April 7th, 2008

ORA-00600 [kgeade_is_0]

Have a look at this nice one (alert log), seen in 10gR2 (10.2.0.4 64bit on Linux):

ORA-00600: Interner Fehlercode, Argumente: [kgeade_is_0], [], [], [], [], [], [], []

And from the trace file:

Current SQL statement for this session:
select tablespace_id, rfno, allocated_space, file_size, file_maxsize, changescn_base, changescn_wrap, flag
from GV$FILESPACE_USAGE
where inst_id != :inst and (changescn_wrap >= :w or (changescn_wrap = :w and changescn_base >= :b))
<stack traces en masse>
kxfpgsg
        Error 600 encountered
error 600 detected in background process
OPIRIP: Uncaught error 447. Error stack:
ORA-00447: fatal error in background process
ORA-00600: Interner Fehlercode, Argumente: [kgeade_is_0], [], [], [], [], [], [], []

Context: One RAC node (of two) worked well, the other one did as described immediately after restarting the instance (I guess during the instance recovery process).

Neither Metalink nor Google seems to know this one. We will see.

Regards
Usn


By usn in Oracle 



You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

3 Responses to “ORA-00600 [kgeade_is_0]”

  1. usn Says:
    April 7th, 2008 at 3:33 pm

    Okay, found out what happened:

    The RAC was in reconfiguration state (node fence due to heavy load), and at synchronous reboots of the two nodes the network stack for the bonded interconnect was initialized too late on Node1.

    Somehow, the instance started, but immediately after that the CRS decided to fence the local node (1) to avoid split brain. But starting of the instance was already in progress, and some background processes died due to a lack of interconnect I guess. That might be where the ORA-447 comes from.

    Weird, but if you need details send me an email.
    Usn

  2. Curtis R. Says:
    April 13th, 2008 at 4:32 am

    Running into this on 3 out of 4 nodes in a RAC. Any more details?

  3. usn Says:
    April 13th, 2008 at 1:36 pm

    Can you cold-out the RAC for some minutes and restart CRS, and make sure that all DB processes live that way? My interpretation of that is the partial death of some processes, the error being just a secondary effect.
    Usn

Leave a Reply