Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2016-11-30

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

All times shown according to UTC.

Time Nick Message
00:08 amontalban joined #salt
00:12 pcdummy hehe, we all hate AD :)
00:18 N-Mi_ joined #salt
00:24 rpb joined #salt
00:29 jas02 joined #salt
00:30 Rumbles joined #salt
00:34 lorengordon joined #salt
00:35 sp0097 How do I obtain a value from a pillar from a custom grain?
00:35 sp0097 I was using __pillar__.get('user_accounts')
00:35 sp0097 I get None
00:36 dxiri joined #salt
00:39 whytewolf I'm not sure pillars are accessable at all in custom grains. I have never heard of it working.
00:39 hemebond Would that make sense?
00:39 UtahDave grains are evaluated before pillar is available.  So __pillar__ is not available within grains
00:41 sp0097 So if I want to have both accessible in a python module, I should use an execution module instead?
00:42 whytewolf or just use the already existing exacution module. pillar.get
00:43 whytewolf https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.pillar.html#salt.modules.pillar.get
00:46 sp0097 yeah, that makes sense.  in an execution module, I would use pillar.get to get at data point.
00:47 sp0097 if I am understanding correctly, custom grains cannot access a pillar item because it isn’t available yet.
00:48 whytewolf that is correct
00:48 sp0097 k, thank you for helping me understand.  :)
00:49 onlyanegg joined #salt
00:51 jas02 joined #salt
00:53 PerilousApricot joined #salt
01:02 eichiro joined #salt
01:04 nickabbey joined #salt
01:16 _KaszpiR_ joined #salt
01:25 btorch joined #salt
01:25 btorch do anyone here mess with salt and windows servers ?
01:27 amontalban joined #salt
01:29 jas02 joined #salt
01:31 pcdummy I would love to hear more about that.
01:31 btorch hehe salt + windows ?
01:31 pcdummy yes :)
01:31 espacio_grande has anyone successfully run an HA saltmaster setup behind an AWS autoscaling group?
01:32 btorch yeah me too, I really want to use salt for windows managing but haven't found much on it
01:32 pcdummy espacio_grande: someone had troubles today, someone recommended to use the TCP transport instead of ZeroMQ.
01:32 pcdummy btorch: but you aren't from Austria, right? :)
01:32 btorch hehe no
01:32 btorch brasil
01:32 UtahDave What do you want to know about Salt on Windows?  It's awesome!
01:33 pcdummy UtahDave:  :)
01:33 btorch UtahDave: hehe idk, I'm still trying to look into it .. ubuntu master and windows server minions
01:34 btorch just really wanted to know what states/modules I could actually use but I guess I'll find out at some point
01:34 UtahDave You can manage files, users.  I wrote a salt package manager for windows.
01:34 espacio_grande @pcdummy hadn't thought of that. thanks
01:34 btorch was just curious right now if anyone had tried anything
01:34 UtahDave There are a bunch of windows specific modules and a lot of the regular states and modules work on Windows
01:35 btorch UtahDave: ah cool, yeah I think it would be mainly things like setup configs (like php, nodejs, custom app confs...) also download packages from dropbox or perhaps some other location
01:37 btorch that's all I can think of right now, still looking into the manual steps of this windows setup
01:42 btorch https://docs.saltstack.com/en/latest/topics/windows/ ?
01:45 UtahDave Yeah, that's regarding the windows package manager.
01:49 dxiri joined #salt
01:50 zer0def joined #salt
01:57 amontalban joined #salt
01:57 amontalban joined #salt
01:58 justanotheruser joined #salt
02:00 Joe630 joined #salt
02:01 Joe630 anyone I can bribe to look at https://github.com/saltstack/salt/issues/37981?
02:01 saltstackbot [#37981][OPEN] Salt-cloud ec2 vpc securitygroupid always returning default | Description of Issue/Question...
02:01 Joe630 looks like security groups arent getting assigned in AWS EC2 in 2016.11.0
02:01 hemebond EC2 is broken in places in 2016.11
02:02 hemebond I would stay away from it if you're using EC2.
02:02 Joe630 I accidently upgraded.
02:02 gtmanfred Joe630: i will be triaging issues withtout a milestone in the morning, i will try to get to that one early
02:03 hemebond I upgraded my master briefly to 2016.11rc2 to test and found it unusable for me. Will be waiting for next release I think.
02:03 gtmanfred please make sure you report issues that make it unusable so we can get them fixed
02:03 hemebond It was already reported.
02:04 hemebond I added a comment that it was still a problem in rc2
02:04 gtmanfred awseome
02:04 hemebond Reported in May
02:04 gtmanfred link?
02:04 hemebond https://github.com/saltstack/salt/issues/33194
02:04 saltstackbot [#33194][OPEN] salt-cloud: EC2 userdata template error | Description of Issue/Question...
02:06 gtmanfred i poked techhat to take a look tomorro
02:06 hemebond 👍  thank you muchly
02:06 raspado in salt-cloud before we used to be able to terminate instances
02:06 hemebond I looked at it but couldn't figure out how compile_template was supposed to work or be used; no comments in the source (from memory)
02:07 raspado then quickly recreate (using ec2) but now with new salt, we cannot create an ec2 instance because the host is terminated
02:07 raspado is this a new feature between 2015 release and now?
02:07 hemebond The host is terminated?
02:07 raspado yeah so if we terminate a host in ec2
02:07 hemebond You terminated AWS?
02:07 hemebond How are you terminating?
02:07 gtmanfred ahh, it should add -DELETED to the name in ec2 right?
02:08 hemebond Yes, it should be renaming
02:08 gtmanfred if you delete it with salt*
02:08 hemebond If you used salt-cloud to terminate
02:08 raspado and try to recreate a host via salt cloud, salt-cloud complains that it cannot create it because the host is either present or terminated
02:08 gtmanfred yes
02:08 gtmanfred raspado: do you delete it using salt-cloud or using the gui/awscli?
02:08 raspado aws/cli
02:08 hemebond Rename it first
02:08 raspado or GUI rather in aws
02:08 gtmanfred rename the server first
02:08 hemebond or use salt-cloud; easier :-)
02:09 gtmanfred that is what salt-cloud does, it just adds -DELETED to the name tag
02:09 hemebond Doesn't it add a little garbage too?
02:09 raspado so rename the host in ui then terminate or use salt-cloud delete option
02:09 gtmanfred yes
02:09 hemebond Correct.
02:09 gtmanfred hemebond: i am not sure, i thought it just tacked on DELETED
02:09 hemebond it does at least that. I thought it added a little random bit too.
02:10 hemebond It adds, e.g., -DEL874df286c85345fcb73e1dae12fab2de
02:11 hemebond No risk of conflicts with all that hanging on.
02:11 gtmanfred yar
02:11 raspado is it salt-cloud -d <instance_name>
02:11 gtmanfred yes
02:11 gtmanfred hemebond: i wouldn't be surprised if it added a random string, that would be smart, but we aren't always smart and think out all the steps
02:12 hemebond :-D
02:12 gtmanfred hemebond: i got bored over the weekend https://github.com/saltstack/salt/pull/37913
02:12 saltstackbot [#37913][MERGED] Add ircbot engine for sending events to the reactor | What does this PR do?...
02:13 hemebond I saw that! Very neat :-D
02:19 pcdummy gtmanfred: that one is nice, especialy for C&C servers :)
02:19 pcdummy A trojan that install Salt to connect to its C&C server :)
02:20 gtmanfred :)
02:21 catpiggest joined #salt
02:25 Bryson joined #salt
02:27 prg3 joined #salt
02:30 jas02 joined #salt
02:35 beardedeagle joined #salt
02:37 nicksloan joined #salt
02:40 sebastian-w joined #salt
02:46 st8less joined #salt
02:46 raspado is this a default grain from salt? "grains['domain']"
02:46 hemebond Seems to be.
02:46 raspado how does salt determine the domain?
02:46 raspado turning out empty on the minions for some reason
02:48 ilbot3 joined #salt
02:48 Topic for #salt is now Welcome to #salt! <+> Latest Versions: 2016.3.4, 2016.11.0 <+> Support: https://www.saltstack.com/support/ <+> Logs: http://irclog.perlgeek.de/salt/ <+> Paste: https://gist.github.com/ (please don't multiline paste into channel) <+> See also: #salt-devel, #salt-offtopic <+> Ask with patience as we are volunteers and may not have immediate answers
02:52 raspado hmm
02:53 raspado wait a minute....
02:53 raspado does salt have an issue with multiple search domains?
02:54 hemebond Issue?
02:54 raspado is that how it determines the domain
02:55 raspado i guess ill find out right now
02:55 hemebond My domain grain is empty.
02:55 hemebond My fqdn is just the hostname.
02:55 hemebond So I'm not sure.
02:55 hemebond If I had to guess, I would say it just takes the fqdn and removes the hostname.
02:59 chamunks This stuff doesn't seem to have official documentation.  https://caktus.github.io/developer-documentation/margarita/custom_states.html
02:59 chamunks I searched the salt docs and didn't find any references to ufw.[allow|enabled|default]
03:00 chamunks http://hastebin.com/fasezomuwo.sql
03:00 chamunks ^ errors.
03:01 chamunks I would do more trial and error but now I'm mucking about with firewall rules I'm really not in a rush to lock myself out.
03:01 raspado hemebond:  yeah i think your right
03:01 raspado is there a way I can see what minion the host is talking to
03:01 raspado from the minion itself
03:02 raspado need to see if the minion is talking to the master by IP or hostname
03:02 hemebond Well, it'll be in the config.
03:02 hemebond The minion will resolve the name to an IP and use that IP.
03:03 chamunks ^
03:03 raspado ah ok
03:05 chamunks Reason: 'ufw.enabled' is not available.
03:08 onlyanegg joined #salt
03:17 chadhs joined #salt
03:19 cliluw joined #salt
03:21 GothAck joined #salt
03:23 evle joined #salt
03:31 VR-Jack-H joined #salt
03:33 jas02 joined #salt
03:42 Zaunei joined #salt
03:43 athaller_ joined #salt
03:46 sebastian-w joined #salt
03:46 bastiandg joined #salt
03:46 zz_capri joined #salt
03:50 zzuf joined #salt
03:51 cyborg-one joined #salt
03:51 dlloyd joined #salt
03:52 raspado hmm so does the grain['domain'] get it from /etc/hostname ?
03:53 hemebond I think it probably uses the hostname command.
03:54 chamunks I suppose there is no ufw module which makes me wonder why that person documented modules that don't seem to exist or work.
03:54 hemebond What is Margarita?
03:55 hemebond Have you installed Margarita?
03:56 andi- joined #salt
03:59 hemebond I'm guessing you haven't if you're getting an error about those states not existing.
03:59 chamunks hemebond https://github.com/jessepeterson/margarita?
03:59 hemebond Yeah
03:59 hemebond no
03:59 hemebond https://github.com/caktus/margarita/
04:00 chamunks Y'know this is the #salt channel right?
04:00 hemebond Yeah
04:00 chamunks ah yours makes more sense.
04:00 hemebond And you're reading documentation on custom Salt modules.
04:01 chamunks hemebond <3 tyvm
04:01 hemebond 👍 Good luck
04:01 chamunks Hopefully I can manage with the docs.
04:01 dxiri joined #salt
04:03 sp0097 joined #salt
04:04 chamunks hemebond do you happen to know where the documentation on how to use this may be?
04:04 hemebond You linked to it
04:04 hemebond https://caktus.github.io/developer-documentation/margarita/margarita.html
04:04 chamunks lol welp
04:04 chamunks The more you know
04:07 karlthane joined #salt
04:09 karlthane joined #salt
04:10 dxiri joined #salt
04:21 raspado anyone sen this issue before?
04:21 raspado when provisioning a host with salt-cloud, it just hangs at "[DEBUG   ] LazyLoaded cloud.cache_node_list "
04:22 writteno1 joined #salt
04:22 karlthane joined #salt
04:33 jas02 joined #salt
04:37 raspado i guess it timed out
04:43 sp0097 joined #salt
04:52 rdas joined #salt
05:17 rem5 joined #salt
05:28 MTecknology finally bumped up to 2016.11.0... took me long enough
05:28 DEger joined #salt
05:33 dxiri joined #salt
05:35 jas02 joined #salt
05:39 keimlink joined #salt
05:43 DEger joined #salt
05:43 preludedrew joined #salt
05:47 lorengordon joined #salt
05:47 Sammichmaker joined #salt
05:52 jas02 joined #salt
05:58 jas02 joined #salt
06:01 lazzurs joined #salt
06:02 ivanjaros joined #salt
06:10 mosen joined #salt
06:14 netcho_ joined #salt
06:34 netcho_ joined #salt
06:36 jas02 joined #salt
06:43 netcho joined #salt
06:47 PerilousApricot joined #salt
07:00 netcho joined #salt
07:01 jas02 joined #salt
07:01 jhauser joined #salt
07:03 teclator joined #salt
07:08 Rumbles joined #salt
07:12 bocaneri joined #salt
07:21 hoonetorg joined #salt
07:23 nidr0x joined #salt
07:23 ivanjaros joined #salt
07:24 teclator joined #salt
07:24 hoonetorg joined #salt
07:27 Rumbles joined #salt
07:27 sjorge joined #salt
07:27 sjorge joined #salt
07:29 DEger joined #salt
07:36 jas02 joined #salt
07:49 jas02 joined #salt
07:52 teclator joined #salt
08:00 jas02 joined #salt
08:00 teclator joined #salt
08:06 felskrone joined #salt
08:08 rofl_____ joined #salt
08:09 rofl_____ is custom grains modules supposed to work with salt-ssh?
08:09 rofl_____ trying to get https://raw.githubusercontent.com/saltstack/salt-contrib/master/grains/digitalocean_metadata.py working
08:09 rofl_____ but i get no grains
08:10 AndreasLutro that's not a very good custom grain
08:10 AndreasLutro don't think requests is available to salt-ssh by default so you'd have to install it system-wide to make that work
08:10 rofl_____ that may be
08:10 AndreasLutro but really just replace requests.get with urllib2
08:10 rofl_____ python-requests is available
08:12 AndreasLutro could be something else then I guess - but custom grains do work with salt-ssh, at leat every time I tested
08:12 rofl_____ how can i debug the blob salt-ssh sends?
08:12 rofl_____ dont see it in /tmp
08:13 rofl_____ ah its put in /var/tmp now
08:14 Talkless joined #salt
08:15 Talkless Hi, is there a state for simply managing file permissions (Linux) without copying/uploading file itself? webutil.user_exists state implicitly creates htpasswd file, but how do I set specific permissions on int?
08:15 rofl_____ AndreasLutro: cant locate my custom grains module in the blob either
08:18 ernescz joined #salt
08:19 rofl_____ AndreasLutro: i deleted the dir in /var/tmp and then it appeared
08:20 AndreasLutro odd. just salt-ssh things...
08:21 alexzel joined #salt
08:23 alexzel hello, I've been playing around with dockerng.sls for testing salt states on docker containers, problem is, when I apply a state that is dependent on map.jinja (using grains.filter_by), but that means map.jinja will always get the OS of the docker host, and I have multiple containers with different distributions
08:24 alexzel Is there any way to force/overwrite this to match the container OS?
08:24 ernescz Hello! Is there is way to see exactly how a state file looks like after Jinja has done its magic on it? Not like state.show_sls but a "preview" if you will, of how it would look like?
08:26 nfahldieck joined #salt
08:28 nfahldieck Hello. :) I'm looking for ways and projects of salt GUIs. So far I tried saltpad, Foreman and unfortunately SaltStack Enterprise is no option for me. Right now I'm putting my hands on SaltShaker. Do you have any other suggestions?
08:30 ronnix joined #salt
08:31 fracklen joined #salt
08:32 fredvd joined #salt
08:34 dariusjs joined #salt
08:34 jhauser joined #salt
08:37 jas02_ joined #salt
08:37 ReV013 joined #salt
08:40 ernescz heh, nevermind  - state.show_low_sls will have to cut it for now :)
08:44 samodid joined #salt
08:48 o1e9 joined #salt
09:02 saffe joined #salt
09:08 mavhq joined #salt
09:08 mikecmpbll joined #salt
09:09 saffe_ joined #salt
09:12 keimlink joined #salt
09:12 s_kunk joined #salt
09:17 ReV013 joined #salt
09:21 netcho joined #salt
09:23 PhilA__ joined #salt
09:23 Rumbles joined #salt
09:24 saffe joined #salt
09:26 perun_ joined #salt
09:30 ozux joined #salt
09:31 TyrfingMjolnir joined #salt
09:38 jas02_ joined #salt
09:38 N-Mi_ joined #salt
09:39 ronnix joined #salt
09:45 Lionel_Debroux joined #salt
09:52 Mattch joined #salt
09:58 orichards joined #salt
09:58 Yog joined #salt
10:00 ronnix_ joined #salt
10:01 amcorreia joined #salt
10:01 dariusjs joined #salt
10:03 orichards joined #salt
10:03 netcho joined #salt
10:10 * mede hi all
10:10 mede guys what is the correct or recommended procedure to upgrade salt-minions to the latest version?
10:11 mede can i do a salt '*' cmd.run 'yum update -y && yum upgrade salt-minion' .. ?  or smth similar?
10:12 fracklen joined #salt
10:12 rubenb IIRC, this will not return a correct response, because the update kills the salt-minion.
10:12 rubenb There where solutions with scheduling the minion install with 'at', but I have no idea what's the current recommended method.
10:13 usernkey joined #salt
10:13 rubenb Also, (at least on debian), salt-common needs to be updated too.
10:15 dynamicudpate joined #salt
10:16 AndreasLutro in recent versions on debian versions with systemd, pkg.upgrade salt-minion will work fine
10:17 mauli joined #salt
10:17 pipps joined #salt
10:18 writtenoff joined #salt
10:19 WKNiGHT_ joined #salt
10:19 k_sze[work] joined #salt
10:20 ronnix joined #salt
10:20 CeBe rubenb: that should be done by apt automatically if you update the minion package.
10:20 CheckYourSix joined #salt
10:20 CruX__ joined #salt
10:20 mrud joined #salt
10:20 mrud joined #salt
10:20 jrklein joined #salt
10:20 descrepes joined #salt
10:20 _KaszpiR_ joined #salt
10:20 fhh joined #salt
10:20 monrad joined #salt
10:21 Laogeodritt joined #salt
10:21 LotR joined #salt
10:21 stiefkind joined #salt
10:21 systeem joined #salt
10:22 mTeK joined #salt
10:22 gmacon_ joined #salt
10:23 antonw joined #salt
10:23 subohm joined #salt
10:24 rideh joined #salt
10:24 FreeSpencer joined #salt
10:24 FreeSpencer joined #salt
10:24 TooLmaN joined #salt
10:24 hoonetorg joined #salt
10:24 kshlm joined #salt
10:24 Nebraskka joined #salt
10:24 TheBigRedButton joined #salt
10:24 devster31 joined #salt
10:24 kutenai joined #salt
10:24 __alex joined #salt
10:25 zifnab06 joined #salt
10:25 berto- joined #salt
10:25 skrobul joined #salt
10:25 saltstackbot joined #salt
10:26 georgemarshall joined #salt
10:27 valkyr2e joined #salt
10:27 inire joined #salt
10:27 bryguy joined #salt
10:27 shalkie joined #salt
10:27 colegatron_origi joined #salt
10:29 Horgix joined #salt
10:31 rdas joined #salt
10:31 oida joined #salt
10:31 cscf joined #salt
10:31 DK2 joined #salt
10:31 krobertson joined #salt
10:32 netcho joined #salt
10:32 mattl joined #salt
10:32 pipps joined #salt
10:32 coredumb joined #salt
10:33 jhauser joined #salt
10:36 Yog joined #salt
10:43 Guest1345 joined #salt
10:44 pipps joined #salt
10:44 netcho joined #salt
10:47 ScottL joined #salt
10:47 ScottL anyone up?
10:48 zer0def joined #salt
10:52 pipps joined #salt
10:53 babilen ScottL: Just ask .. (unless you really are interested in how many people in this channel are "up" ;)
10:54 netcho joined #salt
10:54 ScottL haha good point. :)
10:55 ScottL I'm a super noob when it comes to Dev Ops, still trying to figure out the feature set of Salt Stack
10:56 babilen You'll do that for some time to come :)
10:56 ScottL Haha
10:56 babilen (and they peskily add new features every release)
10:56 ScottL Anyway, I don't want to spend a month learning it and figure out it's not going to do what I imagine is the primary reason I'd want to use it at all.
10:58 Cadmus joined #salt
10:58 ScottL I'm looking at setting up a multi datacenter deployment with each datacenter containing two kinds of servers: Application servers that run through a node balancer and database servers that use cassandra.
10:58 ScottL So the question is, can Salt be configured to automatically provision new servers as needed based on variables like CPU/Memory load on the existing servers?  It would need to do it based on datacenter and node type(application or database)
11:00 AndreasLutro salt won't create the servers in aws or whatever for you if that's what you're asking
11:01 ronnix_ joined #salt
11:01 ScottL Does anyone know of a dev ops solution that will?
11:02 abednarik joined #salt
11:02 AndreasLutro AWS ASG are the closest thing to a pre-built solution to that problem
11:02 ScottL I mean, it feels like a pretty simple ask... watch the CPU/Memory usage of a collection of servers and if they go above a certain threshold for a certain amount of time, use a bash script or something similar to provision new servers until the load is restored to a normal one.
11:02 AndreasLutro otherwise you'll just have to add some hooks in your alerting system to run scripts that create the servers
11:02 mikecmpbll joined #salt
11:02 ScottL wait...so that's possible?
11:02 AndreasLutro you could use salt to run those scripts but it probably doesn't make a difference
11:03 AndreasLutro yes but not with salt alone
11:04 ScottL Thanks for your help by the way... so Salt IS capable of provisioning servers without my involvement?  I just need to configure a custom hook to do it?
11:04 ScottL I'm fine if it's outside the scope of a point and click interface...
11:05 AndreasLutro like I said you can set up salt to run scripts to create servers, or maybe you can use the boto3 states/modules
11:05 Cadmus Hello, I've got a file.managed state which is giving the error message "Unable to manage file 'hash_type' ", what can cause that?
11:05 AndreasLutro but it'll be functionally the same as a bash script that creates your servers
11:06 ScottL Ah makes sense..
11:07 ScottL Ok, thanks for answering my questions. :)
11:07 AndreasLutro "devops" is never about a single tool that does all your work for you. you have to string things together yourself
11:07 Cadmus I appreciate that's a pretty meaningless error message. Oddly there is no has_type for this file, it's dynamically generated so I've got "skip_verify: True" in there.
11:08 flebel joined #salt
11:08 ScottL Yeah, I'm starting to realize that! :D
11:08 ronnix joined #salt
11:08 AndreasLutro Cadmus, do you get a stack trace?
11:09 Cadmus AndreasLutro: It's being called by Jenkins using cherry-api, but thinking about that has given me a place to look.
11:15 ozux joined #salt
11:16 sp0097 joined #salt
11:17 ozux__ joined #salt
11:19 PerilousApricot joined #salt
11:20 Cadmus I think my first bet will just be update everything and see if there's a version mis-match anywehre, it feels like a parsing error of some sort because if I copy-paste it and run it from the master it seems much happier...
11:22 impi joined #salt
11:27 inad922 joined #salt
11:31 amontalban joined #salt
11:31 amontalban joined #salt
11:34 Jonax joined #salt
11:34 jesusaur joined #salt
11:43 Talkless What state could I use to set  Linux file permissons on existing file?
11:44 AndreasLutro file.managed
11:45 jas02_ joined #salt
11:48 bluenemo joined #salt
11:49 samodid joined #salt
12:00 jas02 joined #salt
12:01 pipps joined #salt
12:03 netcho_ joined #salt
12:04 jas02 joined #salt
12:05 justanotheruser joined #salt
12:07 justanotheruser joined #salt
12:08 fracklen joined #salt
12:10 fracklen joined #salt
12:10 dariusjs joined #salt
12:11 justanotheruser joined #salt
12:13 Talkless file.manages needs source? i do not want to upload/copy file, just update existing one
12:14 Talkless AndreasLutro: webutil.user_exists creates file, I jus need to fix it's permissions.
12:14 AndreasLutro no, file.managed does not need source
12:14 Talkless oh
12:15 sp0097 left #salt
12:15 Talkless mind blown, thanks! :)
12:15 netcho joined #salt
12:17 fizmat joined #salt
12:18 pipps joined #salt
12:20 PerilousApricot joined #salt
12:21 jas02 joined #salt
12:24 pipps joined #salt
12:29 rofl_____ how can i make salt execute a command if a iptables.append function is changed?
12:29 rofl_____ trying to think of some watch_in logic
12:29 rofl_____ but its not a service that is run
12:29 rofl_____ only a cmd
12:29 AndreasLutro onchanges
12:30 rofl_____ does onchanges support a cmd.run ?
12:31 AndreasLutro onchanges doesn't "support" anything, it just does things
12:32 AndreasLutro just like require
12:32 rofl_____ i dont want a cmd.run with 100 x onchanges rules
12:32 rofl_____ id rather have a onchanges_in
12:32 AndreasLutro use onchanges_in then
12:35 PhilA__ joined #salt
12:37 bluenemo hi guys. I'm getting "local variable 'repo_obj' referenced before assignment" when I try to service salt-master start on master 2016.11.0 carbon: http://paste.debian.net/hidden/eefffcad/
12:41 abednarik joined #salt
12:41 jas02_ joined #salt
12:43 bluenemo Ref https://github.com/saltstack/salt/issues/37980
12:43 saltstackbot [#37980][OPEN] Having 'git' in fileserver_backends and no gitfs_remotes defined causes a crash | ...
12:44 pipps joined #salt
12:45 numkem joined #salt
12:45 bluenemo imho this should be high prio, as it will mess up many reactor setups
12:46 jas02 joined #salt
12:47 dariusjs joined #salt
12:47 pezus hi
12:47 pezus i am trying to match host in an sls file with a substring of a grain but it doesn't seem to match:
12:47 pezus {% if "Proliant" in grains["productname"] %}
12:48 pezus am i using it wrong?
12:48 jas02__ joined #salt
12:50 jhauser joined #salt
12:50 telx joined #salt
12:51 pezus yes, i am
12:51 pezus the matching is case-sensitive. nothing to see here, go on guys
12:52 bluenemo lol ;)
12:53 bluenemo pezus, there is an upper and a lower jinja filter as far as I remember
12:54 bluenemo wow nifty grain. didnt know that one
12:55 jas02 joined #salt
12:55 pezus it's an awesome grain as these servers have some specifics that only apply to them
12:55 pezus i love grain-matching a lot
12:58 ozux joined #salt
12:58 ozux__ joined #salt
13:02 rofl_____ AndreasLutro: thx, worked :)
13:09 felskrone joined #salt
13:10 ozux joined #salt
13:10 PerilousApricot joined #salt
13:11 aidin joined #salt
13:14 mikecmpbll is there a simple way that I can include a pillar file per node
13:14 mikecmpbll without having to do node1:\  - node1\node2:\  -node2 etc in my pillar topfile
13:14 bluenemo mikecmpbll, use sth like {% if grains['id'] == 'node.example.com' %}
13:15 mikecmpbll oh snap, didn't realise i could use jinja in pillar
13:15 bluenemo you can also use sth like: include: - foo.bar.{{ grains['id'] }} if you like - however grains['id'] might have dots in it, not sure how that behaves then
13:15 bluenemo you could use a different renderer just for that file if thats a problme
13:16 mikecmpbll bluenemo : that's neat 👌🏼
13:18 bluenemo mikecmpbll, if you are new to salt I recommend to read the public saltstack formulas (or those from arnold bechtold preferably) around 30 minutes to an hour per day. Its the best way to get your started with salt imho.
13:18 bluenemo there is also a book about salt, which sbd in this channel wrote, you should check it out if you want an 8 hour spanning full introduction.
13:27 mikecmpb_ joined #salt
13:36 mikecmpbll joined #salt
13:37 jas02 joined #salt
13:37 PerilousApricot joined #salt
13:45 debian112 joined #salt
13:47 jas02_ joined #salt
13:49 Satyajit joined #salt
13:50 jas02 joined #salt
13:51 martoss joined #salt
13:55 lilvim joined #salt
13:56 ronnix joined #salt
14:01 amontalban joined #salt
14:01 amontalban joined #salt
14:02 zer0def when using a `http.query` state with `decode: true` and `decode_type: json`, is there any way to probe the returned dict for particular keyvals?
14:03 lorengordon joined #salt
14:03 babilen zer0def: Why aren't you calling the execution function if you want to work with the data?
14:06 krymzon joined #salt
14:07 Rumbles joined #salt
14:08 bobswarley joined #salt
14:09 ozux joined #salt
14:09 zer0def you mean call it from jinja or module.run?
14:10 babilen The former
14:10 ozux joined #salt
14:11 zer0def i guess i haven't through about it, but it depends on a few objects already existing, which are being created in states
14:12 zer0def which then creates a catch-22 scenario
14:12 zer0def at least initially.
14:12 The_Loeki joined #salt
14:12 zmalone joined #salt
14:12 zer0def i guess i could shell out in this case
14:12 ozux__ joined #salt
14:15 AndreasLutro maybe time to consider writing your own state function
14:15 DEger joined #salt
14:16 ozux joined #salt
14:18 zer0def i might
14:19 ssplatt joined #salt
14:22 ozux joined #salt
14:22 dxiri joined #salt
14:24 ozux__ joined #salt
14:26 nickabbey joined #salt
14:27 ozux joined #salt
14:27 ronnix_ joined #salt
14:28 Cadmus Hmm, upgrading seems to break all my gitfs stuff for pillars, so I'm back to scratching my head over this
14:29 dxiri_ joined #salt
14:29 austin_ joined #salt
14:33 gableroux joined #salt
14:34 pipps joined #salt
14:34 Cadmus Ah, got some logs from the master "Received function _file_find which is unavailable on the master, returning False"
14:36 babilen .oO( Why is every major .0 release broken? )
14:37 dxiri joined #salt
14:42 mede hi guys... i have a state running  on a minion BUT is returning a false positive (i think). When i lookup_jid i get an answer: The function "state.sls" is running as PID 22581 and was started at 2016, Nov 30 15:03:08.503391 with jid 20161130150308503391   .... BUT on the minion that PID is not present
14:42 mede any ideas?
14:44 mede i have the latest version of SALT
14:48 jas02_ joined #salt
14:48 pjy joined #salt
14:49 dxiri joined #salt
14:52 racooper joined #salt
14:54 fracklen joined #salt
14:55 pipps joined #salt
14:59 cscf mede, is that PID on the master?
15:00 mede no ... minion
15:01 pipps joined #salt
15:03 rem5 joined #salt
15:09 nicksloan joined #salt
15:13 pipps joined #salt
15:15 ronnix joined #salt
15:17 jas02 joined #salt
15:17 _aeris_ left #salt
15:21 Brew joined #salt
15:22 sarcasticadmin joined #salt
15:25 pipps joined #salt
15:29 akoumjian joined #salt
15:32 pipps joined #salt
15:34 _JZ_ joined #salt
15:35 Cadmus I don't know if that error makes sense, the one that's killing me is this "Unable to manage file: 'hash_type'",
15:35 Cadmus But I have no idea what could provoke that
15:36 gtmanfred https://github.com/saltstack/salt/issues/37941
15:36 saltstackbot [#37941][OPEN] Outdated documentation for 2016.11.x | Description of Issue/Question...
15:37 numkem joined #salt
15:37 fracklen joined #salt
15:40 pipps joined #salt
15:40 mdhas joined #salt
15:43 raspado joined #salt
15:46 raspado anyon know how the grain['domain'] is added by salt?
15:46 raspado doesnt seem to be loaded when we have multiple search domains on instance boot up
15:48 abednarik joined #salt
15:48 dxiri joined #salt
15:49 jas02_ joined #salt
15:51 _Cyclone_ joined #salt
15:55 JPT joined #salt
15:55 johnkeates joined #salt
15:56 dxiri joined #salt
15:58 Tanta joined #salt
15:59 tristianc_ joined #salt
15:59 coredumb Hey folks
16:00 coredumb may I ask how you folks are delegating rights to use your salt instance ?
16:00 coredumb like allowing teams to write rules for specific groups of servers and the like
16:00 PerilousApricot joined #salt
16:01 pipps joined #salt
16:01 Cadmus coredumb: We're using gitfs as a backend, so people can make pull requests and we check for side-effects
16:01 coredumb I'm feeling something with git could be made how would one ensure that incorrect matching couldn't be made/overwritten in top.sls
16:01 coredumb Cadmus: yeah you still have to check then
16:05 Cadmus coredumb: top is in a different repo (seems to be the recommended way becuase of branching). To be honest it's not come up yet because no-one in any other departments gives a damn :)
16:06 coredumb Cadmus: iirc top is merged from top to down right ?
16:06 Cadmus I think so, but don't quote me
16:06 dxiri joined #salt
16:06 coredumb If I want to let some ppl add their own new files they would need to be able to modify their own top
16:07 coredumb think I would need to write some heavy hooks to prevent major problems ...
16:12 gtmanfred raspado: it is set here https://github.com/saltstack/salt/blob/develop/salt/grains/core.py#L1716
16:12 raspado awesome thx gtmanfred
16:12 ozux__ joined #salt
16:13 XenophonF oh great so the output of key.finger changed between 2015.5 and 2016.11, eh?
16:15 * XenophonF STFWs for the 2016.11 release notes.
16:16 scoates joined #salt
16:16 XenophonF hm, nothing there about minion key management changes...
16:17 netcho joined #salt
16:17 Demos joined #salt
16:18 XenophonF I'm not important enough to MITM, so I'm a little surprised by the key fingerprint differences here.
16:19 g3cko joined #salt
16:19 gtmanfred XenophonF: i have seen this https://github.com/saltstack/salt/issues/37991
16:19 saltstackbot [#37991][OPEN] Invalid master key | Description of Issue/Question...
16:20 XenophonF how might i verify this out of band, without using Salt?
16:20 gtmanfred i do not know
16:20 raspado is there a way to trigger this on a minion? "salt.utils.network.get_fqhostname()"
16:21 XenophonF ah, here we go
16:21 XenophonF i'll just compare the two certificates
16:21 XenophonF /etc/salt/pki/minion
16:22 johnkeates joined #salt
16:22 XenophonF er, i mean /etc/salt/pki/minion/minion.pub on the minion and /etc/salt/pki/master/minion_pre/...minionid... on the master
16:22 gtmanfred raspado:create a module that runs the command and returns the output and put it in /srv/salt/_modules
16:23 tiwula joined #salt
16:23 raspado ah k thx
16:23 gtmanfred raspado: actually, you might be able to just do like
16:24 gtmanfred def get_fqhostname():
16:24 gtmanfred return __utils__['network.get_fqhostname']()
16:24 gtmanfred and call the <filename>.get_fqhostname
16:24 gtmanfred but you still need to make a module to call it
16:24 raspado very nice thx!
16:25 raspado soemthing really weird is going on when ec2 adds two search domains
16:25 raspado the grain is empty and mucks up our network adapter
16:25 raspado or no search domain at all for that matter
16:26 aidin joined #salt
16:27 jorian joined #salt
16:27 XenophonF well i diffed the certificates and they're identical
16:28 XenophonF why would 'salt-call --local key.finger' (on the minion) and 'salt-key -f ...' (on the master) have different output?
16:29 tobiasBora joined #salt
16:33 fracklen joined #salt
16:34 onlyanegg joined #salt
16:34 abednarik joined #salt
16:37 PerilousApricot joined #salt
16:45 Sylvain31 joined #salt
16:46 dxiri joined #salt
16:49 pipps joined #salt
16:50 jas02_ joined #salt
16:52 anotherzero joined #salt
16:56 dyasny_ joined #salt
17:00 Bryson joined #salt
17:02 jas02 joined #salt
17:03 jas02_ joined #salt
17:05 pipps joined #salt
17:06 beowuff joined #salt
17:10 orionx joined #salt
17:12 cyborg-one joined #salt
17:28 NeoXiD joined #salt
17:29 CampusD joined #salt
17:29 whytewolf XenophonF: odd. I just thought i would check on my system both salt-call key.finger and salt-key -f match. but both are running 2016.11 are you using different versions?
17:31 CampusD hi all, question, is there a way to make the required state run first before the state doing the requiring?
17:31 cscf CampusD, - require already does that
17:32 CampusD so if state A requires B, then B will run first then A ?
17:32 cscf Yes.  That's kinda the point, actually
17:32 cscf Also if B fails, A won't run and will return failure
17:32 CampusD cool, I'll give that a try, thanks cscf
17:37 Trauma joined #salt
17:37 beowuff joined #salt
17:40 Trauma joined #salt
17:41 _Cyclone_ joined #salt
17:42 jas02 joined #salt
17:42 nidr0x joined #salt
17:43 Gareth Anyone seen anything weird with the dockerng module in the latest stable release?  Specifically it complaining about a host not being included when pulling an image?
17:49 iggy joined #salt
17:53 Jimlad joined #salt
17:54 Llmiseyhaa joined #salt
17:54 bfrog joined #salt
17:55 dxiri joined #salt
17:56 pipps joined #salt
17:56 pirulo joined #salt
17:57 pirulo Pillar issues: Specified SLS 'default' in environment 'base' is not available on the salt master
17:57 pirulo my master pillar: pillar_roots:   base:     - /srv/pillar
17:58 PerilousApricot joined #salt
17:58 pirulo my top.sls: base:   '*':     - sdc
17:59 Perilous_ joined #salt
18:00 nickabbey joined #salt
18:00 DEger joined #salt
18:01 rbrennan joined #salt
18:05 Edgan joined #salt
18:07 mohae joined #salt
18:10 Aleks3Y joined #salt
18:14 const joined #salt
18:15 const Hi all, how to debug message 'postgres_user' __virtual__ returned False?
18:15 cscf pirulo, and do you have an sls called 'default' anywhere?
18:16 mikecmpbll joined #salt
18:18 pirulo nope
18:19 whytewolf does sdc have an include?
18:19 pirulo here is my sdc.sls
18:19 samodid joined #salt
18:20 pirulo datacenter: sdc env: test
18:24 _Cyclone_ joined #salt
18:25 flebel joined #salt
18:31 Jimlad joined #salt
18:32 Brew joined #salt
18:35 infrmnt joined #salt
18:35 fracklen joined #salt
18:38 lorengordon joined #salt
18:39 djgerm joined #salt
18:41 dyasny joined #salt
18:41 const anyone knows about this '__virtual__'?
18:44 johnkeates joined #salt
18:45 whytewolf __virtual__ is a function that determines if a module should be loaded. and is used by lazy loading. if it returns false then dependencies for a module are not being met.
18:47 const I looked at postgres users state sources and found only one reason not to load - if there is no postgres module
18:47 gtmanfred const: there is a bug about that
18:48 gtmanfred const: https://github.com/saltstack/salt/issues/37935
18:48 saltstackbot [#37935][OPEN] Postgres module regression on 2016.11 | Description of Issue/Question...
18:48 const ahh, thank you
18:48 const this is exactly my case too...
18:53 nickabbey joined #salt
18:54 dxiri joined #salt
19:01 scsinutz joined #salt
19:04 ivanjaros joined #salt
19:04 Salander27 joined #salt
19:17 jas02 joined #salt
19:33 CampusD another question related to my previous, I have the following https://gist.github.com/UtahCampusD/7aaa9d8cc73bb600e3ae69bc19b04e23, the second state seems to run no matter what but I only want to run when the first will run
19:33 CampusD any ideas?
19:33 SaucyElf joined #salt
19:33 CampusD can I use "onlyif" or "unless" in this case?
19:34 SaucyElf joined #salt
19:35 s_kunk joined #salt
19:36 plinnell joined #salt
19:37 cscf CampusD, as in, you want to make a backup any time the file.managed changes something?
19:38 CampusD correct, but right now the backup state runs even when the config state doesn't
19:39 cscf CampusD, file.managed has a setting for that.  - backup: '.bak'
19:39 cscf CampusD, also, "onlyif" or "unless" run bash commands, not salt stuff
19:40 cyborg-one joined #salt
19:40 lunarlamp joined #salt
19:42 CampusD cscf, perhaps my file.managed is a bad generic example but basically I want to avoid a second state from running unless the first state calls it with the required
19:43 cscf CampusD, ah.  Then you want - onchanges
19:44 CampusD cscf, is onchanges similar to watch?
19:44 cscf CampusD, so the second state should have - onchanges: \n - first_state
19:44 CampusD cscf, I see I'll give that a try, thanks!
19:44 cscf CampusD, very similar, but where "watch" says "do extra things if", ie restarting a service, "onchanges" is "execute only if"
19:44 orionx_ joined #salt
19:52 lem2 joined #salt
19:52 lem2 left #salt
19:55 XenophonF whytewolf: i'm running 2015.8 on the master, 2016.11 on the minion, both are centos 7
19:57 gtmanfred XenophonF: ahh, salt does not support having newer minion versions than masters
19:57 gtmanfred https://docs.saltstack.com/en/latest/faq.html#can-i-run-different-versions-of-salt-on-my-master-and-minion
19:59 _Cyclone_ joined #salt
20:02 dps joined #salt
20:17 jas02 joined #salt
20:26 irctc622 joined #salt
20:26 irctc622 Hello is there a way to check for a installed pkg without having that package already managed by salt?
20:27 Sketch you can do anything with cmd.run....
20:27 irctc622 Sketch: sure but that's ugly. I meant something like if salt['file.directory_exists'](...)
20:28 gtmanfred salt-call pkg.list_pkgs <pkgname>
20:28 Sketch you can do that too, if you know what it is that you want to exist
20:28 gtmanfred https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.yumpkg.html#salt.modules.yumpkg.list_pkgs
20:28 cscf Is the gitfs cache timeout configurable?
20:28 gtmanfred you could also do unless or onlyif on 'test -d /path/to/dir'
20:30 Sketch {% if salt['file.file_exists']('/path/to/file') %}
20:30 irctc622 gtmanfred: where's the advantage of onlyif - test -d .. instead of {% if salt['file.directory_exists']('') %}
20:30 gtmanfred not having to use jinja
20:30 whytewolf onlyif keeps the state in the system. jinja renders it out of the system
20:30 irctc622 gtmanfred: i like jinja and i save a fork and an extra binary execution
20:30 gtmanfred cscf: it isn't so much a cahce timeout as how often the master updates it, and i believe that can be configured, it should update it each iteration through the event loop which i believe is configured with https://docs.saltstack.com/en/latest/ref/configuration/master.html#loop-interval
20:31 gtmanfred then use jinja
20:31 irctc622 whytewolf: sure but that's not a problem in this case
20:31 gtmanfred like how Sketch showed you
20:31 cscf gtmanfred, what are the downsides to making it shorter?  CPU usage?
20:31 gtmanfred yeah, it will do all the maintenance tasks more often
20:31 XenophonF gtmanfred: oh that's right, gotta upgrade the master before upgrading minions
20:32 gtmanfred cscf: you could also setup a reactor and salt-api to trigger a refresh on the fileserver when your git repo gets a new commit
20:32 irctc622 Sketch: yes i have it like this right now but thought it would be cleaner if i could do {% if salt['pkg.installed']('ABC') %}
20:32 gtmanfred irctc622: you can do {% if salt['pkg.list_pkgs']('ABC') %}
20:32 gtmanfred that /should/ work
20:33 irctc622 gtmanfred: thanks will try
20:33 jmastron joined #salt
20:33 gtmanfred cscf: https://docs.saltstack.com/en/latest/topics/tutorials/gitfs.html#refreshing-gitfs-upon-push
20:33 cscf gtmanfred, thanks
20:33 gtmanfred irctc622: otherwise, check the different pkg modules to find the function that works for you
20:33 cscf " provided that the git server is also a Salt minion" nope, won't work.
20:34 gtmanfred cscf: update it to fit your needs, that is just one way to do it
20:34 Sketch pkg is a state, not a module, so you'd have to know which package type you're looking at.  yumpkg.version might work
20:34 Sketch Returns a string representing the package version or an empty string if not installed.
20:34 gtmanfred yumpkg.version is not an execution module, it is pkg.version
20:34 Sketch ahh
20:34 gtmanfred because __virtualname__ in yumpkg.py is set to pkg
20:35 gtmanfred so yeah, irctc622 use pkg.version instead of pkg.list_pkgs
20:35 whytewolf what about pkg.info_installed
20:36 gtmanfred any of them work :P
20:36 whytewolf so many options :P
20:36 Brew joined #salt
20:37 irctc622 gtmanred ok will try
20:37 irctc622 gtmanfred: works perfectly
20:37 gtmanfred neat, i got libvirt working with vmwarefusion
20:41 whytewolf btw, fyi irctc622 the jinja actually might be more costly then the onlyif. the modules still run the package commands through shells
20:41 whytewolf nut has extra code added
20:42 whytewolf s/nut/but
20:45 CampusD cscf, I used the onchanges option but I get a recursive requisite now, I am assuming kind of infinite loop.  https://gist.github.com/UtahCampusD/12e280232fa5efc56c949a7c06e2a7eb
20:46 cscf CampusD, well yeah, you need to remove the - require
20:46 cscf CampusD, however, in this case, the backup will run after the file.managed, so it won't be a useful backup
20:46 cscf But in the general case, that's how you do it
20:47 cscf for file.managed, just use the - backup: '.bak' option
20:47 CampusD will the first state still run second
20:47 CampusD oh this is file.managed is just a generic example
20:47 whytewolf CampusD: onchanges is something should only be ran if something changes..... it won't know if that changes if it runs before the item it needs to check
20:47 cscf ^
20:47 jmastron What you're looking for is prereq.
20:48 jmastron It means only run this if the prereq state has changes, and make sure it runs first.
20:48 jmastron *make sure *this* runs first
20:49 jmastron https://docs.saltstack.com/en/latest/ref/states/requisites.html#prereq
20:50 whytewolf honestly, read all of the requisites page ....
20:51 CampusD will do, so I have state A with prereq for state B, if A will have changes, then B will run and then A. What if A has no changes, will B still run? I want to avoid B running unless called by A
20:53 gtmanfred If A has a prereq for B, B will be run first with test=True, if changes are going to be made, then A is run and then B is run as well.
20:53 jmastron If A has the prereq the A runs first, but only if B signals changes.
20:53 jmastron *then
20:53 jmastron So yes
20:53 gtmanfred If B does not have changes then A is not run, and then B is run, but it is already determined that B will not make changes, so it also runs and returns back without making changes
20:54 scsinutz joined #salt
20:55 scsinutz quick question… from https://docs.saltstack.com/en/latest/topics/tutorials/multimaster_pki.html
20:55 scsinutz "When running multiple masters, either the signing key-pair has to be present on all of them, or the master_pubkey_signature has to be pre-computed for each master individually (because they all have different public-keys)."
20:55 scsinutz how do i pre-compute each master's master_pubkey_signature in this scenario? is this done on the minion side? if so, where?
20:56 nickabbey joined #salt
20:57 netcho joined #salt
21:01 ozux joined #salt
21:03 netcho joined #salt
21:05 lorengordon joined #salt
21:07 cscf Can I load a file's contents into pillar, raw, without using file_roots ?
21:07 gtmanfred salt-key -F will give you all the signatures won't it?
21:08 gtmanfred oh, he left
21:14 irctc622 is there an easy way to specifiy templates for pkg.installed which changes the default values?
21:18 jas02 joined #salt
21:21 irctc622 What's wrong with this sls - it does not update the pkg to 0.3-3 version. https://gist.github.com/disaster123/2c99772e97abd4547244114e4e38cf08 but output is https://gist.github.com/disaster123/f3fb7a87dfb14a9b183f280ad3bb2416
21:22 hemebond What does "apt-cache policy toolkit-ph" show?
21:24 netcho joined #salt
21:24 irctc622 hemebond: https://gist.github.com/disaster123/706d34e33658a36055ecad4dde5695f0
21:25 hemebond Strange.
21:26 hemebond Can you install it manually?
21:26 sarcasticadmin joined #salt
21:26 hemebond With that no-recommends thing?
21:27 irctc622 hemebond: sure. also a apt-get --only-upgrade install toolkit-ph installs it fine
21:27 irctc622 homebond: it's just no reommends it's not no dependencies ;-)
21:27 hemebond Sure.
21:28 hemebond I wonder why the default for install_recommends is True.
21:28 hemebond That seems very odd.
21:28 irctc622 hemebond: yes me too
21:28 irctc622 hemebond: output is always:      Comment: All specified packages are already installed - it seems it never checks the version
21:29 hemebond Which version of Salt?
21:30 cscf irctc622, if you want it to update, wouldn't you want pkg.latest ?
21:32 irctc622 cscf: i do not always want to force the latest. I want to force a specific version but only if installed
21:32 irctc622 hemebond: 2016.03
21:34 BattleChicken joined #salt
21:36 ericprice joined #salt
21:37 hemebond irctc622: Is this a custom repo you're trying to install from?
21:37 irctc622 hemebond: yes
21:38 hemebond Have you tried the refresh property?
21:38 hemebond To make sure it refreshes the database?
21:38 patarr_ joined #salt
21:38 irctc622 i can also remove the only_upgrade option and it still does not install the mentioned version and just says pkg installed. i did manuall a apt-get update before while testing. Does refresh do anything else?
21:39 hemebond No, that should be enough.
21:39 irctc622 updated to 2016.11 no change
21:43 fracklen_ joined #salt
21:44 whytewolf irctc622: what does pkg.list_repo_pkgs toolkit-ph show?
21:46 irctc622 whytewolf: got it working by using pkg.installed: pkgs: - toolkit.ph: 0.3-3 instead of -name: toolkit-ph -version: 0.3-3 ? is this a bug?
21:47 whytewolf humm. that i don't know...
21:47 whytewolf apperently list_repo_pkgs isn't in aptpkg so that wouldn't have worked anyway
21:48 irctc622 does anybody know a way to make install_recommends: False as a default in my setup. I never ever want recommends pkgs
21:48 whytewolf although pkg.list_upgrades might have shown it
21:49 irctc622 whytewolf: have enough servers without the update
21:49 irctc622 should i run a salt call list-upgrades?
21:49 whytewolf salt-call pkg.list_upgrades
21:50 whytewolf and look for that package
21:50 irctc622 whytewolf: yes it's shown
21:50 irctc622 also the correct version
21:51 whytewolf humm. maybe file a bug report. at the very least will get confirmation
21:51 irctc622 whytewolf: https://gist.github.com/disaster123/16e11af89f374ce44741a4dfa4cab2e4
21:53 whytewolf salt-call pkg.latest_version toolkit-ph
21:53 whytewolf should also give the correct info
21:54 sarcasticadmin joined #salt
21:54 whytewolf also defintly file a bug report I am pretty sure that version: should try to keep the version correct. but i have a feeling i am missing something that i might have forgot about the way it works
21:56 irctc622 whytewolf: salt-call pkg.latest_version toolkit-ph works too and fine - ok will file a report
21:56 irctc622 whytewolf: does anybody know a way to make install_recommends: False as a default in my setup. I never ever want recommends pkgs
21:56 whytewolf no
21:57 whytewolf i mean you could edit the code for pkg
21:58 whytewolf but you then have to maintain your own copy at that point
21:59 skullone Is there a way to merge Pillar data? say, base system state pillar has a set of users defined, and another pillar for another service has more users to add?
21:59 abednarik joined #salt
22:00 irctc622 whytewolf: not a good idea. Is there a way to extend the pkg.installed? like mypkg.installed: extend pkg.installed - with params: -recommends: False, allow_updates: True, ... and than i use mypkg.installed everywhere?
22:00 skullone ie pillars/base.sls has one set of users, and then pillars/service.sls has more - i read that pillar data can't be merged in this way, though
22:02 skullone or, would each state need its own 'users' function, and lookup user data from its own pillar?
22:02 whytewolf irctc622: no, but all you really need is to make the modification in this file and add that as a custom module
22:02 whytewolf irctc622: https://github.com/saltstack/salt/blob/develop/salt/modules/aptpkg.py#L639-L643
22:05 irctc622 whytewolf: mhm have to think about it. May be it's easier to use a macro
22:05 whytewolf skullone: look into PILLAR_MERGE_LISTS & PILLAR_SOURCE_MERGING_STRATEGY in the master config
22:05 whytewolf skullone: https://docs.saltstack.com/en/latest/ref/configuration/master.html#pillar-merging-options
22:07 ericprice joined #salt
22:12 tmrtn[m] joined #salt
22:13 devtea joined #salt
22:13 ericprice joined #salt
22:15 skullone hrm.. might be best to put a user-add function and matching pillar in its own pillar then
22:19 jas02 joined #salt
22:28 abednarik joined #salt
22:32 freelock[m] joined #salt
22:32 jcl[m] joined #salt
22:32 ThomasJ|m joined #salt
22:32 saintaquinas[m] joined #salt
22:32 jerrykan[m] joined #salt
22:32 Mads[m] joined #salt
22:32 dnull[m] joined #salt
22:32 M-MadsRC joined #salt
22:32 M-liberdiko joined #salt
22:33 Sammichmaker joined #salt
22:34 orionx joined #salt
22:35 orionx_ joined #salt
22:38 wryfi i'm trying to get more information about the new ceph support in 2016.11
22:39 wryfi all i found is this page, so far, which doesn't say much: https://docs.saltstack.com/en/develop/ref/states/all/salt.states.ceph.html
22:39 wryfi does anybody know where to find more information?
22:39 whytewolf the code?
22:40 wryfi i was hoping for, you know, some documentation. but yes, i know i can read the code.
22:41 whytewolf you pointed at the documentation ... it says all there is currently. what were you expecting?
22:41 whytewolf the other place to find out more would be the module instead of the state module
22:41 whytewolf https://docs.saltstack.com/en/develop/ref/modules/all/salt.modules.ceph.html#module-salt.modules.ceph
22:42 wryfi well whytewolf, there are lots of questions that aren't answered by those docs. but i wasn't really looking for an argument, so i will just go read the code.
22:44 wryfi for one thing, the version of python-ceph-cfg doesn't work for ubuntu-14.04 so i was wondering what OSs are actually supported
22:49 whytewolf then the python-ceph-cfg in ubuntu 14.04 is not responding to a import ceph_cfg... it isn't checking the version. just that it can import.
22:49 _Cyclone_ joined #salt
22:50 skullone hrm.. im still confused about iterating users... is the best practice to put user-related tasks in their own states, with their own pillar files?
22:51 anotherzero joined #salt
22:51 whytewolf skullone: what ever works for you
22:51 skullone :p
22:51 whytewolf :b
22:51 skullone in puppet, i found myself adding users in each module, with hiera lookups from 1 big hiera file
22:52 skullone but salt has... more flexibility :p
22:52 hemebond joined #salt
22:52 whytewolf yes, that is the boon and curse of salt. there is a lot of flexability
22:53 wryfi whytewolf: i'm talking about underlying incompatibilities in the code of python-ceph-cfg. for example, if you look here https://github.com/oms4suse/python-ceph-cfg/blob/master/ceph_cfg/service.py
22:54 wryfi you will see that the init_system_sysV is written for RedHat derivs (note the chkconfig commands) and there is nothing for ubuntu pre-systemd
22:54 skullone just seems like id have a state file, with the same exact function, to iterate a pillar to add users
22:54 wryfi so python-ceph-cfg itself is incompatible with unbuntu-14.04
22:55 hemebond probably because Ceph was painful crap on <=14.04
22:55 skullone could i have a generic 'user' state, lookup data from a different pillar?
22:55 whytewolf ceph in all iterations i have tried has always been a painful POS regardless of underlying os.
22:55 wryfi hemebond: then it would be nice if somehwere in the release notes or other docs it says what OSs are compatible
22:56 hemebond I thought there was a note about it.
22:56 wryfi none that i have seen
22:56 hemebond Wait...
22:56 hemebond This isn't even in Salt.
22:57 hemebond or was it merged?
22:57 wryfi my understanding is that the new ceph code in salt relies on this library
22:57 whytewolf hemebond: he is talking about the salt module for ceph
22:57 hemebond Sure, but they linked to something else.
22:57 hemebond I see.
22:58 wryfi https://docs.saltstack.com/en/develop/ref/modules/all/salt.modules.ceph.html#module-salt.modules.ceph
22:58 hemebond yip
22:58 wryfi says "depends python ceph_cfg module"
22:58 wryfi the ceph_cfg module points to python-ceph-cfg
22:59 whytewolf most likely they testing with the pip version of ceph_cfg
22:59 wryfi i *think* this is what got merged into carbon? https://github.com/oms4suse/sesceph
23:00 whytewolf that i wouldn't doubt. most likely that was added as a PR request
23:01 whytewolf wryfi: https://github.com/oms4suse/salt-ceph-roadmap
23:01 wryfi um, what am i intended to do with these xml files?
23:01 wryfi is this docbook or something?
23:02 hemebond I currently don't trust any configuration management system to play nicely with Ceph so I do it all manually.
23:02 hemebond Hope the Salt module becomes useful.
23:02 whytewolf I think it is that guys roadmap for his plans for ceph in salt
23:02 whytewolf never said it was useful info
23:02 wryfi i'm just wondering how to read it ;)
23:03 whytewolf it has a Makefile.... maybe it is docbook
23:03 hemebond It's supposed to be compiled but you should be able to read the files directly.
23:04 wryfi ok, well this gives me a direction, not to use salt for this. thanks whytewolfand hemebond.
23:05 hemebond Yeah, I wouldn't trust anything but the Ceph scripts and binaries to touch Ceph.
23:05 hemebond Even restarting a Ceph service is a pain manually.
23:05 whytewolf yeah, i remeber having to restart ceph. it isn't the most ... os friendly system
23:06 whytewolf the worst was rebalancing
23:06 hemebond Yeah, terrible stuff.
23:06 * wryfi predicts lots more questions in IRC
23:07 hemebond If you use the bootstrap it'll at least get it all up and running properly.
23:07 hemebond but then don't touch it.
23:07 whytewolf I think, I'm not sure though ceph would be one of those systems to REQUIRE using orchestration to function
23:11 brasticstack joined #salt
23:14 krymzon joined #salt
23:15 scoates joined #salt
23:20 jas02 joined #salt
23:20 caedus joined #salt
23:23 hemebond Yes, that's what the Ceph bootstrap does.
23:24 hemebond Though you manually run the commands.
23:24 whytewolf I meant for maint as well
23:24 hemebond I tried their manual instructions several times and never ended up with a working cluster.
23:24 hemebond I would do maintenance manually too, directly using their tools.
23:25 wryfi i do really think that the release notes should be changed, guys
23:25 wryfi "Our friends over at SUSE have delivered a powerful new tool to make the deployment of Ceph storage systems using Salt very easy."
23:26 wryfi ^^ not a true statement
23:26 orionx joined #salt
23:26 gtmanfred they are using the modules to deploy it
23:27 wryfi modules or not, ceph is not simple to deploy (salt or not) and the whole thing depends on using a particular OS
23:28 raspado does modifying the master.conf or files inside master.d/ require a restart?
23:28 Edgan raspado: yes
23:28 whytewolf raspado: yes
23:28 raspado thx
23:35 abednarik joined #salt
23:42 netcho joined #salt
23:51 scoates joined #salt

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