June 21, 2005 at 10:03 am
SQL 2000 SP3a (8.00.818) - Win2000
We log ship several databases from one server to another. This process has been running for approx 2 yrs.
With the last few reboots, there is one standby database which keeps coming up as "suspect" after the reboot.
T-log backups are every minute. Copy happens also every minute (approx 15 seconds after the backup). Restores happen every 15 minutes.
Has anyone experienced this? What's the best way to tell the cause? If SQL tries to shut down in the middle of a t-log backup or a restore, I have assumed that the backups will be either, allowed to finish, or terminated gracefully. But I'm starting to think that has not been a good assumption.
Thanks very much for your time!
June 21, 2005 at 12:29 pm
A shutdown on the production (source) system should not affect the status of the destination (target) server. The only thing that should cause the target server to come up as suspect is a killed restore process. Could this be the culprit?
June 21, 2005 at 1:46 pm
We do not explicitly kill the restore process. I am wondering if: when SQL Server shuts down, if it kills a restore process. I would be surprised if it was that abrasive. But I do not know for sure.
June 22, 2005 at 6:21 am
Hi,
just curious, are you performing custom logshipping, or is it the out-of-the-box procedure?
Are you restoring the failing database from a network share or SAN, or from a local disk?
Can you tell if the failing db is being restored at the moment of the reboot?
Do you shut down the server with the option "force"?
karl
Best regards
karl
June 22, 2005 at 7:43 am
Hi Karl,
We're using SQL Server log-shipping "out of the box," with maintenance plans and the log-shipping monitor.
The transaction logs reside on the local drive at the time they are restored. The log shipping process copies the file to the destination server.
We do not use a "force" option at shutdown. We use this command in a batch file: 'net stop "MSSQLServer"'
I cannot tell yet if it's actually trying to perform restore at the time of shutdown. That may be the $64K question. I do know now that the database is out of sync by approx 1 hr at the time of shutdown. So it's restoring multiple logs per minute (when it's able) so that it will catch up and get back in sync. That increases the likelihood that it's trying to restore at the time of shutdown.
Thanks for your interest.
Ben
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply