September 11, 2006 at 10:00 am
I am using "high safety with automatic failover" database mirroring. This includes the witness principal and mirror. The mirroring has been running without problems for several weeks. We have tested the mirror by manually failing back and forth betweeen nodes, so we know that the mirror has worked before as advertised.
This past weekend, one of our databases automatically failed over to its partner, and then after a while went into a "DBName (principal, disconnected)" state. Its partner (the original principal) is in the following state: DBName (Principal, Disconnected / In Recovery), so it appears as if it is the "warm standby" but the "disconnected" indiciates that no transactions are being sent to it from its partner, which is now the real principal (it has all of the connections). so to re-cap:
SQL-1: DBName (Principal, Disconnected / In Recovery)
SQL-2: DBName (Principal, Disconnected)
SQL-2 is currently the real principal.
So, how do we re-synchronize the database on SQL-1 with its partner without rebuilding the mirror (restore w/no recovery, etc)?
September 13, 2006 at 5:03 pm
Charcoll,
Unless the connection to the witness is broken, the two messages seem to carry a contradiction: Both databases cannot be attempting to be Principal. If you refresh the view, are you getting the same message?
We have also been testing mirroring for several weeks and so far the results have been predictable and stable. We have Alerts and Jobs that force the associated databases to follow whenever one of the group fails-over, and to repair the logins so that the applications can re-connect to the backup server.
The descriptions we get are "Principal, Disconnected" or "Principal Synchronized with Witness".
Elliott
January 29, 2010 at 10:07 am
Elliott Berkihiser (9/13/2006)
Charcoll,
Unless the connection to the witness is broken, the two messages seem to carry a contradiction: Both databases cannot be attempting to be Principal. If you refresh the view, are you getting the same message?...
Logically it should not happen but we have had it happen from time to tome here. Unfortunately we've not had a good plan of attack and I have not found a good reference as to what is the preferred method of dealing with that.
...
-- FORTRAN manual for Xerox Computers --
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply