zhongyuankai

### Changelog category (leave one): - Performance Improvement ### Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md): Implement compress marks and primary key. close #34437.

Add server setting `aggregate_function_group_array_has_limit_size`

4

### Changelog category (leave one): - Improvement ### Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md): Add this `aggregate_function_group_array_has_limit_size`setting to support discarding data in some...

Improve the performance of `MODIFY TTL`

7

```sql CREATE TABLE test_fast_ttl (`id` UInt32, `name` String, `create_time` DateTime) ENGINE = MergeTree ORDER BY id TTL create_time + toIntervalDay(300); INSERT INTO test_fast_ttl SELECT number, 'AAA', date_sub(day, 100, now()) from...

**Describe what's wrong** If the analyzer is not enabled in versions prior to 24.5.3.5, the query will fail in the following cases. In this example https://fiddle.clickhouse.com/5f9fbe55-90ec-4b1d-acbe-f63eef3d8b5f According to my investigation,...

配置了150采集任务，采集延迟很高，原因是inotify event queue溢出，LogInput线程消费不过来，有什么办法能提升采集能力吗？顺便想问一下为什么LogInput中只设计了一个线程来消费inotify event，可以设计为多线程吗？ ``` load:245.71 226.69 210.91 51/11232 1143911 plugin_stats:{ "flushers" : { "flusher_kafka_v2" : 150 }, "inputs" : { "input_file" : 150 }, "processors" : { "processor_add_fields" :...

Strips binary keep a static symbol table

8

Strips binary, we still to keep a static symbol table for good stack traces.

Multi-line parsing supports SIMD optimization

5

Improve the performance of parsing newlines through SIMD. The following are the performance comparison test results under 150 tasks and 180MB/s traffic: Before optimization:   After optimization:  ...