Linux Forum Schweiz
Alles rund um Linux

Startseite » Linux Distributionen » Redhat Linux » CentOS7 mit ORACLE 12c (Script Fehler )
CentOS7 mit ORACLE 12c [Beitrag #2816] Do, 09 November 2017 10:10
Slowlyer ist gerade offline  Slowlyer
Beiträge: 1
Registriert: November 2017
Ort: Zürich
Junior Member
Guete Morge mitenand
ich bin einem Problem auf der Spur und suche Rat bei euch.

Es geht um folgendes:

Betriebssystem: CentOS7 Minimal mit Oracle 12 c Standard Edition 2

Anwendung: rman backup manager

rmanbackup Ordner auf oracle:oinstall 775 Rechte
backup_rman.log datei auf oracle:oinstall 775 rechte

Logge mich als Orcale user ein setze richtige SID.
Führe folgendes Script manuell aus:

#! /bin/bash
#
# /opt/oracle/bin/backup_rman.sh
#
# 2014-Nov-06 ada    changed the way in which it checks for Primary or Standby
# 2013-Jun-06 vl     removed extra memory for rman again
# 2013-Jun-06 pmo    added support for systems running in german language
# 2013-Apr-25 vl     more memory for rman; adjust default bakwin to 7 days to be in sync with default retention policy for controlfile
# 2013-Apr-10 vl     remove filesperset
# 2013-Apr-09 vl     backup retention policy only for primary server
# 2013-Apr-05 vl     remove redundant commands, incremental backup during the week, remove archwin, change bakwin rule
# 2013-Feb-01 ada    added NLS_DATE_FORMAT
# 2012-Sep-04 ada    added "as compressed backupset" from #2278693
# 2012-Aug-28 pmo        changed Mail to mailx
# 2012-Jul-04 pmo        'alter system archive log current' now dependent on database_role
# 2011-Oct-14 dk     changed nomenklatura for backup name. DBID included now
# 2011-Sep-12 vl     add support for red hat Enterprise Linux Server. source .bash_profile
# 2009-Jul-03 pe     bugfix: SLES sets an Environment that avoids a successful rman login, use .profile instead
# 2009-Jan-08 ahu    changed command order, first delete obsoletes and then backup
# 2008-Nov-11 pe     bugfix: SLES 9 sets an Environment that avoids a successful rman login, use .profile instead
# 2008-Oct-27 pe     on Solaris source .profile if available
# 2008-Jun-12 aka    logfile includes history information now (append-write)
# 2008-Feb-26 vl/aka archivelog deletion changed
# 2008-Feb-25 aka    rewritten for full backup -
#                    backup + database id is enough to reconstruct all necessary files
# typical cron:      40 5 * * * test -x /opt/oracle/bin/backup_rman.sh && /opt/oracle/bin/backup_rman.sh
#
# use
#
#   find ~corpus/css/app ~corpus/cscs/app -name "*.xml" | xargs grep -P '<cron\s+pattern='
#
# to checkout cronjobs within censhare. grep option -h suppresses filename output
#
# script parameters - change at installation time
set -x
bakdir=/rmanbackup
bakwin=7                        # backup window in days - must be greater than 7 because only one full backup a week
# keeptime in oracle should be more than bakwin. Check with: "show parameter keep_time;"
# and change if needed: "alter system set control_file_record_keep_time=14 scope=both;"
logfil=/var/log/backup_rman.log
esetup=/etc/profile.d/oracle.sh # Solaris: ~oracle/.profile should contain all variable definitions
mailad=root
dbid=$(echo -e "select dbid from v\$database;" | sqlplus / as sysdba 2>/dev/null | grep [0-9][0-9][0-9][0-9][0-9][0-9] | sed -e 's/ //g')

Error () { echo >&2 "$*"; exit 1; }

# reread environment if necessary (ORACLE_SID) and available
test -z "$ORACLE_SID" -a -f "$esetup" && source "$esetup"
test $(uname) = Linux -a -f /etc/SuSE-release && grep -i "SUSE LINUX Enterprise Server" /etc/SuSE-release >/dev/null && source ~oracle/.profile
test $(uname) = Linux -a -f /etc/redhat-release && grep -i "Red Hat Enterprise Linux Server" /etc/redhat-release >/dev/null && source ~oracle/.bash_profile

# check environment
test $(uname) = SunOS && { test -e ~oracle/.profile && source ~oracle/.profile; }
test -z "$ORACLE_SID"  && Error "\$ORACLE_SID required."
test -z "$ORACLE_HOME" && Error "\$ORACLE_HOME required."
test ! -d "$ORACLE_HOME" && Error "dir ORACLE_HOME='$ORACLE_HOME' not found."
test ! -d "$bakdir"      && Error "dir bakdir='$bakdir' not found."
touch "$logfil" || Error "log '$logfil' not writeable."
rman=$ORACLE_HOME/bin/rman
test ! -x "$rman" && Error "rman program '$rman' not found or not executable."
test "$bakwin" -ge 7 >&/dev/null || Error "bakwin='$bakwin' must greater-equal 7."

# increase memory for rman
# export ORA_RMAN_SGA_SIZE=33554432

# exec rman with logging
exec >>$logfil 2>&1
echo
echo "-----------------------------------------------------------"
echo "$(date +"%Y-%m-%d_%H:%M:%S") starting rman backup..."
# define backup methods depending on day of week
if test "$(date +%a)" = Sat -o "$(date +%a)" = Sa; then
        BACKUPMETHOD="backup as compressed backupset incremental level 0 database plus archivelog;"
else
        BACKUPMETHOD="backup as compressed backupset incremental level 1 database plus archivelog;"
fi
# check db role because BAKWIN can only defined on the primary
DATABASE_ROLE=$(echo -e "whenever sqlerror exit sql.sqlcode;\nconnect / as sysdba;\nselect database_role, open_mode from v\$database;\nquit;" | sqlplus /nolog || Error "cannot determine database_role, open_mode from v\$database;")
echo $DATABASE_ROLE | grep "PHYSICAL STANDBY MOUNTED" > /dev/null &&
BACKUPWIN="" ||
BACKUPWIN="configure retention policy to recovery window of $bakwin days;"

export NLS_DATE_FORMAT='YYYY-MM-DD HH24:MI:SS'
set -x
$rman target / <<-EOF
        configure channel device type disk format '$bakdir/$ORACLE_SID-$dbid-ora-df%t_s%s_s%p';
        configure controlfile autobackup format for device type disk to '$bakdir/$ORACLE_SID%F';
                configure backup optimization on;
        $BACKUPWIN
        run {
            delete noprompt obsolete;
        }
        run {
                        $BACKUPMETHOD
                }
        quit;
EOF
ec=$?
type Mail >&/dev/null && Mail=Mail || Mail=/usr/bin/mailx
test $ec != 0 && echo "see log '$logfil'" | \
    $Mail -s "$HOSTNAME: error '$ec' on rman backup" $mailad
echo "$(date +"%Y-%m-%d_%H:%M:%S") rman backup finished."
~


funktioniert tadellos.

jetzt möchte ich das ganze als cronjob laufen lassen unter User Oracle wie folgt:

(((SID ist manuell festgelegt)))

55 2 * * * /u01/app/oracle/product/12.2.0/dbhome_1/backup_rman.sh


jedoch taucht folgendes Problem auf wenn ich das über einen Cronjob laufen lasse:

Subject: Cron <oracle@EFAGLIVEDB01> /u01/app/oracle/product/12.2.0/dbhome_1/backup_rman.sh
Content-Type: text/plain; charset=UTF-8
Auto-Submitted: auto-generated
Precedence: bulk
X-Cron-Env: <XDG_SESSION_ID=3697>
X-Cron-Env: <XDG_RUNTIME_DIR=/run/user/1000>
X-Cron-Env: <LANG=en_US.UTF-8>
X-Cron-Env: <SHELL=/bin/sh>
X-Cron-Env: <HOME=/u01/app/oracle/product/12.2.0/dbhome_1>
X-Cron-Env: <PATH=/usr/bin:/bin>
X-Cron-Env: <LOGNAME=oracle>
X-Cron-Env: <USER=oracle>
Message-Id: <20171109085501.54970C8001800@EFAGLIVEDB01.softlayer.com>
Date: Thu,  9 Nov 2017 02:55:01 -0600 (CST)

+ bakdir=/rmanbackup
+ bakwin=7
+ logfil=/var/log/backup_rman.log
+ esetup=/etc/profile.d/oracle.sh
+ mailad=root
++ echo -e 'select dbid from v$database;'
++ sqlplus / as sysdba
++ sed -e 's/ //g'
++ grep '[0-9][0-9][0-9][0-9][0-9][0-9]'
+ dbid=
+ test -z '' -a -f /etc/profile.d/oracle.sh
++ uname
+ test Linux = Linux -a -f /etc/SuSE-release
++ uname
+ test Linux = Linux -a -f /etc/redhat-release
+ grep -i 'Red Hat Enterprise Linux Server' /etc/redhat-release
++ uname
+ test Linux = SunOS
+ test -z ''
+ Error '$ORACLE_SID required.'
+ echo '$ORACLE_SID required.'
$ORACLE_SID required.
+ exit 1

manuell funktiniert das und via cronjob nicht?
Was mache ich flasch oder übersehe ich ???

Freue mich auf euer Feedback

Gruss aus ZH
Vorheriges Thema: Fedora 18 - Testbericht
Gehe zum Forum:
  


aktuelle Zeit: Sa Okt 20 02:21:18 CEST 2018

Insgesamt benötigte Zeit, um die Seite zu erzeugen: 0.00841 Sekunden