Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

*:support capacity unit read/write statistics #318

Merged
merged 27 commits into from
May 30, 2019
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion rdsn
11 changes: 11 additions & 0 deletions scripts/falcon_screen.json
Original file line number Diff line number Diff line change
Expand Up @@ -39,6 +39,17 @@
"method": "",
"timespan": 86400
},
{
"title": "集群读写吞吐量(统计最近10s内的读写吞吐量,单位:Capacity Unit)",
"endpoints": ["cluster=${cluster.name} job=collector service=pegasus"],
"counters": [
"collector*app.pegasus*app.stat.recent_read_cu#_all_/cluster=${cluster.name},job=collector,port=${collector.port},service=pegasus",
"collector*app.pegasus*app.stat.recent_write_cu#_all_/cluster=${cluster.name},job=collector,port=${collector.port},service=pegasus"
],
"graph_type": "a",
"method": "",
"timespan": 86400
},
{
"title": "集群Load-Balance状态(待执行的balance操作数、已执行的balance操作数等)",
"endpoints": ["cluster=${cluster.name} job=meta service=pegasus"],
Expand Down
96 changes: 15 additions & 81 deletions src/server/available_detector.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -9,10 +9,13 @@
#include <sstream>

#include "base/pegasus_key_schema.h"
#include "result_writer.h"

namespace pegasus {
namespace server {

DEFINE_TASK_CODE(LPC_DETECT_AVAILABLE, TASK_PRIORITY_COMMON, ::dsn::THREAD_POOL_DEFAULT)

available_detector::available_detector()
: _client(nullptr),
_ddl_client(nullptr),
Expand Down Expand Up @@ -64,6 +67,7 @@ available_detector::available_detector()
}
_client = pegasus_client_factory::get_client(_cluster_name.c_str(), _app_name.c_str());
dassert(_client != nullptr, "Initialize the _client failed");
_result_writer = dsn::make_unique<result_writer>(_client);
_ddl_client.reset(new replication_ddl_client(_meta_list));
dassert(_ddl_client != nullptr, "Initialize the _ddl_client failed");
if (!_alert_email_address.empty()) {
Expand Down Expand Up @@ -118,45 +122,28 @@ available_detector::available_detector()
_pfc_available_minute->set(1000000); // init to 100%
}

available_detector::~available_detector()
{
// don't delete _client, just set _client to nullptr.
_client = nullptr;
stop();
}
available_detector::~available_detector() = default;

void available_detector::start()
{
// available detector delay 60s to wait the pegasus finishing the initialization.
_detect_timer = ::dsn::tasking::enqueue(LPC_DETECT_AVAILABLE,
nullptr,
&_tracker,
std::bind(&available_detector::detect_available, this),
0,
std::chrono::minutes(1));
report_availability_info();
}

void available_detector::stop()
{
for (auto &tptr : _detect_tasks) {
if (tptr != nullptr)
tptr->cancel(true);
}

if (_detect_timer != nullptr)
_detect_timer->cancel(true);

if (_report_task != nullptr)
_report_task->cancel(true);
}
void available_detector::stop() { _tracker.cancel_outstanding_tasks(); }

void available_detector::detect_available()
{
if (!generate_hash_keys()) {
derror("initialize hash_keys failed, do not detect available, retry after 60 seconds");
_detect_timer =
::dsn::tasking::enqueue(LPC_DETECT_AVAILABLE,
nullptr,
&_tracker,
std::bind(&available_detector::detect_available, this),
0,
std::chrono::minutes(1));
Expand All @@ -172,7 +159,7 @@ void available_detector::detect_available()
auto call_func = std::bind(&available_detector::on_detect, this, i);
_detect_tasks[i] =
::dsn::tasking::enqueue_timer(LPC_DETECT_AVAILABLE,
nullptr,
&_tracker,
std::move(call_func),
std::chrono::seconds(_detect_interval_seconds));
}
Expand Down Expand Up @@ -225,7 +212,7 @@ void available_detector::report_availability_info()
};
_report_task = ::dsn::tasking::enqueue_timer(
LPC_DETECT_AVAILABLE,
nullptr,
&_tracker,
std::move(call_func),
std::chrono::minutes(1),
0,
Expand Down Expand Up @@ -426,9 +413,7 @@ void available_detector::on_day_report()
}
}

// set try_count to 3000 (keep on retrying for 3000 minutes) to avoid losting detect result
// if the result table is also unavailable for a long time.
set_detect_result(hash_key, sort_key, value, 3000);
_result_writer->set_result(hash_key, sort_key, value);
}

void available_detector::on_hour_report()
Expand All @@ -452,9 +437,7 @@ void available_detector::on_hour_report()
_pfc_fail_times_hour->set(fail_times);
_pfc_available_hour->set(available);

// set try_count to 3000 (keep on retrying for 3000 minutes) to avoid losting detect result
// if the result table is also unavailable for a long time.
set_detect_result(hash_key, sort_key, value, 3000);
_result_writer->set_result(hash_key, sort_key, value);
}

void available_detector::on_minute_report()
Expand All @@ -478,56 +461,7 @@ void available_detector::on_minute_report()
_pfc_fail_times_minute->set(fail_times);
_pfc_available_minute->set(available);

// set try_count to 3000 (keep on retrying for 3000 minutes) to avoid losting detect result
// if the result table is also unavailable for a long time.
set_detect_result(hash_key, sort_key, value, 3000);
}

void available_detector::set_detect_result(const std::string &hash_key,
const std::string &sort_key,
const std::string &value,
int try_count)
{
_client->async_set(
hash_key,
sort_key,
value,
[this, hash_key, sort_key, value, try_count](int err,
pegasus_client::internal_info &&info) {
if (err != PERR_OK) {
int new_try_count = try_count - 1;
if (new_try_count > 0) {
derror("set_detect_result fail, hash_key = %s, sort_key = %s, value = %s, "
"error = %s, left_try_count = %d, try again after 1 minute",
hash_key.c_str(),
sort_key.c_str(),
value.c_str(),
_client->get_error_string(err),
new_try_count);
::dsn::tasking::enqueue(LPC_DETECT_AVAILABLE,
nullptr,
[this, hash_key, sort_key, value, new_try_count]() {
set_detect_result(
hash_key, sort_key, value, new_try_count);
},
0,
std::chrono::minutes(1));
} else {
derror("set_detect_result fail, hash_key = %s, sort_key = %s, value = %s, "
"error = %s, left_try_count = %d, do not try again",
hash_key.c_str(),
sort_key.c_str(),
value.c_str(),
_client->get_error_string(err),
new_try_count);
}
} else {
ddebug("set_detect_result succeed, hash_key = %s, sort_key = %s, value = %s",
hash_key.c_str(),
sort_key.c_str(),
value.c_str());
}
});
}
_result_writer->set_result(hash_key, sort_key, value);
}
} // namespace
} // namespace server
} // namespace pegasus
15 changes: 7 additions & 8 deletions src/server/available_detector.h
Original file line number Diff line number Diff line change
Expand Up @@ -13,22 +13,18 @@ namespace server {

using ::dsn::replication::replication_ddl_client;

DEFINE_TASK_CODE(LPC_DETECT_AVAILABLE, TASK_PRIORITY_COMMON, ::dsn::THREAD_POOL_DEFAULT)
class result_writer;

class available_detector
{
public:
available_detector();
virtual ~available_detector();
~available_detector();

void start();
void stop();

private:
void set_detect_result(const std::string &hash_key,
const std::string &sort_key,
const std::string &value,
int try_count);
// generate hash_keys that can access every partition.
bool generate_hash_keys();
void on_detect(int32_t idx);
Expand All @@ -41,8 +37,11 @@ class available_detector
void on_minute_report();

private:
dsn::task_tracker _tracker;
std::string _cluster_name;
std::string _app_name;
// for writing detect result
std::unique_ptr<result_writer> _result_writer;
// client to access server.
pegasus_client *_client;
std::shared_ptr<replication_ddl_client> _ddl_client;
Expand Down Expand Up @@ -87,5 +86,5 @@ class available_detector
::dsn::perf_counter_wrapper _pfc_fail_times_minute;
::dsn::perf_counter_wrapper _pfc_available_minute;
};
}
}
} // namespace server
} // namespace pegasus
Loading