Kbase P103881: First occurrence of 1124 while reading the database from a single user session
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
16/10/2008 |
|
Status: Unverified
SYMPTOM(s):
probkup <database-name> <backup-file>
Physical Redo Phase Completed at blk off upd . (7161)
0 bi blocks will be dumped. (6688)
Begin backup of Data file(s). (5461)
SYSTEM ERROR: read wrong dbkey at offset <offset> in file <file> found <dbkey>, expected <dbkey>, retrying. (9445)
Corrupt block detected when reading from database. (4229)
<func-name>: Error occurred in area <num>, block number: <num>, extent<name>: . (10560)
SYSTEM ERROR: wrong dbkey in block. Found <dbkey>, should be <dbkey2> (1124)
CAUSE:
A corrupted block that is detected for the first time on a read, with "probkup" or "dbanalys" for example, is a clear sign that the corruption has been generated by a problem on the machine and not a Progress bug.
We can never dismiss the possibility of an unknown problem with Progress code, but it seems very unlikely that the write of a corrupted block has not been detected by Progress; we do verify the block before the write command. Here the corrupted block was detected when we read it.
Based on customers' experience we know that the machine problem can be at different levels. It can be the disk, the controller, the OS cache or the memory to change, even if no error was seen by the OS tools. Recently some customers had to disable the "write-back" to eliminate the corruption.
FIX:
1. Fix the hardware problem.
2. Restore an earlier valid database backup