puzl.cloud

Kubernetes 1.36 is now supported by kubesdk. Make sure to upgrade kube_models to version 0.5.0 to get access to the new resources. github.com/puzl-cloud/kub…

We've upgraded the runmyjob.io cache action github.com/puzl-cloud/git… Now on Node 24 + ESM, with refreshed tooling and cache round-trip validation. Cleaner, future-ready GitHub Actions caching for jobs running on RunMyJob. Big thanks to @LayZeeDK for contributing 🙌

kubesdk — async-first, fully typed Python SDK for Kubernetes Open-source Python SDK with fully typed models, async client, and multi-cluster support for Kubernetes automation. github.com/puzl-cloud/kub…

kubesdk is a fully typed, async-first Python Kubernetes client with a CLI that generates models from any live cluster or CRD, achieving over 1000 RPS on large, multi-cluster workloads ➜ ku.bz/x-z5V_BPd

Certified innovation for cloud workloads. Puzl Cloud is officially SOC 2 Type II certified. Run your infrastructure on Spike Instances™ to eliminate idle CPU/RAM costs. The new FinOps paradigm of true load-based billing, now validated by rigorous 3rd-party security audits.

#Kubernetes CRD generator from your #Python dataclasses is finally here in the fresh release of kubesdk Build #k8s operator and its CRD in just a few lines of Python, with full IDE autocomplete throughout your code. github.com/puzl-cloud/kub…

kubesdk v0.2.0 is here 🎉 A few improvements, plus a ready-to-run example to help you quickly build your own #Kubernetes operator with a CRD. #k8s #OpenSource #CloudNative github.com/puzl-cloud/kub…

@_smitop @valigo Yeah. That's why we've made CI runners service, where your CPU utilization is the only thing that actually affect billing. And we don't use Microsoft's bash scripts! puzl.cloud/products/run-m…

@valigo Actions usage is billed by minutes of usage; your CPU utilization rate doesn't affect billing

GitHub's official "safe sleep" script: >is not safe >does not even sleep Microsoft just can't stop losing

Launching kubesdk 🚀 A developer-friendly, fully-typed, async #Kubernetes client for #Python, designed for large-scale multi-cluster workloads. ✅ Fast ✅ CRD & API models generator built in ✅ Great DX vs the official client Check it out 👇 github.com/puzl-cloud/kub… #devtools #k8s

We've rolled out updated #GitLab runners for runmyjob.io. - Improved stability and job starts. - @gitlab < 15.10 is deprecated. - Still the best cloud GitLab runner since 2023, trusted by dozens of teams. Follow for updates!

✅We now have a public status page for puzl.cloud! status.puzl.cloud Bookmark it and subscribe for real-time updates on incidents, maintenance windows, and performance metrics.

🎉 RunMyJob now supports GitHub Actions! • Load-based billing — don't pay for idle CPU & memory • Up to 2× faster & 5× cheaper than your runners • Drop-in install: just swap runs-on: puzl-ubuntu-latest, done 👍 Try for free 👉 puzl.cloud/products/run-m…

@semanticbeeng It could work technically, but you will face extra latency with such a scheme. GPU-over-IP is something that @elonmusk is building with his @xai

@puzl_cloud > "... it isn't possible for GPU due to limitations of Nvidia drivers and the way how the GPU memory is allocating for your process" x.com/tom_doerr/stat… Would SCUDA help?

💡Let's talk efficiency! Puzl's Zero Idle cost model is reshaping our approach to #CloudComputing, ensuring you only pay for what you use. Our latest article delves into the details of Zero Idle approach and how it resonates with #FinOps. puzl.cloud/blog/embracing…

@semanticbeeng It's done normally like in other cloud providers. GPU is passing directly to your VM. If you need to stop billing, you should stop your VM.

@puzl_cloud How to stop GPU billing exactly? Stop whole VM or just the processes that allocated the GPU? How to learn more about the technical aspects on how GPU allocation is done? For example: uses MIG? nvidia.com/en-us/technolo…

@semanticbeeng Unfortunately, it isn't possible for GPU due to limitations of Nvidia drivers and the way how the GPU memory is allocating for your process. So such a billing approach works for CPU and memory only.

@puzl_cloud > "Puzl introduces a game-changing approach — Zero Idle costs, ensuring organizations only pay for resources (CPU and memory, in particular) their workloads actually utilized" Also GPU? Can I have an LLM inference node with GPU allocated 24hr and only pay for 8hr when used?

☁️ Enhanced Cloud Storage for #GitLab CI runners! ⚡️Streamline your #CI workflows with shared pipeline job caching across multiple runners. Tackle runner concurrency with ease! 🛠️Create multiple shared persistent directories for seamless data exchange. docs.puzl.cloud/ci-runners-for…

🚀 Exciting updates at Puzl! New Cache Retention Time feature for your #GitLab CI jobs - use #cloud storage wisely & cut your costs! #DevOps #FinOps Details 👉 #spec-resources-persistentStorage-cacheRetentionTime" target="_blank" rel="nofollow noopener">docs.puzl.cloud/api/ci-runners…

@srchvrs I was using @puzl_cloud recently and they seem fairly priced with decent availability on A100-40Gb

🛠️Facing challenges in managing extensive #GitLab runner fleet? Our new #Terraform modules offer an effective solution, complementing areas where GitLab's native tools fall short. It's especially useful if you have more than a dozen of runners to manage. gitlab-pipelines.puzl.cloud/docs/terraform…