| Author |
Topic |
|
sherrys
Starting Member
37 Posts |
 Posted - 2003-09-30 : 09:55:57
|
| We run SQL 2000 on Datacenter 2000. We have a 2-node cluster in active-passive mode. We are hooked up to our SAN through 1GB fiber channel. We are running 3 instances of SQL..We use backup devices formatted as <database>_A and <database>_B (Logs are <database>_A_LOGS, differentials are <database>_A_DIFF, with appropriate B sets), as such:select @command = 'backup database ' + @db + ' to ' + @db + '_' + @dayweek + ' with init'The devices point to a file share on the SAN that is backed up to tape every night.Here is the problem: without rhyme or reason our backup files seem to get corrupted. They pass the ‘restore verify only’ step of the job, but the next cycle we cannot over write the file and receive the following error:The backup data in 'wslogdb50_B' is incorrectly formatted This problem randomly happens on all three instances, in various databases.Is this a known problem? Any advice would be appreciated. |
|
|
tkizer
Almighty SQL Goddess
38200 Posts |
Posted - 2003-09-30 : 12:43:03
|
| Are you running the latest service pack (sp3a) on each of the instances? We've got a very similar environment here and one of the things that we don't do is use backup devices.Tara |
 |
|
|
sherrys
Starting Member
37 Posts |
Posted - 2003-09-30 : 13:28:43
|
| Yes, we have SP3a on all three instances. We don’t have a choice on using devices, that came from upper management and their architect. If we have to stop using devices, do you have any justification that I can give to management? Like would that solve our problem? Thanks |
|
|
|
tkizer
Almighty SQL Goddess
38200 Posts |
Posted - 2003-09-30 : 13:36:13
|
| I don't have any justification except that there is no need to use backup devices anymore. It was necessary in SQL Server 6.5. Why bother using something that isn't needed? Anyway, found this article on MS' knowledge base:[url]http://support.microsoft.com/default.aspx?scid=kb;en-us;290787[/url]Tara |
|
|
|
ejcorcoran
Starting Member
10 Posts |
Posted - 2003-09-30 : 14:52:04
|
| From Ed -- coworker of sherrysI think our original question is getting missed. What we are wondering is the "why" of the corruption. I also create some BCP files on this share and their associated format files are located there. Should I be worried about impending disk doom? We have run checks on the physical disks in the array and they all seem to pass.I am not at the point of loosing sleep over this, but it's getting close."Fixing" the error message with FORMAT or NOSKIP is one thing, but I feel rather shaky having backup sets out there that may or may not be bad. By time I can "prove" them, it's too late. The pure randomness, and the frequency, of it is starting to worry me at a professional, resume writing, level.Thank you for any help you might be able to give us-- Ed |
|
|
|
tkizer
Almighty SQL Goddess
38200 Posts |
Posted - 2003-09-30 : 14:57:10
|
| Did you read the article? (SQL Team URLs do not show semi colons, copy the URL into a browser then put semi colons where you see the two spaces.)The article gives the CAUSE.Tara |
|
|
|
ejcorcoran
Starting Member
10 Posts |
Posted - 2003-09-30 : 16:30:11
|
| Exactly, so does this mean that I should suspect my disk subsystem(media in this case would be the SAN)? As I said I have checked the disks, they are find. Not going across the network, going to fiber so no prob there. Would simulatneous writes to this disk from another job (Say another back up or creating a BCP) recreate the network type problem described by the article. My I/O seems fine so I discounted it out of hand. Maybe a bit cavalier on my part. Could two simulatnaeous writes to different files on the same array cause this error? Would my I/O counts show this?Thanks-- Ed |
|
|
|
tkizer
Almighty SQL Goddess
38200 Posts |
Posted - 2003-09-30 : 17:02:25
|
| I am not sure what would recreate the problem. Not sure about the two simultaneous writes, but I would doubt it. But the only way to find out is to test it or ask the SAN vendor about any potential problems.Tara |
|
|
|
ejcorcoran
Starting Member
10 Posts |
Posted - 2003-09-30 : 17:03:29
|
| Thanks, I'll go beat up on the vendor now. :)-- Ed |
|
|
|
|
corrupt backup file