Load balancing Airbyte workloads across multiple Kubernetes clusters

How do you load balance long-running Kubernetes workloads across multiple clusters?

At Airbyte, as part of supporting multiple geographic regions for data replication workloads, we adopted a control-plane/data-plane architecture. A control-plane orchestrates data movement workloads across multiple data-planes. Technically speaking, each plane is a Kubernetes cluster.

Our solution to load-balancing workloads across multiple data-planes is to push down the responsibility to the data-plane itself. We enqueue workloads in a single job queue and let the data-planes compete for jobs to process if they have capacity to do so. This has the benefit of treating capacity as a problem that is local to a cluster, removes the complexity of planning ahead for available resources, and keeps operations simple when facing cluster downtime.



https://redd.it/1dz4tih
@r_devops

Gitlab - Syft/grype: Are there any GOOD resources to learn how to set up?

I'm new to Devops. I, along with a coworker, am tasked with getting container vulnerability scanning and SBOMs generation set up. I've been looking for a decent video or webpage that goes over the implementation of syft and grype but have failed to do so. Even the one on posted on the documentation section refers to a video that I don't think helped me much. Could be that I just don't understand what exactly I am supposed to take from our AWS EKS images/containers to input into the .gitlab-ci.yml file. Does anyone have any tips and/or sites they can refer me to so I can get a better understanding of the steps involved? And before you ask, no, we don't have the option of using an alternative. This is what THEY want and paid for (Gitlab ultimate).

https://redd.it/1dz2vwc
@r_devops

Cloud Deploy Skaffold overwriting Terraform

Hello, does anyone have experience using Cloud Deploy / Skaffold in conjunction with Terraform?

I'm setting up a Cloud Deploy pipeline for the first time (previously had a simple Cloud Build setup for deployments). However, I noticed that my server configuration defined in Terraform (e.g. scaling, service account, etc.) is being overwritten by new Cloud Deploy releases.


Question: Is there a way for Cloud Deploy / Skaffold to only update the container's image while leaving all other parts of the configuration alone, to be managed by Terraform?


skaffold.yaml:

apiVersion: skaffold/v3alpha1
kind: Config
deploy:
cloudrun: {}
profiles:
- name: development
manifests:
rawYaml:
- run-development.yaml

run-development.yaml

apiVersion: serving.knative.dev/v1
kind: Service
metadata:
name: my-service
spec:
template:
spec:
containers:
- image: image


I can migrate all the config to skaffold, but I'd prefer to keep it in Terraform.

https://redd.it/1dzafii
@r_devops

Miss the ctrl-alt-delete life.

Just had nostalgia now when I thought back to the days when computers became popular. Good old days!

https://redd.it/1dzcktl
@r_devops

OpenTofu 1.8.0-beta1 is out and ready to test!

We've polished up the alpha1 and are hoping to get some feedback on the new features in OpenTofu 1.8.0!

https://opentofu.org/blog/opentofu-1-8-0-beta1/

https://redd.it/1dzcoqo
@r_devops

Dynatrace vs. Datadog with AWS monitoring

Hello all! I am wondering if anybody has experience with both Dynatrace and Datadog in terms of monitoring AWS specifically. I am not able to identify the pros and cons of each since I don’t have access to both tools. Any shared experience would be greatly helpful, thank you!


https://redd.it/1dza6nv
@r_devops

cloud for backend services

I'd like to develop and manage a few backend services for my side future commercial project. I'm proficient in java/sql, I know/seen some reports that even js/nodejs seems te perform better in the cloud thus I open to switch the stack

But on the other hand I'm working with quarkus/kubernetes on daily basis, I love kubernetes and quarkus address some issues with java based containers and so this is my prefered stack. Also I like to have full control how the requests are processed so I',m a bit resistant in sth like functions

I assume the traffic would be pretty low but the availability must be 24/7 at low/reasonable price

I did some reserach but yet cannot decide, switching in the future would be hard especially if I use functions offered by some providers instead of docker containers/quarkus/spring boot

I think I should decide between:

* digital ocean: managed kubernetes / app platform + managed postgres

Looks like a dead simple solution from devops point of view, I know the max cost so I can sleep well at night

* gcp: bigtable + cloudrun looks like an almost ideal solution

The only thing is I have no trust in google and their support, and I prefer working in aws terminal than gcp

* aws: with lambda and dynamodb propably could cost me nothing. rds I guess could be too costly at the beginning

But it requires to spend much more time now and in the future; you know even choosing the right services among many requires some research, also I need to learn how work with dynamodb, design proper schema ,get familiar with other services billings, cloudwatch etc etc
Do you have any strong opinion which way to go/no go from devops point of view?

https://redd.it/1dzdxyk
@r_devops

splitting larger commits into smaller commits

hi all I wanted some advice on how to split a larger commit into smaller commits. The intent is that I want to take the smaller commits and create smaller prs. In a nut shell I created a giant pr that is really hard to review and don't want to stress out other devs.
<br>
These are the steps I was taking to split a larger commit into smaller commits. It has a code smell, so please let me know what I'm doing wrong
<br>
git checkout  <largeCommitHash>
git rebase -i HEAD\~1
The interactive rebase editor will open,
Change the pick command to edit for the commit you want to split
git reset HEAD\^
git add changes I want for first part of the commit
git commit -m ‘smallerCommit1’
git add other changes
git commit -m ‘smallerCommit2’
git rebase --continue
git checkout -b smallerCommits
git checkout original_branch_with_large_commits
git rebase smallerCommits
git push origin original_branch_with_large_commits --force

Thank you. please let me know if anything is unclear

https://redd.it/1dzjveb
@r_devops

Is Serverless becoming a hated company now?

They keep optimizing to make money by not releasing features in v3

https://redd.it/1dzlbed
@r_devops

Need help to create a CI/CD pipeline

I am new in devops and only member in the team. I have to create a CI/CD pipeline to deploy service now code. Have to decide tools for source code, artifact, build tools, testing and quality. I find github for source code and Jenkins as build tool would be ideal. please help me to define all aspects to create a CI/CD pipeline.

https://redd.it/1dzm795
@r_devops

On-premise infrastructure vs. hosting with the hyperscalers

Compare the Total Cost of Ownership (TCO) of running on-premise cloud infrastructure and hosting with the hyperscalers. Use the ShapeBlue calculator to evaluate the costs of using hyperscalers like AWS, Azure, or GCP versus managing your own infrastructure. What savings can you achieve with Cloud Repatriation?

This Cloud Cost Calculator allows you to compare the total costs associated with running workloads on different hyperscalers against on-premises workloads using Apache CloudStack. With the calculator, you can see the TCO for running your own infrastructure for 36 months and compare it against using instances/virtual machines from AWS, Azure or GCP. Operating an on-premises Apache CloudStack infrastructure involves expenses for datacenter facilities, software, hardware, licensing, and support.  The calculator is built in Microsoft Excel and is customisable for your needs.

https://www.shapeblue.com/cloud-cost-calculator-and-cloud-pricing-report/

https://redd.it/1dzq0e7
@r_devops

Mastering GitOps: ArgoCD vs. FluxCD - Complete Guide with Demo

I wrote a blog for beginners, comparing ArgoCD and FluxCD for mastering GitOps in Kubernetes. It covers core principles, key features, installation steps, and best practices: https://www.cloudraft.io/blog/argocd-vs-fluxcd

https://redd.it/1dzrep6
@r_devops

Istio Service Mesh inter-service communication.

So I am pretty much a beginner in DevOps. I have been asked to design architecture such that there are 4 microservices (let's call them A, B, C, D). A and B are public facing. So each of their Deployments in K8s are exposed by a LB Service behind a single Ingress.

However, C and D are not public facing. Only B can communicate to the C and D microservice deployments.

All deployments are autoscaling and the org is using Istio Service Mesh. My question is:
1. Is there any specific library to connect to Proxy Sidecar from Service Pod (apps A and B running in the respective Pods are written in Go and Java while C n D are written in Node)?

2. When I will be trying to communicate to C from B through Istio Proxy, would Istio Load Balance between C's Deployment Pods automatically?

3. A emits event which D needs to listen to. Org has proposed Kafka. How would an app running in K8s Deploy emit and receive event (I know it is a noob question)?

Any help is much appreciated.

https://redd.it/1dzs6fy
@r_devops

Oracle's Container Instances deploy new image CICD?

Hello everyone! I am working with Oracle's `Container Instances` service and I try to create my CICD deployments but I cant find any documentation related.



I cant even find how to deploy a newer version image.

The only thing I thought was create a custom flow with oci-cli like:



* Create a new container with same settings on the same ContainerInstance with new image version

* Make a healthcheck if the new container is up

* If ok kill the original container



Any thoughts?

https://redd.it/1dzt0sf
@r_devops

Do you have a self paid work phone?

I've thought off and on about separating work from personal, by getting a separate phone that will only be for work things.

I currently have an android and I'm noticing my company is getting more serious about security and with Android seems they can admin my phone in thinking it's time to separate devices.

https://redd.it/1dzso5j
@r_devops

What's in your Observability/APM Stack?

I found some odd answers while at KCD New York last months and I'm curious how y'all are doing your production monitoring

For my team:

* OpenTelemetry Rails instrumentation
* Coralogix - [receiving OTel traces](https://coralogix.com/docs/opentelemetry/)/metrics
* Checkly - synthetic monitoring and [visual regression testing](https://www.checklyhq.com/docs/browser-checks/visual-regression-snapshot-testing/?utm_source=chat&utm_medium=link&utm_campaign=synthetics&utm_id=social_button)

We are considering implementing our own prometheus/grafana solution but like a few of the Coralogix features and UI

https://redd.it/1dzts3i
@r_devops

Do you perform house cleaning and delete old container images that are not in use anymore?

Or should you keep the images? Not sure what should we do with old deprecated images. I keep receiving alerts from security team that the deprecated images have security vulnerability

https://redd.it/1dzubau
@r_devops

Serverless vs Managed Kubernetes

I like the ease of serverless, but I like the standards given by Kubernetes better (although, not at the cost of managing everything manually)

Do you have an opinion about the comparison between Serverless and Managed Kubernetes ?

Some interesting comparison criteria in my opinion would be (not exhaustively):

- Ease and speed to first deployment
- Maintainability / How easy and how much time (thus, also how much people) would you need to maintain it
- Future-proofness
- Cost (usual trend between them)
- Portability / Ease of transitioning of vendors
- Performances
- Extensions/Plugins to other services (i.e. GitHub Actions
- Valuation (which would make investors generally like the project more)

https://redd.it/1dzr2iy
@r_devops

Whats the most complex piece of technology in DevOps currently?

So, I understand that tech today is advancing at a very fast pace and we as people haven't been able to catch up to it at the current pace (I'm talking about majority of the people in the current time).
Just wanted to get the idea about what are the technologies that are complex right now.
Thank you!

https://redd.it/1dzzshs
@r_devops

Using Docker in Production

What are your "best practices", guidelines, tips and big no-nos for using Docker in productive environments?


Not looking to containerize own apps, but mainly utilise quick POC-testing, easy deployment for trials of new tools etc...
Is using a specialized host OS like AWS Bottlerocket worth it?
Would love to get some discussions going and see what everyones experience is, thank you all for your input :)

https://redd.it/1e006l2
@r_devops

Action Version: utility to pin github actions versions

Heya folks

I've made a utility for myself for one-off jobs where I need to quickly pin the versions for all the github actions in a repo

I've also made it auto-fix any documentation that uses semver tags instead of hashes, since people copy-paste these things

Maybe useful to others 😊

https://github.com/audunmo/action-version

https://redd.it/1e01lkn
@r_devops