nix-super/src/libstore/remote-store.cc

726 lines
20 KiB
C++
Raw Normal View History

#include "serialise.hh"
#include "util.hh"
2006-11-30 20:35:50 +02:00
#include "remote-store.hh"
#include "worker-protocol.hh"
2006-11-30 22:45:20 +02:00
#include "archive.hh"
#include "affinity.hh"
#include "globals.hh"
#include "derivations.hh"
#include "pool.hh"
2006-11-30 20:35:50 +02:00
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/socket.h>
#include <sys/un.h>
2014-12-09 13:16:27 +02:00
#include <errno.h>
#include <fcntl.h>
#include <unistd.h>
2016-02-23 17:40:16 +02:00
#include <cstring>
2006-11-30 20:35:50 +02:00
namespace nix {
Path readStorePath(Store & store, Source & from)
{
Path path = readString(from);
store.assertStorePath(path);
return path;
}
template<class T> T readStorePaths(Store & store, Source & from)
{
T paths = readStrings<T>(from);
for (auto & i : paths) store.assertStorePath(i);
return paths;
}
template PathSet readStorePaths(Store & store, Source & from);
template Paths readStorePaths(Store & store, Source & from);
/* TODO: Separate these store impls into different files, give them better names */
RemoteStore::RemoteStore(const Params & params)
: Store(params)
, connections(make_ref<Pool<Connection>>(
std::max(1, (int) maxConnections),
2017-03-03 20:35:34 +02:00
[this]() { return openConnectionWrapper(); },
[this](const ref<Connection> & r) {
return
r->to.good()
&& r->from.good()
&& std::chrono::duration_cast<std::chrono::seconds>(
std::chrono::steady_clock::now() - r->startTime).count() < maxConnectionAge;
}
))
2006-12-04 15:28:14 +02:00
{
}
2017-03-03 20:35:34 +02:00
ref<RemoteStore::Connection> RemoteStore::openConnectionWrapper()
{
if (failed)
throw Error("opening a connection to remote store '%s' previously failed", getUri());
2017-03-03 20:35:34 +02:00
try {
return openConnection();
} catch (...) {
failed = true;
throw;
}
}
UDSRemoteStore::UDSRemoteStore(const Params & params)
: Store(params)
, LocalFSStore(params)
, RemoteStore(params)
{
}
std::string UDSRemoteStore::getUri()
{
return "daemon";
}
ref<RemoteStore::Connection> UDSRemoteStore::openConnection()
{
auto conn = make_ref<Connection>();
/* Connect to a daemon that does the privileged work for us. */
2016-06-09 17:15:58 +03:00
conn->fd = socket(PF_UNIX, SOCK_STREAM
#ifdef SOCK_CLOEXEC
| SOCK_CLOEXEC
#endif
, 0);
2016-07-11 22:44:44 +03:00
if (!conn->fd)
throw SysError("cannot create Unix domain socket");
2016-07-11 22:44:44 +03:00
closeOnExec(conn->fd.get());
string socketPath = settings.nixDaemonSocketFile;
struct sockaddr_un addr;
addr.sun_family = AF_UNIX;
if (socketPath.size() + 1 >= sizeof(addr.sun_path))
throw Error(format("socket path '%1%' is too long") % socketPath);
strcpy(addr.sun_path, socketPath.c_str());
2012-07-31 00:13:25 +03:00
if (::connect(conn->fd.get(), (struct sockaddr *) &addr, sizeof(addr)) == -1)
throw SysError(format("cannot connect to daemon at '%1%'") % socketPath);
2016-07-11 22:44:44 +03:00
conn->from.fd = conn->fd.get();
conn->to.fd = conn->fd.get();
conn->startTime = std::chrono::steady_clock::now();
initConnection(*conn);
return conn;
}
void RemoteStore::initConnection(Connection & conn)
{
/* Send the magic greeting, check for the reply. */
try {
conn.to << WORKER_MAGIC_1;
conn.to.flush();
unsigned int magic = readInt(conn.from);
if (magic != WORKER_MAGIC_2) throw Error("protocol mismatch");
conn.from >> conn.daemonVersion;
if (GET_PROTOCOL_MAJOR(conn.daemonVersion) != GET_PROTOCOL_MAJOR(PROTOCOL_VERSION))
throw Error("Nix daemon protocol version not supported");
if (GET_PROTOCOL_MINOR(conn.daemonVersion) < 10)
throw Error("the Nix daemon version is too old");
conn.to << PROTOCOL_VERSION;
if (GET_PROTOCOL_MINOR(conn.daemonVersion) >= 14) {
int cpu = settings.lockCPU ? lockToCurrentCPU() : -1;
if (cpu != -1)
conn.to << 1 << cpu;
else
conn.to << 0;
}
if (GET_PROTOCOL_MINOR(conn.daemonVersion) >= 11)
conn.to << false;
conn.processStderr();
}
catch (Error & e) {
throw Error("cannot open connection to remote store '%s': %s", getUri(), e.what());
}
setOptions(conn);
2006-11-30 20:35:50 +02:00
}
void RemoteStore::setOptions(Connection & conn)
{
conn.to << wopSetOptions
2015-07-20 02:16:16 +03:00
<< settings.keepFailed
<< settings.keepGoing
<< settings.tryFallback
<< verbosity
<< settings.maxBuildJobs
<< settings.maxSilentTime
<< true
<< (settings.verboseBuild ? lvlError : lvlVomit)
<< 0 // obsolete log type
<< 0 /* obsolete print build trace */
<< settings.buildCores
<< settings.useSubstitutes;
2012-07-31 00:13:25 +03:00
if (GET_PROTOCOL_MINOR(conn.daemonVersion) >= 12) {
auto overrides = settings.getSettings(true);
conn.to << overrides.size();
2015-07-20 02:16:16 +03:00
for (auto & i : overrides)
conn.to << i.first << i.second;
2012-08-01 01:19:44 +03:00
}
conn.processStderr();
}
bool RemoteStore::isValidPathUncached(const Path & path)
2006-11-30 20:35:50 +02:00
{
auto conn(connections->get());
conn->to << wopIsValidPath << path;
conn->processStderr();
return readInt(conn->from);
2006-11-30 20:35:50 +02:00
}
PathSet RemoteStore::queryValidPaths(const PathSet & paths, SubstituteFlag maybeSubstitute)
{
auto conn(connections->get());
if (GET_PROTOCOL_MINOR(conn->daemonVersion) < 12) {
PathSet res;
2015-07-17 20:24:28 +03:00
for (auto & i : paths)
if (isValidPath(i)) res.insert(i);
return res;
} else {
conn->to << wopQueryValidPaths << paths;
conn->processStderr();
return readStorePaths<PathSet>(*this, conn->from);
}
}
PathSet RemoteStore::queryAllValidPaths()
{
auto conn(connections->get());
conn->to << wopQueryAllValidPaths;
conn->processStderr();
return readStorePaths<PathSet>(*this, conn->from);
}
PathSet RemoteStore::querySubstitutablePaths(const PathSet & paths)
{
auto conn(connections->get());
if (GET_PROTOCOL_MINOR(conn->daemonVersion) < 12) {
PathSet res;
2015-07-17 20:24:28 +03:00
for (auto & i : paths) {
conn->to << wopHasSubstitutes << i;
conn->processStderr();
if (readInt(conn->from)) res.insert(i);
}
return res;
} else {
conn->to << wopQuerySubstitutablePaths << paths;
conn->processStderr();
return readStorePaths<PathSet>(*this, conn->from);
}
2006-11-30 20:35:50 +02:00
}
download-from-binary-cache: parallelise fetching of NAR info files Getting substitute information using the binary cache substituter has non-trivial latency overhead. A package or NixOS system configuration can have hundreds of dependencies, and in the worst case (when the local info cache is empty) we have to do a separate HTTP request for each of these. If the ping time to the server is t, getting N info files will take tN seconds; e.g., with a ping time of 0.1s to nixos.org, sequentially downloading 1000 info files (a typical NixOS config) will take at least 100 seconds. To fix this problem, the binary cache substituter can now perform requests in parallel. This required changing the substituter interface to support a function querySubstitutablePathInfos() that queries multiple paths at the same time, and rewriting queryMissing() to take advantage of parallelism. (Due to local caching, parallelising queryMissing() is sufficient for most use cases, since it's almost always called before building a derivation and thus fills the local info cache.) For example, parallelism speeds up querying all 1056 paths in a particular NixOS system configuration from 116s to 2.6s. It works so well because the eccentricity of the top-level derivation in the dependency graph is only 9. So we only need 10 round-trips (when using an unlimited number of parallel connections) to get everything. Currently we do a maximum of 150 parallel connections to the server. Thus it's important that the binary cache server (e.g. nixos.org) has a high connection limit. Alternatively we could use HTTP pipelining, but WWW::Curl doesn't support it and libcurl has a hard-coded limit of 5 requests per pipeline.
2012-07-07 02:08:20 +03:00
void RemoteStore::querySubstitutablePathInfos(const PathSet & paths,
SubstitutablePathInfos & infos)
{
if (paths.empty()) return;
auto conn(connections->get());
2012-07-31 00:13:25 +03:00
if (GET_PROTOCOL_MINOR(conn->daemonVersion) < 12) {
2012-07-31 00:13:25 +03:00
2015-07-17 20:24:28 +03:00
for (auto & i : paths) {
SubstitutablePathInfo info;
conn->to << wopQuerySubstitutablePathInfo << i;
conn->processStderr();
unsigned int reply = readInt(conn->from);
if (reply == 0) continue;
info.deriver = readString(conn->from);
if (info.deriver != "") assertStorePath(info.deriver);
info.references = readStorePaths<PathSet>(*this, conn->from);
info.downloadSize = readLongLong(conn->from);
info.narSize = readLongLong(conn->from);
2015-07-17 20:24:28 +03:00
infos[i] = info;
}
2012-07-31 00:13:25 +03:00
} else {
2012-07-31 00:13:25 +03:00
conn->to << wopQuerySubstitutablePathInfos << paths;
conn->processStderr();
size_t count = readNum<size_t>(conn->from);
for (size_t n = 0; n < count; n++) {
Path path = readStorePath(*this, conn->from);
SubstitutablePathInfo & info(infos[path]);
info.deriver = readString(conn->from);
if (info.deriver != "") assertStorePath(info.deriver);
info.references = readStorePaths<PathSet>(*this, conn->from);
info.downloadSize = readLongLong(conn->from);
info.narSize = readLongLong(conn->from);
}
2012-07-31 00:13:25 +03:00
download-from-binary-cache: parallelise fetching of NAR info files Getting substitute information using the binary cache substituter has non-trivial latency overhead. A package or NixOS system configuration can have hundreds of dependencies, and in the worst case (when the local info cache is empty) we have to do a separate HTTP request for each of these. If the ping time to the server is t, getting N info files will take tN seconds; e.g., with a ping time of 0.1s to nixos.org, sequentially downloading 1000 info files (a typical NixOS config) will take at least 100 seconds. To fix this problem, the binary cache substituter can now perform requests in parallel. This required changing the substituter interface to support a function querySubstitutablePathInfos() that queries multiple paths at the same time, and rewriting queryMissing() to take advantage of parallelism. (Due to local caching, parallelising queryMissing() is sufficient for most use cases, since it's almost always called before building a derivation and thus fills the local info cache.) For example, parallelism speeds up querying all 1056 paths in a particular NixOS system configuration from 116s to 2.6s. It works so well because the eccentricity of the top-level derivation in the dependency graph is only 9. So we only need 10 round-trips (when using an unlimited number of parallel connections) to get everything. Currently we do a maximum of 150 parallel connections to the server. Thus it's important that the binary cache server (e.g. nixos.org) has a high connection limit. Alternatively we could use HTTP pipelining, but WWW::Curl doesn't support it and libcurl has a hard-coded limit of 5 requests per pipeline.
2012-07-07 02:08:20 +03:00
}
}
void RemoteStore::queryPathInfoUncached(const Path & path,
std::function<void(std::shared_ptr<ValidPathInfo>)> success,
std::function<void(std::exception_ptr exc)> failure)
{
sync2async<std::shared_ptr<ValidPathInfo>>(success, failure, [&]() {
auto conn(connections->get());
conn->to << wopQueryPathInfo << path;
try {
conn->processStderr();
} catch (Error & e) {
// Ugly backwards compatibility hack.
if (e.msg().find("is not valid") != std::string::npos)
throw InvalidPath(e.what());
throw;
}
if (GET_PROTOCOL_MINOR(conn->daemonVersion) >= 17) {
bool valid; conn->from >> valid;
if (!valid) throw InvalidPath(format("path '%s' is not valid") % path);
}
auto info = std::make_shared<ValidPathInfo>();
info->path = path;
info->deriver = readString(conn->from);
if (info->deriver != "") assertStorePath(info->deriver);
info->narHash = Hash(readString(conn->from), htSHA256);
info->references = readStorePaths<PathSet>(*this, conn->from);
conn->from >> info->registrationTime >> info->narSize;
if (GET_PROTOCOL_MINOR(conn->daemonVersion) >= 16) {
conn->from >> info->ultimate;
info->sigs = readStrings<StringSet>(conn->from);
conn->from >> info->ca;
}
return info;
});
}
void RemoteStore::queryReferrers(const Path & path,
2006-11-30 20:35:50 +02:00
PathSet & referrers)
{
auto conn(connections->get());
conn->to << wopQueryReferrers << path;
conn->processStderr();
PathSet referrers2 = readStorePaths<PathSet>(*this, conn->from);
referrers.insert(referrers2.begin(), referrers2.end());
2006-11-30 20:35:50 +02:00
}
PathSet RemoteStore::queryValidDerivers(const Path & path)
{
auto conn(connections->get());
conn->to << wopQueryValidDerivers << path;
conn->processStderr();
return readStorePaths<PathSet>(*this, conn->from);
}
PathSet RemoteStore::queryDerivationOutputs(const Path & path)
{
auto conn(connections->get());
conn->to << wopQueryDerivationOutputs << path;
conn->processStderr();
return readStorePaths<PathSet>(*this, conn->from);
}
PathSet RemoteStore::queryDerivationOutputNames(const Path & path)
{
auto conn(connections->get());
conn->to << wopQueryDerivationOutputNames << path;
conn->processStderr();
return readStrings<PathSet>(conn->from);
}
Path RemoteStore::queryPathFromHashPart(const string & hashPart)
{
auto conn(connections->get());
conn->to << wopQueryPathFromHashPart << hashPart;
conn->processStderr();
Path path = readString(conn->from);
if (!path.empty()) assertStorePath(path);
return path;
}
void RemoteStore::addToStore(const ValidPathInfo & info, const ref<std::string> & nar,
RepairFlag repair, CheckSigsFlag checkSigs, std::shared_ptr<FSAccessor> accessor)
{
auto conn(connections->get());
if (GET_PROTOCOL_MINOR(conn->daemonVersion) < 18) {
conn->to << wopImportPaths;
StringSink sink;
sink << 1 // == path follows
;
assert(nar->size() % 8 == 0);
sink((unsigned char *) nar->data(), nar->size());
sink
<< exportMagic
<< info.path
<< info.references
<< info.deriver
<< 0 // == no legacy signature
<< 0 // == no path follows
;
StringSource source(*sink.s);
conn->processStderr(0, &source);
auto importedPaths = readStorePaths<PathSet>(*this, conn->from);
assert(importedPaths.size() <= 1);
}
else {
conn->to << wopAddToStoreNar
<< info.path << info.deriver << info.narHash.to_string(Base16, false)
<< info.references << info.registrationTime << info.narSize
<< info.ultimate << info.sigs << info.ca
<< repair << !checkSigs;
conn->to(*nar);
conn->processStderr();
}
}
Path RemoteStore::addToStore(const string & name, const Path & _srcPath,
bool recursive, HashType hashAlgo, PathFilter & filter, RepairFlag repair)
2006-11-30 20:35:50 +02:00
{
if (repair) throw Error("repairing is not supported when building through the Nix daemon");
auto conn(connections->get());
2012-07-31 00:13:25 +03:00
Path srcPath(absPath(_srcPath));
conn->to << wopAddToStore << name
2015-07-20 02:16:16 +03:00
<< ((hashAlgo == htSHA256 && recursive) ? 0 : 1) /* backwards compatibility hack */
<< (recursive ? 1 : 0)
<< printHashType(hashAlgo);
try {
conn->to.written = 0;
conn->to.warn = true;
connections->incCapacity();
2016-03-22 15:21:45 +02:00
dumpPath(srcPath, conn->to, filter);
connections->decCapacity();
conn->to.warn = false;
conn->processStderr();
} catch (SysError & e) {
/* Daemon closed while we were sending the path. Probably OOM
or I/O error. */
if (e.errNo == EPIPE)
try {
conn->processStderr();
} catch (EndOfFile & e) { }
throw;
}
return readStorePath(*this, conn->from);
2006-11-30 20:35:50 +02:00
}
Path RemoteStore::addTextToStore(const string & name, const string & s,
const PathSet & references, RepairFlag repair)
2006-11-30 20:35:50 +02:00
{
if (repair) throw Error("repairing is not supported when building through the Nix daemon");
auto conn(connections->get());
conn->to << wopAddTextToStore << name << s << references;
2012-07-31 00:13:25 +03:00
conn->processStderr();
return readStorePath(*this, conn->from);
2006-11-30 20:35:50 +02:00
}
void RemoteStore::buildPaths(const PathSet & drvPaths, BuildMode buildMode)
2006-11-30 20:35:50 +02:00
{
auto conn(connections->get());
conn->to << wopBuildPaths;
if (GET_PROTOCOL_MINOR(conn->daemonVersion) >= 13) {
conn->to << drvPaths;
if (GET_PROTOCOL_MINOR(conn->daemonVersion) >= 15)
conn->to << buildMode;
else
/* Old daemons did not take a 'buildMode' parameter, so we
need to validate it here on the client side. */
if (buildMode != bmNormal)
throw Error("repairing or checking is not supported when building through the Nix daemon");
} else {
/* For backwards compatibility with old daemons, strip output
identifiers. */
PathSet drvPaths2;
2015-07-17 20:24:28 +03:00
for (auto & i : drvPaths)
drvPaths2.insert(string(i, 0, i.find('!')));
conn->to << drvPaths2;
}
conn->processStderr();
readInt(conn->from);
2006-11-30 20:35:50 +02:00
}
BuildResult RemoteStore::buildDerivation(const Path & drvPath, const BasicDerivation & drv,
BuildMode buildMode)
{
auto conn(connections->get());
conn->to << wopBuildDerivation << drvPath << drv << buildMode;
conn->processStderr();
BuildResult res;
unsigned int status;
conn->from >> status >> res.errorMsg;
res.status = (BuildResult::Status) status;
return res;
}
void RemoteStore::ensurePath(const Path & path)
2006-11-30 20:35:50 +02:00
{
auto conn(connections->get());
conn->to << wopEnsurePath << path;
conn->processStderr();
readInt(conn->from);
2006-11-30 20:35:50 +02:00
}
void RemoteStore::addTempRoot(const Path & path)
{
auto conn(connections->get());
conn->to << wopAddTempRoot << path;
conn->processStderr();
readInt(conn->from);
}
void RemoteStore::addIndirectRoot(const Path & path)
{
auto conn(connections->get());
conn->to << wopAddIndirectRoot << path;
conn->processStderr();
readInt(conn->from);
}
void RemoteStore::syncWithGC()
{
auto conn(connections->get());
conn->to << wopSyncWithGC;
conn->processStderr();
readInt(conn->from);
}
Roots RemoteStore::findRoots()
{
auto conn(connections->get());
conn->to << wopFindRoots;
conn->processStderr();
size_t count = readNum<size_t>(conn->from);
Roots result;
while (count--) {
Path link = readString(conn->from);
Path target = readStorePath(*this, conn->from);
result[link] = target;
}
return result;
}
void RemoteStore::collectGarbage(const GCOptions & options, GCResults & results)
{
auto conn(connections->get());
2012-07-31 00:13:25 +03:00
conn->to
<< wopCollectGarbage << options.action << options.pathsToDelete << options.ignoreLiveness
<< options.maxFreed
/* removed options */
<< 0 << 0 << 0;
2012-07-31 00:13:25 +03:00
conn->processStderr();
2012-07-31 00:13:25 +03:00
results.paths = readStrings<PathSet>(conn->from);
results.bytesFreed = readLongLong(conn->from);
readLongLong(conn->from); // obsolete
{
auto state_(Store::state.lock());
state_->pathInfoCache.clear();
}
}
void RemoteStore::optimiseStore()
{
auto conn(connections->get());
conn->to << wopOptimiseStore;
conn->processStderr();
readInt(conn->from);
}
bool RemoteStore::verifyStore(bool checkContents, RepairFlag repair)
{
auto conn(connections->get());
conn->to << wopVerifyStore << checkContents << repair;
conn->processStderr();
return readInt(conn->from);
}
2016-02-23 17:40:16 +02:00
void RemoteStore::addSignatures(const Path & storePath, const StringSet & sigs)
{
auto conn(connections->get());
conn->to << wopAddSignatures << storePath << sigs;
conn->processStderr();
readInt(conn->from);
}
void RemoteStore::queryMissing(const PathSet & targets,
PathSet & willBuild, PathSet & willSubstitute, PathSet & unknown,
unsigned long long & downloadSize, unsigned long long & narSize)
{
{
auto conn(connections->get());
if (GET_PROTOCOL_MINOR(conn->daemonVersion) < 19)
// Don't hold the connection handle in the fallback case
// to prevent a deadlock.
goto fallback;
conn->to << wopQueryMissing << targets;
conn->processStderr();
willBuild = readStorePaths<PathSet>(*this, conn->from);
willSubstitute = readStorePaths<PathSet>(*this, conn->from);
unknown = readStorePaths<PathSet>(*this, conn->from);
conn->from >> downloadSize >> narSize;
return;
}
fallback:
return Store::queryMissing(targets, willBuild, willSubstitute,
unknown, downloadSize, narSize);
}
void RemoteStore::connect()
{
auto conn(connections->get());
}
void RemoteStore::flushBadConnections()
{
connections->flushBad();
}
2016-02-23 17:40:16 +02:00
RemoteStore::Connection::~Connection()
{
try {
to.flush();
} catch (...) {
ignoreException();
}
}
static Logger::Fields readFields(Source & from)
{
Logger::Fields fields;
size_t size = readInt(from);
for (size_t n = 0; n < size; n++) {
auto type = (decltype(Logger::Field::type)) readInt(from);
if (type == Logger::Field::tInt)
fields.push_back(readNum<uint64_t>(from));
else if (type == Logger::Field::tString)
fields.push_back(readString(from));
else
throw Error("got unsupported field type %x from Nix daemon", (int) type);
}
return fields;
}
void RemoteStore::Connection::processStderr(Sink * sink, Source * source)
{
to.flush();
while (true) {
auto msg = readNum<uint64_t>(from);
if (msg == STDERR_WRITE) {
string s = readString(from);
2007-02-21 18:34:00 +02:00
if (!sink) throw Error("no sink");
2016-05-04 16:46:25 +03:00
(*sink)(s);
2007-02-21 18:34:00 +02:00
}
else if (msg == STDERR_READ) {
if (!source) throw Error("no source");
size_t len = readNum<size_t>(from);
auto buf = std::make_unique<unsigned char[]>(len);
writeString(buf.get(), source->read(buf.get(), len), to);
to.flush();
}
else if (msg == STDERR_ERROR) {
string error = readString(from);
unsigned int status = readInt(from);
throw Error(status, error);
}
else if (msg == STDERR_NEXT)
printError(chomp(readString(from)));
else if (msg == STDERR_START_ACTIVITY) {
auto act = readNum<ActivityId>(from);
auto lvl = (Verbosity) readInt(from);
auto type = (ActivityType) readInt(from);
auto s = readString(from);
auto fields = readFields(from);
auto parent = readNum<ActivityId>(from);
logger->startActivity(act, lvl, type, s, fields, parent);
}
else if (msg == STDERR_STOP_ACTIVITY) {
auto act = readNum<ActivityId>(from);
logger->stopActivity(act);
}
else if (msg == STDERR_RESULT) {
auto act = readNum<ActivityId>(from);
auto type = (ResultType) readInt(from);
auto fields = readFields(from);
logger->result(act, type, fields);
}
else if (msg == STDERR_LAST)
break;
else
throw Error("got unknown message type %x from Nix daemon", msg);
}
}
2006-11-30 20:35:50 +02:00
}