I'm a DevOps engineer and for the past couple of years I've been struggling to wrap my head around Kubernetes et. al. Your content has made all the difference in the world! I usually have to go to a conference to get this sort of knowledge transfer and inspiration. Now the progress I'm making in this space is beyond anything I could have imagined even just six months ago. Please don't be afraid to keep the NERD turned up to 11; your content is so INCREDIBLY helpful :)
Thank you so much for the kind words! It's a tough balance between intro content and going deep. I am glad you appreciate the depth! Keep it up! You got this!
Because of your channel, I compiled my first Blender docker container today. I'm going to try to create a distributed Kubernetes cluster to handle the Blender rendering workload in my home network. My server and several clients can be used for this purpose. I have an engineering background but without such tutorials its much more cumbersome to get it done, just 2 days ago I started with docker and portainer, your tutorials are very helpfull, well presented and easy to understand, thanks man 👍, you're a legend.
I discovered your channel very recently, and I absolutely love your content. I'm still learning a lot about homelabbing, self-hosting, Kubernetes, etc. and your videos have really struck a chord in helping me learn these decently complex concepts much faster and easier. This is my first video of yours that I'm seeing on day 1, and I just wanted to say thanks for the great content, and keep it up!
I‘ve been into ArgoCD a lot due to work, but Flux looks very awesome too. Especially the part where it monitors a Registry for new Artifacts, that seems super helpful. Will probably go the Flux route for my homelab now.
There's ArgoCD Image Updater that does the same thing, but that's a separate application/plugin you need to take care of. So there's tradeoffs with that.
Perfect timing on this video. I just got my cluster up and running using the k8s-at-home template and it uses Flux, Kustomize and Helm. Its actually been pretty nice to work with. Thanks for another awesome video.
Three things: 1. Is it possible to revert a deployment after it autodetects an new image? For example you found a bug and the fix will take some time. 2. Maybe take a look at k9s. Awesome k8s management tool I use. 3. Your content is like cocain to me! ;)
seems very nice. I like that it doesn't allow you to make mistake - no namespace in manifest - "namespace not specified, error: the server could not find the requested resource" and so on
You have no idea how much I've needed this video. Working for huge corporation and basically this is my situation: Ericsson - "You know some combination of programming languages right? Are you a fullstack developer?" Me - "eh.. yes but no? You hired me for my Linux, Network and virtualization skills. I'm a infra guy." Ericsson - "GREAT! Heres a fullstack developer assignment." I love it but as Im building the front-end and the back-end.. Its getting REAAALY messy with each new modification. Im ashamed to say that im basically copying the whole project folder into a backup folder before each major change... This whole developing and deploying things into Kubernetes is such a messy web for me..
Great video 👍🏼 this made my k3s cluster so much better. Had to pauze and go back a couple of times as its a quite fast overview so you easily mis a crucial part.
This all genuinely looks fucking awesome, but there's a point at which this starts to feel as though you are learning new tools to manage the tools which manage your deployments.
I have been using kubernetes agent for gitlab-ci in my homelab for some time, I love it because it's easy and works with both helm and raw manifests. Never really been a fan of Helm, so most of my stuff is in manifests, but I have the option, none the less. The thing I like about flux that I didn't realise until watching this is update management. I use Keel currently because I like to be able to define a per service policy around tags add an approval step (via Slack bot) before an updated image gets rolled out; but this has its downfall in the fact that I often forget to update the tag in the k8s manifest with the new image after keel has done its thing. I then go and make a change to the manifest in my repo, push that change up and inadvertently roll back an image. Thanks, I might have a play later.
Been slowly building out my homelab again from the bottom up (PXE boot all machines to install Flatcar Linux, securely introduce machines to Vault and finally getting to Consul and Nomad to do something different from k8s and ECS at work) but your content is making me want to pivot to k3s, not going to lie. Would love to see a comparison with Argo. Keep up the great work mate.
This was super interesting, Thank You. I am curious what your Infrastructure as code looks like outside of Kubernetes. Say you have a new machine with Proxmox on it. How do you handle the building and deploying of VMs? do you just run Terraform/Ansible locally? or do you have a process like Flux for all things outside of Kubernetes?
Hi Tim, I am a regular infrastructure engineer who specialises primarily in Virtualization and Storage - I am wanting to get into DevOps, I have no experience with coding though I am pushing myself to learn Python, YAML and Go - How do you keep yourself up to date with the latest 'solutions' such as FluxCD, Prometheus/Grafana, as well as others that make you then adopt that service? I would like to know where to go, in order to stay up to date with this sort of information - Great tutorial, I will definitely be tinkering with FluxCD soon on my homelab I'm sure!!
Thank you! Well our Discord is a good place! Lots of infra folks in there! Otherwise I just try to keep my pulse on the cloud since that’s where so much technology is being developed!
@TechnoTim did you miss a step in your docs? I had trouble getting the notifications working until I manually applied both the provider and alerts manifests. It didn't apply through the git commit automatically. Or am I not understanding that part right?
Sad day for me, this channel has gotten too advanced for me. Miss the older days of getting hand held through docker or a network setup. I don't even know what I just watched. Looks like you changed the room you film out of... I guess I don't like change lol
Literally exactly what I was looking for. Tough for a sec engineer to understand CI/CD pipelines. Simple and super useful video. Will be using this for my home lab. Side note. Do you use multiple kubernetes clusters? I.e Rancher/K8s and K3s.
Like mind, think the same. I am working to switch to flux as part of my move away from rancher project. Just got rke2 deployed sort of the way I wanted, so next step is setup flux to deploy things like Prometheus stack, metallb & etc
I've been following your channel and love your content. As a newbie who doesn't do this for a living, I'm making great progress and keeping up (IMHO) thanks to you and your videos. I'm just not able to make the leap from docker-compose to this. Any chance you could run through setting up some apps from k8s-at-home? There are a ton of great home lab examples which use flux and argocd, but I can't make the leap to understand what's going on. I love your HA Kubernetes + Ansible video BTW.
great content as always from what i've read about flux it only works with k8s i'm wondering if there's something like this that would work with things like ansible or other similar tools
Would/Could/Should you run something like this in your production cluster in a release often setup, or would it be for something like test clusters or both?
I think you should cover how to add SSH Public key to nodes for ansible and k3s. Its not explicitly covered and a lot of people will run into that as an issue.
Awesome content! I've been following for a while, and started leaning real hard into devops and trying to pick up as much as I can as fast as I can. While also trying to implement it at work. One question from this video I had was, does this setup sort of replace rancher or supplement it?
It can supplement it! Just worked through installing Rancher with FLUX so it's a really nice way to source control your cluster and deploy it like code!
@@TechnoTim I was just looking for how to upgrade the version of kubernetes. I used your ansible playbook to get mine up and running. Then put rancher on top of it but I can't upgrade it. The whole gitops strategy is looking real nice right now lol!
Kubernetes is a container orchestration framework. Running a couple of containers on one server is easy, when you notice one is down, you can start it back up. But it becomes really difficult to manage when you run 20 services, with lets say 5 replicated containers per service over 10 nodes. To maintain 100 containers by yourself is not fun. So kubernetes task is to ensure desired state. If your service was defined to run 5 replicas per service, kubernetes will ensure that it keeps desired state, if a node goes down, kubernetes will notice it and makes sure that the dead containers are rescheduled. Then theres things like scheduling algorithms, etc.
Did you think about using external secret operator instead of sealed secrets? You can fetch your secrets from a hasicorp vault or a secret management solution of your cloud provider
very tiny suggestion, it is actually a nit since I am very noob in infra topics, in a single project, please favor only .yml or only .yaml but not both at the same time, consistency is really a good thing in a monorepo :)
One thing I'm still trying to figure out is, how to make things work with a cluster that already has a number of applications running. Basically how to retrofit Flux into a working cluster.
Do you have any experience using Flux in a multi-cluster/multi-env infrastructure? I manage 24 K8s clusters across multiple datacenters on numerous continents. I've been looking for an effective GitOps solution that allows me to stagger deployments across environments. Given our breadth of stakeholders, we unfortunately cannot implement a true CI/CD workflow, as multiple internal and external stakeholders may be validating new features in lower environments, and are not yet ready for them to go to prod. The problem with the GitOps model in this scenario is that it requires either managing multiple branches (eww), or multiple repos (ewwer).
That's why the pause button exists, making instructional videos like this great on TH-cam. Tim also links to his show notes in the description, which should get you sorted.
I think I turned the NERD up to 11 with this one!
I'm a DevOps engineer and for the past couple of years I've been struggling to wrap my head around Kubernetes et. al. Your content has made all the difference in the world! I usually have to go to a conference to get this sort of knowledge transfer and inspiration. Now the progress I'm making in this space is beyond anything I could have imagined even just six months ago. Please don't be afraid to keep the NERD turned up to 11; your content is so INCREDIBLY helpful :)
Thank you so much for the kind words! It's a tough balance between intro content and going deep. I am glad you appreciate the depth! Keep it up! You got this!
Because of your channel, I compiled my first Blender docker container today. I'm going to try to create a distributed Kubernetes cluster to handle the Blender rendering workload in my home network. My server and several clients can be used for this purpose. I have an engineering background but without such tutorials its much more cumbersome to get it done, just 2 days ago I started with docker and portainer, your tutorials are very helpfull, well presented and easy to understand, thanks man 👍, you're a legend.
Thank you so much and way to go! That's quite the journey!
I discovered your channel very recently, and I absolutely love your content. I'm still learning a lot about homelabbing, self-hosting, Kubernetes, etc. and your videos have really struck a chord in helping me learn these decently complex concepts much faster and easier. This is my first video of yours that I'm seeing on day 1, and I just wanted to say thanks for the great content, and keep it up!
Thank you so much!
This is so amazing! Like the new camera angle and the new presentation style a lot 🥰
Thanks Christian!
Gemme more ^^ I'd like to see ArgoCD also. Nice video btw, thank you!
Yes, Argo as well please!!
I‘ve been into ArgoCD a lot due to work, but Flux looks very awesome too. Especially the part where it monitors a Registry for new Artifacts, that seems super helpful. Will probably go the Flux route for my homelab now.
There's ArgoCD Image Updater that does the same thing, but that's a separate application/plugin you need to take care of. So there's tradeoffs with that.
The advice at the end is spot on. I am currently have a meltdown learning Golang.
I love the new angle, and your presence is night and day compared to older videos! you're a pro now!
Thank you so much! What a journey! Still more to come!
Tim I love your channel! Being in Devops, I learn so much from you! Thanks dude :)
Glad to hear it!
OMG IVE BEEN DOING THE SAME THING!!!!! I’m a developer too! Love your channel!
Perfect timing on this video. I just got my cluster up and running using the k8s-at-home template and it uses Flux, Kustomize and Helm. Its actually been pretty nice to work with. Thanks for another awesome video.
You are helping me so much learning about kubernetes. This is the only thing i still cant get my head around completely.
I was ready to ask what that UI was, then you answered it seconds later, thank you! Also great video like always!
Three things:
1. Is it possible to revert a deployment after it autodetects an new image? For example you found a bug and the fix will take some time.
2. Maybe take a look at k9s. Awesome k8s management tool I use.
3. Your content is like cocain to me! ;)
There is no revert, only move forward, just commit the old tag :) Also, thank you!
very clear explanation. Thanks a lot.
I would love to see this same video with included Terraform concept in it.
seems very nice. I like that it doesn't allow you to make mistake - no namespace in manifest - "namespace not specified, error: the server could not find the requested resource" and so on
You have no idea how much I've needed this video.
Working for huge corporation and basically this is my situation:
Ericsson - "You know some combination of programming languages right? Are you a fullstack developer?"
Me - "eh.. yes but no? You hired me for my Linux, Network and virtualization skills. I'm a infra guy."
Ericsson - "GREAT! Heres a fullstack developer assignment."
I love it but as Im building the front-end and the back-end.. Its getting REAAALY messy with each new modification.
Im ashamed to say that im basically copying the whole project folder into a backup folder before each major change...
This whole developing and deploying things into Kubernetes is such a messy web for me..
Great video 👍🏼 this made my k3s cluster so much better. Had to pauze and go back a couple of times as its a quite fast overview so you easily mis a crucial part.
This all genuinely looks fucking awesome, but there's a point at which this starts to feel as though you are learning new tools to manage the tools which manage your deployments.
Which then manage tools 😀. I hear you but what this does enable is kubernetes deployments from git that are repeatable and version controlled.
Thanks for the Lens tip! This is so much better then staring at terminals running watch 'kubectl get pods' :D
Great video would love to see more on this topic
I have been using kubernetes agent for gitlab-ci in my homelab for some time, I love it because it's easy and works with both helm and raw manifests. Never really been a fan of Helm, so most of my stuff is in manifests, but I have the option, none the less. The thing I like about flux that I didn't realise until watching this is update management. I use Keel currently because I like to be able to define a per service policy around tags add an approval step (via Slack bot) before an updated image gets rolled out; but this has its downfall in the fact that I often forget to update the tag in the k8s manifest with the new image after keel has done its thing. I then go and make a change to the manifest in my repo, push that change up and inadvertently roll back an image. Thanks, I might have a play later.
Awesome quality. I'm glad this found me. I actually needed this.
Great. Another maintenance window aka weekend fun at home. Ty Tim!
Been slowly building out my homelab again from the bottom up (PXE boot all machines to install Flatcar Linux, securely introduce machines to Vault and finally getting to Consul and Nomad to do something different from k8s and ECS at work) but your content is making me want to pivot to k3s, not going to lie. Would love to see a comparison with Argo. Keep up the great work mate.
Thank you so much! Will do!
Came back to say I'm all in on k3s and flux now lol
Your content is literal fire man. Love your channel! I have to try this out ASAP now lol.
This was super interesting, Thank You. I am curious what your Infrastructure as code looks like outside of Kubernetes. Say you have a new machine with Proxmox on it.
How do you handle the building and deploying of VMs? do you just run Terraform/Ansible locally? or do you have a process like Flux for all things outside of Kubernetes?
Awesome, crystal clear, and well-ordered video! Thanks a lot!
Hi Tim, I am a regular infrastructure engineer who specialises primarily in Virtualization and Storage - I am wanting to get into DevOps, I have no experience with coding though I am pushing myself to learn Python, YAML and Go - How do you keep yourself up to date with the latest 'solutions' such as FluxCD, Prometheus/Grafana, as well as others that make you then adopt that service? I would like to know where to go, in order to stay up to date with this sort of information - Great tutorial, I will definitely be tinkering with FluxCD soon on my homelab I'm sure!!
Thank you! Well our Discord is a good place! Lots of infra folks in there! Otherwise I just try to keep my pulse on the cloud since that’s where so much technology is being developed!
@TechnoTim did you miss a step in your docs? I had trouble getting the notifications working until I manually applied both the provider and alerts manifests. It didn't apply through the git commit automatically. Or am I not understanding that part right?
Would love to see a video on ArgoCD. Also for some teams a gate would be required to push to prod instead of automated sync.
Totally agree!
This are really Good, Great Job Tim 🔥
Johnny Depp teaching software concepts, so cool 😁
Always a joy to watchyour videos
Thank you!
awesome! Please show an overview of all alternatives like you suggested :)
whoa amazing technology, just .... amazing
Sad day for me, this channel has gotten too advanced for me. Miss the older days of getting hand held through docker or a network setup. I don't even know what I just watched. Looks like you changed the room you film out of... I guess I don't like change lol
Sorry!
OMG YOUR USING THE SAME DISCORD NOTIFIER SETUP!!
how do you use multiple values.yaml files in flux and also what if they are in different repo than the helm chart?
Literally exactly what I was looking for. Tough for a sec engineer to understand CI/CD pipelines. Simple and super useful video. Will be using this for my home lab.
Side note. Do you use multiple kubernetes clusters? I.e Rancher/K8s and K3s.
Thanks! I do. One for rancher and one for everything else!
Like mind, think the same. I am working to switch to flux as part of my move away from rancher project. Just got rke2 deployed sort of the way I wanted, so next step is setup flux to deploy things like Prometheus stack, metallb & etc
I've been following your channel and love your content. As a newbie who doesn't do this for a living, I'm making great progress and keeping up (IMHO) thanks to you and your videos. I'm just not able to make the leap from docker-compose to this. Any chance you could run through setting up some apps from k8s-at-home? There are a ton of great home lab examples which use flux and argocd, but I can't make the leap to understand what's going on. I love your HA Kubernetes + Ansible video BTW.
great content as always
from what i've read about flux it only works with k8s
i'm wondering if there's something like this that would work with things like ansible or other similar tools
Not with flux, I think you could just do this in CI though!
we have a Terraform controller that works natively for Flux to help you manage cloud resources.
Not sure if that's what you're looking for.
Would/Could/Should you run something like this in your production cluster in a release often setup, or would it be for something like test clusters or both?
Both. GitOps it’s the way!
What do you think about the Argo CD? I like flux cd but I prefer Argo cd over it. Argo projects have a lot to offer and fancy dashboards.
Soon!
Super useful! Thanks for that
Hi Tim, thank you for this really helpful tutorial. One question, what is the dashboard you are using 0:21? It looks really nice :)
Can you help me understand how do I bootstrap flux in multiple git repos at the same time
I think you should cover how to add SSH Public key to nodes for ansible and k3s. Its not explicitly covered and a lot of people will run into that as an issue.
I would just do an ssh-copy-id to ssh into the machines first
@@TechnoTim yea perhaps put in your web site docs for that step or description as a shown note.
I've been trying to obtain my cert in k8s and wondering whats the difference here between flux and argoCD?
Awesome content! I've been following for a while, and started leaning real hard into devops and trying to pick up as much as I can as fast as I can. While also trying to implement it at work. One question from this video I had was, does this setup sort of replace rancher or supplement it?
It can supplement it! Just worked through installing Rancher with FLUX so it's a really nice way to source control your cluster and deploy it like code!
@@TechnoTim I was just looking for how to upgrade the version of kubernetes. I used your ansible playbook to get mine up and running. Then put rancher on top of it but I can't upgrade it. The whole gitops strategy is looking real nice right now lol!
What is the benefit of using kubernetes over just containers?
Kubernetes is a container orchestration framework. Running a couple of containers on one server is easy, when you notice one is down, you can start it back up. But it becomes really difficult to manage when you run 20 services, with lets say 5 replicated containers per service over 10 nodes. To maintain 100 containers by yourself is not fun. So kubernetes task is to ensure desired state. If your service was defined to run 5 replicas per service, kubernetes will ensure that it keeps desired state, if a node goes down, kubernetes will notice it and makes sure that the dead containers are rescheduled. Then theres things like scheduling algorithms, etc.
Did you think about using external secret operator instead of sealed secrets? You can fetch your secrets from a hasicorp vault or a secret management solution of your cloud provider
I use mozilla sops
HOLY SHIT u went off on this one
very tiny suggestion, it is actually a nit since I am very noob in infra topics, in a single project, please favor only .yml or only .yaml but not both at the same time, consistency is really a good thing in a monorepo :)
When you created the manifest in the "default" folder. Does that folder have any meaning? or just your way of organizing namespace?
Just organizing according to name space.
@@TechnoTim thanks for the clarification. you have any advanced flux stuff? like variable replacement where those vars are stored in the ci/cd tool?
Where can I learn to setup this workflow?
Hi Tim I'm getting an error when bootstrapping "waiting for Kustomization "flux-system/flux-system" to be reconciled" any idea please ?
Check your tokens or look at the logs!
Can you cover nomad?
What are you using for diagrams? Looks sleek!
Diagrams and licid charts!
Hey! Awesome! I just made a video about Argo CD! About the same as Flux. I like Argo better tho.
One thing I'm still trying to figure out is, how to make things work with a cluster that already has a number of applications running. Basically how to retrofit Flux into a working cluster.
use the manifest or the helm chart you were already using
@@regis9596 Yep. I'm understanding Flux better now, thanks!
What's the dev prefix you are using in the terminal? I haven't seen that one before
It’s the folder I am in. I am using zhsell with oh my zsh
@@TechnoTim Haha, that makes sense..
Do you have any experience using Flux in a multi-cluster/multi-env infrastructure? I manage 24 K8s clusters across multiple datacenters on numerous continents. I've been looking for an effective GitOps solution that allows me to stagger deployments across environments. Given our breadth of stakeholders, we unfortunately cannot implement a true CI/CD workflow, as multiple internal and external stakeholders may be validating new features in lower environments, and are not yet ready for them to go to prod. The problem with the GitOps model in this scenario is that it requires either managing multiple branches (eww), or multiple repos (ewwer).
You can manage multiple clusters and multiple envs per cluster with Flux. I think Fleet also does GitOps at scale (millions of clusters)
Why did you pick Flux over Argo?
I like the simplify, low resources, and pull model. Will be reviewing argo too!
You’re basically me just not bald lol
Argocd please Timmy.
Great. Just great content
Thanks!
Wow! Thank you much!!!
cool & perfect 👌🙏🙏🙏
"High availability home cluster" - it can't be both. Some internet or electricity failure and you're out
backup internet and UPS ✅
Thank you
Woooooooooooooooo!!!!!
Listening to this at 0.5x did not help. Far to fast for me to follow
Sorry, you can pause and I have the documentation on my site too.
First
much to fast for me
That's why the pause button exists, making instructional videos like this great on TH-cam. Tim also links to his show notes in the description, which should get you sorted.
Sorry!
I Will watch again and read thé notes. Great content of course as usual. Thank you Tim
Buzzword Soup.
I forgot a few
web3, crypto, nft, ai
@@TechnoTim true. Lol