aphp / py-hdfs-mount Goto Github PK

Python 3

sudo apt-get install fuse libfuse2

pip3 install -r requirements.txt

If you will be using kerberos, install libkrb5-dev:

sudo apt-get install libkrb5-dev

cp example.config.yaml config.yaml
$EDITOR config.yaml

If you are using kerberos, run a kinit:

kinit -kt $USER $USER@REALM

In all cases you then will have to create a new empty directory that with be the mount point:

mkdir /mnt/dest_mount

And finaly you can run py-hdfs-fuse:

python3 hdfs_mount.py [--loglevel LEVEL] config.yaml

Have fun!

Note: if anything goes wrong and you have to kill py-hdfs-mount, you will probably have to run this command on the mounted folder to unlock it:

fusermount -u /mnt/dest_mount
umount -l /mnt/dest_mount

• Vim (open file, edit randomly, save and close)
• cp/mv

• Cached writes (HDFS is an immutable FS (so writes=delete+insert))
• Random writes (slow - because of the immutability of HDFS - but working!)
• Very fast ls (cached directory metadata)
• directory stored as a zip file in HDFS (to solve small files problem)
• directory stored as a avro file in HDFS (to solve small files problem)
• CRC32 checksum
• Load options from configuration file

• access
• chmod
• chown
• getattr
• readdir
• readlink
• mknod
• rmdir
• mkdir
• statfs
• unlink
• symlink
• rename
• link
• utimens

• open
• create
• read
• write (caching is done in memory)
• truncate
• flush (writes the in memory written chunks to a temporary file in the local FS in the right order and calls fsync)
• fsync (send the temporary file to HDFS)
• release