My stuck point for fluent-plugin-s3 configuration

Why not uploaded?

I was really stuck in configuration of fluent-plugin-s3. (Actually, the plugin is not responsible for this. I'll write later)

This is an out-plugin for fluentd in order to upload to S3. The match configuration requires some parameters to access S3 and the path. They are AWS access key id, secret key, bucket name, endpoint and so on.

Here is an example for it, https://github.com/fluent/fluent-plugin-s3#configuration

BTW, I configured it and started fluentd, but it seemed no logs would be uploaded to S3. I was wondering if it's not been uploaded. It was very simple mistake for path parameter.

At first, I configured like this.

s3_object_key_format %{path}%{time_slice}_%{index}.%{file_extension}
path /tmp/logs

This never had worked.

Actually, I had to write like

path tmp/logs

It seems that we must not start with /. Mh...

aws-sdk gem

The plugin uses aws-sdk like this. I checked the behavior. The version of aws-sdk is 1.9.1 provided by rubygems.

## path-test.rb
require "aws-sdk"
bucket = AWS::S3.new.buckets["s3://example.com"]
obj  bucket.objects["/tmpfile"]
p obj 

$ ruby path-test.rb
<AWS::S3::S3Object:s3://example.com//tmpfile>

I tried to create a new object at /tmpfile, but it didn't work. This issue depends on aws-sdk.

Don't forget removing / at head!