From 8781c47d0ddac22192f1c233606d3a2923abd6f9 Mon Sep 17 00:00:00 2001
From: Tim Laszlo <tim.laszlo@gmail.com>
Date: Sat, 5 Jun 2010 13:41:23 +0000
Subject: Performance improvements for bcfg2-admin reports scrub.

Switched object updates to executemany statements.
Added BatchFetch to retreive Django objects in groups.

git-svn-id: https://svn.mcs.anl.gov/repos/bcfg/trunk/bcfg2@5891 ce84e21b-d406-0410-9b95-82705330c041
---
 src/lib/Server/Admin/Reports.py | 25 +++++++++++--------------
 src/lib/Server/Reports/utils.py | 30 ++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 14 deletions(-)
 create mode 100755 src/lib/Server/Reports/utils.py

(limited to 'src')

diff --git a/src/lib/Server/Admin/Reports.py b/src/lib/Server/Admin/Reports.py
index e712ae77d..5d0595869 100755
--- a/src/lib/Server/Admin/Reports.py
+++ b/src/lib/Server/Admin/Reports.py
@@ -9,6 +9,7 @@ import platform
 import sys
 from Bcfg2.Server.Reports.importscript import load_stats
 from Bcfg2.Server.Reports.updatefix import update_database
+from Bcfg2.Server.Reports.utils import *
 from lxml.etree import XML, XMLSyntaxError
 
 try:
@@ -117,7 +118,8 @@ class Reports(Bcfg2.Server.Admin.Mode):
         dup_reasons = []
     
         cmp_reasons = dict()
-        for reason in Reason.objects.all():
+        batch_update = []
+        for reason in BatchFetch(Reason.objects):
             ''' Loop through each reason and create a key out of the data. \
                 This lets us take advantage of a fast hash lookup for \
                 comparisons '''
@@ -129,24 +131,19 @@ class Reports(Bcfg2.Server.Admin.Mode):
             if key in cmp_reasons:
                 self.log.debug("Update interactions from %d to %d" \
                                     % (reason.id, cmp_reasons[key]))
-                try:
-                    Entries_interactions.objects.filter(reason=reason).\
-                                    update(reason=cmp_reasons[key])
-                    dup_reasons.append(reason.id)
-                except Exception, ex:
-                    self.log.error("Failed to update interactions for %d : %s" \
-                                    % (reason.id, ex))
+                dup_reasons.append(reason.id)
+                batch_update.append([cmp_reasons[key], reason.id])
             else:
                 cmp_reasons[key] = reason.id
             self.log.debug("key %d" % reason.id)
     
         self.log.debug("Done with updates, deleting dupes")
-        for dup in dup_reasons:
-            self.log.debug("Deleting %d" % dup)
-            try:
-                Reason.objects.get(id=dup).delete()
-            except Exception, ex:
-                self.log.error("Failed to delete reason %d: %s" % (dup, ex))
+        try:
+            cursor = connection.cursor()
+            cursor.executemany('update reports_entries_interactions set reason_id=%s where reason_id=%s', batch_update)
+            cursor.executemany('delete from reports_reason where id = %s', dup_reasons)
+        except Exception, ex:
+            self.log.error("Failed to delete reasons: %s" % ex)
 
         self.log.info("Found %d dupes out of %d" % (len(dup_reasons), start_count))
 
diff --git a/src/lib/Server/Reports/utils.py b/src/lib/Server/Reports/utils.py
new file mode 100755
index 000000000..2ef21e446
--- /dev/null
+++ b/src/lib/Server/Reports/utils.py
@@ -0,0 +1,30 @@
+'''Helper functions for reports'''
+
+class BatchFetch(object):
+    '''Fetch Django objects in smaller batches to save memory'''
+
+    def __init__(self, obj, step=10000):
+        self.count = 0
+        self.block_count = 0
+        self.obj = obj
+        self.data = None
+        self.step = step
+        self.max = obj.count()
+
+    def __iter__(self):
+        return self
+
+    def next(self):
+        '''Return the next object from our array and fetch from the
+           database when needed'''
+        if self.block_count + self.count - self.step == self.max:
+            raise StopIteration
+        if self.block_count == 0 or self.count == self.step:
+            # Without list() this turns into LIMIT 1 OFFSET x queries
+            self.data = list(self.obj.all()[self.block_count: \
+                                   (self.block_count + self.step)])
+            self.block_count += self.step
+            self.count = 0
+        self.count += 1
+        return self.data[self.count - 1]
+
-- 
cgit v1.2.3-1-g7c22