Wait for housekeeping to finish before bbstored exits

Should fix random test failures due to housekeeping process realising that the master has asked it to terminate, and logging that it has, after the master has already died and the test finished, confusing the test runner with its extra output. Should also help make the occasionally-reported housekeeping crashes more obvious and easier to debug. (cherry picked from commit 0100d31)
boxbackup · Jan 15, 2018 · 2cc729b · 2cc729b
1 parent aaa5301
commit 2cc729b
Show file tree

Hide file tree

Showing 2 changed files with 44 additions and 1 deletion.
diff --git a/lib/bbstored/BBStoreDHousekeeping.cpp b/lib/bbstored/BBStoreDHousekeeping.cpp
@@ -217,6 +217,14 @@ bool BackupStoreDaemon::CheckForInterProcessMsg(int AccountNum, int MaximumWaitT
 		return false;
 	}
 
+	// First, check to see if it's EOF -- this means something has gone wrong, and the housekeeping should terminate.
+	if(mInterProcessComms.IsEOF())
+	{
+		BOX_INFO("Housekeeping process was hungup by main daemon, terminating");
+		SetTerminateWanted();
+		return true;
+	}
+
 	// Get a line, and process the message
 	std::string line;
 

diff --git a/lib/bbstored/BackupStoreDaemon.cpp b/lib/bbstored/BackupStoreDaemon.cpp
@@ -171,6 +171,7 @@ void BackupStoreDaemon::Run()
 	mExtendedLogging = false;
 	const Configuration &config(GetConfiguration());
 	mExtendedLogging = config.GetKeyValueBool("ExtendedLogging");
+	int housekeeping_pid;
 
 	// Fork off housekeeping daemon -- must only do this the first
 	// time Run() is called.  Housekeeping runs synchronously on Win32
@@ -190,7 +191,8 @@ void BackupStoreDaemon::Run()
 		int whichSocket = 0;
 
 		// Fork
-		switch(::fork())
+		housekeeping_pid = ::fork();
+		switch(housekeeping_pid)
 		{
 		case -1:
 			{
@@ -261,6 +263,39 @@ void BackupStoreDaemon::Run()
 		if(IsTerminateWanted())
 		{
 			mInterProcessCommsSocket.Write("t\n", 2);
+
+#ifndef WIN32 // waitpid() only works with fork(), and thus not on Windows
+			// Wait for housekeeping to finish, and report its status, otherwise it
+			// can remain running and write to the console after the main process has
+			// died, which can confuse the tests.
+			int child_state;
+			if(waitpid(housekeeping_pid, &child_state, 0) == -1)
+			{
+				BOX_LOG_SYS_ERROR("Failed to wait for housekeeping child process "
+					"to finish");
+			}
+			else if(WIFEXITED(child_state))
+			{
+				if(WEXITSTATUS(child_state) == 0)
+				{
+					BOX_TRACE("Housekeeping child process exited normally");
+				}
+				else
+				{
+					BOX_ERROR("Housekeeping child process exited with "
+						"status " << WEXITSTATUS(child_state));
+				}
+			}
+			else if(WIFSIGNALED(child_state))
+			{
+				BOX_ERROR("Housekeeping child process terminated with signal " <<
+					WTERMSIG(child_state));
+			}
+			else
+			{
+				BOX_ERROR("Housekeeping child process terminated in unexpected manner");
+			}
+#endif // !WIN32
 		}
 	}
 }