Reputation: 377
I'm using a product called Localstack to mock Amazon S3 locally, which is serving as a streaming file sink for a Flink job.
In the run logs, I can see that Flink disregards Localstack and attempts to contact Amazon S3.
Received error response: org.apache.flink.fs.s3base.shaded.com.amazonaws.services.s3.model.AmazonS3Exception: Service Unavailable
Retrying Request: HEAD https://s3.amazonaws.com /testBucket/
In a flink-conf.yaml, I've specified the following configuration properties:
s3.impl: org.apache.hadoop.fs.s3a.S3AFileSystem
s3.buffer.dir: ./tmp
s3.endpoint: localhost:4566
s3.path.style.access: true
s3.access-key: ***
s3.secret-key: ***
Why might Flink disregard the s3.endpoint
?
Upvotes: 3
Views: 904
Reputation: 135
You are almost right with your config, you will need to add http
when using localstack:
s3.endpoint: http://localhost:4566
and maybe try with extra dummy secrets as environment variables:
AWS_ACCESS_KEY_ID=foo
AWS_SECRET_ACCESS_KEY=bar
Upvotes: 1