Puppet Community

oh dear.  I'm still calling this wrong and a mistake.

and I wonder why their not using pdk anymore?

It was only released yesterday.  Give it chance for some bug reports to come in?

Correct, for this round we are only applying safe/unsafe auto corrections from rubocop.. so you will see todos appear in most repos.

The safe/unsafe aren’t related to ruby compatibility AFAIK. There can be ‘safe’ fixes that are very much not safe if your version of ruby has been dropped.

But that's what it is.  I can't see jruby in Puppetserver 7 getting a bump this late in the lifecycle.

It's not one of those cases where people on really old versions might complain.  This is relevant for users running the very latest puppetserver release.

It's a bit annoying that setting target version to 2.6 limits you to rubocop 1.49 but the latest version is 1.50.x so you won't be missing much (yet).

and I don't think it's worth the pain trying to figure out which cops might be problematic and individually disabling them.  That's literally the job of the `TargetRubyVersion` setting.

I’ve also started a conversation internally 

I don’t see any issue with moving the target version to 2.6 at the moment 

Sure.  It would limit the rubocop version to 1.49 (but 1.50 was only released last week IIRC)

Yeah I’m not too worried about that atm 

Anyway... on `puppet_certificate` I think I've used it on my puppetserver certs, just not the CA server.

The new min ruby is set to be 2.7 across the board for our stuff 

if you got a minute for me, and for Nomad, I made this PR. I also added a ruby script but I'm among the worse ruby coders in the planet:
<https://github.com/voxpupuli/puppet-nomad/pull/85>

and does anyone know what this error could be?
It's happening on 2 tests out of 14, and it's CentOS 8.

I tried already this approach, and had it didn't work:
• I tried the function but, the validation needs to check that a few files are available on the system. hence it has to run on the agent.
• tried deferred function and it didn't work (had a discussion in a long thread with <@U112RM52S> )
• we thought about facts, and the problems it that it will always fail once and runs properly the second time (because the fact is evaluated earlier, when the directory is still missing). 
The `validate_cmd` seems the only viable solution.

you got it in a minute... I wasted time to try it out :exploding_head:

Now you got the experience so you’ll catch this in a minute too next time :-D

`#!/opt/puppetlabs/puppet/bin/ruby` is ok for AIO packages but might cause some issues on custom installations.. not sure how to deal with it better

just commented about that on the PR :smile:

commented a bit more on it (not every OS might have ruby installed and `/opt/puppetlabs/puppet/bin` is not in $PATH usually (IIRC just `/opt/puppetlabs/bin` should be in $PATH)

I think, per suggestion from <@U1PC2RAH0>, I can make the `validate_cmd` parameter customizable. It can even default to `nomad config validate %` , with a mention to the script in the documentation (or the way around.
It can even default to `undef` , or have pre-filled options.

`nomad config validate %` is a sane default IMO