관리-도구
편집 파일: PoolTest.cpp
#include <TestSupport.h> #include <jsoncpp/json.h> #include <Core/ApplicationPool/Pool.h> #include <LoggingKit/Context.h> #include <FileTools/FileManip.h> #include <StrIntTools/StrIntUtils.h> #include <IOTools/MessageSerialization.h> #include <map> #include <vector> #include <cerrno> #include <signal.h> using namespace std; using namespace Passenger; using namespace Passenger::ApplicationPool2; namespace tut { struct Core_ApplicationPool_PoolTest: public TestBase { WrapperRegistry::Registry wrapperRegistry; SpawningKit::Context::Schema skContextSchema; SpawningKit::Context::DebugSupport skDebugSupport; SpawningKit::Context skContext; SpawningKit::FactoryPtr skFactory; Context context; PoolPtr pool; Pool::DebugSupportPtr debug; Ticket ticket; GetCallback callback; SessionPtr currentSession; ExceptionPtr currentException; AtomicInt number; boost::mutex syncher; list<SessionPtr> sessions; bool retainSessions; Core_ApplicationPool_PoolTest() : skContext(skContextSchema) { retainSessions = false; wrapperRegistry.finalize(); skContext.resourceLocator = resourceLocator; skContext.wrapperRegistry = &wrapperRegistry; skContext.integrationMode = "standalone"; skContext.debugSupport = &skDebugSupport; skContext.spawnDir = getSystemTempDir(); skContext.finalize(); context.spawningKitFactory = boost::make_shared<SpawningKit::Factory>(&skContext); context.finalize(); pool = boost::make_shared<Pool>(&context); pool->initialize(); callback.func = _callback; callback.userData = this; Json::Value config; vector<ConfigKit::Error> errors; LoggingKit::ConfigChangeRequest req; config["level"] = "warn"; config["app_output_log_level"] = "debug"; if (LoggingKit::context->prepareConfigChange(config, errors, req)) { LoggingKit::context->commitConfigChange(req); } else { P_BUG("Error configuring LoggingKit: " << ConfigKit::toString(errors)); } } ~Core_ApplicationPool_PoolTest() { // Explicitly destroy these here because they can run // additional code that depend on other fields in this // class. TRACE_POINT(); clearAllSessions(); UPDATE_TRACE_POINT(); pool->destroy(); UPDATE_TRACE_POINT(); pool.reset(); Json::Value config; vector<ConfigKit::Error> errors; LoggingKit::ConfigChangeRequest req; config["level"] = DEFAULT_LOG_LEVEL_NAME; config["app_output_log_level"] = DEFAULT_APP_OUTPUT_LOG_LEVEL_NAME; if (LoggingKit::context->prepareConfigChange(config, errors, req)) { LoggingKit::context->commitConfigChange(req); } else { P_BUG("Error configuring LoggingKit: " << ConfigKit::toString(errors)); } SystemTime::releaseAll(); } void initPoolDebugging() { pool->initDebugging(); debug = pool->debugSupport; } void clearAllSessions() { SessionPtr myCurrentSession; list<SessionPtr> mySessions; { LockGuard l(syncher); myCurrentSession = currentSession; mySessions = sessions; currentSession.reset(); sessions.clear(); } myCurrentSession.reset(); mySessions.clear(); } Options createOptions() { Options options; options.spawnMethod = "dummy"; options.appRoot = "stub/rack"; options.appType = "ruby"; options.appStartCommand = "ruby start.rb"; options.startupFile = "start.rb"; options.loadShellEnvvars = false; options.user = testConfig["normal_user_1"].asCString(); options.defaultUser = testConfig["default_user"].asCString(); options.defaultGroup = testConfig["default_group"].asCString(); return options; } static void _callback(const AbstractSessionPtr &_session, const ExceptionPtr &e, void *userData) { Core_ApplicationPool_PoolTest *self = (Core_ApplicationPool_PoolTest *) userData; SessionPtr session = static_pointer_cast<Session>(_session); SessionPtr oldSession; { LockGuard l(self->syncher); oldSession = self->currentSession; self->currentSession = session; self->currentException = e; self->number++; if (self->retainSessions && session != NULL) { self->sessions.push_back(session); } } // destroy old session object outside the lock. } void sendHeaders(int connection, ...) { va_list ap; const char *arg; vector<StaticString> args; va_start(ap, connection); while ((arg = va_arg(ap, const char *)) != NULL) { args.push_back(StaticString(arg, strlen(arg) + 1)); } va_end(ap); shared_array<StaticString> args_array(new StaticString[args.size() + 1]); unsigned int totalSize = 0; for (unsigned int i = 0; i < args.size(); i++) { args_array[i + 1] = args[i]; totalSize += args[i].size(); } char sizeHeader[sizeof(uint32_t)]; Uint32Message::generate(sizeHeader, totalSize); args_array[0] = StaticString(sizeHeader, sizeof(uint32_t)); gatheredWrite(connection, args_array.get(), args.size() + 1, NULL); } string stripHeaders(const string &str) { string::size_type pos = str.find("\r\n\r\n"); if (pos == string::npos) { return str; } else { string result = str; result.erase(0, pos + 4); return result; } } string sendRequest(const Options &options, const char *path) { int oldNumber = number; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == oldNumber + 1; ); if (currentException != NULL) { P_ERROR("get() exception: " << currentException->what()); abort(); } currentSession->initiate(); sendHeaders(currentSession->fd(), "PATH_INFO", path, "REQUEST_METHOD", "GET", NULL); shutdown(currentSession->fd(), SHUT_WR); string body = stripHeaders(readAll(currentSession->fd(), 1024 * 1024).first); ProcessPtr process = currentSession->getProcess()->shared_from_this(); currentSession.reset(); EVENTUALLY(5, result = process->busyness() == 0; ); return body; } // Ensure that n processes exist. Options ensureMinProcesses(unsigned int n) { Options options = createOptions(); options.minProcesses = n; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); EVENTUALLY(5, result = pool->getProcessCount() == n; ); currentSession.reset(); return options; } void disableProcess(ProcessPtr process, AtomicInt *result) { *result = (int) pool->disableProcess(process->getGupid()); } }; DEFINE_TEST_GROUP_WITH_LIMIT(Core_ApplicationPool_PoolTest, 100); TEST_METHOD(1) { // Test initial state. ensure(!pool->atFullCapacity()); } /*********** Test asyncGet() behavior on a single Group ***********/ TEST_METHOD(2) { // asyncGet() actions on empty pools cannot be immediately satisfied. // Instead a new process will be spawned. In the mean time get() // actions are put on a wait list which will be processed as soon // as the new process is done spawning. Options options = createOptions(); ScopedLock l(pool->syncher); pool->asyncGet(options, callback, false); ensure_equals("(1)", number, 0); ensure("(2)", pool->getWaitlist.empty()); ensure("(3)", !pool->groups.empty()); l.unlock(); EVENTUALLY(5, result = pool->getProcessCount() == 1; ); EVENTUALLY(5, result = number == 1; ); ensure("(4)", currentSession != NULL); ensure("(5)", currentException == NULL); } TEST_METHOD(3) { // If one matching process already exists and it's not at full // capacity then asyncGet() will immediately use it. Options options = createOptions(); // Spawn a process and opens a session with it. pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); // Close the session so that the process is now idle. ProcessPtr process = currentSession->getProcess()->shared_from_this(); currentSession.reset(); ensure_equals(process->busyness(), 0); ensure(!process->isTotallyBusy()); // Verify test assertion. ScopedLock l(pool->syncher); pool->asyncGet(options, callback, false); ensure_equals("callback is immediately called", number, 2); } TEST_METHOD(4) { // If one matching process already exists but it's at full capacity, // and the limits prevent spawning of a new process, // then asyncGet() will put the get action on the group's wait // queue. When the process is no longer at full capacity it will // process the request. // Spawn a process and verify that it's at full capacity. // Keep its session open. Options options = createOptions(); options.appGroupName = "test"; pool->setMax(1); pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); SessionPtr session1 = currentSession; ProcessPtr process = session1->getProcess()->shared_from_this(); currentSession.reset(); ensure_equals(process->sessions, 1); ensure(process->isTotallyBusy()); // Now call asyncGet() again. pool->asyncGet(options, callback); ensure_equals("callback is not yet called", number, 1); ensure_equals("the get action has been put on the wait list", pool->groups.lookupCopy("test")->getWaitlist.size(), 1u); session1.reset(); ensure_equals("callback is called after the process becomes idle", number, 2); ensure_equals("the get wait list has been processed", pool->groups.lookupCopy("test")->getWaitlist.size(), 0u); ensure_equals(process->sessions, 1); } TEST_METHOD(5) { // If one matching process already exists but it's at full utilization, // and the limits and pool capacity allow spawning of a new process, // then get() will put the get action on the group's wait // queue while spawning a process in the background. // Either the existing process or the newly spawned process // will process the action, whichever becomes first available. // Here we test the case in which the existing process becomes // available first. initPoolDebugging(); // Spawn a regular process and keep its session open. Options options = createOptions(); debug->messages->send("Proceed with spawn loop iteration 1"); SessionPtr session1 = pool->get(options, &ticket); ProcessPtr process1 = session1->getProcess()->shared_from_this(); // Now spawn a process that never finishes. pool->asyncGet(options, callback); // Release the session on the first process. session1.reset(); EVENTUALLY(1, result = number == 1; ); ensure_equals("The first process handled the second asyncGet() request", currentSession->getProcess(), process1.get()); debug->messages->send("Proceed with spawn loop iteration 2"); EVENTUALLY(5, result = number == 1; ); } TEST_METHOD(6) { // Here we test the case in which the new process becomes // available first. // Spawn a regular process. Options options = createOptions(); pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); SessionPtr session1 = currentSession; ProcessPtr process1 = currentSession->getProcess()->shared_from_this(); currentSession.reset(); // As long as we don't release process1 the following get // action will be processed by the newly spawned process. pool->asyncGet(options, callback); EVENTUALLY(5, result = pool->getProcessCount() == 2; ); ensure_equals(number, 2); ensure(currentSession->getProcess() != process1.get()); } TEST_METHOD(7) { // If multiple matching processes exist, and one of them is idle, // then asyncGet() will use that. // Spawn 3 processes and keep a session open with 1 of them. Options options = createOptions(); options.minProcesses = 3; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); EVENTUALLY(5, result = pool->getProcessCount() == 3; ); SessionPtr session1 = currentSession; ProcessPtr process1 = currentSession->getProcess()->shared_from_this(); currentSession.reset(); // Now open another session. It should complete immediately // and should not use the first process. ScopedLock l(pool->syncher); pool->asyncGet(options, callback, false); ensure_equals("asyncGet() completed immediately", number, 2); SessionPtr session2 = currentSession; ProcessPtr process2 = currentSession->getProcess()->shared_from_this(); l.unlock(); currentSession.reset(); ensure(process2 != process1); // Now open yet another session. It should also complete immediately // and should not use the first or the second process. l.lock(); pool->asyncGet(options, callback, false); ensure_equals("asyncGet() completed immediately", number, 3); SessionPtr session3 = currentSession; ProcessPtr process3 = currentSession->getProcess()->shared_from_this(); l.unlock(); currentSession.reset(); ensure(process3 != process1); ensure(process3 != process2); } TEST_METHOD(8) { // If multiple matching processes exist, then asyncGet() will use // the one with the smallest utilization number. // Spawn 2 processes, each with a concurrency of 2. skDebugSupport.dummyConcurrency = 2; Options options = createOptions(); options.minProcesses = 2; pool->setMax(2); GroupPtr group = pool->findOrCreateGroup(options); { LockGuard l(pool->syncher); group->spawn(); } EVENTUALLY(5, result = pool->getProcessCount() == 2; ); // asyncGet() selects some process. pool->asyncGet(options, callback); ensure_equals("(1)", number, 1); SessionPtr session1 = currentSession; ProcessPtr process1 = currentSession->getProcess()->shared_from_this(); currentSession.reset(); // The first process now has 1 session, so next asyncGet() should // select the other process. pool->asyncGet(options, callback); ensure_equals("(2)", number, 2); SessionPtr session2 = currentSession; ProcessPtr process2 = currentSession->getProcess()->shared_from_this(); currentSession.reset(); ensure("(3)", process1 != process2); // Both processes now have an equal number of sessions. Next asyncGet() // can select either. pool->asyncGet(options, callback); ensure_equals("(4)", number, 3); SessionPtr session3 = currentSession; ProcessPtr process3 = currentSession->getProcess()->shared_from_this(); currentSession.reset(); // One process now has the lowest number of sessions. Next // asyncGet() should select that one. pool->asyncGet(options, callback); ensure_equals("(5)", number, 4); SessionPtr session4 = currentSession; ProcessPtr process4 = currentSession->getProcess()->shared_from_this(); currentSession.reset(); ensure("(6)", process3 != process4); } TEST_METHOD(9) { // If multiple matching processes exist, and all of them are at full capacity, // and no more processes may be spawned, // then asyncGet() will put the action on the group's wait queue. // The process that first becomes not at full capacity will process the action. // Spawn 2 processes and open 4 sessions. Options options = createOptions(); options.appGroupName = "test"; options.minProcesses = 2; pool->setMax(2); skDebugSupport.dummyConcurrency = 2; vector<SessionPtr> sessions; int expectedNumber = 1; for (int i = 0; i < 4; i++) { pool->asyncGet(options, callback); EVENTUALLY(5, result = number == expectedNumber; ); expectedNumber++; sessions.push_back(currentSession); currentSession.reset(); } EVENTUALLY(5, result = pool->getProcessCount() == 2; ); GroupPtr group = pool->groups.lookupCopy("test"); ensure_equals(group->getWaitlist.size(), 0u); ensure(pool->atFullCapacity()); // Now try to open another session. pool->asyncGet(options, callback); ensure_equals("The get request has been put on the wait list", pool->groups.lookupCopy("test")->getWaitlist.size(), 1u); // Close an existing session so that one process is no // longer at full utilization. sessions[0].reset(); ensure_equals("The get request has been removed from the wait list", pool->groups.lookupCopy("test")->getWaitlist.size(), 0u); ensure(pool->atFullCapacity()); } TEST_METHOD(10) { // If multiple matching processes exist, and all of them are at full utilization, // and a new process may be spawned, // then asyncGet() will put the action on the group's wait queue and spawn the // new process. // The process that first becomes not at full utilization // or the newly spawned process // will process the action, whichever is earlier. // Here we test the case where an existing process is earlier. // Spawn 2 processes and open 4 sessions. skDebugSupport.dummyConcurrency = 2; Options options = createOptions(); options.minProcesses = 2; pool->setMax(3); GroupPtr group = pool->findOrCreateGroup(options); vector<SessionPtr> sessions; int expectedNumber = 1; for (int i = 0; i < 4; i++) { pool->asyncGet(options, callback); EVENTUALLY(5, result = number == expectedNumber; ); expectedNumber++; sessions.push_back(currentSession); currentSession.reset(); } EVENTUALLY(5, result = pool->getProcessCount() == 2; ); // The next asyncGet() should spawn a new process and the action should be queued. ScopedLock l(pool->syncher); skDebugSupport.dummySpawnDelay = 5000000; pool->asyncGet(options, callback, false); ensure(group->spawning()); ensure_equals(group->getWaitlist.size(), 1u); l.unlock(); // Close one of the sessions. Now it will process the action. ProcessPtr process = sessions[0]->getProcess()->shared_from_this(); sessions[0].reset(); ensure_equals(number, 5); ensure_equals(currentSession->getProcess(), process.get()); ensure_equals(group->getWaitlist.size(), 0u); ensure_equals(pool->getProcessCount(), 2u); } TEST_METHOD(11) { // Here we test the case where the newly spawned process is earlier. // Spawn 2 processes and open 4 sessions. Options options = createOptions(); options.minProcesses = 2; pool->setMax(3); GroupPtr group = pool->findOrCreateGroup(options); skDebugSupport.dummyConcurrency = 2; vector<SessionPtr> sessions; int expectedNumber = 1; for (int i = 0; i < 4; i++) { pool->asyncGet(options, callback); EVENTUALLY(5, result = number == expectedNumber; ); expectedNumber++; sessions.push_back(currentSession); currentSession.reset(); } EVENTUALLY(5, result = pool->getProcessCount() == 2; ); // The next asyncGet() should spawn a new process. After it's done // spawning it will process the action. pool->asyncGet(options, callback); EVENTUALLY(5, result = pool->getProcessCount() == 3; ); EVENTUALLY(5, result = number == 5; ); ensure_equals(currentSession->getProcess()->getPid(), 3); ensure_equals(group->getWaitlist.size(), 0u); } TEST_METHOD(12) { // Test shutting down. ensureMinProcesses(2); ensure(pool->detachGroupByName("stub/rack")); ensure_equals(pool->getGroupCount(), 0u); } TEST_METHOD(13) { // Test shutting down while Group is restarting. initPoolDebugging(); debug->messages->send("Proceed with spawn loop iteration 1"); ensureMinProcesses(1); ensure(pool->restartGroupByName("stub/rack")); debug->debugger->recv("About to end restarting"); ensure(pool->detachGroupByName("stub/rack")); ensure_equals(pool->getGroupCount(), 0u); } TEST_METHOD(14) { // Test shutting down while Group is spawning. initPoolDebugging(); Options options = createOptions(); pool->asyncGet(options, callback); debug->debugger->recv("Begin spawn loop iteration 1"); ensure(pool->detachGroupByName("stub/rack")); ensure_equals(pool->getGroupCount(), 0u); } TEST_METHOD(17) { // Test that restartGroupByName() spawns more processes to ensure // that minProcesses and other constraints are met. ensureMinProcesses(1); ensure(pool->restartGroupByName("stub/rack")); EVENTUALLY(5, result = pool->getProcessCount() == 1; ); } TEST_METHOD(18) { // Test getting from an app for which minProcesses is set to 0, // and restart.txt already existed. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; options.minProcesses = 0; initPoolDebugging(); debug->spawning = false; SystemTime::forceAll(1000000); pool->get(options, &ticket); SystemTime::forceAll(20000000); touchFile("tmp.wsgi/tmp/restart.txt", 1); pool->asyncGet(options, callback); debug->debugger->recv("About to end restarting"); debug->messages->send("Finish restarting"); EVENTUALLY(5, result = number == 1; ); ensure_equals(pool->getProcessCount(), 1u); } /*********** Test asyncGet() behavior on multiple Groups ***********/ TEST_METHOD(20) { // If the pool is full, and one tries to asyncGet() from a nonexistant group, // then it will kill the oldest idle process and spawn a new process. Options options = createOptions(); pool->setMax(2); // Get from /foo and close its session immediately. options.appRoot = "/foo"; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); ProcessPtr process1 = currentSession->getProcess()->shared_from_this(); GroupPtr group1 = process1->getGroup()->shared_from_this(); currentSession.reset(); // Get from /bar and keep its session open. options.appRoot = "/bar"; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 2; ); SessionPtr session2 = currentSession; currentSession.reset(); // Get from /baz. The process for /foo should be killed now. options.appRoot = "/baz"; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 3; ); ensure_equals(pool->getProcessCount(), 2u); ensure_equals(group1->getProcessCount(), 0u); } TEST_METHOD(21) { // If the pool is full, and one tries to asyncGet() from a nonexistant group, // and all existing processes are non-idle, then it will // kill the oldest process and spawn a new process. Options options = createOptions(); pool->setMax(2); // Get from /foo and close its session immediately. options.appRoot = "/foo"; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); ProcessPtr process1 = currentSession->getProcess()->shared_from_this(); GroupPtr group1 = process1->getGroup()->shared_from_this(); // Get from /bar and keep its session open. options.appRoot = "/bar"; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 2; ); SessionPtr session2 = currentSession; currentSession.reset(); // Get from /baz. The process for /foo should be killed now. options.appRoot = "/baz"; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 3; ); ensure_equals(pool->getProcessCount(), 2u); ensure_equals(group1->getProcessCount(), 0u); } TEST_METHOD(22) { // Suppose the pool is at full capacity, and one tries to asyncGet() from an // existant group that does not have any processes. It should kill a process // from another group, and the request should succeed. Options options = createOptions(); SessionPtr session; pid_t pid1, pid2; pool->setMax(1); // Create a group /foo. options.appRoot = "/foo"; SystemTime::force(1); session = pool->get(options, &ticket); pid1 = session->getPid(); session.reset(); // Create a group /bar. options.appRoot = "/bar"; SystemTime::force(2); session = pool->get(options, &ticket); pid2 = session->getPid(); session.reset(); // Sleep for a short while to give Pool a chance to shutdown // the first process. usleep(300000); ensure_equals("(1)", pool->getProcessCount(), 1u); // Get from /foo. options.appRoot = "/foo"; SystemTime::force(3); session = pool->get(options, &ticket); ensure("(2)", session->getPid() != pid1); ensure("(3)", session->getPid() != pid2); ensure_equals("(4)", pool->getProcessCount(), 1u); } TEST_METHOD(23) { // Suppose the pool is at full capacity, and one tries to asyncGet() from an // existant group that does not have any processes, and that happens to need // restarting. It should kill a process from another group and the request // should succeed. Options options1 = createOptions(); Options options2 = createOptions(); TempDirCopy dir("stub/wsgi", "tmp.wsgi"); SessionPtr session; pid_t pid1, pid2; pool->setMax(1); // Create a group tmp.wsgi. options1.appRoot = "tmp.wsgi"; options1.appType = "wsgi"; options1.startupFile = "passenger_wsgi.py"; options1.spawnMethod = "direct"; SystemTime::force(1); session = pool->get(options1, &ticket); pid1 = session->getPid(); session.reset(); // Create a group bar. options2.appRoot = "bar"; SystemTime::force(2); session = pool->get(options2, &ticket); pid2 = session->getPid(); session.reset(); // Sleep for a short while to give Pool a chance to shutdown // the first process. usleep(300000); ensure_equals("(1)", pool->getProcessCount(), 1u); // Get from tmp.wsgi. SystemTime::force(3); touchFile("tmp.wsgi/tmp/restart.txt", 4); session = pool->get(options1, &ticket); ensure("(2)", session->getPid() != pid1); ensure("(3)", session->getPid() != pid2); ensure_equals("(4)", pool->getProcessCount(), 1u); } TEST_METHOD(24) { // Suppose the pool is at full capacity, with two groups: // - one that is spawning a process. // - one with no processes. // When one tries to asyncGet() from the second group, there should // be no process to kill, but when the first group is done spawning // it should throw away that process immediately to allow the second // group to spawn. Options options1 = createOptions(); Options options2 = createOptions(); initPoolDebugging(); debug->restarting = false; pool->setMax(1); // Create a group foo. options1.appRoot = "foo"; options1.noop = true; SystemTime::force(1); pool->get(options1, &ticket); // Create a group bar, but don't let it finish spawning. options2.appRoot = "bar"; options2.noop = true; SystemTime::force(2); GroupPtr barGroup = pool->get(options2, &ticket)->getGroup()->shared_from_this(); { LockGuard l(pool->syncher); ensure_equals("(1)", barGroup->spawn(), SR_OK); } debug->debugger->recv("Begin spawn loop iteration 1"); // Now get from foo again and let the request be queued. options1.noop = false; SystemTime::force(3); pool->asyncGet(options1, callback); // Nothing should happen while bar is spawning. SHOULD_NEVER_HAPPEN(100, result = number > 0; ); ensure_equals("(2)", pool->getProcessCount(), 0u); // Now let bar finish spawning. Eventually there should // only be one process: the one for foo. debug->messages->send("Proceed with spawn loop iteration 1"); debug->debugger->recv("Spawn loop done"); debug->messages->send("Proceed with spawn loop iteration 2"); debug->debugger->recv("Spawn loop done"); EVENTUALLY(5, LockGuard l(pool->syncher); vector<ProcessPtr> processes = pool->getProcesses(false); if (processes.size() == 1) { GroupPtr group = processes[0]->getGroup()->shared_from_this(); result = group->getName() == "foo"; } else { result = false; } ); } TEST_METHOD(25) { // Suppose the pool is at full capacity, with two groups: // - one that is spawning a process, and has a queued request. // - one with no processes. // When one tries to asyncGet() from the second group, there should // be no process to kill, but when the first group is done spawning // it should throw away that process immediately to allow the second // group to spawn. Options options1 = createOptions(); Options options2 = createOptions(); initPoolDebugging(); debug->restarting = false; pool->setMax(1); // Create a group foo. options1.appRoot = "foo"; options1.noop = true; SystemTime::force(1); pool->get(options1, &ticket); // Create a group bar with a queued request, but don't let it finish spawning. options2.appRoot = "bar"; SystemTime::force(2); pool->asyncGet(options2, callback); debug->debugger->recv("Begin spawn loop iteration 1"); // Now get from foo again and let the request be queued. options1.noop = false; SystemTime::force(3); pool->asyncGet(options1, callback); // Nothing should happen while bar is spawning. SHOULD_NEVER_HAPPEN(100, result = number > 0; ); ensure_equals("(1)", pool->getProcessCount(), 0u); // Now let bar finish spawning. The request for bar should be served. debug->messages->send("Proceed with spawn loop iteration 1"); debug->debugger->recv("Spawn loop done"); EVENTUALLY(5, result = number == 1; ); ensure_equals(currentSession->getGroup()->getName(), "bar"); // When that request is done, the process for bar should be killed, // and a process for foo should be spawned. currentSession.reset(); debug->messages->send("Proceed with spawn loop iteration 2"); debug->debugger->recv("Spawn loop done"); EVENTUALLY(5, LockGuard l(pool->syncher); vector<ProcessPtr> processes = pool->getProcesses(false); if (processes.size() == 1) { GroupPtr group = processes[0]->getGroup()->shared_from_this(); result = group->getName() == "foo"; } else { result = false; } ); EVENTUALLY(5, result = number == 2; ); } /*********** Test detachProcess() ***********/ TEST_METHOD(30) { // detachProcess() detaches the process from the group. The pool // will restore the minimum number of processes afterwards. Options options = createOptions(); options.appGroupName = "test"; options.minProcesses = 2; pool->asyncGet(options, callback); EVENTUALLY(5, result = pool->getProcessCount() == 2; ); EVENTUALLY(5, result = number == 1; ); ProcessPtr process = currentSession->getProcess()->shared_from_this(); pool->detachProcess(process); { LockGuard l(pool->syncher); ensure(process->enabled == Process::DETACHED); } EVENTUALLY(5, result = pool->getProcessCount() == 2; ); currentSession.reset(); EVENTUALLY(5, result = process->isDead(); ); } TEST_METHOD(31) { // If the containing group had waiters on it, and detachProcess() // detaches the only process in the group, then a new process // is automatically spawned to handle the waiters. Options options = createOptions(); options.appGroupName = "test"; pool->setMax(1); skDebugSupport.dummySpawnDelay = 1000000; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); SessionPtr session1 = currentSession; currentSession.reset(); pool->asyncGet(options, callback); { LockGuard l(pool->syncher); ensure_equals(pool->groups.lookupCopy("test")->getWaitlist.size(), 1u); } pool->detachProcess(session1->getProcess()->shared_from_this()); { LockGuard l(pool->syncher); ensure(pool->groups.lookupCopy("test")->spawning()); ensure_equals(pool->groups.lookupCopy("test")->enabledCount, 0); ensure_equals(pool->groups.lookupCopy("test")->getWaitlist.size(), 1u); } EVENTUALLY(5, result = number == 2; ); } TEST_METHOD(32) { // If the pool had waiters on it then detachProcess() will // automatically create the Groups that were requested // by the waiters. Options options = createOptions(); options.appGroupName = "test"; options.minProcesses = 0; pool->setMax(1); skDebugSupport.dummySpawnDelay = 30000; // Begin spawning a process. pool->asyncGet(options, callback); ensure(pool->atFullCapacity()); // asyncGet() on another group should now put it on the waiting list. Options options2 = createOptions(); options2.appGroupName = "test2"; options2.minProcesses = 0; skDebugSupport.dummySpawnDelay = 90000; pool->asyncGet(options2, callback); { LockGuard l(pool->syncher); ensure_equals(pool->getWaitlist.size(), 1u); } // Eventually the dummy process for "test" is now done spawning. // We then detach it. EVENTUALLY(5, result = number == 1; ); SessionPtr session1 = currentSession; currentSession.reset(); pool->detachProcess(session1->getProcess()->shared_from_this()); { LockGuard l(pool->syncher); ensure(pool->groups.lookupCopy("test2") != NULL); ensure_equals(pool->getWaitlist.size(), 0u); } EVENTUALLY(5, result = number == 2; ); } TEST_METHOD(33) { // A Group does not become garbage collectable // after detaching all its processes. Options options = createOptions(); pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); ProcessPtr process = currentSession->getProcess()->shared_from_this(); currentSession.reset(); GroupPtr group = process->getGroup()->shared_from_this(); pool->detachProcess(process); LockGuard l(pool->syncher); ensure_equals(pool->groups.size(), 1u); ensure(group->isAlive()); ensure(!group->garbageCollectable()); } TEST_METHOD(34) { // When detaching a process, it waits until all sessions have // finished before telling the process to shut down. Options options = createOptions(); options.spawnMethod = "direct"; options.minProcesses = 0; SessionPtr session = pool->get(options, &ticket); ProcessPtr process = session->getProcess()->shared_from_this(); ensure(pool->detachProcess(process)); { LockGuard l(pool->syncher); ensure_equals(process->enabled, Process::DETACHED); } SHOULD_NEVER_HAPPEN(100, LockGuard l(pool->syncher); result = !process->isAlive() || !process->osProcessExists(); ); session.reset(); EVENTUALLY(1, LockGuard l(pool->syncher); result = process->enabled == Process::DETACHED && !process->osProcessExists() && process->isDead(); ); } TEST_METHOD(35) { // When detaching a process, it waits until the OS processes // have exited before cleaning up the in-memory data structures. Options options = createOptions(); options.spawnMethod = "direct"; options.minProcesses = 0; ProcessPtr process = pool->get(options, &ticket)->getProcess()->shared_from_this(); ScopeGuard g(boost::bind(::kill, process->getPid(), SIGCONT)); kill(process->getPid(), SIGSTOP); ensure(pool->detachProcess(process)); { LockGuard l(pool->syncher); ensure_equals(process->enabled, Process::DETACHED); } EVENTUALLY(1, result = process->getLifeStatus() == Process::SHUTDOWN_TRIGGERED; ); SHOULD_NEVER_HAPPEN(100, LockGuard l(pool->syncher); result = process->isDead() || !process->osProcessExists(); ); kill(process->getPid(), SIGCONT); g.clear(); EVENTUALLY(1, LockGuard l(pool->syncher); result = process->enabled == Process::DETACHED && !process->osProcessExists() && process->isDead(); ); } TEST_METHOD(36) { // Detaching a process that is already being detached, works. Options options = createOptions(); options.appGroupName = "test"; options.minProcesses = 0; initPoolDebugging(); debug->restarting = false; debug->spawning = false; debug->detachedProcessesChecker = true; pool->asyncGet(options, callback); EVENTUALLY(5, result = pool->getProcessCount() == 1; ); EVENTUALLY(5, result = number == 1; ); ProcessPtr process = currentSession->getProcess()->shared_from_this(); pool->detachProcess(process); debug->debugger->recv("About to start detached processes checker"); { LockGuard l(pool->syncher); ensure(process->enabled == Process::DETACHED); } // detachProcess() will spawn a new process. Prevent it from being // spawned too soon. debug->spawning = true; pool->detachProcess(process); debug->messages->send("Proceed with starting detached processes checker"); debug->messages->send("Proceed with starting detached processes checker"); debug->messages->send("Proceed with spawn loop iteration 2"); EVENTUALLY(5, result = pool->getProcessCount() == 0; ); currentSession.reset(); EVENTUALLY(5, result = process->isDead(); ); } /*********** Test disabling and enabling processes ***********/ TEST_METHOD(40) { // Disabling a process under idle conditions should succeed immediately. ensureMinProcesses(2); vector<ProcessPtr> processes = pool->getProcesses(); ensure_equals("Disabling succeeds", pool->disableProcess(processes[0]->getGupid()), DR_SUCCESS); LockGuard l(pool->syncher); ensure(processes[0]->isAlive()); ensure_equals("Process is disabled", processes[0]->enabled, Process::DISABLED); ensure("Other processes are not affected", processes[1]->isAlive()); ensure_equals("Other processes are not affected", processes[1]->enabled, Process::ENABLED); } TEST_METHOD(41) { // Disabling the sole process in a group, in case the pool settings allow // spawning another process, should trigger a new process spawn. ensureMinProcesses(1); Options options = createOptions(); SessionPtr session = pool->get(options, &ticket); ensure_equals(pool->getProcessCount(), 1u); ensure(!pool->isSpawning()); skDebugSupport.dummySpawnDelay = 60000; AtomicInt code = -1; TempThread thr(boost::bind(&Core_ApplicationPool_PoolTest::disableProcess, this, session->getProcess()->shared_from_this(), &code)); EVENTUALLY2(100, 1, result = pool->isSpawning(); ); EVENTUALLY(1, result = pool->getProcessCount() == 2u; ); ensure_equals((int) code, -1); session.reset(); EVENTUALLY(1, result = code == (int) DR_SUCCESS; ); } TEST_METHOD(42) { // Disabling the sole process in a group, in case pool settings don't allow // spawning another process, should fail. pool->setMax(1); ensureMinProcesses(1); vector<ProcessPtr> processes = pool->getProcesses(); ensure_equals("(1)", processes.size(), 1u); if (defaultLogLevel == (LoggingKit::Level) DEFAULT_LOG_LEVEL) { // If the user did not customize the test's log level, // then we'll want to tone down the noise. LoggingKit::setLevel(LoggingKit::ERROR); } DisableResult result = pool->disableProcess(processes[0]->getGupid()); ensure_equals("(2)", result, DR_ERROR); ensure_equals("(3)", pool->getProcessCount(), 1u); } TEST_METHOD(43) { // If there are no enabled processes in the group, then disabling should // succeed after the new process has been spawned. initPoolDebugging(); debug->messages->send("Proceed with spawn loop iteration 1"); debug->messages->send("Proceed with spawn loop iteration 2"); Options options = createOptions(); SessionPtr session1 = pool->get(options, &ticket); SessionPtr session2 = pool->get(options, &ticket); ensure_equals(pool->getProcessCount(), 2u); GroupPtr group = session1->getGroup()->shared_from_this(); ProcessPtr process1 = session1->getProcess()->shared_from_this(); ProcessPtr process2 = session2->getProcess()->shared_from_this(); AtomicInt code1 = -1, code2 = -2; TempThread thr(boost::bind(&Core_ApplicationPool_PoolTest::disableProcess, this, process1, &code1)); TempThread thr2(boost::bind(&Core_ApplicationPool_PoolTest::disableProcess, this, process2, &code2)); EVENTUALLY(5, LockGuard l(pool->syncher); result = group->enabledCount == 0 && group->disablingCount == 2 && group->disabledCount == 0; ); session1.reset(); session2.reset(); SHOULD_NEVER_HAPPEN(20, result = code1 != -1 || code2 != -2; ); debug->messages->send("Proceed with spawn loop iteration 3"); EVENTUALLY(5, result = code1 == DR_SUCCESS; ); EVENTUALLY(5, result = code2 == DR_SUCCESS; ); { LockGuard l(pool->syncher); ensure_equals(group->enabledCount, 1); ensure_equals(group->disablingCount, 0); ensure_equals(group->disabledCount, 2); } } TEST_METHOD(44) { // Suppose that a previous disable command triggered a new process spawn, // and the spawn fails. Then any disabling processes should become enabled // again, and the callbacks for the previous disable commands should be called. initPoolDebugging(); debug->messages->send("Proceed with spawn loop iteration 1"); debug->messages->send("Proceed with spawn loop iteration 2"); Options options = createOptions(); options.minProcesses = 2; SessionPtr session1 = pool->get(options, &ticket); SessionPtr session2 = pool->get(options, &ticket); ensure_equals(pool->getProcessCount(), 2u); AtomicInt code1 = -1, code2 = -1; TempThread thr(boost::bind(&Core_ApplicationPool_PoolTest::disableProcess, this, session1->getProcess()->shared_from_this(), &code1)); TempThread thr2(boost::bind(&Core_ApplicationPool_PoolTest::disableProcess, this, session2->getProcess()->shared_from_this(), &code2)); EVENTUALLY(2, GroupPtr group = session1->getGroup()->shared_from_this(); LockGuard l(pool->syncher); result = group->enabledCount == 0 && group->disablingCount == 2 && group->disabledCount == 0; ); SHOULD_NEVER_HAPPEN(20, result = code1 != -1 || code2 != -1; ); if (defaultLogLevel == (LoggingKit::Level) DEFAULT_LOG_LEVEL) { // If the user did not customize the test's log level, // then we'll want to tone down the noise. LoggingKit::setLevel(LoggingKit::CRIT); } debug->messages->send("Fail spawn loop iteration 3"); EVENTUALLY(5, result = code1 == DR_ERROR; ); EVENTUALLY(5, result = code2 == DR_ERROR; ); { GroupPtr group = session1->getGroup()->shared_from_this(); LockGuard l(pool->syncher); ensure_equals(group->enabledCount, 2); ensure_equals(group->disablingCount, 0); ensure_equals(group->disabledCount, 0); } } // TODO: asyncGet() should not select a disabling process if there are enabled processes. // TODO: asyncGet() should not select a disabling process when non-rolling restarting. // TODO: asyncGet() should select a disabling process if there are no enabled processes // in the group. If this happens then asyncGet() will also spawn a new process. // TODO: asyncGet() should not select a disabled process. // TODO: If there are no enabled processes and all disabling processes are at full // utilization, and the process that was being spawned becomes available // earlier than any of the disabling processes, then the newly spawned process // should handle the request. // TODO: A disabling process becomes disabled as soon as it's done with // all its request. TEST_METHOD(50) { // Disabling a process that's already being disabled should result in the // callback being called after disabling is done. ensureMinProcesses(2); Options options = createOptions(); SessionPtr session = pool->get(options, &ticket); AtomicInt code = -1; TempThread thr(boost::bind(&Core_ApplicationPool_PoolTest::disableProcess, this, session->getProcess()->shared_from_this(), &code)); SHOULD_NEVER_HAPPEN(100, result = code != -1; ); session.reset(); EVENTUALLY(5, result = code != -1; ); ensure_equals(code, (int) DR_SUCCESS); } // TODO: Enabling a process that's disabled succeeds immediately. // TODO: Enabling a process that's disabling succeeds immediately. The disable // callbacks will be called with DR_CANCELED. TEST_METHOD(51) { // If the number of processes is already at maximum, then disabling // a process will cause that process to be disabled, without spawning // a new process. pool->setMax(2); ensureMinProcesses(2); vector<ProcessPtr> processes = pool->getProcesses(); ensure_equals(processes.size(), 2u); DisableResult result = pool->disableProcess( processes[0]->getGupid()); ensure_equals(result, DR_SUCCESS); { ScopedLock l(pool->syncher); GroupPtr group = processes[0]->getGroup()->shared_from_this(); ensure_equals(group->enabledCount, 1); ensure_equals(group->disablingCount, 0); ensure_equals(group->disabledCount, 1); } } /*********** Other tests ***********/ TEST_METHOD(60) { // The pool is considered to be at full capacity if and only // if all Groups are at full capacity. Options options = createOptions(); Options options2 = createOptions(); options2.appGroupName = "test"; pool->setMax(2); pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); pool->asyncGet(options2, callback); EVENTUALLY(5, result = number == 2; ); ensure_equals(pool->getProcessCount(), 2u); ensure(pool->atFullCapacity()); clearAllSessions(); pool->detachGroupByName("test"); ensure(!pool->atFullCapacity()); } TEST_METHOD(61) { // If the pool is at full capacity, then increasing 'max' will cause // new processes to be spawned. Any queued get requests are processed // as those new processes become available or as existing processes // become available. Options options = createOptions(); retainSessions = true; pool->setMax(1); pool->asyncGet(options, callback); pool->asyncGet(options, callback); pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); pool->setMax(4); EVENTUALLY(5, result = number == 3; ); ensure_equals(pool->getProcessCount(), 3u); } TEST_METHOD(62) { // Each spawned process has a GUPID, which can be looked up // through findProcessByGupid(). Options options = createOptions(); pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); string gupid = currentSession->getProcess()->getGupid().toString(); ensure(!gupid.empty()); ensure_equals(currentSession->getProcess(), pool->findProcessByGupid(gupid).get()); } TEST_METHOD(63) { // findProcessByGupid() returns a NULL pointer if there is // no matching process. ensure(pool->findProcessByGupid("none") == NULL); } TEST_METHOD(64) { // Test process idle cleaning. Options options = createOptions(); pool->setMaxIdleTime(50000); SessionPtr session1 = pool->get(options, &ticket); SessionPtr session2 = pool->get(options, &ticket); ensure_equals(pool->getProcessCount(), 2u); session2.reset(); // One of the processes still has a session open and should // not be idle cleaned. EVENTUALLY(2, result = pool->getProcessCount() == 1; ); SHOULD_NEVER_HAPPEN(150, result = pool->getProcessCount() == 0; ); // It shouldn't clean more processes than minInstances allows. sessions.clear(); SHOULD_NEVER_HAPPEN(150, result = pool->getProcessCount() == 0; ); } TEST_METHOD(65) { // Test spawner idle cleaning. Options options = createOptions(); options.appGroupName = "test1"; Options options2 = createOptions(); options2.appGroupName = "test2"; retainSessions = true; pool->setMaxIdleTime(50000); pool->asyncGet(options, callback); pool->asyncGet(options2, callback); EVENTUALLY(2, result = number == 2; ); ensure_equals(pool->getProcessCount(), 2u); EVENTUALLY(2, SpawningKit::SpawnerPtr spawner = pool->getGroup("test1")->spawner; result = static_pointer_cast<SpawningKit::DummySpawner>(spawner)->cleanCount >= 1; ); EVENTUALLY(2, SpawningKit::SpawnerPtr spawner = pool->getGroup("test2")->spawner; result = static_pointer_cast<SpawningKit::DummySpawner>(spawner)->cleanCount >= 1; ); } TEST_METHOD(66) { // It should restart the app if restart.txt is created or updated. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; options.statThrottleRate = 0; pool->setMax(1); // Send normal request. ensure_equals(sendRequest(options, "/"), "front page"); // Modify application; it shouldn't have effect yet. writeFile("tmp.wsgi/passenger_wsgi.py", "def application(env, start_response):\n" " start_response('200 OK', [('Content-Type', 'text/html')])\n" " return ['restarted']\n"); ensure_equals(sendRequest(options, "/"), "front page"); // Create restart.txt and send request again. The change should now be activated. touchFile("tmp.wsgi/tmp/restart.txt", 1); ensure_equals(sendRequest(options, "/"), "restarted"); // Modify application again; it shouldn't have effect yet. writeFile("tmp.wsgi/passenger_wsgi.py", "def application(env, start_response):\n" " start_response('200 OK', [('Content-Type', 'text/html')])\n" " return ['restarted 2']\n"); ensure_equals(sendRequest(options, "/"), "restarted"); // Touch restart.txt and send request again. The change should now be activated. touchFile("tmp.wsgi/tmp/restart.txt", 2); ensure_equals(sendRequest(options, "/"), "restarted 2"); } TEST_METHOD(67) { // Test spawn exceptions. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; writeFile("tmp.wsgi/passenger_wsgi.py", "import sys\n" "sys.stderr.write('Something went wrong!')\n" "exit(1)\n"); if (defaultLogLevel == (LoggingKit::Level) DEFAULT_LOG_LEVEL) { // If the user did not customize the test's log level, // then we'll want to tone down the noise. LoggingKit::setLevel(LoggingKit::CRIT); } pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); ensure(currentException != NULL); boost::shared_ptr<SpawningKit::SpawnException> e = dynamic_pointer_cast<SpawningKit::SpawnException>(currentException); ensure(e->getProblemDescriptionHTML().find("Something went wrong!") != string::npos); } TEST_METHOD(68) { // If a process fails to spawn, then it stops trying to spawn minProcesses processes. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; options.minProcesses = 4; writeFile("tmp.wsgi/counter", "0"); chmod("tmp.wsgi/counter", 0666); // Our application starts successfully the first two times, // and fails all the other times. writeFile("tmp.wsgi/passenger_wsgi.py", "import sys\n" "def application(env, start_response):\n" " pass\n" "counter = int(open('counter', 'r').read())\n" "f = open('counter', 'w')\n" "f.write(str(counter + 1))\n" "f.close()\n" "if counter >= 2:\n" " sys.stderr.write('Something went wrong!')\n" " exit(1)\n"); if (defaultLogLevel == (LoggingKit::Level) DEFAULT_LOG_LEVEL) { // If the user did not customize the test's log level, // then we'll want to tone down the noise. LoggingKit::setLevel(LoggingKit::CRIT); } pool->asyncGet(options, callback); EVENTUALLY(10, result = number == 1; ); EVENTUALLY(10, result = pool->getProcessCount() == 2; ); EVENTUALLY(10, result = !pool->isSpawning(); ); SHOULD_NEVER_HAPPEN(500, result = pool->getProcessCount() > 2; ); } TEST_METHOD(69) { // It removes the process from the pool if session->initiate() fails. Options options = createOptions(); options.appRoot = "stub/wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; options.minProcesses = 0; pool->asyncGet(options, callback); EVENTUALLY(5, result = number == 1; ); pid_t pid = currentSession->getPid(); kill(pid, SIGTERM); // Wait until process is gone. EVENTUALLY(5, result = kill(pid, 0) == -1 && (errno == ESRCH || errno == EPERM || errno == ECHILD); ); try { currentSession->initiate(); fail("Initiate is supposed to fail"); } catch (const SystemException &e) { ensure_equals(e.code(), ECONNREFUSED); } ensure_equals(pool->getProcessCount(), 0u); } TEST_METHOD(70) { // When a process has become idle, and there are waiters on the pool, // consider detaching it in order to satisfy a waiter. Options options1 = createOptions(); Options options2 = createOptions(); options2.appRoot = "stub/wsgi"; retainSessions = true; pool->setMax(2); pool->asyncGet(options1, callback); pool->asyncGet(options1, callback); EVENTUALLY(3, result = pool->getProcessCount() == 2; ); pool->asyncGet(options2, callback); ensure_equals(pool->getWaitlist.size(), 1u); ensure_equals(number, 2); currentSession.reset(); sessions.pop_front(); EVENTUALLY(3, result = number == 3; ); ensure_equals(pool->getProcessCount(), 2u); GroupPtr group1 = pool->groups.lookupCopy("stub/rack"); GroupPtr group2 = pool->groups.lookupCopy("stub/rack"); ensure_equals(group1->enabledCount, 1); ensure_equals(group2->enabledCount, 1); } TEST_METHOD(71) { // A process is detached after processing maxRequests sessions. Options options = createOptions(); options.minProcesses = 0; options.maxRequests = 5; pool->setMax(1); SessionPtr session = pool->get(options, &ticket); ensure_equals(pool->getProcessCount(), 1u); pid_t origPid = session->getPid(); session.reset(); for (int i = 0; i < 3; i++) { pool->get(options, &ticket).reset(); ensure_equals(pool->getProcessCount(), 1u); ensure_equals(pool->getProcesses()[0]->getPid(), origPid); } pool->get(options, &ticket).reset(); EVENTUALLY(2, result = pool->getProcessCount() == 0; ); } TEST_METHOD(72) { // If we restart while spawning is in progress, and the restart // finishes before the process is done spawning, then that // process will not be attached and the original spawn loop will // abort. A new spawn loop will start to ensure that resource // constraints are met. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); initPoolDebugging(); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.minProcesses = 3; options.statThrottleRate = 0; // Trigger spawn loop and freeze it at the point where it's spawning // the second process. pool->asyncGet(options, callback); debug->debugger->recv("Begin spawn loop iteration 1"); debug->messages->send("Proceed with spawn loop iteration 1"); debug->debugger->recv("Begin spawn loop iteration 2"); ensure_equals("(1)", pool->getProcessCount(), 1u); // Trigger restart, wait until it's finished. touchFile("tmp.wsgi/tmp/restart.txt", 1); pool->asyncGet(options, callback); debug->messages->send("Finish restarting"); debug->debugger->recv("Restarting done"); ensure_equals("(2)", pool->getProcessCount(), 0u); // The restarter should have created a new spawn loop and // instructed the old one to stop. debug->debugger->recv("Begin spawn loop iteration 3"); // We let the old spawn loop continue, which should drop // the second process and abort. debug->messages->send("Proceed with spawn loop iteration 2"); debug->debugger->recv("Spawn loop done"); ensure_equals("(3)", pool->getProcessCount(), 0u); // We let the new spawn loop continue. debug->messages->send("Proceed with spawn loop iteration 3"); debug->messages->send("Proceed with spawn loop iteration 4"); debug->messages->send("Proceed with spawn loop iteration 5"); debug->debugger->recv("Spawn loop done"); ensure_equals("(4)", pool->getProcessCount(), 3u); } TEST_METHOD(73) { // If a get() request comes in while the restart is in progress, then // that get() request will be put into the get waiters list, which will // be processed after spawning is done. // Spawn 2 processes. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.minProcesses = 2; options.statThrottleRate = 0; pool->asyncGet(options, callback); EVENTUALLY(2, result = pool->getProcessCount() == 2; ); // Trigger a restart. The creation of the new spawner should take a while. skDebugSupport.spawnerCreationSleepTime = 20000; touchFile("tmp.wsgi/tmp/restart.txt"); pool->asyncGet(options, callback); GroupPtr group = pool->findOrCreateGroup(options); ensure_equals("(1)", pool->getProcessCount(), 0u); ensure_equals("(2)", group->getWaitlist.size(), 1u); // Now that the restart is in progress, perform a get(). pool->asyncGet(options, callback); ensure_equals("(3)", group->getWaitlist.size(), 2u); EVENTUALLY(2, result = number == 3; ); ensure_equals("(4) The restart function respects minProcesses", pool->getProcessCount(), 2u); } TEST_METHOD(74) { // If a process fails to spawn, it sends a SpawnException result to all get waiters. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); chmod("tmp.wsgi", 0777); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; pool->setMax(1); writeFile("tmp.wsgi/passenger_wsgi.py", "import os, time, sys\n" "\n" "def file_exists(filename):\n" " try:\n" " os.stat(filename)\n" " return True\n" " except OSError:\n" " return False\n" "\n" "f = open('spawned.txt', 'w')\n" "f.write(str(os.getpid()))\n" "f.close()\n" "while not file_exists('continue.txt'):\n" " time.sleep(0.05)\n" "sys.stderr.write('Something went wrong!')\n" "exit(1)\n"); retainSessions = true; if (defaultLogLevel == (LoggingKit::Level) DEFAULT_LOG_LEVEL) { // If the user did not customize the test's log level, // then we'll want to tone down the noise. LoggingKit::setLevel(LoggingKit::CRIT); } pool->asyncGet(options, callback); pool->asyncGet(options, callback); pool->asyncGet(options, callback); pool->asyncGet(options, callback); EVENTUALLY(5, result = fileExists("tmp.wsgi/spawned.txt"); ); usleep(20000); writeFile("tmp.wsgi/passenger_wsgi.py", unsafeReadFile("stub/wsgi/passenger_wsgi.py")); pid_t pid = (pid_t) stringToLL(unsafeReadFile("tmp.wsgi/spawned.txt")); kill(pid, SIGTERM); EVENTUALLY(5, result = number == 4; ); ensure_equals(pool->getProcessCount(), 0u); ensure(sessions.empty()); } TEST_METHOD(75) { // If a process fails to spawn, the existing processes // are kept alive. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; options.minProcesses = 2; // Spawn 2 processes. retainSessions = true; pool->asyncGet(options, callback); pool->asyncGet(options, callback); EVENTUALLY(10, result = number == 2; ); ensure_equals(pool->getProcessCount(), 2u); // Mess up the application and spawn a new one. writeFile("tmp.wsgi/passenger_wsgi.py", "import sys\n" "sys.stderr.write('Something went wrong!')\n" "exit(1)\n"); if (defaultLogLevel == (LoggingKit::Level) DEFAULT_LOG_LEVEL) { // If the user did not customize the test's log level, // then we'll want to tone down the noise. LoggingKit::setLevel(LoggingKit::CRIT); } try { currentSession = pool->get(options, &ticket); fail("SpawnException expected"); } catch (const SpawningKit::SpawnException &) { ensure_equals(pool->getProcessCount(), 2u); } } TEST_METHOD(76) { // No more than maxOutOfBandWorkInstances process will be performing // out-of-band work at the same time. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; options.maxOutOfBandWorkInstances = 2; initPoolDebugging(); debug->restarting = false; debug->spawning = false; debug->oobw = true; // Spawn 3 processes and initiate 2 OOBW requests. SessionPtr session1 = pool->get(options, &ticket); SessionPtr session2 = pool->get(options, &ticket); SessionPtr session3 = pool->get(options, &ticket); session1->requestOOBW(); session1.reset(); session2->requestOOBW(); session2.reset(); // 2 OOBW requests eventually start. debug->debugger->recv("OOBW request about to start"); debug->debugger->recv("OOBW request about to start"); // Request another OOBW, but this one is not initiated. session3->requestOOBW(); session3.reset(); SHOULD_NEVER_HAPPEN(100, result = debug->debugger->peek("OOBW request about to start") != NULL; ); // Let one OOBW request finish. The third one should eventually // start. debug->messages->send("Proceed with OOBW request"); debug->debugger->recv("OOBW request about to start"); debug->messages->send("Proceed with OOBW request"); debug->messages->send("Proceed with OOBW request"); debug->debugger->recv("OOBW request finished"); debug->debugger->recv("OOBW request finished"); debug->debugger->recv("OOBW request finished"); } TEST_METHOD(77) { // If the getWaitlist already has maxRequestQueueSize items, // then an exception is returned. Options options = createOptions(); options.appGroupName = "test1"; options.maxRequestQueueSize = 3; GroupPtr group = pool->findOrCreateGroup(options); skDebugSupport.dummyConcurrency = 3; initPoolDebugging(); pool->setMax(1); for (int i = 0; i < 3; i++) { pool->asyncGet(options, callback); } ensure_equals(number, 0); { LockGuard l(pool->syncher); ensure_equals(group->getWaitlist.size(), 3u); } try { pool->get(options, &ticket); fail("Expected RequestQueueFullException"); } catch (const RequestQueueFullException &e) { // OK } debug->messages->send("Proceed with spawn loop iteration 1"); debug->messages->send("Spawn loop done"); EVENTUALLY(5, result = number == 3; ); } TEST_METHOD(78) { // Test restarting while a previous restart was already being finalized. // The previous finalization should abort. Options options = createOptions(); initPoolDebugging(); debug->spawning = false; pool->get(options, &ticket); ensure_equals(pool->restartGroupsByAppRoot(options.appRoot), 1u); debug->debugger->recv("About to end restarting"); ensure_equals(pool->restartGroupsByAppRoot(options.appRoot), 1u); debug->debugger->recv("About to end restarting"); debug->messages->send("Finish restarting"); debug->messages->send("Finish restarting"); debug->debugger->recv("Restarting done"); debug->debugger->recv("Restarting aborted"); } TEST_METHOD(79) { // Test sticky sessions. // Spawn 2 processes and get their sticky session IDs and PIDs. ensureMinProcesses(2); Options options = createOptions(); SessionPtr session1 = pool->get(options, &ticket); SessionPtr session2 = pool->get(options, &ticket); int id1 = session1->getStickySessionId(); int id2 = session2->getStickySessionId(); pid_t pid1 = session1->getPid(); pid_t pid2 = session2->getPid(); session1.reset(); session2.reset(); // Make two requests with id1 as sticky session ID. They should // both go to process pid1. options.stickySessionId = id1; session1 = pool->get(options, &ticket); ensure_equals("Request 1.1 goes to process 1", session1->getPid(), pid1); // The second request should be queued, and should not finish until // the first request is finished. ensure_equals(number, 1); pool->asyncGet(options, callback); SHOULD_NEVER_HAPPEN(100, result = number > 1; ); session1.reset(); EVENTUALLY(1, result = number == 2; ); ensure_equals("Request 1.2 goes to process 1", currentSession->getPid(), pid1); currentSession.reset(); // Make two requests with id2 as sticky session ID. They should // both go to process pid2. options.stickySessionId = id2; session1 = pool->get(options, &ticket); ensure_equals("Request 2.1 goes to process 2", session1->getPid(), pid2); // The second request should be queued, and should not finish until // the first request is finished. pool->asyncGet(options, callback); SHOULD_NEVER_HAPPEN(100, result = number > 2; ); session1.reset(); EVENTUALLY(1, result = number == 3; ); ensure_equals("Request 2.2 goes to process 2", currentSession->getPid(), pid2); currentSession.reset(); } // TODO: Persistent connections. // TODO: If one closes the session before it has reached EOF, and process's maximum concurrency // has already been reached, then the pool should ping the process so that it can detect // when the session's connection has been released by the app. /*********** Test previously discovered bugs ***********/ TEST_METHOD(85) { // Test detaching, then restarting. This should not violate any invariants. TempDirCopy dir("stub/wsgi", "tmp.wsgi"); Options options = createOptions(); options.appRoot = "tmp.wsgi"; options.appType = "wsgi"; options.startupFile = "passenger_wsgi.py"; options.spawnMethod = "direct"; options.statThrottleRate = 0; SessionPtr session = pool->get(options, &ticket); string gupid = session->getProcess()->getGupid().toString(); session.reset(); pool->detachProcess(gupid); touchFile("tmp.wsgi/tmp/restart.txt", 1); pool->get(options, &ticket).reset(); } /*****************************/ }