do you want to set it up that way?
11:19 Tiriel It's a bit of an off use case.
11:19 NV then change the formula to use grain data then?
11:20 Tiriel I wanted to know if I could set a master to work as a states server of sorts. A service that could provide pre-configured states for minions to come and get but not to actually control minions with
11:20 NV grains come from the minion, pillar from the master (for a specific minion)
11:20 NV yeah, just change the formulas provided by the salt master to use grain data instead then
11:21 NV kdun
11:21 Tiriel the idea is that users of the service could get a brand new machine, call into this server and have, for example, a LAMP stack configured
11:24 Ryan_Lane joined #salt
11:25 mike251 Tiriel: what about the os deployment? what do you use?
11:25 mike251 salt will manage the lamp install... but the os?
11:26 1JTABBYVA Tiriel: you can use minion with local config
11:26 1JTABBYVA that might work best
11:27 1JTABBYVA states really are about control
11:27 Tiriel That would be out of the scope of the exercise, we assume the minion can be any machine that has salt installed
11:28 blee joined #salt
11:28 Tiriel local config won't cut it either.
11:29 Ertac joined #salt
11:29 NV Tiriel: or the other way around it is you define 'roles' as a grain that selectively enables states and associated pillar data on the server
11:29 NV consider you still need a way for the states to be included in the highstate too (unless you want each state to be manually executed one-by-one, but that's a pita)
11:30 Tiriel Actually the states executed one by one is an option on this setup
11:30 viq Tiriel: not sure "any machine" will work, as I think master would still need to accept the minion's key
11:31 Tiriel "any machine" will work as long as the master is on auto_accept and open_mode, it's not secure but it's a proof of concept
11:31 viq ah
11:31 viq How about all machines having a git checkout of the state tree, and operating in local mode? Or that's out as well?
11:32 Tiriel No, that's an option
11:32 NV Tiriel: consider that even with how you have defined it, the salt master still has root on all the boxes whenever it wants
11:32 Tiriel but getting the pillar data locally would make things easier
11:32 NV so if they are truely third party boxes, some people might take offense to that level of power given over their box to a third party
11:32 viq NV: well, not if they don't run the minon agents
11:33 NV viq: sure, but we're assuming it's on by default etc
11:33 NV compare to a local masterless minion, entirely self contained and no issues if the central salt master gets compromised
11:33 viq NV: do you know if in local mode it would read pillar data off of local filesystem as well, or reach out to master?
11:33 NV include some states to update the local states from a central git repo, kdun
11:33 Tiriel I'm aware of that, again, proof of concept.
11:34 NV viq: never used masterless tbh :P
11:34 viq NV: neither have I ;) But it sounds a bit like a solution here
11:34 Tiriel It would be nice if masterless operation could use git as a backend
11:34 NV indeed
11:34 Tiriel we wouldn't be having this conversation then
11:34 viq Can't it?
11:34 Tiriel not at the moment.
11:35 viq Ah
11:35 NV Tiriel: salt-call state.sls update
11:35 NV where update.sls contains a git.latest state (or two)
11:35 Tiriel It would be worth having a look to see how difficult is it to implement
11:35 Tiriel would that be a local call?
11:36 NV Tiriel: yup
11:36 Tiriel worth looking into
11:37 NV gives the user options to overlay their own states/etc over the top too
11:39 zooz joined #salt
11:44 Tiriel the local flag is shared for sls and pillar, but it shouldn't be too difficult to segregate them
11:45 juasiepo joined #salt
11:46 viq Tiriel: you don't need to if you keep both in say git and push that to machines
11:47 _ikke_ joined #salt
11:48 Tiriel Yes, but I'm trying to make it even easier
11:48 babilen left #salt
11:49 Tiriel best way is if the local flag on the minion was able to pull from git.
11:52 baoboa joined #salt
11:54 sroegner___ joined #salt
11:55 gasbakid joined #salt
11:59 gasbakid_ joined #salt
12:06 Sheco joined #salt
12:18 oz_akan_ joined #salt
12:28 balboah joined #salt
12:33 Destro joined #salt
12:38 mike251 guys... i have a set jinjamap = salt['grains.filter_by']  where i set up a lot of variables depending on the pillar['env']. My question is ... if one of these jinja values can be another {{ another_jinja }} defined before this jinjamap
12:39 dmanser joined #salt
12:39 dmanser left #salt
12:40 scott_w joined #salt
12:43 mike251 it works... i just need to use annother_jinja without the {{}}
12:43 ckao joined #salt
12:52 MrTango joined #salt
12:54 jslatts joined #salt
12:55 ggoZ joined #salt
13:02 bhosmer joined #salt
13:09 ipmb joined #salt
13:09 diegows joined #salt
13:10 bhosmer joined #salt
13:11 jergerber joined #salt
13:14 jfzhu_us joined #salt
13:14 danielbachhuber joined #salt
13:19 bhosmer_ joined #salt
13:22 bhosmer joined #salt
13:22 harobed joined #salt
13:25 oz_akan_ joined #salt
13:26 Ryan_Lane joined #salt
13:28 vejdmn joined #salt
13:28 willist joined #salt
13:29 oz_akan_ joined #salt
13:38 sroegner___ joined #salt
13:39 sroegner____ joined #salt
13:41 mpanetta joined #salt
13:42 amahon joined #salt
13:43 harobed joined #salt
13:44 harobed joined #salt
13:48 tyler-baker joined #salt
13:54 patrek joined #salt
13:55 schristensen joined #salt
13:56 jslatts joined #salt
14:01 jeffrubic joined #salt
14:02 ze- Hi. Is there a way to filter on multiple grains at once ?
14:02 ze- like -G rule1 -G rule2
14:02 ze- and only use hosts matching both rules ?
14:03 mpanetta ze-: Take a look at the -C option
14:03 mpanetta It will do what you want.
14:06 jslatts joined #salt
14:07 JasonSwindle joined #salt
14:08 ze- thanks, checking :)
14:08 mpanetta ze-: No probs :)
14:09 JasonSwindle left #salt
14:10 Gifflen joined #salt
14:10 willist joined #salt
14:10 sroegner joined #salt
14:12 juicer2 joined #salt
14:13 MTecknology http://dpaste.com/1491352/ ... thanks rhel, making life pain for me yet again
14:13 racooper joined #salt
14:13 jankowiak joined #salt
14:15 Ixan MTecknology: if you're using rhel, why not use the package from epel-testing?
14:16 tyler-baker joined #salt
14:17 MTecknology Ixan: I was trying to get zmq3 and it apparently has a weird dep issue with python26-zmq
14:18 wilywonka joined #salt
14:18 Ixan MTecknology: don't think you'll get anything running on python 2.4 regardless
14:18 Ixan MTecknology: you're using rhel 5?
14:18 MTecknology yup
14:19 MTecknology I'll give up on zmq3 then?
14:20 Ixan you wont be able to run salt at all, with python 2.4.
14:20 JulianGindi joined #salt
14:20 MTecknology I have 2.6 on there
14:21 MTecknology salt was working fine up until I updated
14:21 Ixan well, not according to your paste. the easy_install you're using deploys to python2.4
14:21 scott_w_ joined #salt
14:22 MTecknology oooooh
14:22 Ixan if you have 2.6 additionally, check if you have an "easy_install2.6" or something
14:22 brianhicks joined #salt
14:22 amahon joined #salt
14:23 MTecknology heh... I have an easy_install and an easy_install-2.4 but not 2-6
14:24 mike251 guys ... quick Q: if i have a state ... that has only the include with 3 other states A B C. When i run parent state, are the states inside being run first A, then B and at last C ? ... so from top to bottom?
14:25 Ixan mike251: http://docs.saltstack.com/ref/states/ordering.html
14:28 mike251 hmm .. not sure.. that ... that is what i need. ..
14:29 mike251 i have 3 states independent states... that can be run separately ...but when added to a parent state file... they need to run in sequence... in the same order i have added them
14:30 wilywonka joined #salt
14:30 mike251 can i add the ORDER to an include statement?
14:31 mike251 like.. include 1st, then second...etc?
14:31 mike251 that would be good...
14:32 Brew joined #salt
14:33 Brew joined #salt
14:39 mgw joined #salt
14:40 bhosmer joined #salt
14:41 zandy joined #salt
14:42 quickdry21 joined #salt
14:42 renoirb mike251: this is what require, require_in, and others are for.
14:43 zandy_ joined #salt
14:43 mike251 maybe ... i don't understand correctly...  or i didn't explain right
14:43 AdamSewell joined #salt
14:43 AdamSewell joined #salt
14:44 mike251 i have 3 states ... A B C that run independently ... and i want them to be like that.  Then i created another parent.sls .. where i include A B C ... but the order in which they are run ... has to be exactly A B C ... is that possible - without editing A B C states and add require?
14:47 matanya joined #salt
14:47 MTecknology Ixan: I finally got easy_install-2.6 installed, now this... http://dpaste.com/1491377/
14:47 MTecknology this is fun
14:49 whiskybar joined #salt
14:49 aleszoulek joined #salt
14:52 oliv` joined #salt
14:53 Ixan MTecknology: yum install zeromq-devel
14:54 scoates I have a state that patches one of the Minion's salt modules. Is there a clever way to get that to both 1) run first, and 2) restart that minion before applying the rest of the states?
14:54 Ixan if you get bored of compiling, you can use epel 5... latest package for rhel 5 is 0.17.1
14:55 Ixan you'd probably be better off
14:57 teskew joined #salt
14:57 MTecknology heh.. I had that installed but had to update it
14:58 scott_w_ joined #salt
14:59 danielbachhuber joined #salt
14:59 bhosmer_ joined #salt
15:00 MTecknology FRICK!
15:00 lemao joined #salt
15:00 MTecknology It worked... sorta.
15:02 th3reverend joined #salt
15:02 AdamSewell is it possible to an order in which cmd.run commands are executed? my ufw state is enabling itself before my allow ssh and other rules
15:03 mike251 AdamSewell: just create 2 different definitions and... require one or the other
15:04 mike251 each cmd.run a different definition..
15:04 mike251 and i think with a stateful: True... if i am not mistaking
15:04 Furao joined #salt
15:05 MTecknology I finally have salt installed and using zmq3. Now /var/cache/salt/master/minions/minion_name/mine.p is getting created. pillar.data is still returning nothing.
15:05 MTecknology dave_den: wanna do any more playing with my issue now that I got further?
15:06 MrTango joined #salt
15:08 MTecknology dave_den: that mine.p file is created but has only a single character '�' in it. So... the data apparently isn't being rendered but the master isn't crashing anymore, maybe? or maybe just handling it better.
15:10 MTecknology Ixan: or.. maybe you have ideas?
15:12 Ixan i don't know which issue you're having
15:12 kaptk2 joined #salt
15:12 alunduil joined #salt
15:15 mgw joined #salt
15:16 MTecknology Ixan: running state.highstate gives me this - http://dpaste.com/1491400/; When I run salt 'host' pillar.data from the master or salt-call pillar.data on the minion I get nothing returned. However, I set pillar_opts: True in the master config. It should be returning a whole lot of stuff.
15:18 AdamSewell does salt-cloud use something other than the salt ports and ssh when running the default state scripts?
15:18 Ixan is your pillar file targeted to the minion in the pillar topfile?
15:19 MTecknology yup, it's targeted to '*'
15:20 Ixan have you set pillar_roots?
15:22 MTecknology yup- on debian boxes I see a pile of pillar data (all the funky stuff the master generates too), but on rhel boxes, zero pillar data
15:22 jperras joined #salt
15:24 higgs001 joined #salt
15:24 Ryan_Lane joined #salt
15:25 forrest joined #salt
15:27 jsm joined #salt
15:27 th3reverend left #salt
15:35 jkleckner joined #salt
15:36 timoguin joined #salt
15:37 pentabular joined #salt
15:37 marcinkuzminski joined #salt
15:39 abe_music joined #salt
15:41 jalbretsen joined #salt
15:42 MTecknology Ixan: I'm starting to wonder if this is a bug in 0.17.x
15:44 amahon joined #salt
15:45 mohae joined #salt
15:50 alunduil_ joined #salt
15:52 forresta MTecknology, you already broke something this morning??
15:53 MTecknology forresta: no, this is a continuation from yesterday
15:53 forresta ahh
15:54 jperras slight syntax question: is there a way to define a requisite for a built-in state? I'd like to have something be run before any `git.latest` state is called.
15:55 * scoates stares at jperras
15:55 forresta jperras, you mean require?
15:55 opapo joined #salt
15:55 jperras scoates: heh
15:55 mgw is there a convention for what a module function should return if it is not available? i.e., the module itself (and presumably some of its functions) are available, but certain function(s) are not available on some platforms.
15:55 scoates forresta: short version is that we'd like to have a state applied as a prerequestite before any git.latest is applied
15:56 forresta oh so you want to define a sort of 'default' behavior?
15:56 jperras forresta: yes, but I'm unsure as to the syntax for requiring a built-in state
15:56 scoates (ideally without having to explicitly require in each git.latest)
15:57 pdayton joined #salt
15:58 forresta jperras, what about prereq? http://docs.saltstack.com/ref/states/requisites.html#prereq
15:59 teskew joined #salt
15:59 jperras forresta: how would one define that? `- prereq: - state: git.latest` ?
16:00 forresta you'd include the state at the top then - prereq: \n  - sls: state_name
16:00 amahon joined #salt
16:00 forresta that's if you wanted to include a whole whole state
16:00 forresta or does your git.latest live in the same state file?
16:00 nmistry joined #salt
16:01 scoates forresta: we don't want to apply it to a single `git.latest`, we'd like it to be applied to *all* `git.latest`s if possible
16:01 forresta jperras, this is how I do it for my git stuff, https://github.com/gravyboat/hungryadmin-sls/blob/master/salt/hungryadmin/app.sls#L36
16:01 forresta scoates, you can't do a blanket like that as far as I'm aware.
16:02 jperras gotcha
16:02 scoates forresta: ok; that's why we couldn't figure it out. (-;
16:02 jperras this is already for somewhat of a dirty hack, so that's fine :)
16:02 forresta heh
16:02 forresta yea unless there is logic I'm not aware of, requires/prereqs need to be properly defined.
16:02 MTecknology could you write your own state that's basically a wrapper for another state?
16:02 jperras hm
16:03 forresta I think you'd still need to be doing prereq/requires though
16:03 scoates next dirty-hack question: if I attempt to do a cmd.run to restart our minions, I run into a Python recursion error. Is there a safe way to restart a minion within a salt state?
16:03 scoates jperras: oh yeah, we could do that. and `use: fkgitclone` or something
16:04 forresta MTecknology, the state_auto_ordering MIGHT be usable if you did the 'prerequisite' state before hand, and then include that state...
16:04 jsm joined #salt
16:04 jperras heh
16:05 jperras context: we have a patch to salt that allows multiple ssh private keys to be specified, since we need this for checking out a repository with submodules
16:05 forresta scoates, https://github.com/saltstack/salt/issues/5721
16:05 forresta so there is a very dirty workaround for this
16:05 jperras since we don't really want to rebuild the binaries and distribute them ourselves (although we could if needed, since we run our own apt repo server)
16:05 whiskybar joined #salt
16:05 jperras we want salt to patch itself on first encountering a git.latest, then restart itself
16:06 jperras hack-central, basically
16:06 forresta you could install salt-master and salt-ssh on the master server, then use the salt service to 'stop' the minion, then use salt-ssh to start the minion, but I've never tried it. Should work though
16:06 forresta jperras, yea I've gotcha.
16:07 scoates I think I might move our patching up a level to the vagrant provision, then. (-:
16:07 forresta yea that might be a bit cleaner
16:07 jperras cleanest: integrating the patch to canonical master :p
16:10 shinylasers joined #salt
16:18 scoates FWIW: https://github.com/fictivekin/salt/commit/4ab592f29d48a75cc11fc6c4ac21d030f6e6c5dc (I haven't submitted a pull request… yet)
16:19 marcinkuzminski joined #salt
16:24 cachedout joined #salt
16:24 JasonSwindle joined #salt
16:24 Ryan_Lane joined #salt
16:24 JasonSwindle basepi:  Can we get an update on https://github.com/saltstack/salt/issues/6558 ? I just posted in it as well. :)
16:24 cachedout JasonSwindle: Are you around? I have a question for you on an old GitHub issue you filed.
16:25 cachedout Oh, that's a co-incidence. :]
16:25 JasonSwindle I am now, what's up? :)
16:25 cachedout Could I PM you?
16:25 Ahlee forresta: dude. Thanks. That so beats my for x in $(salt-run manage.up); looping over to restart salt
16:26 JasonSwindle cachedout:  Sure!
16:26 forresta Ahlee, lol
16:26 forresta I don't know if it will work
16:26 forresta haven't tried it
16:26 forresta but technically it should
16:26 Ahlee guess it depends on how salt-ssh gets its list of minions
16:26 Ahlee i know there's been work in that recently
16:26 Ahlee well, i should say
16:26 Ahlee I believe there's been work in that recently
16:27 basepi JasonSwindle: None of us here are actively working on that.  Don't know if tomeduarte is working on it or not.  Right now, our biggest in-house pushes are for tests and bugfixes -- we want hydrogen to go better than 0.17 did in terms of stability.
16:27 basepi However, I'm pushing it onto the current milestone so it will be more visible.
16:27 forresta Ahlee, well you would have to dump them into http://docs.saltstack.com/topics/ssh/roster.html
16:27 nmistry joined #salt
16:27 JasonSwindle basepi:  Okie.  It is not a breaking thing…. just bothersome to not have Postgres equal to MySQL. :)
16:28 Ahlee forresta: ah-ha! thanks
16:28 forresta np
16:28 basepi JasonSwindle: totally understand.  hopefully with this test push we can get the bugs under control a little and be able to find more feature-work time.
16:28 forresta Ahlee, when you test it can you let me know how that works?
16:29 davet joined #salt
16:30 JasonSwindle basepi:  I am really looking forward to Hydrogen.
16:30 basepi JasonSwindle: looking to be a great release.
16:31 JasonSwindle Oh yes.   Been reading the commits and wow are you guys busy. :)
16:31 Ahlee forresta: sure, no eta though
16:31 forresta yea np
16:34 mbrgm joined #salt
16:35 mbrgm hey there! does anyone know if having a reboot be part of a state run has already been implemented? i'm referencing this issue: https://groups.google.com/forum/#!msg/salt-users/yP1-852RWIc/ciwIlk2ZsfIJ
16:37 forresta mbrgm, for windows or linux?
16:38 mbrgm forresta: linux
16:38 mbrgm i want to do the exact thing as described in the post: update the kernel, reboot and continue the state execution after reboot
16:38 mbrgm i.e. when the minion is up again
16:38 forresta so we were just talking about this actually, right now there's no support for restars like that.
16:39 mbrgm forresta: are there plans for implementing this in the near to mid future?
16:40 forresta mbrgm, I don't know I brought it up a few months ago, and created this issue: https://github.com/saltstack/salt/issues/6792
16:40 dmanser joined #salt
16:40 forresta and then saw that post on the mailing list so I didn't make another one
16:40 forresta basepi, you know anything about the reboots within states, or if there's been any talk about it?
16:41 marcinkuzminski joined #salt
16:41 forresta terminalmage, awesome on 8226
16:41 dmanser i have a strange issue. when i run salt '*' pkg.list_upgrades repeatedly, i get answers from some machines or others, but the result is not always the same. so for instance, serverA responds when i first invoke it, but not on the second invocation
16:42 forresta dmanser, how many machines?
16:42 dmanser forresta: about 15 i guess
16:42 terminalmage forresta: the yum/rpm python API is the bane of my existence
16:42 forresta terminalmage, lol
16:42 forresta dmanser, hmm, should be ok with the number of file handles
16:42 basepi forresta: you mean, salt rebooting itself?
16:43 forresta basepi, no, rebooting the server in a state run, say after a kernel update or something
16:43 bt joined #salt
16:43 forresta dmanser, can you see what happens when you run the same command with -l debug?
16:44 dmanser forresta: hmm, yeah. nothing that looks strange though
16:48 bhosmer_ joined #salt
16:49 dmanser first invocation: got an answer from 6 minions. second invocation: got an answer form 9 minions. third invocation: 10
16:49 dmanser may i run into some timeout?
16:49 forresta mbrgm, you might want to open another issue on that, I don't know if since what I prefaced it as is really more a question than a feature request.
16:50 forresta dmanser, so the way the timeout works, is by default it takes 5 seconds where it waits, then it checks in with the minion, if the minion doesn't respond after 5 seconds, then it considers to have hit the timeout
16:50 forresta dmanser, you could try changing it as an experiment in your /etc/salt/master file
16:50 basepi forresta: have you checked out `system.reboot`?
16:51 basepi (it's a module, not a state, though)
16:51 MZAWeb joined #salt
16:51 matanya joined #salt
16:51 mbrgm basepi: that's the problem 'bout that. some states would require rebooting the machine, such as a kernel update etc.
16:52 forresta basepi, yea I think we talked about that one last time this was brought up. The thing most people would like I imagine is 'ok update this package, restart the server, continue the state run'
16:52 mbrgm exactly
16:52 basepi right, we currently have no way of resuming a state run after a reboot
16:52 dmanser forresta: hm, no. i set the timeout variable higher, but it doesn't come close to it. but it's always pretty much exactly 15s when i run it with time salt '*' pkg.list_upgrades
16:53 basepi which means that even if you used `module.run` with `order:last`, you probably wouldnt' get the state return (even though it ran)
16:53 forresta basepi, yep
16:53 mbrgm basepi: shall i open a feature request for this or is it in the pipeline anyway?
16:54 basepi mbrgm: open an issue.  as far as i know, it's not in the pipeline at the moment.
16:54 forresta it's almost like the minion needs to return data to the master saying it restarted, at a specific position of the job, then the master stores that as an update, and when it can communicate with the machine again, it pushes the rest of the job back.
16:54 amortimer_mbp joined #salt
16:54 forresta which I believe would totally rework how jobs are currently handled.
16:54 basepi yep, it's going to be a substantial change
16:55 mbrgm basepi: ok, will do that tomorrow (CET). have to leave now ;-)
16:55 basepi and for a time we're steering away from substantial changes to focus on general salt stability
16:55 forresta basepi, yea that's totally fair
16:55 jankowiak joined #salt
16:55 forresta since you keep breaking pkgrepo :P
16:55 basepi ;)
16:55 mbrgm great work btw... saltstack ;-)
16:55 mbrgm awesome tool
16:56 dave_den the minion has the entire highstate data, it would be possible to marshall everything to disk and have the minion check it upon restart after the reboot...
16:56 basepi mbrgm: thanks!
16:56 forresta mbrgm, when you open that can you either do an @gravyboat on there, or let me know in the IRC so I can track it?
16:56 basepi dave_den: possible.  that's probably the most reasonable method
16:56 mbrgm forresta: yep, i'll consider this
16:56 forresta thanks
16:56 dave_den still not a trivial thing, but should be possible
16:56 basepi forresta: why the "a"?
16:57 forresta ?
16:57 basepi you used to be just
16:57 basepi "forrest"
16:57 forresta because the bastard who has forrest registered logged in after being gone for 5 weeks
16:57 basepi aw man.
16:57 mbrgm bye guys
16:57 forresta so I said screw it, I'll register this, then group forrest when he doesn't log in
16:57 basepi bye
16:57 forresta I've been watching the account for almost 6 months
16:57 forresta and he always logs in at just at 10 weeks
16:58 basepi what's the threshold for inactivity?
16:58 forresta 10 weeks
16:58 basepi haha
16:58 forresta /at//
16:58 forresta yea it sucks
16:58 MTecknology basepi: fix my issue!
16:59 pipps_ joined #salt
16:59 forresta basepi, I tried to find him when he was logged in last time, but he's in some private channel somewhere, so there was no way to look it up and ask him if he'd be willing to give it up.
16:59 basepi MTecknology: lol.  which?
16:59 dmanser any other ideas? :-(
16:59 forresta dave_den, you have a minute
16:59 basepi forresta: /msg didn't work?
16:59 dave_den forresta: sure. whats up
16:59 Ryan_Lane joined #salt
17:00 MTecknology Bosse: the one I've been whining about the past two days
17:00 dave_den well, maybe. my kid just woke up from a nap, i might have to leave in a sec
17:00 basepi MTecknology: haha, I'm still catching up with weekend issues!  link?
17:00 forresta dave_den, dmanser has about 15 minions, he's trying to do salt '*' pkg.list_upgrades, and is only getting returns on some minions, any ideas for that?
17:00 forresta he's got plenty of file handles, so that shouldn't be an issue, and there's no weird debug output either
17:00 forresta basepi, if it did, he didn't respond.
17:00 basepi forresta: gotcha
17:00 dave_den forresta: tried increasing the timeout?
17:01 forresta dave_den, yes, that didn't resolve it
17:01 dmanser dave_den: yes, up to 40s
17:01 forresta basepi, yea I'll harass him when he logs in 2 months from now
17:01 dave_den 40s? that's not very long.
17:01 dmanser the command always takes about 15s to run, every time. with very little deviation
17:02 dmanser only a few milliseconds
17:02 basepi dmanser: when you use the 40s timeout, does the CLI wait the full 40 seconds?
17:02 dave_den what happens when you batch it to 2 minions at a time?
17:02 dmanser so i get answers from 6, 9, 10 or 11 minions in the same time frame
17:02 basepi or is it short-circuiting?
17:02 dmanser basepi: no
17:02 basepi dmanser: o.O
17:02 basepi dmanser: how are you targeting?
17:03 chitown joined #salt
17:03 dmanser basepi: time salt '*' pkg.list_upgrades
17:03 basepi can you give me some output with -v ?
17:03 basepi (and a timeout of 40 seconds)
17:03 dave_den salt -b 2 -t 90 '*' pkg.list_upgrades
17:04 MTecknology basepi: On salt 0.17.2 and zmq3, I'm having an issue with centos boxes only not getting pillar data, but every other server gets the pillar data. I put 'pillar_opts: True' in the master config so I should have a lot of stuff showing up. It shows up on debian based boxes, but not rhel. Since upgrading to zmq3 on that box I see /var/cache/salt/master/minions/centos5_box/mine.p populated with only
17:04 MTecknology '�', before it just didn't get created.
17:04 basepi MTecknology: is there an open issue?
17:04 MTecknology dave_den: am I missing anything useful?
17:04 MTecknology not yet- I assumed I broke something
17:04 basepi dmanser: wait, are you using batch mode?
17:05 dmanser basepi: not that i know
17:05 mannyt joined #salt
17:05 dave_den bbiab
17:05 basepi MTecknology: erm, I guess I would want to see your pillar topfile
17:05 basepi dmanser: kk.  the example dave_den used had the -b flag, which I'm pretty sure is broken right now
17:05 basepi unless someone stealth-fixed it
17:06 MTecknology basepi: even without anything from that being set, it should have a lot of stuff from that option in the master file
17:06 dmanser so batch mode is not recommended as of now?
17:06 basepi MTecknology: right.....hrm.....
17:07 dmanser hmm at least, the command from dave_den gives me the same amount of results every time so far...
17:07 basepi dmanser: https://github.com/saltstack/salt/issues/6456  i haven't actually reproduced this, but it seems problematic
17:07 basepi dmanser: i would still be interested in a run with `-v -t 40`
17:08 dmanser @basepi ok, give me a few minutes
17:08 basepi dmanser: no worries.
17:09 dave_den dmanser: is it just possible that some mininos are fully up to date?
17:10 dave_den since you got all returns back when using batching
17:10 dmanser hmm... interesting. with the -t 40 option, the command waits 40 seconds. but not if it is set in the master config
17:10 tyler-baker joined #salt
17:10 tyler-baker joined #salt
17:11 basepi dmanser: oh?  possible that master config option is broken.  what is the exact config option you're using?
17:11 basepi also, you have to restart the master after changing that opt to have it take effect
17:12 dave_den MTecknology: i would try setting up a development environment with your rhel minions and a copy of your master so you can troubleshoot easier.
17:12 dmanser basepi: i did restart the master. the option is called timeout in the /etc/salt/master config file
17:12 dave_den there should be something in the master or minion logs
17:12 basepi hrm, yes, `timeout: 40` should do it.  i'll have to dive this code later and see what's going on.
17:13 UtahDave joined #salt
17:13 basepi dmanser: did they all return with -t 40?
17:14 dmanser basepi: http://pastie.org/8525947
17:15 dmanser so it looks okay with -t 40, but not if the option is set in the master config. let me double-check that and that the master really is restarted...
17:16 basepi dmanser: yes, it may just be a bug with the timeout option
17:16 dmanser basepi: yeah, it stopped after 18 seconds
17:16 dmanser that's an interesting bug
17:18 forresta basepi, maybe related to this from a month ago? https://github.com/saltstack/salt/issues/6881
17:18 forresta basepi, oh nevermind https://github.com/saltstack/salt/issues/8447
17:18 forresta even better :P
17:18 basepi boom.
17:18 basepi there we go
17:18 basepi you're not the only one.  thanks for finding that forresta, totally forgot about it
17:19 dmanser uh
17:19 forresta basepi, yea np.
17:19 forresta I wonderif the job IDs themselves are actually finishing even when the 'timeout' is exceeded.
17:20 KyleG joined #salt
17:20 KyleG joined #salt
17:20 dmanser okay, i'm using salt 0.17.2
17:20 basepi forresta: likely.
17:20 dmanser thank you basepi and forresta
17:20 forresta yea np, thanks for hopping in dave_den
17:21 basepi =D
17:21 dmanser yeah, sorry. thanks to dave_den as well :)
17:21 bhosmer joined #salt
17:21 dave_den cool :)
17:22 forresta basepi, is the plan to do a 0.17.3 release, then from there onto hydrogen?
17:22 basepi yes, 0.17.3 in the next few days, and hopefully the first hydrogen RC in the next week.
17:22 dmanser so what's next on my salt-todo-list: ah, develop a cisco module :-)
17:23 basepi we want to do a full 3-4 release candidates for hydrogen
17:23 basepi and expect the official, non-RC release to be early January.
17:23 forresta basepi ahh ok that makes sense, I was gonna say, next week?? Is no one gonna work on client stuff!
17:23 SEJeff_work Uhoh, salt has a new competitor: http://fuckingshellscripts.org
17:23 SEJeff_work :)
17:24 forresta SEJeff_work, well, I read step 1
17:24 forresta err step 0
17:24 forresta and almost closed the page
17:24 basepi hahaha!
17:24 forresta 'shell scripts! OK NOW INSTALL THIS GEM!'
17:24 basepi ew
17:24 SEJeff_work Scroll to the very bottom for lulz
17:24 forresta heh
17:24 SEJeff_work Step 5: Remove your chef repo and all its contents.
17:24 SEJeff_work rm -rf ~/old_config_management/chef
17:24 SEJeff_work HOLY SHIT THAT WAS EASY.
17:24 dmanser haha :-)
17:25 forresta I like step 3
17:25 forresta want to install ruby2? Here's an example that only takes 13 lines of code and has no comments!
17:25 mattmtl joined #salt
17:25 SEJeff_work or error checking!
17:25 SEJeff_work That's for sissies and real sysadmins
17:25 forresta good point
17:25 SEJeff_work not devops rubyistas
17:27 ncjohnsto joined #salt
17:30 scott_w_ joined #salt
17:32 irctc860 joined #salt
17:33 troyready joined #salt
17:35 mike251 left #salt
17:36 ipmb joined #salt
17:38 higgs001 joined #salt
17:39 teskew joined #salt
17:39 UtahDave nice, SEJeff_work
17:40 rojem joined #salt
17:41 bemehow joined #salt
17:41 Sheco joined #salt
17:43 jdenning joined #salt
17:47 redondos joined #salt
17:57 druonysus joined #salt
17:59 supplicant I have a hostname being used in a state, is it possible to look up the associated ip address in the state to use instead?
18:01 UtahDave supplicant: you probably could with jinja. Have you tried that yet?
18:01 supplicant UtahDave: I'm not sure which module to use, if any. otherwise, I'll just cmd.run getent and use the result
18:02 supplicant UtahDave: I guess my question was more about which module I should use to get this information
18:02 supplicant I could always roll my own, too, I guess
18:05 JasonSwindle joined #salt
18:09 dmanser left #salt
18:12 jacksontj joined #salt
18:19 amckinley joined #salt
18:22 druonysuse joined #salt
18:25 logix812 joined #salt
18:25 rojem joined #salt
18:26 londo_ joined #salt
18:26 bhosmer joined #salt
18:28 cyrusdavid joined #salt
18:29 higgs001 joined #salt
18:31 ajw0100 joined #salt
18:31 druonysuse joined #salt
18:31 druonysuse joined #salt
18:32 dpippenger joined #salt
18:33 sandGorgon joined #salt
18:33 jacksontj joined #salt
18:33 nmistry joined #salt
18:33 [diecast] joined #salt
18:34 pdayton joined #salt
18:41 bhosmer joined #salt
18:42 fllr joined #salt
18:43 bhosmer_ joined #salt
18:50 jslatts joined #salt
18:54 JesseCW joined #salt
18:54 sandGorgon joined #salt
18:55 ccase joined #salt
19:01 millerdev joined #salt
19:02 millerdev I recently submitted a pull request to the salt-bootstrap repository. It fixes some issues on FreeBSD. https://github.com/saltstack/salt-bootstrap/pull/268
19:02 millerdev who do I need to talk to to get this reviewed (and hopefully merged)
19:02 millerdev ?
19:03 mpanetta joined #salt
19:03 JulianGindi joined #salt
19:06 robawt g'morning cool kids
19:06 KyleG good morning.
19:07 KyleG millerdev: I believe cedwards is our FreeBSD maintainer.
19:07 Brew joined #salt
19:08 davet joined #salt
19:10 cedwards looking..
19:10 millerdev thank you!
19:13 mpanetta joined #salt
19:14 ajw0100 joined #salt
19:23 Tropolis joined #salt
19:25 millerdev left #salt
19:26 millerdev joined #salt
19:29 cdcalef joined #salt
19:33 kermit joined #salt
19:44 higgs001 joined #salt
19:47 zooz joined #salt
19:50 zandy joined #salt
19:53 abe_music joined #salt
19:55 KyleG Does anybody know why 'bd0.br* and not bd0.br0.*': does not match bd0.br1.domain.com in my top.sls? I thought I had it right…
19:56 millerdev left #salt
19:56 anitak joined #salt
19:58 mgw KyleG: bd0.br0.* does not match bd0.br1.domain.com…. is that what you're saying?
19:58 KyleG Ahh I think I forgot to add   - match: compound
19:58 KyleG 1 sec let me try compound
19:59 mgw maybe it was a typo, but br0 will not match br1
19:59 KyleG right I don't want it to match that
19:59 KyleG It needs to match bd0.br*
19:59 lynxman joined #salt
19:59 mgw sorry, misunderstand
19:59 mgw s/stand/stood/
20:00 ajw0100 joined #salt
20:00 KyleG woot! It was the compound matcher. :)
20:01 MTecknology Every single rhel based box, some centos, some mlos... all the same issue
20:01 MTecknology I think it's time to go to github
20:03 ajw0100 joined #salt
20:03 KyleG lol damnit
20:03 KyleG Salt outputs the binary for /usr/local/lib/php/20100525/newrelic.so
20:03 KyleG When I highstate
20:03 KyleG That's pretty annoying
20:04 bt joined #salt
20:04 KyleG https://gist.github.com/kylegato/16fba2a27a7da8f0dbe1
20:04 KyleG Is that a bug in salt somehow? Not realizing it's a binary
20:05 erchn joined #salt
20:05 cnelsonsic joined #salt
20:08 alunduil joined #salt
20:09 forresta KyleG, what does the state look like?
20:09 cjh973 joined #salt
20:11 forresta KyleG, because issues with non-ascii characters was fixed: https://github.com/saltstack/salt/issues/7760
20:12 forresta err were fixed, can't grammar today apparently
20:15 scott_w_ joined #salt
20:19 MTecknology forresta: I finally created the issue on github. I'm hoping my thinking can no be over and a magic solution will soon be fount.
20:19 forresta lol
20:19 forresta just wave your hands
20:20 forresta problem solved
20:20 forresta magic
20:25 MTecknology can you imagine no new features until all bugs fixed and tests written for everything followed by no new commits without a test...
20:26 MTecknology time to run off.. back later
20:27 jslatts joined #salt
20:29 lyska joined #salt
20:30 bhosmer joined #salt
20:38 Ahlee salt on windows.  I know it's possible, but anybody got deployment recommendations? I've got a plethora of 2008, 2008R2, 2012, and even some 2k3 systems still kicking it around
20:39 forresta Ahlee, he 64bit installer has been tested on Windows 7 64bit and Windows Server 2008R2 64bit. The 32bit installer has been tested on Windows 2003 Server 32bit.
20:39 forresta that's from the docs
20:39 forresta so maybe?
20:40 Ahlee hrm
20:40 Ahlee i'm very hesitant to roll out until the self-update issue is resolved
20:41 Ahlee as revisiting all linux boxes is easy for me, i don't want to write powershell to rip this out and re do it.
20:41 forresta yea I'm clueless when it comes to the windows release, I only get to mess with even the Linux verison at home :P
20:46 Ahlee guess we add that to next sprint, as well.
20:50 zandy joined #salt
20:52 zandy joined #salt
20:53 dyim joined #salt
20:57 teepark joined #salt
21:01 pipps joined #salt
21:02 Gifflen_ joined #salt
21:03 cachedout joined #salt
21:04 che-arne joined #salt
21:05 cjh973 joined #salt
21:05 bhosmer joined #salt
21:06 jslatts joined #salt
21:12 pipps_ joined #salt
21:12 cewood joined #salt
21:18 nmistry joined #salt
21:24 smccarthy joined #salt
21:25 druonysuse joined #salt
21:25 druonysuse joined #salt
21:33 kermit joined #salt
21:42 harobed_ joined #salt
21:50 Ryan_Lane joined #salt
21:50 zandy joined #salt
22:02 matanya joined #salt
22:03 Ryan_Lane joined #salt
22:06 che-arne_ joined #salt
22:06 matanya joined #salt
22:07 cdcalef tests are for programmers who get paid by the hour
22:07 forresta ?
22:08 JesseC joined #salt
22:08 cdcalef er nothing
22:08 matanya joined #salt
22:09 matanya joined #salt
22:11 JesseCW joined #salt
22:11 zerowolfgang joined #salt
22:11 JesseCW left #salt
22:14 zerowolfgang left #salt
22:15 amckinley joined #salt
22:16 redondos joined #salt
22:16 JesseC joined #salt
22:17 z3uS joined #salt
22:20 scott_w_ joined #salt
22:21 nmistry joined #salt
22:23 forresta off topic, but has anyone set up foreman-proxy on a rhel box?
22:23 pipps joined #salt
22:25 cachedout joined #salt
22:28 thaddeusmt joined #salt
22:30 amckinley joined #salt
22:32 thaddeusmt hello everyone, quick question: is there a way to ensure state execution after a jinja loop? so that, for instance, I can restart a process after looping over a few configuration files?
22:36 bhosmer joined #salt
22:37 forresta thaddeusmt, is your jinja loop looping through state info?
22:38 forresta most of the time if you do that, you want to simply add a watch_in
22:38 forresta http://docs.saltstack.com/ref/states/requisites.html#watch-in
22:41 thaddeusmt forresta: I tried that... let me try again and remember why it didn't work (and hopefully it will just work this time!)
22:41 forresta ok cool
22:44 Ahlee anybody got a cent 5.x server with epel installed and able to do a yum provides "*/libyaml" for me?
22:44 forresta no cent 5 machines Ahlee :(
22:45 Ahlee no problem, just trying to avoid building a new system and the network changes to get it unfettered access to epel
22:46 Ahlee duh. It's probablyin http://dl.fedoraproject.org/pub/epel/5/x86_64/repoview/libyaml.html
22:46 Gareth Are there current modules/states where there are different variables, options used depending on the operating system/distribution?
22:46 Ahlee i searched for yaml on the P page, heh.
22:47 forresta Gareth, yea we use that all the time, check out https://github.com/terminalmage/djangocon2013-sls
22:47 bhosmer joined #salt
22:47 forresta Gareth, that uses a more advanced idea of jinja map files
22:47 terminalmage Ahlee:  hold up
22:47 austin987 joined #salt
22:47 terminalmage I have a cent5 box on rackspace
22:47 Ahlee np, i got it
22:48 terminalmage ok
22:48 forresta Gareth, check out http://docs.saltstack.com/topics/pillar/#declaring-the-master-pillar, about 4 examples down
22:48 forresta that is used.
22:48 Ahlee it's provided by libyaml-0.1.2-3.el5.x86_64.rpm
22:48 terminalmage ahh
22:48 terminalmage the more you know
22:48 Ahlee yeah.  who'd have known libyaml wouldn't show up on a page listing just packages that started with a p?
22:50 zandy joined #salt
22:51 Gareth forresta: background....working on networking modules.  the way debian handles bonded interfaces is totally different than Red Hat.  the way of specifying which devices are slave devices isn't going to work, so I'm thinking of adding another argument that is only used if its a debian system to handle the slaves.
22:52 forresta ahh gotcha
22:52 Ryan_Lane joined #salt
22:52 forresta so you mean actual python stuff, not yaml, sorry
22:54 thaddeusmt forresta:thanks for the tip, it is executing in the proper order - I think I must have been tangled up with "recursive dependency" errors before
22:54 forresta thaddeusmt, ahh gotcha, yea that is always a bit tricky when you get into the loop
22:54 forresta *s
22:55 bhosmer_ joined #salt
22:56 rojem joined #salt
22:56 thaddeusmt executing in the correct order is not solving my problem though, which I hoped it would :/ in the loop, after I configure the Mongod instances defined in my pillar, I am initializing a replica set. but it's failing, since even though Salt thinks the mongo instances have been started, they apparently take a while to be fully "up" and ready for replication
22:56 thaddeusmt is there a way, in Salt, to loop until a condition is met?
22:56 forresta what sort of condition?
22:57 forresta I mean you can use some python style loop logic within jinja
22:57 thaddeusmt some kind of a "wait until x is true" condition? where it keeps checking a script (like and 'unless' statement) until the statement is true (or a timeout is hit)?
22:57 forresta you could use a while loop I guess
22:58 thaddeusmt I want to wait until the mongo instances return a "ready" status, and then fire of the replica set init script... :/
22:58 forresta ahh
22:58 pdayton1 joined #salt
22:58 forresta for that it might be easier to write a script, then envoke the script with salt maybe?
22:58 forresta where the script doesn't 'finish' running until that returns true?
22:59 thaddeusmt hmmm, yeah that might be the best way to do it - have Salt fire off a daemonized shell script, what waits until the mongos are up, then runs the init command
22:59 thaddeusmt *that waits
23:00 forresta you could also look at using the pydsl http://docs.saltstack.com/ref/renderers/all/salt.renderers.pydsl.html
23:00 forresta depending on what language you feel more comfortable in, but that would let you code in pure python, so you could easily just check the status with a subprocess call or something
23:01 forresta it's more work depending on the complexity of your state however.
23:01 Gareth forresta: yeah.
23:02 thaddeusmt forresta: I'll check out pydsl also, thanks for your help!
23:02 forresta yea np
23:03 forresta Gareth, looks like this uses it for windows
23:03 forresta https://github.com/saltstack/salt/blob/develop/salt/states/pkg.py
23:03 matanya joined #salt
23:04 Gareth forresta: cool.  will check that out
23:04 forresta ok
23:07 scott_w_ joined #salt
23:08 diegows joined #salt
23:13 serge joined #salt
23:13 sergelesch left #salt
23:14 sergelesch joined #salt
23:15 Gifflen joined #salt
23:16 faldridge joined #salt
23:17 jcockhren thaddeusmt: pydsl works
23:18 jcockhren thaddeusmt: there's an example formula in github
23:18 jcockhren backuptocloud-formula
23:19 sergelesch hello. I have a small problem with "cp.push" - the checksum on minion differs from checksum on salt master. is it normal behaviour?
23:20 forresta sergelesch, did you diff the files?
23:20 forresta jcockhren, that's a pretty confusing example to start soeone on jcockhren :P
23:20 forresta *someone
23:20 sergelesch I can, but checksum (md5) seems to be sufficient for me
23:23 forresta sergelesch, ok, would just be good to confirm the content/filesize is different, could be as simple as selinux settings if you're on rhel
23:23 jcockhren forresta: that formula is a dsl for those willing to use pydsl
23:23 forresta I know jcockhren, I was just joking
23:23 jcockhren I was making fun oy myself
23:23 jcockhren of*
23:27 sergelesch forresta, I copy binaries... I'm going to do the same with plain text file
23:27 forresta ok
23:27 sergelesch give me a minute - I'll show you the result
23:28 AlcariTheMad joined #salt
23:30 mpanetta joined #salt
23:31 sergelesch forresta: after cp.push the files seems to be either encrypted or corrupted. "/var/cache/salt/master/minions/xxxx/files/tmp/textfile" may be a binary file.  See it anyway?
23:32 forresta weird
23:32 lemoi joined #salt
23:32 sergelesch forresta: is it specific to my environment ?
23:32 lemoi http://dpaste.com/1491895/
23:33 forresta I don't know
23:33 lemoi this is 0.17.3 runninng on gentoo stable python2.7.5
23:33 lemoi :s
23:33 sergelesch my I ask someone to reproduce it ?
23:33 kermit joined #salt
23:34 lemoi err I mean 0.17.2
23:37 alunduil joined #salt
23:38 oz_akan_ joined #salt
23:42 mikef joined #salt
23:46 sergelesch cp.push problem - https://github.com/terminalmage/salt/commit/da3e157e0c3bbf1740dcd60a71923439fb14944e
23:47 mikef im trying out salt and i wanted to get it to start the mongod service, but it does not, here is the config and the error: http://dpaste.com/1491901/
23:48 mikef is there some other name for the mongod service that i should be aware of? i tried mongodb mongod.service unsuccessfully.
23:48 scott_w_ joined #salt
23:50 bhosmer_ joined #salt
23:50 zandy joined #salt
23:50 terminalmage mikef: the list of services available to be managed can be found by running service.get_all
23:51 terminalmage salt minion-id service.get_all
23:52 Ryan_Lane joined #salt
23:53 mikef terminalmage: thanks, seems like mongod is not listed,  i wonder what could cause that,  on the minion i can use os   service commands and the service exists (im on fedora)
23:53 mikef release 19
23:54 terminalmage mikef: is the mongodb server installed on the minion yet?
23:54 mikef yes
23:54 __number5__ mikef: is mongod using system.d?
23:54 mikef __number5__  : um not sure
23:54 nocturn joined #salt
23:55 mikef __number5__  :  how do i check that
23:55 scott_w_ joined #salt
23:55 terminalmage mikef: I think he meant systemd
23:55 terminalmage Fedora 19 should be using systemd
23:56 __number5__ yep, sorry about the dot :P
23:57 mikef terminalmage:  yes your right by default it should
23:57 terminalmage mikef: are you overriding the provider?
23:58 terminalmage if not, can you run from the CLI on the F19 box "stat /run/systemd/system"
23:58 mikef terminalmage: nope, im just getting started,  i dont know how to do that,   using systemd i can start and stop the service, but salt can not even tell it exists ,
23:58 terminalmage that's how salt detects if systemd is running
23:59 mikef http://dpaste.com/1491907/

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