👋 n00b here, trying to adjust a module for teamviewer or more specifically when the module applies to the endpoint, to stop the service, alter a file, start the service but what I have is not working.

You can't to that this way

service has to have a desired state, your catalog would fail to compile due to duplicate declaration

Normally you don't need to stop the service to edit configuration file. Is it really a requirement ?

also, I think you will add that line every single puppet run. you need a match attribute

as per Teamviewer support yes but I just tested and editing + restarting the service seems to work

so just notify service

the normal pattern would be for the

file

and

file_line

resources to have

notify => Service['teamviewerd']

. if these are in place, the service doesn't need the

after

attribute.

If there is a compelling reason to stop service for something. you would have to use puppetlabs/transition module, as a workaround

for instance, you might need it to modify a service user, since os will refuse if service is running

After upgrade to 2023.8.2 our compile masters are running out of inodes and get stuck... most usage is on

/opt/puppetlabs/server/data/puppetserver/filesync/client/versioned-dirs/puppet-code

and I see some unused environments there and double versions of some envs... I'm not able to bring pe-puppetserver up since it gets out of inodes. Beside adding an extra disk is there a way to approach this?

I have a weird problem: on my new PE-2023 infrastructure I can provision new compilers and everything is working, but the next puppet run access to postgresql is denied. One of my compilers is still working, but others are removed from /opt/puppetlabs/server/data/postgresql/14/data/pg_ident.conf When i manually add the missing compilers to this file the puppetdb sync is working, but the next puppet run they are removed again Most probably a very simple setting somewhere, but i cant find it.

is there some task out there that will run puppet agents on the nodes in a node_group ?

I think the whole Bolt/Node interactions could be much expanded. i.e., if its in the console, that should just automatically be available to

-t

by name

there is an API, but I was hoping a task already exists

Oh I know, it just seems to me the value proposition here that Bolt had that things like Ansible or Chef would not have would be to just be able to target a node group already in existence in the console by name. Bolt would do the "glue" on the back end, and it would "just work" bolt command run "foo" -t <node_group_already_in_console>

There’s a task in the puppet_agent module https://github.com/puppetlabs/puppetlabs-puppet_agent/blob/main/tasks/run.rb

And plan https://github.com/puppetlabs/puppetlabs-puppet_agent/blob/main/plans/run.pp

Right, and peadm too, but I mean a classification group in console as a target

not individual node

This one also individual node

puppet task show enterprise_tasks::run_puppet

something that would do:

puppet task show enterprise_tasks::run_puppet node_group='PE Infrastructure Agent'

@vchepkov when you use puppet job run, you cns specifiy a node group (uuid only I think)

You mean in console? Yes, I can. I want from a plan

I meant the cli

Also puppet task run supports node groups

There is --node-group

is there a built-in resource type for Puppet's "Scheduled Tasks"... everytime I search for this sadly I keep ending up down the rabbit hole of "Windows" Scheduled Tasks

basically want to have a module with a task and when I include it on our PE primary it creates a scheduled task... could write shell script to make API requests to create that but figured there should be some sort of way to do this in the Puppet DSL

We usually have systemd timers that start tasks or plans, or gitlab jobs