mirror of
https://github.com/rustfs/rustfs.git
synced 2026-08-09 14:49:25 +00:00
70e1d79dfd
* feat: replace jemalloc with mimalloc * docs: record allocator rounds5 retest * fix(replication): satisfy clippy unwrap lints * docs: keep allocator migration plan local only * feat(profiling): rely on pyroscope cpu profiling * refactor(profiling): centralize unsupported pprof responses * chore(deps): update s3s revision
218 lines
8.8 KiB
Rust
218 lines
8.8 KiB
Rust
// Copyright 2024 RustFS Team
|
|
//
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
// you may not use this file except in compliance with the License.
|
|
// You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
|
|
//! RAII guard for OpenTelemetry provider lifecycle management.
|
|
//!
|
|
//! [`OtelGuard`] holds all runtime resources created during telemetry
|
|
//! initialisation. Dropping it triggers an ordered shutdown:
|
|
//!
|
|
//! 1. Tracer provider — flushes pending spans.
|
|
//! 2. Meter provider — flushes pending metrics.
|
|
//! 3. Logger provider — flushes pending log records.
|
|
//! 4. Profiling agent — flushes pending profiles.
|
|
//! 5. Cleanup task — aborted to prevent lingering background work.
|
|
//! 6. Tracing worker guard — flushes buffered log lines written by
|
|
//! `tracing_appender`.
|
|
//! 7. Stdout worker guard — flushes buffered log lines written to stdout.
|
|
|
|
use opentelemetry_sdk::{logs::SdkLoggerProvider, metrics::SdkMeterProvider, trace::SdkTracerProvider};
|
|
use tracing::{debug, error};
|
|
|
|
#[cfg(all(
|
|
feature = "pyroscope",
|
|
any(target_os = "macos", all(target_os = "linux", target_env = "gnu", target_arch = "x86_64"))
|
|
))]
|
|
pub(crate) type ProfilingAgent = pyroscope::PyroscopeAgent<pyroscope::pyroscope::PyroscopeAgentRunning>;
|
|
#[cfg(not(all(
|
|
feature = "pyroscope",
|
|
any(target_os = "macos", all(target_os = "linux", target_env = "gnu", target_arch = "x86_64"))
|
|
)))]
|
|
pub(crate) type ProfilingAgent = ();
|
|
|
|
const LOG_COMPONENT_OBS: &str = "obs";
|
|
const LOG_SUBSYSTEM_GUARD: &str = "guard";
|
|
const EVENT_OBS_GUARD_SHUTDOWN: &str = "obs_guard_shutdown";
|
|
const STDERR_ERROR_PREFIX: &str = "[ERROR]";
|
|
|
|
fn format_guard_shutdown_stderr_message(resource: &str, error: impl std::fmt::Display) -> String {
|
|
format!("{STDERR_ERROR_PREFIX} observability guard shutdown failed: resource={resource} error={error}")
|
|
}
|
|
|
|
/// RAII guard that owns all active OpenTelemetry providers and the
|
|
/// `tracing_appender` worker guard.
|
|
///
|
|
/// Construct this via the `init_*` functions in [`crate::telemetry`] rather
|
|
/// than directly. The guard must be kept alive for the entire duration of the
|
|
/// application — once dropped, all telemetry pipelines are shut down.
|
|
pub struct OtelGuard {
|
|
/// Optional tracer provider for distributed tracing.
|
|
pub(crate) tracer_provider: Option<SdkTracerProvider>,
|
|
/// Optional meter provider for metrics collection.
|
|
pub(crate) meter_provider: Option<SdkMeterProvider>,
|
|
/// Optional logger provider for OTLP log export.
|
|
pub(crate) logger_provider: Option<SdkLoggerProvider>,
|
|
pub(crate) profiling_agent: Option<ProfilingAgent>,
|
|
/// Handle to the background log-cleanup task; aborted on drop.
|
|
pub(crate) cleanup_handle: Option<tokio::task::JoinHandle<()>>,
|
|
/// Worker guard that keeps the non-blocking `tracing_appender` thread
|
|
/// alive. Dropping it blocks until all buffered records are flushed.
|
|
pub(crate) tracing_guard: Option<tracing_appender::non_blocking::WorkerGuard>,
|
|
/// Optional guard for stdout logging; kept separate to allow independent flushing and shutdown.
|
|
pub(crate) stdout_guard: Option<tracing_appender::non_blocking::WorkerGuard>,
|
|
}
|
|
|
|
impl std::fmt::Debug for OtelGuard {
|
|
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
|
|
let mut s = f.debug_struct("OtelGuard");
|
|
s.field("tracer_provider", &self.tracer_provider.is_some())
|
|
.field("meter_provider", &self.meter_provider.is_some())
|
|
.field("logger_provider", &self.logger_provider.is_some())
|
|
.field("profiling_agent", &self.profiling_agent.is_some())
|
|
.field("cleanup_handle", &self.cleanup_handle.is_some())
|
|
.field("tracing_guard", &self.tracing_guard.is_some())
|
|
.field("stdout_guard", &self.stdout_guard.is_some())
|
|
.finish()
|
|
}
|
|
}
|
|
|
|
impl Drop for OtelGuard {
|
|
/// Shut down all telemetry providers in order.
|
|
///
|
|
/// Errors are emitted before tracing resources are dropped so shutdown
|
|
/// diagnostics remain structured and low-noise.
|
|
fn drop(&mut self) {
|
|
if let Some(provider) = self.tracer_provider.take()
|
|
&& let Err(err) = provider.shutdown()
|
|
{
|
|
if tracing::dispatcher::has_been_set() {
|
|
error!(
|
|
event = EVENT_OBS_GUARD_SHUTDOWN,
|
|
component = LOG_COMPONENT_OBS,
|
|
subsystem = LOG_SUBSYSTEM_GUARD,
|
|
resource = "tracer_provider",
|
|
result = "shutdown_failed",
|
|
error = %err,
|
|
"observability guard shutdown failed"
|
|
);
|
|
} else {
|
|
eprintln!("{}", format_guard_shutdown_stderr_message("tracer_provider", err));
|
|
}
|
|
}
|
|
|
|
if let Some(provider) = self.meter_provider.take()
|
|
&& let Err(err) = provider.shutdown()
|
|
{
|
|
if tracing::dispatcher::has_been_set() {
|
|
error!(
|
|
event = EVENT_OBS_GUARD_SHUTDOWN,
|
|
component = LOG_COMPONENT_OBS,
|
|
subsystem = LOG_SUBSYSTEM_GUARD,
|
|
resource = "meter_provider",
|
|
result = "shutdown_failed",
|
|
error = %err,
|
|
"observability guard shutdown failed"
|
|
);
|
|
} else {
|
|
eprintln!("{}", format_guard_shutdown_stderr_message("meter_provider", err));
|
|
}
|
|
}
|
|
|
|
#[cfg(all(
|
|
feature = "pyroscope",
|
|
any(target_os = "macos", all(target_os = "linux", target_env = "gnu", target_arch = "x86_64"))
|
|
))]
|
|
if let Some(agent) = self.profiling_agent.take() {
|
|
match agent.stop() {
|
|
Err(err) => {
|
|
if tracing::dispatcher::has_been_set() {
|
|
error!(
|
|
event = EVENT_OBS_GUARD_SHUTDOWN,
|
|
component = LOG_COMPONENT_OBS,
|
|
subsystem = LOG_SUBSYSTEM_GUARD,
|
|
resource = "profiling_agent",
|
|
result = "shutdown_failed",
|
|
error = %err,
|
|
"observability guard shutdown failed"
|
|
);
|
|
} else {
|
|
eprintln!("{}", format_guard_shutdown_stderr_message("profiling_agent", err));
|
|
}
|
|
}
|
|
Ok(stopped) => {
|
|
stopped.shutdown();
|
|
}
|
|
}
|
|
}
|
|
|
|
if let Some(handle) = self.cleanup_handle.take() {
|
|
debug!(
|
|
event = EVENT_OBS_GUARD_SHUTDOWN,
|
|
component = LOG_COMPONENT_OBS,
|
|
subsystem = LOG_SUBSYSTEM_GUARD,
|
|
resource = "log_cleaner",
|
|
result = "abort_requested",
|
|
"observability guard resource shutdown requested"
|
|
);
|
|
handle.abort();
|
|
}
|
|
|
|
if let Some(provider) = self.logger_provider.take()
|
|
&& let Err(err) = provider.shutdown()
|
|
{
|
|
// After logger shutdown, the OTLP log bridge is no longer a reliable
|
|
// sink for its own failure path, so fall back to stderr.
|
|
// resource = "logger_provider"
|
|
eprintln!("{}", format_guard_shutdown_stderr_message("logger_provider", err));
|
|
}
|
|
|
|
if let Some(guard) = self.tracing_guard.take() {
|
|
debug!(
|
|
event = EVENT_OBS_GUARD_SHUTDOWN,
|
|
component = LOG_COMPONENT_OBS,
|
|
subsystem = LOG_SUBSYSTEM_GUARD,
|
|
resource = "tracing_guard",
|
|
result = "flush_requested",
|
|
"observability guard resource flush requested"
|
|
);
|
|
drop(guard);
|
|
}
|
|
|
|
if let Some(guard) = self.stdout_guard.take() {
|
|
debug!(
|
|
event = EVENT_OBS_GUARD_SHUTDOWN,
|
|
component = LOG_COMPONENT_OBS,
|
|
subsystem = LOG_SUBSYSTEM_GUARD,
|
|
resource = "stdout_guard",
|
|
result = "flush_requested",
|
|
"observability guard resource flush requested"
|
|
);
|
|
drop(guard);
|
|
}
|
|
}
|
|
}
|
|
|
|
#[cfg(test)]
|
|
mod tests {
|
|
use super::*;
|
|
|
|
#[test]
|
|
fn test_guard_shutdown_stderr_message_is_actionable() {
|
|
let message = format_guard_shutdown_stderr_message("logger_provider", "flush failed");
|
|
|
|
assert!(message.starts_with(STDERR_ERROR_PREFIX));
|
|
assert!(message.contains("resource=logger_provider"));
|
|
assert!(message.contains("error=flush failed"));
|
|
}
|
|
}
|