Amazon s3a returns 400 Bad Request with Spark

Question

For checkout purpose I try to set up an Amazon S3 bucket as checkpoint file.

val checkpointDir = "s3a://bucket-name/checkpoint.txt"
val sc = new SparkContext(conf)
sc.setLocalProperty("spark.default.parallelism", "30")
sc.hadoopConfiguration.set("fs.s3a.access.key", "xxxxx")
sc.hadoopConfiguration.set("fs.s3a.secret.key", "xxxxx")
sc.hadoopConfiguration.set("fs.s3a.endpoint", "bucket-name.s3-website.eu-central-1.amazonaws.com")
val ssc = new StreamingContext(sc, Seconds(10))
ssc.checkpoint(checkpointDir)

but it stops with this exception

Exception in thread "main" com.amazonaws.services.s3.model.AmazonS3Exception: Status Code: 400, AWS Service: Amazon S3, AWS Request ID: 9D8E8002H3BBDDC7, AWS Error Code: null, AWS Error Message: Bad Request, S3 Extended Request ID: Qme5E3KAr/KX0djiq9poGXPJkmr0vuXAduZujwGlvaAl+oc6vlUpq7LIh70IF3LNgoewjP+HnXA=
at com.amazonaws.http.AmazonHttpClient.handleErrorResponse(AmazonHttpClient.java:798)
at com.amazonaws.http.AmazonHttpClient.executeHelper(AmazonHttpClient.java:421)
at com.amazonaws.http.AmazonHttpClient.execute(AmazonHttpClient.java:232)
at com.amazonaws.services.s3.AmazonS3Client.invoke(AmazonS3Client.java:3528)
at com.amazonaws.services.s3.AmazonS3Client.headBucket(AmazonS3Client.java:1031)
at com.amazonaws.services.s3.AmazonS3Client.doesBucketExist(AmazonS3Client.java:994)
at org.apache.hadoop.fs.s3a.S3AFileSystem.initialize(S3AFileSystem.java:154)
at org.apache.hadoop.fs.FileSystem.createFileSystem(FileSystem.java:2596)
at org.apache.hadoop.fs.FileSystem.access$200(FileSystem.java:91)
at org.apache.hadoop.fs.FileSystem$Cache.getInternal(FileSystem.java:2630)
at org.apache.hadoop.fs.FileSystem$Cache.get(FileSystem.java:2612)
at org.apache.hadoop.fs.FileSystem.get(FileSystem.java:370)
at org.apache.hadoop.fs.Path.getFileSystem(Path.java:296)
at org.apache.spark.streaming.StreamingContext.checkpoint(StreamingContext.scala:232)
at com.misterbell.shiva.StreamingApp$.main(StreamingApp.scala:89)
at com.misterbell.shiva.StreamingApp.main(StreamingApp.scala)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.lang.reflect.Method.invoke(Method.java:497)
at org.apache.spark.deploy.SparkSubmit$.org$apache$spark$deploy$SparkSubmit$$runMain(SparkSubmit.scala:664)
at org.apache.spark.deploy.SparkSubmit$.doRunMain$1(SparkSubmit.scala:169)
at org.apache.spark.deploy.SparkSubmit$.submit(SparkSubmit.scala:192)
at org.apache.spark.deploy.SparkSubmit$.main(SparkSubmit.scala:111)
at org.apache.spark.deploy.SparkSubmit.main(SparkSubmit.scala)

I don't understand why I got this error and I can't find any example.

Answer 1

This message correspond to something like "bad endpoint" or bad signature version support.

like seen here frankfurt is the only one that not support signature version 2. And it's the one I picked.

Of course after all my reserch can't say what is signature version, it's not obvious in the documentation. But the V2 seems to work with s3a.

The endpoint seen in the S3 interface is not the real endpoint it's just the web endpoint.

you have to use one of theses endpoint like that sc.hadoopConfiguration.set("fs.s3a.endpoint", "s3.eu-west-1.amazonaws.com")

But it's work by default with US endpoint

Answer 2

If you'd like to anyway use the region that supports Signature V4 in spark you can pass flag -Dcom.amazonaws.services.s3.enableV4 to the driver options and executor options on runtime. For example:

spark-submit --conf spark.driver.extraJavaOptions='-Dcom.amazonaws.services.s3.enableV4' \
    --conf spark.executor.extraJavaOptions='-Dcom.amazonaws.services.s3.enableV4' \
    ... (other spark options)

With this settings Spark is able to write to Frankfurt (and other V4-only regions) even with not-so-fresh AWS sdk version (com.amazonaws:aws-java-sdk:1.7.4 in my case)