ARCHIVE/xahaud
3
0
Fork 0
mirror of https://github.com/Xahau/xahaud.git synced 2025-11-04 02:35:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

added configurable NuDB block size support in xahaud (#601)

Browse Source
This commit is contained in:
Niq Dudfield
2025-10-22 11:15:12 +07:00
committed by GitHub
parent 9378f1a0ad
commit 74477d2c13
5 changed files with 462 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
Builds/CMake/RippledCore.cmake
View File

@@ -965,6 +965,7 @@ if (tests)
src/test/nodestore/Basics_test.cpp
src/test/nodestore/DatabaseShard_test.cpp
src/test/nodestore/Database_test.cpp
src/test/nodestore/NuDBFactory_test.cpp
src/test/nodestore/Timing_test.cpp
src/test/nodestore/import_test.cpp
src/test/nodestore/varint_test.cpp

33
cfg/xahaud-example.cfg
View File

@@ -1127,6 +1127,39 @@
# it must be defined with the same value in both
# sections.
#
# Optional keys for NuDB only:
#
# nudb_block_size EXPERIMENTAL: Block size in bytes for NuDB storage.
# Must be a power of 2 between 4096 and 32768. Default is 4096.
#
# This parameter controls the fundamental storage unit
# size for NuDB's internal data structures. The choice
# of block size can significantly impact performance
# depending on your storage hardware and filesystem:
#
# - 4096 bytes: Optimal for most standard SSDs and
# traditional filesystems (ext4, NTFS, HFS+).
# Provides good balance of performance and storage
# efficiency. Recommended for most deployments.
#
# - 8192-16384 bytes: May improve performance on
# high-end NVMe SSDs and copy-on-write filesystems
# like ZFS or Btrfs that benefit from larger block
# alignment. Can reduce metadata overhead for large
# databases.
#
# - 32768 bytes (32K): Maximum supported block size
# for high-performance scenarios with very fast
# storage. May increase memory usage and reduce
# efficiency for smaller databases.
#
# Note: This setting cannot be changed after database
# creation without rebuilding the entire database.
# Choose carefully based on your hardware and expected
# database size.
#
# Example: nudb_block_size=4096
#
# These keys modify the behavior of online_delete, and thus are only
# relevant if online_delete is defined and non-zero:

9
src/ripple/nodestore/Backend.h
View File

@@ -23,6 +23,7 @@
#include <ripple/nodestore/Types.h>
#include <atomic>
#include <cstdint>
#include <optional>
namespace ripple {
namespace NodeStore {
@@ -175,6 +176,14 @@ public:
virtual int
fdRequired() const = 0;
/** Get the block size for backends that support it
*/
virtual std::optional<std::size_t>
getBlockSize() const
{
return std::nullopt;
}
/** Returns read and write stats.
@note The Counters struct is specific to and only used

63
src/ripple/nodestore/backend/NuDBFactory.cpp
View File

@@ -18,6 +18,7 @@
//==============================================================================
#include <ripple/basics/contract.h>
#include <ripple/beast/core/LexicalCast.h>
#include <ripple/nodestore/Factory.h>
#include <ripple/nodestore/Manager.h>
#include <ripple/nodestore/impl/DecodedBlob.h>
@@ -31,6 +32,7 @@
#include <exception>
#include <memory>
#include <nudb/nudb.hpp>
#include <sstream>
namespace ripple {
namespace NodeStore {
@@ -48,6 +50,7 @@ public:
size_t const keyBytes_;
std::size_t const burstSize_;
std::string const name_;
std::size_t const blockSize_;
nudb::store db_;
std::atomic<bool> deletePath_;
Scheduler& scheduler_;
@@ -62,6 +65,7 @@ public:
, keyBytes_(keyBytes)
, burstSize_(burstSize)
, name_(get(keyValues, "path"))
, blockSize_(parseBlockSize(name_, keyValues, journal))
, deletePath_(false)
, scheduler_(scheduler)
{
@@ -81,6 +85,7 @@ public:
, keyBytes_(keyBytes)
, burstSize_(burstSize)
, name_(get(keyValues, "path"))
, blockSize_(parseBlockSize(name_, keyValues, journal))
, db_(context)
, deletePath_(false)
, scheduler_(scheduler)
@@ -110,6 +115,12 @@ public:
return name_;
}
std::optional<std::size_t>
getBlockSize() const override
{
return blockSize_;
}
void
open(bool createIfMissing, uint64_t appType, uint64_t uid, uint64_t salt)
override
@@ -137,7 +148,7 @@ public:
uid,
salt,
keyBytes_,
nudb::block_size(kp),
blockSize_,
0.50,
ec);
if (ec == nudb::errc::file_exists)
@@ -362,6 +373,56 @@ public:
{
return 3;
}
private:
static std::size_t
parseBlockSize(
std::string const& name,
Section const& keyValues,
beast::Journal journal)
{
using namespace boost::filesystem;
auto const folder = path(name);
auto const kp = (folder / "nudb.key").string();
std::size_t const defaultSize =
nudb::block_size(kp); // Default 4K from NuDB
std::size_t blockSize = defaultSize;
std::string blockSizeStr;
if (!get_if_exists(keyValues, "nudb_block_size", blockSizeStr))
{
return blockSize; // Early return with default
}
try
{
std::size_t const parsedBlockSize =
beast::lexicalCastThrow<std::size_t>(blockSizeStr);
// Validate: must be power of 2 between 4K and 32K
if (parsedBlockSize < 4096 || parsedBlockSize > 32768 ||
(parsedBlockSize & (parsedBlockSize - 1)) != 0)
{
std::stringstream s;
s << "Invalid nudb_block_size: " << parsedBlockSize
<< ". Must be power of 2 between 4096 and 32768.";
Throw<std::runtime_error>(s.str());
}
JLOG(journal.info())
<< "Using custom NuDB block size: " << parsedBlockSize
<< " bytes";
return parsedBlockSize;
}
catch (std::exception const& e)
{
std::stringstream s;
s << "Invalid nudb_block_size value: " << blockSizeStr
<< ". Error: " << e.what();
Throw<std::runtime_error>(s.str());
}
}
};
//------------------------------------------------------------------------------

357
src/test/nodestore/NuDBFactory_test.cpp Normal file
View File

@@ -0,0 +1,357 @@
//------------------------------------------------------------------------------
/*
This file is part of rippled: https://github.com/ripple/rippled
Copyright (c) 2012, 2013 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 <test/nodestore/TestBase.h>
#include <test/unit_test/SuiteJournal.h>
#include <ripple/basics/BasicConfig.h>
#include <ripple/basics/ByteUtilities.h>
#include <ripple/beast/utility/temp_dir.h>
#include <ripple/nodestore/DummyScheduler.h>
#include <ripple/nodestore/Manager.h>
#include <memory>
#include <sstream>
namespace ripple {
namespace NodeStore {
class NuDBFactory_test : public TestBase
{
private:
// Helper function to create a Section with specified parameters
Section
createSection(std::string const& path, std::string const& blockSize = "")
{
Section params;
params.set("type", "nudb");
params.set("path", path);
if (!blockSize.empty())
params.set("nudb_block_size", blockSize);
return params;
}
// Helper function to create a backend and test basic functionality
bool
testBackendFunctionality(
Section const& params,
std::size_t expectedBlocksize)
{
try
{
DummyScheduler scheduler;
test::SuiteJournal journal("NuDBFactory_test", *this);
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
if (!BEAST_EXPECT(backend))
return false;
if (!BEAST_EXPECT(backend->getBlockSize() == expectedBlocksize))
return false;
backend->open();
if (!BEAST_EXPECT(backend->isOpen()))
return false;
// Test basic store/fetch functionality
auto batch = createPredictableBatch(10, 12345);
storeBatch(*backend, batch);
Batch copy;
fetchCopyOfBatch(*backend, &copy, batch);
backend->close();
return areBatchesEqual(batch, copy);
}
catch (...)
{
return false;
}
}
// Helper function to test log messages
void
testLogMessage(
Section const& params,
beast::severities::Severity level,
std::string const& expectedMessage)
{
test::StreamSink sink(level);
beast::Journal journal(sink);
DummyScheduler scheduler;
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
std::string logOutput = sink.messages().str();
BEAST_EXPECT(logOutput.find(expectedMessage) != std::string::npos);
}
public:
void
testDefaultBlockSize()
{
testcase("Default block size (no nudb_block_size specified)");
beast::temp_dir tempDir;
auto params = createSection(tempDir.path());
// Should work with default 4096 block size
BEAST_EXPECT(testBackendFunctionality(params, 4096));
}
void
testValidBlockSizes()
{
testcase("Valid block sizes");
std::vector<std::size_t> validSizes = {4096, 8192, 16384, 32768};
for (auto const& size : validSizes)
{
beast::temp_dir tempDir;
auto params = createSection(tempDir.path(), to_string(size));
BEAST_EXPECT(testBackendFunctionality(params, size));
}
}
void
testInvalidBlockSizes()
{
testcase("Invalid block sizes");
std::vector<std::string> invalidSizes = {
"2048", // Too small
"1024", // Too small
"65536", // Too large
"131072", // Too large
"5000", // Not power of 2
"6000", // Not power of 2
"10000", // Not power of 2
"0", // Zero
"-1", // Negative
"abc", // Non-numeric
"4k", // Invalid format
"4096.5" // Decimal
};
for (auto const& size : invalidSizes)
{
beast::temp_dir tempDir;
auto params = createSection(tempDir.path(), size);
DummyScheduler scheduler;
test::SuiteJournal journal("NuDBFactory_test", *this);
// Should throw exception for invalid sizes
try
{
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
// If we get here, the test failed
BEAST_EXPECT(false);
}
catch (std::exception const& e)
{
// Expected exception
std::string error{e.what()};
BEAST_EXPECT(
error.find("Invalid nudb_block_size") != std::string::npos);
}
}
}
void
testLogMessages()
{
testcase("Log message verification");
// Test valid custom block size logging
{
beast::temp_dir tempDir;
auto params = createSection(tempDir.path(), "8192");
testLogMessage(
params,
beast::severities::kInfo,
"Using custom NuDB block size: 8192");
}
// Test invalid block size exception message
{
beast::temp_dir tempDir;
auto params = createSection(tempDir.path(), "5000");
test::StreamSink sink(beast::severities::kWarning);
beast::Journal journal(sink);
DummyScheduler scheduler;
try
{
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
fail();
}
catch (std::exception const& e)
{
std::string logOutput{e.what()};
BEAST_EXPECT(
logOutput.find("Invalid nudb_block_size: 5000") !=
std::string::npos);
BEAST_EXPECT(
logOutput.find(
"Must be power of 2 between 4096 and 32768") !=
std::string::npos);
}
}
// Test non-numeric value exception message
{
beast::temp_dir tempDir;
auto params = createSection(tempDir.path(), "invalid");
test::StreamSink sink(beast::severities::kWarning);
beast::Journal journal(sink);
DummyScheduler scheduler;
try
{
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
fail();
}
catch (std::exception const& e)
{
std::string logOutput{e.what()};
BEAST_EXPECT(
logOutput.find("Invalid nudb_block_size value: invalid") !=
std::string::npos);
}
}
}
void
testPowerOfTwoValidation()
{
testcase("Power of 2 validation logic");
// Test edge cases around valid range
std::vector<std::pair<std::string, bool>> testCases = {
{"4095", false}, // Just below minimum
{"4096", true}, // Minimum valid
{"4097", false}, // Just above minimum, not power of 2
{"8192", true}, // Valid power of 2
{"8193", false}, // Just above valid power of 2
{"16384", true}, // Valid power of 2
{"32768", true}, // Maximum valid
{"32769", false}, // Just above maximum
{"65536", false} // Power of 2 but too large
};
for (auto const& [size, shouldWork] : testCases)
{
beast::temp_dir tempDir;
auto params = createSection(tempDir.path(), size);
test::StreamSink sink(beast::severities::kWarning);
beast::Journal journal(sink);
DummyScheduler scheduler;
try
{
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
BEAST_EXPECT(shouldWork);
}
catch (std::exception const& e)
{
std::string logOutput{e.what()};
BEAST_EXPECT(
logOutput.find("Invalid nudb_block_size") !=
std::string::npos);
BEAST_EXPECT(!shouldWork);
}
}
}
void
testDataPersistence()
{
testcase("Data persistence with different block sizes");
std::vector<std::string> blockSizes = {
"4096", "8192", "16384", "32768"};
for (auto const& size : blockSizes)
{
beast::temp_dir tempDir;
auto params = createSection(tempDir.path(), size);
DummyScheduler scheduler;
test::SuiteJournal journal("NuDBFactory_test", *this);
// Create test data
auto batch = createPredictableBatch(50, 54321);
// Store data
{
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
backend->open();
storeBatch(*backend, batch);
backend->close();
}
// Retrieve data in new backend instance
{
auto backend = Manager::instance().make_Backend(
params, megabytes(4), scheduler, journal);
backend->open();
Batch copy;
fetchCopyOfBatch(*backend, &copy, batch);
BEAST_EXPECT(areBatchesEqual(batch, copy));
backend->close();
}
}
}
void
run() override
{
testDefaultBlockSize();
testValidBlockSizes();
testInvalidBlockSizes();
testLogMessages();
testPowerOfTwoValidation();
testDataPersistence();
}
};
BEAST_DEFINE_TESTSUITE(NuDBFactory, ripple_core, ripple);
} // namespace NodeStore
} // namespace ripple