February 28, 2010
The last couple of weeks I’ve been working on converting on of my Databases to ASM. As databases grow adding more disk breaks performance if file systems are being used at many levels and to control performance Oracle released ASM which would provide RAW disk performance with dynamic re-balancing. This was precisely one of my use cases. So After performing my due diligence (dev testing) and so forth. I spent the last 3 weeks slowly converting the database to ASM. There were 2 reasons for the slow conversion. one i didnt have the diskto be able to convert it all at once and the second was gettin enough downtime to perform the conversion was a challenge.
Everything went really well till week 3 when the conversion completed . I asked my Sysadmins to give me the last chunk of disk to add to my disk group and executed a power 1 re balance. All went well for 5 days until i got a call from our support informing me of ORA-600[kffmUnlock_3] followed by an instance crash. Instance crash is always something to be worried about but oracle over the years is resilient enough for such issues. The database came up only to crash again in 9 minutes. Why 9 minutes because that was the total time that the re balance operation had left.
Oracle does have a note here that explains the whole situation . There is a apatch but i opted to shutdown the database and allowed te rebal;ance to finish. Since this was the last set of disks for now I’m guessing i will not hit this issues and plan to test the patch and have it in production before then. TIll then i hope all is well for this database.
