Posted by lightning-dave on 03/02/06 11:21

We have a custom logshipping solution set up between two of our SQL
Clusters - both running SQL 2000 Ent Edition.

The process is as follows:

1) A job runs every 15 mins to back up the transaction log on the live
server. After each transaction log backup is complete a flag is set in
a monitoring table to indicate that the log is ready to send.

2) A job runs every 2 mins and checks if there are any transaction log
backups ready to send and, if there are it sends them (using DOS copy)
to the DR cluster. Once the copy process has finished another flag is
set to indicate that the transaction log has been sent.

3) A job runs every 2 mins on the DR server and runs a RESTORE
HEADERONLY on the transaction log backup as a very basic check that the
copy has worked and to check the LSN etc. When this is done another
flag is set to indicate that the transaction log has been received.

4) A job runs at 8:00 and 18:00 to apply all the transaction logs to
the standby database (which is in standby mode and used for read-only
queries during the day). Prior to applying the logs all user
connections in the DB are killed.

The problem we're having is that allocation errors keep appearing on
the DR system. It's happened twice in the past two months. I've run
DBCC checks on the live system to ensure it's not inheriting these
errors from there and it isn't.

The upshot of this is that I have to refresh the DR database from live
which means restoring a 300Gb database from tape.

I would welcome some suggestions on why this is happening. There are no
errors in the logs whilst applying the logs (logs can still be applied
despite this error) and, the only time this is picked up is when I run
a DBCC CHECKDB on the standby database.

[Back to original message]