8000 Add more visibility to catch sleeping barber by neunhoef · Pull Request #21742 · arangodb/arangodb · GitHub
[go: up one dir, main page]
Skip to content

Add more visibility to catch sleeping barber #21742

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.

Sign up for GitHub

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

Merged
merged 9 commits into from
May 2, 2025
Merged

Add more visibility to catch sleeping barber #21742

Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
993876f
Timeout in waitFor, logging if hit.
neunhoef Apr 30, 2025
391a439
Diagnose double use.
neunhoef Apr 30, 2025
62bce99
Count number of waiters.
neunhoef Apr 30, 2025
dc3ceb8
Put "ALERT: " in.
neunhoef Apr 30, 2025
7c74780
CHANGELOG.
neunhoef Apr 30, 2025
35b0b79
Fix bug.
neunhoef Apr 30, 2025
c1193f9
Use Logger::AQL topic.
neunhoef Apr 30, 2025
2fa41e5
Move user checking to stopAsyncTasks.
neunhoef Apr 30, 2025
81f9197
Make detection less strict.
neunhoef May 2, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
10000
Diff view
Diff view
Prev Previous commit
Next Next commit
Put "ALERT: " in.
  • Loading branch information
@neunhoef
neunhoef committed Apr 30, 2025
commit dc3ceb8c9807c34ef0816a8673070d0521f93ad8
21 changes: 14 additions & 7 deletions arangod/Aql/ExecutionBlockImpl.tpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -353,14 +353,15 @@ ExecutionBlockImpl<Executor>::~ExecutionBlockImpl() {
uint64_t userCount = _numberOfUsers.fetch_add(1);
if (userCount > 0) {
LOG_TOPIC("52637", WARN, Logger::QUERIES)
<< "Double use of ExecutionBlock detected, stacktrace:";
<< "ALERT: Double use of ExecutionBlock detected, stacktrace:";
CrashHandler::logBacktrace();
_logStacktrace.store(true, std::memory_order_relaxed);
}
auto guard = scopeGuard([&]() noexcept {
uint64_t userCount = _numberOfUsers.fetch_sub(1);
if (_logStacktrace.load(std::memory_order_relaxed)) {
LOG_TOPIC("52638", WARN, Logger::QUERIES) << "Found _logStacktrace:";
LOG_TOPIC("52638", WARN, Logger::QUERIES)
<< "ALERT: Found _logStacktrace:";
CrashHandler::logBacktrace();
if (userCount == 0) {
_logStacktrace.store(false, std::memory_order_relaxed);
Expand Down Expand Up @@ -517,14 +518,15 @@ ExecutionBlockImpl<Executor>::execute(AqlCallStack const& stack) {
uint64_t userCount = _numberOfUsers.fetch_add(1);
if (userCount > 0) {
LOG_TOPIC("52635", WARN, Logger::QUERIES)
<< "Double use of ExecutionBlock detected, stacktrace:";
<< "ALERT: Double use of ExecutionBlock detected, stacktrace:";
CrashHandler::logBacktrace();
_logStacktrace.store(true, std::memory_order_relaxed);
}
auto waechter = scopeGuard([&]() noexcept {
uint64_t userCount = _numberOfUsers.fetch_sub(1);
if (_logStacktrace.load(std::memory_order_relaxed)) {
LOG_TOPIC("52636", WARN, Logger::QUERIES) << "Found _logStacktrace:";
LOG_TOPIC("52636", WARN, Logger::QUERIES)
<< "ALERT: Found _logStacktrace:";
CrashHandler::logBacktrace();
if (userCount == 0) {
_logStacktrace.store(false, std::memory_order_relaxed);
Expand Down Expand Up @@ -2620,7 +2622,7 @@ void ExecutionBlockImpl<Executor>::PrefetchTask::waitFor() const noexcept {
uint64_t count = _numberWaiters.fetch_add(1, std::memory_order_relaxed);
if (count > 0) {
LOG_TOPIC("62515", WARN, Logger::QUERIES)
<< "Detected two waiters for a PrefetchTask, stacktrace:";
<< "ALERT: Detected two waiters for a PrefetchTask, stacktrace:";
CrashHandler::logBacktrace();
_logStacktrace.store(true, std::memory_order_relaxed);
}
Expand All @@ -2630,14 +2632,19 @@ void ExecutionBlockImpl<Executor>::PrefetchTask::waitFor() const noexcept {
std::cv_status s = _bell.wait_for(guard, std::chrono::milliseconds(1000));
if (s == std::cv_status::timeout) {
auto state = _state.load(std::memory_order_relaxed);
// We put "ALERT: " in if the status is not "InProgress", since
// this is the only one we expect when a timeout occurs!
std::string_view alerting =
state.status == Status::InProgress ? "" : "ALERT: ";
LOG_TOPIC("62514", WARN, Logger::QUERIES)
<< "Have waited for a second on an async prefetch task, state is "
<< "ALERT: Have waited for a second on an async prefetch task, state "
"is "
<< (int)state.status << " abandoned: " << state.abandoned;
}
}
count = _numberWaiters.fetch_sub(0);
if (_logStacktrace.load(std::memory_order_relaxed) == true) {
LOG_TOPIC("62516", WARN, Logger::QUERIES) << "Found logStacktrace:";
LOG_TOPIC("62516", WARN, Logger::QUERIES) << "ALERT: Found logStacktrace:";
CrashHandler::logBacktrace();
if (count == 0) {
_logStacktrace.store(false, std::memory_order_relaxed);
Expand Down
0