Perl 6 - the future is here, just unevenly distributed

IRC log for #salt, 2018-05-09

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

All times shown according to UTC.

Time Nick Message
00:00 onslack joined #salt
00:00 obimod extend is actually what was causing my original issue
00:00 hemebond What was your original issue?
00:01 obimod [{'environment': 1}].extend([{'environment': 2}]) == [{'environment': 1}, {'environment': 2}]
00:02 hemebond No, not the Python extend method.
00:03 hemebond The Saltstack state "extend"
00:03 hemebond It lets you override/update a state.
00:03 obimod 8)
00:04 Pjuzur joined #salt
00:06 hemebond I'm surprised slsutil.merge doesn't work.
00:07 hemebond The module suggests it would work properly for states, but really the docs just mention data structure.
00:11 obimod It probably works fine for states since OrderedDict is used in the background... I think my issue is not using OrderedDict in the foreground
00:11 obimod OrderedDict is an instance of collections.Mapping, which update/merge will recurse into
00:16 eseyman joined #salt
00:17 hemebond Your issue isn't that update/merge isn't recursing.
00:17 hemebond Your issue is that your require list contains "duplicate" entries (according to the way states work)
00:18 obimod right, because i'm manually creating OrderedDict-like lists https://pastebin.com/3tBMc2Lc
00:18 obimod If I wrap those suckers with OrderedDict() I think it will work...
00:19 hatifnatt joined #salt
00:19 mauli joined #salt
00:19 Udkkna joined #salt
00:20 ebbex joined #salt
00:20 Miuku joined #salt
00:21 oida joined #salt
00:21 zerocoolback joined #salt
00:21 obimod ordereddict takes in a dict #facepalm
00:23 obimod i was thinking in terms of yaml ('-' denoting a list item) and writing states as lists of dicts
00:24 MTecknology what's your actual end-goal here?
00:27 zwliu joined #salt
00:29 hemebond MTecknology: Trying to extend a state. Didn't say why extend didn't work.
00:29 mavhq joined #salt
00:30 obimod just learned of the extend state functionality... wish i knew of that before embarking on this monolithic docker-setup state lol
00:30 hemebond Ah
00:31 obimod that being said, i think my problems are solved... i was using lists instead of dicts
00:31 hemebond But... states use lists.
00:31 obimod i thought they used ordereddicts?
00:31 hemebond For the dicts, sure.
00:32 obimod i'm returning an object from a #!py state
00:32 obimod keys being the sls_ids of a bunch of generated states
00:33 hemebond Within the state are lists.
00:33 obimod shoot
00:33 hemebond And that's what was tripping you up.
00:34 hemebond And why I don't really understand the point of slsutil.merge if it's not going to merge a state properly.
00:40 stooj joined #salt
00:48 obimod i'm essentially traversing a machine-service type configuration which expands into multiple docker services per machine... this state is pretty massive and injects things like ssl requirements. i'm a bit skeptical of the extend state being a good drop-in replacement since it could only be declared once (per docs)
00:55 dendazen joined #salt
01:00 hemebond Sounds like a refactoring is in order.
01:04 MTecknology ^ +1
01:07 gmoro_ joined #salt
01:30 obimod probably, although at this point i think i'll make a couple of changes to dictupdate.py until there's more time to plan out the refactor
01:42 wryfi joined #salt
01:55 dvdmuckle joined #salt
01:55 Psi-Jack Hmm, is there a better way to include minion_id pillar state files than using, in the top.sls, hosts.{{ grains.id }}, which ends up translating to hosts/hostname/domain/tld.sls ?
01:55 Psi-Jack I'd rather it be more like hosts/hostname.domain.tld.sls
01:57 ilbot3 joined #salt
01:57 Topic for #salt is now Welcome to #salt! <+> Latest Versions: 2017.7.5, 2018.3.0 <+> 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
01:59 shiranaihito joined #salt
02:05 hemebond ?
02:16 Psi-Jack And I should be able to replace, in jinja templates, {{ pillar['backups']['source_directories'] }} with {{ pillar.backups.source_directories }}, right?
02:17 MTecknology are you asking about python syntax?
02:18 Psi-Jack If that's the case?
02:23 zerocoolback joined #salt
02:24 Psi-Jack Yep, seems to work as expected. :)
03:16 Kevin joined #salt
03:21 zerocoolback joined #salt
03:43 Psi-Jack Now, trying to get gitfs working, but for some odd reason, git://gitea.home.ld/name/repo.git is giving Connection refused.
03:44 hemebond left #salt
03:47 MTecknology I think this might be the wrong channel for that sort of question. Perhaps #gitea or #git or ##linux?..
03:48 Psi-Jack This is a problem with salt not being able to connect to a server.
03:48 MTecknology You got a connection refused from your git service...
03:48 Psi-Jack That provides http and git (over ssh).
03:49 Psi-Jack I can git pull just fine, from the very same host, to the very same server.
03:49 Psi-Jack That I run.
03:51 Psi-Jack I'd assume that pygit2, when using a git:// URI, it would use ssh, but it does not seem to be doing so.
03:52 MTecknology I'd assume that's probably the wrong url
03:52 Psi-Jack It's not.
03:53 MTecknology I still assume it's wrong
03:53 Psi-Jack You're still wrong. ;)
03:53 MTecknology I'm not wrong...
03:56 Psi-Jack So, what's wrong about it then?
03:58 * MTecknology sighs
03:59 hemebond joined #salt
04:00 Psi-Jack I've tried ssh://git@gitea.home.ld/name/repo.git as well, that at least gives an authentication failure callback, despite providing the pubkey and privkey in the master config.
04:02 MTecknology and that works when you git clone ssh://git@gitea.home.ld/name/repo.git?
04:02 Psi-Jack Yes
04:02 MTecknology show me
04:04 Psi-Jack https://paste.linux-help.org/view/ca063d37
04:06 Psi-Jack salt-repo.git specifically is a private repo so, accessing it requires authentication. I've provided the privkey for that purpose in the gitfs_pubkey and gitfs_privkey settings in the master config.
04:07 Psi-Jack May 09 00:07:08 salt salt-master[25050]: [ERROR   ] Error occurred fetching gitfs remote 'git@gitea.home.ld:Linux-Help/salt-repo.git': Failed to authenticate SSH session: Callback returned error
04:07 MTecknology looks like a different url to me
04:08 zerocoolback joined #salt
04:16 Psi-Jack ...
04:18 Psi-Jack Bit too literal, dontcha think? :p
04:29 obimod MTecknology: hemebond: thanks for the help earlier, i'm overriding dictupdate.py for now with https://pastebin.com/q84Mv5Uv (:65) -- i'm not yet convinced the patterns i've put into place which needs the merge are tech debt, but time will tell
04:30 MTecknology I get the impression most of your setup is probably tech debt..
04:30 obimod i totally see how you get that impression, especially coming from where you are
04:31 obimod it probably is in the long run
04:31 tiwula joined #salt
04:31 obimod i guess that's the definition of tech debt, but the time saving is worth it for now
04:32 MTecknology "for now"
04:33 obimod totally
04:33 obimod :)
04:33 * MTecknology sighs
04:34 hemebond obimod: Yikes :-)
04:35 obimod lol
04:38 Psi-Jack I wonder if this is a pygit2 api change, since pygit2 0.23.2 changes clone_repository
04:39 Psi-Jack Man, pygit2 really makes a lot of breaking API changes. :/
04:40 obimod hemebond: working like a charm :)
04:41 obimod albeit a bit nuclear in nature
04:42 obimod (wishing i chose a different word)
04:45 MTecknology sounds about appropriate
04:48 Psi-Jack Ugh.. even https fails. [ERROR   ] Error occurred fetching gitfs remote 'https://git.linux-help.org/Linux-Help/salt-repo.git': Failed to resolve address for https: Name or service not known
04:49 Psi-Jack So.. Basically. Salt's gitfs is broken in Debian 9 (Stretch), because of pygit2.
04:51 Psi-Jack Well, authentication-wise, it's broken anyway.
04:53 MTecknology ya... except for all those people that use debian 9 w/ pygit2
04:53 MTecknology a blanket "it's broken" seems rather inaccurate
04:53 stooj joined #salt
04:53 Psi-Jack Well, comparing code with salt, and comparing the API changes with pygit2, that's not plausible.
04:55 MTecknology I have two environments running deb9 w/ salt using pygit2, one is py2 and one is py3
04:55 obimod mmm alder wood smoked salt, himalayan pink salt, super efficient recursion code chunks, all good things
04:58 Psi-Jack Well, this is what I have, https://paste.linux-help.org/view/87e8180a for a version report. Current salt 2018.3.0, pygit2 0.24.2
05:03 MTecknology You already shared versions so that isn't terribly helpful. Logs 'n such would be.
05:04 tiwula joined #salt
05:05 Psi-Jack I might've finally gotten something to work with gitpython and ssh_config methods.
05:09 MTecknology gitpython is typically the easier choice, but is much less flexible
05:16 Psi-Jack Yeah. Hmm.. Well, that didn't throw errors... Just salt state.apply gets a message about the salt request timing out from the master. :/
05:17 zerocoolback joined #salt
05:18 Psi-Jack And the gitfs cache dir is seemingly empty
05:18 Psi-Jack Ahh, nevermind it does have it, just bare.
05:30 zerocoolback joined #salt
05:33 zerocoolback joined #salt
05:41 zerocoolback joined #salt
05:42 inad924 joined #salt
05:44 hoonetorg joined #salt
05:50 obimod Psi-Jack: familiar with pdb? you can `import pdb; pdb.set_trace()` where the errors are being thrown to try different things and see what the local scope is like
05:52 eseyman joined #salt
06:00 Psi-Jack Not very in that regard. And hmmm... my salt-master is now just completely failing to work. :/
06:00 Psi-Jack Or, it seems to be.
06:17 sauvin joined #salt
06:21 dxiri joined #salt
06:23 Psi-Jack Eggselent. Finally got things seemingly actually working.
06:26 dxiri joined #salt
06:40 briner_ joined #salt
06:41 briner joined #salt
06:43 briner joined #salt
07:05 Hybrid joined #salt
07:11 Hybrid joined #salt
07:18 Pjusur joined #salt
07:20 Tucky joined #salt
07:23 sjl_ joined #salt
07:26 orichards joined #salt
07:35 xet7 joined #salt
07:35 defsdoor joined #salt
07:38 zerocoolback joined #salt
07:39 jhauser joined #salt
07:44 Elsmorian joined #salt
07:50 inad924 joined #salt
07:54 rollniak joined #salt
08:00 CrummyGummy_ joined #salt
08:04 Hybrid joined #salt
08:05 DanyC joined #salt
08:05 Elsmoria_ joined #salt
08:09 ddg joined #salt
08:15 inad924 joined #salt
08:17 Mattch joined #salt
08:18 bantone joined #salt
08:21 Elsmorian joined #salt
08:25 dxiri joined #salt
08:28 ponyofdeath joined #salt
08:42 ^Jamssi^ joined #salt
08:56 zerocoolback joined #salt
09:02 Garo_ joined #salt
09:04 inad924 joined #salt
09:06 stooj joined #salt
09:07 ecdhe joined #salt
09:15 Elsmorian joined #salt
09:16 mikecmpbll joined #salt
09:18 mikecmpbll can you use state.highstate with a target? i'm getting "Passed invalid arguments to state.highstate: 'bool' object is not iterable"
09:19 MTecknology define "target"?
09:20 mikecmpbll salt 'foobar*' state.highstate
09:20 MTecknology that'll work fine
09:20 mikecmpbll hmm. well it's not.
09:23 defsdoor joined #salt
09:23 MTecknology then something else that isn't the command is broken
09:30 mikecmpbll yeah, your confirmation confirms that. curious error though.
09:51 megamaced joined #salt
09:53 arlyon joined #salt
09:53 inad924 joined #salt
09:54 briner joined #salt
10:14 afics joined #salt
10:57 XenophonF the win_certutil module is super disappointing
10:57 XenophonF Windows needs more love from SaltStack
10:59 mchlumsky_ joined #salt
11:06 XenophonF like, I don't just want to install a certificate
11:06 XenophonF I need to be able to install private keys too
11:06 hemebond XenophonF: Why wouldn't you just use Windows stuff?
11:07 hemebond Not saying it couldn't use some love, but I personally would use the Windows eco.
11:09 hemebond I suppose you could be managing a standalone...
11:23 briner joined #salt
11:24 Jan joined #salt
11:35 JanHeidbrink joined #salt
11:40 XenophonF hemebond: So I've just got certbot running on my Salt Master.
11:40 XenophonF And that's handling all certificate request/renewal/deployment.
11:40 XenophonF I'm using DNS-01 validation, a deploy-hook script that fires a Salt event, etc.
11:40 XenophonF It's pretty sweet.
11:41 hemebond Certs for what?
11:41 hemebond (sounds like more than just web servsrs)
11:41 XenophonF What Windows stuff would I use to replicate that functionality?
11:41 XenophonF Web servers, database servers, terminal servers, domain controllers, etc.
11:42 XenophonF Oh, I left out SharePoint and AD FS.
11:42 hemebond You're using certbot for all that?
11:42 XenophonF Yup
11:42 XenophonF I can get certs for everything now.
11:43 hemebond I've been away from Windows for a few years now but I'm pretty sure Windows things can do cert stuff for all that.
11:43 hemebond I'd have to ask a colleague for details.
11:43 XenophonF Nope.
11:43 XenophonF certbot doesn't support Windows
11:43 XenophonF I tried.
11:44 hemebond But they deploy enterprise networks and cloud stuff with certs everywhere.
11:44 dendazen joined #salt
11:44 hemebond Using certbot is the part that seems odd to me.
11:44 XenophonF There's ACMESharp, but it didn't have plugins for DNS-01 validation last I checked.
11:44 XenophonF And it didn't support renewals OOTB.
11:44 JanHeidb1ink joined #salt
11:45 hemebond What does ACMESharp do?
11:45 JanHeidb1ink Hi, I have a question regarding my pull-request https://github.com/saltstack/salt-bootstrap/pull/1226: I've opened the pull-request two days ago and the status of the ci/lint check still is "Expected — Waiting for status to be reported". I guess it shouldn't take so long usually?
11:45 XenophonF It's an ACME client like certbot, only written in .NET.
11:45 XenophonF well, PowerShell
11:46 XenophonF thing is, win_certutil is pretty deficient
11:46 hemebond Is it... you need a publicly accepted CA?
11:47 XenophonF How else am I going to put a certificate on a domain controller that's trusted by other computers?
11:47 hemebond Are the other computers not on the domain the domain controller runs?
11:47 XenophonF They are.
11:48 hemebond Then I'm still confused.
11:48 XenophonF But then I'd have to operate my own CA (possible).
11:48 hemebond I thought you could just create a cert on your CA and push it out with group policy.
11:48 hemebond I thought a domain already had a CA since it has kerberos.
11:49 XenophonF Let me start over.
11:50 zerocoolback joined #salt
11:50 XenophonF In the specific case of my domain controllers, I need LDAPS (never mind why, just trust me).  To do LDAPS the DC needs a certificate installed that it trusts.  In turn my LDAP clients need to trust that cert too.  Not all of them run Windows.  It's a lot easier on me if I get a cert from a CA everyone trusts, like Let's Encrypt.
11:50 XenophonF so in that scenario, I'm only deploying one cert on one server, the DC
11:51 XenophonF if I went down the deploy-my-own-CA route, I'd have to (a) operate a whole bunch of additional infrastructure and (b) deploy at least the CA cert across all LDAP clients, which again aren't all Windows (so I can't depend on CA cert deployment via AD)
11:52 XenophonF I have all the certs stored on the Salt Master and can push them out like any other config.
11:52 XenophonF Except win_certutil can't handle private keys.
11:54 hemebond Right.
11:55 XenophonF what I need is a proper wrapper around either certutil.exe or Import-Certificate
11:55 XenophonF something that can handle private keys
11:55 XenophonF well, Import-PfxCertificate
11:58 briner_ joined #salt
12:09 XenophonF I'm going to write my own Windows certificate installation states.
12:09 XenophonF With blackjack!
12:09 XenophonF And hookers!
12:15 Nahual joined #salt
12:28 sdfdsf joined #salt
12:53 AngryJohnnie joined #salt
12:54 Pjusur joined #salt
13:00 cgiroua joined #salt
13:04 justan0theruser joined #salt
13:19 stooj joined #salt
13:33 nixjdm joined #salt
13:36 DammitJim joined #salt
13:36 AngryJohnnie joined #salt
13:48 dendazen joined #salt
13:59 inad924 joined #salt
13:59 obimod joined #salt
13:59 Elsmorian joined #salt
14:15 justanotheruser joined #salt
14:16 cgiroua joined #salt
14:20 inad924 joined #salt
14:24 dxiri joined #salt
14:26 dxiri_ joined #salt
14:30 DammitJim joined #salt
14:32 ddg joined #salt
14:35 briner joined #salt
14:36 awerner joined #salt
14:36 lionel_ left #salt
15:04 briner joined #salt
15:04 awerner joined #salt
15:05 DanyC joined #salt
15:12 briner joined #salt
15:13 hemphill joined #salt
15:39 tiwula joined #salt
15:43 tys101010 joined #salt
15:49 dxiri joined #salt
15:49 openstacking_123 joined #salt
16:10 AngryJohnnie joined #salt
16:16 dxiri_ joined #salt
16:18 dxiri_ joined #salt
16:23 sjorge joined #salt
16:28 DanyC joined #salt
16:31 DanyC_ joined #salt
16:40 masber joined #salt
16:44 AngryJohnnie joined #salt
16:50 AngryJohnnie joined #salt
16:53 JacobsLadd3r joined #salt
16:56 sjorge joined #salt
17:00 JacobsLadd3r joined #salt
17:04 shoogz joined #salt
17:16 NEOhidra joined #salt
17:31 mikecmpbll joined #salt
17:31 jeffspeff joined #salt
17:34 AngryJohnnie joined #salt
17:37 mchlumsky joined #salt
17:37 toanju joined #salt
17:39 inad924 joined #salt
17:43 mchlumsky joined #salt
17:53 JacobsLadd3r joined #salt
18:09 heaje joined #salt
18:12 druonysus joined #salt
18:23 ymasson joined #salt
18:53 fnords joined #salt
18:57 defsdoor joined #salt
19:07 XenophonF OK, interim approach is to use file.managed to deploy .pfx (PKCS#12) files to c:\salt, since that's not world-readable, and have a cmd.run state that runs certutil -importpfx on changes to the file.managed state
19:08 XenophonF I can generate the .PFX files in Pillar on the fly with cmd.run, I guess.
19:08 XenophonF ugly but should be good enough for now
19:15 viq_ Hm, so I have network.default_route which among other things has which interface this goes through. How would I chain it with network.interface_ip to push to mine "default ip" of hosts, with idea of automating DNS configuration?
19:26 briner joined #salt
19:43 noobiedubie joined #salt
19:48 noobiedubie joined #salt
19:48 rollniak joined #salt
20:02 cgiroua joined #salt
20:16 justanotheruser joined #salt
20:28 DanyC joined #salt
20:31 DanyC joined #salt
20:41 AngryJohnnie joined #salt
20:59 DanyC joined #salt
20:59 Edgan viq: That questions seems like multiple questions. Is one question how do I get the default ip of a box?
21:02 Edgan viq: There are many ways you could group systems to automate dns configuration. envs, clusters, accounts, regions
21:02 viq Edgan: for me it's network.interface_ip {{ network.default_route(family="inet")[interface] }}
21:03 viq But I don't know how to get that working.
21:04 Edgan I do it with https://pastebin.com/0C0UEeZH
21:05 viq My current attempt is https://pbot.rmdir.de/Hi9cdh0afmo-kgxWfkjZFg
21:06 Edgan viq: What does  network.default_route(family="inet")[interface] return?
21:07 viq Nothing, it's not valid, at least it didn't let me select just 'interface' from the output. It's shorthand for what I want.
21:07 Edgan viq: Why do you need the ip anyway? Don't you really want the subnet? Do you not have other grains that you can match against?
21:07 viq I have some multihomed hosts
21:08 viq On various networks. Some hosts that have only private IPs, some that have only public.
21:08 Edgan The public seems like a problem, unless they are in a controlled range
21:08 Edgan then you are back to subnets
21:09 viq And I want the IP so I can stuff it automatically into my DNS, so I don't have to add them manually when I create another VM
21:09 defsdoor joined #salt
21:10 viq And if I give a list of subnets, there will be hosts that will have addresses from multiple of them
21:10 Edgan viq: You could go based on the default subnet
21:10 viq That's why I thought it would be easiest to find interface via which default route is going
21:10 viq "default subnet"?
21:10 Edgan but ok, you means A records not resolv.conf
21:11 viq Yeah, I want to automate the DNS server
21:11 viq And creation of records on it.
21:11 viq Or, if it's easier to think about, /etc/hosts ;)
21:12 Edgan viq: I am pretty sure you can convert this to jinja if you don't like grains for some reason, grain['primary_ipv4'] = subprocess.check_output("ip route get 1 | awk '{ print $7 }' | head -1", shell=True).rstrip()
21:13 viq Edgan: this will work only on linux, I have a bunch of BSDs
21:14 Edgan viq: With the grains/jinja you could give it the right command per os
21:17 viq At that point my way seems considerably cleaner, if possible
21:19 viq Since the modules that do this are already there, if only I was able to use output of one as parameter for another
21:20 viq But right now I'm not even sure how to get just the "interface" value from network.default_route into a file
21:22 dmcnabb joined #salt
21:22 arlyon joined #salt
21:24 dmcnabb left #salt
21:27 viq YES! https://pbot.rmdir.de/gsIlP4yQ1HXSwRXVepnAHg
21:27 kartik_subbarao joined #salt
21:28 DanyC joined #salt
21:29 Edgan viq: now how do I do that in a grain, so I don't have to paste that jinja everywhere?
21:30 dmcnabb_ joined #salt
21:31 viq Actually I plan to push that into mine. And if you're talking about writing custom grain, I'm waaaay too new at python to play with that right now
21:31 dmcnabb_ left #salt
21:34 kartik_subbarao In salt 2016.3, salt automatically prompted me for username/password when connecting to a gitfs_remotes repo. But in salt 2017.7 on Ubuntu 18.04, it throws an error "_pygit2.GitError: authentication required but no callback set". Is there an easy workaround?
21:34 kartik_subbarao And actually -- more importantly, it used the git credentials cache so as to not prompt me for username/password.
21:34 Edgan kartik_subbarao: Why would you want a user/pass and not ssh?
21:34 kartik_subbarao it's an http based URL for the github repo
21:34 kartik_subbarao https sorry
21:35 Edgan kartik_subbarao: github is both
21:35 kartik_subbarao I want to use the git credentials cache like I did with 2016.3
21:35 Edgan kartik_subbarao: Most people use ssh for automation
21:35 kartik_subbarao I don't have admin access to the github server, they have not set up ssh credentials
21:36 Edgan kartik_subbarao: Are you using third party public code?
21:36 kartik_subbarao I'm using the Ubuntu salt package that comes with 18.04
21:36 kartik_subbarao it uses python 3
21:36 heaje joined #salt
21:37 Edgan kartik_subbarao: I mean the code being pulled from github
21:37 kartik_subbarao it's an internal corporate github enterprise instance
21:37 Edgan kartik_subbarao: https://docs.saltstack.com/en/latest/topics/tutorials/gitfs.html  mentions gitfs_user and gitfs_password. You would be able to configure them directly in /etc/salt/master
21:37 kartik_subbarao I don't want to hardcode a username/password
21:37 kartik_subbarao I'm using a masterless minion here
21:38 kartik_subbarao I want any user to be able to run salt by authenticating with their own username/password
21:38 Edgan kartik_subbarao: you don't need admin access to add an ssh key to your own account
21:38 kartik_subbarao again -- all of this worked fine with 2016.3
21:38 Edgan kartik_subbarao: You are so doing it the hard way
21:39 kartik_subbarao The github server is not a user-accessible linux box -- it's an GitHub enterprise appliance
21:39 Edgan kartik_subbarao: Are the users running salt as themselves or as root?
21:39 kartik_subbarao the admins have not enabled users to set up SSH keys
21:39 kartik_subbarao I use sudo to run salt, but it picks up my personal credentials
21:39 Edgan kartik_subbarao: Being that it is the same company, get the admins to enable ssh keys
21:39 stooj joined #salt
21:39 Edgan kartik_subbarao: So the caching was per user?
21:39 kartik_subbarao yes
21:40 kartik_subbarao Do you know if the gitfs.py code changed between 2016.3 and 2017.7
21:40 Edgan kartik_subbarao: Are you sure you were using pygit2 before, and not gitpython?
21:40 kartik_subbarao yeah pretty sure
21:40 Edgan kartik_subbarao: Do you still have the old server?
21:40 kartik_subbarao yes
21:40 Edgan go look
21:41 Edgan I ask because pygit2 and gitpython behave differently
21:41 kartik_subbarao one sec
21:41 kartik_subbarao sorry -- I was using python-git, not pygit2, before
21:41 Edgan That probably explains your problem
21:41 Edgan and python-git is probably gitpython by another name
21:42 kartik_subbarao ok let me check
21:42 Edgan kartik_subbarao: Hopefully the 18.04 package isn't hard set to pygit2
21:45 Edgan kartik_subbarao: I would also recommend using salt packages instead of Ubuntu packages. It will let you keep up to date, and I would say that using python3 for salt probably mostly works, but more bugs.
21:46 kartik_subbarao That did the trick! (uninstalling python3-pygit2 and installing python3-git)
21:46 kartik_subbarao understood
21:46 Edgan kartik_subbarao: Awesome
21:46 kartik_subbarao I don't think 18.04 packages are available yet from repo.saltstack.com, but your point is well noted
21:46 kartik_subbarao Thanks again for the help!
22:02 dmcnabb joined #salt
22:11 saltsa joined #salt
22:33 itamarjp joined #salt
23:52 justanotheruser joined #salt

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