Remove gpexpand --novacuum option, and never vacuum the catalogs.

The vacuum was causing trouble, see github issue #4298. And it seems pretty
pointless or unrelated to gpexpand, anyway.

gpMgmt/bin/gpexpand

@@ -120,8 +120,6 @@ def parseargs():
                       help='remove the expansion schema.')
     parser.add_option('-r', '--rollback', action='store_true',
                       help='rollback failed expansion setup.')
-    parser.add_option('-V', '--novacuum', action='store_true',
-                      help='Do not vacuum catalog tables before creating schema copy.')
     parser.add_option('-a', '--analyze', action='store_true',
                       help='Analyze the expanded table after redistribution.')
     parser.add_option('-d', '--duration', type='duration', metavar='[h][:m[:s]]',
@@ -634,13 +632,12 @@ class SegmentTemplate:
 
     def __init__(self, logger, statusLogger, pool,
                  gparray, masterDataDirectory,
-                 dburl, conn, noVacuumCatalog, tempDir, batch_size,
+                 dburl, conn, tempDir, batch_size,
                  segTarDir='.', schemaTarFile='gpexpand_schema.tar'):
         self.logger = logger
         self.statusLogger = statusLogger
         self.pool = pool
         self.gparray = gparray
-        self.noVacuumCatalog = noVacuumCatalog
         self.tempDir = tempDir
         self.batch_size = batch_size
         self.dburl = dburl
@@ -702,10 +699,6 @@ class SegmentTemplate:
         """Creates the schema template that is used by new segments"""
         self.logger.info('Creating segment template')
 
-        if not self.noVacuumCatalog:
-            self.logger.info('VACUUM FULL on the catalog tables')
-            catalog.vacuum_catalog(self.dburl, self.conn, full=True, utility=True)
-
         MakeDirectory.local('gpexpand create temp dir', self.tempDir)
 
         self._select_src_segment()
@@ -1337,7 +1330,6 @@ Set PGDATABASE or use the -D option to specify the correct database to use.""" %
                                            masterDataDirectory=self.options.master_data_directory,
                                            dburl=self.dburl,
                                            conn=self.conn,
-                                           noVacuumCatalog=self.options.novacuum,
                                            tempDir=self.tempDir,
                                            segTarDir=self.options.tardir,
                                            batch_size=self.options.batch_size)

gpMgmt/bin/gppylib/db/catalog.py

@@ -88,50 +88,3 @@ def get_catalogtable_list(conn):
     finally:
         if cursor:
             cursor.close()
-
-
-def vacuum_catalog(dburl,conn,full=False,utility=False):
-    """ Will use the provided connection to enumerate the list of databases
-        and then connect to each one in turn and vacuum full all of the 
-        catalog files
-        
-        TODO:  There are a few tables that are cluster-wide that strictly speaking
-               don't need to be vacuumed for each database.  These are most likely
-               small and so perhaps isn't worth the added complexity to optimize them.
-    
-        WARNING:  doing a vacuum full on the catalog requires that 
-        there aren't any users idle in a transaction as they typically
-        hold catalog share locks.  The result is this vacuum will wait forever on
-        getting the lock.  This method is best called when no one else
-        is connected to the system.  Our own connections are typically idle
-        in transactions and so are especially bad.
-    """
-    dblist = getDatabaseList(conn)
-    catlist = get_catalogtable_list(conn)
-    conn.commit()
-    
-    for db in dblist:
-        test_url = copy.deepcopy(dburl)
-        test_url.pgdb = db[0]
-        
-        if db[0] == 'template0' or db[0] == 'postgres':
-            continue
-        
-        vac_conn = dbconn.connect(test_url,utility)
-        vac_curs = vac_conn.cursor()
-        vac_curs.execute("COMMIT")
-        vac_curs.execute("SET CLIENT_MIN_MESSAGES='ERROR'")
-        for table in catlist:
-            logger.debug('Vacuuming %s %s' % (db[0],table[0]) )
-            
-            if full:
-                sql = "VACUUM FULL %s" % table[0]
-            else:
-                sql = "VACUUM %s" % table[0]
-            
-            vac_curs.execute(sql)
-            
-        
-        vac_curs.execute(sql)
-        vac_conn.commit()
-        vac_conn.close()

gpMgmt/bin/gppylib/db/test/regress/test_regress_catalog.py

@@ -30,11 +30,6 @@ class catalogTestCase(unittest.TestCase):
         self.conn.close()
         pass
 
-    def test_vacuumcatalog(self):
-        logger.info("test_vacuumcatalog")
-        catalog.vacuum_catalog(self.dburl,self.conn)
-        catalog.vacuum_catalog(self.dburl,self.conn,full=True)
-    
 #------------------------------- Mainline --------------------------------
 if __name__ == '__main__':
     unittest.main()    

gpMgmt/bin/gppylib/db/test/test_catalog.py

@@ -30,11 +30,6 @@ class catalogTestCase(unittest.TestCase):
         self.conn.close()
         pass
     
-    def test_vacuumcatalog(self):
-        logger.info("test_vacuumcatalog")
-        catalog.vacuum_catalog(self.dburl,self.conn)
-        catalog.vacuum_catalog(self.dburl,self.conn,full=True)
-    
 #------------------------------- Mainline --------------------------------
 if __name__ == '__main__':
     unittest.main()    

gpdb-doc/dita/utility_guide/admin_utilities/gpexpand.xml

@@ -8,7 +8,7 @@
     <section id="section2">
       <title>Synopsis</title>
       <codeblock><b>gpexpand</b> [{<b>-f</b>|<b>--hosts-file</b>} <varname>hosts_file</varname>]
-        | {<b>-i</b>|<b>--input</b>} <varname>input_file</varname> [<b>-B</b> <varname>batch_size</varname>] [<b>-V</b>|<b>--novacuum</b>] 
+        | {<b>-i</b>|<b>--input</b>} <varname>input_file</varname> [<b>-B</b> <varname>batch_size</varname>]
         | {{<b>-d</b> | <b>--duration</b>} <varname>hh:mm:ss</varname> | {<b>-e</b>|<b>--end</b>} '<varname>YYYY-MM-DD hh:mm:ss</varname>'} 
           [<b>-a</b>|<b>-analyze</b>] 
           [<b>-n </b> <varname>parallel_processes</varname>]
@@ -187,10 +187,6 @@
           <pt>--version</pt>
           <pd>Display the utility's version number and exit.</pd>
         </plentry>
-        <plentry>
-          <pt>-V | --novacuum</pt>
-          <pd>Do not vacuum catalog tables before creating schema copy.</pd>
-        </plentry>
         <plentry>
           <pt>-? | -h | --help</pt>
           <pd>Displays the online help.</pd>