Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2014-03-14

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

All times shown according to UTC.

Time Nick Message
00:00 eculver_ joined #salt
00:00 hunter____ joined #salt
00:05 KyleG joined #salt
00:05 seapasulli left #salt
00:06 Outlander joined #salt
00:09 abe_music joined #salt
00:09 meteorfox joined #salt
00:11 saltyk joined #salt
00:11 r0rySavag3 joined #salt
00:13 saltyk Hey there, I'm trying to get an environment variable set for Django, specifically DJANGO_SETTINGS_MODULE. I've tried running an export command, but when my other sls runs that instlls my requirements file through pip, it says the variable is not set
00:14 saltyk Both are using the root user, and I made sure to echo to see that it was running, but no luck
00:14 Teknix joined #salt
00:14 saltyk Is there something I'm missing?
00:15 tlyng_ joined #salt
00:17 NV run the command prefixed with the set
00:17 NV ie
00:17 NV BLAH=asdf pip ...
00:18 r2dk1ng2 joined #salt
00:18 saltyk so I can do DJANGO_SETTINGS_MODULE=cardisle_kiosk.settings pip install -r /my/requirements.txt
00:19 googolhash joined #salt
00:19 saltyk Is there a reason why env do not stay set though?
00:20 ajw0100 joined #salt
00:23 Outlander I have 2 debian boxes using the same salt repo, the 7.3 box says salt-minion 0.17.5 is latest, the debian 7.4 box says salt-minion 2014.1.0 is the latest
00:24 jpcw joined #salt
00:24 meteorfox left #salt
00:25 meteorfox joined #salt
00:25 utz left #salt
00:25 meteorfox left #salt
00:25 dstanek_afk joined #salt
00:26 lionel joined #salt
00:27 xt joined #salt
00:27 jacksontj_ joined #salt
00:28 scalability-junk joined #salt
00:28 Outlander how/why are these nodes seeing different packages as latest in the same exact repo?
00:29 eculver joined #salt
00:29 eculver joined #salt
00:30 torrancew Outlander: apt-get update not up-to-date, by chance?
00:31 zain joined #salt
00:31 Outlander Debian 7.3 Node: 3.2.0-4-amd64 #1 SMP Debian 3.2.46-1 x86_64 GNU/Linux ,Debian 7.4 Node: 3.2.0-4-amd64 #1 SMP Debian 3.2.54-2 x86_64 GNU/Linux
00:31 Outlander 7.3 node is AMD, 7.4 node is Intel
00:31 Outlander and yes apt-get update has been run on both
00:31 torrancew Outlander: no, what I mean is, have you run apt-get update on both nodes recently?
00:31 torrancew ok
00:31 torrancew try apt-cache policy on each box
00:32 torrancew see what other options it's seeing, which may give you a lead
00:32 Outlander also cleared entire apt cache on both and re-ran apt-get update
00:32 Outlander cool will try that
00:32 torrancew apt-cache policy $pkgname
00:33 shoemonkey joined #salt
00:33 andrej If I have the result-set of a client.LocalClient call with 3 functions ... how can I "flatten" it (for lack of a better word) to be able to write a csv file where only the minion mame and the values of the embedded dict (omitting their keys) come out?
00:33 sporkd2 Outlander: I had the exact same problem, I ended up using the bootstrap script so I could specify a tag v2014.1.0
00:35 Outlander gross
00:36 torrancew Outlander: any clues?
00:36 Outlander yeah
00:37 shoemonkey joined #salt
00:38 Outlander http://pastebin.com/K4rFepvT
00:38 Outlander different policies
00:39 Outlander the one getting the 2014 package versions is only using remote repo, the 7.3 box is using local cache of some sort first
00:39 Outlander so any chance these packages/versions were pinned somehow?
00:39 Outlander I didn't build the 7.3 box, so unsure what was done to it
00:39 torrancew well, even when pinned, you see the other candidates
00:39 Outlander yeah so that's odd right
00:39 torrancew any chance you have a bad hosts file entry or something equally silly?
00:40 torrancew basically, wondering if the two hosts are /actually/ talking to the same repo
00:40 andr386 joined #salt
00:40 Outlander ah right
00:40 Outlander I'll check that now
00:40 Outlander hosts file looks ok on both
00:40 Outlander but in different data centres, so potentially different dns settings
00:41 torrancew yeah, try pinging the repo host, as well as studying apt-get update output closely
00:41 torrancew perhaps one is 404'ing on that repo
00:41 torrancew and you're stuck with old data re: that repo, as a result
00:41 Outlander resolves to same on both boxes
00:41 Outlander Non-authoritative answer:
00:41 Outlander Name:debian.saltstack.com
00:41 Outlander Address: 198.211.96.193
00:41 dave_den joined #salt
00:41 torrancew ok, so can both actually talk to it?
00:42 Outlander ok so i can clear the apt cache fully and run apt-get update again
00:42 torrancew or is one perhaps being firewalled
00:42 torrancew yeah, try that
00:42 Outlander yeah will test firewalls next
00:42 torrancew but yeah, it sounds like for whatever reason, the 7.3 host isn't actually getting updates from the repo
00:43 ajw0100 joined #salt
00:43 torrancew hell, maybe even a poorly configured http caching proxy or something
00:45 schimmy1 joined #salt
00:47 Katafalk_ joined #salt
00:47 Outlander ok so after moving/clearing apt cache files and re-running apt-get update, a new candidate version appears: http://pastebin.com/5SmBs41x
00:47 funzo_ joined #salt
00:47 shoemonk_ joined #salt
00:48 vbabiy joined #salt
00:49 r0rySavag3 joined #salt
00:50 berto- joined #salt
00:51 Kyle- joined #salt
00:51 KyleG joined #salt
00:51 KyleG joined #salt
00:51 markizano joined #salt
00:51 eculver joined #salt
00:51 eculver joined #salt
00:51 nliadm joined #salt
00:54 madduck joined #salt
00:55 gparent_ joined #salt
00:56 puppet joined #salt
00:56 berto- joined #salt
00:57 oz_akan_ joined #salt
00:58 manicouman joined #salt
00:59 akoumjian joined #salt
01:00 bhosmer joined #salt
01:01 oz_akan__ joined #salt
01:02 berto- joined #salt
01:03 alunduil joined #salt
01:07 ekristen is pillar data stored on the minion? is it encrypted? or sitting plain text?
01:08 ekristen past reboot? or is it re-downloaded from the master when a state is run?
01:08 fxhp ekristen - it lives on the master
01:08 nliadm joined #salt
01:08 ekristen k
01:08 fxhp the master serves pillar data as needed to the minion
01:08 fxhp over the encrypted zermq bus
01:08 KyleG joined #salt
01:08 KyleG joined #salt
01:09 fxhp That said, a user with access to a minion, can query the pillar
01:09 jesusaurus ekristen: pillar data is unencrypted 'at rest'
01:09 ekristen on the master
01:09 ekristen yes I know
01:09 fxhp if that user has access to the user that runs the salt-minion daemon
01:10 jesusaurus ekristen: in the minions cache as well as on the master
01:10 ekristen jesusaurus: so the minions will cache the pillar data locally?
01:10 smcquay_ joined #salt
01:10 fxhp jesusaurus - in the default implementation it is not encrypted at rest, but there is a external pillar API
01:11 jesusaurus ekristen: yes, by default the minions will cache pillar data they are given. but the cache will automatically refresh during a highstate
01:11 ekristen does it go away? do you know where it is stored?
01:12 fxhp jesusaurus - are you sure it is cached?
01:13 fxhp I'm pretty sure it is not cached
01:14 arodizzy_ joined #salt
01:14 andrej joined #salt
01:14 andrej joined #salt
01:14 andrej what am I doing wrong here?  http://pastebin.com/FkTgBcj2
01:14 fxhp Oh no I think it is cached
01:14 andrej If I have test.fib & 1 cmd.run I get both outputs.
01:14 andrej If I use cmd.run twice only the 2nd command returns output
01:15 jesusaurus fxhp: im pretty sure its cached in /var/cache/salt/minion
01:15 * fxhp nods
01:15 fxhp I don't see it
01:16 jesusaurus andrej: each cmd.run needs its own unique id
01:17 berto- joined #salt
01:17 fxhp andrej - it seems like you are clobbering the previous result (each needs a unique ID)
01:17 nliadm joined #salt
01:19 Outlander ok I think I'm all updated :)
01:19 snuffeluffegus joined #salt
01:19 fxhp ekristen: sudo salt-call pillar.data from a minion
01:20 ndrei joined #salt
01:20 Outlander but I think I found a bug in /usr/lib/python2.7/dist-packages/salt/states/debconfmod.py
01:20 Outlander so will go and see if there are any existing bugs in github issues
01:20 Outlander that match this
01:20 Outlander and will also try on another box
01:20 fxhp ekristen: the minion calls home to the master and asks for the pillar data that is targeted to itself, and is then presented with the data
01:20 ekristen fxhp: but it caches it :/
01:20 fxhp ekristen where though?
01:21 fxhp Seems to fetch it each time, takes like 2 secs to compile the data on the master and send it back
01:21 fxhp http://docs.saltstack.com/ref/modules/all/salt.modules.saltutil.html#salt.modules.saltutil.refresh_pillar
01:21 n8n_ joined #salt
01:21 fxhp ^ that makes me believe it might be cached, but I don't see it
01:21 ekristen I thought you sait oh no I think it is cached
01:21 ekristen … /var/cache/salt/minion maybe?
01:22 fxhp That looks the the state tree (file_roots)
01:22 Networkn3rd joined #salt
01:22 xzarth_ joined #salt
01:22 ekristen maybe > highstate.p
01:23 ekristen oh nm
01:23 ekristen fxhp: I’m hoping it isn’t cached ;)
01:24 bretep` joined #salt
01:26 fxhp ekristen jesusaurus - it is cached
01:26 fxhp we were both right
01:26 balboah joined #salt
01:26 ekristen oh?
01:27 fxhp its cached in /var/cache/salt/minion/proc
01:27 fxhp for the duration of the remote execution
01:27 fxhp I determined this
01:27 ekristen and then is deleted?
01:27 redondos joined #salt
01:27 redondos joined #salt
01:27 fxhp by running: watch -n.1 'ls -hal /var/cache/salt/minion/proc'
01:28 fxhp and then running: sudo salt-call pillar.data
01:28 fxhp in another window
01:28 fxhp and for a blip there exists a file
01:28 fxhp so it cleans up the data
01:29 fxhp nope wrong again
01:29 fxhp fun pillar.data jid 20140313212936306759 pid c
01:29 fxhp it just puts that in proc
01:29 fxhp hmm
01:30 andrej Aight ... so how do I make cmd.run unique if I ultimately need to run 4 of them?
01:30 torrancew joined #salt
01:30 * andrej waits for the drumroll
01:30 lahwran joined #salt
01:31 andrej joined #salt
01:31 codekoala joined #salt
01:31 codekoala joined #salt
01:31 markizano joined #salt
01:31 nliadm joined #salt
01:31 andrej joined #salt
01:31 berto- joined #salt
01:31 Saint-N joined #salt
01:32 fxhp andrej - where is that running
01:32 andrej on the master
01:33 fxhp ekristen: I'm pretty confident pillar data is not cached and if fetched from master as needed
01:33 fxhp http://docs.saltstack.com/topics/tutorials/pillar.html
01:34 fxhp additionally different data can have different "ACLs" by targeting particular data to different minions
01:34 andrej I need to gather some auditing info from all minions
01:34 fxhp Pillar is where we put our SSH keys and certs
01:34 andrej so want to run 4 cmd.runs
01:34 torrancew joined #salt
01:34 andrej and was hoping to be able to avoid 4 separate invocations
01:35 fxhp is that a bash file?
01:35 fxhp I woudl run separate invocations
01:36 andrej heh
01:36 andrej ok
01:37 andrej seems kind of counter-intuitive seeing as numerous invocations are possible
01:37 andrej Just not of the same function
01:38 andrej 200 calls instead of 50
01:38 andrej ah well
01:38 fxhp hmm
01:38 andrej will have to re-thing the CSV generation, too
01:38 fxhp how did you import local?
01:39 andrej import salt.client
01:39 andrej local = salt.client.LocalClient()
01:39 Saint-N joined #salt
01:39 codekoala joined #salt
01:39 codekoala joined #salt
01:40 andrej It would work fine if it didn't return a dict but a list, like what it's fed
01:40 nliadm joined #salt
01:40 fxhp andrej: its the difference between the master sending 200 singular jobs, or 50 jobs with 4 commands in the payload, the work on the masters side is negligible
01:41 fxhp and each minion would subscribe to 4 "jobs" instead of one
01:41 fxhp the master publishing 200 jobs isn't a hard task
01:42 puppet joined #salt
01:43 markizano joined #salt
01:44 Nazca__ joined #salt
01:44 andrej alright
01:44 andrej I'll change my approach
01:44 andrej still *feels* weird ;}
01:44 kermit joined #salt
01:44 fxhp andrej: agreed
01:45 fxhp This is neat though, I'm going to mess with your example a bit
01:45 toastedpenguin joined #salt
01:46 andrej :)
01:49 andrej Ok, so w/ the API - how do I get a list of minions to iterate over?
01:50 fxhp hmm,
01:50 manicouman joined #salt
01:50 fxhp I was looking at "manage.up" runner code a few days ago that had a method
01:50 fxhp not sure if that is the best way
01:50 fxhp "manage.status()"
01:50 andrej If I have 4 invocations per minion i wil need to iterate to make the output in CSV useful in some way :}
01:51 HeadAIX joined #salt
01:52 fxhp andrej: https://github.com/saltstack/salt/blob/develop/salt/runners/manage.py#L26
01:53 fxhp not sure if you can import that
01:53 colinbits joined #salt
01:53 fxhp https://github.com/saltstack/salt/blob/develop/salt/runners/manage.py#L119
01:54 yomilk joined #salt
01:54 andrej thanks fxhp - I was lazy :)  for i in local.cmd( '*', ['test.ping'],[]):
01:54 fxhp thats one way
01:54 fxhp : )
01:55 fxhp just steal the codes
01:55 andrej heh
01:55 fxhp andrej: that will only give you a list of "listening" minions
01:56 juicer2 joined #salt
01:56 andrej fxhp - I know ... but running commands against minions that are down is pretty silly, anyway
01:56 fxhp to get the minions the master knows about, you need to use the key api
01:58 andrej Ummm - where do I find that? :)
01:58 andrej It would be nice to hear which minions DIDN'T answer.
01:59 manicouman joined #salt
01:59 ajw0100 joined #salt
02:00 anitak joined #salt
02:05 favadi joined #salt
02:06 kisa joined #salt
02:07 andrej joined #salt
02:07 berto- joined #salt
02:08 Saint-N joined #salt
02:08 nliadm joined #salt
02:09 puppet joined #salt
02:09 alekibango joined #salt
02:10 JordanRinke joined #salt
02:11 smcquay_ joined #salt
02:12 malinoff joined #salt
02:16 andrej joined #salt
02:16 andrej Sweet lord
02:16 andrej IRC is being iffy today
02:17 possibilities joined #salt
02:21 JordanRinke joined #salt
02:22 cpenner461 joined #salt
02:23 rostam joined #salt
02:30 mgw joined #salt
02:33 andrej I'm sure I asked this before ... if I want custom defined grains, e.g. purpose: and owner: which describe the reason for a minions existence, and who the business owner is
02:34 andrej How do I go about that?
02:35 andrej Manually create a grains file on each minion? Or can I do this centrally on the master?
02:35 abe_music joined #salt
02:37 malinoff andrej, http://docs.saltstack.com/ref/modules/all/salt.modules.grains.html?highlight=grains#module-salt.modules.grains
02:41 andrej malinoff thanks; where does a grain I "append" end up? :)
02:42 andrej in /etc/salt/grains on the minion ... to answer my own question
02:45 andrej thanks :)
02:47 ilbot3 joined #salt
02:47 Topic for #salt is now Welcome to #salt | Thanks to everyone who made SaltConf 2014 amazing! | 2014.1.0 is the latest | Please be patient when asking questions as we are volunteers and may not have immediate answers | Channel logs are available at http://irclog.perlgeek.de/salt/
02:47 vejdmn joined #salt
02:49 forrest joined #salt
02:50 londo_ joined #salt
02:51 possibil_ joined #salt
03:00 quickdry21 joined #salt
03:00 malinoff andrej, np :)
03:01 kisa joined #salt
03:02 googolhash joined #salt
03:03 frasergraham joined #salt
03:05 jeremyfelt joined #salt
03:08 anitak joined #salt
03:08 balboah joined #salt
03:09 n8n joined #salt
03:14 fllr joined #salt
03:15 n8n_ joined #salt
03:16 googolhash_j joined #salt
03:19 googolhash joined #salt
03:31 FunkyBob joined #salt
03:33 fllr joined #salt
03:34 tmandry Is it possible to use require_in for an sls file?
03:43 Networkn3rd joined #salt
03:44 mgw joined #salt
03:47 n8n joined #salt
03:54 HeadAIX joined #salt
04:00 ravibhure joined #salt
04:03 nextdoorwarren joined #salt
04:04 fllr joined #salt
04:29 naiyte joined #salt
04:49 balboah joined #salt
05:11 frasergraham joined #salt
05:29 faldridge joined #salt
05:38 l0x3py joined #salt
05:44 googolhash joined #salt
05:48 Nexpro1 joined #salt
05:56 faldridge joined #salt
05:59 ipalreadytaken joined #salt
06:00 andr386 joined #salt
06:17 faldridge joined #salt
06:24 bhosmer joined #salt
06:29 balboah joined #salt
06:31 yomilk joined #salt
06:33 fllr joined #salt
06:35 _fllr_ joined #salt
06:45 ndrei joined #salt
06:46 dccc joined #salt
06:49 Katafalkas joined #salt
06:51 Furao joined #salt
07:08 ndrei joined #salt
07:16 ajw0100 joined #salt
07:35 ml_1 joined #salt
07:37 Iwirada joined #salt
07:40 elfixit1 joined #salt
07:45 n8n joined #salt
07:52 kadel joined #salt
07:54 carlos joined #salt
08:04 harobed_ joined #salt
08:06 balboah joined #salt
08:06 ChaosPsyke joined #salt
08:12 topochan joined #salt
08:12 balboah joined #salt
08:12 johtso joined #salt
08:16 Kenzor joined #salt
08:32 linjan_ joined #salt
08:36 fllr joined #salt
08:38 slav0nic joined #salt
08:38 slav0nic joined #salt
08:47 yomilk joined #salt
08:51 hhenkel Hi all, is there a way to set the environment of a minion on the master?
08:52 hhenkel What I'm thinking of is a way to set the env in one place instead of matching the hostname in various locastions (pillars/states).
08:55 thayne joined #salt
08:59 xl1 joined #salt
08:59 CeBe left #salt
09:08 ndrei joined #salt
09:11 yomilk joined #salt
09:13 babilen joined #salt
09:14 lessismore joined #salt
09:14 bhosmer joined #salt
09:17 mike25ro hhenkel: not sure what you mean
09:22 giantlock joined #salt
09:24 giantlock_ joined #salt
09:30 hhenkel mike25ro: I have the possibility to match have base, dev and so on in both top.sls files.
09:30 ndrei joined #salt
09:31 hhenkel mike25ro: The question is, is there a sane way to set the env that should be matched once or is my only chance to set it on the minion
09:34 flebel joined #salt
09:35 bezaban salt is making much more sense to me using extensive use of pillars and jinja :)
09:37 bezaban what enviroment are you referring to hhenkel? :)
09:37 bezaban environment
09:38 hhenkel bezaban: Yes, i guess the way to go is with jinja and pillars.
09:38 bezaban that was not in regards to that specifically, just a comment to others starting out :)
09:39 hhenkel bezaban: I'm trying to get a good design when using gitfs. Furao here on the channel provided an interesting approach with roles.
09:41 hhenkel bezaban: But it look kind like an hen and egg problem. I can set the roles is pillars, but can I then have something else then a base env in my pillars?
09:41 hhenkel /is/in
09:43 mike25ro hhenkel: i use my environment ... as a variable on the pillar... each minion has its own pillar and it has an environment variable there... for me this was the most logical solution
09:43 bezaban ah, not familiar with that aproach I'm afraid, still trying to sort the best structure myself
09:44 hhenkel mike25ro: okay, but you only got a base env then in the pillar top.sls and match there the host on some id, right?
09:45 hhenkel mike25ro: in the state top.sls you're then matching on the roles set via the pillar, right?
09:46 hhenkel mike25ro: Are you using the gitfs backend for states or pillars?
09:46 mike25ro hhenkel:  one sec please:
09:48 mike25ro http://pastebin.com/hjuc4BNG
09:48 kickerdog joined #salt
09:52 mike25ro hhenkel:  does it make any sense?
09:53 mike25ro then you can target based on the environment
09:53 mike25ro salt -I "environment:office" state.sls my.state.file
09:53 hhenkel mike25ro: yes, I guess. 7 - 21 are all in the fqdn.sls, right?
09:54 mike25ro no ... office-saltminion01.sls
09:54 mike25ro that is the first part of the fqdn
09:54 hhenkel yes, that's what I meant.
09:54 mike25ro an fqdn is like x.y.z ... i need only the x part
09:55 hhenkel okay, so why are you not using grains['host'] then on line 3 ?
09:55 mike25ro good question.. let me see
09:56 mike25ro because i am a moron and i did not see that grain
09:56 mike25ro host is always the short hostname? i presume yes
09:57 hhenkel I thought so.
09:57 hhenkel eh, not your first comment, but the last... ;)
09:57 hhenkel What version of salt are you running?
09:58 zooz joined #salt
09:59 hhenkel I'm also wondering why you're setting the distro_os and distro_version. It seems like lsb_distrib_id and lsb_distrib_release are no longer available - at least on centos with salt 2014
09:59 mike25ro 0,17,1
09:59 malinoff 0.17.1 is quite old :)
10:00 mike25ro yeah .. i can imagine
10:00 mike25ro what would be a good... update procedure?
10:01 hhenkel I can see "os os_family osarch oscodename osfinger osfullname osmajorrelease osrelease" as grains for my system.
10:01 mike25ro just update the minions .. then the master?
10:01 mike25ro hhenkel: i don't have that on my grains
10:02 hhenkel mike25ro: okay, so then it will be interesting to update... ;)
10:02 xmj terminalmage: cedwards: ping
10:02 mike25ro hmm... yeah
10:02 xmj terminalmage: cedwards: does salt-cloud support freebsd Jails, and bhyve ?
10:03 hhenkel mike25ro: Thanks for the pastebin, that is more or less what I thought about. But I'm still not sure how that all fits to the gitfs environment stuff.
10:04 mike25ro sorry i could not help more
10:04 hhenkel With git you can use branches and each branch is mapped to an environment. I'm still currious how to set the environment for a host so that it uses the "dev" version of a file.
10:05 mike25ro now i have a question... in my top.sls pillar file.. can i set up a value that will be passed to all minions? like  x:y
10:07 hhenkel mike25ro: I guess you need to match all hosts and then set the key value pair. IMHO I read something that data will only be provided to hosts matching.
10:08 mike25ro yeah right
10:11 yomilk joined #salt
10:16 hhenkel Is there a way to target files in the top.sls that contain dots?
10:17 hhenkel for example if I got a file under at hosts/host1.example.com.sls ?
10:18 ggoZ joined #salt
10:18 mike25ro hhenkel: that is why i used the grains['fqdn'].split('.')[0] .... and then saved the minion.sls as the host1.sls instead of the full fqdn
10:19 mike25ro dots are not accepted.. as they are parsed as folder delimiters
10:20 hhenkel mike25ro: Yes, thought that was the reason...problem is that I got multiple domains and I'm more satisfied with the fqdn.
10:21 johtso joined #salt
10:21 hhenkel I guess I introduce another folder then with the name of the domain
10:23 mike25ro hhenkel: as a grain you men?
10:24 ede ls
10:24 ede oops.
10:25 hhenkel mike25ro: nope, more something like: hosts/example/com/host.sls
10:25 mike25ro ah right ... good idea hhenkel
10:26 hhenkel That gives me the chances to use the domain grain and the host grain
10:26 mike25ro hhenkel:  now ... i might think at the same thing...
10:26 mike25ro i know what you mean
10:27 hhenkel mike25ro: What do you think, is there the need / possibility to check if the file exists or is it just fine to work with the error /exception?
10:28 mike25ro hhenkel:  i would check if the file exists.. just to be on the safe side.
10:29 mike25ro hhenkel:  i have a web interface where i accept my minions.. and the pillar file is generated automatically - then i have some checkboxes to add roles / environment
10:29 mike25ro i always check ... to see if the automation is not breaking anything
10:29 mike25ro better safe than sorry
10:30 hhenkel mike25ro: yes, that's what I thought. So you're using jinja to test then?
10:30 mike25ro no, i use php
10:30 hhenkel mike25ro: Sounds like a neat thing to have.
10:30 hhenkel ah, okay you're checking within the web app.
10:31 mike25ro hhenkel:  it is a really helpful tool.... i manage all the initial deployments via the web interface
10:31 hhenkel mike25ro: github url? ;)
10:31 mike25ro yeah hhenkel inside the app i do all the checks.
10:31 mike25ro hhenkel:  nope.. is not public...
10:31 hhenkel mike25ro: that's sad. I'm off to lunch. till later.
10:31 mike25ro hhenkel:  .. i am thinking of rewriting it.. using F3 framework ... so maybe others can use it
10:32 mike25ro hhenkel:  enjoy your lunch
10:32 vbabiy joined #salt
10:36 fllr joined #salt
10:39 kickerdog joined #salt
10:40 Nexpro joined #salt
10:42 baboon` joined #salt
10:44 scott_w joined #salt
10:51 Nexpro1 joined #salt
10:54 r2dk1ng2 joined #salt
10:58 manicouman joined #salt
11:00 Kraln joined #salt
11:01 n8n_ joined #salt
11:07 bhosmer joined #salt
11:09 Chrisje joined #salt
11:36 sgflt joined #salt
11:36 sroegner joined #salt
11:40 yomilk joined #salt
11:42 hhenkel mike25ro: would be interesting to see - even as I'm not a big fan of php.
11:44 mike25ro hhenkel: i really don't understand why people are against php.. i know there are apps... that are more or less crappy (php) but still 85%of web pages... maybe more are php based
11:44 voozl the majority isn't always right
11:45 hhenkel Don't wanted to start a flame war....
11:45 mike25ro hhenkel:  i want to ... rebuild the app ... so i can show it to others as well
11:45 hhenkel mike25ro: http://pear.php.net/bugs/bug.php?id=12038
11:45 mike25ro hhenkel: brb my lunch time now
11:45 hhenkel mike25ro: Sure, enjoy your meal!
11:49 carlos joined #salt
11:50 quickdry21 joined #salt
11:53 Networkn3rd joined #salt
11:55 gparent joined #salt
12:10 bhosmer joined #salt
12:23 cpenner461 joined #salt
12:28 sroegner joined #salt
12:31 oz_akan_ joined #salt
12:34 oz_akan_ joined #salt
12:36 Katafalkas joined #salt
12:36 jrdx joined #salt
12:37 fllr joined #salt
12:39 _fllr_ joined #salt
12:40 r0rySavag3 joined #salt
12:42 ipmb joined #salt
12:44 timoguin joined #salt
12:47 Outlander joined #salt
12:48 wkf_ joined #salt
12:48 hhenkel Is it possible to match something within a pillar list (same as https://github.com/saltstack/salt/issues/752 for grains)?
12:51 timoguin hhenkel, from the command line? yes, you can match on pillar: 'I@pdata:foobar'
12:51 timoguin which i think is only documented on the compound matcher page: http://docs.saltstack.com/topics/targeting/compound.html
12:56 hhenkel timoguin: hmm, seems like it's not working for me...maybe I do someting wrong?
12:56 hhenkel I got something like roles: [ "weblogic", "http" ]
12:57 hhenkel but this: " salt 'I@roles:weblogic' test.ping" gives me => No minions matched the target. No command was sent, no jid was assigned.
12:58 hhenkel but looking at at the pillars of my test hosts I got one host with: roles: -weblogic -http
12:59 timoguin yea it could be broken. i've never used it and it didn't work for me either just now
12:59 timoguin 2014.1.0 on Ubuntu
13:00 hhenkel hmm, I'm also on 2014.1.0 with centos.
13:00 timoguin tried a few variations on the command with no luck
13:00 shoemonkey joined #salt
13:02 hhenkel Hmm, maybe someone around with an older version around to see if it works for them?
13:02 hhenkel mike25ro: ^^
13:03 timoguin it just kind-of worked for me.... i did this: salt -I 'users:timoguin:home:/home/timoguin' test.ping
13:03 timoguin so looks like it won't match on the key, but maybe works to match on list items
13:04 hhenkel timoguin: hmm, but do you have multiple entries?
13:05 timoguin no, not for that example. it's matching on a specific key.
13:05 yomilk joined #salt
13:06 hhenkel hmm, okay.
13:06 timoguin matching on 'users:timoguin:*' worked, but matching on 'users:timoguin' didn't
13:06 timoguin and i have other users defined
13:07 mpanetta joined #salt
13:08 cpenner461 joined #salt
13:09 toastedpenguin joined #salt
13:13 x11x joined #salt
13:16 shoemonkey joined #salt
13:16 mike25ro hhenkel: what? :)
13:17 juicer2 joined #salt
13:18 gnugnu9 joined #salt
13:20 GradysGhost joined #salt
13:23 manicouman joined #salt
13:24 pass_by_value joined #salt
13:24 faldridge joined #salt
13:25 timoguin hhenkel, it's working for me now after refreshing pillar
13:25 timoguin try: pillar.get roles
13:25 timoguin it that returns a proper list then this should work: salt -I 'roles:httpd'
13:26 jslatts joined #salt
13:27 hhenkel mike25ro: targeting nodes via a specific pillar from a list of pillars
13:28 mike25ro i haven't played with nodes.... sorry guys... all my crap is depending on pillar and states
13:28 hhenkel timoguin: salt '*' pillar.get roles ?
13:28 timoguin hhenkel, yea, does that work?
13:28 hhenkel mike25ro: hmm, node as is host...?
13:28 hhenkel /is/in/
13:28 mike25ro ah ok :) sorry
13:29 vejdmn joined #salt
13:29 mike25ro hhenkel: salt -I "environment:office" state.sls my.state.file   -> where environment: office   is a key:value in the pillar
13:29 andr386 joined #salt
13:29 mike25ro this is how i target my hosts
13:29 hhenkel timoguin: nope, doesn't look like. I get an empty reply on the get command.
13:30 timoguin hhenkel, okay you need to refresh pillar then. try running saltutil.refresh_pillar
13:30 timoguin and then doing a pillar.get again
13:31 hhenkel mike25ro: Yes, that's fine. I try to achieve this https://github.com/saltstack/salt/issues/752 (described for grains).
13:31 vejdmn joined #salt
13:31 hhenkel timoguin: yes! That works.
13:32 hhenkel timoguin: Any idea why it's necessary to do this?
13:32 timoguin so now you should be able to run -I 'roles:httpd' for matching now
13:32 timoguin hhenkel, pillar data is sent on minion startup. and when it's updated on the master it needs to be sent down to the minion
13:33 timoguin the weird thing for me is that my 'roles' pillar was showing up when calling pillar.items, but not when calling pillar.get KEY
13:33 timoguin not until i did a refresh anyway
13:36 mike25ro the pillar data should refresh pretty fast
13:36 mike25ro or at least on my ancient version 0.17.1
13:36 hhenkel timoguin: Yes, that was the same for me. pillar.items showed me the stuff but the get did not work at first.
13:37 timoguin oh, i know why.
13:37 timoguin pillar.items calls the pillar data on the master and renders it on the fly.
13:38 timoguin pillar.get actually calls it on the minion
13:38 timoguin which explains why pillar.get didn't show until a refresh
13:38 sporkd2 quick question, if I require_in another pkg from another state I just need to import that right?
13:40 andr386 joined #salt
13:40 hhenkel timoguin: yes, that sounds reasonable. Thanks for your help!
13:42 jaimed joined #salt
13:46 tsantero left #salt
13:46 fllr joined #salt
13:48 occup4nt joined #salt
13:48 _fllr_ joined #salt
13:50 quickdry21 joined #salt
13:51 mgw joined #salt
13:53 mortis__ how do i view the value of a single grain? something like .... salt someserver items.grains.somegrain
13:53 mortis__ oh wait
13:53 shoemonkey joined #salt
13:54 mortis__ got it!
13:54 mortis__ hehe
13:54 pkruithof joined #salt
13:54 mortis__ salt someserver items.grain somegrain
13:54 mortis__ :P
13:56 philipsd6 actually, isn't it: salt someserver grains.item somegrain?
13:56 timoguin yes
13:56 mortis__ yeah, sorry :)
13:56 mortis__ was re-typing and failed horribly
13:58 shoemonkey joined #salt
14:01 occup4nt howdy. question about deploy scripts: the examples in the repos are iirc bash, but is there any reason one could not use a python script ?
14:05 ndrei joined #salt
14:05 timoguin occup4nt, you mean the salt-bootstrap BASH script?
14:05 occup4nt yeah
14:06 kermit joined #salt
14:06 timoguin nope, no reason you can't use Python or whatever else you want. that script mainly determines the appropriate install method for the target OS
14:06 timoguin which mostly ends up being using the right package manager
14:07 anitak joined #salt
14:08 HeadAIX joined #salt
14:08 occup4nt gotcha
14:08 occup4nt i'll give it a spin
14:09 manfred joined #salt
14:11 shoemonkey joined #salt
14:12 occup4nt timoguin: thx
14:16 bhosmer joined #salt
14:16 occup4nt oh yeah, re salt-cloud - i thought i saw somewhere in the docs that you could add a virt interface via maybe a grain or somesuch but it was late at the time and well, i can't find it again. possible ?
14:17 pkruithof I want to provision multiple clusters via 1 master, so far I haven't had much success. Is there a good resource on how to do this?
14:17 pkruithof Basically I want the master to use different state files and pillar data for different projects, or am I wrong to want this?
14:18 shoemonkey joined #salt
14:18 manfred pkruithof: you want different environments
14:19 manfred pkruithof: http://docs.saltstack.com/ref/states/top.html
14:19 manfred check file_roots
14:20 shoemonkey joined #salt
14:20 cedwards xmj: i don't believe salt-cloud supports either jails or bhyve.
14:21 pkruithof I tried that before but without success, I can't recall why anymore so I'll give it a try again, thanks for pointing me to this. Do I have to clear caches or something when making changes to the salt states or is restarting the master enough?
14:21 scrootch joined #salt
14:21 manfred that i don't know, i haven't had a chance to really go into environments yet
14:22 kaptk2 joined #salt
14:24 rostam joined #salt
14:27 timoguin_ joined #salt
14:27 sroegner joined #salt
14:28 sporkd2 could someone point me to the right docs on requiring a pkg that was installed w/ pip.installed?
14:35 jcsp joined #salt
14:35 danielbachhuber joined #salt
14:37 roidelapluie joined #salt
14:37 opapo joined #salt
14:38 ChaosPsyke joined #salt
14:38 fllr joined #salt
14:39 colinbits joined #salt
14:40 mgw joined #salt
14:42 faldridge joined #salt
14:43 Signum joined #salt
14:48 renoirb Anybody has experience tracing with salt when it gets slow to run commands?
14:48 renoirb sporkd2: did you find an answer?
14:49 sporkd2 i did
14:49 renoirb ... are you asking to make such check within a state file or as a module?
14:49 sporkd2 ty
14:49 yomilk joined #salt
14:51 ajprog_laptop joined #salt
14:51 sporkd2 I was asking in a state
14:51 renoirb sporkd2, what did you find?   unless statement?
14:52 sporkd2 it didn't seem very intuitive to just use pip: (but thats just me)
14:52 sporkd2 i was trying pip.installed: etc first
14:53 renoirb In fact you have to have a require: - pkg: python-pip
14:54 renoirb Because otherwise it can try to execute the pip.installed before it even get installed sporkd2
14:54 renoirb before [pip] gets installed
14:55 pdayton joined #salt
14:56 sporkd2 true
14:56 sporkd2 thankfully I can change the ordering now :)
14:57 renoirb Ordering is also something to get your head into it
14:57 renoirb did you work yet with require, require_in, watch, watch_in?
14:58 sporkd2 a little
14:58 renoirb I've been using salt since last september and I still make mistakes.
14:58 sporkd2 i need to start using my pillars and grains more in my states
14:58 renoirb yes.
14:58 sporkd2 but I'm still in dev env.
14:58 renoirb Separation is good, but too much not better.
14:58 renoirb Are you using Vagrant?
14:59 sporkd2 for my local devs yes
14:59 sporkd2 i have a masterless vagrant and some ec2 boxes right now
14:59 renoirb What I think is best is to have a vagrant VM with salt states checkout as the development salt master. Then connect other computers with vagrant VMs to reproduce a live environment.
15:00 renoirb That's what I am working toward.
15:00 renoirb I manage a set of 20 VMs through an OpenStack cluster.
15:00 renoirb I'm moving away from developing from that environment.
15:00 _bt joined #salt
15:01 renoirb That's good. If you have a spare computer or two, just install vagrant with multiple boxes and configure them to use your development master that would have a static IP in your LAN.
15:01 renoirb You know how to have more than one VM per Vagrantfile?
15:02 sporkd2 yep
15:02 renoirb k, not everybody knows.
15:04 sporkd2 renoirb: check PM
15:08 spiette joined #salt
15:09 twobitsprite is a version 2014.1.0 master compatible with 0.17.1 minions?
15:10 joehillen joined #salt
15:10 sporkd2 twobitsprite: masters are backwards compatible
15:10 sporkd2 minions are not
15:10 JesseCW joined #salt
15:11 sporkd2 renoirb: that makes sense, i'm just trying to figure out how to logically separate my sls' logically
15:11 renoirb twobitsprite: do not worry about upgrading
15:11 renoirb I ran    salt '*' pkg. install salt-minion --refresh=True   without any problems  from 0.17 last week
15:12 renoirb but beware twobitsprite, pkg.upgrade is NOT the same as pkg.install    when you want to upgrade one given package such as salt-minion
15:12 mpanetta joined #salt
15:13 renoirb pkg.upgrade will upgrade ALL PACKAGES whereas pkg.install pkgname   will upgrade the pkgname package....  and with --refresh=True... it will take care of restarting the salt-minion daemon for you
15:14 renoirb twobitsprite ... and to answer your question directly. Not is not "compatible".   I had a new minion with 2014.1.0  and my master could not do anything with it.
15:14 rainmanjam joined #salt
15:14 renoirb ... had to upgrade the whole fleet.
15:15 rainmanjam joined #salt
15:16 sporkd2 im running 2014.1.0 on my master/minions
15:17 renoirb spork2, Have you thought of having separate salt submodule out of your code workspace.
15:18 renoirb ... I mean. When we develop an application that has separate components, you might want to separate them in separate git repositories.
15:18 ipmb joined #salt
15:18 twobitsprite thanks everyone
15:19 seapasulli joined #salt
15:20 shoemonkey joined #salt
15:21 gadams999 joined #salt
15:22 gadams999 left #salt
15:24 gadams9991 joined #salt
15:24 gadams9991 left #salt
15:27 foxx joined #salt
15:27 sporkd2 renoirb: FYI, if you include another sls, all those states get run
15:28 renoirb sporkd2: is that included file part of the top.sls file?
15:28 sporkd2 nope
15:29 sporkd2 my init has an include
15:29 bhosmer joined #salt
15:29 ckao joined #salt
15:30 renoirb Did you execute that state file through state.highstate
15:30 renoirb or through state.sls mymodule
15:30 sporkd2 renoirb: so what I'm doing is including my regular pkg.install states in init, then my project specific files etc in a secondary sls
15:30 sporkd2 so that you can easily find the project / box specific files
15:30 sporkd2 without having to sort through a bunch of states
15:31 frasergraham joined #salt
15:31 hunter____ joined #salt
15:31 renoirb Try to execute your module directly through state.sls mymodule. Some states you want to have them part of the highstate. Some otters not.
15:31 jeremyBass1 joined #salt
15:32 sporkd2 well right now I'm only dealing with highstate states
15:32 renoirb For example, we have a folder outside of the salt folders that has all deployed code clones in them.  And in the salt states has a "code" state (without init.sls).
15:33 renoirb So in code.nodetypename  I have a file code/nodetypename.sls that has all code rsync requirements.
15:33 ckao joined #salt
15:33 sporkd2 no I get what your saying
15:33 sporkd2 all I'm saying is you do not have to require states if they are included
15:33 renoirb to launch a new instance i do launch it through nova (eventually salt-cloud), then run a state.highstate on it... then state.sls code.nodetypename  (e.g. node_db)
15:34 renoirb I know what you are saying. But in some cases, when executing directly, to my experience, including is not enough.
15:34 renoirb But I might have to review some parts of my state file tree... and I can be wrong too
15:35 sporkd2 well what I'm trying to accomplish is a top env that brings up my app box
15:36 sporkd2 but organizing the state files in a way even someone without salt knowledge can tell where to edit the apache virtualhost (for example)
15:37 utz joined #salt
15:37 conan_the_destro joined #salt
15:37 Networkn3rd joined #salt
15:39 pkruithof is it possible to get pillar data in the same pillar file? For example if I set foo: bar, can I use {{ pillar['foo'] }} on the next line?
15:39 jeremyfelt joined #salt
15:39 ndrei joined #salt
15:40 sporkd2 pkruithof: defining pillars with pillars?
15:41 pkruithof yeah, I got a user which I define first, and later on I need that user as part of a directory, but I don't want to repeat the user value there
15:41 Ryan_Lane joined #salt
15:44 manfred pkruithof: i would be surprised if that works
15:44 utz1 joined #salt
15:46 utz joined #salt
15:48 ndrei joined #salt
15:51 Iwirada left #salt
15:51 JesseCW joined #salt
15:51 vejdmn joined #salt
15:53 forrest joined #salt
15:53 ndrei joined #salt
15:54 utz1 joined #salt
15:55 utz1 joined #salt
15:55 utz1 joined #salt
15:56 utz1 left #salt
15:56 ml_1 joined #salt
15:59 lessismore left #salt
15:59 ndrei joined #salt
16:00 kermit joined #salt
16:00 jalbretsen joined #salt
16:00 ndrei_ joined #salt
16:01 utz joined #salt
16:02 druonysus joined #salt
16:02 druonysus joined #salt
16:05 timoguin joined #salt
16:05 Plam joined #salt
16:05 Plam hi
16:06 Plam a question about generating a config file with all minions and their fqdn
16:06 Plam is there a better way than this: http://stackoverflow.com/questions/17158665/how-to-get-a-list-of-all-salt-minions-in-a-template ?
16:07 Plam (better, understand simpler without Saltmine)
16:08 gadams999 joined #salt
16:15 xmj terminalmage: ping
16:19 Katafalkas joined #salt
16:19 Katafalkas joined #salt
16:20 rainmanjam left #salt
16:20 akl joined #salt
16:20 akl joined #salt
16:21 gnugnu9 joined #salt
16:22 pkruithof manfred: as you expected, does not work :( oh well
16:25 sroegner Plam: i'd say the salt mine is what you want to use
16:27 sroegner Plam: alternatively you could use peer calls but the that doesn't really scale
16:27 elfixit1 joined #salt
16:27 Plam sroegner: ok. I wish I could use a kind of minions list directly in the sls, and then iterate on it
16:28 Plam and for each element apply its selected grain
16:28 sroegner Plam: and you would want to edit that list whenever a minion comes in or out?
16:28 erchn joined #salt
16:28 Plam nope, it's really for all minions
16:28 erchn hi all, good morning (here anyway)
16:29 ndrei joined #salt
16:29 erchn quick question, is it possible to target minions based on a pillar value where the value is a list item?
16:29 ipalreadytaken joined #salt
16:29 ndrei_ joined #salt
16:29 erchn say you have "myapp: [ 'a','b'], I want to match 'myapp:a'
16:30 sroegner Plam: what I do is to import a settings.sls into the actual one - the settings state does the ugly jinja call and returns a dict of values (including the list of minions)
16:31 Plam and how did you get this list of minions?
16:31 sroegner Plam: https://github.com/saltstack-formulas/hadoop-formula/blob/master/hadoop/hdfs/settings.sls
16:32 Plam thx
16:32 chrisjones joined #salt
16:32 bemehow joined #salt
16:33 pdayton joined #salt
16:33 bemehow_ joined #salt
16:34 hunter____ joined #salt
16:34 sroegner Plam: one more thing: the stackoverflow example you quoted advertises specifying the interface in the state (eth0 in that example)
16:34 KyleG joined #salt
16:34 KyleG joined #salt
16:34 timoguin erchn, yes, you can. i did this earlier today. if you're talking about the CLI, you can do this: salt -I 'key:value' test.ping
16:34 erchn sroegner, why not just use config.get wouldn't that keep you from having to look at grains then pillar for the various items?
16:35 erchn timoguin; do you know if that works on 0.17.x also?
16:35 timoguin I think it does. I tried today on 2014.1.0. If doing a pillar.get 'key:value' works from the CLI, the pillar match will probably work too
16:35 erchn timoguin, I ask becuase I'm getting no luck using that syntax
16:35 timoguin The issue we ran into earlier is the pillars hadn't been refreshed, so the minion didn't actually have them.
16:36 sroegner erchn: I try to merge individual items without relying on the framework - not sure config.get would do that
16:36 dave_den erchn: make sure your minion has refreshed its pillar.
16:36 dave_den as timoguin just said  :)
16:36 erchn http://pastebin.com/UUEGixtV
16:36 timoguin try pillar.get
16:36 dave_den salt-call != salt
16:36 timoguin pillar.items grabs pillar data from the master and renders it on the fly
16:37 dave_den you need to tell salt-minion to refresh the pillar.
16:37 dave_den salt 'minionname' saltutil.refresh_pillar
16:38 Signum I'm very new to Salt… Just set up a master and a minion and can run "salt" commands. Can I make minions pull any changes from the master automatically? Like I change an SLS and the minion will pick it up? (I'm used to such behavior from Puppet.)
16:38 erchn ahh I found the issue, turns out the issue was between my ears, lol...my test minion wasn't running, thanks guys
16:38 erchn the VM was, but salt-minion was shutdown
16:39 erchn I forgot to check if it was "plugged in"...jeez
16:40 erchn signum, look at the salt scheduler...
16:40 erchn http://docs.saltstack.com/topics/jobs/schedule.html
16:41 arodizzy_ joined #salt
16:42 Signum erchn: thanks
16:42 Gordonz joined #salt
16:42 jacksontj joined #salt
16:43 Gordonz joined #salt
16:44 ipalreadytaken joined #salt
16:45 n8n joined #salt
16:46 tmandry joined #salt
16:46 jbub joined #salt
16:48 hunter____ joined #salt
16:48 Ryan_Lane joined #salt
16:49 thayne joined #salt
16:49 quantumriff joined #salt
16:52 joehillen joined #salt
16:53 quantumriff If I want to run a command via a state file, and pass in the value of eth0.. is that just {% grains.get['ip_interfaces']['eth0'] %} ?
16:56 schimmy joined #salt
16:56 mgw joined #salt
17:00 quantumriff okay, I have found that this returns ['10.10.10.10'], but how do I make it return the value, minus the bracket and quotes so I can pass in just the value? {{ grains['ip_interfaces']['eth0'] }}
17:01 schimmy joined #salt
17:01 rgbkrk joined #salt
17:02 dave_den ['eth0'][0]
17:03 quantumriff dave_den: thanks  I have lots more python to learn.. I'm more of bash guy :)
17:04 dave_den np
17:04 gcbirzan Hm!
17:05 gcbirzan I'm just guessing for now, but, is it possible that Ic annot use decorated functions for my modules?
17:05 fauxtoniche joined #salt
17:06 Ryan_Lane joined #salt
17:06 Signum erchn: I'm not sure how to properly enable scheduling on the clients. I have added a pillar on the server to enable a highstate scheduler. But it seems like the clients need to have that information. Some people seem to run salt locally via cron and put it into a high-state. What's the right way?
17:07 erchn both are valid methods
17:07 timoguin Signum, there's really no right way. I would probably opt for cron over the Salt scheduler, since cron has a lot more granular control.
17:07 timoguin But you can also initiate state runs on the minions if you wanted.
17:07 erchn signum, once you run saltutil.refresh_pillar it should pick things up
17:08 ml_1 joined #salt
17:08 erchn (I personally use cron with salt-call)
17:08 Ahlee There's an outstanding issue for a future release to make salt's schedulig not braindead, too
17:08 Ahlee for the time being we're stuck with cron or compeltely not predictble schedules
17:09 shoemonk_ joined #salt
17:09 erchn luckily you can control the cron from the highstate cron runs, if you so choose
17:09 erchn s/the cron/the cron entries/
17:09 gcbirzan Except when you need to add it there!
17:10 erchn well, you're going to need to run something on the mast at some point
17:10 erchn s/mast/master/
17:10 Ahlee i suppose i could adopt the web2.0 mentality and install a bunch of servers and run chronos to schedule
17:13 Signum erchn: Great, that seems to work.
17:14 fauxtoniche joined #salt
17:14 erchn np
17:14 scalability-junk joined #salt
17:15 Signum erchn: Is it common to make the minion run the top.sls every few minutes automatically? Or is salt's idea merely to make the sysadmin push the changes (Ansible-style)?
17:15 raizyr joined #salt
17:15 Ahlee Signum: It's entirely up to the admin.  Salt works hard to not dictate that
17:15 erchn signum, I personally don't run highstate on most of my nodes on schedule, I target specific sls files using "state.sls <state_sls_file_to_run>
17:15 ajw0100 joined #salt
17:16 Ahlee That's a common pattern.
17:16 erchn signum, mainly for performance reasons, highstate is generally overkill
17:16 Ahlee One I, too, use but am trying to move away from in favor of more of puppets "the servers will always look like this"
17:16 Signum So the admin could just try changes on a staging server and then distribute the changes to all systems, right?
17:16 Ahlee by running highstate and enforcing consistency
17:16 erchn signum, yea fo rsure
17:16 analogbyte joined #salt
17:17 Signum Ahlee: Yes, that's what I liked in Puppet. On the one hand you were sure that systems always had the right state. On the other hand it was easy to break dozens of servers at once by a careless recipe change. :)
17:17 Ahlee Signum: Indeed.
17:18 Ahlee Right now I use a lot of state.sls and only apply specific items at specific times, so I still don't get the benefit of confidence that my systems are as I expect them to be
17:18 dfinn joined #salt
17:18 Ahlee mostly becuase my memory is bad and I can never remember what's where, and when
17:19 Signum That's a risk, yes. And then when you enable scheduled highstates suddenly things happen that you didn't expect. There's indeed no best way probably.
17:19 dfinn i'm a little confused by http://docs.saltstack.com/topics/releases/2014.1.0.html#saltenv.  we have env used quite a bit in our sls files, like so for a source : - salt://packages/ad_client/files/krb5.conf.{{ env }}.  does that mean we will need to change env to saltenv in those entries as part of the upgrade to 2014.x.x?
17:19 erchn there's certainly nothing stopping you from creating a bunch of periodic states that aren't part of highstate, for deployment reasons or whatever, but still scheduling highstate to run on a schedule
17:19 timoguin dfinn, yes
17:20 Signum I think I'll start with the manual "push" way. I was trying Ansible in the past days but I found it too inflexible for real-life scenarios. So I hope that Salt is the right middle ground between Puppet (mighty and weird) and Ansible (basic and sexy). :)
17:20 Linz joined #salt
17:20 timoguin dfinn, actually, looks like not. support for 'env' won't be removed until a future release, so it should work
17:20 bhosmer joined #salt
17:20 dfinn anyone know how far out in the future that might be?
17:20 erchn signum, I agree with your assessment
17:21 dave_den dfinn: boron, iirc.
17:22 dave_den which is 4 releases away… probably 12+ months out
17:23 hunter____ joined #salt
17:23 rojem joined #salt
17:23 shoemonkey joined #salt
17:23 Plam left #salt
17:26 ndrei joined #salt
17:27 kballou joined #salt
17:31 ChaosPsyke joined #salt
17:37 ajw0100 joined #salt
17:38 thayne joined #salt
17:40 amckinley joined #salt
17:40 zain joined #salt
17:42 arthabaska joined #salt
17:44 Saint-N ya know that summary after you run a salt.highstate
17:44 Saint-N is there a way to ONLY show the failed lines?
17:45 Saint-N from salt-call
17:47 dave_den state_verbose: False
17:47 dave_den i think salt-call should honor that
17:47 tonthon Hi
17:48 tonthon there seem to be recent changes on cmd.wait
17:49 bhosmer joined #salt
17:51 tonthon a single cmd.wait state statement used to fire, but in the latest version it doesn't
17:53 dave_den do you have any other info?
17:53 dave_den and what does your state look like
17:54 analogbyte joined #salt
17:54 redondos joined #salt
17:56 khaije1 joined #salt
17:57 johtso joined #salt
17:57 johtso joined #salt
17:58 johtso joined #salt
17:58 ajw0100 joined #salt
17:58 tonthon dave_den: http://pastealacon.com/34134
17:59 tonthon this states fires on 0.17.5 but not on 2014.1
17:59 johtso joined #salt
17:59 dave_den tonthon: so what's it supposed to wait for?
17:59 tonthon I don't say it's supposed to wait
17:59 dave_den you need a 'atch' requisite so it knows when to run
17:59 tonthon now yes, previously not
17:59 n8n joined #salt
17:59 dave_den cmd.wait waits for something
18:00 dave_den your probably want cmd.run
18:00 kermit joined #salt
18:00 khaije1 does a local-master (file_client: local , master: localhost) minion config have support for multiple environments?
18:01 tonthon dave_den: yes, I wanted to let people know some miswritten states that worked previously could be broken with the latest update
18:01 asmaps left #salt
18:03 dfinn joined #salt
18:05 ajw0100 joined #salt
18:09 hunter____ joined #salt
18:12 ipalreadytaken joined #salt
18:13 smcquay joined #salt
18:14 Katafalkas joined #salt
18:17 zain joined #salt
18:19 philipsd6 If I have state.sls and an example.pillar that I'm developing, how can I display the state.sls after it's been run through the rendering engine?
18:20 repl1cant \
18:20 bhosmer_ joined #salt
18:22 fauxtoniche joined #salt
18:25 thayne joined #salt
18:25 timoguin philipsd6, state.show_sls
18:28 redondos joined #salt
18:28 redondos joined #salt
18:29 jergerber joined #salt
18:30 bryan joined #salt
18:32 kickerdog joined #salt
18:32 googolhash joined #salt
18:33 pkruithof joined #salt
18:34 philipsd6 thanks timoguin-- not sure why I couldn't find that... :(
18:35 nulldude joined #salt
18:36 nix_ joined #salt
18:37 ndrei joined #salt
18:37 ndrei_ joined #salt
18:37 timoguin philipsd6, we all need pushes in the right direction sometimes. :)
18:37 kzx it seems to be job control on windows doesn't work, I spawned salt 'user*' cmd.run "C:/windows/system32/calc.exe" , it spawned calculator on minion, then I issued saltutil.term_job <jobid> from another terminal, it made cmd.run command return but the calculator application stayed up on th minion, am I doing something wrong?
18:43 chrisjones joined #salt
18:43 xmltok joined #salt
18:44 ndrei joined #salt
18:44 ndrei_ joined #salt
18:45 kvbik joined #salt
18:46 kzx term_job doesn't seem to work it seems it kills the terminal with the salt minion running but leave the application (calc) running
18:47 zain joined #salt
18:49 mgw joined #salt
18:52 googolhash joined #salt
18:53 Ryan_Lane joined #salt
18:53 kickerdog joined #salt
18:55 budrose joined #salt
18:55 kzx does anyone know of a better way to starting and stopping windows applications (these are gui applications not services), user powershell invoke-item ?
18:57 r2dk1ng2 joined #salt
18:57 Ahlee i'm pretty sure saltutil.term_job refers to the salt job, not a process on a minion
18:59 philipsd6 I think you're right Ahlee: http://docs.saltstack.com/ref/modules/all/salt.modules.saltutil.html#salt.modules.saltutil.term_job
18:59 Ahlee you'll likely want to look through the windows process list and kill from within powershell or similar
19:00 Ahlee alternately, can windows minions use service? and if so, can you rework your app to be a serivce to be started/stopped?
19:00 xmltok anyone using git_pillar in 2014.1.0? my server appears to be fetching the pillar data over and over again, not sure if i have bad data or if something else is going on
19:00 kzx Ahlee, philipsd6: that helps, i'll give that a try
19:00 xmltok the pillar data looks good
19:00 kzx Ahlee: unfortunatley it is a GUI application that needs to started
19:00 philipsd6 Back to my jinja rendering issue... Why am I getting this? Jinja syntax error: no test named 'mapping'
19:01 Ahlee kzx: Currently, I use a wrapper script on my windows hosts that uses subprocess.Popen to spawn them.  I do not execute that job through salt, yet, though.
19:01 philipsd6 the mapping test is built in to jinja since v2.6
19:01 philipsd6 I've tested a simple template using Jinja directly in my python shell, but it doesn't work in my state.sls
19:02 kzx Ahlee: so you run piece of python, it does subprocess.Popen() then you have the pid of the process and can control it ?
19:04 vejdmn joined #salt
19:04 Ahlee kzx: I don't track the PID. I spawn the windows apps, they run, then at the end of the day I run another process that runs through all the windows hosts and kills the PID if it managed to survive the entire day
19:05 xmltok it appears to be an infinite loop when i define two ext_pillars that use git, if i use either one its ok, but when two are defined it just fetches them over and over again
19:07 diegows joined #salt
19:07 smcquay_ joined #salt
19:09 che-arne joined #salt
19:10 steven__ joined #salt
19:10 MedicalJaneParis joined #salt
19:10 steven__ Hello -- anyone here using salt-cloud w/AWS, and wanted to change root disk size?
19:11 kzx Ahlee: $p=Start-Process <executable> -passthru ; $p.id also seems to give out the pid
19:11 Ahlee kzx: nice, sounds like that's what you need, right?
19:12 kzx pretty close, I was hoping to use salt job for it
19:15 frasergraham joined #salt
19:15 steven__ Because when I try to use the block_device_mappings in cloud.profile, I get a *hang*. Salt-cloud just sits there at :[INFO    ] Salt node data. Private_ip: 10.113.51.82 and goes nowhere.
19:16 steven__ Ah-hah! And now: [ERROR   ] Port connection timed out: 900 Error: There was a profile error: Failed to connect to remote ssh
19:16 Ahlee Right, but that's not what salt's for.  Salt's for starting services, not wrapping services and running them
19:16 Ahlee or, well, yeah.
19:18 kzx Ahlee: I guess I am confused by cmd.run and job behavior, it seems to block, has timeout options, runas, etc, if it is not intended for starting application and keeping them running why even provide term_job and kill_job ?
19:18 Ahlee kzx: again, term_job and kill_job are for salt jobs
19:19 baniir joined #salt
19:20 Ahlee Salt jobs are processes intenral to salt - i.e. the cmd.run is a salt job.  What cmd.run executes is not
19:20 kzx Ahlee: so what is use is controling a salt job resulting from cmd.run by itself without controlling the application it spawned, as in what is the purpose even exposing that to the user?
19:20 shoemonkey joined #salt
19:21 Ahlee I don't follow.  You mean why is it returning the salt jid?
19:21 timoguin every command you run on salt gets a unique JobID to track executions and results.
19:21 Ahlee you can see the job cache in action by salt \* test.sleep 300
19:21 Ahlee er, salt -v \* test.sleep 300
19:21 timoguin the job stuff in salt is for managing these job runs, i.e., what's running, what were the results, kill running jobs, etc.
19:21 Ahlee then salt-run jobs.lookup_jid <jid returned>
19:22 hunter____ joined #salt
19:24 kzx Ahlee: alright, I think it makes sense
19:26 Ahlee also keep in mind, windows minions are rare.
19:26 shoemonkey joined #salt
19:26 Ahlee You will likely find bugs and things that don't work. :)
19:26 rps joined #salt
19:26 kzx Ahlee: hehe, I figured that much by now ;-)
19:27 kzx that is why I want to automate everything and got to using salt, i know how handle linux machine pretty well even without salt, just didn't want to deal with windows, but oh well
19:27 rps is there a way to recursively copy a directory ONLY if the directory doesn't exist yet? file.recurse doesn't seem to honor either "unless" or "replace: False" ...
19:32 timoguin rps, that would be a nice feature for file.recurse to have.
19:32 timoguin maybe you can do something with file.exists
19:33 timoguin like a jinja conditional that checks, and only includes the file.recurse state if it returns false
19:34 Ahlee timoguin: got an example of jinja conditional check?
19:34 rps cute ... if super super hacky :) ... I would probably opt for a tarball install at that point - or add that feature if I had the time ;)
19:34 Ahlee my jinja sucks.
19:34 timoguin yea so does mine. i don't have anything off-hand
19:35 Ahlee rps: open an issue on the tracker, I'm sure it's useful enough to be picked up quickly
19:35 timoguin agreed
19:35 naiyte joined #salt
19:36 Ahlee also, I thought directory.recurse was for permissions?
19:37 thayne joined #salt
19:37 dave_den {% if salt['file.directory_exists']('/path/to/dir') %} should do it
19:37 rps nice
19:37 hunter____ joined #salt
19:38 Ahlee dave_den: nice man, thanks
19:38 dave_den there is an issue open for universal if/unless
19:38 dave_den if i can find it
19:39 schristensen anyone know what this error means when attempting to install a yum package: No information found for "<package name>".
19:39 ipalreadytaken joined #salt
19:39 schristensen ?
19:39 dave_den https://github.com/saltstack/salt/issues/1976
19:39 Ahlee yikes that's old
19:39 dave_den opened 2 years ago, heh
19:40 rps well, UNIVERSAL is complicated
19:40 rps :)
19:41 Ryan_Lane joined #salt
19:42 n8n joined #salt
19:43 kzx what is the raet and flo work happing in development of Salt?
19:43 schristensen rps, keep in mind that the salt['file.directory_exists']() test will be executed on the minion and not the server.
19:44 timoguin kzx, that's a new transport mechanism that's being developed
19:44 Ahlee kzx: afaik optional transport layers that are drop in replacements for zeromq
19:45 Ahlee hrm.  clever way of dtermining via pillar or build in grains if a minion is running on the same host as the master?
19:45 rps of course really what you need is {% if not salt['file.directory_exists']('/path/to/dir') %} ... but it works like a charm!
19:46 kzx timoguin, Ahlee: interesting, looks exciting
19:46 obimod joined #salt
19:49 timoguin Ahlee, just did this on my master, which is a minion of itself: salt 'foo' config.get master
19:49 timoguin which returns 'localhost' for my master
19:50 Ahlee timoguin: yeah, sadly I use cnames for the master host
19:50 Ahlee and checking for a grain that doesn't exist blows up the jinja renderer it appears
19:50 Ahlee hrnm.
19:51 dave_den you could just assign a pillar value to your master.
19:52 dave_den but i don't know if there's a way to do it completely dynamically by default
19:52 Ahlee yeah, i was trying to render a pillar based on the grain if it existed.  I'll just flip that and manually specify
19:52 timoguin if you use grains.get('key', 'defaultvalue') it won't blow, btw
19:52 kzx Ahlee: for grain checking I use get() instead of [] which just returns None instead of blowing up
19:52 timoguin so if you set default to an empty string you could check against that
19:53 kzx Ahlee: I have a case were I am trying to run 2 minions (on windows) and one has a grain set no it, so it does something else based on that grain
19:53 Ahlee oh money.
19:53 Ahlee Thanks all.
19:53 kzx {% if grains.get('role') != 'myrole' %} ...
19:54 Ahlee yeah, that'll work.
19:55 Ahlee i'm grains.setval'ing on install, so this works
19:56 aleszoulek joined #salt
20:00 ndrei joined #salt
20:00 ndrei_ joined #salt
20:00 quickdry21 joined #salt
20:03 vimalloc Is it possible to find the ip of a non-accepted salt key on the master?
20:05 FL1SK joined #salt
20:05 DaveQB joined #salt
20:14 timoguin vimalloc, i'm not seeing a place to do that. looks like salt-key does. and the event bus doesn't show anything related to the IP in the initial auth request
20:14 timoguin err, *looks like salt-key doesn't
20:16 vimalloc Hrm. Alrighty, thanks for checking. Let the which hunt for the mystery minion begin
20:16 timoguin seems like that'd be useful info to have when decided whether or not to accept a key
20:17 vimalloc ++. I'll open a feature request.
20:20 Gifflen joined #salt
20:21 CeBe joined #salt
20:21 linjan_ joined #salt
20:23 Ahlee vimalloc: i accept and then target the mystery minion and ssh to the grains['ip']
20:23 Ahlee but, I trust my environment entirely, so that may not be valid for your use case.
20:24 zz_cro joined #salt
20:24 linjan__ joined #salt
20:30 kballou joined #salt
20:32 kickerdog joined #salt
20:33 hunter____ joined #salt
20:34 schimmy1 joined #salt
20:39 manicouman joined #salt
20:42 shoemonk_ joined #salt
20:44 faldridge joined #salt
20:49 steven__ Trying again, as it's been a while. ;)  Anyone here used salt-cloud to change the root disk in AWS? I try it, and I get a hang until ssh times out.
20:53 fllr joined #salt
20:53 forrest steven__, I haven't tried that before. Is there any sort of output in the logs? And are you trying to do that via salt cloud itself, or after the minion is joined (not super familiar with AWS disk swapping)
20:55 steven__ salt-cloud itself, via the example in the documentation.  I have this in my cloud.profiles: block_device_mappings:     - DeviceName: /dev/sda       Ebs.VolumeSize: 20
20:55 steven__ What I get for an error is a long wait, and then: [ERROR   ] Port connection timed out: 900 Error: There was a profile error: Failed to connect to remote ssh
20:56 forrest steven__, weird
20:56 philipsd6 I still haven't been able to figure this out -- why can't I use the jinja2 mapping test in my sls?
20:56 philipsd6 Rendering SLS "base:test" failed: Jinja syntax error: no test named 'mapping'; line 3
20:56 philipsd6 {% if group is mapping %}    <======================
20:57 steven__ That's what I thought.  It's as if the resize is failing -- since I can't ssh into the boxes via bare ssh -- but I presume someone has tested it at some point. ;)
20:57 DenkBrettl 'lo philipsd6 :-)
20:59 dave_den philipsd6: what version of jinja are you using?
20:59 philipsd6 Hiya DenkBrettl -- I'm on vacation, pretend you didn't see me here ;)
20:59 philipsd6 I'm using jinja 2.7
20:59 philipsd6 mapping test was supported from 2.6 on
20:59 philipsd6 I tested using jinja directly, and it works fine.
21:00 DenkBrettl /ignore philipsd6
21:00 DenkBrettl :-)
21:00 ndrei_ joined #salt
21:00 ndrei joined #salt
21:01 meteorfox joined #salt
21:02 manicouman joined #salt
21:05 bhosmer joined #salt
21:12 frasergraham joined #salt
21:16 frasergr_ joined #salt
21:16 allanparsons joined #salt
21:16 allanparsons for file.*
21:16 allanparsons if i use a url as the file source
21:16 allanparsons it doesnt seem to follow 301/302 redirects
21:16 allanparsons is that by design?
21:18 kermit joined #salt
21:18 johtso joined #salt
21:19 johtso joined #salt
21:19 johtso joined #salt
21:22 schimmy1 joined #salt
21:22 shoemonkey joined #salt
21:23 allanparsons looks like redirects dont work for file.copy but work for file.managed
21:23 Eugene sounds like a bug.
21:25 allanparsons man, its super annoyingto have to use hashes
21:25 allanparsons when downloading files
21:25 allanparsons i get the security aspect of it.  but sheesh.
21:26 allanparsons i should be able to bypass a has if i i know i want to trust the file.
21:26 allanparsons or, i may replace the file.  (i build redis servers and want to restore a .rdb file from s3).  I dont always wan to recalculate the hash if i'm serving that file from my own s3 acct.
21:27 yano joined #salt
21:30 cpenner461 joined #salt
21:34 shoemonkey joined #salt
21:37 DenkBrettl joined #salt
21:39 shoemonk_ joined #salt
21:40 frasergraham joined #salt
21:47 gparent_ joined #salt
21:47 MTecknology joined #salt
21:48 jforest joined #salt
21:49 ndrei joined #salt
21:49 ndrei_ joined #salt
21:50 [diecast] joined #salt
21:50 KyleG1 joined #salt
21:54 xmltok did pyobjects make it into 2014.1.0?
21:54 zain joined #salt
21:59 KyleG joined #salt
21:59 KyleG joined #salt
21:59 googolhash joined #salt
22:05 r0rySavag3 joined #salt
22:10 64MAAB60T joined #salt
22:10 17SAAVE84 joined #salt
22:11 sroegner joined #salt
22:13 derks joined #salt
22:13 derks hello.  I have an issue whre two state id's conflict
22:13 derks For example, a users.sls that defines a 'nginx' user... and an app.sls that defines the 'nginx' package.  both are the same id
22:14 derks is there an alternative way to define something where the 'id' doesn't define the 'package name' to install, or 'user name' to create?
22:16 allanparsons man, does anyone have a working redis upstart script for Ubuntu 12.04
22:16 allanparsons mine starts redis
22:16 allanparsons but doesnt stop it correctly
22:16 derks ahhh.. I think I see... I would need to define "user.present" and "pkg.installed" under the same state id
22:17 dopp derks: yeah, I just ran into a very similar issue
22:17 dopp a user and group that were the same name
22:17 dopp so for the group I just used some other ID, and set '- name: <group>'
22:18 derks dopp, ok I thought that might work also
22:18 dopp I think you could do: nginx-user: as the ID and then do - name: nginx
22:18 allanparsons heres my upstart script:  http://pastebin.com/03FKqfLy
22:19 dopp I might actually get away from implicit IDs altogether, as I've hit the ID conflict situation multiple times
22:20 berto- joined #salt
22:20 jbub joined #salt
22:20 foxx joined #salt
22:21 derks Another issue: "These values could not be changed" on setting the users home dir?
22:22 johtso joined #salt
22:22 krow joined #salt
22:24 meteorfo_ joined #salt
22:25 Gifflen joined #salt
22:28 jidar left #salt
22:29 dopp yeah, I just ran into that, too. I was unable to use a user state to change a user's home dir
22:29 dopp not sure how to work around it.
22:29 dopp (nor am I sure if it should or shouldn't work)
22:30 * dopp a salt newb
22:30 timoguin joined #salt
22:30 xmltok do syndic nodes only need the syndic process or do they also need the master process? and is it possible to configure a syndic to read from multiple masters?
22:33 lionel joined #salt
22:33 Heartsbane basepi: I am going to get you
22:33 basepi xD
22:34 hunter____ joined #salt
22:34 Heartsbane I do not approve of your humor
22:34 basepi i have Friday-afternoon brain.  I can't be held responsible!
22:35 Heartsbane I knwo the feeling that is why we had to go for kawfee
22:35 * Gareth slips a note that reads "Give Gareth $50" in basepi's pocket
22:35 basepi WHOA HOW DID THAT GET THERE
22:38 steven__ salt-cp 'basepi' gareth.50.note /dev/pocket
22:39 hunter____ joined #salt
22:40 basepi salt-call cmd.run 'mv /dev/pocket/gareth.50.note /dev/null'
22:40 basepi Again, Friday brain, pretty sure that command wouldn't work.  But you get the idea!
22:41 steven__ Last try of the day, just in case more salt-cloud experts have wandered by: Anyone here used salt-cloud to change the root disk in AWS? I try it, and I get a hang until ssh times out.
22:43 krow joined #salt
22:44 shoemonkey joined #salt
22:44 thayne joined #salt
22:44 allanparsons @steven_ what are you trying to change on the root disk?
22:45 steven__ root volume and filesystem size, so I can have a decent-sized root partition without having to do the remount/rebuild every time.
22:45 allanparsons hm
22:45 allanparsons no ive never done that
22:45 allanparsons i usually add a volume
22:45 allanparsons and then attach it with an ssh script via salt
22:45 allanparsons lemme check docs
22:45 steven__ THere's an example in the salt-cloud documentation, but when I try and get it to work, it hangs.
22:46 scalability-junk joined #salt
22:47 mgw joined #salt
22:47 allanparsons @steven__ try this
22:47 allanparsons http://pastebin.com/9AdLNvzG
22:47 allanparsons https://github.com/saltstack/salt-cloud/blob/develop/salt_cloud/doc/topics/aws.rst
22:47 allanparsons "You can also use block device mappings to change the size of the root device at the provisioing time. For example, assuming the root device is '/dev/sda', you can set its size to 100G by using the following configuration."
22:48 akoumjian joined #salt
22:48 steven__ That's what I've got, sadly.
22:49 allanparsons really?
22:49 steven__ http://pastebin.com/Bm5PZ1Ez
22:49 scrootch joined #salt
22:50 steven__ (I first tested it at 300, and thought the delay might have been due to the size, so I tried somethign smaller.)
22:50 allanparsons hey, see this ticket:  https://github.com/saltstack/salt-cloud/issues/849
22:51 allanparsons hm
22:51 allanparsons i dunno then
22:51 steven__ Hm. Well, it says the code was merged, and I'm running 2014. Very odd.
22:51 allanparsons i usually just attach an EBS volume
22:51 allanparsons :(
22:51 steven__ I may have to try that. It's just a nuisance repathing everything. Thanks...
22:52 allanparsons i stopped using salt-cloud after seeing i couldng attach instances to elbs
22:52 allanparsons or create IAM roles, etc.
22:52 allanparsons i began using cloudformation
22:52 allanparsons with a user-data script
22:52 allanparsons that attaches the node to ec2
22:53 allanparsons see the second script here:  https://gist.github.com/allanparsons/4ce9645e74c1a5ddda6a
22:55 steven__ Interesting. I'll have to take a look. Salt-cloud's pretty tied in to what we're doing, but...
22:55 jeremyfelt joined #salt
22:55 andrej joined #salt
22:57 scrootch ?
22:58 scrootch joined #salt
22:59 n8n_ joined #salt
23:00 derks This is driving me crazy.  Ubuntu 12.04 LTS .... "start salt-master" says its running... but its not and won't start.  running "salt-master" starts up fine
23:00 steven__ Now *that* is weird. I moved where the configuration was, from after the grains to before the grains. It built, but it didn't change the root fs.
23:01 scrootch joined #salt
23:01 Ryan_Lane joined #salt
23:02 mgw joined #salt
23:03 manicouman joined #salt
23:05 frasergraham joined #salt
23:06 jslatts joined #salt
23:06 manicouman joined #salt
23:06 [diecast] joined #salt
23:06 dopp derks: any errors in /var/log/salt/master?
23:06 allanparsons @derks probably a permission issue or issue w/ your upstart script
23:07 steven__ OK. Ne'er mind. I'd changed something else as well. It works building /dev/sdb, but not altering /dev/sda, at least for me.
23:07 frasergraham joined #salt
23:07 derks dopp, allanparsons when I run "salt-master" manually ... no prob (as root) (and master is configured with all defaults to run as root).... but when i "start salt-master" it says "start/running, process PID" ... nothing in the log
23:07 derks at all
23:08 allanparsons @steven
23:08 allanparsons i wonder if you need to use /dev/xvda
23:08 derks and there is no pid in /var/run/salt/master/
23:08 allanparsons do you get a pid? (ps -ef|grep salt-master)
23:09 derks no
23:10 derks using upstart: "start salt-master" (with log level set to trace) ... it doesn't show anything... like upstart is fucked
23:10 derks not even trying to start it
23:10 allanparsons how did you install your salt-master?
23:10 derks via the salt PPA
23:10 derks I've done the exact installation a dozen times on ubuntu 12.04 ... its just this particular install for some reason is jacked
23:11 allanparsons derks - do you have logs in /var/log/upstart or cat /var/log/syslog | grep salt
23:11 derks i have no idea wtf is going on
23:11 derks allanparsons... ah, yes.  /var/log/upstart/salt-master.log
23:11 derks "salt-master" isn't allowed to be executed.
23:11 derks wtf
23:12 allanparsons find /  | grep salt-master
23:12 allanparsons and then chmod +x it
23:12 krow joined #salt
23:12 derks /usr/bin/salt-master is 755 .. like I said, I can run it manually
23:13 steven__ Worth a try, @allanparsons. I'll test it and see.
23:13 * derks is not the most familiar with upstart obviously
23:13 allanparsons upstart stinks on Ubuntu 12.04
23:13 allanparsons man, i was armwrestling w/ redis upstart the past 2 hours
23:14 derks it's driving me crazy... i've done the same exact install on the same exact cloud image ...
23:14 allanparsons finally said fuck it and moved to the init.d start script
23:14 allanparsons hey, also, the bootstrap is nicer in my opinion
23:14 dopp first thing I do on ubuntu is install sysv-rc-conf :)
23:14 manfred good thing upstart is going away
23:14 krow1 joined #salt
23:14 manfred ... eventually
23:14 dopp it's going away just as othewr distros are starting to use it? figures
23:14 manfred dopp: nope, everything is moving to systemd
23:15 manfred no one was really using it, only ubuntu
23:15 derks MOTHER FUCKER... haha... it's because on this instance we're using sudosh
23:15 manfred redhat had a basterdized version in rhel6
23:15 dopp heh, yes, just about all of the systems I manage are RHEL 6 :)
23:15 allanparsons ha
23:15 dopp I'd just as soon they left the upstart out rather than half use it
23:15 allanparsons i like ubuuntu, but man upstart is nice in theory
23:15 allanparsons but dump in execution
23:15 allanparsons dumb*
23:16 manfred dopp: but you never actually used upstart... always used init.d scripts on rhel6 anyway
23:16 steven__ Looks like xvda is hanging as well, but I'll give it a little more time.
23:16 dopp I do sort of use it, but only the built-in stuff
23:16 dopp like the upstart scripts for setting a serial console based on boot options
23:16 derks wierd though... sudosh wasn't a problem when testing on ubuntu
23:16 derks i mean centos
23:17 steven__ Of course, it made a liar out of me. We'll see if it builds right. ;)
23:17 * dopp would prefer that RHEL just use LSB-format init scripts like SLES
23:18 frasergr_ joined #salt
23:19 allanparsons @steven__ so /dev/xvda seems to be doing the trick?
23:19 steven__ That's really odd. It ran, but it didn't resize.
23:19 derks yeah, that was it... just had to modify sudosh config
23:19 steven__ @allanparsons It booted into 8G, and a resize2fs gives "The filesystem is already 2097152 blocks long.  Nothing to do!"
23:21 thayne joined #salt
23:21 allanparsons hm
23:21 allanparsons im out of ideas then :(
23:21 steven__ I know. Seven different kinds of strange.
23:21 steven__ But I think a little post-install script with a mkfs /dev/sdb, mount /dev/sdb <filesystem> will give the engineering team the space they need.
23:21 allanparsons what size node?
23:22 steven__ m1.large. Which, if I am reading Amazon's lousy docs correctly, should have 400+ GB to *spare*.
23:23 allanparsons yeah, but that's ephimerial
23:23 allanparsons we use that as swap here
23:23 allanparsons i dont know if you can use that as a block device.
23:23 allanparsons this may come in handy:  http://pastebin.com/bYVFVe0B
23:23 steven__ Hm. Save that I can build filesystems on it when not /dev/sda -- I built one on /dev/sdb.
23:24 frasergraham joined #salt
23:24 xmltok my 2014.1.0 minion keeps crashing, it wont run. http://pastebin.com/sTxCyBt0 has anyone else seen this exception?
23:25 che-arne joined #salt
23:25 steven__ Very useful-looking. Thank you!
23:25 rojem joined #salt
23:25 kballou joined #salt
23:25 allanparsons and here's a guide t use that free ephimerial storage:  http://serverfault.com/questions/521790/how-can-i-create-a-swap-partion-on-amazon-ec2-with-ephemeral-storage
23:25 derks left #salt
23:26 mgw joined #salt
23:27 FunkyBob left #salt
23:28 CeBe joined #salt
23:29 xmltok i believe its due to https://github.com/saltstack/salt/commit/64802e207a8feefe7bdb39351463c96db5647bf7 which removes the StopIteration catch
23:30 frasergraham joined #salt
23:32 xmltok nm, if i pull down that version i am good, its just 2014.1.0 that is broken
23:35 faldridg_ joined #salt
23:37 Kizano joined #salt
23:39 DaveQB_ joined #salt
23:41 naiyte_ joined #salt
23:41 amckinley1 joined #salt
23:42 codekoala-simann joined #salt
23:42 codekoala-simann joined #salt
23:42 linjan_ joined #salt
23:44 eqe joined #salt
23:45 KyleG joined #salt
23:45 KyleG joined #salt
23:45 nliadm joined #salt
23:45 frasergraham joined #salt
23:45 babilen joined #salt

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