various drbd changes

ccnode/drbd.py

@@ -124,19 +124,18 @@ class DRBD(object):
     def destroy(self):
         '''
         '''
-        with self._mutex.write:
-            # bring down drbd device
-            Exec.silent([self.DRBDSETUP,
-                         self._path,
-                         'detach'])
-            Exec.silent([self.DRBDSETUP,
-                         self._path,
-                         'down'])
-            # release device from the pool
-            self._manager._release_device(self._minor)
-            # suicide
-            self._path = None
-            self._minor = None
+        # bring down drbd device
+        Exec.silent([self.DRBDSETUP,
+                     self._path,
+                     'detach'])
+        Exec.silent([self.DRBDSETUP,
+                     self._path,
+                     'down'])
+        # release device from the pool
+        self._manager._release_device(self._minor)
+        # suicide
+        self._path = None
+        self._minor = None
     
     def get_minor(self):
         '''
@@ -179,6 +178,10 @@ class DRBD(object):
             status['rdisk'] = self.DSTATES.get(xres.getAttribute('ds2'))
             status['role']  = self.NROLES.get(xres.getAttribute('ro1'))
             status['rrole'] = self.NROLES.get(xres.getAttribute('ro2'))
+            if xres.hasAttribute('resynced_percent'):
+                status['percent'] = xres.getAttribute('resynced_percent')
+            else:
+                status['percent'] = None
             return status
         except Exception as err:
             return {'conn' : None}
@@ -258,11 +261,10 @@ class DRBD(object):
     def disconnect(self):
         '''
         '''
-        with self._mutex.write:
-            # disconnect from remote node
-            Exec.silent([self.DRBDSETUP,
-                         self._path,
-                         'disconnect'])
+        # disconnect from remote node
+        Exec.silent([self.DRBDSETUP,
+                     self._path,
+                     'disconnect'])
     
     def primary(self):
         '''
@@ -321,7 +323,7 @@ class DRBD(object):
             # FIXME magic sleep, seems to be mandatory
             sleep(0.5)
             status = self.status()
-            print status, self.DSTATES.UP_TO_DATE_str
             if (rc != 0 or not status or
-                            status['disk'] != self.DSTATES.UP_TO_DATE):
+                        ('disk' in status and
+                            status['disk'] != self.DSTATES.UP_TO_DATE)):
                 raise DRBDError('synchronization failed')

ccnode/handlers.py

@@ -946,17 +946,22 @@ class NodeHandler(RpcHandler):
         job.drbd_takeover(state)
     
     @pure
-    def drbd_sync(self, res):
+    def drbd_sync_status(self, res):
         '''
-        Wait for the live copy engine to complete a pending initial disk
-        synchronization.
+        Get status information about a running sync
         '''
         # facilities
         jobmgr = self._host_handle.jobmgr
         # get the job
         job = jobmgr.get_job(res['jid'])
-        # wait for sync completion
-        job.drbd_waitsync()
+        # get drbd status
+        status = job.drbd_status()
+        # return status
+        return {
+            'done' : status['disk'] == job._drbd.DSTATES.UP_TO_DATE,
+            'completion' : status['percent'],
+            'speed' : None,
+            }
     
     @pure
     def drbd_shutdown(self, res):
@@ -965,10 +970,9 @@ class NodeHandler(RpcHandler):
         '''
         # facilities
         jobmgr = self._host_handle.jobmgr
-        # get the job
-        job = jobmgr.get_job(res['jid'])
-        # FIXME
-        job.drbd_stop()
+        # cancel the job so it can clean itself
+        # FIXME: check type, we don't want to cancel anything
+        jobmgr.cancel(res['jid'])
     
     ##################################
     #   Job management

ccnode/jobs.py

@@ -610,8 +610,14 @@ class DrbdCopyJob(BaseJob):
         '''
         '''
         # release some pending calls
+        self._event_connected_ready.set()
+        self._event_connected.set()
+        self._event_rolechosen_ready.set()
+        self._event_rolechosen.set()
         self._event_sleep_ready.set()
         self._event_sleep.set()
+        # disconnect the node
+        self._drbd.disconnect()
     
     def _cleanup(self):
         '''
@@ -731,7 +737,7 @@ class DrbdCopyJob(BaseJob):
                                                         self._drbd.get_path())
                 self._drbd.setup(self._copy_path, self._meta_path)
                 self._drbd_table = '0 %d linear %s 0' % (self._source_size / 512
-                                                        , self._source_path)
+                                                    , self._drbd.get_path())
             except Exception as err:
                 self._log.append('failed to configure the drbd device')
                 raise err
@@ -772,6 +778,8 @@ class DrbdCopyJob(BaseJob):
             self._log.append('waiting for initial device sync')
             self._drbd.wait_sync()
         except Exception as err:
+            import traceback
+            traceback.print_exc()
             self._log.append('failed to sync the DRBD')
             raise err
         else:
@@ -828,7 +836,7 @@ class DrbdCopyJob(BaseJob):
         self._log.append('waiting for sync')
         # wait for initial sync and setup to be completed
         self._event_sleep_ready.wait()
-        # wait for DRBD sync
+        # wait for additionnal DRBD sync
         with self._mutex_step:
             self._drbd.wait_sync()
     
@@ -849,16 +857,10 @@ class DrbdCopyJob(BaseJob):
             #
             self._lvm.dm_set_table(self._source_dm, table)
     
-    def drbd_stop(self):
+    def drbd_status(self):
         '''
         '''
-        self._event_sleep_ready.wait(5) # to fix thread sync issue
-        if self._event_sleep_ready.is_set():
-            with self._mutex_step:
-                # unlock job termination
-                self._event_sleep.set()
-        else:
-            raise DrbdCopyJobError('can\'t stop now')
+        return self._drbd.status()
 
 
 class TCPTunnelJob(BaseJob):