dal2: Fix SeekableRead not read correctly

[Xuanwo]

Signed-off-by: Xuanwo github@xuanwo.io

I hereby agree to the terms of the CLA available at: https://databend.rs/dev/policies/cla/

Summary

Part of #3677

The following SQL doesn't works:

create table test as select * from numbers(1234567890);
select count(*) from test;

Meeting EOF errors like:

2022-02-11T12:22:55.395256Z ERROR databend_query::servers::mysql::writers::query_result_writer: OnQuery Error: Code: 1024, displayText = fail to read block _b/4e5ffe0fad65439990109f9a0d9bf5ff.parquet, Code: 1002, displayText = External format error: underlying IO error: unexpected end of file..

   0: common_exception::exception_code::<impl common_exception::exception::ErrorCode>::ParquetError
             at common/exception/src/exception_code.rs:36:66
   1: databend_query::storages::fuse::operations::read::<impl databend_query::storages::fuse::table::FuseTable>::do_read::{{closure}}::{{closure}}::{{closure}}::{{closure}}
             at query/src/storages/fuse/operations/read.rs:88:25
   2: core::result::Result<T,E>::map_err

This is caused by the incorrect SeekableReader implementation.

Changelog

• New Feature

Test Plan

Unit Tests

Stateless Tests

[databend-bot]

Thanks for the contribution!
I have applied any labels matching special text in your PR Changelog.

Please review the labels and make any necessary changes.

[databend-bot]

Thanks for the contribution!
I have applied any labels matching special text in your PR Changelog.

Please review the labels and make any necessary changes.

[vercel]

This pull request is being automatically deployed with Vercel (learn more).
To see the status of your deployment, click below or on the icon next to each commit.

🔍 Inspect: https://vercel.com/databend/databend/3WXALENzejEVq6uBHZrRX1ynnQHL
✅ Preview: Canceled

[Deployment for 9310603 canceled]

[BohuTANG]

I remember we have a test about create table select from ..., why the test passed before?

[Xuanwo]

I remember we have a test about create table select from ..., why the test passed before?

Maybe the dataset is too small that can be fit into buffer so that there's no extra read. (under 1MB)

I will add this case into stateless test too.

[BohuTANG]

It seems we need more unit tests on dal2, like seekable.rs, to make our logic is right.

[Xuanwo]

It seems we need more unit tests on dal2, like seekable.rs, to make our logic is right.

True, I will keep improving this part.

[codecov-commenter]

Codecov Report

Merging #4107 (9310603) into main (85d443b) will increase coverage by 0%.
The diff coverage is 100%.

@@          Coverage Diff           @@
##            main   #4107    +/-   ##
======================================
  Coverage     57%     57%            
======================================
  Files        857     857            
  Lines      46462   46255   -207     
======================================
- Hits       26707   26670    -37     
+ Misses     19755   19585   -170     

Impacted Files	Coverage Δ
common/dal2/src/readers/seekable.rs	97% <100%> (+<1%)	⬆️
common/datavalues/src/arrays/string/builder.rs	43% <0%> (-13%)	⬇️
query/src/common/mod.rs	76% <0%> (-8%)	⬇️
common/dal/src/context.rs	59% <0%> (-5%)	⬇️
common/datavalues/src/arrays/comparison.rs	6% <0%> (-5%)	⬇️
common/functions/src/scalars/function2_factory.rs	91% <0%> (-1%)	⬇️
common/datavalues/src/arrays/primitive/mod.rs	38% <0%> (-1%)	⬇️
common/functions/src/scalars/strings/string.rs	98% <0%> (ø)
metasrv/src/meta_service/raftmeta.rs	49% <0%> (+<1%)	⬆️
... and 8 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 85d443b...9310603. Read the comment docs.

[BohuTANG]

/lgtm

Preferred add the unit tests in the later PR :D