Depending on the size of the application (how much work involved in changing code):
Cftry/cfcatch on database errors would prevent users from noticing anything other than a slight delay on failover? Mvh Helge Hetland WebSite as Den 27. jan. 2011 kl. 03:35 skrev "Kris Jones" <[email protected]>: > > Hi Brian, > > I mistyped -- we are running 2k8 in 2k5 compatibility mode. > > I took the advice of my sysadmin who is also a certified MSSQL guy > (and a really smart one too). With Mirroring you can specify the > failover partner in the connection string. You have a primary, and a > secondary. When the primary fails, the secondary becomes the primary, > and when the failed server comes back online, it becomes secondary. > This is all transparent (as far as I know, and based on what I've been > told and read). I really don't know how a DB cluster would better suit > our needs. > > The failover is plenty fast actually -- it's just that I'd like to > have it so that CF doesn't have to return a connection failure to the > caller before failing over. Naturally it needs to see the failure, but > I'd hope there'd be something in the connection that would re-try once > if it were configured for mirror w/ failover partner? > > We're running a 12-instance CF Cluster, and it returns a failed > connection for each instance. Given a user Joe, and a not so busy > application, let's say that Joe logs on, and hits an instance. The > primary has failed, so Joe gets an error. Being a persistent guy, Joe > then reloads the page, where he hits another instance, and gets > another error. And so on, and so on. > > Luckily, we're talking more than 1 server, with LB w/ sticky session > in front of the cluster. So Joe will only get this a maximum of 4 > times. But this same scenario would have to happen on each of the > instances across the cluster. And I'd rather not have the error return > to Joe (or any user) in the first place. > > Cheers, > Kris > > > On Wed, Jan 26, 2011 at 6:32 PM, Brian Cain <[email protected]> wrote: >> >> Hi Kris, >> >> I don't really have a suggestion for your issue, but I was wondering what >> factored in to your decision to use mirroring instead of a DB cluster. I >> use a 2 node database cluster running MSSQL 2K5 and can drop the primary >> node and see my system recover (including running CF requests) in less than >> 30 seconds. I am not a database expert, but love the setup I have now. >> >> Thanks, >> Brian Cain > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Order the Adobe Coldfusion Anthology now! http://www.amazon.com/Adobe-Coldfusion-Anthology/dp/1430272155/?tag=houseoffusion Archive: http://www.houseoffusion.com/groups/cf-talk/message.cfm/messageid:341503 Subscription: http://www.houseoffusion.com/groups/cf-talk/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/groups/cf-talk/unsubscribe.cfm
