CMFActivity: Use a random value for activity uids

Sequential number generators stored in a fixed-size format eventually run
out of values. But activity queues only care about what activities are
currently present: any uid can be reused as soon as it is available.
So stop using a sequential id generator for activity uids, and instead use
random values.

Vincent Pelletier:
- Commit message.
- Minor formatting changes.
- Do probability computations, and increase activity uid storage size to
  64bits integers, up from 32. Table schema migration happens on first
  activity node which starts on upgraded code.
- Apply to SQLJobLib too.

product/CMFActivity/Activity/SQLBase.py

@@ -28,6 +28,9 @@
 
 import sys
 import transaction
+from random import getrandbits
+import MySQLdb
+from MySQLdb.constants.ER import DUP_ENTRY
 from DateTime import DateTime
 from Shared.DC.ZRDB.Results import Results
 from Shared.DC.ZRDB.DA import DatabaseError
@@ -47,6 +50,19 @@ READ_MESSAGE_LIMIT = 1000
 # TODO: Limit by size in bytes instead of number of rows.
 MAX_MESSAGE_LIST_SIZE = 100
 INVOKE_ERROR_STATE = -2
+# Activity uids are stored as 64 bits unsigned integers.
+# No need to depend on a database that supports unsigned integers.
+# Numbers are far big enough without using the MSb. Assuming a busy activity
+# table having one million activities, the probability of triggering a conflict
+# when inserting one activity with 64 bits uid is 0.5e-13. With 63 bits it
+# increases to 1e-13, which is still very low.
+UID_SAFE_BITSIZE = 63
+# Inserting an activity batch of 100 activities among one million existing
+# activities has a probability of failing of 1e-11. While it should be low
+# enough, retries can help lower that. Try 10 times, which should be short
+# enough while yielding one order of magnitude collision probability
+# improvement.
+UID_ALLOCATION_TRY_COUNT = 10
 
 def sort_message_key(message):
   # same sort key as in SQLBase.getMessageList
@@ -126,8 +142,6 @@ class SQLBase(Queue):
     portal = activity_tool.getPortalObject()
     for i in xrange(0, len(registered_message_list), MAX_MESSAGE_LIST_SIZE):
       message_list = registered_message_list[i:i+MAX_MESSAGE_LIST_SIZE]
-      uid_list = portal.portal_ids.generateNewIdList(self.uid_group,
-        id_count=len(message_list), id_generator='uid')
       path_list = ['/'.join(m.object_path) for m in message_list]
       active_process_uid_list = [m.active_process_uid for m in message_list]
       method_id_list = [m.method_id for m in message_list]
@@ -141,9 +155,14 @@ class SQLBase(Queue):
       for m in message_list:
         m.order_validation_text = x = self.getOrderValidationText(m)
         processing_node_list.append(0 if x == 'none' else -1)
+      for _ in xrange(UID_ALLOCATION_TRY_COUNT):
+        try:
           portal.SQLBase_writeMessageList(
             table=self.sql_table,
-        uid_list=uid_list,
+            uid_list=[
+              getrandbits(UID_SAFE_BITSIZE)
+              for _ in xrange(len(message_list))
+            ],
             path_list=path_list,
             active_process_uid_list=active_process_uid_list,
             method_id_list=method_id_list,
@@ -154,6 +173,13 @@ class SQLBase(Queue):
             tag_list=tag_list,
             processing_node_list=processing_node_list,
             serialization_tag_list=serialization_tag_list)
+        except MySQLdb.IntegrityError, (code, _):
+          if code != DUP_ENTRY:
+            raise
+        else:
+          break
+      else:
+        raise ValueError("Maximum retry for SQLBase_writeMessageList reached")
 
   def getNow(self, context):
     """

product/CMFActivity/Activity/SQLJoblib.py

@@ -26,10 +26,15 @@
 #
 ##############################################################################
 
+from random import getrandbits
 from zLOG import LOG, TRACE, INFO, WARNING, ERROR, PANIC
-from SQLBase import SQLBase, sort_message_key, MAX_MESSAGE_LIST_SIZE
+import MySQLdb
+from MySQLdb.constants.ER import DUP_ENTRY
+from SQLBase import (
+  SQLBase, sort_message_key, MAX_MESSAGE_LIST_SIZE,
+  UID_SAFE_BITSIZE, UID_ALLOCATION_TRY_COUNT,
+)
 from Products.CMFActivity.ActivityTool import Message
-
 from SQLDict import SQLDict
 
 class SQLJoblib(SQLDict):
@@ -69,8 +74,6 @@ class SQLJoblib(SQLDict):
     portal = activity_tool.getPortalObject()
     for i in xrange(0, len(registered_message_list), MAX_MESSAGE_LIST_SIZE):
       message_list = registered_message_list[i:i+MAX_MESSAGE_LIST_SIZE]
-      uid_list = portal.portal_ids.generateNewIdList(self.uid_group,
-        id_count=len(message_list), id_generator='uid')
       path_list = ['/'.join(m.object_path) for m in message_list]
       active_process_uid_list = [m.active_process_uid for m in message_list]
       method_id_list = [m.method_id for m in message_list]
@@ -85,8 +88,13 @@ class SQLJoblib(SQLDict):
       for m in message_list:
         m.order_validation_text = x = self.getOrderValidationText(m)
         processing_node_list.append(0 if x == 'none' else -1)
+      for _ in xrange(UID_ALLOCATION_TRY_COUNT):
+        try:
           portal.SQLJoblib_writeMessage(
-        uid_list=uid_list,
+            uid_list=[
+              getrandbits(UID_SAFE_BITSIZE)
+              for _ in xrange(len(message_list))
+            ],
             path_list=path_list,
             active_process_uid_list=active_process_uid_list,
             method_id_list=method_id_list,
@@ -98,6 +106,13 @@ class SQLJoblib(SQLDict):
             processing_node_list=processing_node_list,
             signature_list=signature_list,
             serialization_tag_list=serialization_tag_list)
+        except MySQLdb.IntegrityError, (code, _):
+          if code != DUP_ENTRY:
+            raise
+        else:
+          break
+      else:
+        raise ValueError("Maximum retry for SQLBase_writeMessageList reached")
 
   def getProcessableMessageLoader(self, activity_tool, processing_node):
     path_and_method_id_dict = {}

product/CMFActivity/skins/activity/SQLBase_createMessageTable.zsql

@@ -9,7 +9,7 @@ class_file:
 </dtml-comment>
 <params>table</params>
 CREATE TABLE <dtml-var table> (
-  `uid` INT UNSIGNED NOT NULL,
+  `uid` BIGINT UNSIGNED NOT NULL,
   `date` DATETIME NOT NULL,
   `path` VARCHAR(255) NOT NULL,
   `active_process_uid` INT UNSIGNED NULL,

product/CMFActivity/skins/activity/SQLJoblib_createMessageTable.zsql

@@ -9,7 +9,7 @@ class_file:
 </dtml-comment>
 <params></params>
 CREATE TABLE message_job (
-  `uid` INT UNSIGNED NOT NULL,
+  `uid` BIGINT UNSIGNED NOT NULL,
   `date` DATETIME NOT NULL,
   `path` VARCHAR(255) NOT NULL,
   `active_process_uid` INT UNSIGNED NULL,