Block Corruption in empty pages in Oracle 11g
in [Oracle]
|
Prev: OCI protocol specification, where?
Next: 100% without investment online part time jobs..(adsense,datawork,neobux..more jobs)
From: zigzagdna on 4 Aug 2010 23:08 I am using Oracle 11.1.0.7.1 on HP UNIX 11i. On several occasions, I have been seen that block corruptions are reported by rman. I have never seen this problem in previous release of databases; but I have seen it more than once in Oracle 11g. However, when I run dbv command; I do not see any corruption in data and index blocks; so my conclusion is block corruption is in empty pages. Anyone has ideas why block corruption occurs in Oracle 11g. Has Oracles algorithm for checking block corruption have changed in 11g; I.e.; these corruptions existed in older Oracle releases but not reported by rman, Another strange thing I noticed: 1. When doing export backup, export backup reports about corrupted blocks. 2. When I do expdp no errors are reported. So question is why is exp command checking empty blocks? Any insights in above will be appreciated.
From: Mladen Gogala on 5 Aug 2010 07:54 On Wed, 04 Aug 2010 20:08:15 -0700, zigzagdna wrote: > Another strange thing I noticed: > 1. When doing export backup, export backup reports about corrupted > blocks. Export is not a backup. > 2. When I do expdp no errors are reported. So question is why is exp > command checking empty blocks? Because it's doing full table scans. Full table scan stops at HWM, not before. Also, exp utility is deprecated in 11.2. Do not use it. -- http://mgogala.byethost5.com
From: Mark D Powell on 5 Aug 2010 08:18 On Aug 4, 11:08 pm, zigzagdna <zigzag...(a)yahoo.com> wrote: > I am using Oracle 11.1.0.7.1 on HP UNIX 11i. > On several occasions, I have been seen that block corruptions are > reported by rman. I have never seen this problem in previous release > of databases; but I have seen it more than once in Oracle 11g. > However, when I run dbv command; I do not see any corruption in data > and index blocks; so my conclusion is block corruption is in empty > pages. > Anyone has ideas why block corruption occurs in Oracle 11g. Has > Oracles algorithm for checking block corruption have changed in 11g; > I.e.; these corruptions existed in older Oracle releases but not > reported by rman, > > Another strange thing I noticed: > 1. When doing export backup, export backup reports about corrupted > blocks. > 2. When I do expdp no errors are reported. > So question is why is exp command checking empty blocks? > > Any insights in above will be appreciated. If the rman backup and expdp are identifying corrupted blocks what does the following query show? set echo off -- -- Display corrupt block information -- -- 20100105 Mark D Powell Save 4 future use -- select * from v$database_block_corruption order by file# ,block# / You can find the object idenfified by this query by looking at dba_extents for the extent that the identified files and blocks are held within. HTH -- Mark D Powell --
From: John Hurley on 5 Aug 2010 08:32 zigzag: > I am using Oracle 11.1.0.7.1 on HP UNIX 11i. > On several occasions, I have been seen that block corruptions are > reported by rman. I have never seen this problem in previous release > of databases; but I have seen it more than once in Oracle 11g. > However, when I run dbv command; I do not see any corruption in data > and index blocks; so my conclusion is block corruption is in empty > pages. > Anyone has ideas why block corruption occurs in Oracle 11g. Has > Oracles algorithm for checking block corruption have changed in 11g; > I.e.; these corruptions existed in older Oracle releases but not > reported by rman, > > Another strange thing I noticed: > 1. When doing export backup, export backup reports about corrupted > blocks. > 2. When I do expdp no errors are reported. > So question is why is exp command checking empty blocks? > > Any insights in above will be appreciated. I have not seen anything similar in 11.1.0.7 ... My experience with any block corruption issues has always been storage related or from unexpected system failures ... in other words shaky hardware in one form or another. Any reason you are not patched up to 11.1.0.7.4 ? I have seen some hpux specific things in the bug lists but not looked closely since we do not run any databases on hpux any longer.
From: zigzagdna on 5 Aug 2010 12:59
On Aug 5, 8:18 am, Mark D Powell <Mark.Powe...(a)hp.com> wrote: > On Aug 4, 11:08 pm, zigzagdna <zigzag...(a)yahoo.com> wrote: > > > > > > > I am using Oracle 11.1.0.7.1 on HP UNIX 11i. > > On several occasions, I have been seen that block corruptions are > > reported by rman. I have never seen this problem in previous release > > of databases; but I have seen it more than once in Oracle 11g. > > However, when I run dbv command; I do not see any corruption in data > > and index blocks; so my conclusion is block corruption is in empty > > pages. > > Anyone has ideas why block corruption occurs in Oracle 11g. Has > > Oracles algorithm for checking block corruption have changed in 11g; > > I.e.; these corruptions existed in older Oracle releases but not > > reported by rman, > > > Another strange thing I noticed: > > 1. When doing export backup, export backup reports about corrupted > > blocks. > > 2. When I do expdp no errors are reported. > > So question is why is exp command checking empty blocks? > > > Any insights in above will be appreciated. > > If the rman backup and expdp are identifying corrupted blocks what > does the following query show? > > set echo off > -- > -- Display corrupt block information > -- > -- 20100105 Mark D Powell Save 4 future use > -- > select > * > from > v$database_block_corruption > order by > file# > ,block# > / > > You can find the object idenfified by this query by looking at > dba_extents for the extent that the identified files and blocks are > held within. > > HTH -- Mark D Powell --- Hide quoted text - > > - Show quoted text - Yes, tis query shows some block corruptions but it does not tell you corrupted blocks are empty. When I run dbv it does not show any corruption in data or indexes. So I am assuming that corruption is in empty pages. |