SShortSingh.
Back to feed

Staging Config in Production Caused Silent Data Split After Column Rename Deploy

0
·2 views

A backend team deployed a database column rename on a Sunday night, expecting a smooth migration, but the migration never ran because production was using a staging configuration file. The file had two dangerous settings: one that disabled Flyway migrations and another that let Hibernate auto-update the schema instead. Hibernate silently added the new column while leaving the old one intact, resulting in a table with two columns where data was split between them. The issue went undetected for eleven minutes since no exceptions were thrown, and it was only flagged when a downstream team noticed customer status fields appearing empty. The incident highlighted the risk of copying staging config files into production without auditing inherited settings, and was resolved by re-enabling Flyway and switching Hibernate to validate mode.

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)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

A Developer Reflects on Navigating an Extrovert-Configured World as an Introvert

A software developer shares their personal experience of identifying as an introvert and how that self-awareness has shaped their daily life and professional interactions. While the realization brought clarity, it also triggered constant self-monitoring during conversations, which the author links to developing an overthinker's mindset. The author notes that even on social media, their behavior mirrors their reserved real-life personality, rarely posting or engaging publicly. In the tech world specifically, they find networking and events like hackathons particularly challenging, despite not considering themselves shy. The piece closes by inviting readers to reflect on which introvert or extrovert traits they wish they possessed.

0
ProgrammingDEV Community ·

How to Pick the Right Kubernetes Alternative for Your Team's Needs

A new guide from Lizard (lizard.build), reviewed in September 2026, outlines how teams can choose a Kubernetes alternative based on which infrastructure tasks they want to stop managing. Options range from managed app platforms like Railway and Render to schedulers like Nomad and server-based tools like Kamal, each covering a different slice of operational responsibility. The guide advises teams to first list every task they currently handle — such as cluster upgrades, networking, and storage — then determine which tasks stem from application needs versus cluster architecture choices. Managed app platforms suit teams running standard web services and workers, while cloud-native options like ECS Fargate and Cloud Run fit those already embedded in a specific cloud provider's ecosystem. The core warning is that switching to a different scheduler or platform does not automatically reduce operational burden unless the new system genuinely eliminates the tasks driving the migration.

0
ProgrammingDEV Community ·

E2B vs Lizard and Rivals: A Cost and Feature Guide to AI Code Sandboxes

A technical guide published by Lizard (lizard.build) compares cloud sandbox providers for AI agents, focusing on pricing, session limits, concurrency, and lifecycle management. E2B offers a free Hobby plan and a Pro tier at $150 per month, with an illustrative 1,000-hour workload costing roughly $315.60 on Pro before taxes and extras. Under the same workload assumptions, Lizard's compute bill comes to approximately $100.05 with no monthly base fee, making it the cheaper option in this rate-based comparison. The guide also lists Daytona and Modal as alternatives, each with distinct billing structures and lifecycle controls worth evaluating. Pricing figures were verified against published documentation on 9 and 15 September 2026, and the calculations reflect stated rates rather than measured real-world performance.