After the cPanel hack last March or April, our VPS was moved from a Centos78 machine to an AlmaLinux8 machine. None of my web apps now work. Some had been working more than four years. The hosting company’s support team has been unable to get it to work. This the the guts of their last email concerning the problem:
We ran a thorough diagnostic pass directly on your VPS and can rule out several things with confidence:
- Not a firewall or network block. Requests made from the server to itself still fail, so nothing external is involved, not our network, not our firewall.
- Not a missing or incompatible library. Your MeterWizard binary loads cleanly with all dependencies resolved. There’s no glibc symbol error, which is what you’d expect to see if this were really a glibc version shortfall.
- Not a container level restriction. We confirmed your container configuration hasn’t changed.
- Not a system zlib mismatch. Your app bundles its own compression library instead of using the OS’s.
- Not resource limits or file capabilities. Both look normal.
What we’re actually seeing is that every Xojo app on the server fails the same way when serving CSS and framework assets. The requests come back with correctly generated headers (proper ETag, content type, cache settings) but a 404 status and an empty body. That means the app recognizes the resource but isn’t returning its content. That’s a specific behavior in how the Xojo web framework serves its embedded assets, and it’s not something we can diagnose or fix from the OS or infrastructure side.
My recommendation is to reach out to a Xojo developer or Xojo’s own support team with these specifics. They’ll have visibility into the framework’s resource serving code that we don’t, and they’re in a much better position to tell you whether this is a known issue on RHEL family distros or something particular to how this app was built or deployed.
Any help with this would be greatly appreciated.