############################################################################## # # Copyright (c) 2002 Nexedi SARL and Contributors. All Rights Reserved. # Jean-Paul Smets-Solanes <jp@nexedi.com> # # WARNING: This program as such is intended to be used by professional # programmers who take the whole responsability of assessing all potential # consequences resulting from its eventual inadequacies and bugs # End users who are looking for a ready-to-use solution with commercial # garantees and support are strongly adviced to contract a Free Software # Service Company # # This program is Free Software; you can redistribute it and/or # modify it under the terms of the GNU General Public License # as published by the Free Software Foundation; either version 2 # of the License, or (at your option) any later version. # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the # GNU General Public License for more details. # # You should have received a copy of the GNU General Public License # along with this program; if not, write to the Free Software # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. # ############################################################################## import random from DateTime import DateTime from Products.CMFActivity.ActivityTool import registerActivity from Queue import VALID, INVALID_ORDER, INVALID_PATH, EXCEPTION, MAX_PROCESSING_TIME, VALIDATION_ERROR_DELAY, SECONDS_IN_DAY from RAMDict import RAMDict from Products.CMFActivity.ActiveObject import DISTRIBUTABLE_STATE, INVOKE_ERROR_STATE, VALIDATE_ERROR_STATE from zLOG import LOG MAX_PRIORITY = 5 priority_weight = \ [1] * 64 + \ [2] * 20 + \ [3] * 10 + \ [4] * 5 + \ [5] * 1 class ActivityFlushError(Exception): """Error during active message flush""" class SQLDict(RAMDict): """ A simple OOBTree based queue. It should be compatible with transactions and provide sequentiality. Should not create conflict because use of OOBTree. """ # Transaction commit methods def prepareQueueMessage(self, activity_tool, m): if m.is_registered: activity_tool.SQLDict_writeMessage( path = '/'.join(m.object_path) , method_id = m.method_id, priority = m.activity_kw.get('priority', 1), broadcast = m.activity_kw.get('broadcast', 0), message = self.dumpMessage(m), date = m.activity_kw.get('at_date', DateTime())) # Also store uid of activity def prepareDeleteMessage(self, activity_tool, m): # Erase all messages in a single transaction path = '/'.join(m.object_path) uid_list = activity_tool.SQLDict_readUidList(path=path, method_id=m.method_id,processing_node=None) uid_list = map(lambda x:x.uid, uid_list) if len(uid_list)>0: activity_tool.SQLDict_delMessage(uid = uid_list) # Registration management def registerActivityBuffer(self, activity_buffer): class_name = self.__class__.__name__ setattr(activity_buffer, '_%s_uid_dict' % class_name, {}) setattr(activity_buffer, '_%s_message_list' % class_name, []) def isMessageRegistered(self, activity_buffer, activity_tool, m): class_name = self.__class__.__name__ uid_dict = getattr(activity_buffer,'_%s_uid_dict' % class_name) return uid_dict.has_key((tuple(m.object_path), m.method_id)) def registerMessage(self, activity_buffer, activity_tool, m): m.is_registered = 1 class_name = self.__class__.__name__ uid_dict = getattr(activity_buffer,'_%s_uid_dict' % class_name) uid_dict[(tuple(m.object_path), m.method_id)] = 1 getattr(activity_buffer,'_%s_message_list' % class_name).append(m) def unregisterMessage(self, activity_buffer, activity_tool, m): m.is_registered = 0 # This prevents from inserting deleted messages into the queue class_name = self.__class__.__name__ uid_dict = getattr(activity_buffer,'_%s_uid_dict' % class_name) if uid_dict.has_key((tuple(m.object_path), m.method_id)): del uid_dict[(tuple(m.object_path), m.method_id)] def getRegisteredMessageList(self, activity_buffer, activity_tool): class_name = self.__class__.__name__ if hasattr(activity_buffer,'_%s_message_list' % class_name): message_list = getattr(activity_buffer,'_%s_message_list' % class_name) return filter(lambda m: m.is_registered, message_list) else: return () # Queue semantic def dequeueMessage(self, activity_tool, processing_node): if hasattr(activity_tool,'SQLDict_readMessage'): now_date = DateTime() # Next processing date in case of error next_processing_date = now_date + VALIDATION_ERROR_DELAY priority = random.choice(priority_weight) # Try to find a message at given priority level which is scheduled for now result = activity_tool.SQLDict_readMessage(processing_node=processing_node, priority=priority, to_date=now_date) if len(result) == 0: # If empty, take any message which is scheduled for now priority = None result = activity_tool.SQLDict_readMessage(processing_node=processing_node, priority=priority, to_date=now_date) if len(result) > 0: line = result[0] path = line.path method_id = line.method_id uid_list = activity_tool.SQLDict_readUidList( path=path, method_id= method_id, processing_node = None, to_date=now_date ) uid_list = map(lambda x:x.uid, uid_list) # Make sure message can not be processed anylonger if len(uid_list) > 0: # Set selected messages to processing activity_tool.SQLDict_processMessage(uid = uid_list) get_transaction().commit() # Release locks before starting a potentially long calculation # This may lead (1 for 1,000,000 in case of reindexing) to messages left in processing state m = self.loadMessage(line.message, uid = line.uid) # Validate message (make sure object exists, priority OK, etc.) validation_state = m.validate(self, activity_tool) if validation_state is not VALID: if validation_state in (EXCEPTION, INVALID_PATH): # There is a serious validation error - we must lower priority if line.priority > MAX_PRIORITY: # This is an error if len(uid_list) > 0: activity_tool.SQLDict_assignMessage(uid = uid_list, processing_node = VALIDATE_ERROR_STATE) # Assign message back to 'error' state #m.notifyUser(activity_tool) # Notify Error get_transaction().commit() # and commit else: # Lower priority if len(uid_list) > 0: # Add some delay before new processing activity_tool.SQLDict_setPriority(uid = uid_list, date = next_processing_date, priority = line.priority + 1) get_transaction().commit() # Release locks before starting a potentially long calculation else: # We do not lower priority for INVALID_ORDER errors but we do postpone execution if len(uid_list) > 0: # Add some delay before new processing activity_tool.SQLDict_setPriority(uid = uid_list, date = next_processing_date, priority = line.priority) get_transaction().commit() # Release locks before starting a potentially long calculation else: # Try to invoke activity_tool.invoke(m) # Try to invoke the message - what happens if read conflict error restarts transaction ? if m.is_executed: # Make sure message could be invoked if len(uid_list) > 0: activity_tool.SQLDict_delMessage(uid = uid_list) # Delete it get_transaction().commit() # If successful, commit if m.active_process: active_process = activity_tool.unrestrictedTraverse(m.active_process) if not active_process.hasActivity(): # Not more activity m.notifyUser(activity_tool, message="Process Finished") # XXX commit bas ??? else: get_transaction().abort() # If not, abort transaction and start a new one if line.priority > MAX_PRIORITY: # This is an error if len(uid_list) > 0: activity_tool.SQLDict_assignMessage(uid = uid_list, processing_node = INVOKE_ERROR_STATE) # Assign message back to 'error' state m.notifyUser(activity_tool) # Notify Error get_transaction().commit() # and commit else: # Lower priority if len(uid_list) > 0: activity_tool.SQLDict_setPriority(uid = uid_list, date = next_processing_date, priority = line.priority + 1) get_transaction().commit() # Release locks before starting a potentially long calculation return 0 get_transaction().commit() # Release locks before starting a potentially long calculation return 1 def hasActivity(self, activity_tool, object, **kw): if hasattr(activity_tool,'SQLDict_readMessageList'): if object is not None: my_object_path = '/'.join(object.getPhysicalPath()) result = activity_tool.SQLDict_hasMessage(path=my_object_path, **kw) if len(result) > 0: return result[0].message_count > 0 else: return 1 # Default behaviour if no object specified is to return 1 until active_process implemented return 0 def flush(self, activity_tool, object_path, invoke=0, method_id=None, commit=0, **kw): """ object_path is a tuple commit allows to choose mode - if we commit, then we make sure no locks are taken for too long - if we do not commit, then we can use flush in a larger transaction commit should in general not be used NOTE: commiting is very likely nonsenses here. We should just avoid to flush as much as possible """ path = '/'.join(object_path) # LOG('Flush', 0, str((path, invoke, method_id))) method_dict = {} if hasattr(activity_tool,'SQLDict_readMessageList'): # Parse each message in registered for m in activity_tool.getRegisteredMessageList(self): if list(m.object_path) == list(object_path) and (method_id is None or method_id == m.method_id): activity_tool.unregisterMessage(self, m) #if not method_dict.has_key(method_id or m.method_id): if not method_dict.has_key(m.method_id): method_dict[m.method_id] = 1 # Prevents calling invoke twice if invoke: # First Validate validate_value = m.validate(self, activity_tool) if validate_value is VALID: activity_tool.invoke(m) # Try to invoke the message - what happens if invoke calls flushActivity ?? if not m.is_executed: # Make sure message could be invoked # The message no longer exists raise ActivityFlushError, ( 'Could not evaluate %s on %s' % (m.method_id , path)) elif validate_value is INVALID_PATH: # The message no longer exists raise ActivityFlushError, ( 'The document %s does not exist' % path) # Parse each message in SQL dict result = activity_tool.SQLDict_readMessageList(path=path, method_id=method_id,processing_node=None) for line in result: path = line.path method_id = line.method_id if not method_dict.has_key(method_id): # Only invoke once (it would be different for a queue) # This is optimisation with the goal to process objects on the same # node and minimize network traffic with ZEO server method_dict[method_id] = 1 m = self.loadMessage(line.message, uid = line.uid) self.deleteMessage(activity_tool, m) if invoke: # First Validate validate_value = m.validate(self, activity_tool) LOG('SQLDict.flush validate_value',0,validate_value) if validate_value is VALID: activity_tool.invoke(m) # Try to invoke the message - what happens if invoke calls flushActivity ?? LOG('SQLDict.flush m.is_executed',0,m.is_executed) if not m.is_executed: # Make sure message could be invoked # The message no longer exists raise ActivityFlushError, ( 'Could not evaluate %s on %s' % (m.method_id , path)) if validate_value is INVALID_PATH: # The message no longer exists raise ActivityFlushError, ( 'The document %s does not exist' % path) def getMessageList(self, activity_tool, processing_node=None): # YO: reading all lines might cause a deadlock message_list = [] if hasattr(activity_tool,'SQLDict_readMessageList'): result = activity_tool.SQLDict_readMessageList(path=None, method_id=None, processing_node=None, to_processing_date=None) for line in result: m = self.loadMessage(line.message, uid = line.uid) m.processing_node = line.processing_node m.priority = line.priority message_list.append(m) return message_list def distribute(self, activity_tool, node_count): processing_node = 1 if hasattr(activity_tool,'SQLDict_readMessageList'): now_date = DateTime() if (now_date - self.max_processing_date) > MAX_PROCESSING_TIME: # Sticky processing messages should be set back to non processing max_processing_date = now_date - MAX_PROCESSING_TIME self.max_processing_date = now_date else: max_processing_date = None result = activity_tool.SQLDict_readMessageList(path=None, method_id=None, processing_node = -1, to_processing_date = max_processing_date) # Only assign non assigned messages get_transaction().commit() # Release locks before starting a potentially long calculation path_dict = {} for line in result: path = line.path broadcast = line.broadcast if broadcast: # Broadcast messages must be distributed into all nodes. uid = line.uid activity_tool.SQLDict_assignMessage(processing_node=1, uid=[uid]) for node in range(2, node_count+1): activity_tool.SQLDict_writeMessage( path = path, method_id = line.method_id, priority = line.priority, broadcast = 1, processing_node = node, message = line.message, date = line.date) elif not path_dict.has_key(path): # Only assign once (it would be different for a queue) path_dict[path] = 1 activity_tool.SQLDict_assignMessage(path=path, processing_node=processing_node, uid=None, broadcast=0) get_transaction().commit() # Release locks immediately to allow processing of messages processing_node = processing_node + 1 if processing_node > node_count: processing_node = 1 # Round robin # Validation private methods def _validate_after_method_id(self, activity_tool, message, value): # Count number of occurances of method_id if type(value) == type(''): value = [value] result = activity_tool.SQLDict_validateMessageList(method_id=value, message_uid=None, path=None) LOG('SQLDict._validate_after_method_id, method_id',0,value) LOG('SQLDict._validate_after_method_id, result[0].uid_count',0,result[0].uid_count) if result[0].uid_count > 0: return INVALID_ORDER return VALID def _validate_after_path(self, activity_tool, message, value): # Count number of occurances of path result = activity_tool.SQLDict_validateMessageList(method_id=None, message_uid=None, path=value) if result[0].uid_count > 0: return INVALID_ORDER return VALID def _validate_after_message_uid(self, activity_tool, message, value): # Count number of occurances of message_uid result = activity_tool.SQLDict_validateMessageList(method_id=None, message_uid=value, path=None) if result[0].uid_count > 0: return INVALID_ORDER return VALID # Required for tests (time shift) def timeShift(self, activity_tool, delay): """ To simulate timeShift, we simply substract delay from all dates in SQLDict message table """ activity_tool.SQLDict_timeShift(delay = delay * SECONDS_IN_DAY) registerActivity(SQLDict)