Skip to main content

​
MinIO HDFS Gateway Slack

MinIO HDFS gateway adds Amazon S3 API support to Hadoop HDFS filesystem. Applications can use both the S3 and file APIs concurrently without requiring any data migration. Since the gateway is stateless and shared-nothing, you may elastically provision as many MinIO instances as needed to distribute the load.
NOTE: Intention of this gateway implementation it to make it easy to migrate your existing data on HDFS clusters to MinIO clusters using standard tools like mc or aws-cli, if the goal is to use HDFS perpetually we recommend that HDFS should be used directly for all write operations.

​
Run MinIO Gateway for HDFS Storage

​
Using Binary

Namenode information is obtained by reading core-site.xml automatically from your hadoop environment variables $HADOOP_HOME
You can also override the namenode endpoint as shown below.

​
Using Docker

Using docker is experimental, most Hadoop environments are not dockerized and may require additional steps in getting this to work properly. You are better off just using the binary in this situation.

​
Setup Kerberos

MinIO supports two kerberos authentication methods, keytab and ccache. To enable kerberos authentication, you need to set hadoop.security.authentication=kerberos in the HDFS config file.
MinIO will load krb5.conf from environment variable KRB5_CONFIG or default location /etc/krb5.conf.
If you want MinIO to use ccache for authentication, set environment variable KRB5CCNAME to the credential cache file path, or MinIO will use the default location /tmp/krb5cc_%{uid}.
If you prefer to use keytab, with automatically renewal, you need to config three environment variables:
  • KRB5KEYTAB: the location of keytab file
  • KRB5USERNAME: the username
  • KRB5REALM: the realm
Please note that the username is not principal name.

​
Test using MinIO Browser

MinIO gateway comes with an embedded web based object browser. Point your web browser to http://127.0.0.1:9000 to ensure that your server has started successfully. Screenshot

​
Test using MinIO Client mc

mc provides a modern alternative to UNIX commands such as ls, cat, cp, mirror, diff etc. It supports filesystems and Amazon S3 compatible cloud storage services.

​
Configure mc

​
List buckets on hdfs

​
Known limitations

Gateway inherits the following limitations of HDFS storage layer:
  • No bucket policy support (HDFS has no such concept)
  • No bucket notification APIs are not supported (HDFS has no support for fsnotify)
  • No server side encryption support (Intentionally not implemented)
  • No server side compression support (Intentionally not implemented)
  • Concurrent multipart operations are not supported (HDFS lacks safe locking support, or poorly implemented)

​
Explore Further