Hey,
I ran accross an interesting issue. It seems as though Radiator never
checks for the timeout of the connection to the database before trying to
execute a delete query. So, what I did, is simply implemented the
ping() check in the reconnect routine. Following is the patch to
SqlDb.pm.
--- /usr/src/Radiator-2.16.1/Radius/SqlDb.pm Mon Jun 12 18:47:33 2000
+++ SqlDb.pm Wed Jun 21 23:46:31 2000
@@ -98,7 +98,7 @@
# Implement backoff strategy in case of database failure
return 0 if time < $self->{backoff_until};
- if (!defined $self->{dbh})
+ if (!defined $self->{dbh} || !$self->{dbh}->ping())
{
# A new connection is required, try all the
- # ones in the $self->{DBSource} in order til we
+ # ones in the $self->{DBSource} in order till we
That did it for me. However, it does rely on a ping() DBI call which has
to be implemented by the underlying DBD driver, and not all of them
do. MySQL seems to have that, not sure about the others.
-Garry
===
Archive at http://www.starport.net/~radiator/
Announcements on [EMAIL PROTECTED]
To unsubscribe, email '[EMAIL PROTECTED]' with
'unsubscribe radiator' in the body of the message.
