allow different shifts for books and keys. default to 10 and 20 respectively

This commit is contained in:
CJ Cobb
2021-05-06 20:13:04 +00:00
parent 20b8059151
commit 144ad5eb19
5 changed files with 293 additions and 69 deletions

View File

@@ -2,7 +2,8 @@
namespace Backend { namespace Backend {
BackendIndexer::BackendIndexer(boost::json::object const& config) BackendIndexer::BackendIndexer(boost::json::object const& config)
: shift_(config.at("indexer_shift").as_int64()) : keyShift_(config.at("indexer_key_shift").as_int64())
, bookShift_(config.at("indexer_book_shift").as_int64())
{ {
work_.emplace(ioc_); work_.emplace(ioc_);
ioThread_ = std::thread{[this]() { ioc_.run(); }}; ioThread_ = std::thread{[this]() { ioc_.run(); }};
@@ -65,11 +66,59 @@ BackendIndexer::deleteBookOffer(
} }
void void
writeFlagLedger( writeKeyFlagLedger(
uint32_t ledgerSequence,
uint32_t shift,
BackendInterface const& backend,
std::unordered_set<ripple::uint256> const& keys)
{
uint32_t nextFlag = ((ledgerSequence >> shift << shift) + (1 << shift));
ripple::uint256 zero = {};
BOOST_LOG_TRIVIAL(info)
<< __func__
<< " starting. ledgerSequence = " << std::to_string(ledgerSequence)
<< " nextFlag = " << std::to_string(nextFlag)
<< " keys.size() = " << std::to_string(keys.size());
while (true)
{
try
{
auto [objects, curCursor, warning] =
backend.fetchLedgerPage({}, nextFlag, 1);
if (!warning)
{
BOOST_LOG_TRIVIAL(warning)
<< __func__ << " flag ledger already written. sequence = "
<< std::to_string(ledgerSequence)
<< " next flag = " << std::to_string(nextFlag)
<< "returning";
return;
}
break;
}
catch (DatabaseTimeout& t)
{
;
}
}
auto start = std::chrono::system_clock::now();
backend.writeKeys(keys, nextFlag);
backend.writeKeys({zero}, nextFlag);
auto end = std::chrono::system_clock::now();
BOOST_LOG_TRIVIAL(info)
<< __func__
<< " finished. ledgerSequence = " << std::to_string(ledgerSequence)
<< " nextFlag = " << std::to_string(nextFlag)
<< " keys.size() = " << std::to_string(keys.size())
<< std::chrono::duration_cast<std::chrono::seconds>(end - start)
.count();
}
void
writeBookFlagLedger(
uint32_t ledgerSequence, uint32_t ledgerSequence,
uint32_t shift, uint32_t shift,
BackendInterface const& backend, BackendInterface const& backend,
std::unordered_set<ripple::uint256> const& keys,
std::unordered_map< std::unordered_map<
ripple::uint256, ripple::uint256,
std::unordered_set<ripple::uint256>> const& books) std::unordered_set<ripple::uint256>> const& books)
@@ -81,15 +130,14 @@ writeFlagLedger(
<< __func__ << __func__
<< " starting. ledgerSequence = " << std::to_string(ledgerSequence) << " starting. ledgerSequence = " << std::to_string(ledgerSequence)
<< " nextFlag = " << std::to_string(nextFlag) << " nextFlag = " << std::to_string(nextFlag)
<< " keys.size() = " << std::to_string(keys.size())
<< " books.size() = " << std::to_string(books.size()); << " books.size() = " << std::to_string(books.size());
while (true) while (true)
{ {
try try
{ {
auto [objects, curCursor, warning] = auto [objects, curCursor, warning] =
backend.fetchLedgerPage({}, nextFlag, 1); backend.fetchBookOffers(zero, nextFlag, 1);
if (!(warning || objects.size() == 0)) if (!warning)
{ {
BOOST_LOG_TRIVIAL(warning) BOOST_LOG_TRIVIAL(warning)
<< __func__ << " flag ledger already written. sequence = " << __func__ << " flag ledger already written. sequence = "
@@ -109,16 +157,11 @@ writeFlagLedger(
backend.writeBooks(books, nextFlag); backend.writeBooks(books, nextFlag);
backend.writeBooks({{zero, {zero}}}, nextFlag); backend.writeBooks({{zero, {zero}}}, nextFlag);
BOOST_LOG_TRIVIAL(debug) << __func__ << " wrote books. writing keys ...";
backend.writeKeys(keys, nextFlag);
backend.writeKeys({zero}, nextFlag);
auto end = std::chrono::system_clock::now(); auto end = std::chrono::system_clock::now();
BOOST_LOG_TRIVIAL(info) BOOST_LOG_TRIVIAL(info)
<< __func__ << __func__
<< " finished. ledgerSequence = " << std::to_string(ledgerSequence) << " finished. ledgerSequence = " << std::to_string(ledgerSequence)
<< " nextFlag = " << std::to_string(nextFlag) << " nextFlag = " << std::to_string(nextFlag)
<< " keys.size() = " << std::to_string(keys.size())
<< " books.size() = " << std::to_string(books.size()) << " time = " << " books.size() = " << std::to_string(books.size()) << " time = "
<< std::chrono::duration_cast<std::chrono::seconds>(end - start) << std::chrono::duration_cast<std::chrono::seconds>(end - start)
.count(); .count();
@@ -132,7 +175,80 @@ BackendIndexer::clearCaches()
} }
void void
BackendIndexer::populateCaches( BackendIndexer::doBooksRepair(
BackendInterface const& backend,
std::optional<uint32_t> sequence)
{
if (!sequence)
{
auto rng = backend.fetchLedgerRangeNoThrow();
if (!rng)
return;
sequence = rng->maxSequence;
}
BOOST_LOG_TRIVIAL(info)
<< __func__ << " sequence = " << std::to_string(*sequence);
ripple::uint256 zero = {};
while (true)
{
try
{
auto [objects, cursor, warning] =
backend.fetchBookOffers(zero, *sequence, 1);
if (!warning)
{
BOOST_LOG_TRIVIAL(warning)
<< __func__ << " flag ledger already written. sequence = "
<< std::to_string(*sequence) << "returning";
return;
}
else
{
uint32_t lower = (*sequence - 1) >> bookShift_ << bookShift_;
doBooksRepair(backend, lower);
}
break;
}
catch (DatabaseTimeout& t)
{
;
}
}
std::optional<ripple::uint256> cursor;
while (true)
{
try
{
auto [objects, curCursor, warning] =
backend.fetchLedgerPage(cursor, *sequence, 2048);
BOOST_LOG_TRIVIAL(debug) << __func__ << " fetched a page";
cursor = curCursor;
for (auto& obj : objects)
{
if (isOffer(obj.blob))
{
auto book = getBook(obj.blob);
booksRepair[book].insert(obj.key);
}
}
if (!cursor)
break;
}
catch (DatabaseTimeout const& e)
{
BOOST_LOG_TRIVIAL(warning)
<< __func__ << " Database timeout fetching keys";
std::this_thread::sleep_for(std::chrono::seconds(2));
}
}
writeBookFlagLedger(*sequence, bookShift_, backend, booksRepair);
booksRepair = {};
BOOST_LOG_TRIVIAL(info)
<< __func__ << " finished. sequence = " << std::to_string(*sequence);
}
void
BackendIndexer::doKeysRepair(
BackendInterface const& backend, BackendInterface const& backend,
std::optional<uint32_t> sequence) std::optional<uint32_t> sequence)
{ {
@@ -152,16 +268,65 @@ BackendIndexer::populateCaches(
{ {
auto [objects, curCursor, warning] = auto [objects, curCursor, warning] =
backend.fetchLedgerPage(cursor, *sequence, 2048); backend.fetchLedgerPage(cursor, *sequence, 2048);
if (warning) // no cursor means this is the first page
if (!cursor)
{
// if there is no warning, we don't need to do a repair
// warning only shows up on the first page
if (!warning)
{
BOOST_LOG_TRIVIAL(info)
<< __func__
<< " flag ledger already written. returning";
return;
}
else
{
uint32_t lower = (*sequence - 1) >> keyShift_ << keyShift_;
doKeysRepair(backend, lower);
}
}
BOOST_LOG_TRIVIAL(debug) << __func__ << " fetched a page";
cursor = curCursor;
for (auto& obj : objects)
{
keysRepair.insert(obj.key);
}
if (!cursor)
break;
}
catch (DatabaseTimeout const& e)
{ {
BOOST_LOG_TRIVIAL(warning) BOOST_LOG_TRIVIAL(warning)
<< __func__ << " performing index repair"; << __func__ << " Database timeout fetching keys";
uint32_t lower = (*sequence - 1) >> shift_ << shift_; std::this_thread::sleep_for(std::chrono::seconds(2));
populateCaches(backend, lower);
writeFlagLedger(
lower, shift_, backend, keysCumulative, booksCumulative);
clearCaches();
} }
}
writeKeyFlagLedger(*sequence, keyShift_, backend, keysRepair);
keysRepair = {};
BOOST_LOG_TRIVIAL(info)
<< __func__ << " finished. sequence = " << std::to_string(*sequence);
}
void
BackendIndexer::populateCaches(BackendInterface const& backend)
{
auto rng = backend.fetchLedgerRangeNoThrow();
if (!rng)
return;
uint32_t sequence = rng->maxSequence;
BOOST_LOG_TRIVIAL(info)
<< __func__ << " sequence = " << std::to_string(sequence);
doBooksRepair(backend, sequence);
doKeysRepair(backend, sequence);
std::optional<ripple::uint256> cursor;
while (true)
{
try
{
auto [objects, curCursor, warning] =
backend.fetchLedgerPage(cursor, sequence, 2048);
BOOST_LOG_TRIVIAL(debug) << __func__ << " fetched a page"; BOOST_LOG_TRIVIAL(debug) << __func__ << " fetched a page";
cursor = curCursor; cursor = curCursor;
for (auto& obj : objects) for (auto& obj : objects)
@@ -189,7 +354,6 @@ BackendIndexer::populateCaches(
std::unique_lock lck(mtx); std::unique_lock lck(mtx);
populatingCacheAsync = false; populatingCacheAsync = false;
} }
auto tip = backend.fetchLatestLedgerSequence();
for (auto& key : deletedKeys) for (auto& key : deletedKeys)
{ {
deleteKey(key); deleteKey(key);
@@ -212,9 +376,7 @@ BackendIndexer::populateCaches(
<< " finished. keys.size() = " << std::to_string(keysCumulative.size()); << " finished. keys.size() = " << std::to_string(keysCumulative.size());
} }
void void
BackendIndexer::populateCachesAsync( BackendIndexer::populateCachesAsync(BackendInterface const& backend)
BackendInterface const& backend,
std::optional<uint32_t> sequence)
{ {
if (keysCumulative.size() > 0) if (keysCumulative.size() > 0)
{ {
@@ -226,11 +388,8 @@ BackendIndexer::populateCachesAsync(
std::unique_lock lck(mtx); std::unique_lock lck(mtx);
populatingCacheAsync = true; populatingCacheAsync = true;
} }
BOOST_LOG_TRIVIAL(info) BOOST_LOG_TRIVIAL(info) << __func__;
<< __func__ << " seq = " << (sequence ? std::to_string(*sequence) : ""); boost::asio::post(ioc_, [this, &backend]() { populateCaches(backend); });
boost::asio::post(ioc_, [this, sequence, &backend]() {
populateCaches(backend, sequence);
});
} }
void void
@@ -243,7 +402,25 @@ BackendIndexer::waitForCaches()
} }
void void
BackendIndexer::writeFlagLedgerAsync( BackendIndexer::writeKeyFlagLedgerAsync(
uint32_t ledgerSequence,
BackendInterface const& backend)
{
BOOST_LOG_TRIVIAL(info)
<< __func__
<< " starting. sequence = " << std::to_string(ledgerSequence);
waitForCaches();
auto keysCopy = keysCumulative;
boost::asio::post(ioc_, [=, this, &backend]() {
writeKeyFlagLedger(ledgerSequence, keyShift_, backend, keysCopy);
});
BOOST_LOG_TRIVIAL(info)
<< __func__
<< " finished. sequence = " << std::to_string(ledgerSequence);
}
void
BackendIndexer::writeBookFlagLedgerAsync(
uint32_t ledgerSequence, uint32_t ledgerSequence,
BackendInterface const& backend) BackendInterface const& backend)
{ {
@@ -253,9 +430,8 @@ BackendIndexer::writeFlagLedgerAsync(
waitForCaches(); waitForCaches();
auto booksCopy = booksCumulative; auto booksCopy = booksCumulative;
auto keysCopy = keysCumulative;
boost::asio::post(ioc_, [=, this, &backend]() { boost::asio::post(ioc_, [=, this, &backend]() {
writeFlagLedger(ledgerSequence, shift_, backend, keysCopy, booksCopy); writeBookFlagLedger(ledgerSequence, bookShift_, backend, booksCopy);
}); });
BOOST_LOG_TRIVIAL(info) BOOST_LOG_TRIVIAL(info)
<< __func__ << __func__
@@ -269,21 +445,23 @@ BackendIndexer::finish(uint32_t ledgerSequence, BackendInterface const& backend)
<< __func__ << __func__
<< " starting. sequence = " << std::to_string(ledgerSequence); << " starting. sequence = " << std::to_string(ledgerSequence);
bool isFirst = false; bool isFirst = false;
uint32_t index = getIndexOfSeq(ledgerSequence); uint32_t keyIndex = getKeyIndexOfSeq(ledgerSequence);
uint32_t bookIndex = getKeyIndexOfSeq(ledgerSequence);
auto rng = backend.fetchLedgerRangeNoThrow(); auto rng = backend.fetchLedgerRangeNoThrow();
if (!rng || rng->minSequence == ledgerSequence) if (!rng || rng->minSequence == ledgerSequence)
{ {
isFirst = true; isFirst = true;
index = ledgerSequence; keyIndex = bookIndex = ledgerSequence;
} }
backend.writeKeys(keys, index); backend.writeKeys(keys, keyIndex);
backend.writeBooks(books, index); backend.writeBooks(books, bookIndex);
if (isFirst) if (isFirst)
{ {
ripple::uint256 zero = {}; ripple::uint256 zero = {};
backend.writeBooks({{zero, {zero}}}, ledgerSequence); backend.writeBooks({{zero, {zero}}}, ledgerSequence);
backend.writeKeys({zero}, ledgerSequence); backend.writeKeys({zero}, ledgerSequence);
writeFlagLedgerAsync(ledgerSequence, backend); writeBookFlagLedgerAsync(ledgerSequence, backend);
writeKeyFlagLedgerAsync(ledgerSequence, backend);
} }
keys = {}; keys = {};
books = {}; books = {};

View File

@@ -68,7 +68,8 @@ class BackendIndexer
std::mutex mutex_; std::mutex mutex_;
std::optional<boost::asio::io_context::work> work_; std::optional<boost::asio::io_context::work> work_;
std::thread ioThread_; std::thread ioThread_;
uint32_t shift_ = 16; uint32_t keyShift_ = 20;
uint32_t bookShift_ = 10;
std::unordered_set<ripple::uint256> keys; std::unordered_set<ripple::uint256> keys;
std::unordered_set<ripple::uint256> keysCumulative; std::unordered_set<ripple::uint256> keysCumulative;
std::unordered_map<ripple::uint256, std::unordered_set<ripple::uint256>> std::unordered_map<ripple::uint256, std::unordered_set<ripple::uint256>>
@@ -80,6 +81,9 @@ class BackendIndexer
std::unordered_set<ripple::uint256> deletedKeys; std::unordered_set<ripple::uint256> deletedKeys;
std::unordered_map<ripple::uint256, std::unordered_set<ripple::uint256>> std::unordered_map<ripple::uint256, std::unordered_set<ripple::uint256>>
deletedBooks; deletedBooks;
std::unordered_set<ripple::uint256> keysRepair;
std::unordered_map<ripple::uint256, std::unordered_set<ripple::uint256>>
booksRepair;
std::mutex mtx; std::mutex mtx;
std::condition_variable cv_; std::condition_variable cv_;
@@ -95,13 +99,9 @@ public:
~BackendIndexer(); ~BackendIndexer();
void void
populateCachesAsync( populateCachesAsync(BackendInterface const& backend);
BackendInterface const& backend,
std::optional<uint32_t> sequence = {});
void void
populateCaches( populateCaches(BackendInterface const& backend);
BackendInterface const& backend,
std::optional<uint32_t> sequence = {});
void void
clearCaches(); clearCaches();
// Blocking, possibly for minutes // Blocking, possibly for minutes
@@ -123,26 +123,56 @@ public:
void void
finish(uint32_t ledgerSequence, BackendInterface const& backend); finish(uint32_t ledgerSequence, BackendInterface const& backend);
void void
writeFlagLedgerAsync( writeKeyFlagLedgerAsync(
uint32_t ledgerSequence, uint32_t ledgerSequence,
BackendInterface const& backend); BackendInterface const& backend);
void
writeBookFlagLedgerAsync(
uint32_t ledgerSequence,
BackendInterface const& backend);
void
doKeysRepair(
BackendInterface const& backend,
std::optional<uint32_t> sequence);
void
doBooksRepair(
BackendInterface const& backend,
std::optional<uint32_t> sequence);
uint32_t uint32_t
getShift() getBookShift()
{ {
return shift_; return bookShift_;
} }
uint32_t uint32_t
getIndexOfSeq(uint32_t seq) const getKeyShift()
{ {
if (isFlagLedger(seq)) return keyShift_;
}
uint32_t
getKeyIndexOfSeq(uint32_t seq) const
{
if (isKeyFlagLedger(seq))
return seq; return seq;
auto incr = (1 << shift_); auto incr = (1 << keyShift_);
return (seq >> shift_ << shift_) + incr; return (seq >> keyShift_ << keyShift_) + incr;
} }
bool bool
isFlagLedger(uint32_t ledgerSequence) const isKeyFlagLedger(uint32_t ledgerSequence) const
{ {
return (ledgerSequence % (1 << shift_)) == 0; return (ledgerSequence % (1 << keyShift_)) == 0;
}
uint32_t
getBookIndexOfSeq(uint32_t seq) const
{
if (isBookFlagLedger(seq))
return seq;
auto incr = (1 << bookShift_);
return (seq >> bookShift_ << bookShift_) + incr;
}
bool
isBookFlagLedger(uint32_t ledgerSequence) const
{
return (ledgerSequence % (1 << bookShift_)) == 0;
} }
}; };
@@ -164,16 +194,28 @@ public:
} }
std::optional<uint32_t> std::optional<uint32_t>
getIndexOfSeq(uint32_t seq) const getKeyIndexOfSeq(uint32_t seq) const
{ {
if (indexer_.isFlagLedger(seq)) if (indexer_.isKeyFlagLedger(seq))
return seq; return seq;
auto rng = fetchLedgerRange(); auto rng = fetchLedgerRange();
if (!rng) if (!rng)
return {}; return {};
if (rng->minSequence == seq) if (rng->minSequence == seq)
return seq; return seq;
return indexer_.getIndexOfSeq(seq); return indexer_.getKeyIndexOfSeq(seq);
}
std::optional<uint32_t>
getBookIndexOfSeq(uint32_t seq) const
{
if (indexer_.isBookFlagLedger(seq))
return seq;
auto rng = fetchLedgerRange();
if (!rng)
return {};
if (rng->minSequence == seq)
return seq;
return indexer_.getBookIndexOfSeq(seq);
} }
bool bool
@@ -183,8 +225,10 @@ public:
auto commitRes = doFinishWrites(); auto commitRes = doFinishWrites();
if (commitRes) if (commitRes)
{ {
if (indexer_.isFlagLedger(ledgerSequence)) if (indexer_.isBookFlagLedger(ledgerSequence))
indexer_.writeFlagLedgerAsync(ledgerSequence, *this); indexer_.writeBookFlagLedgerAsync(ledgerSequence, *this);
if (indexer_.isKeyFlagLedger(ledgerSequence))
indexer_.writeKeyFlagLedgerAsync(ledgerSequence, *this);
} }
return commitRes; return commitRes;
} }

View File

@@ -399,7 +399,7 @@ CassandraBackend::fetchLedgerPage(
std::uint32_t ledgerSequence, std::uint32_t ledgerSequence,
std::uint32_t limit) const std::uint32_t limit) const
{ {
auto index = getIndexOfSeq(ledgerSequence); auto index = getKeyIndexOfSeq(ledgerSequence);
if (!index) if (!index)
return {}; return {};
LedgerPage page; LedgerPage page;
@@ -449,11 +449,11 @@ CassandraBackend::fetchLedgerPage(
page.objects.push_back({std::move(key), std::move(obj)}); page.objects.push_back({std::move(key), std::move(obj)});
} }
} }
if (keys.size() && !cursor && !keys[0].isZero()) if (!keys.size() || (!cursor && !keys[0].isZero()))
page.warning = "Data may be incomplete"; page.warning = "Data may be incomplete";
return page; return page;
} }
return {{}, {}}; return {{}, {}, "Data may be incomplete"};
} }
std::vector<Blob> std::vector<Blob>
CassandraBackend::fetchLedgerObjects( CassandraBackend::fetchLedgerObjects(
@@ -498,7 +498,7 @@ CassandraBackend::fetchBookOffers(
{ {
CassandraStatement statement{selectBook_}; CassandraStatement statement{selectBook_};
statement.bindBytes(book); statement.bindBytes(book);
auto index = getIndexOfSeq(sequence); auto index = getBookIndexOfSeq(sequence);
if (!index) if (!index)
return {}; return {};
BOOST_LOG_TRIVIAL(info) << __func__ << " index = " << std::to_string(*index) BOOST_LOG_TRIVIAL(info) << __func__ << " index = " << std::to_string(*index)
@@ -517,7 +517,7 @@ CassandraBackend::fetchBookOffers(
BOOST_LOG_TRIVIAL(debug) << __func__ << " - got keys"; BOOST_LOG_TRIVIAL(debug) << __func__ << " - got keys";
std::vector<ripple::uint256> keys; std::vector<ripple::uint256> keys;
if (!result) if (!result)
return {{}, {}}; return {{}, {}, "Data may be incomplete"};
do do
{ {
keys.push_back(result.getUInt256()); keys.push_back(result.getUInt256());
@@ -535,17 +535,17 @@ CassandraBackend::fetchBookOffers(
results.push_back({keys[i], objs[i]}); results.push_back({keys[i], objs[i]});
} }
std::optional<std::string> warning; std::optional<std::string> warning;
if (keys[0].isZero()) if (!cursor && !keys[0].isZero())
warning = "Data may be incomplete"; warning = "Data may be incomplete";
if (keys.size() == limit) if (keys.size() == limit)
return {results, keys[keys.size() - 1], warning}; return {results, keys[keys.size() - 1], warning};
else else
return {results, {}, warning}; return {results, {}, warning};
return {{}, {}};
} }
else if (!cursor)
return {{}, {}, "Data may be incomplete"};
return {{}, {}}; return {};
} }
struct WriteBookCallbackData struct WriteBookCallbackData
{ {

View File

@@ -329,7 +329,7 @@ PostgresBackend::fetchLedgerPage(
std::uint32_t ledgerSequence, std::uint32_t ledgerSequence,
std::uint32_t limit) const std::uint32_t limit) const
{ {
auto index = getIndexOfSeq(ledgerSequence); auto index = getKeyIndexOfSeq(ledgerSequence);
if (!index) if (!index)
return {}; return {};
PgQuery pgQuery(pgPool_); PgQuery pgQuery(pgPool_);
@@ -366,6 +366,8 @@ PostgresBackend::fetchLedgerPage(
return {results, returnCursor, "Data may be incomplete"}; return {results, returnCursor, "Data may be incomplete"};
return {results, returnCursor}; return {results, returnCursor};
} }
if (!cursor)
return {{}, {}, "Data may be incomplete"};
return {}; return {};
} }

View File

@@ -300,7 +300,7 @@ ReportingETL::buildNextLedger(org::xrpl::rpc::v1::GetLedgerResponse& rawData)
flatMapBackend_->writeAccountTransactions(std::move(accountTxData)); flatMapBackend_->writeAccountTransactions(std::move(accountTxData));
accumTxns_ += rawData.transactions_list().transactions_size(); accumTxns_ += rawData.transactions_list().transactions_size();
bool success = true; bool success = true;
if (accumTxns_ > txnThreshold_) if (accumTxns_ >= txnThreshold_)
{ {
auto start = std::chrono::system_clock::now(); auto start = std::chrono::system_clock::now();
success = flatMapBackend_->finishWrites(lgrInfo.seq); success = flatMapBackend_->finishWrites(lgrInfo.seq);