Syntax for dropping standby redo logs

In one of my primary databases, I had several old standby redo log files that were forgotten from a previous relocation using Data Guard. They could all be dropped at this point.

set lines 200
col member format a50
select s.group#,s.thread#,s.sequence#,s.archived,s.status,f.member, f.type
from v$standby_log s, v$logfile f
where f.type = 'STANDBY'
and s.group# = f.group#;

    GROUP#    THREAD#  SEQUENCE# ARC STATUS     MEMBER                                             TYPE
---------- ---------- ---------- --- ---------- -------------------------------------------------- -------
         6          1          0 YES UNASSIGNED /data03/oradata/prod01/onlinelog/stb_redo06.log         STANDBY
         7          1          0 YES UNASSIGNED /data03/oradata/prod01/onlinelog/stb_redo07.log         STANDBY
         8          1          0 YES UNASSIGNED /data03/oradata/prod01/onlinelog/stb_redo08.log         STANDBY
         9          1          0 YES UNASSIGNED /data03/oradata/prod01/onlinelog/stb_redo09.log         STANDBY
        10          1          0 YES UNASSIGNED /data03/oradata/prod01/onlinelog/stb_redo10.log         STANDBY
        11          1          0 YES UNASSIGNED /data03/oradata/prod01/onlinelog/stb_redo11.log         STANDBY

Syntax for dropping:

SQL> alter database drop standby logfile group n;

In my case

alter database drop standby logfile group 6;
alter database drop standby logfile group 7;
etc etc.



The official Oracle 19c Alter Database documentation is here

Workaround for error adstpall.sh: too few arguments specified when stopping the EBS 12.2 server processes

ebs@ebsserver1.oric.no 35642926]$ $ADMIN_SCRIPTS_HOME/adstpall.sh –skipNM -skipAdmin
 
You are running adstpall.sh version 120.22.12020000.7
 
 
Enter the WebLogic Server password:
adstpall.sh: too few arguments specified.
 
USAGE:  adstpall.sh  [-skipNM] [-skipAdmin] [-nothreading]
        adstpall.sh  -secureapps [-skipNM] [-skipAdmin]
        adstpall.sh -nodbchk [-skipNM] [-skipAdmin]
        adstpall.sh -mode=allnodes
 
 
adstpall.sh: exiting with status 1

Reason: You have copied the text directly from step 5 in the documentation

Turns out the docs use a different dash character than the expected one. The one used in the documentation is most likely copied from MS word or similar, which is interpreted differently than the regular dash character:

Incorrect: –skipNM
Correct:   -skipNM

In other words, the documented command below is incorrect:

$ $ADMIN_SCRIPTS_HOME/adstpall.sh –skipNM -skipAdmin

Workaround: execute it with the correct dash

$ $ADMIN_SCRIPTS_HOME/adstpall.sh -skipNM -skipAdmin

Thanks to Marco DeDecker from Oracle Netherlands for finding and pointing out this error.

How to write custom messages to the alert log file

exec sys.dbms_system.ksdwrt(2,'Your message here')

The first argument can be either

1 -> Write to the trace file

2 -> Write to the database alert log file

3 -> Write to both of the above files

Syntax for applying and removing a patch on the WLS server Home in an EBS 12.2 installation

The bsu (or "Smart update") utility:

cd $MW_HOME/utils/bsu
bsu.sh -version
Oracle Smart Update. Version: 3.3.0.0

Here's an example on how to install a patch:

./bsu.sh -remove -patch_download_dir=$MW_HOME/utils/bsu/cache_dir -patchlist=1LRI -prod_dir=$MW_HOME/wlserver_10.3 -verbose -log=install_1LRI.log

Here's an example on how to remove a patch:

./bsu.sh -install -patch_download_dir=$MW_HOME/utils/bsu/cache_dir -patchlist=E7HI -prod_dir=$MW_HOME/wlserver_10.3 -verbose -log=install_E7HI.log

Script sniplet to differentiate between cdb and pdb environment settings in a shell script

This little sniplet could be used if you find yourself in a situation where you need to differentiate between the cdb environment and the pdb environment on a server using multitenant setup.

I am setting a commonly used alias for the usage of sqlplus called "sql" depending on the ORACLE_SID being set to the cdb or the pdb.

It differs slightly since I you can only do a bequath session when connecting to the former, and not the latter:

When connecting to the pdb as sysdba, you need to go through the listener, and therefore supply the password.

When connecting to the cdb you can do a so called bequeath session and connect directly to the oracle instance through IPC.

read -p 'Will you be administrating the cdb or the PDB? ' container
case $container in
        CDB|cdb)
        . $ORACLE_HOME/cdb_settings.env;
        alias sql='sqlplus / as sysdba';;
        PDB1|pdb1)
        . $ORACLE_HOME/pdb1.env;
        alias sql='sqlplus sys@pdb1 as sysdba';;
        *) echo "You must select cdb or pdb1";
         exit 1;;
esac

How to generate a cold backup script for a database

Probably many ways to to this, but here is how I generated a simple file that after being made executable will copy all files to a specific folder. The database must be shut down before running the script, so in other words, this will be a good, old-fashioned cold backup!

set lines 200
set pages 0
set trimspool on
set heading off
set echo off
set feedback off
set verify off

spool cp_files.sh


SELECT    'cp '
       || a.file_name
       || ' /u01/oracle/cold_backup/'
       || (SELECT TRIM (SUBSTR (b.file_name,
                                (  INSTR (b.file_name,
                                          '/',
                                          -1,
                                          1)
                                 + 1)))
             FROM cdb_data_files b
            WHERE a.file_name = b.file_name)
  FROM cdb_data_files a
UNION
SELECT    'cp '
       || a.MEMBER
       || ' /u01/oracle/cold_backup/'
       || (SELECT TRIM (SUBSTR (b.MEMBER,
                                (  INSTR (b.MEMBER,
                                          '/',
                                          -1,
                                          1)
                                 + 1)))
             FROM v$logfile b
            WHERE a.MEMBER = b.MEMBER)
FROM v$logfile A
UNION
SELECT    'cp '
       || a.name
       || ' /u01/oracle/cold_backup/'
       || (SELECT TRIM (SUBSTR (b.name,
                                (  INSTR (b.name,
                                          '/',
                                          -1,
                                          1)
                                 + 1)))
             FROM v$tempfile b
            WHERE a.name = b.name)
  FROM v$tempfile a
UNION
SELECT    'cp '
       || a.name
       || ' /u01/oracle/cold_backup/'
       || (SELECT TRIM (SUBSTR (b.name,
                                (  INSTR (b.name,
                                          '/',
                                          -1,
                                          1)
                                 + 1)))
             FROM v$controlfile b
            WHERE a.name = b.name)
  FROM v$controlfile a;

How to use strace to figure out what files are being accessed by a shell script

I had a situation where an ebs-supplied script adstrtal.sh would not start - it kept throwing error

ORA-12541: TNS:no listener

Although $TNS_ADMIN was correctly set, and sqlplus and tnsping would confirm that the database was open and the listener was up, accepting connections on the desired port.

The sqlnet.log file created in the same directory from which I executed adstrtal.sh displayed the connection being attempted:

Fatal NI connect error 12541, connecting to:
 (DESCRIPTION=(CONNECT_DATA=(SERVICE_NAME=EBS32.oric.no)(CID=(PROGRAM=sqlplus)(HOST=oric-ebsapp-utv.oric.no)(USER=ebs)))
 (ADDRESS=(PROTOCOL=TCP)(HOST=162.20.5.225)(PORT=1521)))

We are not using default port 1521, but a different port.

I then used strace to find the source of the error, like this:

strace -f -o /tmp/strace.out ./adstrtal.sh apps/****

When going through the /tmp/strace.out file, I was pointed in the right direction:

openat(AT_FDCWD, "$INST_TOP/ora/10.1.3/network/admin/EBS32_oric-ebsapp-utv_ifile.ora", O_RDONLY|O_LARGEFILE) = -1 ENOENT (No such file or directory)

Turns out that adstrtal.sh was looking for a tnsnames.ora entry not in the tnsnames.ora in $TNS_ADMIN ($INST_TOP/ora/10.1.2/network/admin) but rather in $INST_TOP/ora/10.1.3/network/admin.

As soon as I had added the correct tnsnames.ora entry, the adstrtall.sh script worked.