問題文
A theater chain discovered that a flow which creates ticket refunds had been failing for 11 days. Nobody looked because nobody knew. The operations lead wants this class of problem found sooner. Where should the effort go?
選択肢
- Into the automation itself: add fault paths that send an email, and additionally rely on the fact that a flow with a fault path continues along that path instead of rolling back, and no refund is ever lost while the alert is being investigated.
- Into a scheduled flow that runs every hour and re-creates any refund that is missing, so the failure repairs itself and no monitoring is needed because a discrepancy never lasts an hour.
- Into debug logs: raise the log level for the users who run the flow so that the next failure is captured in enough detail for the whole team to read afterwards.
- Into the automation itself: add fault paths that send an email including the values of the flow's resources, and direct those error emails to the addresses that actually watch them rather than the last person who edited the flow.