Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2017-08-21

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

All times shown according to UTC.

Time Nick Message
00:24 sh123124213 joined #salt
00:33 GMAzrael joined #salt
00:48 snc joined #salt
00:59 justanotheruser joined #salt
01:02 ssplatt joined #salt
01:28 masber joined #salt
01:31 snc joined #salt
01:51 ilbot3 joined #salt
01:51 Topic for #salt is now Welcome to #salt! <+> Latest Versions: 2016.11.7, 2017.7.1 <+> Support: https://www.saltstack.com/support/ <+> Logs: http://irclog.perlgeek.de/salt/ <+> Paste: https://gist.github.com/ <+> See also: #salt-devel, #salt-offtopic <+> We are volunteers and may not have immediate answers
02:08 gnomethrower joined #salt
02:19 zerocool_ joined #salt
02:21 GMAzrael joined #salt
02:24 lbv joined #salt
02:37 onlyanegg joined #salt
02:41 brianthelion joined #salt
02:44 JPT_ joined #salt
02:47 justanotheruser joined #salt
02:55 zulutango joined #salt
02:59 evle joined #salt
03:03 jerichowasahoax joined #salt
03:04 jerichowasahoax My salt master is running like ten different /usr/bin/salt-master processes. Is that supposed to happen?
03:05 whytewolf only 10?
03:05 jerichowasahoax that sounds like a yes
03:06 jerichowasahoax whytewolf: combination of "small network" and "i just rebooted this box", i guess
03:06 whytewolf yes. it is a yes. if you run ps -ef it will actually list out what each of the processes do. [at least it does for me]
03:07 whytewolf such as ProcessManager, ZeroMQPubServerChannel, Maintenance, ect
03:14 omie888777 joined #salt
03:15 mavhq joined #salt
03:22 GMAzrael joined #salt
03:22 dxiri joined #salt
03:26 Hipikat joined #salt
03:40 donmichelangelo joined #salt
03:50 MTecknology pstree
03:58 teratoma joined #salt
04:03 GMAzrael joined #salt
04:03 preludedrew joined #salt
04:10 brianthelion joined #salt
04:16 whytewolf honestly, not sure why pstree even exists.
04:16 whytewolf ps auxf is a tree format already
04:18 jerichowasahoax i first saw it in htop, and thought it was some nonsense about threads
04:22 teratoma joined #salt
04:24 whytewolf htop? they you really should see a lot more then 10. as htop displayes both threads and processes.
04:26 whytewolf each mworker having a couple of threads under it. the mworkerqueue would have several under it. ect.
04:26 whytewolf i have about 30 for my saltmaster if we count threads of procs
04:31 Ni3mm4nd joined #salt
04:41 honestly I don't think htop displays threads by default, you need to turn that on, right?
04:42 honestly nevermind - it's on by default but you can turn it off
04:50 GMAzrael joined #salt
04:54 hoonetorg joined #salt
05:06 GMAzrael joined #salt
05:08 pualj joined #salt
05:15 mavhq joined #salt
05:40 felskrone joined #salt
05:43 ProT-0-TypE joined #salt
05:58 justan0theruser joined #salt
05:59 justan0theruser joined #salt
05:59 oida_ joined #salt
06:07 GMAzrael joined #salt
06:08 shanth__ joined #salt
06:09 pualj joined #salt
06:18 varuntej Hi guys, I want to manage 10000 network devices using salt-proxy minions , What should be my minion & master system requirements ?
06:20 lkolstad joined #salt
06:26 Ricardo1000 joined #salt
06:38 evle joined #salt
06:41 coredumb joined #salt
06:46 EthPyth joined #salt
06:48 darioleidi joined #salt
07:07 ProT-0-TypE joined #salt
07:07 GMAzrael joined #salt
07:12 JohnnyRun joined #salt
07:13 Elsmorian joined #salt
07:17 o1e9 joined #salt
07:20 pppingme joined #salt
07:25 sh123124213 joined #salt
07:45 rgrundstrom joined #salt
07:45 mavhq joined #salt
07:47 jhauser joined #salt
07:47 rgrundstrom Good morning
07:52 absolutejam Morning
07:52 absolutejam Anyone know if file.serialize templates first?
07:52 absolutejam Can't see anything about it
07:53 absolutejam would be good if I could use file.serialize to tempalte with jinja first, then consume the dataset
07:55 manwe178 joined #salt
07:55 whytewolf absolutejam: well, since you feed it a dataset. you can build that dataset with jinja in the sls before hand.
07:57 absolutejam What I'm doing is getting a list of all files in a specific folder then feeding them to file.serialize in my state (win a for loop)
07:57 absolutejam I'm doing that via. {{ import_yaml <file> as yaml_file }}
07:57 absolutejam and using that for the dataset: parameter
07:58 absolutejam but I assume {{ import_yaml }} is choking because it doesn't template
07:58 absolutejam And i just realised that was my issue :P
07:58 absolutejam not file.serialize
07:58 absolutejam Haven't had my coffee yet
07:58 manwe178 Hi all, does anyone know what the process is for merging PRs between branches?  I submitted a few back in July which are all now in the 2017.7 branch, but only one of them has been merged to the 2017.7.1 branch, so I'm not sure how I'm supposed to get fixes into actual releases of Salt, as presumably 2017.7 won't be re-released.  Will the maintainers just merge this over eventually, or do I need to submit another PR to get it into the
07:58 absolutejam Alright, second question - Is there any way to template a file in memory? a file.managed but without saving to disk
07:59 absolutejam otherwise I'll have to save to disk in /tmp/ and remove afterwards :[
07:59 whytewolf manwe178: when they go to release they will pull a new version from 2017.7 branch
08:00 whytewolf absolutejam: instead of output as a yaml how about importing the jinja vars directly
08:01 whytewolf https://docs.saltstack.com/en/latest/topics/jinja/index.html#including-context-during-include-import
08:01 absolutejam oh, that makes sense
08:03 manwe178 whytewolf: Thanks for the info, so I guess I was just unlucky to miss the date that 2017.7.1 was branched from 2017.7?
08:06 MTecknology whytewolf: Am I missing something or should this now be closed?  https://github.com/saltstack/salt/issues/19054
08:07 kedare joined #salt
08:07 * whytewolf shrugs.
08:08 GMAzrael joined #salt
08:09 whytewolf i should really be in bed :(
08:09 MTecknology agreed..
08:09 MTecknology I've been hacking on gitea.debian.net all day and it's been really fun... but hard on the sleep.
08:10 viq shanth__: not really, with my personal stuff my focus is mostly on OpenBSD, and at work linux
08:11 MTecknology It's running redundant front-end gitea instances, LE certs w/ acmetool, cert-based auth from Debian's SSO, automatic user-creation, few more bugs created, wiki page (soon) relatively useful... it was too much excitement to sleep
08:13 _KaszpiR_ joined #salt
08:14 viq MTecknology: niiice :)
08:15 viq "cert-based auth from Debian's SSO" sounds intriguing, where can I read more about that? ;)
08:16 MTecknology sso.debian.org
08:16 viq thanks
08:19 Akkarin joined #salt
08:20 coredumb MTecknology: I'm interested to hear how you've made gitea redundant
08:21 Akkarin Hey, I just ran into #40490 on Arch. Does anybody have a workaround handy? downgrading M2Crypto sadly doesn't work :-/
08:23 bdrung_work joined #salt
08:27 MTecknology coredumb: mostly thanks to nfs
08:28 MTecknology feel free to hop into #salt-offtopic if you wanna hear details :)
08:29 Mattch joined #salt
09:03 saintpablo joined #salt
09:07 schasi joined #salt
09:10 saintpablo joined #salt
09:11 Hybrid joined #salt
09:12 GMAzrael joined #salt
09:14 izibi joined #salt
09:14 izibi joined #salt
09:14 izibi hi, where can I find a security contact for salt?
09:15 Cadmus joined #salt
09:15 CrummyGummy joined #salt
09:18 Cadmus Hello, I seem to be having a dilly of a problem with a file.managed that pulls from a repo of ours. So it's an HTTP GET of some url like http://repo.example.org/binary?LATEST which gives you a redirect to the latest binary.
09:18 izibi https://saltstack.com/contact/ should really mention https://docs.saltstack.com/en/latest/security/ ...
09:19 Cadmus But later runs say "file is already in the correct state" clearing the cache makes it work. So either I need to tell the statement "never cache this managed file" or turn off the cache on my minions or something
09:20 absolutejam hm, how do I updaet a dict value in jinja?
09:20 absolutejam doesn't like {% set foo['bar'] = 'xyz' %}
09:20 absolutejam or foo.bar
09:23 aldevar joined #salt
09:25 Cadmus absolutejam: I'm not so up on jinja, what if you set foo the empty dict first? `set foo = {}`
09:26 varuntej Hi guys whats the best way manage 10000 devices using proxy-minion using salt? do we need to start 10000 proxy processes in a minion or do we have any alternative?
09:31 Hybrid joined #salt
09:34 kukacz joined #salt
09:37 absolutejam I'm trying to append to a dict property
09:37 absolutejam I'm actuallying trying to update { 'foo': [ ] }
09:38 seena joined #salt
09:38 Elsmorian joined #salt
09:43 absolutejam aha
09:43 absolutejam got it
09:44 absolutejam {{ mydict.update({ 'foo': bar }) }}
09:44 absolutejam where bar is the appended list
09:44 zulutango joined #salt
09:49 Naresh joined #salt
09:50 babilen absolutejam: That doesn't merge nested datastructures, you might be interested in https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.defaults.html#salt.modules.defaults.merge
09:51 seena hi
09:52 seena I am facing issue with the python version whenever installing/upgrading salt-minion via rpm on Amazon linux. Which python version salt minion will take ?
09:56 absolutejam thanks. Not too big a deal because it's a fresh dict I'm making, but I'll check
09:56 absolutejam seems to be working okay now after I pass it to file.serialize
09:58 seena https://gist.github.com/seenae/61d63338b32cc3941842a31243555077
09:59 kukacz joined #salt
09:59 babilen absolutejam: Sure, just though I'd mention it in case you wonder why some data is being replaced and not merged
10:04 seena @babilen can u help?
10:10 zerocoolback joined #salt
10:11 EthPyth joined #salt
10:13 GMAzrael joined #salt
10:13 pbandark joined #salt
10:14 schasi When I run state.orch with a sls that has "- highstate: True" and a host that is offline, the state returns successfully o_0
10:18 absolutejam I have hit that issue before I think haha babilen
10:19 schasi I actually found the bug in the bug database
10:21 Valfor joined #salt
10:21 Valfor joined #salt
10:24 sh123124213 joined #salt
10:31 smartalek joined #salt
10:47 bluenemo joined #salt
10:51 vtolstov joined #salt
10:56 vtolstov hi all. may be im try very complicated things.. but i cant understand how can i do something like this: i have 1000 nodes, each node have some roles like relay traffic, boot server, web app, dns server.... for eac role i need to provide some generic attributes dependent on dc, and some attributes depends on server
10:57 vtolstov i'm try to use pillarstack all fine, but i can't include pillar depends on some pillar value
10:57 vtolstov i think about to provide on node pillar roles list and include needed pillar files, but as i see i can' to pillar.get inside pillar (result always empty).
10:58 vtolstov i'm use salt 2016.11
11:03 babilen vtolstov: I'm off for lunch now, but shouldn't you be able to get pillar data defined "higher" in the stack?
11:04 babilen So, complete one set first and then build on that?
11:04 Ricardo1000 joined #salt
11:07 vtolstov @babilen - i can't, i simply definen roles: [ boot, relay ] and in stack config file try to {% for role in __salt__['pillar.get']('roles', None) %}
11:07 vtolstov and always have none
11:07 vtolstov but pillar.items says that roles defined
11:10 pbandark1 joined #salt
11:13 babilen vtolstov: Please keep in mind that there are various bits at play here ... First you have the question if you render "normal" pillars first (i.e. before pillarstack) or afterwards and then you have the various "layers" in the pillarstack stack. You should be able to access data defined in "lower/earlier" layers
11:14 wireknot joined #salt
11:16 vtolstov @babilen: i'm not use other pillars only pillarstack
11:17 vtolstov but how can i get pillar data in current stack? as i understand this can be possible becasue pillar is the big python dict
11:18 babilen Pillars are empty in the configuration file if you don't define them in "normal" pillars, but you should be able to access pillar data in the files that make up later layers in the pillarstack itself
11:23 johnj joined #salt
11:27 kedare joined #salt
11:31 zerocoolback joined #salt
11:34 vtolstov @babilen: but i need to access in curent pillar layer =(
11:34 babilen Why?
11:35 babilen That is not possible as you cannot access data in the current layer as that hasn't been rendered yet
11:35 babilen You could define a suitable datastructure in the SLS that you query if you have to do that
11:35 evle1 joined #salt
11:35 babilen (e.g. a dictionary)
11:41 EthPyth joined #salt
11:43 Guest73 joined #salt
11:46 GMAzrael joined #salt
11:46 mavhq joined #salt
11:48 vtolstov @babilen: =( but i'm not understand how structure sls in my case if pillar data inside pillar not supported
11:51 babilen It is, you just have to make sure that the data you reference is rendered before you access it
11:57 haam3r_ Hi, will file.replace work recursively if I give it a name like *.py or a folder?
11:59 johnkeates joined #salt
11:59 flughafen_ joined #salt
12:00 flughafen_ does jinja work with "starting".startswith?
12:01 flughafen does jinja work with "starting".startswith?
12:04 ProT-0-TypE joined #salt
12:04 EthPyth joined #salt
12:12 vtolstov @babilen: so in ext_pillar data not rendered?
12:13 Tucky joined #salt
12:17 numkem joined #salt
12:24 ixs joined #salt
12:27 manwe178 Does anyone use Salt to manage the salt package version itself?  Currently I'm finding that when applying the state, the minion doesn't return and yum ends up with unfinished transactions.  I'm thinking of just making the update of the Salt packages a manual process.
12:28 johnkeates it should run in a systemd run context unit and finish regardless of salt minion getting stopped for upgrades
12:28 johnkeates if you use legacy init it doesn't work
12:28 gmacon joined #salt
12:29 uncool joined #salt
12:29 jdipierro joined #salt
12:32 babilen vtolstov: Pillarstack has multiple "layers" in that earlier data is available .. you might also want to look into https://devopsimply.com/2016/09/24/problems-with-salt-pillar-stack/
12:33 vtolstov @babilen: thanks i'm solve my issue - i'm provide two files to stack pillar so i can reference older vars by stack dict
12:34 babilen vtolstov: Yes, that is exactly what I meant by "earlier layers"
12:34 babilen That's the reason why its called pillarSTACK as you can access data defined in earlier files/layers
12:35 vtolstov =) thanks!
12:36 babilen One is rendered, data is available to all files/layers rendered afterwards
12:36 pualj joined #salt
12:36 vtolstov @babilen: thanks, last question
12:36 babilen How do you access it now? Mind showing me an example on one of http://paste.debian.net, https://gist.github.com, http://sprunge.us, … ?
12:37 vtolstov @babilen: common file: https://gist.github.com/vtolstov/99d8423703fd7d2201c98ec7f46afd08
12:38 vtolstov @babilen: roles file: https://gist.github.com/vtolstov/6937867d1622b8f3ee960fa35932f78c
12:38 vtolstov @babilen: last qeustion - does it possible in top.sls file check that some states dir contains files?
12:39 vtolstov @babilen: my top.sls https://gist.github.com/vtolstov/3e70ec49970cc5336ade4e40d37b5d18
12:39 vtolstov my problem with top.sls that i need to create empty sls file inside all relevant dirs
12:39 vtolstov because pillarstack silently ignore dirs that ebsent and not contains data, but when salt render states file it fails for missing states
12:41 babilen ignore_missing: True
12:42 edrocks joined #salt
12:42 babilen lunch
12:43 vtolstov this is only for pillar and not states
12:43 vtolstov https://github.com/saltstack/salt/issues/16687
12:47 mavhq joined #salt
12:48 manwe178 johnkeates: Thanks for the info.  We are using systemd but still seem to have this problem :/
12:49 johnkeates maybe you are using stricter kill policies? it seems that your process chain gets killed when the parent gets closed
12:49 manwe178 Will have to google what you mean by that
12:49 manwe178 I think it behaves better when I use 'salt-call' than when I use 'salt'
12:50 babilen vtolstov: See https://github.com/saltstack/salt/issues/20010 for a workaround for states
12:51 AvengerMoJo joined #salt
12:52 kedare joined #salt
12:54 tapoxi joined #salt
12:58 mchlumsky joined #salt
13:01 TK__ joined #salt
13:02 johnkeates i think i have a visibility bug for salt; when running postgres states, some failures have useful information in the log but simply return "failed" in the state
13:03 johnkeates do i report it on the salt project or on the formula/module/state project? it's not a problem in the formula, but i can imagine someone watching that has it covered
13:29 beardedeagle joined #salt
13:30 numkem joined #salt
13:38 Guest73 joined #salt
13:41 brianthelion joined #salt
13:46 beardedeagle joined #salt
13:46 k1412 joined #salt
13:47 cgiroua joined #salt
13:51 numkem joined #salt
13:53 Shados joined #salt
13:55 Ricardo1000 joined #salt
13:56 evle joined #salt
13:56 ssplatt joined #salt
13:56 bildz I'd like to use a directory structure like the following:  https://pastebin.com/BRQpS8SA      How do you reference a windows repository with multiple base paths?  Do you need to create one for each and how are they referenced in the state?
14:05 brianthelion joined #salt
14:07 muxdaemon joined #salt
14:32 edrocks joined #salt
14:36 Ch3LL joined #salt
14:46 vtolstov how can i get currrent dir in sls file ?
14:46 vtolstov i'm try slspath but it empty
14:46 vtolstov (try in top.sls)
14:49 ntropy vtolstov: i don't think there's a way to do that, but can you explain what you're trying to do?  im surprised you need this
14:49 ixs curious question: when using grains.filter_by, is there actually any difference between the default and the base parameter? googling for examples shows https://github.com/saltstack/salt/issues/28606#issuecomment-224981792 where base is used in the same way I would expect default to be used.
14:51 sarcasticadmin joined #salt
14:54 XenophonF are there conventions for orchestration formulas?
14:54 Shirkdog joined #salt
14:57 XenophonF i want to create an orchestration that deploys something like the AARC pilot platform - https://aarc-project.eu/wp-content/uploads/2017/06/AARC-BPA-2016.png
15:04 vtolstov @ntropy: i need to skip empty dirs (i'm use wildcard includes in top.sls)
15:04 vtolstov for example i have dir environment/{dc1,dc2}
15:04 vtolstov in top sls i have - states.environment.dc1.*
15:05 vtolstov but in some cases dc1 may be empty dir and in this case salt fails
15:06 babilen vtolstov: Did the workaround not work .... around?
15:15 vtolstov @babilen it works =) but i'm try to avoid loops in file_roots =)
15:19 Guest73 joined #salt
15:21 ProT-0-TypE joined #salt
15:25 hax404_ joined #salt
15:32 fatal_exception joined #salt
15:43 tiwula joined #salt
15:44 ProT-0-TypE joined #salt
15:46 JPT joined #salt
15:47 omie888777 joined #salt
15:51 wireknot hi all, i'm trying to save the result data to redis server. I get 'ret:xxxxx' values in the redis - how can i convert it to a usable output?
15:51 wireknot https://dpaste.de/9em6
15:53 wireknot what am i missing? i thought that the result will be saved as string or something...
15:57 Hybrid joined #salt
16:06 _KaszpiR_ joined #salt
16:11 whytewolf wireknot: what is the result of the second key key? the ret:<jobid> one. the other one looks like it returned the jobid for that task.
16:13 _JZ_ joined #salt
16:14 wireknot whytewolf: https://dpaste.de/Yd1J
16:15 wireknot whytewolf: i trired that before but i assumed it's the value of the key that needs to be converted to some other type (e.g. dict) when i got the exception
16:16 wireknot it's the first time i use redis so guessing a lot..
16:19 whytewolf unforchantly i have never used redis directly. only used apps that used redis. so honestly not really sure myself
16:19 wireknot np, thanks for checking it out
16:20 cyborg-one joined #salt
16:22 fritz09 joined #salt
16:26 rgrundstrom joined #salt
16:26 rgrundstrom Good evening.
16:27 numkem joined #salt
16:35 bildz is anyone using a s3 bucket as a package repository?  I have the credentials configured on the master, however, if I run a salt 'server' s3.get  it says invalid credentials
16:36 bildz https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.s3.html  I have everything setup correctly per this link
16:48 XenophonF Did you try setting the s3 credentials on the minion?
16:51 sh123124213 wireknot: I think the data type is different from the one you are trying to get
16:51 sh123124213 try list of set
16:52 sh123124213 or*
16:56 sh123124213 type ret:20170821163727770333
16:59 WesleyTech joined #salt
17:06 Lionel_Debroux_ joined #salt
17:11 pipps joined #salt
17:11 alvinstarr1 joined #salt
17:12 WesleyTech joined #salt
17:21 pipps joined #salt
17:24 numkem joined #salt
17:36 pipps joined #salt
17:36 _KaszpiR_ joined #salt
17:39 pipps99 joined #salt
17:53 schemanic joined #salt
17:56 jerichowasahoax joined #salt
18:10 ipmb joined #salt
18:11 Brew joined #salt
18:11 ipmb joined #salt
18:25 Yodle joined #salt
18:25 shanth__ is anyone managing windows machines using salt in a corporate setting? seems like the windows stuff for salt barely works
18:26 shanth__ seeing if it is worth exploring
18:33 beardedeagle joined #salt
18:37 ChubYann joined #salt
18:37 gh34 joined #salt
18:38 brd joined #salt
18:43 shanth_ joined #salt
18:43 johnkeates joined #salt
18:51 shanth_ joined #salt
18:52 smartalek joined #salt
18:57 sjorge joined #salt
19:14 debian1121 joined #salt
19:17 tapoxi joined #salt
19:21 _JZ_ joined #salt
19:27 XenophonF I'm using Salt on Windows.
19:27 XenophonF Right now it's just super basic stuff like local account creation and software deployment.
19:28 XenophonF Complicating things on Windows is the frequent need to resort to orchestration to do anything real.
19:31 jrra joined #salt
19:32 bildz XenophonF: I got the s3.get to work by creating a s3fs config under minion.d
19:32 bildz XenophonF: have you been able to use a bucket for holding all the win_repo files?
19:33 gtmanfred bildz: if you want to only set the configs on the master, you will need to enable `pillar_opts: True` on the master, to pass the master's configs to the minion in pillar data
19:33 bildz gtmanfred: thanks that is helpful info
19:34 gtmanfred not advisable though, because it would have all the master configs there
19:34 bildz yeah its easier to push out a minion config
19:34 whytewolf isn't s3fs one of those items that gets it's configs through config.get
19:34 gtmanfred yes
19:34 whytewolf so, could do pillar than
19:35 gtmanfred well, not s3fs
19:35 gtmanfred s3.get is
19:35 whytewolf ahh
19:35 bildz unable to cache salt://s3bucket/RubrikBackupService.msi
19:35 gtmanfred s3fs will pull the files down from the master
19:35 bildz or do I need s3://s3bucket/<file> ?
19:36 gtmanfred for s3.get? or are you trying to pull files from s3fs?
19:37 shanth__ joined #salt
19:37 gtmanfred if you have s3fs configured
19:37 bildz gtmanfred: it works with s3.get.   I am trying to use this in winrepo_ng
19:37 gtmanfred you should only need to check cp.cache_file `salt://RubrikBackupService.msi`
19:37 gtmanfred because the salt-master will run s3.get on s3://s3bucket/RubrikBackupService.msi
19:37 gtmanfred and send it to the minion
19:37 gtmanfred check out `cp.list_master`
19:38 Hybrid joined #salt
19:40 gtmanfred whatever files you see from `cp.list_master` you can get from the master by putting a `salt://` in front of it, and it will come through the fileserver
19:40 noname joined #salt
19:41 gtmanfred if you setup s3fs, this will allow the master to use s3 as a fileserver backend, and the master will do the s3.get.
19:41 gtmanfred I do believe it is also possible to prefix files with s3:// on the minion, and have it connect to s3 directly, (also think swift:// is available for that)
19:41 gtmanfred https://github.com/saltstack/salt/blob/2017.7/salt/fileclient.py#L515
19:42 gtmanfred here is the salt one https://github.com/saltstack/salt/blob/2017.7/salt/fileclient.py#L494
19:45 ToeSnacks joined #salt
19:46 copelco_ joined #salt
19:46 Awesomecase joined #salt
19:46 ople_ joined #salt
19:47 mihait joined #salt
19:47 phobosd__ joined #salt
19:48 moy joined #salt
19:49 bildz gtmanfred: thanks that totally worked.  Its just really slow pulling it down from s3 master -> minion
19:49 gtmanfred yeah it is slower
19:50 gtmanfred cause it pulls it down from the master, and then transfers it to the minion
19:50 gtmanfred you can speed up the transfer to the minion by increasing the chunk size
19:50 bildz yeah a 12MB file took a while
19:50 gtmanfred https://docs.saltstack.com/en/latest/ref/configuration/master.html#file-buffer-size
19:51 fritz09 joined #salt
19:51 gtmanfred I believe that is the setting
19:52 gtmanfred by default the buffer is about 1Kb, i think if you increase that, it should make the transfer speed faster
19:53 gtmanfred bildz: ^^
19:54 whytewolf maybe also force the fileserver cache to update before trying to grab it on the minion.
19:54 gtmanfred actually that default is 1Mb
19:55 Hybrid joined #salt
19:56 simondodsley joined #salt
20:01 lordcirth_work how do I target all minions except *421* ?
20:02 gtmanfred salt -C 'not *421*'
20:02 bildz are there any best practices for the file_buffer_size ?
20:03 gtmanfred the best practice is the one it is set to by default
20:03 gtmanfred but you can speed up the transfer by increasing it
20:03 bildz yeah heh
20:03 gtmanfred at the same time
20:03 gtmanfred the file server sends file chunks defaulting to 1 meg, that option allows you to send larger single zmq message chunks
20:03 gtmanfred raising that value can make the master less responsive as the mworker processes will block longer on the IO
20:03 gtmanfred so, pick your poison :P
20:04 bildz i dont really see a huge benefit to using the s3 backend.  It's slower and also causes the state to time out (i guess i could up the timeout level)
20:04 gtmanfred so, it doesn't have any benefits to transfering large files
20:04 bildz just use s3 to keep the repo in sync
20:04 gtmanfred you could reference it in for the fileclient, using s3://, and then it won't be cached on the minion, and the minion will pull it down directly
20:05 gtmanfred you could periodically run `salt-run fileserver.update backend=s3fs` and the file will be cached on the master
20:05 gtmanfred then it will only be the transfer to the minion that it has to do
20:05 lordcirth_work gtmanfred, thanks!
20:06 bildz gtmanfred: thanks!
20:06 gtmanfred no problem!
20:08 shanth__ can i use file.blockreplace to manage a file in the sense that i have one section is that managed but allow the user to edit below that and keep those changes too?
20:08 shanth__ that is*
20:08 gtmanfred yes
20:08 shanth__ excellent
20:09 gtmanfred blockreplace only touches stuff between the commented section
20:09 gtmanfred the user can do anything else
20:09 shanth__ just making sure thats the best module for that
20:09 whytewolf that would atually be the best use of blockreplace :P
20:10 gtmanfred yes
20:10 shanth__ sweet
20:13 alvinstarr1 joined #salt
20:19 eseyman joined #salt
20:29 shanth__ file.blockreplace marker_end says Note: you can use file.accumulated and target this state - does that mean i can use file.accumulated for example to have 2 states add text to a managed block?
20:29 shanth__ doesnt show how to target the markers under file.accumulated
20:30 gtmanfred so
20:31 impi joined #salt
20:32 shanth__ oh wait it might be under the block replace section
20:32 gtmanfred yeah, you would use them together
20:33 shanth__ ah i see
20:33 shanth__ let me play with it and see how it goes
20:33 gtmanfred The quick brown fox{% for animal in accumulator['animals_doing_things'] %}{{ animal }}{% endfor %}
20:34 gtmanfred i can't think of a reason to use the accumulator
20:35 whytewolf i have never found a viable use case fo accumulator that can't be handeled in a template some other way
20:35 astronouth7303 is there a reference to the "lowstate" expected by salt-api?
20:36 astronouth7303 i'm finding examples, but nothing exhaustive
20:37 DeepThinking joined #salt
20:39 whytewolf what is asking for a lowstate?
20:40 whytewolf a lowstate is the final dataset before exacution.
20:42 astronouth7303 that's how libpepper refers to the datastructer sent to `/` and `/run`
20:44 whytewolf have you tried something like [{"client": "local", "tgt": "*", "fun": "test.ping", "username": "saltdev", "password": "saltdev", "eauth": "auto"}]
20:46 astronouth7303 i mean, that's the example
20:46 whytewolf yes it is
20:46 astronouth7303 i'm working on heavily modifying libpepper and i wanted a definitive reference for what could be in it
20:46 whytewolf it isn't a lowstate but it is what /run expects
20:48 DeepThinking joined #salt
20:49 DeepThinking joined #salt
20:49 lkolstad joined #salt
20:52 Elsmorian joined #salt
20:56 wavded joined #salt
20:58 Guest89125 left #salt
20:59 moy joined #salt
21:02 OliverMT left #salt
21:04 bildz gtmanfred: if you have environments split up, do you need to specify a winrepo for each of them, or can they all default to base?
21:04 woodtablet joined #salt
21:05 gtmanfred depends on how you specify your environments for the minion
21:05 gtmanfred if you lock the minion in with `environment: dev` then win/repo-ng needs to be available in the dev environment on the fileserver
21:07 schemanic joined #salt
21:18 omie888777 joined #salt
21:23 shanth__ joined #salt
21:30 teratoma joined #salt
21:31 brokensyntax joined #salt
21:33 teratoma joined #salt
21:34 johnkeates joined #salt
21:34 aldevar left #salt
21:37 teratoma joined #salt
21:41 Guest73 joined #salt
21:43 avinson joined #salt
21:45 avinson is there a way i can do nested pillar data like this: {{ salt['vault'].read_secret('secret/app/{{ pillar['stack_name'] }}/universe/vpn/server.key', 'value') | yaml_encode }}
21:45 avinson i'm sure there is something obvious i'm missing here
21:46 astronouth7303 try `{{ salt['vault'].read_secret('secret/app/' + pillar['stack_name'] + '/universe/vpn/server.key', 'value') | yaml_encode }} ?
21:46 astronouth7303 within {{}} it's kinda python-ish
21:46 avinson cool thanks
21:47 avinson thought catenating might be the way to do this
21:47 avinson will try now and report back
21:48 rpb joined #salt
21:52 cyborg-one joined #salt
21:53 rpb joined #salt
21:54 noraatepernos joined #salt
21:54 avinson @astronouth7303 nice that worked.. thanks again
21:56 LeProvokateur joined #salt
21:57 felskrone joined #salt
21:59 hammer065 joined #salt
22:02 Eris joined #salt
22:03 Eris left #salt
22:14 pipps joined #salt
22:16 ssplatt joined #salt
22:28 cro joined #salt
22:29 lkolstad joined #salt
22:31 _KaszpiR_ joined #salt
22:38 mechleg joined #salt
22:40 GMAzrael joined #salt
22:42 jrra left #salt
22:42 felskrone joined #salt
22:47 pipps joined #salt
23:02 keldwud joined #salt
23:02 keldwud joined #salt
23:03 debian112 joined #salt
23:18 vtolstov joined #salt
23:19 vtolstov hi again =). i have a problems with autogenerated top.sls file
23:19 vtolstov my top.sls https://gist.github.com/vtolstov/d4f77aaf6959eee8dec79545e472c814
23:20 vtolstov i'm use syndic to control minions
23:20 vtolstov when this top file runs on master all fine - i'm generate top.sls with needed states.
23:21 vtolstov but in case of minions each minion cache old states in /var/cache/salt/files/base/
23:21 vtolstov does it possible not generate top.sls on minion side and generate it on master and send to minion?
23:21 vtolstov in my case if minion does not have cached states top file is empty =(
23:22 nidr0x joined #salt
23:31 fatal_exception joined #salt
23:31 pipps joined #salt
23:35 whytewolf is this a state top file or a pillar top file.
23:35 pipps99 joined #salt
23:39 nidr0x joined #salt
23:41 GMAzrael joined #salt

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