Design the break-glass procedure and test it
Task
Federation creates a dependency: if the identity provider is unavailable, nobody can sign in, including the people who would fix it. Design the exception, then test it -- because an untested break-glass account is an assumption.
Steps
- Write
lab/breakglass.mdspecifying: how many accounts, where their credentials live, what second factor they use, who may use them, under what circumstances, and what must be recorded afterwards. - State explicitly where the credentials are stored and confirm it is OUTSIDE the environment they recover. If they live in a password manager that itself uses the federated identity, write why that fails.
- Configure an alert that fires on ANY use of the break-glass identity, to more than one recipient.
- Test it: simulate the identity provider being unavailable, sign in with the break-glass account, confirm the alert fired, and record the elapsed time in
lab/breakglass-test.md. - Rotate the credential afterwards and record that you did, because a tested credential is a used credential.
Verify
grep -Eci 'outside' lab/breakglass.md
grep -Eci 'alert' lab/breakglass.md lab/breakglass-test.md
grep -Eci 'rotat' lab/breakglass-test.md
grep -Ec '[0-9]+ ?(s|sec|min)' lab/breakglass-test.md
All four non-zero. The rotation record is the step people skip: after a test, the credential has been used and seen, and leaving it in place makes the test itself the weakness.
This is an independent study companion for CompTIA Cloud+ CV0-004 and is not produced by or endorsed by CompTIA.