Hi Dave, Thanks for the feedback, after some more searching last night, I found a thread that showed how to add an alternateServers=(server2:1433) to the connection string which would configure the datasource to automatically contact the second / alternate server is the first was down.
http://forums.datadirect.com/ddforums/message.jspa?messageID=3685 Adobe also has an article about this and about how those settings when inputed directly in the datasource config in cfadmin would be lost and need to be manually added to the neo-query.xml file http://www.macromedia.com/go/25b0de5e I was able to add this parameter and have been testing it but am not quite there yet. I don't actually even have two database servers set up yet..so.. I did manage to hack together this code to test the datasource configuration and see if the failover works. This is basically the same code from the article above combined with some insights gained from an adobe article: ColdFusion MX: Creating a pure Java connection to a data source http://www.adobe.com/cfusion/knowledgebase/index.cfm?id=tn_19607 <cftry> <!--- This cfobject creates a java object of type Class. Leave this statement as it is. ---> <cfobject type="JAVA" action="Create" name="Class" class="java.lang.Class"> <!--- The java class name for your driver is the argument for Class.forName ---> <cfset Class.forName("macromedia.jdbc.MacromediaDriver")> <!--- This cfobject creates a DriverManager java object which provides your connection ---> <cfobject type="JAVA" action="Create" name="DriverManager" class="java.sql.DriverManager"> <!--- The getConnection method takes the JDBC URL for your driver as its arguement ---> <cfset con = DriverManager.getConnection("jdbc:macromedia:sqlserver://myServerName:1433;d atabaseName=myDB;SelectMethod=direct;sendStringParametersAsUnicode=false;Max PooledStatements=1000;alternateServers=(failoverDBName:1433)", "username", "password")> <cfset DatabaseMetaData = con.getMetaData()> <cfoutput> #con.getWarnings()#<BR><BR> Driver Name: #DatabaseMetaData.getDriverName()#<BR> Driver Version: #DatabaseMetaData.getDriverVersion()#<BR> Database Name:#DatabaseMetaData.getDatabaseProductName()#<BR> Database Version: #DatabaseMetaData.getDatabaseProductVersion()#<BR> </cfoutput> <cfcatch> <h2>Error - info below</h2> <cfdump var="#cfcatch#"><cfabort> </cfcatch> </cftry> <cftry> <cfset con.close()> <cfcatch> <h2>Close Error - info below</h2> <cfdump var="#cfcatch#"><cfabort> </cfcatch> </cftry> I'm gonna keep trying.. Blists -----Original Message----- From: Dave Watts [mailto:[EMAIL PROTECTED] Sent: October 4, 2006 8:37 PM To: CF-Talk Subject: RE: SQl Server 2005 w/ Mirroring. How does CF know when to 'failo ver' to the mirror? > I am just doing some research into this. With SQL Server 2005 > and mirroring, the mirror will automatically take over if the > primary/principle server fails. But if you the datasource > references the principle SQL Servers name, and that server > was down, how would it failover? > > The only thing I can think of is that you may need to setup > two datasources for the principle and mirror and set up a > connection check within the CF application code that checks > the server and changes the #datasource# var used in the app > to the mirror if the principle is down. > > But this seems incorrect, because in which state would we > consider the princible down? And would those states match the > mirrored states? And what about client variable storage, > would that need the same type of thing? Any have a mirrored > SQL Server and coldfusion running together? How do you do it? SQL Server 2005's mirroring is somewhat like log shipping in SQL Server 2000. With that, you'd have a primary and a standby server, and transaction logs would be shipped periodically from the primary to the standby. If the primary failed, you'd have to manually reconfigure everything to point to the standby - you'd typically do this by changing the IP address of the standby to that of the primary, and restart things as necessary. You would lose whatever data had been added to the primary since the last log shipping operation. SQL Server 2005's mirroring includes replication functionality as well - in SQL Server 2000, replication and log shipping are completely separate, although many DBAs would use them together. Mirroring is also easier to implement than replication or log shipping in 2000, although I don't know (yet) whether it's more or less reliable and/or fragile. SQL Server 2005 has an additional feature which I've only read a bit about - the ability to have a monitor server that's used to automatically flip connections to the standby when the primary fails. I have no idea how that works exactly, because I haven't used it. Finally, SQL Server 2000 and 2005 both support clustering, which presents a single virtual IP to your database clients. This requires some pretty specific hardware support, and if that hardware fails, you get a big-time failure - cluster members essentially lose all network connectivity. Ouch! Dave Watts, CTO, Fig Leaf Software http://www.figleaf.com/ Fig Leaf Software provides the highest caliber vendor-authorized instruction at our training centers in Washington DC, Atlanta, Chicago, Baltimore, Northern Virginia, or on-site at your location. Visit http://training.figleaf.com/ for more information! ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Introducing the Fusion Authority Quarterly Update. 80 pages of hard-hitting, up-to-date ColdFusion information by your peers, delivered to your door four times a year. http://www.fusionauthority.com/quarterly Archive: http://www.houseoffusion.com/groups/CF-Talk/message.cfm/messageid:255702 Subscription: http://www.houseoffusion.com/groups/CF-Talk/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=89.70.4
