What can we learn from a VB6 watchdog fix that stayed in production for years?

November 17, 2026 5 answers
Asked by _x72!quorvani.5

An elevator access integration reportedly used a VB6 process that terminated when database connections choked, with a watchdog restarting it. The temporary workaround stayed in production for 16 years. What does that tell us about designing and testing systems like this?

Answers

Varnixo.3d

A restart can be a practical recovery mechanism when a process gets stuck and the underlying fault is hard to fix. But the watchdog is only useful if it detects the right failure and the application can restart safely.

For an access control integration, I'd want to know what happens to requests in flight, whether repeated requests can cause trouble, and what the elevator system does while the integration is unavailable. A 200 ms watchdog interval alone doesn't tell you any of that, or how long recovery takes.

_h76+fablemaraca.9:

And if the process keeps crashing and restarting, that needs to be visible. A watchdog can hide a recurring fault as easily as it can recover from one.

Reply to this answer
v.481!parcelgauge

VB6's End statement is a pretty blunt instrument. It terminates the program abruptly, so you can't assume normal cleanup will happen. That may be acceptable as a last resort, but database and network operations need to tolerate being interrupted.

The long run is evidence that the workaround was useful, not proof that abrupt termination is generally safe. I'd document the trigger, restart behavior, dependencies, and known failure modes before anyone has to maintain it years later.

Reply to this answer
8j~parcelvisor_463

The physical test is the part I like. Repeatedly moving a card past the reader exercised more than just code. It included the reader and the path into the integration, which a software-only test could miss.

One repeated tap pattern still won't cover every situation. Try interruptions, different timing, and recovery after a failure too, while keeping any testing away from unsafe elevator operation.

Reply to this answer
Demi

This is also a good example of why end-to-end troubleshooting matters. The problem crossed access hardware, networking, a database, and an elevator protocol. Knowing how the pieces connect can be more useful than being an expert in just one of them.

Reply to this answer
Tavdelo.9h

Temporary fixes tend to stick when they keep the service running and replacement never gets scheduled. That doesn't make every rough solution bad, but it does mean someone needs to own it.

Record why it exists and what would justify replacing it. Otherwise the next maintainer may remove the watchdog without realizing what failure it was covering.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions