Apache Zeppelin JDBC notebooks

Start with the shared setup and dataset. The versioned observations below use the MatrixOne build identified in the overview.

Configure the interpreter

Install Apache Zeppelin; the tested version was 0.12.0 with MySQL Connector/J 8.4.0.

  1. Install com.mysql:mysql-connector-j:8.4.0 in the JDBC interpreter’s dependency settings, or place that driver JAR in the interpreter’s JDBC classpath as described upstream. Restart the interpreter after changing dependencies.

  2. Configure these JDBC interpreter properties:

default.driver = com.mysql.cj.jdbc.Driver
default.url = jdbc:mysql://MO_HOST:6001/bi_tutorial?useSSL=false&connectionCollation=utf8mb4_general_ci&serverTimezone=UTC
default.user = MO_USER
default.password = REPLACE_WITH_PASSWORD
common.max_count = 100000
zeppelin.jdbc.maxRows = 100000

The URL reproduces the local non-TLS test. For remote use, choose Connector/J SSL options with CA/hostname verification. serverTimezone=UTC defines the driver setting; confirm the business reporting time zone as well.

  1. The retest increased interpreter result limits to 100,000 rows and the server’s zeppelin.interpreter.output.limit to 10,485,760 bytes. Set limits appropriate to your workload; these caps do not guarantee every result is complete. Restart affected services/interpreters and verify the effective settings.

  2. Create a notebook bound to the JDBC interpreter. Run a paragraph containing %jdbc followed by SELECT 1;.

Build saved analyses

Create and save this paragraph:

%jdbc
SELECT category, COUNT(*) AS paid_orders,
       COUNT(DISTINCT customer_id) AS buyers,
       SUM(net_amount) AS net_amount, SUM(quantity) AS units
FROM bi_tutorial.order_metrics
GROUP BY category ORDER BY category;

Verify A=130.000000, B=280.000000. Add a paragraph for the shared daily/running-total SQL, save the notebook, reopen it and execute the paragraphs again. This makes the SQL and its result checks inspectable together. Notebook table or chart controls can be used after validating the raw result; browser chart rendering was not part of the compatibility run.

Result formatting boundary

Actual saved notebook paragraphs and JDBC interpreter execution were verified. An edge case with a newline inside a cell broke the raw tab/newline-delimited TABLE message even though the JDBC values were correct. Use an appropriate escaped/export representation when values contain tabs or newlines; do not report that formatting discrepancy as incorrect MatrixOne query data.

Check both row limits and the byte output limit before treating a notebook result as the complete dataset. JDBC URL options alone do not address notebook truncation or formatting.