tasks/online-compute/FlinkSQL/flinksql_008_en/task.md

---
id: online-compute_FlinkSQL_flinksql_008
name: Session Window User Behavior Aggregation
category: online-compute/FlinkSQL
timeout_seconds: 600
modality: pure-text
engine: flink-sql
---
## Prompt
I need you to write a Flink SQL that uses a datagen built-in table to simulate a user behavior stream, performs session window aggregation to compute behavioral metrics per user within each session, and outputs the results to the console.

**Business Background and Objective**: Simulate a user behavior data stream using a datagen built-in table that generates 50 records per second, containing four fields: `userId` (user ID, string of length 6), `action` (action type, string of length 8), `action_time` (action occurrence time, timestamp), and `ftime` (event time, timestamp). Based on this datagen table, using `ftime` as the event time (with a 5-second watermark delay), perform a 30-minute session window (`SESSION`) aggregation grouped by `userId`. Compute the following metrics for each user within each session window: session start time, session end time, total action count, number of distinct action types, and the last action time (the maximum value of `action_time`). Output the results to a built-in console table.

**Source Table Definition**:
- `user_source` (user behavior stream, datagen connector):
  - `userId VARCHAR`: User ID, randomly generated, 6 characters in length
  - `action VARCHAR`: Action type, randomly generated, 8 characters in length
  - `action_time TIMESTAMP(3)`: Action occurrence time
  - `ftime`: Uses `LOCALTIMESTAMP` to generate event time, with a WATERMARK delay of 5 seconds
  - Generation rate: `rows-per-second = 50`

**Output Table Definition**:
- `console_output` (print connector):
  - `user_id VARCHAR`: User ID
  - `session_start TIMESTAMP`: Session start time
  - `session_end TIMESTAMP`: Session end time
  - `action_count BIGINT`: Total action count
  - `unique_actions BIGINT`: Number of distinct action types
  - `last_action_time TIMESTAMP`: Last action time

**Query Logic**:
- Window: `SESSION(ftime, INTERVAL '30' MINUTE)` session window
- Grouping: Group by `userId` and the session window
- Aggregation: `COUNT(*)` for total action count, `COUNT(DISTINCT action)` for distinct action type count, `MAX(action_time)` for last action time
- Window functions: Use `SESSION_START` / `SESSION_END` to obtain the window start and end times

**Output Requirements**:
- Use `INSERT INTO console_output` to output the results.
- Output field order: `user_id`, `session_start`, `session_end`, `action_count`, `unique_actions`, `last_action_time`