merge?

tools/python/xen/lowlevel/xc/xc.c

@@ -220,6 +220,9 @@ static PyObject *pyxc_domain_getinfo(PyObject *self,
         return PyErr_NoMemory();
 
     nr_doms = xc_domain_getinfo(xc->xc_handle, first_dom, max_doms, info);
+
+    if (nr_doms < 0)
+        return PyErr_SetFromErrno(xc_error);
     
     list = PyList_New(nr_doms);
     for ( i = 0 ; i < nr_doms; i++ )

tools/python/xen/web/httpserver.py

@@ -273,6 +273,9 @@ class HttpServer:
         self.interface = interface
         self.port = port
         self.root = root
+        # ready indicates when we are ready to begin accept connections
+        # it should be set after a successful bind
+        self.ready = False
 
     def getRoot(self):
         return self.root
@@ -283,6 +286,7 @@ class HttpServer:
     def run(self):
         self.bind()
         self.listen()
+        self.ready = True
         self.requestLoop()
 
     def stop(self):

tools/python/xen/web/tcp.py

@@ -18,6 +18,7 @@
 import sys
 import socket
 import types
+import time
 
 from connection import *
 from protocol import *
@@ -35,8 +36,25 @@ class TCPListener(SocketListener):
     def createSocket(self):
         sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
         sock.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
-        addr = (self.interface, self.port)
-        sock.bind(addr)
+
+        # SO_REUSEADDR does not always ensure that we do not get an address
+        # in use error when restarted quickly
+        # we implement a timeout to try and avoid failing unnecessarily
+
+        timeout = time.time() + 30
+        again = True
+        while again and time.time() < timeout:
+            again = False
+            try:
+                sock.bind((self.interface, self.port))
+            except socket.error, (errno, strerrno):
+                if errno == 98:
+                    again = True
+                else:
+                    raise socket.error(errno, strerrno)
+        if again:
+            raise socket.error(98, "address in use")
+                
         return sock
 
     def acceptConnection(self, sock, protocol, addr):

tools/python/xen/xend/XendDomainInfo.py

@@ -110,9 +110,13 @@ def dom_get(dom):
     @param dom: domain id
     @return: info or None
     """
-    domlist = xc.domain_getinfo(dom, 1)
-    if domlist and dom == domlist[0]['dom']:
-        return domlist[0]
+    try:
+        domlist = xc.domain_getinfo(dom, 1)
+        if domlist and dom == domlist[0]['dom']:
+            return domlist[0]
+    except Exception, err:
+        # ignore missing domain
+        log.exception("domain_getinfo(%d) failed, ignoring", dom)
     return None
 
 class XendDomainInfo:
@@ -349,7 +353,12 @@ class XendDomainInfo:
     def update(self, info=None):
         """Update with  info from xc.domain_getinfo().
         """
-        self.info = info or dom_get(self.domid)
+        if info:
+            self.info = info
+        else:
+            di = dom_get(self.domid)
+            if not di:
+                return 
         self.memory = self.info['mem_kb'] / 1024
         self.ssidref = self.info['ssidref']
 

tools/python/xen/xend/server/SrvDaemon.py

@@ -137,13 +137,6 @@ class Daemon:
         else:
             return 0
 
-    def onSIGCHLD(self, signum, frame):
-        if self.child > 0: 
-            try: 
-                pid, sts = os.waitpid(self.child, os.WNOHANG)
-            except os.error, ex:
-                pass
-
     def fork_pid(self, pidfile):
         """Fork and write the pid of the child to 'pidfile'.
 
@@ -200,15 +193,29 @@ class Daemon:
             # Trying to run an already-running service is a success.
             return 0
 
-        signal.signal(signal.SIGCHLD, self.onSIGCHLD)
+        ret = 0
+
+        # we use a pipe to communicate between the parent and the child process
+        # this way we know when the child has actually initialized itself so
+        # we can avoid a race condition during startup
+        
+        r,w = os.pipe()
         if self.fork_pid(XEND_PID_FILE):
-            #Parent. Sleep to give child time to start.
-            time.sleep(1)
+            os.close(w)
+            r = os.fdopen(r, 'r')
+            s = r.read()
+            r.close()
+            if not len(s):
+                ret = 1
+            else:
+                ret = int(s)
         else:
+            os.close(r)
             # Child
             self.tracing(trace)
-            self.run()
-        return 0
+            self.run(os.fdopen(w, 'w'))
+
+        return ret
 
     def tracing(self, traceon):
         """Turn tracing on or off.
@@ -290,7 +297,7 @@ class Daemon:
     def stop(self):
         return self.cleanup(kill=True)
 
-    def run(self):
+    def run(self, status):
         _enforce_dom0_cpus()
         try:
             log.info("Xend Daemon started")
@@ -298,12 +305,14 @@ class Daemon:
             relocate.listenRelocation()
             servers = SrvServer.create()
             self.daemonize()
-            servers.start()
+            servers.start(status)
         except Exception, ex:
             print >>sys.stderr, 'Exception starting xend:', ex
             if XEND_DEBUG:
                 traceback.print_exc()
             log.exception("Exception starting xend (%s)" % ex)
+            status.write('1')
+            status.close()
             self.exit(1)
             
     def exit(self, rc=0):

tools/python/xen/xend/server/SrvServer.py

@@ -48,6 +48,7 @@ from xen.xend import XendRoot; xroot = XendRoot.instance()
 from xen.xend import Vifctl
 from xen.xend.XendLogging import log
 from xen.web.SrvDir import SrvDir
+import time
 
 from SrvRoot import SrvRoot
 
@@ -59,7 +60,7 @@ class XendServers:
     def add(self, server):
         self.servers.append(server)
 
-    def start(self):
+    def start(self, status):
         Vifctl.network('start')
         threads = []
         for server in self.servers:
@@ -67,6 +68,25 @@ class XendServers:
             thread.start()
             threads.append(thread)
 
+
+        # check for when all threads have initialized themselves and then
+        # close the status pipe
+
+        threads_left = True
+        while threads_left:
+            threads_left = False
+
+            for server in self.servers:
+                if not server.ready:
+                    threads_left = True
+                    break
+
+            if threads_left:
+                time.sleep(.5)
+
+        status.write('0')
+        status.close()
+
         for t in threads:
             t.join()