02267_file_globs_schema_inference is broken on AARCH64

[Algunenano]

Run: https://s3.amazonaws.com/clickhouse-test-reports/0/91b70b412043bdda6d7f322ecc0ae94156faa0f3/stateless_tests__aarch64_.html

Error is only found in the runlog:

2022-09-29 11:29:45 02267_file_globs_schema_inference:                                      [ FAIL ] 0.13 sec. - return code: 27
2022-09-29 11:29:45 Expected server error code: 117 but got: 27 (query: select * from file('02267_data*.jsonl') settings schema_inference_use_cache_for_file=0; --{serverError INCORRECT_DATA}).
2022-09-29 11:29:45 Received exception from server (version 22.10.1):
2022-09-29 11:29:45 Code: 27. DB::Exception: Received from localhost:9000. DB::ParsingException. DB::ParsingException: Cannot parse input: expected '{' before: '1\n[1,2,3]\n': While executing JSONEachRowRowInputFormat: While executing File: (in file/uri /var/lib/clickhouse/user_files/02267_data1.jsonl): (at row 1)
2022-09-29 11:29:45 . (CANNOT_PARSE_INPUT_ASSERTION_FAILED)
2022-09-29 11:29:45 (query: select * from file('02267_data*.jsonl') settings schema_inference_use_cache_for_file=0; --{serverError INCORRECT_DATA})
2022-09-29 11:29:45 , result:
2022-09-29 11:29:45 
2022-09-29 11:29:45 1
2022-09-29 11:29:45 \N
2022-09-29 11:29:45 1
2022-09-29 11:29:45 \N
2022-09-29 11:29:45 
2022-09-29 11:29:45 stdout:
2022-09-29 11:29:45 1
2022-09-29 11:29:45 \N
2022-09-29 11:29:45 1
2022-09-29 11:29:45 \N
2022-09-29 11:29:45 
2022-09-29 11:29:45 
2022-09-29 11:29:45 Settings used in the test: --max_insert_threads=0 --group_by_two_level_threshold=1152921504606846976 --group_by_two_level_threshold_bytes=1152921504606846976 --distributed_aggregation_memory_efficient=1 --fsync_metadata=1 --output_format_parallel_formatting=0 --input_format_parallel_parsing=0 --min_chunk_bytes_for_parallel_parsing=11984955 --max_read_buffer_size=907339 --prefer_localhost_replica=0 --max_block_size=14478 --max_threads=38 --optimize_or_like_chain=0 --optimize_read_in_order=1 --read_in_order_two_level_merge_threshold=10 --optimize_aggregation_in_order=1 --aggregation_in_order_max_block_bytes=37993372 --use_uncompressed_cache=1 --min_bytes_to_use_direct_io=169177437 --min_bytes_to_use_mmap_io=0 --local_filesystem_read_method=pread_threadpool --remote_filesystem_read_method=threadpool --local_filesystem_read_prefetch=1 --remote_filesystem_read_prefetch=0 --compile_expressions=1 --compile_aggregate_expressions=0 --compile_sort_description=1 --merge_tree_coarse_index_granularity=18 --optimize_distinct_in_order=0 --optimize_sorting_by_input_stream_properties=1
2022-09-29 11:29:45 

There isn't a single good run in master in the last 30 days.

[tavplubix]

#41877

[Algunenano]

Looking into this test there are several problem:

• Can't be rerun since it hardcodes new files and won't delete them.
• The test assumes '02267_data*.jsonl' reads the files in alphabetical order but that's not the case. Depending on which file is read first a different error will be thrown.

I'll open a PR to both guarantee the files are read in alphabetical order and that the test can be re-run multiple times and in parallel (since it's simple to do in shell tests).

[KochetovNicolai]

https://s3.amazonaws.com/clickhouse-test-reports/42319/bd88f0b1d9e35403ec20b737fdc0a6dd79fa52d1/stateless_tests__aarch64_.html

[Algunenano]

Should be fixed by #41877