Most Hadoop functionality of HDP 2.5 Sandbox can be used through Toad. All checks are green, just the real access to HDFS data is broken.
SYSTEM ENVIRONMENT: HDP Sandbox 2.5 (Docker) on Windows for Docker
STEPS TO REPRODUCE:
- Open HDFS View
- Select any file and double click to open
- Wait (file content should be shown)
Instead of showing file content, the system waits and waits and the operation stops without results
BACKGROUND
- The sandbox can be reached as sandbox.hortonworks.com, which resolves as 127.0.0.1 (localhost)
- All relevant ports (like 10000 for hive or 8020 for the nameserver) are portforwared into the virtual linux machine and then into the docker container, in which hadoop is running
- After accessing sandbox.hortonworks.com:8020 for the nameserver, Toad for Hadoop tries to connect to the data node to fetch file data
- Instead of accessing sandbox.hortonworks.com:50010, Toad tries to reach the datanode via IP address (172.17.0.2:50010).
- As the internal docker IP address is not know on the host, the connection attempt times out, which leads to the issue
ALTERNATIVE SOLUTIONS
- Enhance Toad for Hadoop to support WebHDFS / HTTPFS
Is there any way how to circumvent the issue? Is it possible to change the HDFS behavior on either TOAD or HADOOP side?
Jan