Hi, I have set up Less3 on sever with domain and port xxxx, S3 Browser 10.0.9 works and I can upload image to bucket.

However when using the AWS SDK with the following code, host is not found.

var s3Config = new AmazonS3Config();
s3Config.ServiceURL = "http://storage.abc.co.za:3435";

            using (var cli = new AmazonS3Client("default","default",s3Config))
            {                    
                var utility = new TransferUtility(cli);
                await utility.UploadAsync(filePic.Path, "mibbmedia");
            }

Any pointers would be a great help.

I have gotten Less3 running (visual studio 2019 - windows), connect to Sql/Server 2019.
For now, I will be mounting these with the fuse-s3fs, and likely later as docker storage.

I am quite the beginner on S3, so likely missing things.

It did not seem to create me any default user or default, account, so I created my own... Watson did create all the schemas for the tables, so not hard to figure out.

INSERT INTO [dbo].[users] ([guid] ,[name] ,[email]) VALUES (newid() ,'thor' ,'[email protected]') GO

INSERT INTO [dbo].[credential] ([guid] ,[userguid] ,[description] ,[accesskey] ,[secretkey] ,[isbase64]) VALUES (newid() ,'E86D8681-A474-4876-A710-B4464D66CFFE' ,'Thor Test Account' ,'ABCDEFG' ,'ReindeerFlotilla' ,0) GO

I may have to create ACLs for this user.

I am reasonably baffled by the accesskey and the secretkey. All of the examples I see out of them show them as a 16 digit key and a binary (probably base64 string), like they were some component of an encryption pair.

So, I need to setup buckets and such, and so downloaded S3 browser as you suggest.

But, I get HMAC errors...

So, it rather seems me like it's trying to sign things with those keys, and that my choices for keys and secrets are bogus, though your examples are default/default, seemly just strings like I used.

I have been unable to find a command like tool to generate an AWS like key/secret set (without getting them from AWS itself). Seems like there would be one out there, or some trick to allow me to do it in openssl or similar. I'm not unfamiliar with crypto and hmac, but don't quite know what's being used here.

I'm ultimately wanting to write my own storage classes for Less3 to abstract several filesystems that I have, but have not yet gotten to "hello world" yet.

Thanks for your efforts!

---Thor Halbert

Good Day. Firstly thank you for a great s3 server

I have one small issue i cant seem to get past.

I setup the less3 server on a ubuntu box with mysql as db. this works great but when traffic picks up then mysql load skyrockets to 100% and when viewing mysql processes this is displayed.

Can you give me any advise on how to resolve this. Apologies for the bother and thank you once again

Hi, Im try to move my data from minio to less3, but the command line tool are not working. Seems because less3 output xml in latin1 but only utf8 is supported by mc.

mc ls less3
mc.exe: Unable to list folder. xml: encoding "iso-8859-1" declared but Decoder.CharsetReader is nil

I think this is really a WatsonWebserver issue but it manifests in Less3.

The AWS S3 SDK PutObjectRequest class has a boolean UseChunkEncoding attribute and by default this is set to true. This influences the request that the S3 SDK makes to the S3 API and the S3ServerLibrary (another dependency of Less3) correctly reads that and sets the Chunked attribute of the S3Request class:
https://github.com/jchristn/S3Server/blob/b449b11b9159fa0f8183b8c360c995eeba41dc4f/S3Server/S3Request.cs#L517-L524
if (HeaderExists("x-amz-content-sha256", false)) { string sha256content = RetrieveHeaderValue("x-amz-content-sha256"); if (!String.IsNullOrEmpty(sha256content)) { if (sha256content.Contains("STREAMING")) Chunked = true; } }

Then the Less3 code will act on that attribute of the S3Request class to determine if ReadChunk should be called:

And that ReadChunk function is implemented in WatsonWebserver where the first line of the function is:
if (!ChunkedTransfer) throw new IOException("Request is not chunk transfer-encoded.");

Where that ChunkedTransfer attribute is set from request headers:
https://github.com/jchristn/WatsonWebserver/blob/b3833c2edc7bd27d4c686e49efa2b0b412905e83/WatsonWebserver/HttpRequest.cs#L220-L223
if (curr.Key.ToLower().Equals("transfer-encoding")) { if (curr.Value.ToLower().Contains("chunked")) ChunkedTransfer = true;

But the AWS S3 SDK does not seem to set transfer-encoding header despite actually chunking the data. Which resulted in the first attempt to use my AWS S3 SDK code hitting that "Request is not chunk transfer-encoded" exception. So there's a conflict between the attribute that determines if Less3 calls ReadChunk and the attribute that WatsonWebserver expects to be set when ReadChunk is called.

Yes, there is a workaround of specifically setting UseChunkEncoding to false in the PutObjectRequest class but my preference would be to use the default behaviour of the AWS S3 SDK and not have to specifically set UseChunkEncoding to false.

I am very happy that I found this project and also thank you so much for publishing your work.
I want to know if this implementation supports resume for uploading and downloading?

Does it support multipart uploads? (doc)
Does it support range header for downloads? (doc)
Thanks.

Could definitely be me doing something wrong, but I'm having trouble with deletion. Creating buckets and uploading objects to them has been working.