This message was deleted.

Hi Team, I'm looking at the supported OSs for Puppet Server 7 (https://puppet.com/docs/puppet/7/server/install_from_packages.html#supported-operating-systems) and I see that SLES15 is not listed. There seem to be server packages available at yum.puppet.com for those platforms. Is the list out of date or is SLES15 not officially supported yet?

This message was deleted.

This message was deleted.

the other option would be using hiera with deep merge

In regards to templating it depends. 😁 Couple of things to consider • In some cases it's really nice to go through the system and delete some_file.txt.rhel6 when you retire that OS. I've often optimized for this in systems that have a lot of static files. Also sometimes it's easier to bring less cruft along when you force a new file creation for the new OS. • Push it all into data as Lumeire suggests above. It's a very clean implementation, but sometimes too magical for general troubleshooting. I'd certainly use it if this is more of a set and forget sort of case. • concat is really nice and also easy to see in logs, catalog reports, etc. I'd lean towards this option as a starting point. • if os == 8 is also a fine starting point too. No need to overthink it until the complexity actually shows up.

In regards to templating it depends. 😁 Couple of things to consider • In some cases it's really nice to go through the system and delete some_file.txt.rhel6 when you retire that OS. I've often optimized for this in systems that have a lot of static files. Also sometimes it's easier to bring less cruft along when you force a new file creation for the new OS. • Push it all into data as Lumeire suggests above. It's a very clean implementation, but sometimes too magical for general troubleshooting. I'd certainly use it if this is more of a set and forget sort of case. • concat is really nice and also easy to see in logs, catalog reports, etc. I'd lean towards this option as a starting point. • if os == 8 is also a fine starting point too. No need to overthink it until the complexity actually shows up.

I think the template + concat might be the way to go. I need to keep the comments and stuff in the file so the deep merge might be too magical. 😉 If the Apache folks rely on template/concat, then I think that might be my best course. Thanks!

Note that you can’t use a

file

resource +

concat_fragment

to manage the same file though. So I just want to clarify that “template + concat” works as long as you don’t do

file

+

concat_fragment

. It’ll need to be

concat

+

concat_fragment

(for the main part of the file) +

concat_fragment

(for the rhel8 part)

waiting 🧑‍🏫Puppet Core Team is about to start up in #CFD8Z9A4T

Yes, I misspoke. It's the concat plus concat::fragment that I need to use. I've never used concat from scratch before. concat::fragment does accept "content => template("path/to/template")" and that was my confusion.

Thanks for clarifying that for me.

Cool, and yup, sounds like you’ve got the right idea.

concat

+ `concat::fragment`’s

Though I do recommend using EPP templates instead of ERB.

content => epp('path/to/template.epp', $context),

Though I do recommend using EPP templates instead of ERB.

content => epp('path/to/template.epp', $epp_context),

Though I do recommend using EPP templates instead of ERB.

content => epp('path/to/template.epp', $context),

Oh, geeze....more stuff I have to learn after being out of the puppet scene for four years.

😉

I think I'm going to ask for forgiveness for using erb templates as I've got a deadline to meet.

message has been deleted

This message was deleted.

Is there a document that describes what is happening at each of these levels (i.e. what is happening during "Node retrieval" or "Config retrieval")?

https://tickets.puppetlabs.com/browse/PUP-10531 suggests to me, the answer is no

I guess my question: what is happening during "Node retrieval" and "Config retrieval"?

• “Node retrieval”: How long it takes to find the agent’s

Node

object. Literally, how long the

GET /puppet/v3/node/<certname>

HTTP request takes. This usually translates into how long your ENC takes to run + the HTTP overhead of sending it. • “Config retrieval”: How long it takes to get the node’s

Catalog

. Literally, how long the

POST /puppet/v3/catalog/<certname>

HTTP request takes. This is the time it takes to compile a catalog + the HTTP overhead of sending it.

398 seconds for Node Retrieval is incredibly slow.

🙂 hence my question

so this points to the server being bogged down and either spending time in iowait or something else?

Specifically, this points to the ENC being slow. Not necessarily “the server”. What do you use for your ENC?

(also thank you for the answer)