September 23, 2012 at 5:57 pm
Comments posted to this topic are about the item The Importance of Validating Backups
Paul Randal
CEO, SQLskills.com: Check out SQLskills online training!
Blog:www.SQLskills.com/blogs/paul Twitter: @PaulRandal
SQL MVP, Microsoft RD, Contributing Editor of TechNet Magazine
Author of DBCC CHECKDB/repair (and other Storage Engine) code of SQL Server 2005
September 24, 2012 at 2:40 am
Thanks Paul for writing such a nice article & clarifying the importance of "WITH CHECKSUM" option when taking the backups! 🙂
September 24, 2012 at 3:21 am
Great stuff as usual by the SQL legend 🙂
M&M
September 24, 2012 at 3:32 am
.
September 24, 2012 at 3:57 am
Paul Brewer (9/24/2012)
cannot cope with databases with more than 1 data file at the moment.
Hi Paul,
Been working on a similar script myself - I use RESTORE FILELISTONLY to get the data files & also to generate the MOVE statements.
Happy to send you over some code, although it'll probably need modifying to fit your process.
My starting point is the backup files themselves, rather than msdb database.
Cheers
Gaz
September 24, 2012 at 3:59 am
Well you know I loved this article. I've been ranting about this issue for quite some time. It's nice to have yet another resource to point to in order to try to get the point across. Thanks Paul.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
September 24, 2012 at 4:10 am
Yes, thanks for the article Paul.
Is there actually an easy way to check the output from DBCC CHECKDB?
I currently scan the SQL log using xp_readerrorlog for the DBCC output info, another way is to use TABLERESULTS clause, and check for the completion line.
However, both methods could be called undocumented - you won't find either in BOL!
September 24, 2012 at 4:32 am
Great article.
Is it possible to use DBCC CHECKDB to check the database first before backing up (with all your other recommendations), so that we don't have to restore the backup to perform that test?
September 24, 2012 at 6:13 am
I recently talked my boss into getting a drive big enough to restore all of our DBs so we can pull them from tape and do tests on a few each week. Now just comes the job of automating that which is likely to be very fun. It would also appear that we need to do some work to secure our backups a little better.
charles.wong (9/24/2012)
Great article.Is it possible to use DBCC CHECKDB to check the database first before backing up (with all your other recommendations), so that we don't have to restore the backup to perform that test?
You can run CHECKDB first but there's no guarantee that the DB doesn't get corrupt before the backup is done. It also won't guarantee that the backup file itself doesn't get corrupted. Some people recommend doing an integrity check and if the DB is corrupt not to back it up (assuming that this means you may be deleting the last good backup.) Our process here is to backup the database since we send the backups to another location so we shouldn't be destroying a good backup. It also means that we have a copy of the corrupt DB so if we need to restore we can then do some comparisons to try and recover some of the changes. The thing you need to be careful of is making sure you're not destroying the last good backup because then you have a problem.
September 24, 2012 at 6:29 am
Glad you all like the article!
Some of the questions:
1) If you can run CHECKDB on the prod system, go for it. But also do the backup/copy/restore/CHECKDB process too. Defense in depth - just like with security.
2) Did CHECKDB complete and find errors? Check the value of @@ERROR afterwards - guarantees to be non-zero if CHECKDB found/had a problem.
Cheers
Paul Randal
CEO, SQLskills.com: Check out SQLskills online training!
Blog:www.SQLskills.com/blogs/paul Twitter: @PaulRandal
SQL MVP, Microsoft RD, Contributing Editor of TechNet Magazine
Author of DBCC CHECKDB/repair (and other Storage Engine) code of SQL Server 2005
September 24, 2012 at 7:40 am
Paul,
I have a few automated backups running in various places.
They typically are started via a .bat file launching sqlcmd.exe with the -i parameter pointing to the .sql file issuing the backup command.
The occassional restore I had to do never presented a problem. But I prefere to be on the safe side.
Once I have put the WITH CHECKSUM into the .sql file, how is an error signaled?
What sort of message is generated? If so, I suppose it's in the sqlcmd.exe output which I would have to pipe into a log.
Does sqlcmd.exe return a DOS error code of some sort in this case?
Thanks.
September 24, 2012 at 8:19 am
Restore verify only with checksum can only be used when the backup file which queried with checksum. If we do not do backup file with checksum, we cannot use the command Restore verify only with checksum
September 24, 2012 at 8:51 am
.
September 24, 2012 at 9:49 am
deleted
September 24, 2012 at 10:02 am
Thanks for the great article, Paul. I was one of those "people around the world asking for help with corrupt databases, where their recent backups also contained the corruption" whom you mercifully helped years ago!
QUESTION: The database maintenance plan dialogs in Microsoft SQLServer2000 (yes, we're still running it) has a "Verify the integrity of the backup upon completion" option. Is this the same as the "WITH CHECKSUM" option?
Viewing 15 posts - 1 through 15 (of 65 total)
You must be logged in to reply to this topic. Login to reply