[Cluster-devel] conga/luci/site/luci/Extensions cluster_adapte ...

[rmccabe@sourceware.org <rmccabe@sourceware.org>]

CVSROOT:	/cvs/cluster
Module name:	conga
Changes by:	rmccabe at sourceware.org	2006-10-24 14:08:50

Modified files:
	luci/site/luci/Extensions: cluster_adapters.py 

Log message:
	more logging/exception code for debugging

Patches:
http://sourceware.org/cgi-bin/cvsweb.cgi/conga/luci/site/luci/Extensions/cluster_adapters.py.diff?cvsroot=cluster&r1=1.125&r2=1.126

--- conga/luci/site/luci/Extensions/cluster_adapters.py	2006/10/20 22:29:22	1.125
+++ conga/luci/site/luci/Extensions/cluster_adapters.py	2006/10/24 14:08:50	1.126
@@ -443,7 +443,7 @@
 		try:
 			resObj = resourceAddHandler[res_type](self, dummy_form)
 		except:
-			luci_log
+			luci_log('res type %d is invalid' % res_type)
 			resObj = None
 
 		if resObj is None:
@@ -2009,6 +2009,7 @@
 		try:
 			clustername = request.form['clusterName']
 		except:
+			luci_log.debug('missing cluster name for NTP')
 			return None
 
 	try:
@@ -2017,20 +2018,21 @@
 		try:
 			nodename = request.form['nodename']
 		except:
+			luci_log.debug('missing nodename name for NTP')
 			return None
 
 	try:
 		task = request['task']
-		if not task:
-			raise
 	except KeyError, e:
 		try:
 			task = request.form['task']
 		except:
+			luci_log.debug('missing task for NTP')
 			return None
 
 	nodename_resolved = resolve_nodename(self, clustername, nodename)
 	if not nodename_resolved or not nodename or not task or not clustername:
+		luci_log.debug('resolve_nodename failed for NTP')
 		return None
 
 	if task != NODE_FENCE:
@@ -2078,22 +2080,42 @@
 			return None
 
 	if task == NODE_LEAVE_CLUSTER:
-		batch_number, result = nodeLeaveCluster(rc)
+		path = str(CLUSTER_FOLDER_PATH + clustername + "/" + nodename_resolved)
 
-		path = CLUSTER_FOLDER_PATH + clustername + "/" + nodename_resolved
-		nodefolder = self.restrictedTraverse(path)
+		try:
+			nodefolder = self.restrictedTraverse(path)
+			if not nodefolder:
+				raise Exception, 'cannot find directory at %s' % path
+		except Exception, e:
+			luci_log.debug('node_leave_cluster err: %s' % str(e))
+			return None
+
+		objname = str(nodename_resolved + "____flag")
+
+		fnpresent = noNodeFlagsPresent(self, nodefolder, objname, nodename_resolved)
+		if fnpresent is None:
+			luci_log.debug('An error occurred while checking flags for %s' \
+				% nodename_resolved)
+			return None
+
+		if fnpresent == False:
+			luci_log.debug('flags are still present for %s -- bailing out' \
+				% nodename_resolved)
+			return None
+
+		batch_number, result = nodeLeaveCluster(rc)
 		batch_id = str(batch_number)
-		objname = nodename_resolved + "____flag"
-		if noNodeFlagsPresent(self, nodefolder, objname, nodename_resolved) == False:
-			raise UnknownClusterError("Fatal", "An unfinished task flag exists for node %s" % nodename)
 
-		nodefolder.manage_addProduct['ManagedSystem'].addManagedSystem(objname)
-		#Now we need to annotate the new DB object
-		objpath = path + "/" + objname
-		flag = self.restrictedTraverse(objpath)
-		flag.manage_addProperty(BATCH_ID,batch_id, "string")
-		flag.manage_addProperty(TASKTYPE,NODE_LEAVE_CLUSTER, "string")
-		flag.manage_addProperty(FLAG_DESC,"Node \'" + nodename + "\' leaving cluster", "string")
+		objpath = str(path + "/" + objname)
+		try:
+			nodefolder.manage_addProduct['ManagedSystem'].addManagedSystem(objname)
+			#Now we need to annotate the new DB object
+			flag = self.restrictedTraverse(objpath)
+			flag.manage_addProperty(BATCH_ID, batch_id, "string")
+			flag.manage_addProperty(TASKTYPE,NODE_LEAVE_CLUSTER, "string")
+			flag.manage_addProperty(FLAG_DESC,"Node \'" + nodename + "\' leaving cluster", "string")
+		except:
+			luci_log.debug('An error occurred while setting flag %s' % objpath)
 
 		response = request.RESPONSE
 		#Is this correct? Should we re-direct to the cluster page?
@@ -2670,22 +2692,29 @@
       try:
         cluname = req.form['clusterName']
       except:
+        luci_log.debug_verbose('No cluster name -- returning empty map')
         return map
 
   path = CLUSTER_FOLDER_PATH + cluname
   try:
     clusterfolder = self.restrictedTraverse(str(path))
     if not clusterfolder:
-      raise
-  except:
+      raise Exception, 'clusterfolder is None'
+  except Exception, e:
+    luci_log.debug_verbose('cluster %s [%s] folder missing: %s -- returning empty map' % (cluname, path, str(e)))
     return map
+  except:
+    luci_log.debug_verbose('cluster %s [%s] folder missing: returning empty map' % (cluname, path))
 
   try:
     items = clusterfolder.objectItems('ManagedSystem')
     if not items or len(items) < 1:
       return map  #This returns an empty map, and should indicate not busy
+  except Exception, e:
+    luci_log.debug('An error occurred while looking for cluster %s flags at path %s: %s' % (cluname, path, str(e)))
+    return map
   except:
-    luci_log.debug('An error occurred while looking for cluster %s flags' % cluname)
+    luci_log.debug('An error occurred while looking for cluster %s flags@path %s' % (cluname, path))
     return map
     
   map['busy'] = "true"
@@ -2716,14 +2745,30 @@
       node_report['desc'] = item[1].getProperty(FLAG_DESC) 
       batch_xml = None
       ricci = item[0].split("____") #This removes the 'flag' suffix
+
       try:
         rc = RicciCommunicator(ricci[0])
-        batch_xml = rc.batch_report(item[1].getProperty(BATCH_ID))
-        if batch_xml != None:
-          (creation_status, total) = batch_status(batch_xml)
+      except RicciError, e:
+        rc = None
+        luci_log.debug_verbose('ricci returned error in iCB for %s: %s' \
+          % (cluname, str(e)))
       except:
-        creation_status = RICCI_CONNECT_FAILURE  #No contact with ricci (-1000)
-        batch_xml = "bloody_failure" #set to avoid next if statement
+        rc = None
+        luci_log.info('ricci connection failed for cluster %s' % cluname)
+
+      if rc is not None:
+        try:
+          batch_xml = rc.batch_report(item[1].getProperty(BATCH_ID))
+          if batch_xml != None:
+            (creation_status, total) = batch_status(batch_xml)
+          else:
+            luci_log.debug_verbose('batch report for cluster %s, item %s is None' % (cluname, item[0]))
+        except:
+          creation_status = RICCI_CONNECT_FAILURE  #No contact with ricci (-1000)
+          batch_xml = "bloody_failure" #set to avoid next if statement
+      else:
+          creation_status = RICCI_CONNECT_FAILURE  #No contact with ricci (-1000)
+          batch_xml = "bloody_failure" #set to avoid next if statement
 
       if batch_xml == None:  #The job is done and gone from queue
         if redirect_message == False: #We have not displayed this message yet
@@ -2732,6 +2777,8 @@
           node_report['errormessage'] = ""
           nodereports.append(node_report)
           redirect_message = True
+
+        luci_log.debug_verbose('batch job is done -- deleting %s' % item[0])
         clusterfolder.manage_delObjects(item[0])
         continue
 
@@ -3620,7 +3667,7 @@
 		items = nodefolder.objectItems('ManagedSystem')
 	except:
 		luci_log.debug('An error occurred while trying to list flags for cluster ' + nodefolder[0])
-		return False
+		return None
 
 	for item in items:
 		if item[0] != flagname:
@@ -3631,7 +3678,7 @@
 			rc = RicciCommunicator(hostname)
 		except RicciError, e:
 			luci_log.info('Unable to connect to the ricci daemon: %s' % str(e))
-			return False
+			return None
 
 		if not rc.authed():
 			try:
@@ -3649,11 +3696,12 @@
 			except Exception, e:
 				luci_log.info('manage_delObjects for %s failed: %s' \
 					% (item[0], str(e)))
-				return False
+				return None
 			return True
 		else:
 			#Not finished, so cannot remove flag
 			return False
+
 	return True
 
 def getModelBuilder(rc, isVirtualized):