This is an automated email from the ASF dual-hosted git repository.
yiguolei pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/doris.git
The following commit(s) were added to refs/heads/master by this push:
new 25ac68e2319 [opt](metrics) More precise task execution and wait worker
time (#48750)
25ac68e2319 is described below
commit 25ac68e231905ddc8ce4f45aebd0cb076a2034b7
Author: zhiqiang <hezhiqi...@selectdb.com>
AuthorDate: Wed Mar 12 00:00:08 2025 +0800
[opt](metrics) More precise task execution and wait worker time (#48750)
### What problem does this PR solve?
Add `task_execution_count_total` and `task_wait_worker_count_total =
nullptr`.
We can calculate avg task execution time and avg wait worker time by:
```
irate(doris_be_task_wait_worker_time_ns_total{thread_pool_name="Scan_normal"}[$__rate_interval])
/(irate(doris_be_task_wait_worker_count_total{thread_pool_name="Scan_normal"}[$__rate_interval])+1)
```
and
```
irate(doris_be_task_execution_time_ns_total{thread_pool_name="Scan_normal"}[$__rate_interval])
/(irate(doris_be_task_execution_count_total{thread_pool_name="Scan_normal"}[$__rate_interval])+1)
```
<img width="996" alt="image"
src="https://github.com/user-attachments/assets/37004f23-defd-4811-94ba-7afb588aca84";
/>
---
be/src/util/threadpool.cpp | 22 ++++++++++++++++------
be/src/util/threadpool.h | 6 ++++--
2 files changed, 20 insertions(+), 8 deletions(-)
diff --git a/be/src/util/threadpool.cpp b/be/src/util/threadpool.cpp
index f0f504b4dc2..4744ae2af1c 100644
--- a/be/src/util/threadpool.cpp
+++ b/be/src/util/threadpool.cpp
@@ -46,8 +46,12 @@ DEFINE_GAUGE_METRIC_PROTOTYPE_2ARG(thread_pool_queue_size,
MetricUnit::NOUNIT);
DEFINE_GAUGE_METRIC_PROTOTYPE_2ARG(thread_pool_max_queue_size,
MetricUnit::NOUNIT);
DEFINE_GAUGE_METRIC_PROTOTYPE_2ARG(thread_pool_max_threads,
MetricUnit::NOUNIT);
DEFINE_COUNTER_METRIC_PROTOTYPE_2ARG(thread_pool_submit_failed,
MetricUnit::NOUNIT);
-DEFINE_COUNTER_METRIC_PROTOTYPE_2ARG(task_execution_time_ns_total,
MetricUnit::NANOSECONDS);
-DEFINE_COUNTER_METRIC_PROTOTYPE_2ARG(task_wait_worker_time_ns_total,
MetricUnit::NANOSECONDS);
+DEFINE_COUNTER_METRIC_PROTOTYPE_2ARG(thread_pool_task_execution_time_ns_total,
+ MetricUnit::NANOSECONDS);
+DEFINE_COUNTER_METRIC_PROTOTYPE_2ARG(thread_pool_task_execution_count_total,
MetricUnit::NOUNIT);
+DEFINE_COUNTER_METRIC_PROTOTYPE_2ARG(thread_pool_task_wait_worker_time_ns_total,
+ MetricUnit::NANOSECONDS);
+DEFINE_COUNTER_METRIC_PROTOTYPE_2ARG(thread_pool_task_wait_worker_count_total,
MetricUnit::NOUNIT);
using namespace ErrorCode;
using std::string;
@@ -309,8 +313,10 @@ Status ThreadPool::init() {
INT_GAUGE_METRIC_REGISTER(_metric_entity, thread_pool_max_threads);
INT_GAUGE_METRIC_REGISTER(_metric_entity, thread_pool_queue_size);
INT_GAUGE_METRIC_REGISTER(_metric_entity, thread_pool_max_queue_size);
- INT_COUNTER_METRIC_REGISTER(_metric_entity, task_execution_time_ns_total);
- INT_COUNTER_METRIC_REGISTER(_metric_entity,
task_wait_worker_time_ns_total);
+ INT_COUNTER_METRIC_REGISTER(_metric_entity,
thread_pool_task_execution_time_ns_total);
+ INT_COUNTER_METRIC_REGISTER(_metric_entity,
thread_pool_task_execution_count_total);
+ INT_COUNTER_METRIC_REGISTER(_metric_entity,
thread_pool_task_wait_worker_time_ns_total);
+ INT_COUNTER_METRIC_REGISTER(_metric_entity,
thread_pool_task_wait_worker_count_total);
INT_COUNTER_METRIC_REGISTER(_metric_entity, thread_pool_submit_failed);
_metric_entity->register_hook("update", [this]() {
@@ -596,7 +602,9 @@ void ThreadPool::dispatch_thread() {
DCHECK_EQ(ThreadPoolToken::State::RUNNING, token->state());
DCHECK(!token->_entries.empty());
Task task = std::move(token->_entries.front());
-
task_wait_worker_time_ns_total->increment(task.submit_time_wather.elapsed_time());
+ thread_pool_task_wait_worker_time_ns_total->increment(
+ task.submit_time_wather.elapsed_time());
+ thread_pool_task_wait_worker_count_total->increment(1);
token->_entries.pop_front();
token->_active_threads++;
--_total_queued_tasks;
@@ -614,7 +622,9 @@ void ThreadPool::dispatch_thread() {
// with this threadpool, and produce a deadlock.
task.runnable.reset();
l.lock();
-
task_execution_time_ns_total->increment(task_execution_time_watch.elapsed_time());
+ thread_pool_task_execution_time_ns_total->increment(
+ task_execution_time_watch.elapsed_time());
+ thread_pool_task_execution_count_total->increment(1);
// Possible states:
// 1. The token was shut down while we ran its task. Transition to
QUIESCED.
// 2. The token has no more queued tasks. Transition back to IDLE.
diff --git a/be/src/util/threadpool.h b/be/src/util/threadpool.h
index 73f0ae33c79..ac1f21efea8 100644
--- a/be/src/util/threadpool.h
+++ b/be/src/util/threadpool.h
@@ -413,8 +413,10 @@ private:
IntGauge* thread_pool_queue_size = nullptr;
IntGauge* thread_pool_max_queue_size = nullptr;
IntGauge* thread_pool_max_threads = nullptr;
- IntCounter* task_execution_time_ns_total = nullptr;
- IntCounter* task_wait_worker_time_ns_total = nullptr;
+ IntCounter* thread_pool_task_execution_time_ns_total = nullptr;
+ IntCounter* thread_pool_task_execution_count_total = nullptr;
+ IntCounter* thread_pool_task_wait_worker_time_ns_total = nullptr;
+ IntCounter* thread_pool_task_wait_worker_count_total = nullptr;
IntCounter* thread_pool_submit_failed = nullptr;
};
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@doris.apache.org
For additional commands, e-mail: commits-h...@doris.apache.org
