ARCHIVE/rippled
1
0
Fork 0
mirror of https://github.com/XRPLF/rippled.git synced 2025-12-06 17:27:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

Performance logging and counters:

Browse Source 
* Tally and duration counters for Job Queue tasks and RPC calls
    optionally rendered by server_info and server_state, and
    optionally printed to a distinct log file.
    - Tally each Job Queue task as it is queued, starts, and
      finishes running. Track total duration queued and running.
    - Tally each RPC call as it starts and either finishes
      successfully or throws an exception. Track total running
      duration for each.
  * Track currently executing Job Queue tasks and RPC methods
    along with durations.
  * Json-formatted performance log file written by a dedicated
    thread, for above-described data.
  * New optional parameter, "counters", for server_info and
    server_state. If set, render Job Queue and RPC call counters
    as well as currently executing tasks.
  * New configuration section, "[perf]", to optionally control
    performance logging to a file.
  * Support optional sub-second periods when rendering human-readable
    time points.
This commit is contained in:
Mark Travis
2018-01-13 04:02:43 -08:00
committed by Nikolaos D. Bougalis
parent ef3bc92b82
commit 8eb8c77886
45 changed files with 10379 additions and 577 deletions
Download Patch File Download Diff File Expand all files Collapse all files

528
src/ripple/basics/impl/PerfLogImp.cpp Normal file
View File

@@ -0,0 +1,528 @@
//------------------------------------------------------------------------------
/*
This file is part of rippled: https://github.com/ripple/rippled
Copyright (c) 2018 Ripple Labs Inc.
Permission to use, copy, modify, and/or distribute this software for any
purpose with or without fee is hereby granted, provided that the above
copyright notice and this permission notice appear in all copies.
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
ANY SPECIAL , DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
//==============================================================================
#include <ripple/basics/BasicConfig.h>
#include <ripple/basics/impl/PerfLogImp.h>
#include <ripple/beast/core/CurrentThreadName.h>
#include <ripple/beast/utility/Journal.h>
#include <ripple/json/json_writer.h>
#include <ripple/json/to_string.h>
#include <boost/optional.hpp>
#include <atomic>
#include <cstdint>
#include <cstdlib>
#include <iostream>
#include <iterator>
#include <sstream>
#include <stdexcept>
#include <string>
#include <unordered_map>
#include <utility>
namespace ripple {
namespace perf {
PerfLogImp::Counters::Counters(std::vector<char const*> const& labels,
JobTypes const& jobTypes)
{
{
// populateRpc
rpc_.reserve(labels.size());
for (std::string const label : labels)
{
auto const elem = rpc_.emplace(label, Rpc());
if (!elem.second)
{
// Ensure that no other function populates this entry.
assert(false);
}
}
}
{
// populateJq
jq_.reserve(jobTypes.size());
for (auto const& job : jobTypes)
{
auto const elem = jq_.emplace(job.first, Jq(job.second.name()));
if (!elem.second)
{
// Ensure that no other function populates this entry.
assert(false);
}
}
}
}
Json::Value
PerfLogImp::Counters::countersJson() const
{
Json::Value rpcobj(Json::objectValue);
// totalRpc represents all rpc methods. All that started, finished, etc.
Rpc totalRpc;
for (auto const& proc : rpc_)
{
Json::Value p(Json::objectValue);
{
auto const sync =
[&proc]() ->boost::optional<Counters::Rpc::Sync> {
std::lock_guard<std::mutex> lock(proc.second.mut);
if (!proc.second.sync.started &&
!proc.second.sync.finished &&
!proc.second.sync.errored)
{
return boost::none;
}
return proc.second.sync;
}();
if (!sync)
continue;
p[jss::started] = std::to_string(sync->started);
totalRpc.sync.started += sync->started;
p[jss::finished] = std::to_string(sync->finished);
totalRpc.sync.finished += sync->finished;
p[jss::errored] = std::to_string(sync->errored);
totalRpc.sync.errored += sync->errored;
p[jss::duration_us] = std::to_string(sync->duration.count());
totalRpc.sync.duration += sync->duration;
}
rpcobj[proc.first] = p;
}
if (totalRpc.sync.started)
{
Json::Value totalRpcJson(Json::objectValue);
totalRpcJson[jss::started] = std::to_string(totalRpc.sync.started);
totalRpcJson[jss::finished] = std::to_string(totalRpc.sync.finished);
totalRpcJson[jss::errored] = std::to_string(totalRpc.sync.errored);
totalRpcJson[jss::duration_us] = std::to_string(
totalRpc.sync.duration.count());
rpcobj[jss::total] = totalRpcJson;
}
Json::Value jqobj(Json::objectValue);
// totalJq represents all jobs. All enqueued, started, finished, etc.
Jq totalJq("total");
for (auto const& proc : jq_)
{
Json::Value j(Json::objectValue);
{
auto const sync =
[&proc]() ->boost::optional<Counters::Jq::Sync> {
std::lock_guard<std::mutex> lock(proc.second.mut);
if (!proc.second.sync.queued &&
!proc.second.sync.started &&
!proc.second.sync.finished)
{
return boost::none;
}
return proc.second.sync;
}();
if (!sync)
continue;
j[jss::queued] = std::to_string(sync->queued);
totalJq.sync.queued += sync->queued;
j[jss::started] = std::to_string(sync->started);
totalJq.sync.started += sync->started;
j[jss::finished] = std::to_string(sync->finished);
totalJq.sync.finished += sync->finished;
j[jss::queued_duration_us] = std::to_string(
sync->queuedDuration.count());
totalJq.sync.queuedDuration += sync->queuedDuration;
j[jss::running_duration_us] = std::to_string(
sync->runningDuration.count());
totalJq.sync.runningDuration += sync->runningDuration;
}
jqobj[proc.second.label] = j;
}
if (totalJq.sync.queued)
{
Json::Value totalJqJson(Json::objectValue);
totalJqJson[jss::queued] = std::to_string(totalJq.sync.queued);
totalJqJson[jss::started] = std::to_string(totalJq.sync.started);
totalJqJson[jss::finished] = std::to_string(totalJq.sync.finished);
totalJqJson[jss::queued_duration_us] = std::to_string(
totalJq.sync.queuedDuration.count());
totalJqJson[jss::running_duration_us] = std::to_string(
totalJq.sync.runningDuration.count());
jqobj[jss::total] = totalJqJson;
}
Json::Value counters(Json::objectValue);
// Be kind to reporting tools and let them expect rpc and jq objects
// even if empty.
counters[jss::rpc] = rpcobj;
counters[jss::job_queue] = jqobj;
return counters;
}
Json::Value
PerfLogImp::Counters::currentJson() const
{
auto const present = steady_clock::now();
Json::Value jobsArray(Json::arrayValue);
auto const jobs = [this]{
std::lock_guard<std::mutex> lock(jobsMutex_);
return jobs_;
}();
for (auto const& j : jobs)
{
if (j.first == jtINVALID)
continue;
Json::Value jobj(Json::objectValue);
auto const e = jq_.find(j.first);
if (e == jq_.end())
{
assert(false);
continue;
}
// label is const and created before multi-threading so needs no lock.
jobj[jss::job] = e->second.label;
jobj[jss::duration_us] = std::to_string(
std::chrono::duration_cast<microseconds>(
present - j.second).count());
jobsArray.append(jobj);
}
Json::Value methodsArray(Json::arrayValue);
std::vector<MethodStart> methods;
{
std::lock_guard<std::mutex> lock(methodsMutex_);
methods.reserve(methods_.size());
for (auto const& m : methods_)
methods.push_back(m.second);
}
for (auto m : methods)
{
Json::Value methodobj(Json::objectValue);
methodobj[jss::method] = m.first;
methodobj[jss::duration_us] = std::to_string(
std::chrono::duration_cast<microseconds>(
present - m.second).count());
methodsArray.append(methodobj);
}
Json::Value current(Json::objectValue);
current[jss::jobs] = jobsArray;
current[jss::methods] = methodsArray;
return current;
}
//-----------------------------------------------------------------------------
void
PerfLogImp::openLog()
{
if (! setup_.perfLog.empty())
{
if (logFile_.is_open())
logFile_.close();
auto logDir = setup_.perfLog.parent_path();
if (!boost::filesystem::is_directory(logDir))
{
boost::system::error_code ec;
boost::filesystem::create_directories(logDir, ec);
if (ec)
{
JLOG(j_.fatal()) << "Unable to create performance log "
"directory " << logDir << ": " << ec.message();
signalStop_();
return;
}
}
logFile_.open(setup_.perfLog.c_str(), std::ios::out | std::ios::app);
if (! logFile_)
{
JLOG(j_.fatal()) << "Unable to open performance log " <<
setup_.perfLog << ".";
signalStop_();
}
}
}
void
PerfLogImp::run()
{
beast::setCurrentThreadName("perflog");
lastLog_ = system_clock::now();
while (true)
{
{
std::unique_lock<std::mutex> lock(mutex_);
if (stop_)
{
return;
}
if (rotate_)
{
openLog();
rotate_ = false;
}
cond_.wait_until (lock, lastLog_ + setup_.logInterval);
}
report();
}
}
void
PerfLogImp::report()
{
if (! logFile_)
// If logFile_ is not writable do no further work.
return;
auto const present = system_clock::now();
if (present < lastLog_ + setup_.logInterval)
return;
lastLog_ = present;
Json::Value report(Json::objectValue);
report[jss::time] = to_string(date::floor<microseconds>(present));
report[jss::workers] = counters_.workers_;
report[jss::hostid] = hostname_;
report[jss::counters] = counters_.countersJson();
auto cur = counters_.currentJson();
report[jss::current_activities] = counters_.currentJson();
logFile_ << Json::Compact{std::move(report)} << std::endl;
}
PerfLogImp::PerfLogImp(Setup const& setup,
Stoppable& parent,
beast::Journal journal,
std::function<void()>&& signalStop)
: Stoppable ("PerfLogImp", parent)
, setup_ (setup)
, j_ (journal)
, signalStop_ (std::move (signalStop))
{
openLog();
}
PerfLogImp::~PerfLogImp()
{
onStop();
}
void
PerfLogImp::rpcStart(std::string const& method, std::uint64_t const requestId)
{
auto counter = counters_.rpc_.find(method);
if (counter == counters_.rpc_.end())
{
assert(false);
return;
}
{
std::lock_guard<std::mutex> lock(counter->second.mut);
++counter->second.sync.started;
}
std::lock_guard<std::mutex> lock(counters_.methodsMutex_);
counters_.methods_[requestId] = {
counter->first.c_str(),
steady_clock::now()
};
}
void
PerfLogImp::rpcEnd(std::string const& method,
std::uint64_t const requestId,
bool finish)
{
auto counter = counters_.rpc_.find(method);
if (counter == counters_.rpc_.end())
{
assert(false);
return;
}
steady_time_point startTime;
{
std::lock_guard<std::mutex> lock(counters_.methodsMutex_);
auto const e = counters_.methods_.find(requestId);
if (e != counters_.methods_.end())
{
startTime = e->second.second;
counters_.methods_.erase(e);
}
else
{
assert(false);
}
}
std::lock_guard<std::mutex> lock(counter->second.mut);
if (finish)
++counter->second.sync.finished;
else
++counter->second.sync.errored;
counter->second.sync.duration +=
std::chrono::duration_cast<microseconds>(
steady_clock::now() - startTime);
}
void
PerfLogImp::jobQueue(JobType const type)
{
auto counter = counters_.jq_.find(type);
if (counter == counters_.jq_.end())
{
assert(false);
return;
}
std::lock_guard<std::mutex> lock(counter->second.mut);
++counter->second.sync.queued;
}
void
PerfLogImp::jobStart(JobType const type,
microseconds dur,
steady_time_point startTime,
int instance)
{
auto counter = counters_.jq_.find(type);
if (counter == counters_.jq_.end())
{
assert(false);
return;
}
{
std::lock_guard<std::mutex> lock(counter->second.mut);
++counter->second.sync.started;
counter->second.sync.queuedDuration += dur;
}
std::lock_guard<std::mutex> lock(counters_.jobsMutex_);
if (instance >= 0 && instance < counters_.jobs_.size())
counters_.jobs_[instance] = {type, startTime};
}
void
PerfLogImp::jobFinish(JobType const type, microseconds dur,
int instance)
{
auto counter = counters_.jq_.find(type);
if (counter == counters_.jq_.end())
{
assert(false);
return;
}
{
std::lock_guard<std::mutex> lock(counter->second.mut);
++counter->second.sync.finished;
counter->second.sync.runningDuration += dur;
}
std::lock_guard<std::mutex> lock(counters_.jobsMutex_);
if (instance >= 0 && instance < counters_.jobs_.size())
counters_.jobs_[instance] = {jtINVALID, steady_time_point()};
}
void
PerfLogImp::resizeJobs(int const resize)
{
std::lock_guard<std::mutex> lock(counters_.jobsMutex_);
counters_.workers_ = resize;
if (resize > counters_.jobs_.size())
counters_.jobs_.resize(resize, {jtINVALID, steady_time_point()});
}
void
PerfLogImp::rotate()
{
if (setup_.perfLog.empty())
return;
std::lock_guard<std::mutex> lock(mutex_);
rotate_ = true;
cond_.notify_one();
}
void
PerfLogImp::onStart()
{
if (setup_.perfLog.size())
thread_ = std::thread(&PerfLogImp::run, this);
}
void
PerfLogImp::onStop()
{
if (thread_.joinable())
{
{
std::lock_guard<std::mutex> lock(mutex_);
stop_ = true;
cond_.notify_one();
}
thread_.join();
}
if (areChildrenStopped())
stopped();
}
void
PerfLogImp::onChildrenStopped()
{
onStop();
}
//-----------------------------------------------------------------------------
PerfLog::Setup
setup_PerfLog(Section const& section, boost::filesystem::path const& configDir)
{
PerfLog::Setup setup;
std::string perfLog;
set(perfLog, "perf_log", section);
if (perfLog.size())
{
setup.perfLog = boost::filesystem::path(perfLog);
if (setup.perfLog.is_relative())
{
setup.perfLog = boost::filesystem::absolute(
setup.perfLog, configDir);
}
}
std::uint64_t logInterval;
if (get_if_exists(section, "log_interval", logInterval))
setup.logInterval = std::chrono::seconds(logInterval);
return setup;
}
std::unique_ptr<PerfLog>
make_PerfLog(
PerfLog::Setup const& setup,
Stoppable& parent,
beast::Journal journal,
std::function<void()>&& signalStop)
{
return std::make_unique<PerfLogImp>(
setup, parent, journal, std::move(signalStop));
}
} // perf
} // ripple

219
src/ripple/basics/impl/PerfLogImp.h Normal file
View File

@@ -0,0 +1,219 @@
//------------------------------------------------------------------------------
/*
This file is part of rippled: https://github.com/ripple/rippled
Copyright (c) 2018 Ripple Labs Inc.
Permission to use, copy, modify, and/or distribute this software for any
purpose with or without fee is hereby granted, provided that the above
copyright notice and this permission notice appear in all copies.
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
ANY SPECIAL , DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
//==============================================================================
#ifndef RIPPLE_BASICS_PERFLOGIMP_H
#define RIPPLE_BASICS_PERFLOGIMP_H
#include <ripple/basics/chrono.h>
#include <ripple/basics/PerfLog.h>
#include <ripple/beast/utility/Journal.h>
#include <ripple/core/Stoppable.h>
#include <ripple/protocol/JsonFields.h>
#include <ripple/rpc/impl/Handler.h>
#include <boost/asio/ip/host_name.hpp>
#include <condition_variable>
#include <cstdint>
#include <fstream>
#include <memory>
#include <mutex>
#include <string>
#include <thread>
#include <type_traits>
#include <unordered_map>
#include <utility>
#include <vector>
namespace ripple {
namespace perf {
/**
* Implementation class for PerfLog.
*/
class PerfLogImp
: public PerfLog, Stoppable
{
/**
* Track performance counters and currently executing tasks.
*/
struct Counters
{
public:
using MethodStart = std::pair<char const*, steady_time_point>;
/**
* RPC performance counters.
*/
struct Rpc
{
// Keep all items that need to be synchronized in one place
// to minimize copy overhead while locked.
struct Sync
{
// Counters for each time a method starts and then either
// finishes successfully or with an exception.
std::uint64_t started {0};
std::uint64_t finished {0};
std::uint64_t errored {0};
// Cumulative duration of all finished and errored method calls.
microseconds duration {0};
};
Sync sync;
mutable std::mutex mut;
Rpc() = default;
Rpc(Rpc const& orig)
: sync (orig.sync)
{}
};
/**
* Job Queue task performance counters.
*/
struct Jq
{
// Keep all items that need to be synchronized in one place
// to minimize copy overhead while locked.
struct Sync
{
// Counters for each time a job is enqueued, begins to run,
// finishes.
std::uint64_t queued {0};
std::uint64_t started {0};
std::uint64_t finished {0};
// Cumulative duration of all jobs' queued and running times.
microseconds queuedDuration {0};
microseconds runningDuration {0};
};
Sync sync;
std::string const label;
mutable std::mutex mut;
Jq(std::string const& labelArg)
: label (labelArg)
{}
Jq(Jq const& orig)
: sync (orig.sync)
, label (orig.label)
{}
};
// rpc_ and jq_ do not need mutex protection because all
// keys and values are created before more threads are started.
std::unordered_map<std::string, Rpc> rpc_;
std::unordered_map<std::underlying_type_t<JobType>, Jq> jq_;
std::vector<std::pair<JobType, steady_time_point>> jobs_;
int workers_ {0};
mutable std::mutex jobsMutex_;
std::unordered_map<std::uint64_t, MethodStart> methods_;
mutable std::mutex methodsMutex_;
Counters(std::vector<char const*> const& labels,
JobTypes const& jobTypes);
Json::Value countersJson() const;
Json::Value currentJson() const;
};
Setup const setup_;
beast::Journal j_;
std::function<void()> const signalStop_;
Counters counters_ {ripple::RPC::getHandlerNames(), JobTypes::instance()};
std::ofstream logFile_;
std::thread thread_;
std::mutex mutex_;
std::condition_variable cond_;
system_time_point lastLog_;
std::string const hostname_ {boost::asio::ip::host_name()};
bool stop_ {false};
bool rotate_ {false};
void openLog();
void run();
void report();
void rpcEnd(std::string const& method,
std::uint64_t const requestId,
bool finish);
public:
PerfLogImp(Setup const& setup,
Stoppable& parent,
beast::Journal journal,
std::function<void()>&& signalStop);
~PerfLogImp() override;
void rpcStart(
std::string const& method, std::uint64_t const requestId) override;
void rpcFinish(
std::string const& method,
std::uint64_t const requestId) override
{
rpcEnd(method, requestId, true);
}
void rpcError(std::string const& method,
std::uint64_t const requestId) override
{
rpcEnd(method, requestId, false);
}
void jobQueue(JobType const type) override;
void jobStart(
JobType const type,
microseconds dur,
steady_time_point startTime,
int instance) override;
void jobFinish(
JobType const type,
microseconds dur,
int instance) override;
Json::Value
countersJson() const override
{
return counters_.countersJson();
}
Json::Value
currentJson() const override
{
return counters_.currentJson();
}
void resizeJobs(int const resize) override;
void rotate() override;
// Stoppable
void onPrepare() override {}
// Called when application is ready to start threads.
void onStart() override;
// Called when the application begins shutdown.
void onStop() override;
// Called when all child Stoppable objects have stopped.
void onChildrenStopped() override;
};
} // perf
} // ripple
#endif //RIPPLE_BASICS_PERFLOGIMP_H

82
src/ripple/basics/impl/Time.cpp
View File

@@ -1,82 +0,0 @@
//------------------------------------------------------------------------------
/*
This file is part of rippled: https://github.com/ripple/rippled
Copyright (c) 2012, 2013 Ripple Labs Inc.
Permission to use, copy, modify, and/or distribute this software for any
purpose with or without fee is hereby granted, provided that the above
copyright notice and this permission notice appear in all copies.
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
ANY SPECIAL , DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
//==============================================================================
#include <ripple/beast/clock/chrono_util.h>
#include <ripple/basics/chrono.h>
#include <iomanip>
#include <sstream>
#include <tuple>
namespace ripple {
static
std::tuple<int, unsigned, unsigned>
civil_from_days(int z) noexcept
{
z += 719468;
const int era = (z >= 0 ? z : z - 146096) / 146097;
const unsigned doe = static_cast<unsigned>(z - era * 146097); // [0, 146096]
const unsigned yoe = (doe - doe/1460 + doe/36524 - doe/146096) / 365; // [0, 399]
const int y = static_cast<int>(yoe) + era * 400;
const unsigned doy = doe - (365*yoe + yoe/4 - yoe/100); // [0, 365]
const unsigned mp = (5*doy + 2)/153; // [0, 11]
const unsigned d = doy - (153*mp+2)/5 + 1; // [1, 31]
const unsigned m = mp + (mp < 10 ? 3 : -9); // [1, 12]
return std::tuple<int, unsigned, unsigned>(y + (m <= 2), m, d);
}
std::string
to_string(std::chrono::system_clock::time_point tp)
{
const char* months[] =
{
"Jan", "Feb", "Mar", "Apr", "May", "Jun",
"Jul", "Aug", "Sep", "Oct", "Nov", "Dec"
};
using namespace std::chrono;
auto s = floor<seconds>(tp.time_since_epoch());
auto sd = floor<days>(s); // number of days
s -= sd; // time of day in seconds
auto h = floor<hours>(s);
s -= h;
auto m = floor<minutes>(s);
s -= m;
int y;
unsigned mn, d;
std::tie(y, mn, d) = civil_from_days(static_cast<int>(sd.count()));
// Date-time in y/mn/d h:m:s
std::ostringstream str;
str.fill('0');
str.flags(std::ios::dec | std::ios::right);
using std::setw;
str << y << '-' << months[mn-1] << '-' << setw(2) << d << ' '
<< setw(2) << h.count() << ':'
<< setw(2) << m.count() << ':'
<< setw(2) << s.count();
return str.str();
}
std::string
to_string(NetClock::time_point tp)
{
using namespace std::chrono;
return to_string(system_clock::time_point{tp.time_since_epoch() + 946684800s});
}
} // ripple