AWS: client credentials provider NPE

[devinrsmith]

Apache Iceberg version

1.11.0 (latest release)

Query engine

None

Please describe the bug 🐞

As of 1.11.0 (#15249), AwsClientProperties#credentialsProvider unconditionally does a clientCredentialsProviderProperties.put(VendedCredentialsProvider.URI, refreshCredentialsEndpoint) if "client.credentials-provider" is set; this may result in null being put into the properties map. Other similar code paths check first to make sure the value is not null or empty before putting the value.

This does not play well with custom credentials providers / org.apache.iceberg.aws.AwsClientProperties:

java.lang.NullPointerException
	at java.base/java.util.Objects.requireNonNull(Objects.java:233)
	at java.base/java.util.stream.Collectors.lambda$uniqKeysMapAccumulator$1(Collectors.java:180)
	at java.base/java.util.stream.ReduceOps$3ReducingSink.accept(ReduceOps.java:169)
	at java.base/java.util.stream.ReferencePipeline$2$1.accept(ReferencePipeline.java:179)
	at java.base/java.util.HashMap$EntrySpliterator.forEachRemaining(HashMap.java:1858)
	at java.base/java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:509)
	at java.base/java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:499)
	at java.base/java.util.stream.ReduceOps$ReduceOp.evaluateSequential(ReduceOps.java:921)
	at java.base/java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
	at java.base/java.util.stream.ReferencePipeline.collect(ReferencePipeline.java:682)
	at org.apache.iceberg.util.PropertyUtil.filterProperties(PropertyUtil.java:191)
	at org.apache.iceberg.aws.AwsClientProperties.<init>(AwsClientProperties.java:116)
	at io.deephaven.iceberg.util.DeephavenS3ClientCredentialsProvider.create(DeephavenS3ClientCredentialsProvider.java:29)
	at java.base/java.lang.reflect.Method.invoke(Method.java:580)
	at org.apache.iceberg.common.DynMethods$UnboundMethod.invokeChecked(DynMethods.java:60)
	at org.apache.iceberg.common.DynMethods$UnboundMethod.invoke(DynMethods.java:73)
	at org.apache.iceberg.common.DynMethods$StaticMethod.invoke(DynMethods.java:186)
	at org.apache.iceberg.aws.AwsClientProperties.createCredentialsProvider(AwsClientProperties.java:314)
	at org.apache.iceberg.aws.AwsClientProperties.credentialsProvider(AwsClientProperties.java:296)
	at org.apache.iceberg.aws.AwsClientProperties.credentialsProvider(AwsClientProperties.java:218)
	at org.apache.iceberg.aws.s3.S3FileIOProperties.getCredentialsProvider(S3FileIOProperties.java:996)
	at org.apache.iceberg.aws.s3.S3FileIOProperties.applyCredentialConfigurations(S3FileIOProperties.java:985)
	at org.apache.iceberg.aws.AwsClientFactories$DefaultAwsClientFactory.lambda$s3$0(AwsClientFactories.java:116)
	at software.amazon.awssdk.utils.builder.SdkBuilder.applyMutation(SdkBuilder.java:61)
	at org.apache.iceberg.aws.AwsClientFactories$DefaultAwsClientFactory.s3(AwsClientFactories.java:115)
	at io.deephaven.iceberg.util.DeephavenAwsClientFactory.s3(DeephavenAwsClientFactory.java:49)

I'm unclear the exact semantics of nulls in property maps, and if this should be considered an upstream issue (ie, null should not be put in the map), or a downstream issue (AwsClientProperties should be tolerant of entries that have null values).

Willingness to contribute

• I can contribute a fix for this bug independently
• I would be willing to contribute a fix for this bug with guidance from the Iceberg community
• I cannot contribute a fix for this bug at this time

[devinrsmith]

It's possible this bug is only relevant for my custom implementation of AwsCredentialsProvider; it seems like #14608 changed the semantics of the properties that were being passed in. (That said, it may still be reasonable to consider the nullability of property values in the original context of the issue.)