org.bouncycastle.crypto.InvalidCipherTextException: mac check in GCM failed

[a1ekseev]

I'm trying to run a test on GitLab CI, but have run into a problem.

    org.bouncycastle.crypto.InvalidCipherTextException: mac check in GCM failed
    	at org.bouncycastle.crypto.modes.GCMBlockCipher.doFinal(Unknown Source)
    	at io.minio.admin.Crypto.decrypt(Crypto.java:189)
    	at io.minio.admin.MinioAdminClient.listUsers(MinioAdminClient.java:242)

So far it has not been possible to reproduce the problem locally.
Do you have any thoughts?

[balamurugana]

Seems like a network issue. Refer bcgit/bc-java#1019

[pbrw]

This is possibly the same issue as in Python SDK minio/minio-py#1343

[rafiulahad]

I am having the same error with minio Java admin client (https://github.com/minio/minio-java/blob/master/adminapi/src/main/java/io/minio/admin/MinioAdminClient.java). The server is running on my MacBook Pro with Apple M3 max chip on MacOS Sequoia 15.0.1. The Java version is openjdk 21.0.2 2024-01-16. The call to listUsers() in the following code returns the error org.bouncycastle.crypto.InvalidCipherTextException: mac check in GCM failed'
All other method calls are okay.
public boolean userExists(String userName) {
Map<String,UserInfo> userInfoMap;
try {
userInfoMap = minioAdminClient.listUsers();
return userInfoMap != null && userInfoMap.get(userName) != null;
} catch (NoSuchAlgorithmException | InvalidKeyException | IOException | InvalidCipherTextException e) {
throw new RuntimeException(e);
}
}

[balamurugana]

I am able to reproduce the bug with current minio-java code.

[rafiulahad]

Also, In my opinion, UserInfo getUserInfo(String accessKey) should not throw an exception when the user does not exist. It should return null.

[marktheunissen]

@balamurugana here's a clue, user reporting that changing the MinIO build architecture from ARM to x86 fixes the problem, in this discussion here: minio/minio#20658

[balamurugana]

@balamurugana here's a clue, user reporting that changing the MinIO build architecture from ARM to x86 fixes the problem, in this discussion here: minio/minio#20658

@marktheunissen There is a different issue I found in my local testing.