yams/tests/unit/wal/wal_manager_catch2_test.cpp
trvon a636204f42
fix(wal): preserve damaged zero-prefix segments
Only prune a nonempty segment after every byte is proven zero. A zero header alone is not evidence that recovery data is absent. Scan with bounded memory and preserve the file on read failure.

Signed-off-by: trvon <me@trevon.dev>
2026-09-10 16:26:08 -06:00

250 lines
8.1 KiB
C++

#include <catch2/catch_test_macros.hpp>
#include <catch2/generators/catch_generators.hpp>
#include <chrono>
#include <filesystem>
#include <fstream>
#include <string>
#include <yams/core/types.h>
#include <yams/wal/wal_manager.h>
using namespace yams;
using namespace yams::wal;
namespace {
struct TempDir {
TempDir() {
auto stamp = std::to_string(std::chrono::steady_clock::now().time_since_epoch().count());
path = std::filesystem::temp_directory_path() / ("yams_wal_manager_test_" + stamp);
std::filesystem::create_directories(path);
}
~TempDir() {
std::error_code ec;
std::filesystem::remove_all(path, ec);
}
std::filesystem::path path;
};
std::string makeHash(char fill) {
return std::string(HASH_SIZE, fill);
}
} // namespace
TEST_CASE("WALManager initialize picks max sequence from existing log names",
"[unit][wal][manager]") {
TempDir temp;
auto walDir = temp.path / "wal";
std::filesystem::create_directories(walDir);
{
std::ofstream plain((walDir / "wal_20240101_010101_42.log").string(), std::ios::binary);
REQUIRE(plain.good());
}
{
std::ofstream compressed((walDir / "wal_20240101_010102_123.log.zst").string(),
std::ios::binary);
REQUIRE(compressed.good());
}
{
std::ofstream ignored((walDir / "wal_invalid.log").string(), std::ios::binary);
REQUIRE(ignored.good());
}
WALManager::Config cfg;
cfg.walDirectory = walDir;
cfg.compressOldLogs = false;
cfg.enableGroupCommit = false;
cfg.syncInterval = 1;
WALManager manager(cfg);
REQUIRE(manager.initialize());
CHECK(manager.getCurrentSequence() == 123);
auto payload = WALEntry::DeleteBlockData::encode(makeHash('s'));
WALEntry entry(WALEntry::OpType::DeleteBlock, 0, 0, payload);
auto writeResult = manager.writeEntry(entry);
REQUIRE(writeResult);
CHECK(writeResult.value() == 124);
CHECK(manager.getCurrentSequence() == 124);
REQUIRE(manager.shutdown());
}
TEST_CASE("WALManager pruneLogs removes old inactive logs and keeps active log",
"[unit][wal][manager]") {
TempDir temp;
auto walDir = temp.path / "wal";
WALManager::Config cfg;
cfg.walDirectory = walDir;
cfg.compressOldLogs = false;
cfg.enableGroupCommit = false;
cfg.syncInterval = 1;
WALManager manager(cfg);
REQUIRE(manager.initialize());
auto activeLogPath = manager.getCurrentLogPath();
REQUIRE(!activeLogPath.empty());
REQUIRE(std::filesystem::exists(activeLogPath));
auto oldInactiveLog = walDir / "wal_19990101_000000_1.log";
{
std::ofstream oldLog(oldInactiveLog.string(), std::ios::binary);
REQUIRE(oldLog.good());
}
auto oldTimestamp = std::filesystem::file_time_type::clock::now() - std::chrono::hours(24);
std::filesystem::last_write_time(oldInactiveLog, oldTimestamp);
REQUIRE(manager.pruneLogs(std::chrono::hours(0)));
CHECK(std::filesystem::exists(activeLogPath));
CHECK_FALSE(std::filesystem::exists(oldInactiveLog));
REQUIRE(manager.shutdown());
}
TEST_CASE("WALManager transaction lifecycle updates stats and enforces state",
"[unit][wal][manager][transaction]") {
TempDir temp;
auto walDir = temp.path / "wal";
WALManager::Config cfg;
cfg.walDirectory = walDir;
cfg.compressOldLogs = false;
cfg.enableGroupCommit = false;
cfg.syncInterval = 1;
WALManager manager(cfg);
REQUIRE(manager.initialize());
auto txn = manager.beginTransaction();
REQUIRE(txn);
CHECK(txn->getState() == Transaction::State::Active);
CHECK(manager.getStats().activeTransactions == 1);
REQUIRE(txn->storeBlock(makeHash('a'), 123, 1));
REQUIRE(txn->commit());
CHECK(txn->getState() == Transaction::State::Committed);
CHECK(manager.getStats().activeTransactions == 0);
auto secondCommit = txn->commit();
REQUIRE_FALSE(secondCommit);
CHECK(secondCommit.error().code == ErrorCode::InvalidOperation);
auto addAfterCommit = txn->deleteBlock(makeHash('b'));
REQUIRE_FALSE(addAfterCommit);
CHECK(addAfterCommit.error().code == ErrorCode::InvalidOperation);
{
auto autoRollbackTxn = manager.beginTransaction();
REQUIRE(autoRollbackTxn);
REQUIRE(autoRollbackTxn->updateReference(makeHash('c'), 2));
CHECK(manager.getStats().activeTransactions == 1);
}
CHECK(manager.getStats().activeTransactions == 0);
REQUIRE(manager.shutdown());
}
TEST_CASE("WALManager initialize preserves nonempty segments with a zero prefix",
"[unit][wal][manager]") {
TempDir temp;
const auto walDir = temp.path / "wal";
std::filesystem::create_directories(walDir);
const auto damagedLog = walDir / "wal_20240101_010159_150.log";
const auto size = GENERATE(std::size_t{3}, std::size_t{5}, std::size_t{64 * 1024},
std::size_t{128 * 1024 + 1});
std::string original(size, '\0');
original.back() = 'x'; // Include short reads, buffer boundaries, and a trailing partial block.
{
std::ofstream out(damagedLog, std::ios::binary);
out.write(original.data(), static_cast<std::streamsize>(original.size()));
REQUIRE(out.good());
}
WALManager::Config cfg;
cfg.walDirectory = walDir;
cfg.compressOldLogs = false;
cfg.enableGroupCommit = false;
WALManager manager(cfg);
REQUIRE(manager.initialize());
CHECK(manager.getCurrentSequence() == 150);
REQUIRE(manager.shutdown());
REQUIRE(std::filesystem::exists(damagedLog));
REQUIRE(std::filesystem::file_size(damagedLog) == original.size());
std::ifstream in(damagedLog, std::ios::binary);
std::string preserved(original.size(), '\0');
in.read(preserved.data(), static_cast<std::streamsize>(preserved.size()));
REQUIRE(in.good());
CHECK(preserved == original);
}
TEST_CASE("WALManager initialize rejects an unconfigured directory", "[unit][wal][manager]") {
WALManager::Config cfg;
cfg.enableGroupCommit = false;
REQUIRE(cfg.walDirectory.empty());
WALManager manager(cfg);
auto result = manager.initialize();
REQUIRE_FALSE(result);
CHECK(result.error().code == ErrorCode::InvalidArgument);
}
TEST_CASE("WALManager initialize prunes empty log segments but keeps their sequence",
"[unit][wal][manager]") {
TempDir temp;
auto walDir = temp.path / "wal";
std::filesystem::create_directories(walDir);
for (int i = 0; i < 50; ++i) {
std::ofstream empty((walDir / ("wal_20240101_0101" + std::to_string(10 + i) + "_" +
std::to_string(100 + i) + ".log"))
.string(),
std::ios::binary);
REQUIRE(empty.good());
}
{
// A killed writer leaves the mmap preallocation behind: 1 MiB of zeros, no entry.
std::ofstream preallocated((walDir / "wal_20240101_010159_150.log").string(),
std::ios::binary);
std::string zeros(1024 * 1024, '\0');
preallocated.write(zeros.data(), static_cast<std::streamsize>(zeros.size()));
REQUIRE(preallocated.good());
}
{
std::ofstream nonEmpty((walDir / "wal_20240101_010200_7.log").string(), std::ios::binary);
nonEmpty << "not-a-real-entry";
REQUIRE(nonEmpty.good());
}
WALManager::Config cfg;
cfg.walDirectory = walDir;
cfg.compressOldLogs = false;
cfg.enableGroupCommit = false;
WALManager manager(cfg);
REQUIRE(manager.initialize());
CHECK(manager.getCurrentSequence() == 150);
std::size_t logFiles = 0;
bool nonEmptySurvived = false;
for (const auto& entry : std::filesystem::directory_iterator(walDir)) {
if (entry.path().extension() == ".log") {
++logFiles;
if (entry.path().filename() == "wal_20240101_010200_7.log") {
nonEmptySurvived = true;
}
}
}
CHECK(nonEmptySurvived);
// The non-empty segment plus the freshly opened active log.
CHECK(logFiles == 2);
REQUIRE(manager.shutdown());
}