Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2018-02-12

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

All times shown according to UTC.

Time Nick Message
00:19 oida joined #salt
01:02 shiranaihito joined #salt
01:03 Guest73 joined #salt
01:12 Guest73 joined #salt
01:14 exarkun joined #salt
01:39 Guest73 joined #salt
01:41 snakegums joined #salt
02:00 Guest73 joined #salt
02:15 Guest73 joined #salt
02:17 KevinAn2757 joined #salt
02:27 Yee joined #salt
02:32 Yee left #salt
02:32 onslack <randy> just for my own edification, have we solved the whole “pick up where we left off after rebooting mid state” thing?
02:33 onslack <randy> I’ve been out of the game for awhile so I don’t know
02:33 YewR joined #salt
02:34 YewR left #salt
02:54 yidhra joined #salt
02:54 nomeed joined #salt
02:56 ilbot3 joined #salt
02:56 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
03:12 fxhp joined #salt
03:15 lompik joined #salt
03:21 evle joined #salt
03:31 Guest73 joined #salt
03:51 JPT joined #salt
03:58 Edgan joined #salt
04:35 exarkun joined #salt
04:38 snakegums joined #salt
05:18 Guest73 joined #salt
05:35 Guest73 joined #salt
05:45 zerocoolback joined #salt
05:46 zerocoolback joined #salt
06:01 alexlist joined #salt
06:14 inad922 joined #salt
06:15 exarkun joined #salt
06:21 Edgan joined #salt
06:27 pualj joined #salt
06:39 Edgan joined #salt
06:49 evle joined #salt
06:58 gnomethrower joined #salt
07:08 colttt joined #salt
07:22 Guest73 joined #salt
07:33 darioleidi joined #salt
07:39 Ricardo1000 joined #salt
07:47 Tucky joined #salt
07:51 irated joined #salt
07:51 irated joined #salt
07:55 exarkun joined #salt
08:09 aldevar joined #salt
08:10 pualj joined #salt
08:16 pualj joined #salt
08:20 Hybrid joined #salt
08:22 pualj joined #salt
08:28 pualj joined #salt
08:29 pualj joined #salt
08:30 pualj joined #salt
08:40 vali joined #salt
08:41 vali morning
08:44 vb29 joined #salt
08:44 onslack <mthibaut858> heya
08:44 vb29 left #salt
08:48 mikecmpbll joined #salt
08:50 pualj joined #salt
09:06 mikecmpbll joined #salt
09:09 sol7 joined #salt
09:09 munhitsu__ joined #salt
09:10 GrisKo joined #salt
09:13 pbandark joined #salt
09:15 GnuLxUsr joined #salt
09:18 copelco__ joined #salt
09:18 bdrung_work joined #salt
09:21 mattfoxxx joined #salt
09:24 inad922 joined #salt
09:26 evle joined #salt
09:28 Mattch joined #salt
09:29 petems joined #salt
09:31 mihait joined #salt
09:32 davedash joined #salt
09:35 exarkun joined #salt
09:38 onslack <ryan.walder> I'm guessing thorium doesn't have a gitfs option yet?
09:44 darioleidi joined #salt
09:47 mikecmpb_ joined #salt
09:50 m0nky joined #salt
09:58 Awesomecase joined #salt
10:19 oida joined #salt
10:21 ToeSnacks joined #salt
10:35 aldevar left #salt
10:37 aldevar joined #salt
10:40 pualj joined #salt
10:43 pualj joined #salt
10:44 colttt Hello, I use debian stretch and tab-copmpletion, but when i for example type : salt '*' pkg.<TAB><TAB>, it takes around 25s
10:44 pualj joined #salt
10:44 pualj joined #salt
10:46 colttt is that nromal ?
10:50 babilen That appears to be the case, unfortunately
10:50 * babilen isn't sure why they don't generate completions during the release phase, but apparently do so dynamically
10:55 yujunz joined #salt
11:06 inad922 joined #salt
11:07 Cadmus joined #salt
11:08 sjorge joined #salt
11:08 babilen colttt: Maybe file a bug about that and suggest to use static tab-completion?
11:09 colttt ok..
11:14 exarkun joined #salt
11:26 ipmb joined #salt
11:39 aerbax joined #salt
11:49 viq ryan.walder: correct, there's a ticket about it but nothing in place yet.
11:52 pualj joined #salt
12:16 pualj joined #salt
12:20 pualj joined #salt
12:33 petererer joined #salt
12:33 mike25de joined #salt
12:33 inad922 joined #salt
12:36 mike25de left #salt
12:36 gmoro joined #salt
12:36 mike25de joined #salt
12:41 pugcs joined #salt
12:43 mikecmpbll joined #salt
12:47 pcn We are running a first highstate and trying to track when a node is ready by asking the jobs runner if the node has completed its highstate via jobs.* functions on the master.
12:48 pcn We've noticed that there are times when a job starts, but never has a completed state.
12:48 pcn Like, it just gets lost.
12:56 exarkun joined #salt
13:00 zerocoolback joined #salt
13:05 xet7 joined #salt
13:08 Guest73 joined #salt
13:20 `mist joined #salt
13:21 snakegums joined #salt
13:22 viq pcn: not that it answers your question at all, but I'm recently playing with reactors, so in theory you could be emiting events on the bus, and either watching for them, or program some reactions to them.
13:23 Nahual joined #salt
13:24 onslack <vicviq> @ryan.walder: https://github.com/saltstack/salt/issues/29081
13:27 Hybrid joined #salt
13:32 Woefdram joined #salt
13:36 onslack <ryan.walder> Cheers
13:39 Naresh joined #salt
13:42 pualj joined #salt
13:54 Hybrid joined #salt
13:55 jhill_ joined #salt
14:11 pualj_ joined #salt
14:24 BitBandit joined #salt
14:26 gh34 joined #salt
14:34 exarkun joined #salt
14:37 pcn viq every action on the bus should be recorded in the event cache.  When I look at mine, I don't see events corresponding to the end of a highstate, so I think that a rector wouldn't quite do the job.
14:38 onslack <mts-salt> you have to turn on per-state event reporting, it's not on by default
14:40 pcn The following highstate will almost always show that it's completed when this happens; and it's not consistent.
14:45 ecdhe joined #salt
14:56 pugcs left #salt
14:58 simondodsley joined #salt
15:01 mchlumsky joined #salt
15:03 Nahual Is it common to have the salt-minion running on a salt "super-master" or controller of syndic nodes? Seems I have some odd connectivity behavior when that happens where minions are unable to respond to test.ping but if I launch it from the syndics directly, no trouble. Soon as I killed the minion process on the controller nodes I don't see any issues. I could run salt there in a local mode but I would prefer not. Even if a single controlle
15:03 Nahual r node has salt-minion running and I execute test.ping from the primary controller node, I get connectivity issues.
15:07 _JZ_ joined #salt
15:07 Ryan_Lane joined #salt
15:09 nomad_fr I'm looking for a way to put a priv key in one user dir
15:10 benner how to disable salt-minion logging to file?
15:11 nomad_fr I will do it with file.managed
15:11 cgiroua joined #salt
15:16 onslack <gtmanfred> set log_level: quiet
15:19 Guest73 joined #salt
15:19 viq nomad_fr: if there was a question there then I missed it
15:21 racooper joined #salt
15:21 babilen "How do I ... put a priv key in ..."
15:22 babilen (I guess)
15:22 XenophonF nomad_fr: check out https://github.com/saltstack-formulas/users-formula
15:22 XenophonF by priv key you might mean SSH private key
15:22 viq Also depends what kind of key, since there are also the x509 modules
15:23 viq Or are they now named tls?
15:23 XenophonF users-formula also lets you manage the contents of a users' home directory
15:23 XenophonF or you can DIY using file.managed/file.recurse
15:26 babilen butterfly.managed / magnets.arrange / emacs.dwim
15:26 rockey i went down a rabbit hole yesterday, when trying to use a random formula with salt-ssh.. according to minion, they couldn't find for instance <formula>/map.jinja and other included files within the templating system
15:26 babilen Yeah
15:27 babilen That's a pain in the arse ..
15:27 rockey had to work extensivly with extra_filerefs and "add" files manually
15:27 babilen Which totally shouldn't be necessary
15:27 rockey how come salt-ssh doesn't have the plain support of a regular file server that a master uses?
15:27 rockey yes, pain in the arse would describe my frustration yesterday :)
15:28 onslack <mts-salt> because salt-ssh collects all the files together into a tarball and sends them all over together
15:28 rockey got it to work, but holy hell
15:28 babilen salt-ssh should just collect all files in file_tree and send those
15:28 rockey mts-salt: yes i kinda got that, but why's the system for including files within the tarball limited?
15:28 onslack <mts-salt> i would suggest if it's not behaving the way you'd expect then you could file an issue
15:29 onslack <mts-salt> that i don't know, i'm not a salt developer :)
15:29 rockey there is already plenty of issues, but it doesn't seem that the main issue at hand is taken care of
15:29 rockey for instance: https://github.com/saltstack/salt/issues/26585
15:30 rockey and:
15:30 rockey https://github.com/saltstack/salt/issues/21370
15:31 rockey the last one had me looking into extra_filerefs params, which lead to adding a list to my master file, since salt-ssh doesn't support multiple --extra-filerefs= apparently
15:31 onslack <mts-salt> i'm sure they'd love a pr if you can fix it :D
15:31 rockey oh well, the rabbit-hole went deep :)
15:31 babilen Maybe its a secret plot by Ansible to ensure that Salt's SSH transport remains unusable?
15:31 rockey haha
15:31 rockey maybe
15:32 onslack <mts-salt> heh, and the final comment on that issue is from babilen... :)
15:32 babilen It is indeed
15:32 babilen Well, we use salt-ssh every now and then, but most of our more advanced states are unusable with it and I refure to maintain extra_filerefs
15:33 rockey mts-salt: i'd love too, but guessing it will take a little too much of my time to re-write the include function within salt-ssh :P
15:33 rockey but i'm with babilen, salt-ssh should collect all files in the file_tree and send it
15:33 rockey would've been some sort of solution at least :)
15:33 babilen To me it seems as if its "optimised" in such a way that it tries to figure out which bits to include in the upload rather than simply including the entirety of file_tree
15:34 babilen And they prune files that are needed
15:34 babilen And .. really .. its uploaded once and if you cared about speed you wouldn't use the SSH transport anyway
15:34 onslack <mts-salt> some installations have a massive set of files, only some of which are used on any given minion. i can understand not including everything
15:35 onslack <mts-salt> the issue is more about load, as the master has to create and compress the tarball each time
15:35 babilen I'm all for optimising it, but if you exclude files that are needed you are doing it wrong
15:35 onslack <mts-salt> i'd hate to do that on a 10k minion installation
15:35 babilen I'd rather have a system that works, than a system that doesn't work optimally
15:35 rockey well, maybe don't include the full list of files in every module, but if i'm trying to apply a users-formula, add those files atleast :)
15:36 Nahual Problem was that master_id: was not set on the controlling masters. All good now.
15:36 onslack <mts-salt> the "solution" given by basepi is that individual states can reference jinja files, rather than including them in the master config
15:37 onslack <mts-salt> https://github.com/saltstack/salt/issues/21370#issuecomment-143901228
15:37 rockey babilen: after the experience yesterday with salt-ssh, i'll probably just go for proper master provisioning even on nodes less secure than internal ones
15:37 rockey and bring big ass firewalls instead
15:38 babilen Masterless is another option
15:38 babilen There are a couple of nice talks by the Lift people on how to design that
15:39 babilen Basic idea is that you have a minion that downloads states/pillars from some repo and then runs masterless
15:39 babilen You obviously miss out on the orchestration bit
15:39 rockey hm, and i quite fancy the orchestration bit a whole lot
15:39 babilen Which Lift doesn't care about as they just create new boxes when they need it
15:40 rockey since there is where i do most of my work and business value
15:40 babilen Sure, it's not a perfect solution for everything
15:41 babilen And then .. salt-ssh and ansible aren't really "agentless" either as you require a SSH server on the remote
15:41 onslack <mts-salt> and as someone reported, you need auth that works too :)
15:42 babilen *nod*
15:45 carlwgeorge joined #salt
15:51 Guest73 joined #salt
15:52 evle1 joined #salt
15:52 rockey always good to have it confirmed by others too though
15:53 rockey then i have to rework my way of thinking, since salt-ssh doesn't work properly with my use-cases
15:53 onslack <mts-salt> you can add extra_filerefs to any state that needs them
15:54 rockey you mean in runtime add '--extra-filerefs='?
15:55 onslack <mts-salt> no, in the sls itself
15:55 onslack <gtmanfred> in oxygen, you will be able to target salt-ssh minions from your regular salt master
15:55 onslack <mts-salt> it can be provided as kwargs to any state, says basepi
15:55 raptelan joined #salt
15:55 onslack <mts-salt> is this issue fixed in oxygen then? :)
15:56 onslack <gtmanfred> what issue?
15:56 onslack <mts-salt> https://github.com/saltstack/salt/issues/21370
15:56 rockey mts-salt: seems like really annoying work tbh, since kinda complex templating could have several imports and the filerefs list could be very long
15:56 onslack <mts-salt> jinja files not being included in salt-ssh tarball
15:56 onslack <gtmanfred> no, you would still need to add `extra_filerefs=salt://` for it to work
15:57 rockey yes, it only includes what's mentioned within salt:// links, but if that file imports another template, it will bork out
15:57 onslack <mts-salt> then salt-ssh will remain crippled for complex cases, which is the bulk of this discussion
15:58 onslack <gtmanfred> that is fine, but at least in oxygen you won’t have to use two seperate commands
15:58 rockey ... and also applied on your "how to write formulas" best practices for more advance formulas tbh
15:58 onslack <mts-salt> i can only conclude that if you need to use salt-ssh then you have to design your states differently to try to avoid these cases, or put up with needing to add the extra_filerefs
15:59 rockey the additional work for using salt-ssh with formulas listed on gh.com/saltstack-formulas for instance is kinda a long list, which is undocumented in a fashion which is easily accessable
16:00 rockey mts-salt: i agree with you, have to re-think when applying cases where salt-ssh might be an option
16:00 onslack <mts-salt> my own setup doesn't lean on formulas - yet. we've rolled everything we need from the ground up
16:01 rockey formulas is a nice way of not having to re-invent the wheel time and time again :)
16:01 onslack <mts-salt> and we also don't use salt-ssh. so i'm just participating to try to ensure i understand what to avoid :)
16:02 rockey i recommend trying it out, just not with salt-ssh :)
16:02 onslack <mts-salt> so far we haven't come across a use case for it, but it's something i'll bear in mind. likewise thorium and orchestration
16:14 exarkun joined #salt
16:16 mikecmpbll joined #salt
16:18 edrocks joined #salt
16:19 Guest73 joined #salt
16:19 nixjdm joined #salt
16:21 XenophonF rockey: if you want more salt-ssh pain, try using GPG-encrypted Pillar data
16:23 coredumb doesn't look so fun
16:25 Guest73 joined #salt
16:34 evle1 joined #salt
16:36 frdy joined #salt
16:53 jrenner left #salt
16:55 heaje joined #salt
16:57 jrenner joined #salt
17:05 Nahual Well, issue is not resolved as much as I would like it to be. Appears that syndic_forward_all_events is creating duplicate returns or doesn't return at all. If I disable that and leave master_id set, everything works accordingly but of course, jobs are only returned to the requesting master.
17:09 DammitJim joined #salt
17:10 DammitJim would a grain or a pillar tell me the version of a software package installed?
17:10 DammitJim I want to make a decision on a state file to use a particular conf file (file.managed) for different versions of mysql
17:10 babilen That's normally tied to the release
17:10 DammitJim release of the OS?
17:10 babilen You can check the mysql-formula on how to do that for mysql/mariadb
17:10 DammitJim lst_release grain?
17:10 babilen (which remdinds me that I should PR my changes for mariadb)
17:11 Cadmus Hello, I have a question of style. I have a big orchestrator which does things over a cluster. I now need to do a single node version, should I just write a second orchestrator where each bit has the same single target or is there a cleverer way?
17:11 babilen oscodename et al
17:11 babilen Cadmus: Doesn't sound too wrong
17:12 Cadmus DammitJim: In a perfect world you'd set the version you're working on somewhere, and have states that ensure that version *is* the one installed, and configure to suit. Ofc course that's harder when you're salting an existing envrionment
17:12 Cadmus babilen: Thanks :)
17:12 DammitJim Cadmus, exactly
17:12 DammitJim I think I'm going to go OS route for now
17:12 DammitJim thanks!
17:12 babilen In the end, you use salt to "do the right thing" in a prescriptable manner, not a reactive one (i.e. you tell it to use version foo, rather than react to version foo being available)
17:13 babilen prescriptivist rather :)
17:15 babilen And as long as you are deploying on defined environments (i.e. distribution releases) there would be only a single version anyway
17:15 DammitJim prescriptivist
17:15 babilen The question is: Do you want that?
17:15 DammitJim I like that title
17:15 DammitJim I don't know if I want that
17:15 babilen It's a term used in linguistics
17:16 DammitJim but with the limited knowledge I have at this time, I think I have to move forward
17:16 DammitJim funny how sometimes it's just not possible to have the big picture and that kinda stuff gets in the way of getting something done
17:16 DammitJim yes, it is configuration debt later on, though
17:16 babilen (as in: "You can't say that, it's wrong. We clearly define what's a language and what isn't" vs. "I can say what I want and language is what people use")
17:18 babilen To get back to the version thing
17:18 DammitJim si
17:18 aldevar left #salt
17:19 babilen For me: If I deploy a system on a VM that's based on a trusted distribution that has specific releases: I ensure that $SOFTWARE is configured correctly and just use the version that comes with that release
17:19 babilen If I need a different version I need a different release
17:20 babilen If people want to fine tune their environment to the point that point releases of software is specified, I feel that container based/microservice deployments are better suited (mostly on k8s)
17:20 bluenemo joined #salt
17:20 babilen I personally don't like the "Water down the distribution by including a bazillion third party repositories with a gazillion versions in them that haven't been tested in the context of other software in that release" approach
17:26 XenophonF amen
17:30 pipps joined #salt
17:35 jeffspeff joined #salt
17:36 jeffspeff what is the command to update custom modules on the minion? I know that running a highstate will do it, but I think I remember another command that will just update the modules from the master.
17:36 onslack <gtmanfred> saltutil.sync_modules
17:36 jeffspeff thank you
17:36 onslack <gtmanfred> https://docs.saltstack.com/en/latest/ref/modules/all/salt.modules.saltutil.html
17:38 keltim joined #salt
17:48 tiwula joined #salt
17:50 Fladi joined #salt
17:55 exarkun joined #salt
17:59 oida joined #salt
18:00 skeezix-hf joined #salt
18:01 edrocks joined #salt
18:01 pipps joined #salt
18:02 jeffspeff I'm working on a module to activate ms office. here's the code https://gist.github.com/jeffclay/0f16f7d76292963b7d8229ce53c37614  when I run "salt.MINION msolicense.manual_kms_host host=ABC" I get the following error: Input Error: There is no file extension in "c:\Program". where am i messing up at?
18:03 onslack <gtmanfred> are you probably need to quote the c:\Program Files path inside the cmd.run
18:03 onslack <gtmanfred> ` set_kms_host_cmd = r'cscript {0}\ospp.vbs /sethst:{1}'.format(office_path(), host)`
18:03 onslack <gtmanfred> actually
18:04 jeffspeff set_kms_host_cmd = r'cscript "' + office_path() + '"\ospp.vbs /sethst:{0}'.format(host)    ?
18:04 onslack <gtmanfred> `set_kms_host_cmd = r'cscript "{0}"\ospp.vbs /sethst:{1}'.format(office_path(), host)`
18:04 onslack <gtmanfred> yeah, you could do that
18:04 onslack <gtmanfred> but i would put the second ” after ospp.vbs
18:04 onslack <gtmanfred> `set_kms_host_cmd = r'cscript "{0}\ospp.vbs" /sethst:{1}'.format(office_path(), host)`
18:04 onslack <gtmanfred> that one
18:04 onslack <gtmanfred> same thing in activate and other places
18:05 jeffspeff ok, thanks
18:05 onslack <gtmanfred> well, i guess just activate
18:07 pipps joined #salt
18:08 jeffspeff gtmanfred, i updated the gist code. I get this error now  Input Error: There is no file extension in "C:\c:\Program".
18:09 zerocoolback joined #salt
18:10 onslack <gtmanfred> well that is fun.  i don’t know
18:10 jeffspeff lol
18:11 onslack <gtmanfred> but the problem is path handling…  also worth mentioning, the windows cmd prompt is smart enough to know / is C:\… so you could just switch your stuff to use /Program Files/ https://gist.github.com/gtmanfred/5218d7e64b08ce19531248b6f21f5831
18:11 onslack <gtmanfred> File uploaded https://saltstackcommunity.slack.com/files/U7KH2HJ0K/F97NELAG3/gistfile1.txt / undefined
18:19 wongster80 joined #salt
18:20 pipps joined #salt
18:23 pipps joined #salt
18:26 K0HAX joined #salt
18:39 edrocks joined #salt
18:43 v0rtex joined #salt
18:46 doubletwist What are the advantages of running envs on a single master vs having a separate master for each env [dev/test/prod]
18:47 XenophonF complexity?
18:54 Cluepon single repo with a shared base env?
18:57 pipps joined #salt
19:00 doubletwist Well, the way I'm working on would have a single repo anyway, just have each env checked out to "release-${env}" branch
19:02 pipps99 joined #salt
19:02 cgiroua_ joined #salt
19:08 ensbro joined #salt
19:18 doubletwist That said, while I'm not using gitfs yet - I might consider it. But reading info on it, I'm not clear how to get gitfs to use a specific branch
19:18 pipps joined #salt
19:21 whytewolf doubletwist: specific branch for what?
19:21 pipps joined #salt
19:23 whytewolf https://docs.saltstack.com/en/latest/topics/tutorials/gitfs.html#per-saltenv-configuration-parameters that?
19:23 ymasson joined #salt
19:32 doubletwist yeah
19:34 pipps joined #salt
19:35 exarkun joined #salt
19:36 cyborg-one joined #salt
19:39 whytewolf doubletwist: don't see what is difficult about it.
19:39 Guest73 joined #salt
19:40 pipps joined #salt
19:47 pipps joined #salt
19:47 aldevar joined #salt
19:50 pipps99 joined #salt
19:50 hammer065 joined #salt
19:51 doubletwist whytewolf: nothing now that I've seen that :)
19:51 doubletwist I got ahead of myself though. Right now I'm trying to get git_pillar working
19:56 edrocks joined #salt
19:57 inad922 joined #salt
20:01 Vaelatern joined #salt
20:04 lordcirth_work Suppressing comments in diffs, for all/most formats, is surprisingly difficult.  I'm considering adding an arg 'hide_regex' and letting people specify their own regex per file :P
20:10 MTecknology considering the massively different options that exist for comments... I don't see how you could do it any other way
20:11 MTecknology In some configs, a semi-colon means end of directive/options, in some it means begin a comment
20:13 major joined #salt
20:16 doubletwist hrm. Seeing auth failure for git_pillar - using the same ssh keypair that I can use to manually clone the repo: https://paste.fedoraproject.org/paste/GRVXE75~HAaakkWDpysgSQ
20:17 doubletwist Am I missing something?
20:21 doubletwist same issue for gitfs_remotes.conf
20:22 MTecknology another pastebin service that renders empty white pages when javascript is disabled, eh?
20:23 whytewolf doubletwist: what git library do you have installed?
20:27 stooj Is anyone able to point to where I am going wrong using lxc with salt-cloud? I'm using the saltstack formula; here's the relevant pillar excerpt, the command and the output. https://dpaste.de/fYG6 I don't understand why I need an image, a profile *and* a template when they are mututally exclusive afaik, and I *think* I've provided a template and a profile to use
20:27 pipps joined #salt
20:32 doubletwist python-pygit2
20:34 Vaelatern joined #salt
20:35 MTecknology doubletwist: there's extra docs for pygit2 (namely, auth)
20:36 lordcirth_work MTecknology, oh I was always going to let them specify the comment string, just not the whole regex.  But I figured it out, I think!
20:38 pipps joined #salt
20:38 doubletwist MTecknology: other than here? https://docs.saltstack.com/en/latest/topics/tutorials/gitfs.html#id5
20:39 pipps99 joined #salt
20:42 doubletwist maybe I'd be better off using gitpython?
20:46 pipps joined #salt
20:49 pipps99 joined #salt
20:50 pipps_ joined #salt
20:56 pipps joined #salt
20:57 pipps99 joined #salt
21:00 keltim joined #salt
21:00 pipps_ joined #salt
21:03 onslack joined #salt
21:03 XenophonF i heartly endorse gitpython
21:06 whytewolf I endorse what ever works for you... I personally don't have issues with pygit2. but i hand compile the dependencies myself and pip install instead of os install it.
21:14 leev joined #salt
21:16 exarkun joined #salt
21:22 oida joined #salt
21:28 pipps joined #salt
21:29 mrueg joined #salt
21:35 sjorge joined #salt
21:37 doubletwist I just wish I could see more detail about what it's trying to do that's not working
21:37 whytewolf salt-run fileserver.update -l trace
21:37 doubletwist I'm not even getting to that point
21:37 whytewolf or salt-run git_pillar.update -l trace
21:38 doubletwist just starting the salt-master service logs that error
21:38 whytewolf try it
21:39 cyteen joined #salt
21:40 whytewolf those commands tell the master to try updating or fetching. so even if the master can't at the start those commands should help in debugging why not
21:43 lordcirth_work modules/file.check_file_meta() takes both 'sfn' and 'source' arguments, and the example only shows it taking one.  What's the difference?
21:45 doubletwist whytewolf: results of the salt-run git_pillar.update -l trace: https://paste.fedoraproject.org/paste/1jQg28WTiD1TehVO~h-HWg
21:47 doubletwist And here's what it does if I just do a git clone from the command line as root: https://paste.fedoraproject.org/paste/Ih~w80kKRIVpKhF8OTBr9A
21:50 lordcirth_work Ah, it seems like 'sfn' is the file path to the local cache of the file from 'source'.
21:51 whytewolf doubletwist: what branch are you trying to clone off the git repo?
21:51 whytewolf that needs to go before the repo
21:52 whytewolf such as - master ssh://git@bitbucket.example.com:7999/saltproject/pillar.git
21:52 doubletwist release-dev but the docs say that's not needed if you set git_pillar_branch [and/or] git_pillar_base - I can't remember which but they are both set
21:52 doubletwist but I'll try it
21:53 doubletwist same error
21:56 whytewolf is your master running as root? and are the permissions set to 600 on the key and cert? [and the directory set to 700]
21:57 doubletwist yes and yes
21:58 whytewolf did you read this? https://docs.saltstack.com/en/latest/topics/tutorials/gitfs.html#id1
22:01 doubletwist yup
22:03 whytewolf have you tried GitPython
22:03 doubletwist not yet
22:03 doubletwist docs said pygit2 is preferred so I though I'd make sure I'm not missing something stupid before moving on
22:04 Hybrid joined #salt
22:06 whytewolf well I have no idea what is happening in your enviroment. pygit2 works for me. but it can be a pain to setup.. GitPython should also work but you need to understand what it is doing. Since it actually calls the git exacutable instead of having a library link into it
22:07 xet7 joined #salt
22:14 Guest73 joined #salt
22:15 doubletwist argh
22:15 doubletwist trying gitpython
22:15 doubletwist Now I'm getting: "git_pillar_remote 'ssh://git@bitbucket.example.com:7999:/saltproject/pillar.git requires authentication, but no authentication configured"
22:16 doubletwist I set up /root/.ssh/config as noted in the docs
22:17 Guest73 joined #salt
22:21 mbologna joined #salt
22:23 Hybrid joined #salt
22:23 mbologna joined #salt
22:25 matti joined #salt
22:25 matti joined #salt
22:30 doubletwist Interesting. It works for formulas but not for the pillar
22:30 doubletwist It's a start at least
22:39 doubletwist So am I imagining things or is using gitfs for formulas "OhMyGodSlow" ?
22:40 pipps joined #salt
22:50 doubletwist like state.apply has gone from a few seconds per host to 3+ minutes per host. That's not going to work
22:51 cgiroua joined #salt
22:52 dfinn joined #salt
22:55 Guest73 joined #salt
22:56 dfinn joined #salt
22:56 exarkun joined #salt
23:06 pipps99 joined #salt
23:21 aviau joined #salt
23:23 * MTecknology kinda likes pygit2, but gitpy has been enough for me.
23:24 MTecknology granted.. I kinda have some hacks in place that I no longer need if I switch to pygit2  :S
23:26 MTecknology doubletwist: I always suggest avoiding forumulas for anything but trial/demo.
23:27 mikecmpbll joined #salt
23:34 doubletwist MTecknology: Why? They seem exceedingly useful to me.
23:38 XenophonF doubletwist: I use salt-formula to manage my Salt Masters' configs.  It includes an SLS that manages formulas.  A complete example config is at https://github.com/irtnog/salt-pillar-example/tree/master/salt/example/com/
23:38 whytewolf a. because it is better to learn how to build things like them on your own. b. because a lot of them have a lot of bloat if what you are doing is a simple install, configure, start service C. because they are really generic and might not actually fit your use case and it is better to build something you need instead of trying to get it to fit your edge cases
23:38 Aikar joined #salt
23:38 Aikar joined #salt
23:39 copec I don't know if anyone cares, but on https://docs.saltstack.com/en/getstarted/fundamentals/states.html the code snippits say state.apply but the terminal screengrabs say state.sls
23:40 XenophonF <luke> I care! </luke>
23:40 pipps joined #salt
23:40 XenophonF copec: you should file a doc bug in https://github.com/saltstack/salt
23:41 copec okay
23:41 whytewolf there is a note there #state.apply was added in 2015.5, so if you are using an earlier version call state.sls instead.
23:47 copec well, that was dumb of me
23:47 copec sorry
23:47 copec I glanced the beginning of the note and didn't scroll the snippit window right
23:48 whytewolf no problem. it happens. live and learn
23:49 XenophonF is 2015.5 still supported?
23:49 XenophonF that's pretty old...
23:49 whytewolf nope
23:50 doubletwist whytewolf: It sounds like you're suggesting to use formulas just not the ones on github salt-formulas?
23:50 oida joined #salt
23:52 whytewolf doubletwist: I"m saying there is no right or wrong answer. you have to decide. but if you use it. know how it works.
23:58 Guest73 joined #salt

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