【转】应用程序的脚本Segmentation fault 导致的Oracle数据库实例 crash

转自：
An Oracle DB instance running on Oracle Linux 6 crashed with the following error: abrt[xxx]: Saved core dump of pid zzzz to core.zzzz at /x/home/oracle (yyyyy bytes) (文档 ID 2524220.1)

 

APPLIES TO:

Linux OS - Version Oracle Linux 6.7 and later
Linux x86-64

SYMPTOMS

The customer's Oracle DB instance crashed when running on an Oracle Linux 6.7 server, and generated the below error message:

Dec 18 13:39:00 <host>abrt[71604]: Saved core dump of pid 97591 to core.97591 at /x/home/oracle (31709372416 bytes)

CAUSE

The DB instance crashed due to an application process segfaulting and was not due to an issue with the operating system itself.
 
The core file back trace from gdb shows it was an application, not an OS process that segfaulted:

# gdb /x/home/oracle/core.97591

<snip>
[New Thread 67514]
[New Thread 97591]
[New Thread 67313]
Core was generated by `/x/home/oracle/product/11.2.0.4/perl/bin/perl /x/home/oracle/admin/scripts/hist'. Program terminated with signal 11, Segmentation fault. <----- application provided version of perl ran an application script which segfaulted
#0 0x00000032ad40f65b in ?? ()
"/x/home/oracle/core.97591" is a core file.
Please specify an executable to debug.
(gdb) bt
#0 0x00000032ad40f65b in ?? ()
#1 0x00007f070b13b4da in ?? ()
#2 0x0000000000000000 in ?? ()
(gdb)

 In this case, a perl script was run which caused the segfault, and thus the DB instance to crash, however it was an application provided version of perl (as seen from the path), rather than the OS supplied version of perl (which is /usr/bin/perl )

SOLUTION

Please check the script and application and any application logs for any issues that could have caused the segfault to occur.