ARCHIVE/rippled
1
0
Fork 0
mirror of https://github.com/XRPLF/rippled.git synced 2026-07-24 07:30:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor(telemetry): consensus + peer-receive spans use spanContext()/freshRoot, drop detach

Browse Source 
With unscoped SpanGuard + spanContext() (own-span capture), the consensus
round/establish/accept sites just capture and drop the detach dance;
openSpan + peer proposal/validation receive are thread-free handoffs (no
detach); rootSpan->freshRoot.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Pratik Mankawde
2026-07-21 22:09:36 +01:00
parent 30ee25a28b
commit 7fc145ebd2
4 changed files with 52 additions and 68 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
src/xrpld/app/consensus/RCLConsensus.cpp
View File

@@ -230,8 +230,8 @@ RCLConsensus::Adaptor::share(RCLCxTx const& tx)
void
RCLConsensus::Adaptor::propose(RCLCxPeerPos::Proposal const& proposal)
{
// Child of the round span via its captured context (roundSpan_ is detached,
// so it is no longer the thread's ambient parent).
// Child of the round span via its captured context (roundSpan_ is a
// thread-free SpanGuard, so parent explicitly via its context).
auto span = telemetry::SpanGuard::childSpan(
telemetry::consensus::span::proposalSend, roundSpanContext_);
span.setAttribute(
@@ -347,8 +347,8 @@ RCLConsensus::Adaptor::onClose(
{
namespace cs = telemetry::consensus::span;
// Child of the round span via its captured context (roundSpan_ is detached,
// so it is no longer the thread's ambient parent).
// Child of the round span via its captured context (roundSpan_ is a
// thread-free SpanGuard, so parent explicitly via its context).
auto span = telemetry::SpanGuard::childSpan(cs::ledgerClose, roundSpanContext_);
span.setAttribute(cs::attr::ledgerSeq, static_cast<int64_t>(ledger.ledger->header().seq) + 1);
span.setAttribute(cs::attr::mode, toDisplayString(mode).c_str());
@@ -534,14 +534,12 @@ RCLConsensus::Adaptor::makeAcceptSpan(Result const& result)
// "validation follows acceptance" causal model).
if (*span)
{
acceptSpanContext_ = span->captureContext();
// The accept span is moved into the JtAccept worker (onAccept) and
// ended there. Detach its Scope now, on this thread, AFTER the capture
// above, so it is not popped on the wrong thread. accept.apply parents
// via acceptSpanContext_ (captured above), so no ambient child is
// orphaned on either the sync (onForceAccept) or async (onAccept) path.
// detachInPlace rebuilds the shared_ptr (move-assign is deleted).
span = telemetry::detachInPlace(std::move(span));
// span is a thread-free SpanGuard handed to the JtAccept worker
// (onAccept), which ends it there. spanContext() captures the guard's
// own span, so accept.apply parents via acceptSpanContext_ regardless
// of which thread runs the sync (onForceAccept) or async (onAccept)
// path -- no scope work is needed.
acceptSpanContext_ = span->spanContext();
}
return span;
}
@@ -584,11 +582,10 @@ RCLConsensus::Adaptor::doAccept(
closeTimeCorrect = true;
}
// Parent accept.apply via the captured accept context (acceptSpanContext_)
// rather than the accept guard's ambient Scope: the accept span is detached
// (its Scope no longer sits on this thread), so an explicit context is used
// for both the sync (onForceAccept) and async (onAccept) paths. Falls back
// to the round context if the accept span was null.
// Parent accept.apply via the captured accept context (acceptSpanContext_):
// the accept span is a thread-free SpanGuard, so an explicit context is
// used for both the sync (onForceAccept) and async (onAccept) paths. Falls
// back to the round context if the accept span was null.
auto doAcceptSpan = acceptSpanContext_.isValid()
? telemetry::SpanGuard::childSpan(cs::acceptApply, acceptSpanContext_)
: telemetry::SpanGuard::childSpan(cs::acceptApply, roundSpanContext_);
@@ -1075,9 +1072,10 @@ RCLConsensus::Adaptor::onModeChange(ConsensusMode before, ConsensusMode after)
{
namespace cs = telemetry::consensus::span;
// Child of the round span via its captured context (roundSpan_ is detached,
// so it is no longer the thread's ambient parent). A mode change outside a
// round leaves roundSpanContext_ invalid, yielding a null guard (no-op).
// Child of the round span via its captured context (roundSpan_ is a
// thread-free SpanGuard, so parent explicitly via its context). A mode
// change outside a round leaves roundSpanContext_ invalid, yielding a null
// guard (no-op).
auto span = telemetry::SpanGuard::childSpan(cs::modeChange, roundSpanContext_);
span.setAttribute(cs::attr::modeOld, toDisplayString(before).c_str());
span.setAttribute(cs::attr::modeNew, toDisplayString(after).c_str());
@@ -1323,14 +1321,11 @@ RCLConsensus::Adaptor::startRoundTracing(RCLCxLedger const& prevLgr)
roundSpan_->addEvent(cs::event::phaseOpen);
roundSpanContext_ = roundSpan_->captureContext();
// roundSpanContext_ (captured above) is the durable handle that child
// spans on other threads link to. The guard itself is reset() on a
// different worker than it was emplaced on, so detach its Scope now --
// AFTER the context capture -- to avoid a wrong-thread Scope pop.
// detachInPlace re-emplaces it (move-assignment is deleted).
telemetry::detachInPlace(roundSpan_);
// roundSpanContext_ is the durable handle that child spans on other
// threads parent to. roundSpan_ is a thread-free SpanGuard that is
// reset() on a different worker than it was emplaced on, so spanContext()
// captures its own span and no scope work is needed.
roundSpanContext_ = roundSpan_->spanContext();
}
std::optional<telemetry::SpanGuard>

49
src/xrpld/consensus/Consensus.h
View File

@@ -650,24 +650,21 @@ private:
/**
* Span for the establish phase of consensus.
* Created when the ledger closes and we enter phaseEstablish;
* cleared (ended) when consensus is reached. Stored detached (its Scope
* is stripped right after its context is captured) because it is emplaced
* and reset on different job workers.
* cleared (ended) when consensus is reached. A thread-free SpanGuard,
* emplaced and reset() on different job workers.
*/
std::optional<xrpl::telemetry::SpanGuard> establishSpan_;
/**
* Captured context of establishSpan_, snapshotted while it was still
* scoped. Same-thread children (update_positions, check) build from this
* explicit context instead of the (now detached) ambient establish span.
* Captured context of establishSpan_ (its own span context). Children
* (update_positions, check) build from this explicit context.
*/
xrpl::telemetry::SpanContext establishSpanContext_;
/**
* Span for the open phase of consensus.
* Created in startRoundInternal(); cleared (ended) in closeLedger().
* Stored detached: emplaced and reset on different job workers;
* detached() prevents wrong-thread Scope pop.
* A thread-free SpanGuard, emplaced and reset() on different job workers.
*/
std::optional<xrpl::telemetry::SpanGuard> openSpan_;
@@ -779,17 +776,14 @@ Consensus<Adaptor>::startRoundInternal(
// early-returns when establishSpan_ is populated).
establishSpan_.reset();
establishSpanContext_ = telemetry::SpanContext{};
// Child of the round span via its captured context: the round span is
// detached (no longer the thread's ambient parent), so parent phase.open
// explicitly under roundSpanContext_ rather than the ambient stack. An
// invalid round context (round span not yet created) yields a null guard.
// Detached in turn: emplaced here on one job worker and reset() on
// another, so strip the thread-local Scope to avoid a wrong-thread pop.
// No same-thread child span nests under openSpan_.
// Child of the round span via its captured context: parent phase.open
// explicitly under roundSpanContext_. An invalid round context (round span
// not yet created) yields a null guard. openSpan_ is a thread-free
// SpanGuard emplaced here on one job worker and reset() on another; no
// scope to strip.
openSpan_.emplace(
telemetry::SpanGuard::childSpan(
telemetry::consensus::span::phaseOpen, adaptor_.roundSpanContext())
.detached());
telemetry::consensus::span::phaseOpen, adaptor_.roundSpanContext()));
// On the Recovered path, fire phase.open here because startRoundTracing
// (which fires it for the Initial path) is not called on re-entry. On
// the Initial path this is a no-op because the round span hasn't been
@@ -1631,8 +1625,8 @@ Consensus<Adaptor>::updateOurPositions(std::unique_ptr<std::stringstream> const&
// NOLINTBEGIN(bugprone-unchecked-optional-access) assert above
using namespace telemetry;
// Child of the establish span via its captured context (establishSpan_ is
// detached, so it is no longer the thread's ambient parent). Null context
// (establish not started) yields a null guard, same as before.
// a thread-free SpanGuard, so parent explicitly via its context). Null
// context (establish not started) yields a null guard, same as before.
auto span = SpanGuard::childSpan(consensus::span::updatePositions, establishSpanContext_);
span.setAttribute(
consensus::span::attr::convergePercent, static_cast<int64_t>(convergePercent_));
@@ -1841,7 +1835,7 @@ Consensus<Adaptor>::haveConsensus(std::unique_ptr<std::stringstream> const& clog
// NOLINTBEGIN(bugprone-unchecked-optional-access) assert above
using namespace telemetry;
// Child of the establish span via its captured context (establishSpan_ is
// detached, so it is no longer the thread's ambient parent).
// a thread-free SpanGuard, so parent explicitly via its context).
auto span = SpanGuard::childSpan(consensus::span::check, establishSpanContext_);
// CHECKME: should possibly count unacquired TX sets as disagreeing
@@ -2101,22 +2095,19 @@ Consensus<Adaptor>::startEstablishTracing()
{
if (establishSpan_)
return;
// Child of the round span via its captured context: the round span is
// detached (no longer the thread's ambient parent), so parent establish
// Child of the round span via its captured context: parent establish
// explicitly under roundSpanContext_. An invalid round context (round span
// not yet created) yields a null guard.
establishSpan_.emplace(
telemetry::SpanGuard::childSpan(
telemetry::consensus::span::establish, adaptor_.roundSpanContext()));
// Capture the establish context while the guard is still scoped, then
// detach. Same-thread children (update_positions, check) link to this
// captured context explicitly instead of relying on establishSpan_ being
// the ambient parent -- the guard is reset() on a different worker than it
// is emplaced on, so it must not keep a thread-local Scope.
// Capture the establish context; children (update_positions, check) parent
// to it explicitly via establishSpanContext_. establishSpan_ is a
// thread-free SpanGuard reset() on a different worker than it is emplaced
// on -- no scope work.
if (*establishSpan_)
{
establishSpanContext_ = establishSpan_->captureContext();
telemetry::detachInPlace(establishSpan_);
establishSpanContext_ = establishSpan_->spanContext();
}
}

14
src/xrpld/overlay/detail/PeerImp.cpp
View File

@@ -1849,10 +1849,9 @@ PeerImp::onMessage(std::shared_ptr<protocol::TMProposeSet> const& m)
// Create a receive span that links to the sender's trace context
// (if propagated). shared_ptr keeps it alive across the job boundary.
// Detach the guard's Scope on this peer thread so it is not popped on the
// job worker thread (which would leak this thread's context stack).
auto span =
std::make_shared<telemetry::SpanGuard>(telemetry::proposalReceiveSpan(set).detached());
// The receive span is a thread-free SpanGuard handed to the job worker;
// no scope to strip.
auto span = std::make_shared<telemetry::SpanGuard>(telemetry::proposalReceiveSpan(set));
span->setAttribute(telemetry::consensus::span::attr::proposalTrusted, isTrusted);
span->setAttribute(
telemetry::consensus::span::attr::round, static_cast<int64_t>(set.proposeseq()));
@@ -2447,10 +2446,9 @@ PeerImp::onMessage(std::shared_ptr<protocol::TMValidation> const& m)
// Create a receive span that links to the sender's trace context
// (if propagated). shared_ptr keeps it alive across the job boundary.
// Detach the guard's Scope on this peer thread so it is not popped on
// the job worker thread (which would leak this thread's context stack).
auto span =
std::make_shared<telemetry::SpanGuard>(telemetry::validationReceiveSpan(*m).detached());
// The receive span is a thread-free SpanGuard handed to the job worker;
// no scope to strip.
auto span = std::make_shared<telemetry::SpanGuard>(telemetry::validationReceiveSpan(*m));
span->setAttribute(telemetry::consensus::span::attr::validationTrusted, isTrusted);
if (val->isFieldPresent(sfLedgerSequence))
{

6
src/xrpld/telemetry/ConsensusReceiveTracing.h
View File

@@ -19,7 +19,7 @@
* +--- has trace_context? ----+
* | yes | no
* v v
* SpanGuard::hashSpan() with SpanGuard::rootSpan()
* SpanGuard::hashSpan() with SpanGuard::freshRoot()
* extracted parent context (fresh trace root)
*
* When XRPL_ENABLE_TELEMETRY is not defined, the functions return
@@ -85,7 +85,7 @@ proposalReceiveSpan([[maybe_unused]] protocol::TMProposeSet const& msg)
}
#endif
// No propagated context — start a fresh trace root (not an ambient child).
return SpanGuard::rootSpan(
return SpanGuard::freshRoot(
TraceCategory::Consensus, seg::consensus, consensus::span::op::proposalReceive);
}
@@ -123,7 +123,7 @@ validationReceiveSpan([[maybe_unused]] protocol::TMValidation const& msg)
}
#endif
// No propagated context — start a fresh trace root (not an ambient child).
return SpanGuard::rootSpan(
return SpanGuard::freshRoot(
TraceCategory::Consensus, seg::consensus, consensus::span::op::validationReceive);
}