@dkocher commented

Sounds like #3125. The hostname is configurable now but explicit support for Eucalyptus Walrus is currently disabled.

151c472 commented

I had not seen that and in principle it is the same. I was specifically talking about fs3 http://fs3.sourceforge.net/ which is similar in functionality. I am aware that it would need some updates to become fully compatible which I was going to use this wonderful application to work with. In the end there would be a better was to manage files in a scalable multi-server environment.

I was simply thinking that if one can set the "base" hostname and S3 is treated like a protocol only, then this would be the best tool ever for managing them and it would promote use of centralize file services as well.

In any event, a hat-tip you the author for coming up with the idea to integrate S3 into the file management tool.

@dkocher commented

I would love I could test this out and give some feedback. In theory, if fs3 is API compatible it should just work.

151c472 commented

fs3 is still under development (by me) and if there is a way that I can set the base hostname for use by Cyberduck, I will definitely make the modifications necessary to support Cyberduck and recommend it as well. I agree if fs3 is compatible it should just work, but fs3 is a "subset" of the S3 service and I need to make sure it has what is required by your tool.

@dkocher commented

Replying to [comment:5 sunsetbrew@…]:

fs3 is still under development (by me) and if there is a way that I can set the base hostname for use by Cyberduck, I will definitely make the modifications necessary to support Cyberduck and recommend it as well. I agree if fs3 is compatible it should just work, but fs3 is a "subset" of the S3 service and I need to make sure it has what is required by your tool.

It should work just fine by entering another hostname in the connection or bookmark setting.

151c472 commented

I had tried that but it seemed to assume the hostname I provided was a bucket name instead. It prepended it to the amazon S3 hostname. For example, if i enter fs3.somewhere.com in the server box, it tries to connect to fs3.somewhere.com.s3.amazonaws.com

151c472 commented

Here is how to recreate it.

• open cyberduck

• toggle log drawer to open it
• click open connectin
• select amazon s3
• enter carbon.expressobeans.com
• select anonymous login
• connection

The dialog result is
I/O Error: Connection failed
https//anonymous@carbon.expressobeans.com/
Bucket not available: carbon.expressobeans.com.

Log reads:
HEAD / HTTP/1.1[\r][\n]
Date: Fri, 29 Jan 2010 02:30:33 GMT[\r][\n]
Content-Type: [\r][\n]
User-Agent: Cyberduck/3.4.1 (5780) (Mac OS X/10.6.2) (i386)[\r][\n]
Host: carbon.expressobeans.com.s3.amazonaws.com[\r][\n]
[\r][\n]
HTTP/1.1 404 Not Found[\r][\n]
HTTP/1.1 404 Not Found[\r][\n]
x-amz-request-id: 6A541EFFEB853A1B[\r][\n]
x-amz-id-2: jrri9PzAjla8gYa/hypX1WjPiFAqhgdaHFG7W8pMLnU8iXRx4UuksBQmE2/RhEUZ[\r][\n]
Content-Type: application/xml[\r][\n]
Transfer-Encoding: chunked[\r][\n]
Date: Fri, 29 Jan 2010 02:30:18 GMT[\r][\n]
Server: AmazonS3[\r][\n]
[\r][\n]

Summary: It is clear from the log that it never connected to the correct server, but instead connected to amazon's service and amazon treated the entered server name as a bucket (expected them to think it was bucket).

Desired result is to make the request to the server of my choice.

-** Never expected carbon (using fs3) to work especially since login is required, but connectivity would be a starting point.

@dkocher commented

Replying to [comment:7 sunsetbrew@…]:

I had tried that but it seemed to assume the hostname I provided was a bucket name instead. It prepended it to the amazon S3 hostname. For example, if i enter fs3.somewhere.com in the server box, it tries to connect to fs3.somewhere.com.s3.amazonaws.com

I fixed this in 58c56fb. A new snapshot is available in a few minutes.

151c472 commented

Thank you!

I will get started on this and let you know when it is completed.