Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I want to mount the Cloudera Hadoop as a Linux file system over the LAN.

As a setup, I already have the hadoop cluster running on a set of Ubuntu machines. But now I need to be able to use it as a normal file system from a Fedora system over the LAN.
I tried FUSe but two things:
1. Cloudera says FUSE loses data (click here for that comment by a Cloudera employee on the official Cloudera support site)
2. I've had no success making it work the way we want

As a point of clarification, I am using Hadoop ONLY for the file system, not for its other capabilities.

share|improve this question
1  
If you're just using Haddoop for HDFS why not have a look at GlusterFS. It may be a better solution for your needs. –  3dinfluence Jan 2 '10 at 21:24
    
MapR is an even better solution. It provides complete Hadoop compatibility and very high performance. –  Ted Dunning Jul 15 '11 at 23:59

2 Answers 2

FUSE is really your only option for mounting exotic file systems such as HDFS. For my own needs, I've found that using the Java API directly was a much better option than a mounted FS.

Unfortunately HDFS client APIs for languages other than Java either don't exist or are very ugly (as in dependent on JNI).

This page has interesting info on the various options available to you. If you do somehow succeed in making FUSE work, you can export the resulting mount point using NFS to other machines that need to use the same HDFS. This obviously carries the same risks and disadvantages as any NFS set up.

With regards to APIs, there is apparently a Thrift interface, FWIW. See here for more info.

share|improve this answer

HDFS isn't really a first class file system. As such, mounting it via FUSE or anything similar is likely to lead to real frustration because of the lack of file update.

Can you say more about your needs? Why not just use MapR (see mapr.com)? MapR provides first class NFS support for the clustered file system while maintaining full compatibility with Hadoop.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.