Measure how long your rollback actually takes
Task
Write a cutover plan with a rollback, then TIME the rollback. A rollback that takes longer than the remaining window is not available to you, and the only way to know is to have done it once.
Steps
- Run a small service in container A, with a name resolving to it. Confirm a client reaches it by name.
- Write
lab/cutover.md: the timeline, the point of no return marked explicitly, the go/no-go criteria, the named decision maker, and the rollback trigger agreed in advance. - Stand up container B with the same service and a copy of the data. Cut over by changing the name to point at B, and time how long the client takes to follow -- then repeat the measurement with a long TTL and with a short one, and record both.
- Write a record to B so the two now differ. Then execute the rollback and time it, recording in
lab/rollback.mdwhat happened to the record written at B. - Write the sentence stating where the real point of no return was, and compare it against the one you wrote in step 2.
Verify
grep -Eci 'ttl' lab/cutover.md lab/rollback.md
grep -Ec '[0-9]+ ?(s|sec|second|min)' lab/rollback.md
grep -Eci 'point of no return|first write' lab/rollback.md
A measured rollback duration must appear. The TTL comparison is the point of step 3: the long TTL slows the rollback, which is when speed actually matters.
This is an independent study companion for CompTIA Cloud+ CV0-004 and is not produced by or endorsed by CompTIA.