memgraph/src/memgraph_bolt.cpp

181 lines
6.1 KiB
C++
Raw Normal View History

#include <experimental/filesystem>
#include <iostream>
#include <gflags/gflags.h>
#include <glog/logging.h>
#include "communication/bolt/v1/session.hpp"
#include "communication/server.hpp"
#include "io/network/network_endpoint.hpp"
#include "io/network/network_error.hpp"
#include "io/network/socket.hpp"
#include "utils/flag_validation.hpp"
#include "utils/scheduler.hpp"
#include "utils/signals/handler.hpp"
#include "utils/stacktrace.hpp"
#include "utils/sysinfo/memory.hpp"
2017-01-13 17:47:17 +08:00
#include "utils/terminate_handler.hpp"
#include "version.hpp"
namespace fs = std::experimental::filesystem;
using endpoint_t = io::network::NetworkEndpoint;
using socket_t = io::network::Socket;
using session_t = communication::bolt::Session<socket_t>;
using result_stream_t =
communication::bolt::ResultStream<communication::bolt::Encoder<
communication::bolt::ChunkedEncoderBuffer<socket_t>>>;
using session_data_t = communication::bolt::SessionData<result_stream_t>;
using bolt_server_t =
communication::Server<session_t, socket_t, session_data_t>;
DEFINE_string(interface, "0.0.0.0",
"Communication interface on which to listen.");
DEFINE_string(port, "7687", "Communication port on which to listen.");
DEFINE_VALIDATED_int32(num_workers,
std::max(std::thread::hardware_concurrency(), 1U),
"Number of workers", FLAG_IN_RANGE(1, INT32_MAX));
DEFINE_string(log_file, "",
"Path to where the log should be stored.");
DEFINE_string(log_link_basename, "",
"Basename used for symlink creation to the last log file.");
DEFINE_uint64(memory_warning_threshold, 1024,
"Memory warning treshold, in MB. If Memgraph detects there is "
"less available RAM available it will log a warning. Set to 0 to "
"disable.");
// Load flags in this order, the last one has the highest priority:
// 1) /etc/memgraph/memgraph.conf
// 2) ~/.memgraph/config
// 3) env - MEMGRAPH_CONFIG
// 4) command line flags
void load_config(int &argc, char **&argv) {
std::vector<fs::path> configs = {fs::path("/etc/memgraph/memgraph.conf")};
if (getenv("HOME") != nullptr)
configs.emplace_back(fs::path(getenv("HOME")) /
fs::path(".memgraph/config"));
{
auto memgraph_config = getenv("MEMGRAPH_CONFIG");
if (memgraph_config != nullptr) {
auto path = fs::path(memgraph_config);
CHECK(fs::exists(path))
<< "MEMGRAPH_CONFIG environment variable set to nonexisting path: "
<< path.generic_string();
configs.emplace_back(path);
}
}
std::vector<std::string> flagfile_arguments;
for (const auto &config : configs)
if (fs::exists(config)) {
flagfile_arguments.emplace_back(
std::string("--flag-file=" + config.generic_string()));
}
int custom_argc = static_cast<int>(flagfile_arguments.size()) + 1;
char **custom_argv = new char *[custom_argc];
custom_argv[0] = strdup(std::string("memgraph").c_str());
for (int i = 0; i < static_cast<int>(flagfile_arguments.size()); ++i) {
custom_argv[i + 1] = strdup(flagfile_arguments[i].c_str());
}
// setup flags from config flags
gflags::ParseCommandLineFlags(&custom_argc, &custom_argv, false);
// unconsumed arguments have to be freed to avoid memory leak since they are
// strdup-ed.
for (int i = 0; i < custom_argc; ++i) free(custom_argv[i]);
delete[] custom_argv;
// setup flags from command line
gflags::ParseCommandLineFlags(&argc, &argv, true);
}
int main(int argc, char **argv) {
google::SetUsageMessage("Memgraph database server");
gflags::SetVersionString(version_string);
load_config(argc, argv);
google::InitGoogleLogging(argv[0]);
google::SetLogDestination(google::INFO, FLAGS_log_file.c_str());
google::SetLogSymlink(google::INFO, FLAGS_log_link_basename.c_str());
// Unhandled exception handler init.
std::set_terminate(&terminate_handler);
// Signal handling init.
SignalHandler::register_handler(Signal::SegmentationFault, []() {
Fix errors with handling SIGSEGV and SIGABRT Summary: There were a couple of issues with handling the above 2 signals. 1) Calling `std::exit` from a signal handler is undefined behaviour. The only defined way for a signal handler to stop the program is calling one of: `std::_Exit`, `std::abort` or `std::quick_exit`. Neither of them will completely clean the resources, so a clean exit is not possible. Since SIGSEGV and SIGABRT happen in extraordinary circumstances that we wish to debug 99% of the time, it makes sense to generate a core dump which can be inspected by a debugger. Of the 3 termination functions, only `std::abort` will generate a core dump, so it makes sense to use that to stop the program. Also, since we are now aborting as is the default behaviour on SIGSEGV and SIGABRT, it becomes questionable why have a custom handler at all. 2) Raising an exception inside a signal handler is undefined behaviour Although the handler by itself does not raise an exception, it is possible for the logging facility to raise one. This is a real case when logging a stack trace in particular. Stack trace is generated by creating a string "<function name> <line location>". It is possible that a function name will contain '{}' somewhere inside. This is usually the case with anonymous functions. The generated string is then passed to logging, which uses the `fmt` library to fill '{}' with remaining arguments. Since only a single argument (the stack trace string) is passed for formatting, naturally the `fmt::format` throws an exception, that it is missing a format argument. We could provide an overload which takes a single string, but that defeats the purpose of `fmt::format` raising an exception in regular code if we forget to pass an argument. Another solution is to escape the whole stack trace string, so it is valid for formatting, but this only complicates the handler even further. The simplest solution is to send the stack trace to `stderr` and avoid logging altogether. Simplify Shutdown, so it can be used in a signal handler Reviewers: florijan, mferencevic, buda, mislav.bradac Reviewed By: mferencevic, buda Subscribers: pullbot Differential Revision: https://phabricator.memgraph.io/D474
2017-06-14 22:37:23 +08:00
// Log that we got SIGSEGV and abort the program, because returning from
// SIGSEGV handler is undefined behaviour.
std::cerr << "SegmentationFault signal raised" << std::endl;
std::abort(); // This will continue into our SIGABRT handler.
});
SignalHandler::register_handler(Signal::Abort, []() {
Fix errors with handling SIGSEGV and SIGABRT Summary: There were a couple of issues with handling the above 2 signals. 1) Calling `std::exit` from a signal handler is undefined behaviour. The only defined way for a signal handler to stop the program is calling one of: `std::_Exit`, `std::abort` or `std::quick_exit`. Neither of them will completely clean the resources, so a clean exit is not possible. Since SIGSEGV and SIGABRT happen in extraordinary circumstances that we wish to debug 99% of the time, it makes sense to generate a core dump which can be inspected by a debugger. Of the 3 termination functions, only `std::abort` will generate a core dump, so it makes sense to use that to stop the program. Also, since we are now aborting as is the default behaviour on SIGSEGV and SIGABRT, it becomes questionable why have a custom handler at all. 2) Raising an exception inside a signal handler is undefined behaviour Although the handler by itself does not raise an exception, it is possible for the logging facility to raise one. This is a real case when logging a stack trace in particular. Stack trace is generated by creating a string "<function name> <line location>". It is possible that a function name will contain '{}' somewhere inside. This is usually the case with anonymous functions. The generated string is then passed to logging, which uses the `fmt` library to fill '{}' with remaining arguments. Since only a single argument (the stack trace string) is passed for formatting, naturally the `fmt::format` throws an exception, that it is missing a format argument. We could provide an overload which takes a single string, but that defeats the purpose of `fmt::format` raising an exception in regular code if we forget to pass an argument. Another solution is to escape the whole stack trace string, so it is valid for formatting, but this only complicates the handler even further. The simplest solution is to send the stack trace to `stderr` and avoid logging altogether. Simplify Shutdown, so it can be used in a signal handler Reviewers: florijan, mferencevic, buda, mislav.bradac Reviewed By: mferencevic, buda Subscribers: pullbot Differential Revision: https://phabricator.memgraph.io/D474
2017-06-14 22:37:23 +08:00
// Log the stacktrace and let the abort continue.
Stacktrace stacktrace;
std::cerr << "Abort signal raised" << std::endl
<< stacktrace.dump() << std::endl;
});
// Initialize bolt session data (Dbms and Interpreter).
session_data_t session_data;
// Initialize endpoint.
endpoint_t endpoint;
try {
endpoint = endpoint_t(FLAGS_interface, FLAGS_port);
} catch (io::network::NetworkEndpointException &e) {
LOG(FATAL) << e.what();
}
// Initialize socket.
socket_t socket;
if (!socket.Bind(endpoint)) {
LOG(FATAL) << "Cannot bind to socket on " << FLAGS_interface << " at "
<< FLAGS_port;
}
if (!socket.SetNonBlocking()) {
LOG(FATAL) << "Cannot set socket to non blocking!";
}
if (!socket.Listen(1024)) {
LOG(FATAL) << "Cannot listen on socket!";
}
// Initialize server.
bolt_server_t server(std::move(socket), session_data);
// register SIGTERM handler
SignalHandler::register_handler(Signal::Terminate,
[&server, &session_data]() {
server.Shutdown();
session_data.dbms.Shutdown();
});
// register SIGINT handler
SignalHandler::register_handler(Signal::Interupt, [&server, &session_data]() {
server.Shutdown();
session_data.dbms.Shutdown();
});
// Start memory warning logger.
Scheduler mem_log_scheduler;
if (FLAGS_memory_warning_threshold > 0) {
mem_log_scheduler.Run(std::chrono::seconds(3), [] {
auto free_ram_mb = utils::AvailableMem() / 1024;
if (free_ram_mb < FLAGS_memory_warning_threshold)
LOG(WARNING) << "Running out of available RAM, only " << free_ram_mb
<< " MB left.";
});
}
// Start worker threads.
server.Start(FLAGS_num_workers);
return 0;
}