Software Updates and Breakage

Recently in Ubuntu MATE 16.04 there was a kernel update to 4.4.0-143. This broke VirtualBox. Browsing the web indicates the kernel update breaks any driver that needs to be recompiled.

On another Ubuntu MATE system recently updated, when booting to run level 3 (multi-user target), the system refused to fully boot, always hanging with the repeated message /dev/getty: No such file or directory.

I accept that computers are incredibly complex. Almost hopelessly complex. I accept and presume that upstream developers work hard, are sincere, and do their best each day.

Yet every little patch these days, regardless of the operating system, seems to cause some kind of butterfly effect. Online forums and ticket systems are filled with reports of this kind of breakage. Daily.

In this case the problem could very well be VirtualBox and not the kernel update. Regardless, straightforward testing should have revealed breakage. Virtualbox should have been patched before releasing the kernel update.

The /dev/getty error? I have no idea. Some days later the bug disappeared.

I am aware of the traditional smug response that users and admins should read change logs before applying patches. That advice fails because reading a change log will not predict the breakage. This is supposed to be the purpose of upstream quality assurance testing.

Modern software quality assurance is a mess.

Posted: Category: Usability Tagged: General

Next: Migrating a Business to Linux — 15

Previous: Musings About RAID