SparksFyz opened a new pull request, #3790:
URL: https://github.com/apache/celeborn/pull/3790
### What changes were proposed in this pull request?
Preserve Spark's effective IO cipher transformation when
`SparkCryptoHandler`
builds its minimized `SparkConf`.
Add bidirectional interoperability tests between `SparkCryptoHandler` and
Spark's native `CryptoStreamUtils` using `AES/CBC/PKCS5Padding`.
### Why are the changes needed?
`CryptoStreamUtils.toCryptoConf()` only extracts
`spark.io.encryption.commons.config.*`, while Spark reads
`IO_CRYPTO_CIPHER_TRANSFORMATION` separately from `SparkConf`.
As a result, `SparkCryptoHandler` previously dropped a non-default
transformation and silently fell back to Spark's default
`AES/CTR/NoPadding`.
Existing self-round-trip tests did not detect this because both encryption
and decryption used the same minimized configuration.
### Does this PR resolve a correctness bug?
- [x] Yes
### Does this PR introduce _any_ user-facing change?
- [x] Yes
A non-default Spark IO cipher transformation is now preserved by Celeborn.
The default `AES/CTR/NoPadding` behavior remains unchanged.
### How was this patch tested?
- Added Celeborn encryption -> Spark `CryptoStreamUtils` decryption
coverage.
- Added Spark `CryptoStreamUtils` encryption -> Celeborn decryption
coverage.
- Spark 3.5 `SparkCryptoHandlerSuiteJ`: 10 tests passed.
- Spark 3.0 / Scala 2.12.10 / JDK 11 compilation passed.
- Spark 4.0 / Scala 2.13.16 / JDK 17 compilation passed.
- Spark 3.5 Spotless check passed.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
