> Unit tests are by their definition not the kind of thing that simulates this kind of failure.
Thorough unit and integration tests include failure modes. Mine include things like "what happens if the OS reports that the storage was unmounted during read/write" (because that was a failure often seen in production with some flaky SAN devices) and "what happens if the server stops responding" (because networks are generally unreliable) and "what happens if invalid (random) data is given" because data corruption is a frequent occurrence for similar reasons.
> Perhaps you have a false sense of security about what you’ve really been testing?
Perhaps. On the other hand I've seen a _lot_ of other developers test only the happy path and call it a day then spend days/weeks/months debugging failures.
Thorough unit and integration tests include failure modes. Mine include things like "what happens if the OS reports that the storage was unmounted during read/write" (because that was a failure often seen in production with some flaky SAN devices) and "what happens if the server stops responding" (because networks are generally unreliable) and "what happens if invalid (random) data is given" because data corruption is a frequent occurrence for similar reasons.
> Perhaps you have a false sense of security about what you’ve really been testing?
Perhaps. On the other hand I've seen a _lot_ of other developers test only the happy path and call it a day then spend days/weeks/months debugging failures.