I will sound silly but I haven't worked on our Puppet code for a while and now we get this error:

Error: Evaluation Error: Error while evaluating a Resource Statement, Could not find declared class augeasproviders::instances (file: /etc/puppetlabs/code/environments/production/site-modules/profile/manifests/sysctl.pp, line: 21, column: 3) on node blabla

Where should I start? We have this in our Puppetfile:

mod 'herculesteam/augeasproviders', '2.4.1'
mod 'herculesteam/augeasproviders_core', '2.6.0'
mod 'herculesteam/augeasproviders_sysctl', '2.5.1'

https://forge.puppet.com/modules/herculesteam/augeasproviders#augeasprovidersinstances That class comes from the

augeasproviders

module.

I’d double check that your code deploys are working and that that module really does exist on your Puppetservers.

Hmmm, I don't see the

augeasproviders

dir in modules

I use r10k

INFO     -> Deploying module to /etc/puppetlabs/code/environments/production/modules/augeasproviders
INFO     -> Deploying module to /etc/puppetlabs/code/environments/production/modules/augeasproviders_core
INFO     -> Deploying module to /etc/puppetlabs/code/environments/production/modules/augeasproviders_sysctl

But:

# ll /etc/puppetlabs/code/environments/production/modules/augeasproviders
ls: cannot access '/etc/puppetlabs/code/environments/production/modules/augeasproviders': No such file or directory

# ll -d /etc/puppetlabs/code/environments/production/modules/augeas*
drwxr-xr-x. 6 root root 222 May 14  2021 /etc/puppetlabs/code/environments/production/modules/augeas
drwxr-xr-x. 6 root root 204 May 14  2021 /etc/puppetlabs/code/environments/production/modules/augeasproviders_core
drwxr-xr-x. 4 root root 155 May 14  2021 /etc/puppetlabs/code/environments/production/modules/augeasproviders_sysctl

This is a server-less setup, is it safe to delete

/etc/puppetlabs/code/environments/production/

and start again?

Yeah, I’d start from scratch by deleting that code dir and re-running r10k

Currently running r10k in debug mode

Try adding

-v debug2

to your r10k command

Running as

/opt/puppetlabs/bolt/bin/r10k deploy environment production --puppetfile --verbose debug

for now.

Oh I got it:

Error during concurrent deploy of a module: Unable to connect to <https://forgeapi.puppet.com> (for request /v3/modules/herculesteam-augeasproviders): execution expired

r10k must have forgotten that it must go through a proxy

Yes, that was it. Our

/etc/puppetlabs/r10k/r10k.yaml

file is managed manually. I don't know why it lost the

proxy

line.

Thanks a lot again, nate, for your invaluable help.

Funny to see that the file was last edited may 14, 2021. But I'm pretty sure it was working earlier this week. Oh well.

This message was deleted.

I probably need to read a manual somewhere but... Is it possible to make a client apply a cached catalogue if it is unable to contact the master? Have I misunderstood the use_cache_on_failure setting?

That’s exactly what

usecacheonfailure

is

at some point in history, I seem to recall that being a default setting, no?

Yeah, it defaults to

true

.

That’s exactly what

usecacheonfailure

is

that’s what I thought. Last catalog continues to apply until connection to server is restored.

@Andy Fry But note that if you’re running puppet with

puppet agent -t

, the

-t

option includes

--no-usecacheonfailure

-t

is short for

--test

. Refer to https://puppet.com/docs/puppet/7/man/agent.html#options to see what all

--test

includes.

@Andy Fry But note that if you’re running puppet with

puppet agent -t

, the

-t

option includes

--no-usecacheonfailure

-t

is short for

--test

. Refer to https://puppet.com/docs/puppet/7/man/agent.html#options to see what all

--test

includes.

@natemccurdy that explains the behaviour I was seeing using 'puppet agent -t' Thanks for that. Again....

@natemccurdy is there a way to cache reports too? Or do we need to have agents send reports direct to puppetdb to achieve that?

The latest report is cached locally on the agent. For more history, you’ll have to setup your Puppetservers to forward agent reports to PuppetDB or some other custom report collection tool.

if you have

vagrant

available, https://github.com/puppet-bootstrap/basic-aio is a quick way to get a test puppet environment running

@natemccurdy yeah the puppet master is forwarding reports to puppetdb. This isn't crucial but would be nice I guess.