← Back to Portfolio
E-Commerce Site Recovery Migration Fix

Rattenkrieg

A live e-commerce site brought back from the dead - migrated, debugged, and fully restored after a hosting move left it completely offline.

Brief / Summary

A broken migration, rebuilt to fully working - with every customer's purchase history intact

After Rattenkrieg's storefront was moved to a new host, it went completely dark - every visitor hit an "Internal Server Error," and the customer login system that gates purchased game credits was dead. We diagnosed the site layer by layer instead of rebuilding from scratch: restoring missing framework files, correcting every scrambled database connection, and untangling a credit system that had been reading data across two separate stores. The result is a fully live site where customers can log in and see the exact credits they paid for, with not a single purchase record lost in the process.

The Recovery

The Highlights

Six specific problems from the live recovery that we'd point to as proof of the work.

Live Site
rattenkrieg.com
Rattenkrieg storefront, restored and back online

From "Internal Server Error" to a working homepage

Diagnosis

The site greeted every visitor with a blank server error. We traced it through the hosting layer - a leftover configuration line from the old server incompatible with the new one, an outdated routing setup, and a stray debug line left in the site's core file that was halting the entire application before it could start.

Key insight
The migration hadn't just moved the site wrong - it had stacked three separate failures on top of each other. Peeling them back in the right order was the difference between a rebuild and a repair.

The right software environment

Diagnosis

The new host had auto-upgraded the site to a modern software version its older codebase simply couldn't run on, producing a flood of errors. We identified the exact compatible version and pinned it, instantly clearing hundreds of errors at once.

Key insight
Newer isn't always better. Matching the environment to the code the client already had - rather than forcing the code to change - got the site stable in minutes instead of days.

Restoring the missing pieces

Diagnosis

The migration had silently failed to copy several essential framework files - the site's session handler and its entire database engine were simply gone. We restored them from the exact matching version of the framework, file for file.

Key insight
This is why the site failed in stages rather than all at once: each missing piece only broke the moment a visitor reached the feature that needed it. Login was the last domino.

Reconnecting three databases

Diagnosis

The site actually relies on three separate databases - the main store, a copy of historical purchase records, and the blog. Every one was pointing at credentials from the old server. We remapped each connection to the new host and, in the process, rotated passwords that had been sitting exposed in plain text.

Key insight
The new host labels its databases one way but names them another internally - a trap that had defeated earlier fix attempts. Spotting that distinction was the key that unlocked the whole login system.

Rescuing the customer credit system

Diagnosis

The login worked, but the feature that mattered most - showing each customer the credits they'd bought - was silently returning nothing. We traced it through the purchase-history data, confirmed every customer's orders had survived the move, and reconnected the logic that turns past purchases into available credits.

Key insight
We verified against real customer accounts that the credit totals were correct - not just that the page loaded. For a business whose customers have paid money, "looks fixed" and "is fixed" are very different things.

A roadmap, not just a patch

Diagnosis

With the site live, we documented what was fixed, flagged the security and future-proofing issues we found along the way, and mapped out the bigger decision ahead: how the credit system should work once the connected store finishes its own move to a new platform.

Key insight
An emergency fix that leaves the client blind to the next problem isn't finished work. The handover doc means they know exactly where they stand and what comes next.
YOUR TURN

Site down after a migration?

We can help. No obligation.