feat(node): enable DHT node cache

CMakeLists.txt

@@ -15,7 +15,7 @@ list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/cmake")
 ############################
 #  Find required packages  #
 ############################
-find_package(opendht 1.2.0 REQUIRED)
+find_package(opendht 1.8.2 REQUIRED)
 find_package(CURLpp REQUIRED)
 find_package(glibmm REQUIRED)
 find_package(B64 REQUIRED)

README.md

@@ -79,7 +79,7 @@ Milis Linux: mps kur dpaste (https://github.com/milisarge/malfs-milis/blob/maste
 
 ## Dependencies
 
-- [OpenDHT](https://github.com/savoirfairelinux/opendht/) (minimal version: 1.2.0)
+- [OpenDHT](https://github.com/savoirfairelinux/opendht/) (minimal version: 1.8.2)
 - [msgpack-c](https://github.com/msgpack/msgpack-c)
 - [gpgmepp](https://github.com/KDE/gpgmepp)
 - [json.hpp](https://github.com/nlohmann/json) (required version for CMake: 2.1.1)
@@ -89,6 +89,16 @@ Milis Linux: mps kur dpaste (https://github.com/milisarge/malfs-milis/blob/maste
 - [catch](https://github.com/catchorg/Catch2) for unit tests
 - getopt (util-linux)
 
+## Caching
+
+To avoid the multi-second DHT cold start on every invocation, `dpaste` caches
+the OpenDHT node state (including its routing table) on disk. The cache lives
+in `${XDG_CACHE_HOME}/dpaste` (usually `~/.cache/dpaste`) and can be relocated
+with the `DPASTE_CACHE_DIR` environment variable. The state is refreshed at
+the end of every run, allowing known peers to be reused and improving
+bootstrap resilience. The first run remains slow, and random-key lookups on
+the public DHT still cannot be eliminated.
+
 ## Pastebin over DHT
 
 A DHT is efficient and requires no infrastructure. In practice, you can always
@@ -115,4 +125,3 @@ not likely to be "down".
 
 - Simon Désaulniers <sim.desaulniers@gmail.com>
 - Adrien Béraud <adrien.beraud@savoirfairelinux.com>
-

configure.ac

@@ -18,7 +18,7 @@ AS_IF([test "x$enable_debug" = "xyes"],
 AC_PROG_CXX
 AC_PROG_RANLIB
 
-PKG_CHECK_MODULES([OpenDHT], [opendht >= 1.2])
+PKG_CHECK_MODULES([OpenDHT], [opendht >= 1.8.2])
 PKG_CHECK_MODULES([CURLPP], [curlpp])
 PKG_CHECK_MODULES([GLIBMM], [glibmm-2.4])
 
@@ -45,4 +45,3 @@ AM_COND_IF([DPASTE_TEST],
 AC_OUTPUT
 
 # vim: set ts=2 sw=2 tw=120 et :
-

src/bin.cpp

@@ -36,7 +36,7 @@ namespace dpaste {
 
 const constexpr uint8_t Bin::PROTO_VERSION;
 
-Bin::Bin() {
+Bin::Bin() : node_ready_(node.run()) {
     /* load dpaste config */
     auto config_file = conf::ConfigurationFile();
     config_file.load();
@@ -48,7 +48,6 @@ Bin::Bin() {
         conv >> port;
     }
 
-    node.run();
     http_client_ = std::make_unique<HttpClient>(conf_.at("host"), port);
 }
 
@@ -69,13 +68,16 @@ std::pair<bool, std::string> Bin::get(std::string&& code, bool no_decrypt) {
     std::vector<uint8_t> data {data_str.begin(), data_str.end()};
 
     /* if fail, then perform request from local node */
-    if (data.empty()) {
+    if (data.empty() and node_ready_) {
         /* get a pasted blob */
         auto values = node.get(lcode);
         if (not values.empty())
             data = values.front();
     }
 
+    if (data.empty() and not node_ready_)
+        return {false, ""};
+
     if (not data.empty()) {
         Packet p;
         try {
@@ -182,7 +184,7 @@ std::string Bin::paste(std::vector<uint8_t>&& data, std::unique_ptr<crypto::Para
     DPASTE_MSG("Pasting data...");
     auto bin_packet = p.serialize();
     auto success = http_client_->put(code, {bin_packet.begin(), bin_packet.end()});
-    if (not success)
+    if (not success and node_ready_)
         success = node.paste(code, std::move(bin_packet));
 
     return success ? DPASTE_URI_PREFIX+code+pwd  : "";
@@ -225,4 +227,3 @@ void Bin::Packet::deserialize(const std::vector<uint8_t>& pbuffer) {
 } /* dpaste  */
 
 /* vim:set et sw=4 ts=4 tw=120: */
-

src/bin.h

@@ -124,9 +124,9 @@ class Bin {
     /* transport */
     std::unique_ptr<HttpClient> http_client_ {};
     Node node {};
+    bool node_ready_ {false};
 };
 
 } /* dpaste */
 
 /* vim:set et sw=4 ts=4 tw=120: */
-

src/node.cpp

@@ -21,15 +21,61 @@
 #include <algorithm>
 #include <random>
 #include <future>
+#include <cstdlib>
+#include <filesystem>
+#include <iostream>
+#include <optional>
 
 #include <opendht.h>
+#include <glibmm.h>
 
 #include "node.h"
+#include "log.h"
 
 namespace dpaste {
 
 const constexpr char* Node::DPASTE_USER_TYPE;
 
+namespace {
+
+/**
+ * Directory holding the on-disk OpenDHT node state.
+ * Can be overridden with the DPASTE_CACHE_DIR environment variable
+ * (e.g. for tests). Defaults to ${XDG_CACHE_HOME}/dpaste.
+ */
+std::optional<std::filesystem::path> create_cache_dir() {
+    const char* env = std::getenv("DPASTE_CACHE_DIR");
+    const std::filesystem::path cache_dir = env and *env ? env : Glib::get_user_cache_dir() + "/dpaste";
+    std::error_code ec;
+    std::filesystem::create_directories(cache_dir, ec);
+    if (ec) {
+        DPASTE_MSG("Failed to create cache directory '%s': %s", cache_dir.string().c_str(), ec.message().c_str());
+        return std::nullopt;
+    }
+    return cache_dir;
+}
+
+} /* anonymous namespace */
+
+bool Node::run(uint16_t port, std::string bootstrap_hostname, std::string bootstrap_port) {
+    if (running_)
+        return true;
+
+    const auto cache_dir = create_cache_dir();
+    if (not cache_dir)
+        return false;
+    /* Ask OpenDHT to load its state (routing table) on start and save it on
+     * shutdown; this reuses known peers and improves bootstrap resilience. */
+    dht::DhtRunner::Config config;
+    config.dht_config.node_config.persist_path = (*cache_dir / "nodes").string();
+    config.threaded = true;
+    node_.run(port, config);
+
+    node_.bootstrap(bootstrap_hostname, bootstrap_port);
+    running_ = true;
+    return true;
+}
+
 bool Node::paste(const std::string& code, dht::Blob&& blob, dht::DoneCallbackSimple&& cb) {
     auto v = std::make_shared<dht::Value>(std::forward<dht::Blob>(blob));
     v->user_type = DPASTE_USER_TYPE;

src/node.h

@@ -54,15 +54,22 @@ class Node {
     static const constexpr char* DPASTE_USER_TYPE = "dpaste";
 
     Node() {}
-    virtual ~Node () {}
-
-    void run(uint16_t port = 0, std::string bootstrap_hostname = DEFAULT_BOOTSTRAP_NODE, std::string bootstrap_port = DEFAULT_BOOTSTRAP_PORT) {
+    virtual ~Node() {
+        /* Persist the node state (routing table) on disk so the next run can
+         * reuse known peers instead of cold-bootstrapping the DHT. */
         if (running_)
-            return;
-        node_.run(port, dht::crypto::generateIdentity(), true);
-        node_.bootstrap(bootstrap_hostname, bootstrap_port);
-        running_ = true;
-    };
+            stop();
+    }
+
+    /**
+     * Start the DHT node. Its routing state is cached on disk (see
+     * DPASTE_CACHE_DIR below) so subsequent runs can connect more quickly.
+     *
+     * @param port              Local port to bind (0 for random).
+     * @param bootstrap_hostname Hostname of the bootstrap node.
+     * @param bootstrap_port    Port of the bootstrap node.
+     */
+    bool run(uint16_t port = 0, std::string bootstrap_hostname = DEFAULT_BOOTSTRAP_NODE, std::string bootstrap_port = DEFAULT_BOOTSTRAP_PORT);
 
     void stop() {
         std::condition_variable cv;
@@ -124,4 +131,3 @@ class Node {
 };
 
 } /* dpaste */
-

tests/node.cpp

@@ -20,6 +20,10 @@
 
 #include <catch2/catch.hpp>
 
+#include <cstdlib>
+#include <stdexcept>
+#include <unistd.h>
+
 #include "tests.h"
 #include "node.h"
 
@@ -35,13 +39,58 @@ class PirateNodeTester {
     bool is_running(const dpaste::Node& n) const { return n.running_; }
 };
 
+class CacheFileGuard {
+    int fd_ {-1};
+    std::string path_;
+    bool had_previous_value_ {false};
+    std::string previous_value_;
+
+public:
+    CacheFileGuard() {
+        const char* previous = std::getenv("DPASTE_CACHE_DIR");
+        had_previous_value_ = previous != nullptr;
+        if (had_previous_value_)
+            previous_value_ = previous;
+
+        char path[] = "/tmp/dpaste-cache-XXXXXX";
+        fd_ = mkstemp(path);
+        if (fd_ == -1)
+            throw std::runtime_error("could not create temporary cache file");
+        path_ = path;
+
+        if (setenv("DPASTE_CACHE_DIR", path_.c_str(), 1) != 0) {
+            close(fd_);
+            unlink(path_.c_str());
+            throw std::runtime_error("could not set DPASTE_CACHE_DIR");
+        }
+    }
+
+    ~CacheFileGuard() {
+        if (had_previous_value_)
+            setenv("DPASTE_CACHE_DIR", previous_value_.c_str(), 1);
+        else
+            unsetenv("DPASTE_CACHE_DIR");
+        close(fd_);
+        unlink(path_.c_str());
+    }
+};
+
+TEST_CASE("Node refuses a cache path that is a regular file", "[Node][cache]") {
+    PirateNodeTester pt;
+    CacheFileGuard cache;
+
+    dpaste::Node node {};
+    REQUIRE_FALSE(node.run());
+    REQUIRE_FALSE(pt.is_running(node));
+}
+
 TEST_CASE("Node get/paste on DHT", "[Node][get][paste]") {
     PirateNodeTester pt;
 
     const std::string PIN = random_pin();
     std::vector<uint8_t> data = {0, 1, 2, 3, 4};
     dpaste::Node node {};
-    node.run();
+    REQUIRE(node.run());
 
     SECTION ( "pasting data {0,1,2,3,4}" ) {
         REQUIRE ( node.paste(PIN, std::vector<uint8_t> {data}) );
@@ -60,4 +109,3 @@ TEST_CASE("Node get/paste on DHT", "[Node][get][paste]") {
 } /* dpaste */
 
 /* vim: set ts=4 sw=4 tw=120 et :*/
-