February 28, 2003 at 7:41 pm
One of the problems with the "high availability" model of clustering, as distinct from "fault tolerant," is that when a virtual server fails-over the SQL Server service rolls back the "in process" transactions when it restarts on a new node.
Does anybody know if there's any technical reason why a transaction couldn't be specified as "restartable" (by marking it as such in the log file before execution starts) meaning that, on failover, after the database has been recovered the "lost" transaction is restarted automatically?
I'm thinking of a "BEGIN RESTARTABLE TRANSACTION" type of syntax. If it is restarted it should re-execute as a "normal" transaction. (In case it was the cause of the failure.)
Any thoughts?
March 1, 2003 at 11:08 am
not that I am aware of. Any time we have transactions that need to be reissued from a failure ether the application does it or someone has to issue it manually. I'll do some checking and see what I can find.
Wes
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply