Signs your working system has quietly become a liability

A system can run without errors for years and still have become a liability — uptime measures whether it’s working, not whether it’s safe to keep depending on. These are the signs worth paying attention to.

Nobody wants to be on-call for it

If the on-call rotation quietly reshuffles around who’s willing to touch a particular system, that’s a signal the risk is already understood informally, even if it’s never been said out loud in a planning meeting.

Every change needs a specialist

A healthy system can be safely modified by most of the team. A liability requires one specific person, every time, because the rest of the team doesn’t have enough context to change it without meaningful risk of breaking something non-obvious.

The vendor or platform underneath is aging out

Systems built on a database version, language runtime, or platform that’s approaching end of life carry a deadline whether or not anyone’s tracking it. The system working today doesn’t mean it will keep working once security patches stop shipping for what it’s built on.

The workarounds have workarounds

When the operational runbook includes steps like “restart it before it hits this known limit” or “don’t run the batch job on days when X,” the system is being kept alive by institutional knowledge rather than by design. That knowledge doesn’t scale, and it doesn’t transfer well.

What to do with this list

None of these show up in an uptime report. If two or more apply to a system your business depends on, it’s worth a proper risk audit — not because the system is about to fail, but because nobody currently knows how close it is to failing, and that’s the actual liability. See the migration risk audit, and why it comes first for what that actually involves.