Hello!
Well, I'm encountering an Oracle internal error, and I can't find any topic about it in MOS.
Error is *ORA-00600: internal error code, arguments: [knlcfrevas:1], [], [], [], [], [], [], []* and is launched by a capture process.
The ORA-600 lookup tool in MOS gives only one note which applies to 10.2.0.4 on Solaris while I'm using 10.2.0.5 on Linux x64.
Well, before giving details, my real question is: I'd like to know if one can teach me how I could try to find the reason of this error (the trace file is pretty hard to understand...)
Context:
On my box (CentOS 6.2 x64), I have 2 databases which are target databases of 2 different source database. Capture and Apply process are both on the target databases.
The streams replication never worked on one of these database (while the other works perfectly). The capture process does everything until "CREATING LCR" and crash when it enter in the "ENQUEUING MESSAGE" phase.
I tried to drop the database and create it again from scratch and I finally get the same ORA-600.
I've checked tablespaces free space, owner of different queues, dba_capture table (which contains the same error), monitored the box memory and opened files, everything seems OK. Even I change these kind of paramaters the capture crash at the same point.
Alert log shows nothing but the ORA-600 above, and the trace file ends with:
OPIRIP: Uncaught error 447. Error stack:
ORA-00447: fatal error in background process
ORA-00600: internal error code, arguments: [knlcfrevas:1], [], [], [], [], [], [], []
In this file, I could see the line below. Does it means the IO pool size is full? (but I still have a lot of free file-descriptors on the linux box...)
ub8 ksfd_current_iopool_size_ [060009F98, 060009FA0) = 00000000 00000000
I don't know what kind of information could be interesting in this trace (which is about 5MB). So, is there somebody who could tell me where I should look to find more informations?
Thanks a lot!