Cache read component path resolutions

scripts/run_cli_tests.sh

@@ -75,5 +75,6 @@ if [[ "$action" == "run" ]]; then
 	scripts/test_mysql_persistent_pool.sh
 	scripts/test_log_timeliness.sh
 	scripts/test_raw_http_request_log.sh
+	scripts/test_component_resolution_ttl.sh
 	scripts/test_socket_activation.sh
 fi

scripts/test_component_resolution_ttl.sh

@@ -0,0 +1,101 @@
+#!/usr/bin/env bash
+set -euo pipefail
+cd "$(dirname "$0")/.."
+
+test_name="component-resolution-ttl-test-$$"
+site_directory="${UCE_TEST_SITE_DIRECTORY:-site}"
+worker_count="${UCE_TEST_WORKER_COUNT:-4}"
+if [[ -r /etc/uce/settings.cfg ]]; then
+	if [[ -z "${UCE_TEST_SITE_DIRECTORY:-}" ]]; then
+		configured_site_directory=$(awk -F= '/^[[:space:]]*SITE_DIRECTORY[[:space:]]*=/ {gsub(/^[[:space:]]+|[[:space:]]+$/, "", $2); print $2; exit}' /etc/uce/settings.cfg)
+		site_directory="${configured_site_directory:-$site_directory}"
+	fi
+	if [[ -z "${UCE_TEST_WORKER_COUNT:-}" ]]; then
+		configured_worker_count=$(awk -F= '/^[[:space:]]*WORKER_COUNT[[:space:]]*=/ {gsub(/^[[:space:]]+|[[:space:]]+$/, "", $2); print $2; exit}' /etc/uce/settings.cfg)
+		worker_count="${configured_worker_count:-$worker_count}"
+	fi
+fi
+http_host="${UCE_TEST_HTTP_HOST:-uce.openfu.com}"
+source_dir="$site_directory/$test_name"
+
+cleanup() {
+	rm -rf "$source_dir"
+}
+trap cleanup EXIT
+mkdir -p "$source_dir"
+
+printf '%s\n' 'RENDER(Request& context) { print("parent:", request_perf()["worker_pid"].to_string(), ":", component("chosen", context)); }' >"$source_dir/parent.uce"
+printf '%s\n' 'COMPONENT(Request& context) { print("a"); }' >"$source_dir/a.uce"
+printf '%s\n' 'COMPONENT(Request& context) { print("b"); }' >"$source_dir/b.uce"
+ln -s a.uce "$source_dir/chosen.uce"
+
+declare -A seen
+collect_workers() {
+	local method="$1"
+	local marker="$2"
+	local path="${3:-parent.uce}"
+	local deadline=$((SECONDS + 30))
+	seen=()
+	while (( SECONDS < deadline && ${#seen[@]} < worker_count )); do
+		local response status body pid
+		response=$(curl -sS --max-time 10 -X "$method" -H "Host: $http_host" -w $'\n%{http_code}' "http://127.0.0.1/$test_name/$path")
+		status=${response##*$'\n'}
+		body=${response%$'\n'*}
+		if [[ "$status" == "503" ]]; then
+			sleep 0.1
+			continue
+		fi
+		if [[ "$status" != "200" || "$body" != parent:*":$marker" ]]; then
+			echo "component resolution TTL probe failed: method=$method status=$status expected=$marker body=$body" >&2
+			exit 1
+		fi
+		pid=${body#parent:}
+		pid=${pid%%:*}
+		seen["$pid"]=1
+	done
+	if (( ${#seen[@]} != worker_count )); then
+		echo "component resolution TTL probe reached ${#seen[@]}/$worker_count workers" >&2
+		exit 1
+	fi
+}
+
+collect_missing_workers() {
+	local deadline=$((SECONDS + 30))
+	seen=()
+	while (( SECONDS < deadline && ${#seen[@]} < worker_count )); do
+		local response status body pid
+		response=$(curl -sS --max-time 10 -H "Host: $http_host" -w $'\n%{http_code}' "http://127.0.0.1/$test_name/missing.uce")
+		status=${response##*$'\n'}
+		body=${response%$'\n'*}
+		if [[ "$status" == "503" ]]; then
+			sleep 0.1
+			continue
+		fi
+		if [[ "$status" != "200" || "$body" != parent:* || "$body" != *"component not found: later"* ]]; then
+			echo "missing component TTL probe failed: status=$status body=$body" >&2
+			exit 1
+		fi
+		pid=${body#parent:}
+		pid=${pid%%:*}
+		seen["$pid"]=1
+	done
+	if (( ${#seen[@]} != worker_count )); then
+		echo "missing component TTL probe reached ${#seen[@]}/$worker_count workers" >&2
+		exit 1
+	fi
+}
+
+collect_workers GET a
+ln -sfn b.uce "$source_dir/chosen.uce"
+collect_workers GET a
+collect_workers POST b
+collect_workers GET a
+sleep 10.2
+collect_workers GET b
+
+printf '%s\n' 'RENDER(Request& context) { print("parent:", request_perf()["worker_pid"].to_string(), ":", component("later", context)); }' >"$source_dir/missing.uce"
+collect_missing_workers
+printf '%s\n' 'COMPONENT(Request& context) { print("created"); }' >"$source_dir/later.uce"
+collect_workers GET created missing.uce
+
+echo "component resolution TTL passed"

src/wasm/worker.cpp

@@ -1167,7 +1167,13 @@ private:
 		bool stale = false;
 		String source_generation;
 	};
+	struct ComponentResolutionState
+	{
+		std::chrono::steady_clock::time_point checked_at;
+		String resolved;
+	};
 	std::map<String, ComponentFreshnessState> component_freshness;
+	std::map<String, ComponentResolutionState> component_resolutions;
 
 	static String cached_wasm_path(const String& wasm_path)
 	{
@@ -2412,6 +2418,9 @@ private:
 		String error;
 		size_t unit_index = 0;
 		bool loaded_reuse = false;
+		String method = context ? to_upper(trim(context->params["REQUEST_METHOD"])) : String("");
+		bool cacheable_read = method == "GET" || method == "HEAD";
+		bool read_request = cacheable_read || method == "OPTIONS";
 		auto loaded = units_by_source.find(file_name);
 		if(loaded == units_by_source.end())
 		{
@@ -2427,7 +2436,30 @@ private:
 			component_loaded_reuse_count++;
 		}
 		else
-			resolved = resolve_source_path(file_name, current_unit);
+		{
+			String cache_key = entry_dir + "\t" + resolve_key;
+			auto cached = worker.component_resolutions.find(cache_key);
+			auto now = std::chrono::steady_clock::now();
+			if(cacheable_read && cached != worker.component_resolutions.end() &&
+				std::chrono::duration_cast<std::chrono::seconds>(now - cached->second.checked_at).count() < 10)
+				resolved = cached->second.resolved;
+			else
+			{
+				resolved = resolve_source_path(file_name, current_unit);
+				if(cacheable_read && resolved != "")
+				{
+					if(cached == worker.component_resolutions.end() && worker.component_resolutions.size() >= 4096)
+					{
+						auto oldest = worker.component_resolutions.begin();
+						for(auto it = worker.component_resolutions.begin(); it != worker.component_resolutions.end(); ++it)
+							if(it->second.checked_at < oldest->second.checked_at)
+								oldest = it;
+						worker.component_resolutions.erase(oldest);
+					}
+					worker.component_resolutions[cache_key] = { now, resolved };
+				}
+			}
+		}
 		component_path_total_us += (u64)std::chrono::duration_cast<std::chrono::microseconds>(
 			std::chrono::steady_clock::now() - probe_start).count();
 		if(resolved == "")
@@ -2447,8 +2479,6 @@ private:
 			auto artifact_start = std::chrono::steady_clock::now();
 			bool artifact_exists = file_exists_host(worker.unit_wasm_path(resolved));
 			bool can_serve_stale = compiler_request_can_serve_stale_artifact(context);
-			String method = context ? to_upper(trim(context->params["REQUEST_METHOD"])) : String("");
-			bool read_request = method == "GET" || method == "HEAD" || method == "OPTIONS";
 			if(can_serve_stale && read_request && !component_source_generation_checked)
 			{
 				component_source_generation = compiler_source_generation(context);