AWS: Support multiple storage credential prefixes #12799
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
nastra
force-pushed
the
s3fileio-multi-prefix-support
branch
4 times, most recently
from
e12e4ff to
c443354
Compare
singhpk234
reviewed
Apr 15, 2025
| private final AtomicBoolean isResourceClosed = new AtomicBoolean(false); | ||
| private transient StackTraceElement[] createStack; | ||
| private List<StorageCredential> storageCredentials = ImmutableList.of(); | ||
| private transient volatile Cache<String, PrefixedS3Client> s3ClientCache; |
Contributor
There was a problem hiding this comment.
Would it be better if we use Execution interceptor doc which gives us the handle of the request, so that we can inspect the request and see which path it refers to ? rather than making a client per perfix ?
Contributor
Author
There was a problem hiding this comment.
I'm not really familiar with ExecutionInterceptor so I'll have to do some digging on how a potential solution would look with this. @singhpk234 is there an easy way to pass the storage credentials (that we currently pass to FileIO) to a given ExecutionInterceptor?
But generally speaking, we're only expecting a handful of different storage credentials in practice, so the cost of maintaining < 5 different client instances should be quite low.
Contributor
There was a problem hiding this comment.
Here is what i had in mind and have seen people doing in past
- Execution interceptor can be configured as part of ClientOverrideConfiguration when we are creating a S3 SDK client, I am assuming we wire the VendedCredProvider for the same way but giving the creds map to the class constructor, so we can do the same way here for the ExecutionInterceptor and pass it in SDK client creation
- Now how to provide / have refreshed creds is this
ExecutionInterceptorwill give us the handle of modify request and you can meddle with SDK header
something like this
@Override
public SdkHttpRequest modifyHttpRequest(Context.BeforeTransmission context, ExecutionAttributes executionAttributes) {
// you can also inspect from the request which path it has in this request
// and then based on the path you can wire in the creds which suits best for the path here :
SdkHttpRequest.Builder builder = context.httpRequest().toBuilder();
AwsCredentials credentials = credentialsProvider.get();
if (credentials != null && credentials.accessKeyId() != null && credentials.secretAccessKey() != null) {
builder.putHeader("X-Amz-Access-Key", credentials.accessKeyId());
builder.putHeader("X-Amz-Secret-Key", credentials.secretAccessKey());
if (credentials.sessionToken().isPresent()) {
builder.putHeader("X-Amz-Security-Token", credentials.sessionToken().get());
}
} else {
System.err.println("Warning: AWS credentials not found.");
}
return builder.build();
}
The main reason for not having one client per path is each client will have a connection pool and we may have to tune it like what is the min pool size for HTTP, if we can by gaurantee it will < 5 (which based on my understanding we can't right now) i think then we can think more.
Would love to know your thoughts for the same considering above
Contributor
Author
There was a problem hiding this comment.
thanks for the suggestion @singhpk234 and I really like that idea. I've opened #12827 to explore using a custom ExecutionInterceptor
nastra
force-pushed
the
s3fileio-multi-prefix-support
branch
3 times, most recently
from
c3ad9bb to
c4080aa
Compare
nastra
force-pushed
the
s3fileio-multi-prefix-support
branch
from
c4080aa to
2c3df22
Compare
nastra
commented
Apr 24, 2025
nastra
force-pushed
the
s3fileio-multi-prefix-support
branch
from
33fbac8 to
aa82932
Compare
aws/src/main/java/org/apache/iceberg/aws/s3/PrefixedS3Client.java
Outdated
Show resolved
Hide resolved
danielcweeks
reviewed
May 1, 2025
danielcweeks
reviewed
May 1, 2025
nastra
force-pushed
the
s3fileio-multi-prefix-support
branch
3 times, most recently
from
e36b267 to
f8470d5
Compare
nastra
force-pushed
the
s3fileio-multi-prefix-support
branch
from
f8470d5 to
395cf9f
Compare
3 tasks
amogh-jahagirdar
approved these changes
May 22, 2025
Contributor
amogh-jahagirdar
left a comment
amogh-jahagirdar
left a comment
There was a problem hiding this comment.
I thought I already approved last week for some reason, but this has looked right to me since the recent changes. Thank you @nastra!
|
Hey. I had a problem with this commit. As I understand my path start with s3a://, not s3://. How to configure it? |
Contributor
Author
|
@and124578963 can you please open a new issue with some additional details about your catalog configuration? |
3 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.