From: Ivan Kohler <ivan@freeside.biz>
Date: Sun, 6 May 2012 08:04:01 +0000 (-0700)
Subject: some ideas on speeding up the cust_main upgrade for big databases
X-Git-Url: http://git.freeside.biz/gitweb/?p=freeside.git;a=commitdiff_plain;h=4feff1c4a6a53f879d9c0dfeedfdd6a70bdf1401

some ideas on speeding up the cust_main upgrade for big databases
---

diff --git a/FS/FS/cust_main.pm b/FS/FS/cust_main.pm
index cba28bc9f..49b76cecd 100644
--- a/FS/FS/cust_main.pm
+++ b/FS/FS/cust_main.pm
@@ -5086,6 +5086,16 @@ sub process_censustract_update {
   return;
 }
 
+#starting to take quite a while for big dbs
+# - seq scan of h_cust_main (yuck), but not going to index paycvv, so
+# - seq scan of cust_main on signupdate... index signupdate?  will that help?
+# - seq scan of cust_main on paydate... index on substrings?  maybe set an
+#    upgrade journal flag now that we have that, yyyy-m-dd paydates are ancient
+# - seq scan of cust_main on payinfo.. certainly not going toi ndex that...
+#    upgrade journal again?  this is also an ancient problem
+# - otaker upgrade?  journal and call it good?  (double check to make sure
+#    we're not still setting otaker here)
+
 sub _upgrade_data { #class method
   my ($class, %opts) = @_;