EBU kills caller ??

From: <twod_at_not.valid>
Date: 18 May 1998 18:40:40 GMT
Message-ID: <6jpvb8$fna$1_at_vnetnews.value.net>



We have a problem with a script that is used to kick off the EBU cold backups.

The PERL5.004 script does the following:

shutdown the database
call ebu (with parameters/files for cold backup) startup the database.

When the script is called from cron the shutdown of the database is completed correctly and the ebu job is started. The ebu job then chugs away and then appears to be killed - although it's logfile shows that it completes correctly. The killing of the ebu job also kills the calling script such that the database is not restarted :(

When called from the command line all works as expected and the database is restarted.

The PERL script uses system() although we tried altering this to `` (backticks), but again the cron-run job failed.

The PERL script was then rewritten to use the plain Bourne shell and the exact same problem occurs.

To add further to this mystery we wrote a separate shell script that simply write 'ps -ef' output to a file for 30 mins whilst the ebu job runs. This script, which has it's own cron entry is also killed when the ebu job terminates !

Our vital statistics : EBU version 2.2.0.5; Solaris 2.6 (patched to latest revs) ; RDBMS 7.3.4; PERL5.004;

Has anyone seen this sort of thing before ?

IAP

--
I am using anti-spam measures, please replace 'not.valid' with 'value.net'
Received on Mon May 18 1998 - 20:40:40 CEST

Original text of this message