Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2018-02-16

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

All times shown according to UTC.

Time Nick Message
00:02 stash__ joined #salt
00:08 jpsharp wom 3
00:08 jpsharp dammit.
00:14 pipps joined #salt
00:21 Vaelatern joined #salt
00:21 pipps joined #salt
00:33 stash__ joined #salt
00:43 pipps joined #salt
00:50 Edgan gtmanfred: ran into a new bug with the zookeeper znode code
00:57 Edgan babilen: When you mean easy to add custom filters, do you mean by adding them to your salt states/formulas, or do you have to modify the salt python code?
00:58 Edgan gtmanfred: nm, had to upgrade kazoo. fpm for the win
01:02 pipps joined #salt
01:06 stash__ joined #salt
01:25 taylorbyte joined #salt
01:32 stash___ joined #salt
01:50 Trauma joined #salt
01:51 cyteen joined #salt
01:55 tiwula joined #salt
02:01 exarkun joined #salt
02:03 stash__ joined #salt
02:38 zulutango joined #salt
02:42 stash__ joined #salt
02:55 ilbot3 joined #salt
02:55 Topic for #salt is now Welcome to #salt! <+> Latest Versions: 2016.11.9, 2017.7.3 <+> Support: https://www.saltstack.com/support/ <+> Logs: http://irclog.perlgeek.de/salt/ <+> Paste: https://gist.github.com/ <+> See also: #salt-devel, #salt-offtopic, and https://saltstackcommunity.herokuapp.com (for slack) <+> We are volunteers and may not have immediate answers
03:08 evle joined #salt
03:13 stash__ joined #salt
03:18 lompik joined #salt
03:41 exarkun joined #salt
03:43 stash__ joined #salt
04:08 zulutango joined #salt
04:11 Church- viq: Ya'll around?
04:18 stash__ joined #salt
04:24 MTecknology it's usually better to just ask your question
04:29 Church- Hmm?
04:29 Church- Woops, meant to do this in -offtopic.
04:39 justan0theruser joined #salt
04:45 stash___ joined #salt
04:50 Edgan joined #salt
05:21 stash__ joined #salt
05:21 exarkun joined #salt
05:31 djinni` joined #salt
05:43 lompik joined #salt
05:49 stash___ joined #salt
05:53 hoonetorg joined #salt
06:07 Edgan joined #salt
06:09 Hybrid joined #salt
06:18 hemebond left #salt
06:23 LocaMocha joined #salt
06:24 stash__ joined #salt
06:26 Hybrid joined #salt
06:30 snakegums joined #salt
06:42 lkthomas folks, what is the best way to push file to minions except using salt-cp ?
06:45 hemebond joined #salt
06:55 stash__ joined #salt
07:01 exarkun joined #salt
07:06 pualj joined #salt
07:15 mechleg lkthomas:  depending on size of the file or security concerns, one idea would be to put file contents into a pillar and distribute using a state
07:17 lkthomas mechleg, I think cp.get_file cli would work
07:22 gnomethrower joined #salt
07:25 stash___ joined #salt
07:30 zerocool_ joined #salt
07:56 Tucky joined #salt
08:00 stash__ joined #salt
08:02 Ricardo1000 joined #salt
08:14 aldevar joined #salt
08:19 aviau joined #salt
08:21 pualj joined #salt
08:25 Hybrid joined #salt
08:30 stash___ joined #salt
08:30 Hybrid joined #salt
08:40 exarkun joined #salt
08:54 saltnoob58 joined #salt
08:55 saltnoob58 hello
08:55 saltnoob58 i wish upon you especially tasty coffee today
09:02 pbandark joined #salt
09:07 pualj_ joined #salt
09:07 jrenner joined #salt
09:25 pualj_ joined #salt
09:51 KolK left #salt
10:10 losh joined #salt
10:16 viq Church-: you're not present in -offtopic ;)
10:19 peters-tx joined #salt
10:21 exarkun joined #salt
10:35 peters-tx joined #salt
10:51 peters-tx joined #salt
11:27 zerocoolback joined #salt
11:50 megamaced joined #salt
12:01 exarkun joined #salt
12:44 oarmstrong joined #salt
12:44 oarmstrong joined #salt
12:44 oarmstrong joined #salt
13:04 _JZ_ joined #salt
13:13 snakegums joined #salt
13:16 Nahual joined #salt
13:25 pualj_ joined #salt
13:30 cgiroua joined #salt
13:33 _JZ_ joined #salt
13:39 onslack <netcho> hey all, i have a gpg encrypted password in pillar, password starts with `!`
13:39 onslack <netcho> when rendering i get error
13:39 onslack <netcho> https://gist.github.com/vlebo/3ba482fda067aba5a4b4f19a8d6fcbe1
13:40 onslack <netcho> was encrypted with `gpgcommand \!MyPass`
13:41 exarkun joined #salt
13:42 onslack <netcho> so it's not possible to have passwords starting with ! or?
13:45 gmacon netcho: the leading exclamation is significant for yaml, you need to add some quotes
13:47 rockey netcho: as gmacon mention, you need either password: '!MyPass' or escape your special characters (ie: \!MyPass) which is used within the yaml engine as function triggers.
13:47 onslack <netcho> stupid me :)
13:48 onslack <netcho> `  - password: "{{ args['password'] }}"`
13:48 onslack <netcho> this worked
13:49 rockey we've all been there.. wait until you get to references in yaml, then you probably will kill yourself :)
14:01 babilen At that point you might just write your state in Python and be done with it
14:10 CrummyGummy joined #salt
14:11 onslack <netcho> :slightly_smiling_face:
14:12 netcho joined #salt
14:12 netcho joined #salt
14:12 netcho still not that hardcore i'm afraid :/
14:14 inetpro joined #salt
14:15 ProT-0-TypE joined #salt
14:29 inad922 joined #salt
14:32 cgiroua joined #salt
14:37 nixjdm joined #salt
14:39 rockey babilen: giant yaml subset of ip's for a full 10.0.0.0/8 subnet where i brought it down to 2k rows instead of 11k, just by adding a flattening function and added references as an list object
14:39 rockey babilen: not trivial to work with yaml references, i miss so much functionality :P
14:42 snakegums joined #salt
14:43 DammitJim joined #salt
14:45 Miuku joined #salt
14:48 evle joined #salt
14:51 irated joined #salt
15:02 racooper joined #salt
15:12 DammitJim joined #salt
15:15 gh34 joined #salt
15:18 Kelsar joined #salt
15:23 pualj_ joined #salt
15:33 tiwula joined #salt
15:58 pualj_ joined #salt
16:10 snakegums joined #salt
16:11 ingy netcho: `  - password: "{{ args['password'] }}"` will bite you if there are \ chars in the pw
16:13 ingy and `  - password: '{{ args["password"] }}'` will error with ' chars :\
16:26 rivyn joined #salt
16:28 demize t
16:32 DammitJim joined #salt
16:39 alfie joined #salt
16:42 lordcirth_work Can you pass a hashed password instead?
16:47 pualj_ joined #salt
16:48 Pulp joined #salt
16:49 gtmanfred yes
17:05 aldevar left #salt
17:44 inad922 joined #salt
18:14 Trauma joined #salt
18:22 pipps joined #salt
18:23 pipps joined #salt
18:28 pipps99 joined #salt
18:31 pualj_ joined #salt
18:41 Hybrid joined #salt
18:52 rubenb joined #salt
18:55 pualj_ joined #salt
18:56 MTecknology lordcirth_work: This is one of my very anctient pillar files from a previous life - https://gist.github.com/anonymous/716c919adc7cb22a884d6f3209d3ef27
18:57 lordcirth_work eyebleach required
18:59 ingy :)
19:00 Hybrid joined #salt
19:02 MTecknology I still don't know how I would make that any prettier than it is.
19:05 lordcirth_work LDAP? :P
19:06 wongster80 joined #salt
19:06 MTecknology for root?!
19:06 MTecknology hell nah
19:08 MTecknology A better solution that requires lots of extra steps would be something like hashi vault which would let you generate a unique root password for every host, but that's a lot of extra work.
19:08 lordcirth_work Well, also not using root, but yes
19:08 lordcirth_work If you have that many zones of access, you need an access control system other than a list of passwords
19:09 MTecknology zones of access?
19:09 ymasson joined #salt
19:09 lordcirth_work security domains, whatever you want to call them
19:09 MTecknology how does any of what I showed imply security domains?
19:09 MTecknology that is for one single security domain.
19:11 Bryson joined #salt
19:13 pipps99 joined #salt
19:14 edrocks joined #salt
19:14 lordcirth_work Then why do they have different passwords?
19:15 MTecknology because .... did you read the jinja?
19:15 lordcirth_work A security domain means you want a given key/user/whatever to be able to access one and not the other
19:15 MTecknology no.. that's not a security domain
19:16 MTecknology security domains are a networking topology thing, not a user-account thing
19:16 lordcirth_work Security domains are an abstract concept commonly implemented in networking topology, I would say
19:19 pipps joined #salt
19:20 pipps_ joined #salt
19:25 targettroublesss joined #salt
19:25 edrocks joined #salt
19:26 snakegums joined #salt
19:27 pipps joined #salt
19:30 pipps joined #salt
19:35 pipps_ joined #salt
19:37 rivyn "YAML uses a fixed indentation scheme to represent relationships between data layers. Salt requires that the indentation for each level consists of exactly two spaces. Do not use tabs."
19:37 rivyn However, I've used tabs without any issue?
19:39 rivyn Why have I not encountered any issue even though the documentation says that I should?
19:41 whytewolf because you are lucky? or your text editor changes tabs to spaces.
19:42 fxhp joined #salt
19:42 rivyn no it definitely does not
19:43 rivyn I really dislike space indentation.
19:44 * whytewolf shrugs
19:44 whytewolf as long as it is consistent it shouldn't be a problem either way.
19:46 rivyn In looking up the YAML specification for clarification, it appears there is both block and flow styles.  Block style, which my SLS files follow, are supposed to use two space levels of indentation.  But that's exactly where I've used tabs.
19:46 rivyn Flow style uses curly braces and allows for tab indentation.  I reckon maybe I should just update my SLS files to use that, as it seems to work as well.
19:47 pipps joined #salt
19:49 doubletwist In a piller if I do a  "set var" from a grain - if the grain isn't defined on the minion - will 'var' still be defined [as empty I presume] or will it be undefined?
19:53 hunter joined #salt
19:54 MTecknology iirc, yaml (like python) only cares about consistency
19:55 MTecknology rivyn: If you feel driven to use tabs, in the context of yaml, that sounds horrible.  The warning is because salt defines a standard set and if it gets mixed, you can expect things to not work as expected.
20:01 rivyn oops, I was testing against a different version of the YAML inadvertently
20:02 pipps99 joined #salt
20:02 rivyn tab indentation in block style does indeed not work - the salt master process ended up stuck and I was unable to kill -9 it, had to force restart the VM.
20:02 rivyn that could be due to other changes I made that I thought I was testing but hadn't though :/
20:04 rivyn writing a simple test SLS with flow style YAML works fine, but salt throws an error when I change the space indentation to tabs regardless of it's validity.
20:04 rivyn so I guess the server locking up was due to some other problem...I was working on different copies of the files than I thought I was yesterday afternoon
20:05 pualj_ joined #salt
20:09 aldevar joined #salt
20:11 MTecknology I would assume some weird rendering stuff would happen when jinja was applied..
20:11 onslack joined #salt
20:12 MTecknology gtmanfred: You happen to have a link for an issue tracker for onslack?  Looks like it needs a bug report filed.
20:13 onslack <gtmanfred> https://github.com/ekmartin/slack-irc
20:13 onslack <gtmanfred> that is what I am using
20:13 onslack <gtmanfred> but eventually i will write my own, one day when i have time
20:13 onslack <gtmanfred> in python3.6 with asyncio
20:13 whytewolf time? whats time?
20:13 onslack <gtmanfred> srsl
20:14 ingy rivyn: there should be no yaml implementation anywhere that allows tabs as indentation
20:14 rivyn ingy: not true.
20:15 ingy what's not true?
20:15 rivyn what you just said.  per the YAML spec.
20:15 rivyn See the first response here for instance:  https://stackoverflow.com/questions/25974485/yaml-as-a-json-superset-and-tab-characters
20:15 rivyn I also tested that YAML using flow style and tab indentation passes validation.
20:16 rivyn it's definitely more picky than I realized though
20:16 MTecknology It's valid.. just a very bad idea
20:16 rivyn I just assumed that whitespace was whitespace.
20:16 rivyn MTecknology: why bad?  sounds like biased opinion.
20:17 MTecknology read my previous statements?..
20:17 ingy tabs are not allowed as block scoping indentation
20:17 rivyn oh, within salt context, sure.
20:17 rivyn well it simply doesn't work in salt, as salt throws an error as soon as it sees any tab in the file - it's more restrictive than the spec.
20:17 ingy tabs are allowed as throwaway whitespace
20:17 MTecknology uhm... we're /in/ #salt, ya?
20:18 rivyn yes, but until I chanced upon that bit in the salt docs, I just assumed it needed to be valid yaml
20:18 rivyn I didn't even realize that yaml had block style indentation character restriction, so that was my bad
20:18 rivyn it would be cool if salt allowed json as well
20:19 whytewolf you can use json
20:19 rivyn yeah?  for the sls files?
20:19 ingy it should because JSON is a subset of YAML
20:19 ingy so any JSON is valid YAML
20:19 rivyn but you can't if your json is formatted with tabs
20:20 ingy yes you can
20:20 rivyn because the salt parser disallows it
20:20 rivyn I tried.
20:20 whytewolf #!jinja|json
20:20 whytewolf https://docs.saltstack.com/en/latest/ref/renderers/
20:21 rivyn oh, cool!
20:21 rivyn I'll play around a bit
20:22 lordcirth_work The command line example for file.check_file_meta seems out of date
20:25 XenophonF I don't want to jinx it, but 2017.7.3 might have fixed it...
20:26 rivyn whytewolf: is there any example?
20:26 whytewolf of using json?
20:26 rivyn yes
20:27 whytewolf not that i know of
20:27 rivyn not really sure how to format it
20:27 MTecknology whytewolf: Thanks!  (fwiw- https://github.com/ekmartin/slack-irc/issues/224)
20:27 MTecknology err. gtmanfred *
20:27 whytewolf lol. I was wondering why you thanked me :P
20:29 whytewolf rivyn: basicly it should be the data version of what you would be building in yaml. with dicts and lists.
20:30 onslack <gtmanfred> thanks!
20:30 MTecknology Is json any easier than python data structures?
20:30 rivyn not figuring out how to make the list part recognized by salt
20:31 onslack <daniel.sulbaran> Hey all, new user starting out, having issue with my only minion not being targeted by `'*'` - What's a good way to start troubleshooting this? Is it possible to see why a minion does/doesn't land on a certain target?
20:32 MTecknology 'salt-key -L' and 'salt-run manage.down' are probably decent starting points.
20:34 onslack <daniel.sulbaran> thanks! seems like it's accepted and not down, I can ping the minion
20:34 onslack <daniel.sulbaran> it's my master as well if that matters any
20:34 DammitJim joined #salt
20:35 rivyn whytewolf: none of these work.  Yeah, I'm stabbing in the dark a bit...  http://dpaste.com/09K4XXD
20:35 onslack <daniel.sulbaran> using gitfs, I think I just don't have my structure setup right
20:35 onslack <daniel.sulbaran> I can see top.sls on `salt-run fileserver.file_list`
20:35 rivyn likewise for [ "name: postgresql" ]
20:36 MTecknology "salt '*' test.ping" "salt 'foo' test.ping"
20:36 MTecknology granted... the runner really already covers that
20:37 onslack <tmacey> You can also specify the render pipeline for any SLS file to use whichever renderer you want
20:37 rivyn oh, I got it
20:37 rivyn need braces inside of square brackets, not either or
20:38 whytewolf yes, it is a list of dicts
20:38 pipps joined #salt
20:38 rivyn cool, I definitely prefer this.  I had another question for you, but let me get my saltstate fixed first. :)
20:39 rivyn gotta figure out what I broke yesterday afternoon
20:39 rivyn (aside from the tabs)
20:39 whytewolf I need to spend less time in this channel
20:39 MTecknology too distracting from real work?
20:39 rivyn :'( but you're such a valuable asset!
20:40 rivyn heheh
20:40 whytewolf no, it is the fact I need to stop giving free tech support and start pointing people to paid support
20:41 rivyn whytewolf: I believe you did that too, actually.
20:41 MTecknology ah, funny
20:41 rivyn I sent the link to the salt training you recommended to my boss
20:41 rivyn pretty likely I'll be able to go
20:41 whytewolf nice! can you talk them into paying for enterprise as well :P
20:42 rivyn I'm not sure how to tell, but I think we probably already have that.
20:42 MTecknology I did that to my previous employer so I could snag a ticket to saltconf :P
20:42 MTecknology err.. not enterprise, but support
20:42 MTecknology sounds like enterprise will soon be bundled w/ support?
20:42 whytewolf enterprise == support
20:43 whytewolf anyway, lunch time
20:44 MTecknology Is there more to enterprise than a pretty web UI?
20:45 whytewolf better rbac, support, the eAPI. I'm sure there is more but i was just about to head out the door.
20:46 rivyn I imagine we have that.  But I also imagine that getting something up through the proper person to go through the official support channels would be very involved here. :(
20:47 edrocks joined #salt
20:48 MTecknology I'm struggling to picture a scenario when you have saltstack enterprise but aren't aware of that fact definitively. If you don't know for a fact you have it, it's hard to believe you do.
20:48 rivyn I'm just going to switch back to the recommended default format to not be inconsistent with the docs.  Good to know about that functionality though
20:48 rivyn MTecknology: I just started at a large company that already has salt in place
20:48 rivyn I took over a couple saltstates, but there are dozens written by various departments
20:48 MTecknology that really doesn't mean anything, though...
20:49 rivyn we are also located within a few minutes of Salt HQ, and the company is big into support contracts from what I've seen
20:49 MTecknology I once tried to get access to a saltstack enterprise demo, but my requests kept winding up in the bit bucket, until I just decided I don't care anymore.
20:49 rivyn but yeah, I don't know how to check aside from asking around
20:49 MTecknology so- I can't really help you deduce which option you have.
20:50 coredumb I once tried .... until I saw the price
20:50 coredumb :D
20:50 MTecknology You thought it was expensive?!
20:50 viq gtmanfred: if there are problems with the slack bridge currently in use, maybe https://github.com/42wim/matterbridge will prove to work better? I have no personal experience with it, I just know a friend was using it to connect matrix to something
20:50 coredumb clearly the most expensive of any other cfg mgt tool
20:51 coredumb that you could pay support for
20:51 coredumb oh well I had the demo
20:51 MTecknology Is it? I actually thought it was exceptionally reasonably priced.
20:51 rivyn I don't know very much about salt yet, so I could be completely mistaken of course - it's just a guess
20:51 rivyn anyways they seemed willing and eager to pay for salt training :)
20:51 MTecknology rivyn: Do you have a pretty web ui or do you run manual salt commands?
20:52 rivyn manual commands
20:52 coredumb MTecknology: I guess we haven't had the same pricing then
20:52 MTecknology you're likely not using enterprise
20:52 MTecknology (doesn't mean they're not paying for support)
20:52 rivyn fair enough.  I'll ask my boss about it next time we chat
20:52 coredumb well didn't they stop the OSS support last year?
20:52 viq coredumb: are you at liberty to talk about the price? The only reference I have ATM is chef, at ~130 USD per node
20:53 MTecknology coredumb: I guess I was also dealing with a 501(c)3 corp...
20:53 MTecknology and ya... this was a couple years ago
20:53 coredumb viq: 20 more $ per node
20:53 MTecknology Ah.. ya, it's the OSS support that I was thinking of.
20:53 viq coredumb: thanks
20:54 coredumb and while the interface is slick
20:54 coredumb it doesn't justify the price
20:54 coredumb imho
20:54 viq Also huge props to chef for having clear pricing straight up on their page
20:54 MTecknology cripes... $130 per minion would be $65k/yr at $previous_employer
20:55 viq MTecknology: and coredumb says salt enterprise costs *more*
20:55 coredumb MTecknology: it was 150 per node for 500 nodes
20:55 MTecknology The OSS support was just a nice easy flat fee for something like X,000 to Y,000 nodes.
20:55 rivyn yeah, we have a whole crapload of VM nodes here too.
20:56 MTecknology Most of the ones I'm referring to were physical.
20:56 rivyn I'm pretty sure there's no way we'd have as many nodes as we do if they had to buy physical hardware for it :P
20:56 coredumb the only interesting stuff in the interface is the job result
20:56 coredumb so ....
20:56 coredumb :D
20:56 rivyn they spin up one-off VM's for every little random task
20:57 MTecknology There were ~200 remote skilled nursing facilities w/ their own network stack, most connected via T1.
20:57 MTecknology man... that was a heck of a learning experience
20:57 rivyn that's what I hate about VM's most - they're cool for their convenience, but then that convenience gets taken for granted and abused
20:57 coredumb s/VM/container/
20:58 coredumb let's be modern please
20:58 coredumb :D
20:58 exarkun pft containers are so 2017
20:58 viq Ah, puppet actually also has up-front pricing available
20:58 rivyn I'm not modern enough to know all the newfangled terminology! :P
20:58 rivyn salt is pretty wild stuff for me ;)
20:58 exarkun go small or go home
20:58 coredumb you should take care cause it's all the rage
20:59 coredumb it's the new way of deploying apps
20:59 MTecknology We should push this rambling/chatter to #salt-offtopic ... since this is kinda fun but also not all that on topic.
20:59 coredumb you put all your shits in a chroot
20:59 rivyn alright, I'll join :)
20:59 coredumb and BAM
20:59 coredumb magic
20:59 coredumb \o/
20:59 coredumb true we're a bit OT
21:00 rivyn coredumb: you're not in -offtopic :P
21:00 rivyn (yet)
21:00 coredumb yeah no
21:05 racooper joined #salt
21:35 rivyn whytewolf: okay, let me ask an on-topic question, if you're willing...  Given this SLS (https://gist.github.com/caseyallenshobe/c46f116416833c08f778e0b49912a9ac), the postgresql_service function is watching postgresql_package.  Is that correct or did I make an error?  Logically, I don't see how it knows to restart the service after database_cluster, but that seems to be what's happening (which is desirable)
21:36 rivyn the good thing is that it works so far as everything gets set up and started correctly - I can connect to the new DB after running it, and I can re-execute it to deploy config file changes
21:37 whytewolf might want to add database_cluster to that watch, but yes. that is correct. it will restart based on if the state being watched reports that it ran, and there are changes
21:37 rivyn I just feel like I made a mistake in the watch because I don't logically understand how it's working
21:38 rivyn installing the package without the service.running also starts the service, but it didn't get restarted to pick up the new cluster - it was in a "started" state with no clusters
21:38 rivyn and thus no actual running PG instances
21:38 rivyn how does execution of database_cluster trigger a service restart as it is now?
21:38 rivyn I don't see how they connect
21:39 whytewolf oh, nevermind i didn't actually see what it was. that seems to be something that would very rarely change
21:40 rivyn also I wasn't sure how it manages to restart after cluster creation at all, since I assumed that service.running wouldn't do anything since the service should already be running by that point.
21:40 whytewolf watch changes the behavoir of service.running
21:40 rivyn the only time the cluster should be created is upon initial install
21:40 whytewolf service.running makes sure that it is running. watch tells the state ok this state changed so we need to restart
21:41 rivyn ok, so probably I have the wrong watch there...I assume I should change it to watch: database_cluster
21:41 rivyn it only works because during the initial run of the saltstate, the package also gets installed, and the restart happens regardless
21:41 rivyn right?
21:41 whytewolf well, you probley also want the pkg.installed watched cause if a new package is installed it should be restarted
21:42 whytewolf you can watch more then one state
21:42 rivyn ahh, that's a good point
21:42 whytewolf same for config files
21:42 rivyn hmm?
21:43 rivyn a package upgrade shouldn't change config files.
21:43 whytewolf no, but you might
21:43 rivyn that's why I have reload_service watching configuration_files
21:43 rivyn although I *do* need to split postgresql.conf into two files - one including the other, with the parameters that require a restart when changed in a separate file
21:44 rivyn once I do that then I can watch the restart-required file from postgresql_service I assume
21:44 rivyn I was hoping to use this to deploy config file changes with a reload when possibel
21:44 rivyn instead of restart
21:45 whytewolf you can add reload: true to the service.running with the watch and watch will reload instead of restart
21:45 rivyn ok, so I should change out the cmd.run in reload_service to service.running with reload: true thon?
21:45 rivyn I thought I tried that, but I'll try again
21:46 whytewolf remember the watch part. that is what triggers the reload vs not reloading
21:46 rivyn oh, I used onchanges: - configuration_files for the reload
21:47 rivyn don't remember now why I needed to use that (or felt I did) instead of watch
21:47 cgiroua joined #salt
21:47 whytewolf onchanges is like watch with out the extra features that watch gives
21:47 whytewolf onchanges keeps the state from running while watch changes the behavour of the state
21:49 rivyn I think it's because I thought I needed onchanges to monitor the files for being changed or not
21:49 rivyn which would be better here?
21:49 rivyn I'm not clear on the difference
21:50 rivyn watch will always run on the initial run, onchanges won't?
21:50 rivyn in that case, I think onchanges is what I want
21:50 whytewolf no, watch changes the behavour based on what is happening on the system. onchanges will keep the state from running at all
21:51 rivyn service.running with reload: true ends up doing:  The service postgresql is already running
21:51 whytewolf it needs a watch other wise the reload means nothing.
21:52 whytewolf and the thing being watched needs to change
21:52 rivyn with the cmd.run version and onchanges, it does a service reload when I edit a config file
21:53 rivyn not sure why, but that's the desired behavior
21:53 pbandark joined #salt
21:54 rivyn it looks like service.running with reload: true and watch: instead of onchanges also does the same thing
21:54 whytewolf yes
21:54 whytewolf that is what i have been trying to say
21:54 rivyn sorry, I didn't grasp that from what was said
21:55 rivyn well, the watch method does a reload upon initial install as well, the onchanges method did not.
21:56 rivyn I thought you were suggesting that the onchanges version was wrong.
21:56 whytewolf no, onchanges is different
21:56 whytewolf not wrong
21:56 MTecknology https://docs.saltstack.com/en/latest/ref/states/requisites.html
21:56 rivyn I was mistaken, the onchanges one does get run at initial install as well.
21:57 rivyn it doesn't really matter, but is there a way to prevent the reload from running on the first pass when the service restart ends up happening anyways?
21:57 MTecknology typically- you only want -watch or -require unless you know you need something else
21:58 rivyn ok, I'll stick with watch then
21:58 whytewolf well can you get away with just using reload?
21:58 rivyn meaning?
21:58 rivyn a reload won't cause a new cluster to start up - it will only cause currently-runnig cluster(s) to reload the config
21:58 MTecknology If you care about avoiding useless service restarts and -reload is not enough, then -watch usually covers things.
22:00 edrocks joined #salt
22:00 rivyn I'm happy to ditch the only cmd.run that I had :)
22:00 rivyn the reload happening right after initial install doesn't hurt anything
22:02 exarkun joined #salt
22:04 swa_work joined #salt
22:14 zulutango joined #salt
22:19 druonysus joined #salt
22:25 druonysus joined #salt
22:30 druonysus joined #salt
22:30 druonysus joined #salt
22:30 druonysus joined #salt
22:30 druonysus joined #salt
22:33 mauli joined #salt
22:40 fxhp joined #salt
23:37 zulutango joined #salt
23:55 Nachos11 joined #salt
23:55 Nachos11 left #salt
23:56 aviau joined #salt

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