Fix read parquet for different encodings across row groups #9129
Conversation
This file contains 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
facebook-github-bot
added
the
CLA Signed
✅ Deploy Preview for meta-velox canceled.
|
|
@Yuhta has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
|
@mbasmanova @majetideepak @Yuhta Could you help to review? Looks like the CI failure is irrelevant. Thanks! |
marin-ma
force-pushed
the
fix-read-different-encodings
branch
from
0946b0c
to
031f7e0
Compare
27 tasks
|
@Yuhta Can you help to review and merge? Result mismatch is a fatal issue to customer. |
Joe-Abraham
pushed a commit
to Joe-Abraham/velox
that referenced
this pull request
Jun 7, 2024
…ncubator#9129) Summary: When the parquet file contains a column that uses both plain and dictionary encodings across row groups, the output after filter can be wrong. The root cause is that `returnReaderNulls_` in `SelectiveColumnReader` may not be reset when the column encoding switches from dictionary encoding to plain encoding. Here is the parquet metadata for the test file to reproduce this issue. `n_2` is dictionary encoded in row group 2 and is plain encoded in row group 3. ``` file: file:/home/sparkuser/github/oap-project/velox/velox/dwio/parquet/tests/examples/different_encodings_with_filter.parquet creator: parquet-cpp-arrow version 15.0.0 extra: ARROW:schema = /////9gAAAAQAAAAAAAKAAwABgAFAAgACgAAAAABBAAMAAAACAAIAAAABAAIAAAABAAAAAMAAABwAAAAMAAAAAQAAACs////AAABBRAAAAAYAAAABAAAAAAAAAADAAAAbl8yAAQABAAEAAAA1P///wAAAQIQAAAAFAAAAAQAAAAAAAAAAwAAAG5fMQDE////AAAAASAAAAAQABQACAAGAAcADAAAABAAEAAAAAAAAQIQAAAAHAAAAAQAAAAAAAAAAwAAAG5fMAAIAAwACAAHAAgAAAAAAAABIAAAAAAAAAA= file schema: schema -------------------------------------------------------------------------------- n_0: OPTIONAL INT32 R:0 D:1 n_1: OPTIONAL INT32 R:0 D:1 n_2: OPTIONAL BINARY O:UTF8 R:0 D:1 row group 1: RC:5 TS:185 OFFSET:4 -------------------------------------------------------------------------------- n_0: INT32 SNAPPY DO:0 FPO:4 SZ:72/71/0.99 VC:5 ENC:RLE,PLAIN ST:[min: 1, max: 5, num_nulls: 0] n_1: INT32 SNAPPY DO:0 FPO:143 SZ:65/71/1.09 VC:5 ENC:RLE,PLAIN ST:[min: 1, max: 2, num_nulls: 0] n_2: BINARY SNAPPY DO:0 FPO:276 SZ:45/43/0.96 VC:5 ENC:RLE,PLAIN ST:[min: A, max: B, num_nulls: 3] row group 2: RC:5 TS:202 OFFSET:369 -------------------------------------------------------------------------------- n_0: INT32 SNAPPY DO:0 FPO:369 SZ:72/71/0.99 VC:5 ENC:RLE,PLAIN ST:[min: 1, max: 5, num_nulls: 0] n_1: INT32 SNAPPY DO:0 FPO:509 SZ:65/71/1.09 VC:5 ENC:RLE,PLAIN ST:[min: 1, max: 2, num_nulls: 0] n_2: BINARY SNAPPY DO:642 FPO:668 SZ:64/60/0.94 VC:5 ENC:RLE,RLE_DICTIONARY,PLAIN ST:[min: A, max: B, num_nulls: 3] row group 3: RC:5 TS:195 OFFSET:766 -------------------------------------------------------------------------------- n_0: INT32 SNAPPY DO:0 FPO:766 SZ:73/71/0.97 VC:5 ENC:RLE,PLAIN ST:[min: 6, max: 10, num_nulls: 0] n_1: INT32 SNAPPY DO:0 FPO:907 SZ:72/71/0.99 VC:5 ENC:RLE,PLAIN ST:[min: 1, max: 3, num_nulls: 0] n_2: BINARY SNAPPY DO:0 FPO:1047 SZ:55/53/0.96 VC:5 ENC:RLE,PLAIN ST:[min: F, max: J, num_nulls: 1] ``` Pull Request resolved: facebookincubator#9129 Reviewed By: mbasmanova Differential Revision: D55018054 Pulled By: Yuhta fbshipit-source-id: 83dbab02575321930262628ea60cd4ad3e7b8892
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
When the parquet file contains a column that uses both plain and dictionary encodings across row groups, the output after filter can be wrong. The root cause is that
returnReaderNulls_inSelectiveColumnReadermay not be reset when the column encoding switches from dictionary encoding to plain encoding.Here is the parquet metadata for the test file to reproduce this issue.
n_2is dictionary encoded in row group 2 and is plain encoded in row group 3.