server side Mongodb GridFS file copying

Question

Per business requirements I need provide possibility to copy content of some file on GridFS. Of course it can be done over domain-specific layer. But in this case I can see some overhead:

• take stream from mongo-server
• allocate memory on business-layer
• read
• place back to mongo-server

Obvious solution is write mongo-side JavaScript that will perform copying in bound of single server. So my questions:

• Where is the description of API to manage GridFS on JavaScript?
• Is there any issues if my GridFS is sharded?
• Is there any issues if my GridFS is replicated?

Thank you in advance

Answer 1

gridfs_session = gridfs.GridFS(mongo.session)
out_file = gridfs_session.get(file_id)
new_copy = gridfs_session.put(out_file, content_type=out_file.content_type)

Answer 2

You never need to copy a GridFS file within a single server, because GridFS files are immutable: you can create, read, or delete them, but not modify them. So there's no reason to make a copy.

Copying from one server to another should be done via a driver; there's no built-in support for copying directly from a MongoDB server to another.

Answer 3

The 'normal' js driver does not support GridFS.

You could do it in Node.js. The documentation is here:

http://mongodb.github.com/node-mongodb-native/markdown-docs/gridfs.html

For replica-sets the documentation can be found here for Node.js:

http://mongodb.github.com/node-mongodb-native/markdown-docs/replicaset.html

For simple one-time copying of some files you can use mongofiles on command line (using a temporary file :( ) :

http://www.mongodb.org/display/DOCS/GridFS+Tools

mongofiles --host HOST get currentfilename
mongofiles --host HOST put -l currentfilename newfilename
rm currentfilename

I however don't know how well mongofiles works with sharding and replicas but I would expect it to work.