Hi,
I had posted a problem few days ago (copy below), but have not found any convincing reply so far. An important point mentioned is "if I run the same script on a non-cluster host DS10 m/c with same Unix, it works fine". Hence, the script is OK and it happens only in a cluster environment. Any more suggestions please?
P.S: Unix versions on all m/cs are v5.1 with patchkit 2.
Harihar K
*************************************************************************
Have a problem with respect to tar on a TruCluster5.1.
We tried backup to tape using tar commond on our 2-node DS20 Cluster running Tru64 & TruCluster v5.1. Herein our script creates a list of files to be backed up (these files cover 8-10 mounted RAID file systems holding Oracle 8.1.7 database files - Oracle Parallel Server being used) using the following:
find /u01 -print > /u111/logfiles/bkup.lst
find /u02 -print >> /u111/logfiles/bkup.lst
...
find /u06 -print >> /u111/logfiles/bkup.lst
find /u07 -print >> /u111/logfiles/bkup.lst
However, when we run this script with "tar cv </u111/logfiles/bkup.lst" to the default tape drive, it backs up files in /u01, and then goes back to backing up /u01 again (though the there is nothing wrong with the list of files in "bkup.lst")! This happens a few times before the backup process moves to /u02. As a result even before reaching /u03, it reaches the end of tape. It may be noted that the cumulative files size in all these directories together do not exceed half the size of the tape. It may also be noted that if I run the same script on a non-cluster host DS10 m/c with same Unix, it works fine. Hence, the problem seems to be with the Cluster set up only. Why could this happen?
NOTE: Also, if we use "cpio -ocvB" instead of tar in the same Cluster, backup is ok.
*************************************************************************
--------------------------------------------------------------------
Mail2Web - Check your email from the web at
http://www.mail2web.com/ .
Received on Fri Jun 22 2001 - 06:06:28 NZST