RE: Cron management...
Date: Mon, 13 Apr 2015 11:28:52 -0700
Message-ID: <1428949732.63591.YahooMailBasic_at_web184806.mail.gq1.yahoo.com>
> isn't managing NFS mounts correctly but periodically they go stale. When that
> happens, in our case the only way to fix the issue is to reboot the affected server
David,
I suspect you don't have intr option for the mount command. By default, the hard option is assumed and intr is not enabled. With intr, you can abort any command that would otherwise hang (ls, cd, etc.) But of course the problem on the NFS server side still has to be addressed.
> re: <time> <script> > /tmp/jobname.out 2>&1
>
> How about including the timestamp in the name of the log file; for example,
> /u01/joblogs/jobname.$(date +%y%m%d%H%M%S)
Iggy,
I thought about that. We don't append a timestamp to this log file name because the timestamp of the file itself serves the purpose. Well, the time you suggested is the beginning moment of running the job while the file timestamp is the last moment. Regardless, if the time info is important, whoever writes the script is responsible for recording that in the log, with either Linux `date' or sqlplus select sysdate from dual. Most of our jobs have their own logs in appropriate directories. Separate from those logs, the /tmp/jobname.out file is a mechanism to catch ANY REMAINING output. 99% of the time, this output, if not empty at all, is useless and very uninteresting. But when there's a problem, it may become valuable.
Yong Huang
-- http://www.freelists.org/webpage/oracle-lReceived on Mon Apr 13 2015 - 20:28:52 CEST