ClickHouse/src/Storages/System/StorageSystemStackTrace.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

374 lines
14 KiB
C++
Raw Normal View History

2019-12-23 18:56:57 +00:00
#ifdef OS_LINUX /// Because of 'sigqueue' functions and RT signals.
#include <csignal>
2019-12-22 20:17:16 +00:00
#include <poll.h>
#include <mutex>
#include <filesystem>
#include <unordered_map>
2021-10-02 07:13:14 +00:00
#include <base/scope_guard.h>
#include <Storages/System/StorageSystemStackTrace.h>
#include <Storages/VirtualColumnUtils.h>
#include <Columns/ColumnString.h>
#include <Columns/ColumnsNumber.h>
#include <DataTypes/DataTypeString.h>
#include <DataTypes/DataTypesNumber.h>
#include <DataTypes/DataTypeArray.h>
#include <IO/ReadHelpers.h>
2021-05-17 04:44:57 +00:00
#include <IO/ReadBufferFromFile.h>
2019-12-22 20:17:16 +00:00
#include <Common/PipeFDs.h>
2020-07-09 01:00:16 +00:00
#include <Common/CurrentThread.h>
#include <Common/HashTable/Hash.h>
2022-04-27 15:05:45 +00:00
#include <Common/logger_useful.h>
#include <Interpreters/Context.h>
#include <Processors/Sources/SourceFromSingleChunk.h>
#include <QueryPipeline/Pipe.h>
#include <base/getThreadId.h>
namespace DB
{
namespace ErrorCodes
{
extern const int CANNOT_SIGQUEUE;
2019-12-22 20:17:16 +00:00
extern const int CANNOT_MANIPULATE_SIGSET;
extern const int CANNOT_SET_SIGNAL_HANDLER;
extern const int CANNOT_READ_FROM_FILE_DESCRIPTOR;
extern const int LOGICAL_ERROR;
}
namespace
{
2020-11-30 08:53:07 +00:00
// Initialized in StorageSystemStackTrace's ctor and used in signalHandler.
2020-12-01 12:33:42 +00:00
std::atomic<pid_t> expected_pid;
2019-12-22 20:17:16 +00:00
const int sig = SIGRTMIN;
2019-12-22 20:37:29 +00:00
2020-03-23 16:25:21 +00:00
std::atomic<int> sequence_num = 0; /// For messages sent via pipe.
2020-12-15 22:43:07 +00:00
std::atomic<int> data_ready_num = 0;
2020-12-23 01:40:23 +00:00
std::atomic<bool> signal_latch = false; /// Only need for thread sanitizer.
2019-12-23 19:23:54 +00:00
2020-12-15 22:43:07 +00:00
/** Notes:
* Only one query from the table can be processed at the moment of time.
* This is ensured by the mutex in fillData function.
* We obtain information about threads by sending signal and receiving info from the signal handler.
* Information is passed via global variables and pipe is used for signaling.
* Actually we can send all information via pipe, but we read from it with timeout just in case,
* so it's convenient to use is only for signaling.
*/
StackTrace stack_trace{NoCapture{}};
2019-12-22 20:37:29 +00:00
2020-03-09 03:14:24 +00:00
constexpr size_t max_query_id_size = 128;
2019-12-22 20:37:29 +00:00
char query_id_data[max_query_id_size];
size_t query_id_size = 0;
2019-12-22 20:17:16 +00:00
LazyPipeFDs notification_pipe;
2020-12-15 22:43:07 +00:00
void signalHandler(int, siginfo_t * info, void * context)
2019-12-22 20:17:16 +00:00
{
2021-01-12 14:34:50 +00:00
DENY_ALLOCATIONS_IN_SCOPE;
2020-03-01 14:23:43 +00:00
auto saved_errno = errno; /// We must restore previous value of errno in signal handler.
2019-12-22 20:17:16 +00:00
/// In case malicious user is sending signals manually (for unknown reason).
/// If we don't check - it may break our synchronization.
if (info->si_pid != expected_pid)
return;
2019-12-23 19:23:54 +00:00
/// Signal received too late.
2020-12-15 22:43:07 +00:00
int notification_num = info->si_value.sival_int;
if (notification_num != sequence_num.load(std::memory_order_acquire))
2019-12-23 19:23:54 +00:00
return;
2020-12-22 08:16:36 +00:00
bool expected = false;
if (!signal_latch.compare_exchange_strong(expected, true, std::memory_order_acquire))
return;
2019-12-22 20:17:16 +00:00
/// All these methods are signal-safe.
const ucontext_t signal_context = *reinterpret_cast<ucontext_t *>(context);
2020-12-15 22:43:07 +00:00
stack_trace = StackTrace(signal_context);
2023-03-15 21:12:29 +00:00
auto query_id = CurrentThread::getQueryId();
query_id_size = std::min(query_id.size(), max_query_id_size);
if (!query_id.empty())
memcpy(query_id_data, query_id.data(), query_id_size);
2019-12-22 20:37:29 +00:00
2020-12-15 22:43:07 +00:00
/// This is unneeded (because we synchronize through pipe) but makes TSan happy.
2020-12-22 08:16:36 +00:00
data_ready_num.store(notification_num, std::memory_order_release);
2020-12-15 22:43:07 +00:00
2019-12-23 19:23:54 +00:00
ssize_t res = ::write(notification_pipe.fds_rw[1], &notification_num, sizeof(notification_num));
2019-12-23 18:58:42 +00:00
2019-12-22 20:17:16 +00:00
/// We cannot do anything if write failed.
2019-12-23 18:58:42 +00:00
(void)res;
2020-03-01 14:23:43 +00:00
errno = saved_errno;
2020-12-22 08:16:36 +00:00
signal_latch.store(false, std::memory_order_release);
2019-12-22 20:17:16 +00:00
}
2019-12-22 20:37:29 +00:00
/// Wait for data in pipe and read it.
2020-12-15 22:43:07 +00:00
bool wait(int timeout_ms)
2019-12-22 20:17:16 +00:00
{
while (true)
{
2019-12-22 20:17:16 +00:00
int fd = notification_pipe.fds_rw[0];
pollfd poll_fd{fd, POLLIN, 0};
int poll_res = poll(&poll_fd, 1, timeout_ms);
if (poll_res < 0)
{
if (errno == EINTR)
{
--timeout_ms; /// Quite a hacky way to update timeout. Just to make sure we avoid infinite waiting.
if (timeout_ms == 0)
return false;
continue;
}
throwFromErrno("Cannot poll pipe", ErrorCodes::CANNOT_READ_FROM_FILE_DESCRIPTOR);
}
if (poll_res == 0)
return false;
2019-12-23 19:23:54 +00:00
int notification_num = 0;
ssize_t read_res = ::read(fd, &notification_num, sizeof(notification_num));
2019-12-22 20:17:16 +00:00
if (read_res < 0)
{
if (errno == EINTR)
continue;
throwFromErrno("Cannot read from pipe", ErrorCodes::CANNOT_READ_FROM_FILE_DESCRIPTOR);
}
2019-12-23 19:23:54 +00:00
if (read_res == sizeof(notification_num))
{
2020-03-23 16:25:21 +00:00
if (notification_num == sequence_num.load(std::memory_order_relaxed))
2019-12-23 19:23:54 +00:00
return true;
else
continue; /// Drain delayed notifications.
}
throw Exception(ErrorCodes::LOGICAL_ERROR, "Logical error: read wrong number of bytes from pipe");
}
2019-12-22 20:17:16 +00:00
}
ColumnPtr getFilteredThreadIds(ASTPtr query, ContextPtr context)
{
MutableColumnPtr all_thread_ids = ColumnUInt64::create();
std::filesystem::directory_iterator end;
/// There is no better way to enumerate threads in a process other than looking into procfs.
for (std::filesystem::directory_iterator it("/proc/self/task"); it != end; ++it)
{
pid_t tid = parse<pid_t>(it->path().filename());
all_thread_ids->insert(tid);
}
Block block { ColumnWithTypeAndName(std::move(all_thread_ids), std::make_shared<DataTypeUInt64>(), "thread_id") };
VirtualColumnUtils::filterBlockWithQuery(query, block, context);
return block.getByPosition(0).column;
}
using ThreadIdToName = std::unordered_map<UInt64, String, DefaultHash<UInt64>>;
ThreadIdToName getFilteredThreadNames(ASTPtr query, ContextPtr context, const PaddedPODArray<UInt64> & thread_ids)
{
ThreadIdToName tid_to_name;
MutableColumnPtr all_thread_names = ColumnString::create();
for (UInt64 tid : thread_ids)
{
std::filesystem::path thread_name_path = fmt::format("/proc/self/task/{}/comm", tid);
String thread_name;
if (std::filesystem::exists(thread_name_path))
{
constexpr size_t comm_buf_size = 32; /// More than enough for thread name
ReadBufferFromFile comm(thread_name_path.string(), comm_buf_size);
readEscapedStringUntilEOL(thread_name, comm);
comm.close();
}
tid_to_name[tid] = thread_name;
all_thread_names->insert(thread_name);
}
Block block { ColumnWithTypeAndName(std::move(all_thread_names), std::make_shared<DataTypeString>(), "thread_name") };
VirtualColumnUtils::filterBlockWithQuery(query, block, context);
ColumnPtr thread_names = std::move(block.getByPosition(0).column);
std::unordered_set<String> filtered_thread_names;
for (size_t i = 0; i != thread_names->size(); ++i)
{
const auto & thread_name = thread_names->getDataAt(i);
filtered_thread_names.emplace(thread_name);
}
for (auto it = tid_to_name.begin(); it != tid_to_name.end();)
{
if (!filtered_thread_names.contains(it->second))
it = tid_to_name.erase(it);
else
++it;
}
return tid_to_name;
}
2019-12-22 20:17:16 +00:00
}
2020-12-15 22:43:07 +00:00
StorageSystemStackTrace::StorageSystemStackTrace(const StorageID & table_id_)
: IStorage(table_id_)
, log(&Poco::Logger::get("StorageSystemStackTrace"))
2019-12-22 20:17:16 +00:00
{
StorageInMemoryMetadata storage_metadata;
storage_metadata.setColumns(ColumnsDescription({
{ "thread_name", std::make_shared<DataTypeString>() },
{ "thread_id", std::make_shared<DataTypeUInt64>() },
{ "query_id", std::make_shared<DataTypeString>() },
{ "trace", std::make_shared<DataTypeArray>(std::make_shared<DataTypeUInt64>()) },
}, { /* aliases */ }));
setInMemoryMetadata(storage_metadata);
2019-12-22 20:17:16 +00:00
notification_pipe.open();
2019-12-22 20:17:16 +00:00
/// Setup signal handler.
2020-12-01 12:33:42 +00:00
expected_pid = getpid();
2019-12-22 20:17:16 +00:00
struct sigaction sa{};
sa.sa_sigaction = signalHandler;
sa.sa_flags = SA_SIGINFO;
2019-12-22 20:17:16 +00:00
if (sigemptyset(&sa.sa_mask))
throwFromErrno("Cannot set signal handler.", ErrorCodes::CANNOT_MANIPULATE_SIGSET);
2019-12-22 20:17:16 +00:00
if (sigaddset(&sa.sa_mask, sig))
throwFromErrno("Cannot set signal handler.", ErrorCodes::CANNOT_MANIPULATE_SIGSET);
2019-12-22 20:17:16 +00:00
if (sigaction(sig, &sa, nullptr))
throwFromErrno("Cannot set signal handler.", ErrorCodes::CANNOT_SET_SIGNAL_HANDLER);
}
Pipe StorageSystemStackTrace::read(
const Names & column_names,
const StorageSnapshotPtr & storage_snapshot,
SelectQueryInfo & query_info,
ContextPtr context,
QueryProcessingStage::Enum /*processed_stage*/,
const size_t /*max_block_size*/,
const size_t /*num_streams*/)
2019-12-22 20:17:16 +00:00
{
storage_snapshot->check(column_names);
2019-12-22 20:17:16 +00:00
int pipe_read_timeout_ms = static_cast<int>(
context->getSettingsRef().storage_system_stack_trace_pipe_read_timeout_ms.totalMilliseconds());
2019-12-22 20:17:16 +00:00
/// It shouldn't be possible to do concurrent reads from this table.
std::lock_guard lock(mutex);
/// Create a mask of what columns are needed in the result.
NameSet names_set(column_names.begin(), column_names.end());
Block sample_block = storage_snapshot->metadata->getSampleBlock();
bool send_signal = names_set.contains("trace") || names_set.contains("query_id");
bool read_thread_names = names_set.contains("thread_name");
MutableColumns res_columns = sample_block.cloneEmptyColumns();
2019-12-22 20:17:16 +00:00
/// Send a signal to every thread and wait for result.
/// We must wait for every thread one by one sequentially,
/// because there is a limit on number of queued signals in OS and otherwise signals may get lost.
/// Also, non-RT signals are not delivered if previous signal is handled right now (by default; but we use RT signals).
2019-12-22 20:17:16 +00:00
/// Obviously, results for different threads may be out of sync.
ColumnPtr thread_ids = getFilteredThreadIds(query_info.query, context);
const auto & thread_ids_data = assert_cast<const ColumnUInt64 &>(*thread_ids).getData();
ThreadIdToName thread_names;
if (read_thread_names)
thread_names = getFilteredThreadNames(query_info.query, context, thread_ids_data);
2019-12-22 20:17:16 +00:00
for (UInt64 tid : thread_ids_data)
{
size_t res_index = 0;
2021-05-17 04:44:57 +00:00
String thread_name;
if (read_thread_names)
{
if (auto it = thread_names.find(tid); it != thread_names.end())
thread_name = it->second;
else
continue; /// was filtered out by "thread_name" condition
}
2021-05-17 04:44:57 +00:00
if (!send_signal)
2019-12-22 20:17:16 +00:00
{
res_columns[res_index++]->insert(thread_name);
res_columns[res_index++]->insert(tid);
res_columns[res_index++]->insertDefault();
res_columns[res_index++]->insertDefault();
}
else
{
sigval sig_value{};
sig_value.sival_int = sequence_num.load(std::memory_order_acquire);
if (0 != ::sigqueue(static_cast<int>(tid), sig, sig_value))
{
/// The thread may has been already finished.
if (ESRCH == errno)
continue;
2019-12-22 20:17:16 +00:00
throwFromErrno("Cannot send signal with sigqueue", ErrorCodes::CANNOT_SIGQUEUE);
}
2019-12-23 19:23:54 +00:00
/// Just in case we will wait for pipe with timeout. In case signal didn't get processed.
if (send_signal && wait(pipe_read_timeout_ms) && sig_value.sival_int == data_ready_num.load(std::memory_order_acquire))
{
size_t stack_trace_size = stack_trace.getSize();
size_t stack_trace_offset = stack_trace.getOffset();
Array arr;
arr.reserve(stack_trace_size - stack_trace_offset);
for (size_t i = stack_trace_offset; i < stack_trace_size; ++i)
arr.emplace_back(reinterpret_cast<intptr_t>(stack_trace.getFramePointers()[i]));
res_columns[res_index++]->insert(thread_name);
res_columns[res_index++]->insert(tid);
res_columns[res_index++]->insertData(query_id_data, query_id_size);
res_columns[res_index++]->insert(arr);
}
else
{
LOG_DEBUG(log, "Cannot obtain a stack trace for thread {}", tid);
res_columns[res_index++]->insert(thread_name);
res_columns[res_index++]->insert(tid);
res_columns[res_index++]->insertDefault();
res_columns[res_index++]->insertDefault();
}
/// Signed integer overflow is undefined behavior in both C and C++. However, according to
/// C++ standard, Atomic signed integer arithmetic is defined to use two's complement; there
/// are no undefined results. See https://en.cppreference.com/w/cpp/atomic/atomic and
/// http://eel.is/c++draft/atomics.types.generic#atomics.types.int-8
++sequence_num;
}
}
UInt64 num_rows = res_columns.at(0)->size();
Chunk chunk(std::move(res_columns), num_rows);
return Pipe(std::make_shared<SourceFromSingleChunk>(sample_block, std::move(chunk)));
}
}
2019-12-23 18:56:57 +00:00
#endif