Database / REDIS
Explain the execution flow of a Redis transaction using MULTI/EXEC?
Unlike a database transaction that begins actual execution immediately, a Redis MULTI block queues commands on the client's connection without running them, and only EXEC triggers the whole batch to actually execute, uninterrupted.
flowchart TD
A[Client sends MULTI] --> B[Redis enters queuing mode for this connection]
B --> C[Client sends SET, INCR, etc.]
C --> D{Command has a syntax error?}
D -- Yes --> E[Command rejected immediately, flagged for this transaction]
D -- No --> F[Command queued, reply QUEUED sent back]
F --> C
C --> G[Client sends EXEC]
G --> H{Any command was flagged with a syntax error earlier?}
H -- Yes --> I[Whole transaction aborted, nothing executes]
H -- No --> J[All queued commands execute back-to-back, no other client interleaved]
J --> K[Array of each command's individual result returned to the client]
The key nuance for interviews: a runtime error (like incrementing a key that holds a string) is different from a syntax error — runtime errors are only discovered during execution in step J, and don't stop the remaining queued commands from still running, since Redis doesn't inspect a command's effect on data until it actually executes it.
What does MULTI do to subsequent commands, before EXEC is sent?
Does a runtime error in one queued command stop the rest of the transaction from executing?
More Related questions...
