Handling load errors in Snowflake - Commands & Configuration

Introduction

When loading data into Snowflake, sometimes errors happen because of bad data or format issues. Handling load errors means finding and fixing these problems so your data loads correctly.

When you load CSV files and some rows have wrong data types.

When you want to skip bad records but still load the good ones.

When you want to see which rows failed and why.

When you want to automatically log load errors for later review.

When you want to prevent your entire load from failing due to a few bad rows.

Commands

Create a simple table to load data into. This table has three columns: id, name, and age.

Terminal

CREATE OR REPLACE TABLE my_table (id INT, name STRING, age INT);

Expected Output

Table MY_TABLE successfully created.

Upload the local CSV file data.csv to the Snowflake internal stage for the my_table table.

Terminal

PUT file:///tmp/data.csv @%my_table;

Expected Output

100% |################################| 1 MB | 1 MB/s | 00:01 | Upload successful.

Load data from the staged CSV file into my_table. The ON_ERROR='CONTINUE' option tells Snowflake to skip bad rows and continue loading the rest.

Terminal

COPY INTO my_table FROM @%my_table/data.csv FILE_FORMAT = (TYPE = 'CSV' FIELD_OPTIONALLY_ENCLOSED_BY='"') ON_ERROR = 'CONTINUE';

Expected Output

COPY INTO my_table executed successfully.

Rows loaded: 95

Rows skipped: 5

Errors logged in table LOAD_ERRORS.

→

ON_ERROR='CONTINUE' - Skip bad rows and continue loading good rows.

Check which rows failed during the last load job by using the VALIDATE function with the job ID from the load history.

Terminal

SELECT * FROM TABLE(VALIDATE(TABLE_NAME => 'MY_TABLE', JOB_ID => (SELECT MAX(LAST_LOAD_JOB_ID) FROM INFORMATION_SCHEMA.LOAD_HISTORY WHERE TABLE_NAME = 'MY_TABLE')));

Expected Output

ROW_NUMBER | ERROR_CODE | ERROR_MESSAGE

3          | 1001       | Invalid integer value in column 'age'

7          | 1002       | Missing required field 'id'

Key Concept

If you remember nothing else from this pattern, remember: use ON_ERROR options to control how Snowflake handles bad data during loads and use VALIDATE to find load errors.

Common Mistakes

Not using ON_ERROR option and expecting the load to skip bad rows automatically.

Without ON_ERROR='CONTINUE' or similar, Snowflake stops loading at the first error.

Always specify ON_ERROR='CONTINUE' or ON_ERROR='SKIP_FILE' to handle errors gracefully.

Not checking load errors after COPY INTO command.

You might miss which rows failed and why, leading to incomplete or wrong data.

Use VALIDATE function or check load history tables to review errors.

Summary

Create a target table before loading data.

Upload data files to Snowflake internal stage using PUT.

Use COPY INTO with ON_ERROR='CONTINUE' to skip bad rows during load.

Use VALIDATE function to find and review load errors.

Practice

1. What does the ON_ERROR option do when loading data into Snowflake?

easy

A. It controls how Snowflake handles errors during data loading.

B. It encrypts the data before loading.

C. It speeds up the data loading process.

D. It automatically deletes duplicate rows.

Solution

1. Step 1: Understand the purpose of ON_ERROR

   The ON_ERROR option is used to specify how Snowflake should behave when it encounters errors during data loading.

2. Step 2: Identify the correct behavior

   It can skip bad rows, skip files, or stop the load depending on the setting, thus controlling error handling.

3. Final Answer:

   It controls how Snowflake handles errors during data loading. -> Option A

4. Quick Check:

   ON_ERROR controls error handling [OK]

Hint: ON_ERROR sets error handling behavior during load [OK]

Common Mistakes:

• Confusing ON_ERROR with encryption settings
• Thinking ON_ERROR speeds up loading
• Assuming ON_ERROR deletes duplicates

2. Which of the following is the correct syntax to skip bad rows during a Snowflake COPY INTO command?

easy

A. COPY INTO table_name FROM @stage FILE_FORMAT = (TYPE = 'CSV') ON_ERROR = 'ignore_error';

B. COPY INTO table_name FROM @stage FILE_FORMAT = (TYPE = 'CSV') ON_ERROR = 'skip_file';

C. COPY INTO table_name FROM @stage FILE_FORMAT = (TYPE = 'CSV') ON_ERROR = 'abort_load';

D. COPY INTO table_name FROM @stage FILE_FORMAT = (TYPE = 'CSV') ON_ERROR = 'skip_row';

Solution

1. Step 1: Recall valid ON_ERROR options

   Snowflake supports options like 'skip_file', 'skip_row', and 'abort_load' for ON_ERROR.

2. Step 2: Identify option to skip bad rows

   'skip_row' tells Snowflake to skip only the bad rows, not the entire file.

3. Final Answer:

   ON_ERROR = 'skip_row' -> Option D

4. Quick Check:

   Skip bad rows = skip_row [OK]

Hint: Use 'skip_row' to skip bad rows in ON_ERROR [OK]

Common Mistakes:

• Using 'skip_file' to skip rows instead of files
• Using invalid ON_ERROR values like 'ignore_error'
• Confusing 'abort_load' with skipping errors

3. Given this COPY command:

COPY INTO my_table FROM @my_stage FILE_FORMAT = (TYPE = 'CSV') ON_ERROR = 'skip_file';

If one file has 5 bad rows, what happens?

medium

A. The entire file with bad rows is skipped, other files load normally.

B. Only the 5 bad rows are skipped, rest of the file loads.

C. The load stops immediately with an error.

D. All files are skipped regardless of errors.

Solution

1. Step 1: Understand ON_ERROR = 'skip_file'

   This option skips the entire file if any error occurs in it.

2. Step 2: Apply to scenario

   Since one file has 5 bad rows, Snowflake skips that whole file but continues loading other files.

3. Final Answer:

   The entire file with bad rows is skipped, other files load normally. -> Option A

4. Quick Check:

   skip_file skips whole file on error [OK]

Hint: 'skip_file' skips whole file if any error found [OK]

Common Mistakes:

• Thinking only bad rows are skipped with 'skip_file'
• Assuming load stops on first error
• Believing all files skip on one bad file

4. You run a COPY INTO command with ON_ERROR = 'skip_row' but still see the load failing. What is a likely cause?

medium

A. ON_ERROR only works for JSON files, not CSV.

B. The file format is incorrect causing parsing errors.

C. You must set ON_ERROR to 'skip_file' to avoid failures.

D. The target table does not exist.

Solution

1. Step 1: Understand ON_ERROR limitations

   ON_ERROR skips bad rows but cannot fix fundamental file format or parsing errors.

2. Step 2: Identify cause of failure

   If file format is wrong, Snowflake cannot parse data, causing load failure despite ON_ERROR.

3. Final Answer:

   The file format is incorrect causing parsing errors. -> Option B

4. Quick Check:

   Wrong file format causes failure despite ON_ERROR [OK]

Hint: Check file format if ON_ERROR skip_row still fails [OK]

Common Mistakes:

• Thinking ON_ERROR fixes all errors
• Believing ON_ERROR only works for JSON
• Ignoring table existence errors

5. You want to load multiple CSV files but skip any file with more than 10 bad rows, while loading others fully. Which ON_ERROR setting should you use?

hard

A. ON_ERROR = 'continue' with MAX_ERROR = 10

B. ON_ERROR = 'skip_row' with MAX_ERROR = 10

C. ON_ERROR = 'skip_file' with MAX_ERROR = 10

D. ON_ERROR = 'abort_load' with MAX_ERROR = 10

Solution

1. Step 1: Understand requirement

   Skip entire files only if bad rows exceed 10, otherwise load fully.

2. Step 2: Match ON_ERROR and MAX_ERROR

   Using ON_ERROR = 'skip_file' with MAX_ERROR = 10 skips files exceeding 10 errors, loads others fully.

3. Final Answer:

   ON_ERROR = 'skip_file' with MAX_ERROR = 10 -> Option C

4. Quick Check:

   Skip files over 10 errors = skip_file + MAX_ERROR [OK]

Hint: Use skip_file with MAX_ERROR to limit bad rows per file [OK]

Common Mistakes:

• Using skip_row which skips rows, not files
• Assuming 'continue' skips files
• Thinking abort_load allows skipping