deb (99651968) · Commits · tuxist / mediadb

build_output.log

0 → 100644

+16 −0

Original line number	Diff line number	Diff line
		[1/16] Building CXX object CMakeFiles/mediadb_test.dir/src/cluster.cpp.o
		[2/16] Building CXX object CMakeFiles/mediadb_perf_test.dir/src/cluster.cpp.o
		[3/16] Building CXX object CMakeFiles/mediadb.dir/src/cluster.cpp.o
		[4/16] Building CXX object CMakeFiles/mediadb_cluster_perf_test.dir/src/cluster.cpp.o
		[5/16] Building CXX object CMakeFiles/mediadb.dir/src/main.cpp.o
		[6/16] Building CXX object CMakeFiles/mediadb.dir/src/server.cpp.o
		[7/16] Building CXX object CMakeFiles/mediadb_cluster_perf_test.dir/test/mediadb_cluster_perf_test.cpp.o
		[8/16] Building CXX object CMakeFiles/mediadb_perf_test.dir/src/backend.cpp.o
		[9/16] Building CXX object CMakeFiles/mediadb_test.dir/src/backend.cpp.o
		[10/16] Building CXX object CMakeFiles/mediadb.dir/src/backend.cpp.o
		[11/16] Building CXX object CMakeFiles/mediadb.dir/src/app.cpp.o
		[12/16] Linking CXX executable mediadb_perf_test
		[13/16] Linking CXX executable mediadb_test
		[14/16] Building CXX object CMakeFiles/mediadb_cluster_perf_test.dir/src/backend.cpp.o
		[15/16] Linking CXX executable mediadb
		[16/16] Linking CXX executable mediadb_cluster_perf_test

+5 −0

Original line number	Diff line number	Diff line
		@@ -3,6 +3,11 @@ mediadb (20260422+59) unstable; urgency=low
		* Fix stuck import state: clean up orphaned H2 import sessions on
		client disconnect so import_running_ is correctly reset.
		* Show import status (In Progress / Idle) on cluster status page.
		* Fix false DEGRADED during import: health probe retries once before
		counting a peer as offline; suppress DEGRADED transition while
		import_active_ is set so busy peers are not marked as down.
		* Notify Cluster of import state via setImportRunning() so the health
		loop can tolerate transient probe failures during replication.

		-- Jan Koester <jan.koester@tuxist.de> Wed, 22 Apr 2026 00:00:00 +0200

+4 −0

Original line number	Diff line number	Diff line
		@@ -448,6 +448,7 @@ HttpResponse App::start_import(const std::uint8_t* data, std::size_t len) {
		import_ok_.store(false);
		import_error_.clear();
		}
		if (g_Cluster) g_Cluster->setImportRunning(true);

		std::cerr << "[import] starting synchronous import (" << len << " bytes)" << "\n";
		bool ok = db_.import_db_from_buffer(data, len);
		@@ -477,6 +478,7 @@ HttpResponse App::start_import_file(const std::string& path) {
		import_ok_.store(false);
		import_error_.clear();
		}
		if (g_Cluster) g_Cluster->setImportRunning(true);

		std::cerr << "[import] starting file-based import from " << path << "\n";
		bool ok = db_.import_db(path);
		@@ -510,6 +512,7 @@ std::unique_ptr<ImportSession> App::begin_import() {
		import_ok_.store(false);
		import_error_.clear();
		}
		if (g_Cluster) g_Cluster->setImportRunning(true);
		auto session = db_.begin_import();
		if (!session) {
		std::cerr << "[import] begin_import: db_.begin_import() returned null\n";
		@@ -525,6 +528,7 @@ void App::finish_import(bool ok, const std::string& error) {
		import_ok_.store(ok);
		if (!ok && !error.empty()) import_error_ = error;
		import_running_.store(false);
		if (g_Cluster) g_Cluster->setImportRunning(false);
		}

		// ---- main router ----

+24 −5

Original line number	Diff line number	Diff line
		@@ -500,9 +500,17 @@ void Cluster::health_loop() {
		continue;
		}
		std::vector<uint64_t> dummy;
		if (pclient_->list_groups_on_node(it->second, dummy)) {
		++online;
		} else {
		// Retry once after short delay — peers may be busy
		// during import or scrub and miss the first probe
		std::this_thread::sleep_for(std::chrono::milliseconds(500));
		dummy.clear();
		if (pclient_->list_groups_on_node(it->second, dummy))
		++online;
		}
		}
		server_->set_peer_status_callback([online, total]() -> std::pair<uint32_t, uint32_t> {
		return {online, total};
		});
		@@ -562,11 +570,22 @@ void Cluster::health_loop() {
		}).detach();
		}
		} else if (online >= k) {
		// During active imports, peers may be busy handling
		// replication traffic and miss health probes. Suppress
		// the DEGRADED transition so it doesn't trigger a
		// recovery-scrub when the import finishes normally.
		if (!import_active_.load()) {
		if (!degraded_) {
		std::cerr << "[CLUSTER] DEGRADED — " << online << "/"
		<< n << " nodes online\n";
		}
		degraded_ = true;
		} else {
		if (!degraded_) {
		std::cerr << "[CLUSTER] probe missed " << (n - online) << " node(s)"
		<< " during import — suppressing DEGRADED\n";
		}
		}
		critical_ = false;
		healthy_cycles = 0;
		} else {

+7 −0

Original line number	Diff line number	Diff line
		@@ -176,11 +176,18 @@ public:

		bool isScrubRunning() const { return scrub_running_.load(); }

		// External flag: set by App when an import is in progress, so the
		// health loop can tolerate transient probe failures (peers busy
		// handling import replication traffic).
		void setImportRunning(bool v) { import_active_.store(v); }
		bool isImportRunning() const { return import_active_.load(); }

		private:
		ClusterConfig cfg_;
		std::atomic<bool> running_{false};
		std::atomic<bool> degraded_{false};
		std::atomic<bool> critical_{false};
		std::atomic<bool> import_active_{false};
		std::thread server_thread_;
		std::thread health_thread_;
		std::mutex scrub_mutex_;