Companion code for Zero-Downtime Database Migrations: Expand-Contract in Practice with Spring Boot: a full expand/migrate-writes/migrate-reads/contract sequence run as an actual rolling deploy across two live replicas, with a load generator sending continuous HTTP traffic through all four deploys (99.98% success, every residual error traced to a root cause rather than left unexplained). Findings include a real NOT NULL constraint trap in the expand migration, a backfill-window bug in the read switch, H2's AUTO_SERVER=TRUE single-point-of-failure behavior under a rolling restart, the drain-before-SIGTERM fix needed to close a health-check gap during graceful shutdown, and H2 silently discarding a concurrently committed INSERT during an ALTER TABLE ADD/DROP COLUMN rebuild - confirmed, by primary source, to be an H2-specific behavior rather than a property of the technique itself. Co-Authored-By: Claude Sonnet 5 <[email protected]> Claude-Session: https://claude.ai/code/session_019Fb7vW8vLyLKngBc4R3huA
67 lines
3.0 KiB
Java
67 lines
3.0 KiB
Java
package com.ankurm.expandcontract;
|
|
|
|
import com.ankurm.expandcontract.customer.CustomerService;
|
|
import org.junit.jupiter.api.Test;
|
|
import org.junit.jupiter.api.io.TempDir;
|
|
import org.springframework.jdbc.core.simple.JdbcClient;
|
|
|
|
import java.nio.file.Path;
|
|
|
|
import static org.assertj.core.api.Assertions.assertThat;
|
|
|
|
/**
|
|
* A trap this article's own first draft of the V2 migration fell into: adding
|
|
* {@code email_address} is not the whole story of "expand" when the column being
|
|
* retired is {@code NOT NULL}. Stage 4 code never writes "email" - so if "email" is
|
|
* still mandatory when Stage 4 starts, every create fails, in production, the moment
|
|
* that deploy reaches its first replica. The fix is one more line in the SAME
|
|
* migration: {@code ALTER TABLE customers ALTER COLUMN email DROP NOT NULL}. See
|
|
* docs/06-the-not-null-trap.md, and compare
|
|
* {@link com.ankurm.expandcontract.customer.CustomerService#create} - the shipped
|
|
* V2 migration (with the fix) is what every other test in this module runs against.
|
|
*/
|
|
class NotNullConstraintTrapTest {
|
|
|
|
@Test
|
|
void stage4FailsIfTheOldColumnIsStillMandatory(@TempDir Path tmp) throws Exception {
|
|
Path db = tmp.resolve("not-null-trap");
|
|
TestSupport.migrateTo(db, "1");
|
|
JdbcClient jdbc = TestSupport.jdbcClient(db);
|
|
Transcript t = Transcript.start("06-not-null-trap",
|
|
"The NOT NULL trap: expand without relaxing the old column's constraint");
|
|
|
|
// The NAIVE version of V2: add the column, backfill, stop there. This is
|
|
// exactly V2 minus its last line.
|
|
jdbc.sql("ALTER TABLE customers ADD COLUMN email_address VARCHAR(320)").update();
|
|
jdbc.sql("UPDATE customers SET email_address = email WHERE email_address IS NULL").update();
|
|
|
|
CustomerService stage4 = new CustomerService(jdbc, 4);
|
|
Throwable thrown = null;
|
|
try {
|
|
stage4.create("Too Early", "[email protected]");
|
|
} catch (Throwable ex) {
|
|
thrown = ex;
|
|
}
|
|
t.section("Stage 4 create() against the NAIVE migration (no DROP NOT NULL)");
|
|
assertThat(thrown).isNotNull();
|
|
Throwable root = thrown;
|
|
while (root.getCause() != null) {
|
|
root = root.getCause();
|
|
}
|
|
t.line(thrown.getClass().getName() + ": " + thrown.getMessage());
|
|
t.line("root cause: " + root.getClass().getName() + ": " + root.getMessage());
|
|
assertThat(root.getMessage()).contains("NULL not allowed for column \"EMAIL\"");
|
|
|
|
t.section("Stage 4 create() against the SHIPPED V2 migration (DROP NOT NULL included)");
|
|
Path fixedDb = tmp.resolve("not-null-fixed");
|
|
TestSupport.migrateTo(fixedDb, "2");
|
|
CustomerService fixedStage4 = new CustomerService(TestSupport.jdbcClient(fixedDb), 4);
|
|
long id = fixedStage4.create("On Time", "[email protected]");
|
|
var result = fixedStage4.findById(id).orElseThrow();
|
|
t.line(result.toString());
|
|
assertThat(result.email()).isEqualTo("[email protected]");
|
|
|
|
t.write();
|
|
}
|
|
}
|