#packages - February 2020

@Zachary Loeber any reason we don't use the download_zip macro?

@Zachary Loeber has joined the channel

I noticed the latest PRs don't leverage it

Honestly... copypaste-itus

I didn't realize it was there to use 🙂

I'm getting a better sense of the backend macros now that I did the wrapper utility though

In fact, I just added a generic template generation based on the selections a person makes when using make helper to pull in github apps. It leverages all of the macros (based on the user selection it uses the proper macro)

you're going to town on it! 😃

selfish reasons but trying to not just be a consumer ya know?

you use any of the silly whiptail wizards I threw together?

I haven't added any new packages lately

mostly just fixing the auto-PRs for updating packages

maintainers keep mucking with their package names! 😃

(and version numbers)

That's something I keep thinking about. There should be a better programatic way to capture the right template to use for downloads that capture these things

fuzzy-logic style 🙂

Oh, there's a away that will work 100%

just use the API

ha, I do scrape that in my helper script

I had to do that for one of the packages which uses a unique git sha in each one

but it's still not 100% for architecture, os, and all that other stuff

in the release name?

thats nuts

If you want to reject the prs for hashicorp apps I'll get them updated to use the zip macro (if you didn't already)

I was on a sales/ideation pitch the first three days of this week and I feels like I gave a three day long technical grilling where I was squeezed for all I was worth

You ever do any Kafka work btw? I guess its hot stuff right now

lol

worked on some kafka stuff years ago, but not recenlty

I've been running it in Kubernetes for a bit for a client that asked for it (ask and thou shall receive).

kafka operator?

what a pain to get right. I'm seriously looking at NSQ as a recommendation instead.

the official operator wasn't released at the time so no, just charts (confluent ones though for the connect component specifically)

Since I have your ear for a sec.... The EKS module of yours, is that pretty much superceded by the aws provider at this point?

?

i don't follow....

I may bounce over to Amazon for an engagement. They want managed kube, I'm looking at the terraform for it and there is your module

oh, so here's how to understand that

(even with managed node pools, you still want to use terraform to manage it)

there the EKS cluster, which is really just the masters

we have a terraform module for that

then you'll want node pools. there are all kinds of different ways to do nodepools

self-managed with autoscale groups

managed node pools (supported by provider), called by terraform

and fargate profiles (like running fargate)

there are many limitations and tradeoffs with each flavor

best to ask on #office-hours hours and I can go into details

I'll pester you more on the next one if other things don't take priority for your time

After using the provider for azure to create aks clusters over the last year

I've seen the provider released half a dozen times over to fix and generally overcome a good deal of my prior workarounds and such

so now the default provider is pretty fully fleshed out (finally)

the thing is nothing is ever easy on AWS

it is very low-level/raw

I'm going to have to say that is the same for Azure. The moment you need to do anything beyond the basics you almost instantly hit a wall

or maybe I'm just never happy with the basics and its a matter of perspective....

your modules are impressive btw, no i'm not hitting on you, I have a wife.

So short answer, starting net new, would you still use your module over default provider code?

100%

it's always our pattern though

modules are for business logic

almost never is a raw resource sufficient.

That's sensible to me good sir, thanks a bunch

https://livebook.manning.com/book/gitops-and-kubernetes <- worth keeping an eye on

@Zachary Loeber i am curious - how are you using gonsul? haven't heard/seen of this before.

I'm not yet. But I was planning on putting it to use in some pipeline tasks to bootstrap/update key-pair values for projects with consul. Not a good idea?

I looked into consul configuration and the idea of manually loading it with values defeats the purpose of config as code to me so I found this thing and figured I'd give it a whirl.

I recorded an intro video of how our packages repo works

It's far from complete, but a start

https://share.vidyard.com/watch/SkGi3ejGmEL3Bm4mZjETXR?

I better watch this one for certain 🙂

My PRs are piling up. Have I been missing something on these that make them unworthy?

@Maxim Mironenko (Cloud Posse) ☝️ :-)

@Maxim Mironenko (Cloud Posse) has joined the channel

I had to delegate these, but Maxim has a question for me as he is new to the repo

I think we are going to review with him tonight

Let’s do just one, then I will take care of others

@Maxim Mironenko (Cloud Posse) are you free now?

yes, let’s do a call

@Zachary Loeber expect a flood of merge notifications 🙂

@Zachary Loeber please, take a loot at <https://github.com/cloudposse/packages/pull/329>

there are some changes required to make this package happens

@UUB28NLDS help keep tabs! 😉

@Zachary Loeber all PRs are merged and here is the release: <https://github.com/cloudposse/packages/releases/tag/0.129.0>

That's cool of you Maxim, thanks much!