@clickzetta/cz-cli-darwin-x64 0.3.92 → 0.3.94

package/bin/skills/clickzetta-dynamic-table/dt-creator/references/refresh-history-guide.md

@@ -1,260 +0,0 @@
-# Dynamic Table Incremental Refresh History Query Guide
-
-There are three ways to view the incremental refresh history of a DT/MV, each suited to different scenarios.
-
----
-
-## Method 1: SHOW DYNAMIC TABLE REFRESH HISTORY
-
-View job-level information for DT refreshes, including the status, duration, trigger type, and refresh mode of each refresh.
-
-### Syntax
-
-```sql
--- Filter by WHERE (name column matches table name)
-SHOW DYNAMIC TABLE REFRESH HISTORY WHERE name = 'my_dt';
-
--- Combine WHERE + LIMIT
-SHOW DYNAMIC TABLE REFRESH HISTORY WHERE name = 'my_dt' AND state = 'SUCCEED' LIMIT 20;
-
--- MV supports the same syntax
-SHOW MATERIALIZED VIEW REFRESH HISTORY WHERE name = 'my_mv' LIMIT 10;
-```
-
-### Output Columns
-
-| Column | Type | Description |
-|------|------|------|
-| workspace_name | STRING | Workspace the DT/MV belongs to |
-| schema_name | STRING | Schema the DT/MV belongs to |
-| name | STRING | DT/MV name |
-| virtual_cluster | STRING | Virtual cluster that executed the refresh |
-| start_time | TIMESTAMP | Refresh start time |
-| end_time | TIMESTAMP | Refresh end time (NULL while running) |
-| duration | INTERVAL | Refresh duration (shows elapsed time while running) |
-| state | STRING | Refresh status (SUCCEED / FAILED / RUNNING, etc.) |
-| refresh_trigger | STRING | Trigger type: `SYSTEM_SCHEDULED` (auto-triggered by system) or `MANUAL` (user manually triggered REFRESH) |
-| refresh_mode | STRING | Refresh mode; see detailed description below |
-| error_message | STRING | Error message on failure (NULL on success) |
-| source_tables | ARRAY<MAP<STRING,STRING>> | Source table list; each element is a MAP with keys `workspace`, `schema`, `table_name` |
-| stats | MAP<STRING,STRING> | Refresh statistics, including `rows_inserted` and `rows_deleted` |
-| job_id | STRING | Corresponding Job ID; can be used to join `information_schema.job_history` for more details |
-
-### refresh_mode Details
-
-`refresh_mode` is the key field for determining whether incremental computation is working:
-
-| Value | Meaning | Description |
-|----|------|------|
-| `INCREMENTAL` | Incremental refresh | The incremental engine successfully generated an incremental plan and only processed change data from source tables |
-| `FULL` | Full refresh | Fell back to full recomputation. Possible reasons: first refresh, dimension table change, incremental plan generation failed, user forced full, etc. |
-| `NO_DATA` | No data changes | No new data changes in source tables since the last refresh; this refresh skipped computation |
-
-### source_tables Details
-
-The `source_tables` column returns information about all input tables involved in this refresh; each element is a MAP:
-
-```
-[
-  {"workspace": "my_ws", "schema": "public", "table_name": "orders"},
-  {"workspace": "my_ws", "schema": "public", "table_name": "dim_product"}
-]
-```
-
-### stats Details
-
-The `stats` column returns write statistics for the target table from this refresh:
-
-```
-{"rows_inserted": "1000", "rows_deleted": "50"}
-```
-
-- `rows_inserted`: number of rows inserted into the target table in this refresh
-- `rows_deleted`: number of rows deleted from the target table in this refresh (in incremental mode, update operations produce delete + insert)
-
-### Typical Usage
-
-```sql
--- View failed refresh records
-SHOW DYNAMIC TABLE REFRESH HISTORY WHERE name = 'my_dt' AND state = 'FAILED';
-
--- Check if it fell back to full refresh (troubleshoot whether incremental is working)
-SHOW DYNAMIC TABLE REFRESH HISTORY WHERE name = 'my_dt' AND refresh_mode = 'FULL';
-
--- View no-data-change refreshes (appears when source table has no new data)
-SHOW DYNAMIC TABLE REFRESH HISTORY WHERE name = 'my_dt' AND refresh_mode = 'NO_DATA';
-
--- View system auto-scheduled refreshes
-SHOW DYNAMIC TABLE REFRESH HISTORY WHERE name = 'my_dt' AND refresh_trigger = 'SYSTEM_SCHEDULED';
-```
-
----
-
-## Method 2: DESC HISTORY
-
-View version-level history of a table, including row count, bytes, and operation type for each version. Useful for understanding data change granularity.
-
-### Syntax
-
-```sql
--- View version history of a DT
-DESC HISTORY my_dt;
-
--- View version history of a source table
-DESC HISTORY source_table;
-
--- Supports WHERE filtering
-DESC HISTORY my_dt WHERE version > 10;
-
--- Supports LIMIT
-DESC HISTORY my_dt LIMIT 20;
-```
-
-### Output Columns
-
-For regular tables (DESC_TABLE_HISTORY):
-
-| Column | Type | Description |
-|------|------|------|
-| sequence | BIGINT | Sequence number |
-| version | BIGINT | Version number |
-| time | TIMESTAMP | Version creation time |
-| total_rows | BIGINT | Total row count for this version |
-| total_bytes | BIGINT | Total bytes for this version |
-| user | STRING | User who performed the operation |
-| operation | STRING | Operation type (INSERT / COMPACTION / REFRESH, etc.) |
-| job_id | STRING | Corresponding Job ID |
-
-For DT/MV (DESC_MV_HISTORY), additionally includes:
-
-| Column | Type | Description |
-|------|------|------|
-| source_tables | ARRAY<MAP<STRING,STRING>> | Source tables and their corresponding version information |
-
-DESC HISTORY's `source_tables` for DT/MV is more detailed than SHOW REFRESH HISTORY, including snapshot information for each source table at that version:
-
-```
-[
-  {"table_name": "orders", "workspace": "my_ws", "schema": "public", "version": "123", "sequence": "5", "commit_time": "2025-01-15 10:30:00"},
-  {"table_name": "dim_product", "workspace": "my_ws", "schema": "public", "version": "456", "sequence": "2", "commit_time": "2025-01-15 08:00:00"}
-]
-```
-
-- `version`: snapshot_id of the source table
-- `sequence`: sequence number of the source table
-- `commit_time`: commit time of that version of the source table
-
-This information can be used to trace which version of source table data was read in a given refresh.
-
-### Typical Usage
-
-```sql
--- View recent version changes of a DT; confirm compaction is executing normally
-DESC HISTORY my_dt LIMIT 10;
-
--- View version history of a source table; determine data write frequency
-DESC HISTORY source_table LIMIT 20;
-
--- View compaction records of a DT
-DESC HISTORY my_dt WHERE operation = 'COMPACTION';
-```
-
----
-
-## Method 3: information_schema.materialized_view_refresh_history
-
-Query refresh history from information_schema; suitable for cross-table batch analysis, integration with other systems, or long-term trend monitoring. Data is partitioned by day (pt_date); retention days are determined by system configuration.
-
-### Syntax
-
-```sql
--- View refresh history for a specific DT
-SELECT *
-FROM information_schema.materialized_view_refresh_history
-WHERE materialized_view_name = 'my_dt'
-ORDER BY start_time DESC
-LIMIT 10;
-
--- View refresh status of all DTs on a given day
-SELECT materialized_view_name, status, start_time, end_time, error_message
-FROM information_schema.materialized_view_refresh_history
-WHERE pt_date = '2025-01-15'
-ORDER BY start_time DESC;
-
--- View failed refreshes
-SELECT materialized_view_name, error_code, error_message, start_time
-FROM information_schema.materialized_view_refresh_history
-WHERE status = 'FAILED' AND pt_date >= '2025-01-01'
-ORDER BY start_time DESC;
-```
-
-### Output Columns
-
-| Column | Type | Description |
-|------|------|------|
-| workspace_name | STRING | Workspace the DT/MV belongs to |
-| schema_name | STRING | Schema the DT/MV belongs to |
-| materialized_view_name | STRING | DT/MV name |
-| cru | DOUBLE | Compute resource units consumed |
-| virtual_cluster_name | STRING | Virtual cluster that executed the refresh |
-| status | STRING | Refresh status |
-| scheduled_start_time | TIMESTAMP | Scheduled start time |
-| start_time | TIMESTAMP | Actual start time |
-| end_time | TIMESTAMP | End time |
-| error_code | STRING | Error code |
-| error_message | STRING | Error message |
-| pt_date | STRING | Partition date |
-
-### Typical Usage
-
-```sql
--- Calculate the refresh success rate of a DT over the last 7 days
-SELECT
-    pt_date,
-    COUNT(*) AS total,
-    SUM(CASE WHEN status = 'SUCCEED' THEN 1 ELSE 0 END) AS success,
-    SUM(CASE WHEN status = 'FAILED' THEN 1 ELSE 0 END) AS failed
-FROM information_schema.materialized_view_refresh_history
-WHERE materialized_view_name = 'my_dt'
-  AND pt_date >= DATE_FORMAT(DATEADD(DAY, -7, CURRENT_DATE()), '%Y-%m-%d')
-GROUP BY pt_date
-ORDER BY pt_date;
-
--- View refreshes that consumed the most CRU
-SELECT materialized_view_name, cru, start_time, end_time
-FROM information_schema.materialized_view_refresh_history
-WHERE pt_date >= '2025-01-01'
-ORDER BY cru DESC
-LIMIT 10;
-```
-
-### Difference from information_schema.job_history
-
-`information_schema.job_history` records all types of Jobs (SQL queries, DML, DDL, etc.), while `materialized_view_refresh_history` specifically records DT/MV refresh history with more targeted fields.
-
-If you need to view complete information about a refresh Job (e.g., job_text, input_bytes, etc.), you can join via job_id:
-
-```sql
--- Get job_id from SHOW DYNAMIC TABLE REFRESH HISTORY, then look up details in job_history
-SELECT *
-FROM information_schema.job_history
-WHERE job_id = '<job_id from SHOW REFRESH HISTORY>'
-  AND pt_date = '2025-01-15';
-```
-
----
-
-## Comparison of Three Methods
-
-| Feature | SHOW REFRESH HISTORY | DESC HISTORY | information_schema |
-|------|---------------------|--------------|-------------------|
-| Granularity | Refresh job level | Table version level | Refresh job level |
-| Refresh mode (incremental/full/no data) | ✅ refresh_mode | ❌ | ❌ |
-| Trigger type (scheduled/manual) | ✅ refresh_trigger | ❌ | ❌ |
-| Write statistics (inserted/deleted) | ✅ stats | ❌ | ❌ |
-| Source table list | ✅ table name level | ✅ includes version/sequence/commit_time | ❌ |
-| Version number / total rows / total bytes | ❌ | ✅ version/total_rows/total_bytes | ❌ |
-| CRU consumption | ❌ | ❌ | ✅ cru |
-| Cross-table batch queries | ❌ (single table) | ❌ (single table) | ✅ (batch supported) |
-| Compaction records | ❌ | ✅ | ❌ |
-| Applicable scenarios | Troubleshoot whether incremental is working; refresh status | View data version changes; trace source table versions | Batch analysis / monitoring / CRU statistics |

package/bin/skills/clickzetta-dynamic-table/dynamic-table-alter/SKILL.md

@@ -1,191 +0,0 @@
----
-name: dynamic-table-alter
-description: |
-  Modify the structure and properties of ClickZetta Dynamic Tables. Supports direct ALTER operations
-  (suspend, resume, rename_column, set_comment, set_column_comment, set/unset properties) as well as
-  CREATE OR REPLACE rebuild operations (modify refresh interval, compute cluster, add column, drop column,
-  change column type, change SQL definition). Triggers when the user says "modify dynamic table",
-  "add column to dynamic table", "change refresh interval", or "suspend dynamic table".
----
-
-# Dynamic Table Modification Workflow
-
-## Instructions
-
-### Step 1: Confirm the Dynamic Table Exists and Retrieve Its Current Definition
-Execute `SHOW CREATE TABLE schema_name.table_name` to get the current definition of the Dynamic Table.
-If unsure whether it is a Dynamic Table, first use `SHOW TABLES WHERE is_dynamic` to view the list.
-
-### Step 2: Determine the Operation Type and Choose the Execution Method
-
-Dynamic Table modification operations fall into two categories:
-
-**A. Direct ALTER operations** (6 types, can be executed directly):
-
-1. **suspend** — Pause the scheduled refresh:
-```sql
-ALTER DYNAMIC TABLE dt_name SUSPEND;
-```
-
-2. **resume** — Start the scheduled refresh:
-```sql
-ALTER DYNAMIC TABLE dt_name RESUME;
-```
-
-3. **set_comment** — Modify the table comment:
-```sql
-ALTER DYNAMIC TABLE dt_name SET COMMENT 'comment';
-```
-
-4. **rename_column** — Rename a column:
-```sql
-ALTER DYNAMIC TABLE dt_name RENAME COLUMN old_name TO new_name;
-```
-
-5. **set_column_comment** — Modify a column comment (note: use CHANGE COLUMN):
-```sql
-ALTER DYNAMIC TABLE dt_name CHANGE COLUMN column_name COMMENT 'comment';
-```
-
-6. **set/unset properties** — Modify table properties (currently reserved parameters):
-```sql
--- Set a property
-ALTER DYNAMIC TABLE dt_name SET PROPERTIES('key' = 'value');
--- Remove a property
-ALTER DYNAMIC TABLE dt_name UNSET PROPERTIES('key');
-```
-
-**B. CREATE OR REPLACE operations** (6 types, require rebuilding the Dynamic Table):
-
-> ⚠️ **The following operations do not support ALTER syntax.** Syntax like `ALTER DYNAMIC TABLE ... SET REFRESH INTERVAL` does not exist and will cause a syntax error. You must use `CREATE OR REPLACE DYNAMIC TABLE` to rebuild.
-
-These operations involve changes to SQL query logic and cannot be completed via ALTER directly:
-
-7. **Modify refresh interval** — ❌ `ALTER ... SET REFRESH INTERVAL` is not supported
-8. **Modify compute cluster** — ❌ `ALTER ... SET VCLUSTER` is not supported
-9. **Add column**
-10. **Drop column**
-11. **Modify column type**
-12. **Modify SQL definition**
-
-### Step 3: Execute CREATE OR REPLACE Rebuild (Type B operations only)
-
-1. Execute `SHOW CREATE TABLE schema_name.table_name` to get the original DDL
-   > ⚠️ `SHOW CREATE TABLE` does not support LIMIT/WHERE clauses; execute it directly
-2. Parse out: column definitions, REFRESH clause, AS SELECT clause, COMMENT, etc.
-3. Modify the relevant parts according to the operation
-4. Execute the rebuild SQL
-
-**About full refresh triggers:**
-- Simple drop column / add column (where the added column is simply passed through from the source table without participating in JOIN keys, GROUP keys, or other computations) → **incremental refresh**
-- Changes involving computation logic (modifying WHERE conditions, modifying aggregation logic, new column participates in computation, etc.) → **full refresh**
-- Compatible type changes (e.g., INT → BIGINT) → **incremental refresh**
-
-### Step 4: Verify the Modification
-Use `DESC TABLE dt_name` to confirm the modification took effect.
-
----
-
-## Examples
-
-### Example 1: Modify Refresh Interval
-
-```sql
--- Original table
-CREATE DYNAMIC TABLE dt_name
-REFRESH INTERVAL 10 MINUTE vcluster DEFAULT
-AS SELECT * FROM student02;
-
--- After modification (changed to 20 minutes)
-CREATE OR REPLACE DYNAMIC TABLE dt_name
-REFRESH INTERVAL 20 MINUTE vcluster DEFAULT
-AS SELECT * FROM student02;
-```
-
-### Example 2: Modify Compute Cluster
-
-```sql
--- Original table
-CREATE DYNAMIC TABLE dt_name
-REFRESH INTERVAL 10 MINUTE vcluster DEFAULT
-AS SELECT * FROM student02;
-
--- After modification (changed to alter_vc cluster)
-CREATE OR REPLACE DYNAMIC TABLE dt_name
-REFRESH INTERVAL 10 MINUTE vcluster alter_vc
-AS SELECT * FROM student02;
-```
-
-### Example 3: Add Column
-
-```sql
--- Original table
-CREATE DYNAMIC TABLE change_table (i, j)
-AS SELECT * FROM dy_base_a;
-
--- Add column col (involves computation logic; next refresh will be a full refresh)
-CREATE OR REPLACE DYNAMIC TABLE change_table (i, j, col)
-AS SELECT i, j, j * 1 FROM dy_base_a;
-
-REFRESH DYNAMIC TABLE change_table;
-```
-
-### Example 4: Drop Column
-
-```sql
--- Original table has columns i, j
-CREATE DYNAMIC TABLE change_table (i, j)
-AS SELECT * FROM dy_base_a;
-
--- Drop column (simple pass-through; incremental refresh)
-CREATE OR REPLACE DYNAMIC TABLE change_table (i)
-AS SELECT i FROM dy_base_a;
-```
-
-### Example 5: Modify SQL Definition
-
-```sql
--- Modify WHERE filter condition (full refresh)
-CREATE OR REPLACE DYNAMIC TABLE change_table (i, j)
-AS SELECT * FROM dy_base_a WHERE i > 3;
-
-REFRESH DYNAMIC TABLE change_table;
-```
-
-### Example 6: Modify Column Type
-
-```sql
--- INT → BIGINT (compatible type; incremental refresh)
-CREATE OR REPLACE DYNAMIC TABLE change_table (i, j)
-AS SELECT CAST(i AS BIGINT), j FROM dy_base_a;
-
-REFRESH DYNAMIC TABLE change_table;
-```
-
----
-
-## Platform-Specific Knowledge
-
-- **CHANGE COLUMN syntax**: set a column comment with `CHANGE COLUMN col COMMENT 'xxx'`, not `ALTER COLUMN`
-- **RENAME COLUMN syntax**: `RENAME COLUMN old TO new`
-- **DML restrictions**: Dynamic Tables do not support UPDATE/DELETE/MERGE by default (due to hidden column MV__KEY); to use DML, first execute `SET cz.sql.dt.allow.dml = true;`
-- **REFRESH format**: `REFRESH INTERVAL <N> MINUTE vcluster <name>`, supports SECOND/MINUTE/HOUR/DAY
-- **CREATE OR REPLACE risk**: changes involving computation logic will trigger a full refresh, which may take a long time for large tables
-- **Schema prefix**: all ALTER/CREATE statements should include the schema prefix in the table name
-- **Column definitions can omit types**: `CREATE DYNAMIC TABLE dt (i, j) AS SELECT ...` — types are inferred from SELECT
-- **DROP syntax**: must use `DROP DYNAMIC TABLE dt_name`; `DROP TABLE dt_name` will cause an error
-- **UNDROP syntax**: must use `UNDROP TABLE dt_name`; `UNDROP DYNAMIC TABLE dt_name` is not supported
-- **DESC syntax**: use `DESC TABLE dt_name` for Dynamic Tables; do not write `DESC DYNAMIC TABLE dt_name EXTENDED` (EXTENDED is not supported)
-
-## Troubleshooting
-
-| Error | Cause | Solution |
-|---|---|---|
-| ALTER reports "Syntax error at or near 'REFRESH'" | `ALTER ... SET REFRESH INTERVAL` syntax does not exist | Use `CREATE OR REPLACE DYNAMIC TABLE ... REFRESH INTERVAL ...` to rebuild |
-| ALTER reports "unsupported operation" | Attempted a Type B ALTER operation on a Dynamic Table | Use CREATE OR REPLACE to rebuild |
-| `DROP TABLE dt_name` fails | Dynamic Tables must use `DROP DYNAMIC TABLE` | Change to `DROP DYNAMIC TABLE dt_name` |
-| `UNDROP DYNAMIC TABLE` fails | UNDROP does not support the DYNAMIC TABLE keyword | Change to `UNDROP TABLE dt_name` |
-| `DESC DYNAMIC TABLE ... EXTENDED` fails | EXTENDED parameter is not supported | Change to `DESC TABLE dt_name` (without EXTENDED) |
-| UPDATE/DELETE reports "MV__KEY" related error | Dynamic Tables have a hidden column MV__KEY; DML is disabled by default | First execute `SET cz.sql.dt.allow.dml = true;` |
-| Data is empty after CREATE OR REPLACE | The AS SELECT clause references an incorrect source table or column | Verify that the SELECT clause returns data first |
-| Full refresh triggered after CREATE OR REPLACE | The new column participates in computation logic (JOIN key, GROUP key, etc.) | Expected behavior; wait for the full refresh to complete |