Compare commits

..

16 Commits

Author SHA1 Message Date
Ed Hennis
72c700c3e0 Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-28 16:23:38 -04:00
Ed Hennis
4589fcbcfc Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-25 14:45:54 -04:00
Ed Hennis
86d840f53d Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-23 15:56:11 -04:00
Ed Hennis
741b61cdf3 Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-22 23:40:28 -04:00
Ed Hennis
6bb0989c9f Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-22 14:49:12 -04:00
Ed Hennis
9120506613 Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-22 13:10:44 -04:00
Ed Hennis
3b3de96bd4 Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-21 18:48:39 -04:00
Ed Hennis
c9ab6ab25f Merge remote-tracking branch 'XRPLF/develop' into ximinez/fix/validator-cache
* XRPLF/develop:
  chore: Remove empty Taker.h (6984)
  chore: Enable clang-tidy modernize checks (6975)
  ci: Upload clang-tidy git diff (6983)
  fix: Add rounding to Vault invariants (6217) (6955)
2026-04-21 18:46:58 -04:00
Ed Hennis
fb0605cfd3 Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-20 17:49:47 -04:00
Ed Hennis
156553bb5e Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-20 15:45:04 -04:00
Ed Hennis
781b56849b Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-20 11:39:06 -04:00
Ed Hennis
278c02bebb Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-17 18:11:31 -04:00
Ed Hennis
1d6fedf9a2 Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-16 13:44:37 -04:00
Ed Hennis
2e8de499aa Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-15 19:06:29 -04:00
Ed Hennis
0bce3639a6 Merge branch 'develop' into ximinez/fix/validator-cache 2026-04-15 14:28:55 -04:00
Ed Hennis
8f329e3bc6 Use Validator List (VL) cache files in more scenarios
- If any [validator_list_keys] are not available after all
  [validator_list_sites] have had a chance to be queried, then fall
  back to loading cache files. Currently, cache files are only used if
  no sites are defined, or the request to one of them has an error. It
  does not include cases where not enough sites are defined, or if a
  site returns an invalid VL (or something else entirely).
- Resolves #5320
2026-04-13 19:46:06 -04:00
18 changed files with 132 additions and 418 deletions

View File

@@ -1,139 +0,0 @@
//------------------------------------------------------------------------------
/*
This file is part of rippled: https://github.com/ripple/rippled
Copyright (c) 2024 Ripple Labs Inc.
Permission to use, copy, modify, and/or distribute this software for any
purpose with or without fee is hereby granted, provided that the above
copyright notice and this permission notice appear in all copies.
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
ANY SPECIAL , DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
//==============================================================================
#ifndef RIPPLE_BASICS_CANPROCESS_H_INCLUDED
#define RIPPLE_BASICS_CANPROCESS_H_INCLUDED
#include <functional>
#include <mutex>
#include <set>
/** RAII class to check if an Item is already being processed on another thread,
* as indicated by it's presence in a Collection.
*
* If the Item is not in the Collection, it will be added under lock in the
* ctor, and removed under lock in the dtor. The object will be considered
* "usable" and evaluate to `true`.
*
* If the Item is in the Collection, no changes will be made to the collection,
* and the CanProcess object will be considered "unusable".
*
* It's up to the caller to decide what "usable" and "unusable" mean. (e.g.
* Process or skip a block of code, or set a flag.)
*
* The current use is to avoid lock contention that would be involved in
* processing something associated with the Item.
*
* Examples:
*
* void IncomingLedgers::acquireAsync(LedgerHash const& hash, ...)
* {
* if (CanProcess check{acquiresMutex_, pendingAcquires_, hash})
* {
* acquire(hash, ...);
* }
* }
*
* bool
* NetworkOPsImp::recvValidation(
* std::shared_ptr<STValidation> const& val,
* std::string const& source)
* {
* CanProcess check(
* validationsMutex_, pendingValidations_, val->getLedgerHash());
* BypassAccept bypassAccept =
* check ? BypassAccept::no : BypassAccept::yes;
* handleNewValidation(app_, val, source, bypassAccept, m_journal);
* }
*
*/
class CanProcess
{
public:
template <class Mutex, class Collection, class Item>
CanProcess(Mutex& mtx, Collection& collection, Item const& item)
: cleanup_(insert(mtx, collection, item))
{
}
~CanProcess()
{
if (cleanup_)
cleanup_();
}
CanProcess(CanProcess const&) = delete;
CanProcess&
operator=(CanProcess const&) = delete;
explicit
operator bool() const
{
return static_cast<bool>(cleanup_);
}
private:
template <bool useIterator, class Mutex, class Collection, class Item>
std::function<void()>
doInsert(Mutex& mtx, Collection& collection, Item const& item)
{
std::unique_lock<Mutex> lock(mtx);
// TODO: Use structured binding once LLVM 16 is the minimum supported
// version. See also: https://github.com/llvm/llvm-project/issues/48582
// https://github.com/llvm/llvm-project/commit/127bf44385424891eb04cff8e52d3f157fc2cb7c
auto const insertResult = collection.insert(item);
auto const it = insertResult.first;
if (!insertResult.second)
return {};
if constexpr (useIterator)
return [&, it]() {
std::unique_lock<Mutex> lock(mtx);
collection.erase(it);
};
else
return [&]() {
std::unique_lock<Mutex> lock(mtx);
collection.erase(item);
};
}
// Generic insert() function doesn't use iterators because they may get
// invalidated
template <class Mutex, class Collection, class Item>
std::function<void()>
insert(Mutex& mtx, Collection& collection, Item const& item)
{
return doInsert<false>(mtx, collection, item);
}
// Specialize insert() for std::set, which does not invalidate iterators for
// insert and erase
template <class Mutex, class Item>
std::function<void()>
insert(Mutex& mtx, std::set<Item>& collection, Item const& item)
{
return doInsert<true>(mtx, collection, item);
}
// If set, then the item is "usable"
std::function<void()> cleanup_;
};
#endif

View File

@@ -197,7 +197,7 @@ public:
/** Add a suppression peer and get message's relay status.
* Return pair:
* element 1: true if the key is added.
* element 1: true if the peer is added.
* element 2: optional is seated to the relay time point or
* is unseated if has not relayed yet. */
std::pair<bool, std::optional<Stopwatch::time_point>>

View File

@@ -20,6 +20,10 @@ removeTokenOffersWithLimit(
Keylet const& directory,
std::size_t maxDeletableOffers);
/** Returns tesSUCCESS if NFToken has few enough offers that it can be burned */
TER
notTooManyOffers(ReadView const& view, uint256 const& nftokenID);
/** Finds the specified token in the owner's token directory. */
std::optional<STObject>
findToken(ReadView const& view, AccountID const& owner, uint256 const& nftokenID);

View File

@@ -35,8 +35,6 @@ struct LedgerHeader
// If validated is false, it means "not yet validated."
// Once validated is true, it will never be set false at a later time.
// NOTE: If you are accessing this directly, you are probably doing it
// wrong. Use LedgerMaster::isValidated().
// VFALCO TODO Make this not mutable
bool mutable validated = false;
bool accepted = false;

View File

@@ -185,7 +185,7 @@ public:
virtual bool
isFull() = 0;
virtual void
setMode(OperatingMode om, char const* reason) = 0;
setMode(OperatingMode om) = 0;
virtual bool
isBlocked() = 0;
virtual bool

View File

@@ -621,6 +621,33 @@ removeTokenOffersWithLimit(ApplyView& view, Keylet const& directory, std::size_t
return deletedOffersCount;
}
TER
notTooManyOffers(ReadView const& view, uint256 const& nftokenID)
{
std::size_t totalOffers = 0;
{
Dir const buys(view, keylet::nft_buys(nftokenID));
for (auto iter = buys.begin(); iter != buys.end(); iter.next_page())
{
totalOffers += iter.page_size();
if (totalOffers > maxDeletableTokenOfferEntries)
return tefTOO_BIG;
}
}
{
Dir const sells(view, keylet::nft_sells(nftokenID));
for (auto iter = sells.begin(); iter != sells.end(); iter.next_page())
{
totalOffers += iter.page_size();
if (totalOffers > maxDeletableTokenOfferEntries)
return tefTOO_BIG;
}
}
return tesSUCCESS;
}
bool
deleteTokenOffer(ApplyView& view, std::shared_ptr<SLE> const& offer)
{

View File

@@ -130,12 +130,7 @@ public:
}
void
acquireAsync(
JobType type,
std::string const& name,
uint256 const& hash,
std::uint32_t seq,
InboundLedger::Reason reason) override
acquireAsync(uint256 const& hash, std::uint32_t seq, InboundLedger::Reason reason) override
{
}

View File

@@ -1,165 +0,0 @@
//------------------------------------------------------------------------------
/*
This file is part of rippled: https://github.com/ripple/rippled
Copyright (c) 2012-2016 Ripple Labs Inc.
Permission to use, copy, modify, and/or distribute this software for any
purpose with or without fee is hereby granted, provided that the above
copyright notice and this permission notice appear in all copies.
THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
ANY SPECIAL , DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
//==============================================================================
#include <xrpl/basics/CanProcess.h>
#include <xrpl/beast/unit_test.h>
#include <memory>
namespace ripple {
namespace test {
struct CanProcess_test : beast::unit_test::suite
{
template <class Mutex, class Collection, class Item>
void
test(
std::string const& name,
Mutex& mtx,
Collection& collection,
std::vector<Item> const& items)
{
testcase(name);
if (!BEAST_EXPECT(!items.empty()))
return;
if (!BEAST_EXPECT(collection.empty()))
return;
// CanProcess objects can't be copied or moved. To make that easier,
// store shared_ptrs
std::vector<std::shared_ptr<CanProcess>> trackers;
// Fill up the vector with two CanProcess for each Item. The first
// inserts the item into the collection and is "good". The second does
// not and is "bad".
for (int i = 0; i < items.size(); ++i)
{
{
auto const& good =
trackers.emplace_back(std::make_shared<CanProcess>(mtx, collection, items[i]));
BEAST_EXPECT(*good);
}
BEAST_EXPECT(trackers.size() == (2 * i) + 1);
BEAST_EXPECT(collection.size() == i + 1);
{
auto const& bad =
trackers.emplace_back(std::make_shared<CanProcess>(mtx, collection, items[i]));
BEAST_EXPECT(!*bad);
}
BEAST_EXPECT(trackers.size() == 2 * (i + 1));
BEAST_EXPECT(collection.size() == i + 1);
}
BEAST_EXPECT(collection.size() == items.size());
// Now remove the items from the vector<CanProcess> two at a time, and
// try to get another CanProcess for that item.
for (int i = 0; i < items.size(); ++i)
{
// Remove the "bad" one in the second position
// This will have no effect on the collection
{
auto const iter = trackers.begin() + 1;
BEAST_EXPECT(!**iter);
trackers.erase(iter);
}
BEAST_EXPECT(trackers.size() == (2 * items.size()) - 1);
BEAST_EXPECT(collection.size() == items.size());
{
// Append a new "bad" one
auto const& bad =
trackers.emplace_back(std::make_shared<CanProcess>(mtx, collection, items[i]));
BEAST_EXPECT(!*bad);
}
BEAST_EXPECT(trackers.size() == 2 * items.size());
BEAST_EXPECT(collection.size() == items.size());
// Remove the "good" one from the front
{
auto const iter = trackers.begin();
BEAST_EXPECT(**iter);
trackers.erase(iter);
}
BEAST_EXPECT(trackers.size() == (2 * items.size()) - 1);
BEAST_EXPECT(collection.size() == items.size() - 1);
{
// Append a new "good" one
auto const& good =
trackers.emplace_back(std::make_shared<CanProcess>(mtx, collection, items[i]));
BEAST_EXPECT(*good);
}
BEAST_EXPECT(trackers.size() == 2 * items.size());
BEAST_EXPECT(collection.size() == items.size());
}
// Now remove them all two at a time
for (int i = items.size() - 1; i >= 0; --i)
{
// Remove the "bad" one from the front
{
auto const iter = trackers.begin();
BEAST_EXPECT(!**iter);
trackers.erase(iter);
}
BEAST_EXPECT(trackers.size() == (2 * i) + 1);
BEAST_EXPECT(collection.size() == i + 1);
// Remove the "good" one now in front
{
auto const iter = trackers.begin();
BEAST_EXPECT(**iter);
trackers.erase(iter);
}
BEAST_EXPECT(trackers.size() == 2 * i);
BEAST_EXPECT(collection.size() == i);
}
BEAST_EXPECT(trackers.empty());
BEAST_EXPECT(collection.empty());
}
void
run() override
{
{
std::mutex m;
std::set<int> collection;
std::vector<int> const items{1, 2, 3, 4, 5};
test("set of int", m, collection, items);
}
{
std::mutex m;
std::set<std::string> collection;
std::vector<std::string> const items{"one", "two", "three", "four", "five"};
test("set of string", m, collection, items);
}
{
std::mutex m;
std::unordered_set<char> collection;
std::vector<char> const items{'1', '2', '3', '4', '5'};
test("unorderd_set of char", m, collection, items);
}
{
std::mutex m;
std::unordered_set<std::uint64_t> collection;
std::vector<std::uint64_t> const items{100u, 1000u, 150u, 4u, 0u};
test("unordered_set of uint64_t", m, collection, items);
}
}
};
BEAST_DEFINE_TESTSUITE(CanProcess, ripple_basics, ripple);
} // namespace test
} // namespace ripple

View File

@@ -159,8 +159,10 @@ RCLConsensus::Adaptor::acquireLedger(LedgerHash const& hash)
// Tell the ledger acquire system that we need the consensus ledger
acquiringLedger_ = hash;
app_.getInboundLedgers().acquireAsync(
jtADVANCE, "GetConsL1", hash, 0, InboundLedger::Reason::CONSENSUS);
app_.getJobQueue().addJob(jtADVANCE, "GetConsL1", [id = hash, &app = app_, this]() {
JLOG(j_.debug()) << "JOB advanceLedger getConsensusLedger1 started";
app.getInboundLedgers().acquireAsync(id, 0, InboundLedger::Reason::CONSENSUS);
});
}
return std::nullopt;
}
@@ -1050,7 +1052,7 @@ void
RCLConsensus::Adaptor::updateOperatingMode(std::size_t const positions) const
{
if ((positions == 0u) && app_.getOPs().isFull())
app_.getOPs().setMode(OperatingMode::CONNECTED, "updateOperatingMode: no positions");
app_.getOPs().setMode(OperatingMode::CONNECTED);
}
void

View File

@@ -128,8 +128,12 @@ RCLValidationsAdaptor::acquire(LedgerHash const& hash)
{
JLOG(j_.warn()) << "Need validated ledger for preferred ledger analysis " << hash;
app_.getInboundLedgers().acquireAsync(
jtADVANCE, "GetConsL2", hash, 0, InboundLedger::Reason::CONSENSUS);
Application* pApp = &app_;
app_.getJobQueue().addJob(jtADVANCE, "GetConsL2", [pApp, hash, this]() {
JLOG(j_.debug()) << "JOB advanceLedger getConsensusLedger2 started";
pApp->getInboundLedgers().acquireAsync(hash, 0, InboundLedger::Reason::CONSENSUS);
});
return std::nullopt;
}

View File

@@ -26,12 +26,7 @@ public:
// Queue. TODO review whether all callers of acquire() can use this
// instead. Inbound ledger acquisition is asynchronous anyway.
virtual void
acquireAsync(
JobType type,
std::string const& name,
uint256 const& hash,
std::uint32_t seq,
InboundLedger::Reason reason) = 0;
acquireAsync(uint256 const& hash, std::uint32_t seq, InboundLedger::Reason reason) = 0;
virtual std::shared_ptr<InboundLedger>
find(LedgerHash const& hash) = 0;

View File

@@ -385,14 +385,7 @@ InboundLedger::onTimer(bool wasProgress, ScopedLockType&)
if (!wasProgress)
{
if (checkLocal())
{
// Done. Something else (probably consensus) built the ledger
// locally while waiting for data (or possibly before requesting)
XRPL_ASSERT(isDone(), "ripple::InboundLedger::onTimer : done");
JLOG(journal_.info()) << "Finished while waiting " << hash_;
return;
}
checkLocal();
mByHash = true;

View File

@@ -6,7 +6,6 @@
#include <xrpld/overlay/PeerSet.h>
#include <xrpl/basics/Blob.h>
#include <xrpl/basics/CanProcess.h>
#include <xrpl/basics/DecayingSample.h>
#include <xrpl/basics/Log.h>
#include <xrpl/basics/Slice.h>
@@ -84,15 +83,12 @@ public:
(reason != InboundLedger::Reason::CONSENSUS))
return {};
std::stringstream ss;
bool isNew = true;
std::shared_ptr<InboundLedger> inbound;
{
ScopedLockType sl(mLock);
if (stopping_)
{
JLOG(j_.debug()) << "Abort(stopping): " << ss.str();
return {};
}
@@ -111,61 +107,47 @@ public:
++mCounter;
}
}
ss << " IsNew: " << (isNew ? "true" : "false");
if (inbound->isFailed())
{
JLOG(j_.debug()) << "Abort(failed): " << ss.str();
return {};
}
if (!isNew)
inbound->update(seq);
if (!inbound->isComplete())
{
JLOG(j_.debug()) << "InProgress: " << ss.str();
return {};
}
JLOG(j_.debug()) << "Complete: " << ss.str();
return inbound->getLedger();
};
using namespace std::chrono_literals;
return perf::measureDurationAndLog(doAcquire, "InboundLedgersImp::acquire", 500ms, j_);
std::shared_ptr<Ledger const> ledger =
perf::measureDurationAndLog(doAcquire, "InboundLedgersImp::acquire", 500ms, j_);
return ledger;
}
void
acquireAsync(
JobType type,
std::string const& name,
uint256 const& hash,
std::uint32_t seq,
InboundLedger::Reason reason) override
acquireAsync(uint256 const& hash, std::uint32_t seq, InboundLedger::Reason reason) override
{
if (auto check = std::make_shared<CanProcess const>(acquiresMutex_, pendingAcquires_, hash);
*check)
std::unique_lock lock(acquiresMutex_);
try
{
app_.getJobQueue().addJob(type, name, [check, name, hash, seq, reason, this]() {
JLOG(j_.debug()) << "JOB acquireAsync " << name << " started ";
try
{
acquire(hash, seq, reason);
}
catch (std::exception const& e)
{
JLOG(j_.warn()) << "Exception thrown for acquiring new "
"inbound ledger "
<< hash << ": " << e.what();
}
catch (...)
{
JLOG(j_.warn()) << "Unknown exception thrown for acquiring new "
"inbound ledger "
<< hash;
}
});
if (pendingAcquires_.contains(hash))
return;
pendingAcquires_.insert(hash);
scope_unlock const unlock(lock);
acquire(hash, seq, reason);
}
catch (std::exception const& e)
{
JLOG(j_.warn()) << "Exception thrown for acquiring new inbound ledger " << hash << ": "
<< e.what();
}
catch (...)
{
JLOG(j_.warn()) << "Unknown exception thrown for acquiring new inbound ledger " << hash;
}
pendingAcquires_.erase(hash);
}
std::shared_ptr<InboundLedger>

View File

@@ -965,9 +965,8 @@ LedgerMaster::checkAccept(std::shared_ptr<Ledger const> const& ledger)
return;
}
JLOG(m_journal.info()) << "Advancing accepted ledger to " << ledger->header().seq << " ("
<< to_short_string(ledger->header().hash) << ") with >= " << minVal
<< " validations";
JLOG(m_journal.info()) << "Advancing accepted ledger to " << ledger->header().seq
<< " with >= " << minVal << " validations";
ledger->setValidated();
ledger->setFull();

View File

@@ -25,8 +25,7 @@ TimeoutCounter::TimeoutCounter(
QueueJobParameter&& jobParameter,
beast::Journal journal)
: app_(app)
, sink_(journal, to_short_string(hash) + " ")
, journal_(sink_)
, journal_(journal)
, hash_(hash)
, timerInterval_(interval)
, queueJobParameter_(std::move(jobParameter))
@@ -42,7 +41,6 @@ TimeoutCounter::setTimer(ScopedLockType& sl)
{
if (isDone())
return;
JLOG(journal_.debug()) << "Setting timer for " << timerInterval_.count() << "ms";
timer_.expires_after(timerInterval_);
timer_.async_wait([wptr = pmDowncast()](boost::system::error_code const& ec) {
if (ec == boost::asio::error::operation_aborted)
@@ -50,10 +48,6 @@ TimeoutCounter::setTimer(ScopedLockType& sl)
if (auto ptr = wptr.lock())
{
JLOG(ptr->journal_.debug())
<< "timer: ec: " << ec
<< " (operation_aborted: " << boost::asio::error::operation_aborted << " - "
<< (ec == boost::asio::error::operation_aborted ? "aborted" : "other") << ")";
ScopedLockType sl(ptr->mtx_);
ptr->queueJob(sl);
}

View File

@@ -3,7 +3,6 @@
#include <xrpld/app/main/Application.h>
#include <xrpl/beast/utility/Journal.h>
#include <xrpl/beast/utility/WrappedSink.h>
#include <xrpl/core/Job.h>
#include <boost/asio/basic_waitable_timer.hpp>
@@ -104,7 +103,6 @@ protected:
// Used in this class for access to boost::asio::io_context and
// xrpl::Overlay. Used in subtypes for the kitchen sink.
Application& app_;
beast::WrappedSink sink_;
beast::Journal journal_;
mutable std::recursive_mutex mtx_;

View File

@@ -35,7 +35,6 @@
#include <xrpld/rpc/MPTokenIssuanceID.h>
#include <xrpld/rpc/ServerHandler.h>
#include <xrpl/basics/CanProcess.h>
#include <xrpl/basics/Log.h>
#include <xrpl/basics/ToString.h>
#include <xrpl/basics/UnorderedContainers.h>
@@ -486,7 +485,7 @@ public:
isFull() override;
void
setMode(OperatingMode om, char const* reason) override;
setMode(OperatingMode om) override;
bool
isBlocked() override;
@@ -924,7 +923,7 @@ NetworkOPsImp::strOperatingMode(bool const admin /* = false */) const
inline void
NetworkOPsImp::setStandAlone()
{
setMode(OperatingMode::FULL, "setStandAlone");
setMode(OperatingMode::FULL);
}
inline void
@@ -1067,7 +1066,7 @@ NetworkOPsImp::processHeartbeatTimer()
{
if (mMode != OperatingMode::DISCONNECTED)
{
setMode(OperatingMode::DISCONNECTED, "Heartbeat: insufficient peers");
setMode(OperatingMode::DISCONNECTED);
std::stringstream ss;
ss << "Node count (" << numPeers << ") has fallen "
<< "below required minimum (" << minPeerCount_ << ").";
@@ -1091,7 +1090,7 @@ NetworkOPsImp::processHeartbeatTimer()
if (mMode == OperatingMode::DISCONNECTED)
{
setMode(OperatingMode::CONNECTED, "Heartbeat: sufficient peers");
setMode(OperatingMode::CONNECTED);
JLOG(m_journal.info()) << "Node count (" << numPeers << ") is sufficient.";
CLOG(clog.ss()) << "setting mode to CONNECTED based on " << numPeers << " peers. ";
}
@@ -1102,11 +1101,11 @@ NetworkOPsImp::processHeartbeatTimer()
CLOG(clog.ss()) << "mode: " << strOperatingMode(origMode, true);
if (mMode == OperatingMode::SYNCING)
{
setMode(OperatingMode::SYNCING, "Heartbeat: check syncing");
setMode(OperatingMode::SYNCING);
}
else if (mMode == OperatingMode::CONNECTED)
{
setMode(OperatingMode::CONNECTED, "Heartbeat: check connected");
setMode(OperatingMode::CONNECTED);
}
auto newMode = mMode.load();
if (origMode != newMode)
@@ -1810,7 +1809,7 @@ void
NetworkOPsImp::setAmendmentBlocked()
{
amendmentBlocked_ = true;
setMode(OperatingMode::CONNECTED, "setAmendmentBlocked");
setMode(OperatingMode::CONNECTED);
}
inline bool
@@ -1841,7 +1840,7 @@ void
NetworkOPsImp::setUNLBlocked()
{
unlBlocked_ = true;
setMode(OperatingMode::CONNECTED, "setUNLBlocked");
setMode(OperatingMode::CONNECTED);
}
inline void
@@ -1941,7 +1940,7 @@ NetworkOPsImp::checkLastClosedLedger(Overlay::PeerSequence const& peerList, uint
if ((mMode == OperatingMode::TRACKING) || (mMode == OperatingMode::FULL))
{
setMode(OperatingMode::CONNECTED, "check LCL: not on consensus ledger");
setMode(OperatingMode::CONNECTED);
}
if (consensus)
@@ -2029,8 +2028,8 @@ NetworkOPsImp::beginConsensus(
// this shouldn't happen unless we jump ledgers
if (mMode == OperatingMode::FULL)
{
JLOG(m_journal.warn()) << "beginConsensus Don't have LCL, going to tracking";
setMode(OperatingMode::TRACKING, "beginConsensus: No LCL");
JLOG(m_journal.warn()) << "Don't have LCL, going to tracking";
setMode(OperatingMode::TRACKING);
CLOG(clog) << "beginConsensus Don't have LCL, going to tracking. ";
}
@@ -2158,7 +2157,7 @@ NetworkOPsImp::endConsensus(std::unique_ptr<std::stringstream> const& clog)
// validations we have for LCL. If the ledger is good enough, go to
// TRACKING - TODO
if (!needNetworkLedger_)
setMode(OperatingMode::TRACKING, "endConsensus: check tracking");
setMode(OperatingMode::TRACKING);
}
if (((mMode == OperatingMode::CONNECTED) || (mMode == OperatingMode::TRACKING)) &&
@@ -2171,7 +2170,7 @@ NetworkOPsImp::endConsensus(std::unique_ptr<std::stringstream> const& clog)
if (registry_.get().getTimeKeeper().now() <
(current->header().parentCloseTime + 2 * current->header().closeTimeResolution))
{
setMode(OperatingMode::FULL, "endConsensus: check full");
setMode(OperatingMode::FULL);
}
}
@@ -2183,7 +2182,7 @@ NetworkOPsImp::consensusViewChange()
{
if ((mMode == OperatingMode::FULL) || (mMode == OperatingMode::TRACKING))
{
setMode(OperatingMode::CONNECTED, "consensusViewChange");
setMode(OperatingMode::CONNECTED);
}
}
@@ -2487,7 +2486,7 @@ NetworkOPsImp::pubPeerStatus(std::function<Json::Value(void)> const& func)
}
void
NetworkOPsImp::setMode(OperatingMode om, char const* reason)
NetworkOPsImp::setMode(OperatingMode om)
{
using namespace std::chrono_literals;
if (om == OperatingMode::CONNECTED)
@@ -2507,12 +2506,11 @@ NetworkOPsImp::setMode(OperatingMode om, char const* reason)
if (mMode == om)
return;
auto const sink = om < mMode ? m_journal.warn() : m_journal.info();
mMode = om;
accounting_.mode(om);
JLOG(sink) << "STATE->" << strOperatingMode() << " - " << reason;
JLOG(m_journal.info()) << "STATE->" << strOperatingMode();
pubServer();
}
@@ -2521,24 +2519,36 @@ NetworkOPsImp::recvValidation(std::shared_ptr<STValidation> const& val, std::str
{
JLOG(m_journal.trace()) << "recvValidation " << val->getLedgerHash() << " from " << source;
std::unique_lock lock(validationsMutex_);
BypassAccept bypassAccept = BypassAccept::no;
try
{
CanProcess const check(validationsMutex_, pendingValidations_, val->getLedgerHash());
try
if (pendingValidations_.contains(val->getLedgerHash()))
{
BypassAccept bypassAccept = check ? BypassAccept::no : BypassAccept::yes;
handleNewValidation(registry_.app(), val, source, bypassAccept, m_journal);
bypassAccept = BypassAccept::yes;
}
catch (std::exception const& e)
else
{
JLOG(m_journal.warn()) << "Exception thrown for handling new validation "
<< val->getLedgerHash() << ": " << e.what();
}
catch (...)
{
JLOG(m_journal.warn())
<< "Unknown exception thrown for handling new validation " << val->getLedgerHash();
pendingValidations_.insert(val->getLedgerHash());
}
scope_unlock const unlock(lock);
handleNewValidation(registry_.get().getApp(), val, source, bypassAccept, m_journal);
}
catch (std::exception const& e)
{
JLOG(m_journal.warn()) << "Exception thrown for handling new validation "
<< val->getLedgerHash() << ": " << e.what();
}
catch (...)
{
JLOG(m_journal.warn()) << "Unknown exception thrown for handling new validation "
<< val->getLedgerHash();
}
if (bypassAccept == BypassAccept::no)
{
pendingValidations_.erase(val->getLedgerHash());
}
lock.unlock();
pubValidation(val);

View File

@@ -161,7 +161,11 @@ ValidatorSite::load(
{
try
{
sites_.emplace_back(uri);
// This is not super efficient, but it doesn't happen often.
bool found = std::ranges::any_of(
sites_, [&uri](auto const& site) { return site.loadedResource->uri == uri; });
if (!found)
sites_.emplace_back(uri);
}
catch (std::exception const& e)
{
@@ -222,7 +226,17 @@ ValidatorSite::setTimer(
std::lock_guard<std::mutex> const& site_lock,
std::lock_guard<std::mutex> const& state_lock)
{
auto next = std::ranges::min_element(
if (!sites_.empty() && //
std::ranges::all_of(
sites_, [](auto const& site) { return site.lastRefreshStatus.has_value(); }))
{
// If all of the sites have been handled at least once (including
// errors and timeouts), call missingSite, which will load the cache
// files for any lists that are still unavailable.
missingSite(site_lock);
}
auto const next = std::ranges::min_element(
sites_, [](Site const& a, Site const& b) { return a.nextRefresh < b.nextRefresh; });
if (next != sites_.end())
@@ -333,7 +347,7 @@ ValidatorSite::onRequestTimeout(std::size_t siteIdx, error_code const& ec)
// processes a network error. Usually, this function runs first,
// but on extremely rare occasions, the response handler can run
// first, which will leave activeResource empty.
auto const& site = sites_[siteIdx];
auto& site = sites_[siteIdx];
if (site.activeResource)
{
JLOG(j_.warn()) << "Request for " << site.activeResource->uri << " took too long";
@@ -341,6 +355,9 @@ ValidatorSite::onRequestTimeout(std::size_t siteIdx, error_code const& ec)
else
JLOG(j_.error()) << "Request took too long, but a response has "
"already been processed";
if (!site.lastRefreshStatus)
site.lastRefreshStatus.emplace(
Site::Status{clock_type::now(), ListDisposition::invalid, "timeout"});
}
std::lock_guard const lock_state{state_mutex_};