#theforeman

hello everyone

Hey All - just sent an email to the google group - wanted to poke in here as well in case anyone was willing/able to assist. Doing a puppet 3 to 4 upgrade and 70% or so of runs having errors: Execution of '/etc/puppetlabs/puppet/node.rb FQDN' returned 2: You need the `json` gem to use the Foreman ENC script"

Only seems to be under load - if i remove that puppet master and hit it manually via puppet agent -t - its all happy through multiple executions

jlang1: read that 😉so the first thing that pops to mind is that puppetserver 4+ runs it's own ruby stack

thus checking if the json gem is available in the system ruby isn't really telling you wanything

from what im reading that is correct - it should be isolated in its own software collection entirely?

tfm-rubygem?

its not an SCL

puppetlabs run their own ruby stack in /opt

(this is different to Foreman, which does use an SCL)

that said, the json gem is usually included so there's definitely some weirdness going on there

so when "im" running node.rb FQDN im using "my" ruby/json/whatever and when the puppetserver is doing it - its specifically (or should be specifically) using the one it lays down as part of the PC1 collection?

right

so you have no issues using node.rb

it appears no matter how many times i run it, it always succeeds with no errors

well thats reassuring from the foreman side, the real question is why your puppetserver can't load the json gem

on my install (debian) the puppetserver package soes seem to come with an IRB interpreter

without understanding everything (i don't) id say it is loaded though, all i need to do is isolate that puppet master and do manual runs against it and its happy every time as well

well your error mesasge is clearly a failure to execute "require 'json'" in the puppetserver's environment

oh im with you - its the intermittent thats confusing me. in my mind - its either loaded or not loaded (100% or 0%)

yeah, its definitely weird

My candlepin is broken in Katello with a qpidd error. "notice SSL plugin not enabled". Any hints?

you say you have several masters - is this a load balanced setup or just a separated network (so the same clients always talk to the same master)?

tamarin: can you pastebin the full error?

this is a load balanced setup. nodes are behind an F5 load balancer

by nodes i mean puppetservers

hmm, that makes it tricky to try to isolate if its a specific agent or master

i've been editing my puppet.conf on agents and just pointing to a specific master

also aggregating the logs and seeing the json errors across all nodes

jsherrill: I can if you want, but it seems that Candlepin task fails because (candlepin.log) can't reach qpidd (connection refused) and qpidd is saying "[Security] notice SSL plugin not enabled, you must set --ssl-cert-db to enable it."

jlang1: sadly i have to run, so hopefully you'll get an answer on the list

no worries man

still poking around :)

tamarin: ahhh so qpidd isn't starting?

jsherrill: It starts, it just doesn't listen on ssl. I dug up an old /etc/qpid/qpidd.conf from backup and it looks like I had ssl-* settings but now I don't.

hm

strange

so - my puppetserver is still logging a ton of node.rb returned error: You need the `json` gem to use the Foreman ENC script"

However - the GEM is installed per puppetserve

Does anyone know why im getting inconsistent results:

first try returns "true" and all subsequent tries return false?

What is the "@profiles" I see in the default "PXELinux Global default" template?

@jlang1: That is normal ruby behavior. Once imported, re-import returns false.

oh okay

so if im returning that module just fine via IRB - any idea what i might look at (or towards) in regards to why puppetserver is telling me it cannot load it?

No idea on that, sorry. :(

I'm lost on this qpidd thing. I have an nssdb directory that works fine with certutil -L, the password, etc. But the certs I see in -L don't include a "broker" although I have two private keys for "broker".

Hiya folks, running Katello 3.4. Is there any way to remove a repo from a product that has once been involved in a content view, but is no longer?

I get the message, "<blah> cannot be deleted since it has already been included in a published Content View."

jpavel: you've got to remove the old view versions that contain the repository

Yeah, I was pretty sure we had.... but you're telling me that if no content views contain the repo, I can delete it?

I can work with that.

Is there any way to query the content views to figure out where that repo is referenced?

I'm getting a "8428ad6b [app] [I] Failed to save:" (error msg is blank) when creating a new LDAP auth source [Active Directory] on a fresh install of 1.16.0-RC2. The other two work [IPA/POSIX], but those are the two I don't need :) Any thoughts on troubleshooting?

and ^ "79d27e82 [app] [D] Unpermitted parameters: utf8, _method, authenticity_token, _ie_support, id, locale"