跳到主要内容
跳到主要内容

flameGraph

聚合函数,通过堆栈跟踪列表构建一个 flamegraph。输出一个字符串数组,可以被 flamegraph.pl utility 用于渲染 flamegraph 的 SVG。

Syntax

flameGraph(traces, [size], [ptr])

Parameters

  • traces — 一个堆栈跟踪。Array(UInt64)。
  • size — 内存分析的分配大小。(可选 - 默认 1)。UInt64
  • ptr — 内存分配地址。(可选 - 默认 0)。UInt64
备注

ptr != 0 的情况下,flameGraph 将使用相同的 size 和 ptr 将分配(size > 0)和释放(size < 0)进行映射。 只显示未被释放的分配。未映射的释放将被忽略。

Returned value

Examples

Building a flamegraph based on a CPU query profiler

SET query_profiler_cpu_time_period_ns=10000000;
SELECT SearchPhrase, COUNT(DISTINCT UserID) AS u FROM hits WHERE SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY u DESC LIMIT 10;
clickhouse client --allow_introspection_functions=1 -q "select arrayJoin(flameGraph(arrayReverse(trace))) from system.trace_log where trace_type = 'CPU' and query_id = 'xxx'" | ~/dev/FlameGraph/flamegraph.pl  > flame_cpu.svg

Building a flamegraph based on a memory query profiler, showing all allocations

SET memory_profiler_sample_probability=1, max_untracked_memory=1;
SELECT SearchPhrase, COUNT(DISTINCT UserID) AS u FROM hits WHERE SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY u DESC LIMIT 10;
clickhouse client --allow_introspection_functions=1 -q "select arrayJoin(flameGraph(trace, size)) from system.trace_log where trace_type = 'MemorySample' and query_id = 'xxx'" | ~/dev/FlameGraph/flamegraph.pl --countname=bytes --color=mem > flame_mem.svg

Building a flamegraph based on a memory query profiler, showing allocations which were not deallocated in query context

SET memory_profiler_sample_probability=1, max_untracked_memory=1, use_uncompressed_cache=1, merge_tree_max_rows_to_use_cache=100000000000, merge_tree_max_bytes_to_use_cache=1000000000000;
SELECT SearchPhrase, COUNT(DISTINCT UserID) AS u FROM hits WHERE SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY u DESC LIMIT 10;
clickhouse client --allow_introspection_functions=1 -q "SELECT arrayJoin(flameGraph(trace, size, ptr)) FROM system.trace_log WHERE trace_type = 'MemorySample' AND query_id = 'xxx'" | ~/dev/FlameGraph/flamegraph.pl --countname=bytes --color=mem > flame_mem_untracked.svg

Build a flamegraph based on memory query profiler, showing active allocations at the fixed point of time

SET memory_profiler_sample_probability=1, max_untracked_memory=1;
SELECT SearchPhrase, COUNT(DISTINCT UserID) AS u FROM hits WHERE SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY u DESC LIMIT 10;
  • 1 - 每秒内存使用情况
SELECT event_time, m, formatReadableSize(max(s) AS m) FROM (SELECT event_time, sum(size) OVER (ORDER BY event_time) AS s FROM system.trace_log WHERE query_id = 'xxx' AND trace_type = 'MemorySample') GROUP BY event_time ORDER BY event_time;
  • 2 - 找到内存使用量最大的时间点
SELECT argMax(event_time, s), max(s) FROM (SELECT event_time, sum(size) OVER (ORDER BY event_time) AS s FROM system.trace_log WHERE query_id = 'xxx' AND trace_type = 'MemorySample');
  • 3 - 在固定时间点固定活动分配
clickhouse client --allow_introspection_functions=1 -q "SELECT arrayJoin(flameGraph(trace, size, ptr)) FROM (SELECT * FROM system.trace_log WHERE trace_type = 'MemorySample' AND query_id = 'xxx' AND event_time <= 'yyy' ORDER BY event_time)" | ~/dev/FlameGraph/flamegraph.pl --countname=bytes --color=mem > flame_mem_time_point_pos.svg
  • 4 - 在固定时间点找到释放情况
clickhouse client --allow_introspection_functions=1 -q "SELECT arrayJoin(flameGraph(trace, -size, ptr)) FROM (SELECT * FROM system.trace_log WHERE trace_type = 'MemorySample' AND query_id = 'xxx' AND event_time > 'yyy' ORDER BY event_time desc)" | ~/dev/FlameGraph/flamegraph.pl --countname=bytes --color=mem > flame_mem_time_point_neg.svg