Implement flush locking for nodes

Implements a locking mechanism to prevent clobbering of node
flushes. When a flush begins, a global cluster lock is placed
which is freed once the flush completes. While the lock is in place,
other flush events queue waiting for the lock to free before
proceeding.

Modifies the CLI output flow when the `--wait` option is specified.
First, if a lock exists when running the command, the message is
tweaked to indicate this, and the client will wait first for the
lock to free, and then for the flush as normal. Second, the wait
depends on the active lock rather than the domain_status for
consistency purposes.

Closes #32

client-cli/pvc_init.py

@@ -650,6 +650,8 @@ def init_zookeeper(zk_host):
     transaction.create('/ceph', ''.encode('ascii'))
     transaction.create('/ceph/osds', ''.encode('ascii'))
     transaction.create('/ceph/pools', ''.encode('ascii'))
+    transaction.create('/locks', ''.encode('ascii'))
+    transaction.create('/locks/flush_lock', 'False'.encode('ascii'))
     transaction.commit()
 
     # Close the Zookeeper connection

client-common/node.py

@@ -134,24 +134,33 @@ def flush_node(zk_conn, node, wait):
     if not common.verifyNode(zk_conn, node):
         return False, 'ERROR: No node named "{}" is present in the cluster.'.format(node)
 
+    if zkhandler.readdata(zk_conn, '/locks/flush_lock') == 'True':
+        retmsg = 'Flushing hypervisor {} of running VMs. A flush lock currently exists; flush will continue once the lock is freed.'.format(node)
+        lock_wait = True
+    else:
         retmsg = 'Flushing hypervisor {} of running VMs.'.format(node)
+        lock_wait = False
         
     # Wait cannot be triggered from the API
     if wait:
         click.echo(retmsg)
         retmsg = ""
+        if lock_wait:
+            time.sleep(1)
+            while zkhandler.readdata(zk_conn, '/locks/flush_lock') == 'True':
+                time.sleep(1)
+            click.echo('Previous flush completed. Proceeding with flush.')
 
     # Add the new domain to Zookeeper
     zkhandler.writedata(zk_conn, {
         '/nodes/{}/domainstate'.format(node): 'flush'
     })
 
-    if wait == True:
+    # Wait cannot be triggered from the API
-        while True:
+    if wait:
+        time.sleep(1)
+        while zkhandler.readdata(zk_conn, '/locks/flush_lock') == 'True':
             time.sleep(1)
-            node_state = zkhandler.readdata(zk_conn, '/nodes/{}/domainstate'.format(node))
-            if node_state == "flushed":
-                break
 
     return True, retmsg
 

node-daemon/pvcd/NodeInstance.py

@@ -316,6 +316,17 @@ class NodeInstance(object):
 
     # Flush all VMs on the host
     def flush(self):
+        # Wait indefinitely for the flush_lock to be freed
+        time.sleep(0.5)
+        while zkhandler.readdata(self.zk_conn, '/locks/flush_lock') == 'True':
+            time.sleep(2)
+
+        # Acquire the flush lock
+        zkhandler.writedata(self.zk_conn, {
+            '/locks/flush_lock'.format(node): 'True'
+        })
+
+        # Begin flush
         self.inflush = True
         self.logger.out('Flushing node "{}" of running VMs'.format(self.name), state='i')
         self.logger.out('Domain list: {}'.format(', '.join(self.domain_list)))
@@ -347,6 +358,11 @@ class NodeInstance(object):
         zkhandler.writedata(self.zk_conn, { '/nodes/{}/domainstate'.format(self.name): 'flushed' })
         self.inflush = False
 
+        # Release the flush lock
+        zkhandler.writedata(self.zk_conn, {
+            '/locks/flush_lock'.format(node): 'False'
+        })
+
     def unflush(self):
         self.inflush = True
         self.logger.out('Restoring node {} to active service.'.format(self.name), state='i')