Solaris User Signal 2 interruptions

Hi all,

A group of Solaris 8 servers that I support are experiencing some unusual errors. This occurs mostly in the middle of cron jobs. A User Signal 2 message is received, prematurely terminating the jobs. For example, an oracle export cron job killed recently:

. . exporting table METHODBODY 8171 rows exported
. . exporting table MGRCONF 603 rows exported
. . exporting table MGRCONF_20050922 56 rows exported
. . exporting table MGRSTART 14 rows exported
. . exporting table MO 1092873 rows exported
. . exporting table MOATT/export/home/oracle/dba/scripts/exportdb.sh[34]: 19937 User Signal 2
Error exporting westnx4 database.
Done with export script at Thu Aug 10 05:00:50 PDT 2006

We've tried running these scripts with "set -vx", hoping that it would reveal something, but there doesn't seem to be a specific command or error.

My initial thought was that this might have been an LD_LIBRARY_PATH issue, related to oracle. However, on one server recently I was sitting idle in a vi session, looking at oracle's .profile, when I received a User Signal 2 message and my vi session was terminated, exiting me to the command line.

Any thoughts on what these random User Signal 2 errors might be and what can be done to fix them would be greatly appreciated.

This is occurrring on multiple Solaris 8 servers, under two IDs, one with a default shell of /bin/sh, another with a default shell of /bin/csh.

$ uname -a
SunOS server2 5.8 Generic_117350-11 sun4u sparc SUNW,Netra-T12

$ uname -a
SunOS server3 5.8 Generic_117350-37 sun4u sparc SUNW,Netra-T12

$ uname -a
SunOS server4 5.8 Generic_117350-37 sun4u sparc SUNW,Netra-T12

Thanks,

Lou Avrami

--
http://www.freelists.org/webpage/oracle-l