diff options
author | Eelco Dolstra <eelco.dolstra@logicblox.com> | 2016-09-14T14·00+0200 |
---|---|---|
committer | Eelco Dolstra <eelco.dolstra@logicblox.com> | 2016-09-14T14·36+0200 |
commit | 90ad02bf626b885a5dd8967894e2eafc953bdf92 (patch) | |
tree | 7af8764fe95e2093f9c99fc5827a2f9ffde8dd5d /src/libstore/download.cc | |
parent | a75d11a7e6984b3df15da9677fbd49ee8de7a9c3 (diff) |
Enable HTTP/2 support
The binary cache store can now use HTTP/2 to do lookups. This is much more efficient than HTTP/1.1 due to multiplexing: we can issue many requests in parallel over a single TCP connection. Thus it's no longer necessary to use a bunch of concurrent TCP connections (25 by default). For example, downloading 802 .narinfo files from https://cache.nixos.org/, using a single TCP connection, takes 11.8s with HTTP/1.1, but only 0.61s with HTTP/2. This did require a fairly substantial rewrite of the Downloader class to use the curl multi interface, because otherwise curl wouldn't be able to do multiplexing for us. As a bonus, we get connection reuse even with HTTP/1.1. All downloads are now handled by a single worker thread. Clients call Downloader::enqueueDownload() to tell the worker thread to start the download, getting a std::future to the result.
Diffstat (limited to 'src/libstore/download.cc')
-rw-r--r-- | src/libstore/download.cc | 550 |
1 files changed, 379 insertions, 171 deletions
diff --git a/src/libstore/download.cc b/src/libstore/download.cc index ed7e124d25f4..2aca0a975bfe 100644 --- a/src/libstore/download.cc +++ b/src/libstore/download.cc @@ -5,10 +5,15 @@ #include "store-api.hh" #include "archive.hh" +#include <unistd.h> +#include <fcntl.h> + #include <curl/curl.h> #include <iostream> #include <thread> +#include <cmath> +#include <random> namespace nix { @@ -30,225 +35,428 @@ std::string resolveUri(const std::string & uri) struct CurlDownloader : public Downloader { - CURL * curl; - ref<std::string> data; - string etag, status, expectedETag, effectiveUrl; - - struct curl_slist * requestHeaders; + CURLM * curlm = 0; - bool showProgress; - double prevProgressTime{0}, startTime{0}; - unsigned int moveBack{1}; + std::random_device rd; + std::mt19937 mt19937; - size_t writeCallback(void * contents, size_t size, size_t nmemb) + struct DownloadItem : public std::enable_shared_from_this<DownloadItem> { - size_t realSize = size * nmemb; - data->append((char *) contents, realSize); - return realSize; - } + CurlDownloader & downloader; + DownloadRequest request; + DownloadResult result; + bool done = false; // whether the promise has been set + std::promise<DownloadResult> promise; + CURL * req = 0; + bool active = false; // whether the handle has been added to the multi object + std::string status; + + bool showProgress = false; + double prevProgressTime{0}, startTime{0}; + unsigned int moveBack{1}; + + unsigned int attempt = 0; + + /* Don't start this download until the specified time point + has been reached. */ + std::chrono::steady_clock::time_point embargo; + + struct curl_slist * requestHeaders = 0; + + DownloadItem(CurlDownloader & downloader, const DownloadRequest & request) + : downloader(downloader), request(request) + { + showProgress = + request.showProgress == DownloadRequest::yes || + (request.showProgress == DownloadRequest::automatic && isatty(STDERR_FILENO)); + + if (!request.expectedETag.empty()) + requestHeaders = curl_slist_append(requestHeaders, ("If-None-Match: " + request.expectedETag).c_str()); + } - static size_t writeCallbackWrapper(void * contents, size_t size, size_t nmemb, void * userp) - { - return ((CurlDownloader *) userp)->writeCallback(contents, size, nmemb); - } + ~DownloadItem() + { + if (req) { + if (active) + curl_multi_remove_handle(downloader.curlm, req); + curl_easy_cleanup(req); + } + if (requestHeaders) curl_slist_free_all(requestHeaders); + try { + if (!done) + fail(DownloadError(Transient, format("download of ‘%s’ was interrupted") % request.uri)); + } catch (...) { + ignoreException(); + } + } - size_t headerCallback(void * contents, size_t size, size_t nmemb) - { - size_t realSize = size * nmemb; - string line = string((char *) contents, realSize); - printMsg(lvlVomit, format("got header: %1%") % trim(line)); - if (line.compare(0, 5, "HTTP/") == 0) { // new response starts - etag = ""; - auto ss = tokenizeString<vector<string>>(line, " "); - status = ss.size() >= 2 ? ss[1] : ""; - } else { - auto i = line.find(':'); - if (i != string::npos) { - string name = trim(string(line, 0, i)); - if (name == "ETag") { // FIXME: case - etag = trim(string(line, i + 1)); - /* Hack to work around a GitHub bug: it sends - ETags, but ignores If-None-Match. So if we get - the expected ETag on a 200 response, then shut - down the connection because we already have the - data. */ - printMsg(lvlDebug, format("got ETag: %1%") % etag); - if (etag == expectedETag && status == "200") { - printMsg(lvlDebug, format("shutting down on 200 HTTP response with expected ETag")); - return 0; + template<class T> + void fail(const T & e) + { + promise.set_exception(std::make_exception_ptr(e)); + done = true; + } + + size_t writeCallback(void * contents, size_t size, size_t nmemb) + { + size_t realSize = size * nmemb; + result.data->append((char *) contents, realSize); + return realSize; + } + + static size_t writeCallbackWrapper(void * contents, size_t size, size_t nmemb, void * userp) + { + return ((DownloadItem *) userp)->writeCallback(contents, size, nmemb); + } + + size_t headerCallback(void * contents, size_t size, size_t nmemb) + { + size_t realSize = size * nmemb; + std::string line((char *) contents, realSize); + printMsg(lvlVomit, format("got header for ‘%s’: %s") % request.uri % trim(line)); + if (line.compare(0, 5, "HTTP/") == 0) { // new response starts + result.etag = ""; + auto ss = tokenizeString<vector<string>>(line, " "); + status = ss.size() >= 2 ? ss[1] : ""; + result.data = std::make_shared<std::string>(); + } else { + auto i = line.find(':'); + if (i != string::npos) { + string name = toLower(trim(string(line, 0, i))); + if (name == "etag") { + result.etag = trim(string(line, i + 1)); + /* Hack to work around a GitHub bug: it sends + ETags, but ignores If-None-Match. So if we get + the expected ETag on a 200 response, then shut + down the connection because we already have the + data. */ + if (result.etag == request.expectedETag && status == "200") { + debug(format("shutting down on 200 HTTP response with expected ETag")); + return 0; + } } } } + return realSize; } - return realSize; - } - static size_t headerCallbackWrapper(void * contents, size_t size, size_t nmemb, void * userp) - { - return ((CurlDownloader *) userp)->headerCallback(contents, size, nmemb); - } + static size_t headerCallbackWrapper(void * contents, size_t size, size_t nmemb, void * userp) + { + return ((DownloadItem *) userp)->headerCallback(contents, size, nmemb); + } - int progressCallback(double dltotal, double dlnow) - { - if (showProgress) { - double now = getTime(); - if (prevProgressTime <= now - 1) { - string s = (format(" [%1$.0f/%2$.0f KiB, %3$.1f KiB/s]") - % (dlnow / 1024.0) - % (dltotal / 1024.0) - % (now == startTime ? 0 : dlnow / 1024.0 / (now - startTime))).str(); - std::cerr << "\e[" << moveBack << "D" << s; - moveBack = s.size(); + int progressCallback(double dltotal, double dlnow) + { + if (showProgress) { + double now = getTime(); + if (prevProgressTime <= now - 1) { + string s = (format(" [%1$.0f/%2$.0f KiB, %3$.1f KiB/s]") + % (dlnow / 1024.0) + % (dltotal / 1024.0) + % (now == startTime ? 0 : dlnow / 1024.0 / (now - startTime))).str(); + std::cerr << "\e[" << moveBack << "D" << s; + moveBack = s.size(); + std::cerr.flush(); + prevProgressTime = now; + } + } + return _isInterrupted; + } + + static int progressCallbackWrapper(void * userp, double dltotal, double dlnow, double ultotal, double ulnow) + { + return ((DownloadItem *) userp)->progressCallback(dltotal, dlnow); + } + + void init() + { + // FIXME: handle parallel downloads. + if (showProgress) { + std::cerr << (format("downloading ‘%1%’... ") % request.uri); std::cerr.flush(); - prevProgressTime = now; + startTime = getTime(); } + + if (!req) req = curl_easy_init(); + + curl_easy_reset(req); + curl_easy_setopt(req, CURLOPT_URL, request.uri.c_str()); + curl_easy_setopt(req, CURLOPT_FOLLOWLOCATION, 1L); + curl_easy_setopt(req, CURLOPT_NOSIGNAL, 1); + curl_easy_setopt(req, CURLOPT_USERAGENT, ("Nix/" + nixVersion).c_str()); + curl_easy_setopt(req, CURLOPT_PIPEWAIT, 1); + curl_easy_setopt(req, CURLOPT_HTTP_VERSION, CURL_HTTP_VERSION_2TLS); + curl_easy_setopt(req, CURLOPT_WRITEFUNCTION, DownloadItem::writeCallbackWrapper); + curl_easy_setopt(req, CURLOPT_WRITEDATA, this); + curl_easy_setopt(req, CURLOPT_HEADERFUNCTION, DownloadItem::headerCallbackWrapper); + curl_easy_setopt(req, CURLOPT_HEADERDATA, this); + + curl_easy_setopt(req, CURLOPT_PROGRESSFUNCTION, progressCallbackWrapper); + curl_easy_setopt(req, CURLOPT_PROGRESSDATA, this); + curl_easy_setopt(req, CURLOPT_NOPROGRESS, 0); + + curl_easy_setopt(req, CURLOPT_HTTPHEADER, requestHeaders); + + if (request.head) + curl_easy_setopt(req, CURLOPT_NOBODY, 1); + + if (request.verifyTLS) + curl_easy_setopt(req, CURLOPT_CAINFO, getEnv("SSL_CERT_FILE", "/etc/ssl/certs/ca-certificates.crt").c_str()); + else { + curl_easy_setopt(req, CURLOPT_SSL_VERIFYPEER, 0); + curl_easy_setopt(req, CURLOPT_SSL_VERIFYHOST, 0); + } + + result.data = std::make_shared<std::string>(); } - return _isInterrupted; - } - static int progressCallbackWrapper(void * userp, double dltotal, double dlnow, double ultotal, double ulnow) - { - return ((CurlDownloader *) userp)->progressCallback(dltotal, dlnow); - } + void finish(CURLcode code) + { + if (showProgress) + //std::cerr << "\e[" << moveBack << "D\e[K\n"; + std::cerr << "\n"; - CurlDownloader() - : data(make_ref<std::string>()) - { - requestHeaders = 0; + long httpStatus = 0; + curl_easy_getinfo(req, CURLINFO_RESPONSE_CODE, &httpStatus); - curl = curl_easy_init(); - if (!curl) throw nix::Error("unable to initialize curl"); - } + char * effectiveUrlCStr; + curl_easy_getinfo(req, CURLINFO_EFFECTIVE_URL, &effectiveUrlCStr); + if (effectiveUrlCStr) + result.effectiveUrl = effectiveUrlCStr; - ~CurlDownloader() - { - if (curl) curl_easy_cleanup(curl); - if (requestHeaders) curl_slist_free_all(requestHeaders); - } + debug(format("finished download of ‘%s’; curl status = %d, HTTP status = %d, body = %d bytes") + % request.uri % code % httpStatus % (result.data ? result.data->size() : 0)); - bool fetch(const string & url, const DownloadOptions & options) - { - showProgress = - options.showProgress == DownloadOptions::yes || - (options.showProgress == DownloadOptions::automatic && isatty(STDERR_FILENO)); + if (code == CURLE_WRITE_ERROR && result.etag == request.expectedETag) { + code = CURLE_OK; + httpStatus = 304; + } + + if (code == CURLE_OK && + (httpStatus == 200 || httpStatus == 304 || httpStatus == 226 /* FTP */ || httpStatus == 0 /* other protocol */)) + { + result.cached = httpStatus == 304; + promise.set_value(result); + done = true; + } else { + Error err = + (httpStatus == 404 || code == CURLE_FILE_COULDNT_READ_FILE) ? NotFound : + httpStatus == 403 ? Forbidden : + (httpStatus == 408 || httpStatus == 500 || httpStatus == 503 + || httpStatus == 504 || httpStatus == 522 || httpStatus == 524 + || code == CURLE_COULDNT_RESOLVE_HOST) ? Transient : + Misc; - curl_easy_reset(curl); + attempt++; - curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L); - curl_easy_setopt(curl, CURLOPT_USERAGENT, ("Nix/" + nixVersion).c_str()); - curl_easy_setopt(curl, CURLOPT_FAILONERROR, 1); + auto exc = + httpStatus != 0 + ? DownloadError(err, format("unable to download ‘%s’: HTTP error %d") % request.uri % httpStatus) + : DownloadError(err, format("unable to download ‘%s’: %s (%d)") % request.uri % curl_easy_strerror(code) % code); + + /* If this is a transient error, then maybe retry the + download after a while. */ + if (err == Transient && attempt < request.tries) { + int ms = request.baseRetryTimeMs * std::pow(2.0f, attempt - 1 + std::uniform_real_distribution<>(0.0, 0.5)(downloader.mt19937)); + printMsg(lvlError, format("warning: %s; retrying in %d ms") % exc.what() % ms); + embargo = std::chrono::steady_clock::now() + std::chrono::milliseconds(ms); + downloader.enqueueItem(shared_from_this()); + } + else + fail(exc); + } + } + }; - curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, writeCallbackWrapper); - curl_easy_setopt(curl, CURLOPT_WRITEDATA, (void *) this); + struct State + { + bool quit = false; + std::vector<std::shared_ptr<DownloadItem>> incoming; + }; - curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, headerCallbackWrapper); - curl_easy_setopt(curl, CURLOPT_HEADERDATA, (void *) this); + Sync<State> state_; - curl_easy_setopt(curl, CURLOPT_PROGRESSFUNCTION, progressCallbackWrapper); - curl_easy_setopt(curl, CURLOPT_PROGRESSDATA, (void *) this); - curl_easy_setopt(curl, CURLOPT_NOPROGRESS, 0); + /* We can't use a std::condition_variable to wake up the curl + thread, because it only monitors file descriptors. So use a + pipe instead. */ + Pipe wakeupPipe; - curl_easy_setopt(curl, CURLOPT_NOSIGNAL, 1); + std::thread workerThread; - curl_easy_setopt(curl, CURLOPT_URL, url.c_str()); + CurlDownloader() + { + static std::once_flag globalInit; + std::call_once(globalInit, curl_global_init, CURL_GLOBAL_ALL); - if (options.verifyTLS) - curl_easy_setopt(curl, CURLOPT_CAINFO, getEnv("SSL_CERT_FILE", "/etc/ssl/certs/ca-certificates.crt").c_str()); - else { - curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0); - curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0); - } + curlm = curl_multi_init(); - data = make_ref<std::string>(); + curl_multi_setopt(curlm, CURLMOPT_PIPELINING, CURLPIPE_MULTIPLEX); + curl_multi_setopt(curlm, CURLMOPT_MAX_TOTAL_CONNECTIONS, 25); // FIXME: configurable - if (requestHeaders) { - curl_slist_free_all(requestHeaders); - requestHeaders = 0; - } + wakeupPipe.create(); + fcntl(wakeupPipe.readSide.get(), F_SETFL, O_NONBLOCK); + + workerThread = std::thread([&]() { workerThreadEntry(); }); + } - if (!options.expectedETag.empty()) { - this->expectedETag = options.expectedETag; - requestHeaders = curl_slist_append(requestHeaders, ("If-None-Match: " + options.expectedETag).c_str()); + ~CurlDownloader() + { + /* Signal the worker thread to exit. */ + { + auto state(state_.lock()); + state->quit = true; } + writeFull(wakeupPipe.writeSide.get(), " "); - curl_easy_setopt(curl, CURLOPT_HTTPHEADER, requestHeaders); + workerThread.join(); - if (options.head) - curl_easy_setopt(curl, CURLOPT_NOBODY, 1); + if (curlm) curl_multi_cleanup(curlm); + } - if (showProgress) { - std::cerr << (format("downloading ‘%1%’... ") % url); - std::cerr.flush(); - startTime = getTime(); - } + void workerThreadMain() + { + std::map<CURL *, std::shared_ptr<DownloadItem>> items; + + bool quit; + + std::chrono::steady_clock::time_point nextWakeup; + + while (!quit) { + checkInterrupt(); + + /* Let curl do its thing. */ + int running; + CURLMcode mc = curl_multi_perform(curlm, &running); + if (mc != CURLM_OK) + throw nix::Error(format("unexpected error from curl_multi_perform(): %s") % curl_multi_strerror(mc)); + + /* Set the promises of any finished requests. */ + CURLMsg * msg; + int left; + while ((msg = curl_multi_info_read(curlm, &left))) { + if (msg->msg == CURLMSG_DONE) { + auto i = items.find(msg->easy_handle); + assert(i != items.end()); + i->second->finish(msg->data.result); + curl_multi_remove_handle(curlm, i->second->req); + i->second->active = false; + items.erase(i); + } + } - CURLcode res = curl_easy_perform(curl); - if (showProgress) - //std::cerr << "\e[" << moveBack << "D\e[K\n"; - std::cerr << "\n"; - checkInterrupt(); - if (res == CURLE_WRITE_ERROR && etag == options.expectedETag) return false; - - long httpStatus = -1; - curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &httpStatus); - - if (res != CURLE_OK) { - Error err = - httpStatus == 404 ? NotFound : - httpStatus == 403 ? Forbidden : - (httpStatus == 408 || httpStatus == 500 || httpStatus == 503 - || httpStatus == 504 || httpStatus == 522 || httpStatus == 524 - || res == CURLE_COULDNT_RESOLVE_HOST) ? Transient : - Misc; - if (res == CURLE_HTTP_RETURNED_ERROR && httpStatus != -1) - throw DownloadError(err, format("unable to download ‘%s’: HTTP error %d") - % url % httpStatus); - else - throw DownloadError(err, format("unable to download ‘%s’: %s (%d)") - % url % curl_easy_strerror(res) % res); - } + /* Wait for activity, including wakeup events. */ + int numfds = 0; + struct curl_waitfd extraFDs[1]; + extraFDs[0].fd = wakeupPipe.readSide.get(); + extraFDs[0].events = CURL_WAIT_POLLIN; + extraFDs[0].revents = 0; + auto sleepTimeMs = + nextWakeup != std::chrono::steady_clock::time_point() + ? std::max(0, (int) std::chrono::duration_cast<std::chrono::milliseconds>(nextWakeup - std::chrono::steady_clock::now()).count()) + : 1000000000; + //printMsg(lvlVomit, format("download thread waiting for %d ms") % sleepTimeMs); + mc = curl_multi_wait(curlm, extraFDs, 1, sleepTimeMs, &numfds); + if (mc != CURLM_OK) + throw nix::Error(format("unexpected error from curl_multi_wait(): %s") % curl_multi_strerror(mc)); + + nextWakeup = std::chrono::steady_clock::time_point(); + + /* Add new curl requests from the incoming requests queue, + except for requests that are embargoed (waiting for a + retry timeout to expire). FIXME: should use a priority + queue for the embargoed items to prevent repeated O(n) + checks. */ + if (extraFDs[0].revents & CURL_WAIT_POLLIN) { + char buf[1024]; + auto res = read(extraFDs[0].fd, buf, sizeof(buf)); + if (res == -1 && errno != EINTR) + throw SysError("reading curl wakeup socket"); + } - char *effectiveUrlCStr; - curl_easy_getinfo(curl, CURLINFO_EFFECTIVE_URL, &effectiveUrlCStr); - if (effectiveUrlCStr) - effectiveUrl = effectiveUrlCStr; + std::vector<std::shared_ptr<DownloadItem>> incoming, embargoed; + auto now = std::chrono::steady_clock::now(); + + { + auto state(state_.lock()); + for (auto & item: state->incoming) { + if (item->embargo <= now) + incoming.push_back(item); + else { + embargoed.push_back(item); + if (nextWakeup == std::chrono::steady_clock::time_point() + || item->embargo < nextWakeup) + nextWakeup = item->embargo; + } + } + state->incoming = embargoed; + quit = state->quit; + } - if (httpStatus == 304) return false; + for (auto & item : incoming) { + debug(format("starting download of %s") % item->request.uri); + item->init(); + curl_multi_add_handle(curlm, item->req); + item->active = true; + items[item->req] = item; + } + } - return true; + debug("download thread shutting down"); } - DownloadResult download(string url, const DownloadOptions & options) override + void workerThreadEntry() { - size_t attempt = 0; + try { + workerThreadMain(); + } catch (Interrupted & e) { + } catch (std::exception & e) { + printMsg(lvlError, format("unexpected error in download thread: %s") % e.what()); + } - while (true) { - try { - DownloadResult res; - if (fetch(resolveUri(url), options)) { - res.cached = false; - res.data = data; - } else - res.cached = true; - res.effectiveUrl = effectiveUrl; - res.etag = etag; - return res; - } catch (DownloadError & e) { - attempt++; - if (e.error != Transient || attempt >= options.tries) throw; - auto ms = options.baseRetryTimeMs * (1 << (attempt - 1)); - printMsg(lvlError, format("warning: %s; retrying in %d ms") % e.what() % ms); - std::this_thread::sleep_for(std::chrono::milliseconds(ms)); - } + { + auto state(state_.lock()); + state->incoming.clear(); + state->quit = true; } } + + void enqueueItem(std::shared_ptr<DownloadItem> item) + { + { + auto state(state_.lock()); + if (state->quit) + throw nix::Error("cannot enqueue download request because the download thread is shutting down"); + state->incoming.push_back(item); + } + writeFull(wakeupPipe.writeSide.get(), " "); + } + + std::future<DownloadResult> enqueueDownload(const DownloadRequest & request) override + { + auto item = std::make_shared<DownloadItem>(*this, request); + enqueueItem(item); + return item->promise.get_future(); + } }; +ref<Downloader> getDownloader() +{ + static std::shared_ptr<Downloader> downloader; + static std::once_flag downloaderCreated; + std::call_once(downloaderCreated, [&]() { downloader = makeDownloader(); }); + return ref<Downloader>(downloader); +} + ref<Downloader> makeDownloader() { return make_ref<CurlDownloader>(); } +DownloadResult Downloader::download(const DownloadRequest & request) +{ + return enqueueDownload(request).get(); +} + Path Downloader::downloadCached(ref<Store> store, const string & url_, bool unpack, string name, const Hash & expectedHash, string * effectiveUrl) { auto url = resolveUri(url_); @@ -303,9 +511,9 @@ Path Downloader::downloadCached(ref<Store> store, const string & url_, bool unpa if (!skip) { try { - DownloadOptions options; - options.expectedETag = expectedETag; - auto res = download(url, options); + DownloadRequest request(url); + request.expectedETag = expectedETag; + auto res = download(request); if (effectiveUrl) *effectiveUrl = res.effectiveUrl; |