Puppet Community

as soon as we get el9 and ubuntu 22.04 support :slightly_smiling_face:

RHEL9 was supported back in Puppet 2019.8 if I recall... Same with Ubuntu 22.04

for the agent. ubuntu 22.04 puppetserver packages aren't available yet

Meanwhile we’ve all been using Ubuntu 22.04 for months and months

you need to provide an fqdn to the `--node` option

and is that an puppet agent or is there also a puppetserver on that box?

there isn't. and i get the same error when i add the fqdn

`puppet lookup` only works on puppetservers

so run it on a box with puppetserver installed :slightly_smiling_face:

i thought I could use it to lookup on the puppetmaster from an agent

Another option to view your Hiera Data is to use HDM <https://github.com/betadots/hdm/blob/main/README.md>

How can I clear out a cert in `puppetserver ca list --all` ?  I have the local puppet servers agent saying the cert is revoked and I want to redo just the agent for the puppet master server itself.

puppetserver ca clean --certname &lt;certname&gt;

See <https://github.com/voxpupuli/puppet-nodejs#npm-packages>

It provides the nodejs installation and a package provider extention for npm.

Regarding `npx` vs `npm` i can not help.

I think `npx` is kind of a convenience thing for running a node module's commands in the context of the local `package.json` , probably if you just use an `exec` resource and set the current directory to the directory of the `package.json`, you'll get the behavior you need. You may need to put the standard idempotency gates on that `exec`, though (i.e. figure out what file to look for with `creates` or what to check with `onlyif` or `unless`)

Thanks <@U02B17Z2MJ7>  That was the path I'm currently going down.

What am I doing wrong.  This is failing and not actually putting the contents of the variables in the command.

are the variables `$::db_name`, `$::db_user`, etc top-scope variables? or where are they defined?

They are defined in another profile file that is being included in the role file.  Sorry if that is not clear...but I'm still new to Puppet. <@U012ASYJN0L>

if they are defined in another profile, you probably need to namespace them to reflect that ie: `${profile::other::dbname}` etc

for reference: <https://www.puppet.com/docs/puppet/7/lang_variables.html#lang_variables-variables-scope>

this is my role file:
```class role::strapi {
  include profile::nodejs::install
  include profile::postgresql::install
  include profile::postgresql::config_strapi_db
  include profile::nodejs::strapi_install
}```
..and  those variables are in the `profile::postgresql::config_strapi_db`

ok, so `${profile::postgresql::config_strapi_db::db_name}` ... etc

That still didn't work.  I get this error:

```Error: '/usr/bin/npx --yes create-strapi-app@latest my-proj                 --dbclient postgres --dbname                  --dbhost 127.0.0.1 --dbport 5432                 --dbusername  --dbpassword ' returned 1 instead of one of [0]
Error: /Stage[main]/Profile::nodejs::Strapi_install/Exec[npx create-strapi-app@latest my-proj]/returns: change from 'notrun' to ['0'] failed: '/usr/bin/npx --yes create-strapi-app@latest my-proj                 --dbclient postgres --dbname                  --dbhost 127.0.0.1 --dbport 5432                 --dbusername  --dbpassword ' returned 1 instead of one of [0] (corrective)```

are those variables set or derived in the config profile? if they are derived it's possible that ordering means the config profile hasn't run yet so it hasn't set those values yet

the `db_name`, `db_user` are just a variable at top of file with a string value

The `db_pass` is pulled in from and ENC eyaml file

if the config profile always must run first (before `profile::nodejs::strapi_install`, then you could possibly add a require to your strapi_install profile

so they're set in the config profile, they're not pulled from other params that are already set

you could try adding `require profile::postgresql::config_strapi_db`  to `profile::nodejs::strapi_install`  (ie: after `class rofile::nodejs::strapi_install {` , before your exec)

OK, does that re-run config_strapi_db profile again?

but that could interfere with other dependencies etc depending on what other interactions you have with the other profiles

no, that specifies ordering ie: config::strapi_db must run before strapi_install

so the order they are listed in the role file doesn't do that?

see <https://www.puppet.com/docs/puppet/7/lang_relationships.html#lang_relationships>

so keep the long variable names `profile::.....::db_user`

yes you need to keep the namespacing otherwise it won't know where those variables are from

because they've been declared/set in another class, so you need to tell it specifically -- see the prev variables link and also <https://www.puppet.com/docs/puppet/7/lang_scope.html#lang_scope> for further info

OK, I'm closer.  I did see that it was using the variables that were just defined as strings in the other  profile working ( `db_user` and `db_name`) however I'm getting this error now:

```Error: sh: 1: S: not found

Error: /Stage[main]/Profile::nodejs::Strapi_install/Exec[npx create-strapi-app@latest my-proj]/returns: change from 'notrun' to ['0'] failed: sh: 1: S: not found
 (corrective)```

so the script/command that the exec is running is returning not found?

or possibly when it's switching to the strapi user?

I'm confused as this error doen't make any sense to me..  Maybe I force the `provider =&gt; shell` ?

you shouldn't need to specify provider =&gt; shell unless you're trying to use specific shell features (which are generally only if you need legacy behaviour)

what happens if you su - strapi on the host manually?

Well, this was running before...although the variables were wrong ( before I asked for help in here) , then I change the variables but they were referenced wrong.  Now I have the require and correct path to variables...but getting this error that  makes no sense to me :confused:

Error: sh: 1: S: not found....

that error is saying the command that `exec` is running is returning "S: not found" -- ie: something called by npx (or the user profile that you're switching to) is passing `S` to the shell and causing it to return not found

if it was running before (but failing due to vars not being set), it's possible that npx is running now and getting further - ie: it wasn't getting the not found error before as it was failing before that point

I'm going to manually run that npx command as the strapi user to see what happens.  I thought I did this before and it was fine...but I'll test again.

yeah i'd check that first. the other possibility is if the dbname/user/password has spaces or special chars they may need to be escaped for shell

Could special characters in the password be messing it up?  I do see a `S` in the password...but there are some special characters too.  So, I'm wondering if that is why it's messing up and returning the `S` ?

Funny..you just said that.. I was typing that and not reading yours until I finished typing :smile:

do you need to quote the password when you pass it to shell manually?

ok, so try adding quotes to your command for dbpassword -- replace the end of your command with:

`--dbpassword \"${profile::postgresql::config_strapi_db::db_pass}\"",`

hopefully slack hasn't mangled quotes :confused:

ie: around the variable you should have `\"`

I actually tried single quotes before you typed that......

currently running and I think might be working....

single quotes could work as well but i would probably escape them to be safe

Thanks for your help <@U012ASYJN0L> Appreciate it

Problem is with modules, need to catch up

Bit if you’d like to cut some edges then “puppet apply path/to/file.pp”. Not guaranteed it’ll work as expected though.. but you may try

that said, there's a puppet run module in ansible

<https://docs.ansible.com/ansible/latest/collections/community/general/puppet_module.html> the last example

yea, my org has used ansible for orchestration for 10 years or so, so we do configuration management in puppet and orchestration in ansible

I am never going to introduce bolt here because of that

Somehow r10k doesn’t download the latest version of the module available on the forge. I don’t specify any version range, so I would expect it to use the latest version. Anybody else noticed that? I’m using r10k 3.15.4 and Puppet 8.1.0

How can I locally test a custom fact I'm writing? Is there a way to test it on a given node locally and have it access its own facts?

Thank your for your response. We are using it in a CI setup, so we’d like to use the latest versions here, but thanks for the warning.

The documentation states

If no version is specified the latest version available at the time will be installed, and will be kept at that version.
```mod 'puppetlabs/apache'```
No other version of the module is yet installed. It is a fresh install. So I would think it would select the latest version.

After some time it actually does. So it seems that r10k or the forge needs some time after getting a new version of a module to deliver it.

I don’t remember having this before. But anyway…. It works now.

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

correct, if fresh install you will get the latest, but it won't be updated after

but I noticed the delay between publishing module and api recognizing it as latest as well

<@U2S1MUUQ7>
Strange thing is that Puppet module install and upgrade however did.

And after some time ( I guess 15 min to 20 min) r10k now does download the latest version. I thought they where using the same  api

<#C11LCKKQ9|> is probably a better source, but today is a national holiday in USA, so folks might not reply right away. I am just bored :)

right. and I was deliberate with USA, since there is United States of Mexico :sunglasses:

r10k does the equivalent to puppet's ensure =&gt; present

if you want to really really bleed, use git instead of forge with branch =&gt; main :slightly_smiling_face:

I don't think that was implied. but it should be to puppet module install though

<@U020LK7D0JU> I wanted the latest version that was published on the forge. I expected r10k to use the same info as puppet module install. But somehow it doesn’t. There seem to be a sync time lapse between the data r10k is using and the data puppet module install is using.

R10k does some caching too, that might be it

Always pin your versions even if it’s to “latest”

<https://github.com/puppetlabs/r10k/blob/main/doc/puppetfile.mkd#puppet-forge>

You need to explicitly specify `:latest` (I wouldn't recommend it)

Thanks – got it working. Way faster than pushing an running puppet every time…

if I'm using a module that requires stdlib between 4.6 and 5, but we have v8 installed will that lead to an unknown resource type error for that module? The metatdata.json is also puppet version requirement of  `"version_requirement": "&gt;= 4.7.0 &lt; 6.0.0"` and we have puppet 7 but I didn't think that would be an issue

`Evaluation Error: Error while evaluating a Resource Statement, Unknown resource type`

stdlib provides very limited list of resources. most likely it's unrelated. What is the resource type?

<https://forge.puppet.com/modules/kpn/advanced_audit_policy/readme>

puppet should indicate, what line it fails

we had internally forked it to update something and are trying to go back to the forge. I'll make sure our deploys are working properly.
Thanks for the assist! I had some people questioning the metadata stuff, so wanted to verify

I don't think anything uses that metadata string besides `puppet module` command

if module was just deployed, maybe puppetserver needs to be reloaded to see new types

using forge module in a puppetfile without version or latest, will only fetch once and then use the version from cache

What is the proper way to pull in a variable into a `exec` `command`?

`/usr/bin/echo ${::db_name}`  &lt;- will this not work? It doesn't seem to work for me.

Can the `$db_name` be pulled in from another profile/module that has been included in the role file?  I have created a new module that I want to pull in the `$db_name` from, but this variable is declared in another module/profile that is being included with the role that is being ran.

Hi Team,
How to ensure puppet server added in uCMDB?

sounds like a question for uCMDB people, not puppet people?