diff options
| -rw-r--r-- | users/flokli/archeology/parse_bucket_logs.rs | 7 |
1 files changed, 5 insertions, 2 deletions
diff --git a/users/flokli/archeology/parse_bucket_logs.rs b/users/flokli/archeology/parse_bucket_logs.rs index 1fbba0506821..c794222f5b7d 100644 --- a/users/flokli/archeology/parse_bucket_logs.rs +++ b/users/flokli/archeology/parse_bucket_logs.rs @@ -29,8 +29,11 @@ fn main() -> ExitCode { 'Regexp', 'owner String , bucket String, timestamp_str String, remote_ip String, requester LowCardinality(String), request_id String, operation LowCardinality(String), key String, request_uri String, http_status String, error_code String, bytes_sent_str String, object_size_str String, total_time String, turn_around_time String, referer String, user_agent String, version_id String, host_id String, signature_version String, cipher_suite String, authentication_type String, host_header String, tls_version String, access_point_arn String, acl_required String' ) - SETTINGS format_regexp = '(\\S+) (\\S+) \\[(.*)\\] (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) ((?:\\S+ \\S+ \\S+)|\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+).*' - INTO OUTFILE '{}' COMPRESSION 'zstd' FORMAT Parquet"#, input_files, output_file)); + ORDER BY timestamp ASC + SETTINGS + format_regexp = '(\\S+) (\\S+) \\[(.*)\\] (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) ((?:\\S+ \\S+ \\S+)|\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+) (\\S+).*', + output_format_parquet_compression_method = 'zstd' + INTO OUTFILE '{}' FORMAT Parquet"#, input_files, output_file)); cmd.status().expect("clickhouse-local failed"); |