ORA-600 errors are thrown by Oracle when trying to drop cache while hugepage is in use.
Issue
- You are running an Oracle Database, single-instance or RAC. You have the SGA backed by hugepages.
-
Now a sysadmin comes along and executes a sysctl -w vm.drop_caches=1 (or 3).
-
This silently corrupts the database hugepages, and various ORA-600 errors are thrown by Oracle, including:
ORA-600 [kghlkrem1], or ORA-600 [kghfrunp: no duration]
- Oracle Internal errors have been experienced as follows:
Oct 14 2016 14:01: ORA-00600 reported in US SPOKE
Dec 10 2016 23:36: ORA-00600 reported in US SPOKE
Jan 31 2017 07:00:5: ORA-07445 (as per Oracle this points to memory corruption)
Jan 31 2017 07:54:57: ORA-00600 [KGHLKREM1] (as per Oracle this points to memory corruption)
- OGG REPLICAT was abended due to below error.
2017-01-31 08:26:59 INFO OGG-00996 Oracle GoldenGate Delivery for Oracle, rnmdsmf1.prm: REPLICAT RNMDSMF1 started.
2017-01-31 08:26:59 ERROR OGG-01082 Oracle GoldenGate Delivery for Oracle, rnmdsmf1.prm: The call to the DBOCI_check_for_virtual_column: OCI error (600-ORA-00600:
internal error code, arguments: [KGHLKREM1], [0x67C000020], [], [], [], [], [], [], [], [], [], []) determining if column is virtual. Table: MDSUSER.MDS_FIELD_FINAL_BUC
KET, Column DATASET_SID function from get_odbc_table_def failed with return code 0.
Environment
- Red Hat Enterprise Linux 5.4
- kernel-2.6.18-164.11.1 and previous versions
- oracle
- hugepage
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.