Puppet Community

Sorry, I didn't want to offend anyone.  I know it is delicate to ask a question about another tool/system like Ansible in a Puppet help channel.

It wouldn't surprise me if a lot of people use Ansible for orchestration and Puppet for configuration management.

Definitely, I just wanted to see if I was better off going with Bolt because I know Puppet, but which represents challenges for an org of our size, or go with Ansible, which I don't know at all, but is more popular on the market, especially for small orgs like us.

I absolutely use ansible for orchestration

my teams are all python people, so it's generally easier

SaltStack is another CM/orchestration tool. (Yes, it uses Python.)

It's still a bit hard for me to distinguish different topic like configuration management, orchestration, provisionning, IaC, cloud provisionning.

I thought at first that Ansible was CM, but it looks like it can do more than that.

<@U020LK7D0JU> can I count this as a vote for Ansible for my current need?

from me, yea.  So, I don't consider Ansible to be configuration management because it isn't designed to be maintain state.  Ansible Tower/AWT gets closer, but if you apply a state that breaks login you can't fix it... so I'd never use it

idempotency and pull are requirements for me to call it CM

(although, yaml as a programming language should get people thrown in jail)

I agree that Puppet's idempotency is something that I love.  It is now achievable with Ansible with AWX or RH Ansible Automation Platform though.

So what kind of backend do you use for Hiera if you don't use yaml?  ENC?

but yaml for writing ansible playbooks makes me feel dirty

Don't know about AWX.  You see how much I know about Ansible

Oh Ansible's equivalent of Puppet's manifests are "written" in YAML?

it's clunky and has no sane way to munge data

Well, for my need I think I should be OK.

it does all the service checks and serially or in groups brings nodes down for patching and then monitors them coming back to service

I guess I'll give Ansible a shot.  Honestly, one of the hope I had with Bolt was the re-use of the Puppet agent for authentication/access because I didn't want to create a user for Ansible on my Linux VMs, but apparently I should use my own account.

because we've been using ansible for like 7-8 years now

I don't even know what the current version of Ansible is

Thanks, but my question was more Bolt vs Ansible than how to do it with Bolt.

Can bolt use the puppet agent for authentication to the VM or do I have to create a separate user with a public key like I would on Ansible?

you wouldn't create a separate user with ansible, just use your normal user account

<@UEQ6B9JAH> Does the Ansible user need to be sudo NOPASSWD?

My normal user requires password when sudoing

no. see the ansible documentation on `become`

<@UF7CLS0JD> <https://forge.puppet.com/modules/encore/patching|https://forge.puppet.com/modules/encore/patching> already has taking vm snapshots as part of patching workflow.

Thanks <@U02819QR9BQ>, my question was more "Bolt or Ansible" than how to actually do it.

(considering that we're using Puppet but not Bolt nor Ansible)

There’s no magic solution. OS patching is challenging enough 

Most teams are using Ansible as there’s already ready made solution. For Bolt you are all on your own. Usually sysadmin community hates doing any kind of coding

What do you mean by "ready made solution"? Something like <https://forge.puppet.com/modules/encore/patching>, but in Ansible?  You mean that there are more publicly available "recipies" for Ansible than Bolt?

Bolt is the same in regards above of user account and as sudo.

I'd say for bolt v ansible honestly it would depend on much deeper questions of integrations, how much you are using puppet and would you see you expanding it to other cases..

For this use case it would be a matter of taste what you prefered to learn

We only use Puppet for CM for Linux VMs.  We don't see any expansion plans.  We are on-prem right now with a vague plan to go to the cloud.  I see Terraform+Puppet in the cloud if we do a fork+lift with EC2 instances or VMWare on AWS.

I don't really have a preference. I would like to choose the one that has the most publicly available resources and is easiest to implement (hence the question about if it can leverage the puppet agent).

In terms of published resources ansible will have more than bolt implementations. Bolt as you would expect has more direct Puppet based integrations such as being able to look up data from hiera or puppetdb and being able to leverage puppet communication protocol (PCP) with puppet agents. It should be clarified that PCP communication is a puppet enterprise only feature so you wouldn't be able to communicate directly with a puppet agent in open source.

Oh, we use Puppet open-source. And only hiera, no PuppetDB.

So Bolt cannot use the open-source agent to access the target nodes?

It would need to be one of the other transports such as ssh or winrm

I see. So the only benefit is to use hiera

How many folks are using r10k? I'm trying to determine whether to deep dive into r10k or teamcity first to evaluate which solution would be better for my homelab environment.

Mainly i'm looking to automate code that's been prompted to the master branch of a git repo getting transferred into the prod environment within puppet.

Although I'm guessing TeamCity would be closer to Bamboo which is being considered at work?

1% use g10k and the rest have obscure requirements or legacy environments

How would I need to structure things? right now my repositories are divided like this.

<https://puppet.com/docs/puppet/7/modules_fundamentals.html#module_structure>

you have one main repo, called control repo. it contains a Puppetfile which is basically a list of all other repos/modules and their version/ref

yep, I use Puppetfile that pulls other modules/repositories, repo itself is pretty empty

<https://puppet.com/docs/pe/2021.7/r10k_deploy_env.html>

It's really not that complicated, <https://github.com/puppetlabs/r10k/blob/main/doc/puppetfile.mkd>

Forge module
```mod 'puppetlabs/puppet_agent', '4.12.1'```
Advanced config for hiera
```mod 'data',
  :git    =&gt; '<mailto:git@example.com|git@example.com>:puppet-hiera.git',
  :branch =&gt; :control_branch,
  :default_branch =&gt; 'development',
  :install_path   =&gt; '.'```

See the `puppetlabs-puppet_agent` module at <https://forge.puppet.com/puppetlabs/puppet_agent?src=slack&amp;channel=puppet>

Can you share the URL of the vid you're looking at?

I found <https://www.youtube.com/watch?v=DO77GgC9u48|this video> helpful.

each branch in that name will be used to create a separate puppet endvironment

and, r10k can use multiple control repos as well

Say you need different code deployed for windows and unix

we created a yaml file spec here and the Puppetfile is just ruby code that loads the yaml file and parses it