v3.2.1
Seed and cleanup diagnostics make the exact fixture namespace and test boundary visible when an asynchronous listener reads Mongo.
Highlights
- 01
Seed diagnostics at DEBUG
Each fixture application reports its phase, source, resolved database.collection and inserted-document count without logging document contents.
Read the guide → - 02
Kafka listener troubleshooting
The new guide explains co-resident Kafka harnesses, listener-thread Mongo clients, tenant-routing name mismatches and late-event cleanup boundaries.
Read the guide → - 03
Dotted MongoSeed targets
A collection value such as billing.invoices now targets the billing database, consistently with the rest of the harness.
Read the guide →
Full notes
Added
- MongoSeeder and SpringMongoSeeder now emit DEBUG seed diagnostics with the phase, fixture source, resolved database.collection and document count. The Spring harness also logs class-start order and each drop-databases cleanup/reapply boundary, without exposing fixture contents.
- The Bonsai skill now includes a Kafka listener and co-resident harness guide covering seed ordering, multi-tenant database-name diagnostics, consumer-thread clients, late events and correlation-id awaits.
Fixed
- MongoSeed now resolves a dotted collection value at its first dot when database is empty, so billing.invoices seeds invoices in billing. Supplying both database and a dotted collection fails fast rather than choosing a namespace silently.
- Embedded Spring tests now resolve their default database from spring.mongodb.database, then the database segment of spring.mongodb.uri, then test, matching Spring Boot. A suite that relied on URI-only configuration resolving to test should set spring.mongodb.database=test to retain that value.