SnowPro Advanced: Administrator free question 6 / 10 · Domain 3: Data and Object Management
Hourly task T_BIG runs a large aggregation on user-managed warehouse WH_TASK (Medium). Since data grew last week, every run is canceled at about 60 minutes. No timeout value has ever been set on T_BIG or the warehouse. Which diagnosis and remedy follow the documented troubleshooting?
- The run is canceled when the next scheduled time arrives, so widen the schedule to every two hours
- A single task run has a default 60-minute limit; raise that limit or use a larger warehouse
- The warehouse STATEMENT_TIMEOUT_IN_SECONDS defaults to 60 minutes, so raise the warehouse value
- Runs over 60 minutes count as failures, so remove SUSPEND_TASK_AFTER_NUM_FAILURES
Show the answer and explanation
Answer: B
Why this is correct
The troubleshooting page says there is a 60-minute default limit on a single run of a task, implemented as a safeguard against non-terminating tasks. When a task is canceled or exceeds its window, check TASK_HISTORY; the cause is often an undersized warehouse, so consider increasing the size to fit within the window or the one-hour limit. Alternatively, raise the limit with ALTER TASK ... SET USER_TASK_TIMEOUT_MS. If SHOW PARAMETERS returns no record for it, the task has the default 3600000 millisecond (60 minute) timeout.
Why the other options are wrong
"The warehouse STATEMENT_TIMEOUT_IN_SECONDS defaults to 60 minutes, so raise the warehouse value" … STATEMENT_TIMEOUT_IN_SECONDS defaults to 172800 seconds (2 days), not 60 minutes. The 60-minute limit is the default on a single task run, so raising a warehouse value leaves the task's own default limit in place.
"The run is canceled when the next scheduled time arrives, so widen the schedule to every two hours" … Only one instance of a scheduled task runs at a time, and if a run is still going at the next scheduled time, that scheduled time is skipped. The running instance is not canceled by the next schedule.
"Runs over 60 minutes count as failures, so remove SUSPEND_TASK_AFTER_NUM_FAILURES" … SUSPEND_TASK_AFTER_NUM_FAILURES decides when the task is suspended after runs fail or time out; it does not set or lift the time limit of a run. Removing it does not stop the cancellation at 60 minutes.
Where it goes wrong
If the cause is a query parallelization problem, a larger warehouse or a longer limit may not help. In that case, review how the SQL run by the task is written.