Applies to:

Oracle Server - Enterprise Edition - Version 10.2.0.4 and later
Information in this document applies to any platform.
10.2.0.1 thru 10.2.0.5
Potentially 11.1
Fixed in 11.2

Symptoms

Any of the following errors are seen in the alert.log and/or trace files

Generic example from alert log:

Errors in file :

ORA-19883: Unused Block Optimization stopped for datafile
ORA-01115: IO error reading block from file (block # )
ORA-27091: unable to queue I/O
ORA-27069: attempt to do I/O beyond the range of the file
Additional information:
Additional information: 1
Additional information:

Also the trace file may show:

"Error: Bitmap for file at block of datafile
has incorrect block type, expected: 30, got:block type like 6>"

The errors can occur within any of several datafiles and tablespaces during the backups

e.g.
...
ORA-01115: IO error reading block from file +MT_DATA_01/MT/datafile/abd.451.677765651 (block # 2899969)
ORA-17510: Attempt to do i/o beyond file size

Another symptom is evidence of TAIL BITMAP block optimization after the backup

*** ACTION NAME:(0000145 STARTED16) 2010-12-30 23:09:54.192
*** MODULE NAME:(backup full datafile) 2010-12-30 23:09:54.192
*** SERVICE NAME:(abc_backup) 2010-12-30 23:09:54.192
...
...
Bitmap 2 for file xx at block xxxx of datafile ...
Rereading bitmap...
Backup Optimization context dump (krbodmpcx)

You can find several key indicators in the Trace file
Bold indicates key identifiers or characteristics of the problem

*** 2010-12-30 23:09:54.192
*** ACTION NAME:(0000145 STARTED16) 2010-12-30 23:09:54.192
*** MODULE NAME:(backup full datafile) 2010-12-30 23:09:54.192
*** SERVICE NAME:(abc_backup) 2010-12-30 23:09:54.192
*** SESSION ID:(939.28844) 2010-12-30 23:09:54.192
Bitmap 2 for file 179 at block 2891777 of datafile 179 has incorrect block type, expected: 30, got:6
Rereading bitmap 2 for file 179 relocated at block 2899969 in datafile 179
Backup Optimization context dump (krbodmpcx)

============================================
Information about file being optimized
Datafile name : +ABC_DATA_01/ABC/datafile/X2_idx_h1.451.677765651
Absolute file number : 179
Relative file number : 179
Tablespace number : 176
Tablespace blocksize : 16384
Tablespace is SFT : FALSE
Chkpoint SCN of datafile: 193955971853
Last block in backup : 2891776 <
Information about datafile holding bitmaps
Absolute file number : 179
Relative file number : 179
Chkpoint SCN bitmap file: 193955971853
Blocks mapped per bit : 4
Blocks mapped per byte : 32
Blocks mapped per bitmap: 516096
Bitmaps related information
Number of bitmaps used : 6
Number of head bitmaps : 2
Number of tail bitmaps : 4 <
Number of bitmaps read : 2
Space header related information
SCN of Space Header : 193956083782
Last deallocation scn : 170474335925
Last block in file : 2899972
Space header rel-fno(hc): 179
Size of file in blocks : 2899972
End of data blocks : 2899968 <969 reporting the ORA-1115 and ORA-17510
Max size of file : 3276804
Information about current bitmap
SCN bitmap : 172874328713
Current bitmap number : 2
Valid bitmap? : NO
First block in bitmap : 516101
Last block in bitmap : 1032196
First block in bytemap : 1032165
Statistics fields
Number of blocks read : 1030943
Number of unread blocks : 64
Status fields
All bitmaps processed? : NO
============================================

ORA-19883: Message 19883 not found; product=RDBMS; facility=ORA
; arguments: [+ABC_DATA_01.../X2_idx_h1.451.677765651]
ORA-01115: IO error reading block from file +ABC_DATA_01.../X2_idx_h1.451.677765651 (block # 2899969)
ORA-17510: Attempt to do i/o beyond file size

Cause

This problem is due to bug: 8354642

If there is an extension of a LMT datafile during a backup operation, you may find the bitmap used by LMT is incorrectly evaluated leading to

ORA-17510: Attempt to do i/o beyond file size error.

The error does not impact the backup: the problem is during the file size evaluation after the backup has completed and we attempt to determine the size of the datafile using incorrect information after the datafile extension and backup

Solution

FIX

The best is to go to 11.2 or higher
or if available apply a single-patch for 8354642

Workaround

One possible simple workaround is to pre-allocate extents in LMTs that may be likely to extend during a backup time window.

References

BUG:8354642 - INTERMITTENT ORA-01115/ORA-27091/ORA-27069 WHILE RUNNING RMAN BACKUPS
NOTE:1088018.1 - Master Note for Handling Oracle Database Corruption Issues