June 28, 2012 at 7:40 am
I was tempted with the first answer as it seemed correct. But a little bit of logic indicated me that there's no reason for the database backups to be incomplete so that it would be necessary to do a full backup. For the contrary, the log wouldn't be reliable after changing the recovery model from full to simple and back to full.
Nice question, I'm glad to have read all the answers before making a choice.
June 28, 2012 at 9:23 am
I actually just dealt with this in a slightly different way. We're doing a purge on a large DB and the app owner wanted a quick way to roll back and re-establish the log chain if anything went wrong. The decision was to take a snapshot, restore from that if there was an issue, and then do a diff. I actually wrote an article about it but haven't submitted it yet because it's still being reviewed by someone else before I send it in.
June 28, 2012 at 9:34 am
Thanks Steve.
June 28, 2012 at 9:46 am
Good question Steve.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
June 28, 2012 at 9:50 am
Definitely a gotcha!
June 28, 2012 at 10:06 am
I have been looking at log backups this week so I was glad for this question to confirm what I have been reading and my understanding.
However, there was no mentioning of the previous full backup in the question. Am I guessing that a similar error message would occur if attempting to perform a log backup if there was no full backup? A previous post only mentioned attempting a differential if there was no full backup.
June 28, 2012 at 10:18 am
Nice one. Thanks Steve.
June 28, 2012 at 10:25 am
Learned something new, thanks Steve
June 28, 2012 at 10:29 am
Drenlin (6/28/2012)
I have been looking at log backups this week so I was glad for this question to confirm what I have been reading and my understanding.However, there was no mentioning of the previous full backup in the question. Am I guessing that a similar error message would occur if attempting to perform a log backup if there was no full backup? A previous post only mentioned attempting a differential if there was no full backup.
If there's no full you won't be able to take a differential backup so that doesn't have to exist. I suppose it could be argued that since there was no statement that there was a full backup that a full may be required. However, in most people's environments it's a safe assumption (at least when it comes to answering a QotD) that a full backup exists.
June 28, 2012 at 2:36 pm
Glad you liked it, but I can't take credit for discovering this. Mr Randal, of SQL Skills, taught me this. Now I really want to get to one of his Immersion classes.
June 28, 2012 at 4:05 pm
Great question. If I wanted to start performing tlog backups on a database yesterday I would automatically run a full backup after setting to full recovery model. Today I know better.
June 28, 2012 at 5:59 pm
Normally I answer these without looking up BOL first but this time I thought I'd double check.
Unfortunately this myth comes straight from the horse's mouth:
"Before you can create the first log backup, you must create a full backup. "
June 28, 2012 at 6:16 pm
davoscollective (6/28/2012)
Normally I answer these without looking up BOL first but this time I thought I'd double check.Unfortunately this myth comes straight from the horse's mouth:
"Before you can create the first log backup, you must create a full backup. "
Took a bit to find this reference, but if I read it correctly they may be talking about when a database is first created. Based on the question, we could make a practical assumption that there was a full backup taken earlier.
I created a new database and both a t-log backup and a differential backup failed. I needed to take a full backup before either of those worked.
Restoring a full backup, i could take either after words.
June 28, 2012 at 6:38 pm
Thanks for the clarification Lynn, it makes more sense in that context.
July 4, 2012 at 10:55 am
Interesting question! Thank you!
But I was confused with the alternative:
"Take either a full OR differential backup and then a log backup"
Shouldn't this be:
"Take either a full AND differential backup and then a log backup"?
One question though: even if one could back up the log after only a differential backup, wouldn't that be useless without the full backup (the diff will not restore)?
Best regards,
Andre Guerreiro Neto
Database Analyst
http://www.softplan.com.br
MCITPx1/MCTSx2/MCSE/MCSA
Viewing 15 posts - 16 through 30 (of 45 total)
You must be logged in to reply to this topic. Login to reply