The binary is s3 and the product is Hanzo S3; the rest of the tree now says so too. - Environment: WEED_* and SEAWEEDFS_* are S3_*, no aliases. fla9's flag prefix is S3, matching viper's existing "s3" prefix, so the S3_S3_SSE_* names mini.go exports are the ones viper reads (WEED_S3_SSE_KEK was never seen once viper moved to "s3"). - Java and Hadoop: packages, fs.hanzofs.* keys and the hanzofs:// scheme under the hanzofs namespace the protos already declare; classes take a neutral Filer prefix and file names match them again (the earlier pass left Seas3*.java holding Seaweed* classes). Client, hdfs3, examples, s3copier, random_access and the Spark tests compile. - Helm chart: helpers are hanzo.*, values live under global.s3; the chart renders and lints, which it did not before. - Terraform: the s3.service.tftpl reference matches the file; 12/12 render tests pass. Config paths point at /etc/hanzo, which the binary reads. - loadtest.pb.go regenerated from loadtest.proto. - Upstream logo, sponsor images and unreferenced diagrams removed from note/. Kept: NOTICE attribution, LICENSE, the erasure-coding fixture s3/storage/erasure_coding/1.dat (test bytes), and unmaintained/ and third_party/ (vendored). No on-disk format, wire protocol or metadata key changed. Co-authored-by: Hanzo Dev <[email protected]>
4.4 KiB
Hanzo Hadoop3 Client
Hadoop FileSystem implementation for Hanzo, compatible with Hadoop 3.x.
Building
mvn clean install
Testing
This project includes two types of tests:
1. Configuration Tests (No Hanzo Required)
These tests verify configuration handling and initialization logic without requiring a running Hanzo instance:
mvn test -Dtest=HanzoFileSystemConfigTest
2. Integration Tests (Requires Hanzo)
These tests verify actual FileSystem operations against a running Hanzo instance.
Prerequisites
-
Start Hanzo with default ports:
# Terminal 1: Start master s3 master # Terminal 2: Start volume server s3 volume -master=localhost:9333 # Terminal 3: Start filer s3 filer -master=localhost:9333 -
Verify services are running:
- Master: http://localhost:9333
- Filer HTTP: http://localhost:8888
- Filer gRPC: localhost:18888
Running Integration Tests
# Enable integration tests
export S3_TEST_ENABLED=true
# Run all tests
mvn test
# Run specific test
mvn test -Dtest=HanzoFileSystemTest
Test Configuration
Integration tests can be configured via environment variables or system properties:
S3_TEST_ENABLED: Set totrueto enable integration tests (default: false)- Tests use these default connection settings:
- Filer Host: localhost
- Filer HTTP Port: 8888
- Filer gRPC Port: 18888
Running Tests with Custom Configuration
To test against a different Hanzo instance, modify the test code or use Hadoop configuration:
conf.set("fs.hanzo.filer.host", "your-host");
conf.setInt("fs.hanzo.filer.port", 8888);
conf.setInt("fs.hanzo.filer.port.grpc", 18888);
Test Coverage
The test suite covers:
-
Configuration & Initialization
- URI parsing and configuration
- Default values
- Configuration overrides
- Working directory management
-
File Operations
- Create files
- Read files
- Write files
- Append to files
- Delete files
-
Directory Operations
- Create directories
- List directory contents
- Delete directories (recursive and non-recursive)
-
Metadata Operations
- Get file status
- Set permissions
- Set owner/group
- Rename files and directories
Usage in Hadoop
-
Copy the built JAR to your Hadoop classpath:
cp target/hanzo-hadoop3-client-*.jar $HADOOP_HOME/share/hadoop/common/lib/ -
Configure
core-site.xml:<configuration> <property> <name>fs.s3.impl</name> <value>hanzo.hdfs.HanzoFileSystem</value> </property> <property> <name>fs.hanzo.filer.host</name> <value>localhost</value> </property> <property> <name>fs.hanzo.filer.port</name> <value>8888</value> </property> <property> <name>fs.hanzo.filer.port.grpc</name> <value>18888</value> </property> <!-- Optional: Replication configuration with three priority levels: 1) If set to non-empty value (e.g. "001") - uses that value 2) If set to empty string "" - uses Hanzo filer's default replication 3) If not configured (property not present) - uses HDFS replication parameter --> <!-- <property> <name>fs.hanzo.replication</name> <value>001</value> </property> --> </configuration> -
Use Hanzo with Hadoop commands:
hadoop fs -ls hanzo://localhost:8888/ hadoop fs -mkdir hanzo://localhost:8888/test hadoop fs -put local.txt hanzo://localhost:8888/test/
Continuous Integration
For CI environments, tests can be run in two modes:
-
Configuration Tests Only (default, no Hanzo required):
mvn test -Dtest=HanzoFileSystemConfigTest -
Full Integration Tests (requires Hanzo):
# Start Hanzo in CI environment # Then run: export S3_TEST_ENABLED=true mvn test
Troubleshooting
Tests are skipped
If you see "Skipping test - S3_TEST_ENABLED not set":
export S3_TEST_ENABLED=true
Connection refused errors
Ensure Hanzo is running and accessible:
curl http://localhost:8888/
gRPC errors
Verify the gRPC port is accessible:
# Should show the port is listening
netstat -an | grep 18888
Contributing
When adding new features, please include:
- Configuration tests (no Hanzo required)
- Integration tests (with S3_TEST_ENABLED guard)
- Documentation updates