Bug-ul de mai jos apare ca rezolvat in versiunea 10.2.0.4.2 de Oracle RDBMS ( database - Patch 8833280 ) . Versiuni afectate pana acum : 10.2.0.3 si 10.2.0.4 .
Pentru windows, bug-ul pare a fi fixat in 10.2.0.4 Patch bundle 25 .
Poate introduce corupere la nivel de redo log, facand astfel imposibil recover-ul folosindu-se rman-ul. Mai mult decat atat, in nota de pe metalink, se specifica clar ca respectivul patch nu rezolva eventualele coruperi deja existente.
De pe metalink.oracle.com (nota 861965.1) :
Bug 7662491 - Array Update can corrupt a row. Errors OERI[kghstack_free1] or OERI[kddummy_blkchk][6110] [Note ID 7662491.8] .
Description : With compatibilty set to 10.2 or higher, a row can be corrupted if it is updated with an Array Update (eg: with plsql FORALL ... UPDATE or an UPDATE of several rows) .
If db_block_checking is set to MEDIUM or FULL we can prevent the corruption going to disk.
Media Recovery like RMAN Blockrecover or "RECOVER DATAFILE" don't fix the corruption as it is already introduced in the Redo/Archived Logs.
Workaround-uri: restore pana in momentul aparitiei coruperii sau incercarea de recuperare a datelor din tabela afectata, prin dbms_repair .
Niciun comentariu:
Trimiteți un comentariu