sfba.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
A Mastodon instance for the San Francisco Bay Area. Come on in and join us!

Server stats:

2.3K
active users

#ansible

31 posts29 participants3 posts today

On one of the terminals I use (Blink on iOS) I can't discern yellow on white, which is why I configure this in #Ansible

export ANSIBLE_COLOR_CHANGED="dark gray"

It looks a bit unusual because it's not yellow, but it helps me.

Hi all. I'm hating to be writing this post, but here we are.

A little while ago, I was laid off from my SRE job at a government contractor.

If anyone is looking for an SRE with:

- #Linux experience (a lot)
- #BSD experience (some)
- Plan 9 and Haiku experience (a little)
- Experience with a large #Ansible codebase
- Experience managing thousands of servers
- Experience with bare metal and VM management

and who isn't afraid of code - I'm your person!

I try to live my life by moving slow and fixing things - I'd be especially interested in roles around labor organization, #climate action or with a socially conscious #coop if anyone from those communities is looking.

Boosts welcome!

Hunting down easter eggs & putting them all in a single basket? Hopefully you're not doing the same with your VMs - the new ProxLB v1.1.1 is out, now!

#ProxLB is a loadbalancer for #Proxmox clusters that balances guests across your nodes based on cpu, memory or (local) disk size. It also comes with additional features like:
* Affinity / anti-affinity rules
* Maintenance mode
* Node evacuation (based on best resource usage)
* Best node evaluation for CI/CD (e.g. #Ansible or #Terraform)

Continued thread

OK, I managed to improve lots of things in those setups and make the setup more reliable (even in case it takes really really long for everything to be up).

codeberg.org/johanneskastl/git

Now with four branches, one for Gitlab installed via helm chart and one using the Gitlab Operator.
And each of them with and without a Gitlab Runner being installed into the cluster.

Summary card of repository johanneskastl/gitlab_on_k3s_vagrant_libvirt_ansible
Codeberg.orggitlab_on_k3s_vagrant_libvirt_ansibleVagrant-libvirt setup that creates a VM with k3s and installs GitLab in the cluster

Sometimes it feels like Ansible is a lot more friendly with home lab and self hosting than Terraform. A lot of things I want to do with Terraform do not have official providers, while Ansible does have a lot of official modules.

I'm not looking for a solution, just making an observation.

EDIT: To be fair, I have been asked for examples, so here is at least one (I could list a few more, but no need).

bitwarden.com/help/ansible-int
registry.terraform.io/search/p

BitwardenAnsible | BitwardenThis article will walk you through setting up and using the Bitwarden Secrets Manager Ansible integration.

The #s390x open source software team at IBM confirms the latest versions of various software packages run well on #Linux on #IBMZ & #LinuxONE

In March 2025 validation was maintained for over 30 projects, including #Ansible, #Apache Tomcat, & #Erlang 🎉

Plus, community CI was added for segyio, and Rust's linux_syscall project now publishes s390x releases on crates.io

Full report + how your project can apply for a s390x VM: community.ibm.com/community/us

community.ibm.com · Linux on IBM Z and LinuxONE Open Source Software Report: March 2025
Continued thread

I'm currently also working on the #powerdns (authoritative) podman plugin impelementation and blocky (#DNS caching) because every good home network needs those things.

I also wrote a pretty opinionated CA #Ansible role that automatically creates and deploys server certificates to all my hosts.

I mean when I automate psql I will have to support TLS out of the box as well, right?

Isn't anybody into those things? I love to have another maintainer for my MMN collection.

What do you guys do about breaking home directories in #Linux.

I mean every desktop env after some time gets flaky and then you feel like you wanted a reset but you don't want a reset because it's a lot of work to reconfigure everything and you will probably forget about details.
I mean there's #nixos but despite me liking the idea I don't like the implementation

Are there highly mature #Ansible roles for like #gnome #kde or whatever?

How do you tackle this?

Continued thread

OK, I found a way that seems to work in all phases:
- the migrations pod is not yet started (PodInitializing)
- the migrations pod is running
- the migrations pod has been Completed

This part waits for the pod to exist (no matter its state):
codeberg.org/johanneskastl/git

This part waits for the pod to be in Completed state:
codeberg.org/johanneskastl/git

I did not use the module's wait functionality as I could not get it to work the way I want. So I used what I often use: Ansble's `until` together with a `json_query` filter.

Now that the Debian netboot/preseed is sorted, my next task was to write an Ansible playbook to rebuild the master nodes, doing the careful step of removing the node, removing from etcd, reinstalling, then re-introducing the node.

About 15 minutes in I realised that this task is actually taking longer than if I just ran this by hand. Sometimes you just have to know when to stop. How often is this process going to run? This is why I used to have XKCD 1205 as a printout on my desk.

https://xkcd.com/1205/

xkcdIs It Worth the Time?
Continued thread

Fifth and (it really seems) last round for the #NixOS #Ansible bubble:

Thanks to @nebucatnetzer I tried installing ansible "the other way round". Rather than trying to install Ansible and have a customized python3 (with hvac and kubernetes modules etc.) as a "build input" I have tried the other way round: Adding ansible and ansible-core to the python package:

```
(pkgs.python3.withPackages (python-pkgs: [
python-pkgs.ansible
python-pkgs.ansible-core
python-pkgs.hvac
python-pkgs.kubernetes
]))
```

A short test was successful, tasks delegated to localhost found the kubernetes module and could successfully do things! Hooray!

Today, #IPv6 to the rescue.

I broke DHCP for the legacy IP space on my pfSense router today but thanks to running dual stack I could reconnect using the IPv6 address and fix things. I was trying to move to using #Ansible for managing pfSense and somehow despite getting zero errors and Ansible reporting things as idempotent, things were unhappy.

I managed to get back into the router and deleted the new records, recreating them manually and everything is working again. No clue what I broke yet, but yay for IPv6.