How to Deploy, Update, and Scale a Custom App on an On-Premise Kubernetes Cluster

·1 views

Part 6 of this Kubernetes on-premise series walks through building a Docker image for a custom Spring Boot REST API and publishing it to Docker Hub or a private registry. The guide demonstrates creating a Deployment using the kubectl command line and exposing it via a LoadBalancer Service with a manually assigned external IP, as on-premise environments lack cloud-native load balancer provisioning. It covers rolling updates using kubectl set image, which replaces the running container image with a new version while recording the change in revision history. In case of issues, a rollback to the previous revision can be performed instantly with kubectl rollout undo. The article also explains both manual replica scaling with kubectl scale and automatic scaling through a Horizontal Pod Autoscaler configured to maintain CPU utilization below a defined threshold.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Developer builds reusable AWS EventBridge scheduling primitive for deferred workflows

A developer building automated workflows for the Ready, Set, Cloud platform needed a simpler way to schedule deferred events without relying on AWS Step Functions Wait states. The solution involved placing an EventBridge Scheduler behind an EventBridge rule, allowing any service that can call PutEvents to request a delayed event at a specified time. A Lambda function was added as an intermediary to upsert schedules, validate payloads, and handle cases where the scheduled time has already passed. The approach was integrated into a shared platform repository, making the scheduling primitive reusable across multiple applications in the same AWS account. The developer noted that while the Lambda layer adds complexity, the tradeoff enables a more flexible and maintainable scheduling experience across their entire app ecosystem.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Green Dashboards, Angry Customers: Why Infrastructure Metrics Miss the Real Picture

A software engineering team once spent nearly an hour troubleshooting a major customer-facing outage — failed payments, endless loading screens, and social media complaints — while every infrastructure metric showed normal. The root cause was not a server failure but a breakdown in the actual user experience that traditional monitoring tools were never designed to detect. This incident prompted the team to shift from measuring system health to monitoring complete user journeys, such as whether a customer could successfully sign up, search, or complete a purchase. Engineers often default to tracking what is easiest to measure, like CPU and memory usage, rather than what matters most to end users. The key takeaway is that green dashboards reflect infrastructure status, not customer success, and teams that conflate the two risk silent, costly erosions of user trust.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Google Offers Three Product Visibility Tools Depending on Inventory Type

Google provides three distinct tools for retailers to display products online: Google Business Profile's Product Editor, Local Inventory, and Merchant Center. Businesses selling items without barcodes can manually upload product details and images through the Product Editor, while those with barcode-based stock can automate listings via Local Inventory using UPC or EAN identifiers connected to a point-of-sale system. Merchant Center serves retailers seeking broader distribution across Google surfaces such as Search, Shopping, Maps, and YouTube, including multi-store management. The choice between tools depends on whether a business has standardized product identifiers, requires store-level inventory automation, or wants visibility beyond a single local profile. Importantly, lacking barcodes does not prevent product visibility, as the Product Editor offers a manual entry path for eligible products.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Moonshot AI Releases Kimi K3 Open Weights: A 1.56 TB Sparse MoE Model on Hugging Face

Moonshot AI last week published the Kimi K3 model weights on Hugging Face, making available a 2.8-trillion-parameter sparse mixture-of-experts model compressed in MXFP4 format. The full checkpoint spans roughly 1.56 TB across 96 safetensors shards, accompanied by configuration, tokenizer, and evaluation metadata files. Serving the model requires compatible software such as vLLM or SGLang, with support varying depending on the user's GPU generation. The model is released under a custom, non-OSI license that permits commercial use but imposes revenue-tier conditions and lists prohibited use cases. Developers are advised to pin a specific repository revision before downloading to avoid inconsistencies if the upstream files are updated.

0 comments Read more at DEV Community