Chapter 17 Troubleshooting RMAN

Download Report

Transcript Chapter 17 Troubleshooting RMAN

Chapter 17
Troubleshooting RMAN
1
Background



2
Authors thought this topic was often glazed
over or not covered well
Knew that every topic can’t be covered, that
could be a very long list
Tried to cover the most common types of
RMAN issues
Sources of Information





3
Backup and recovery problems are stressful
Usually you can find somebody else who has
already encountered the issue
Start at bottom of error stack and work your
way up
Syntax issues are common
Check alert.log and trace files
Sources of Information (continued)





4
Ask for help
Search Oracle’s documentation
Use RMAN forum
Open a SR with Oracle
Enable output logging (Oracle Support will
often request this)
Resolving Connection Issues




5
Common problem
Usually caused by not using an
authenticated user
If using a password file, ensure correct
username/password are being used
Attempt to connect to SQL*Plus to
troubleshoot
Handling Disk Space Issues


Usually caused by running out of disk space
To correct:
–
–
–
–
–
6
Change location of the backup
Add disk space (if possible)
If using a FRA, either add space or move
Change retention policy
Delete old files
Dealing with RMAN-06059 Error




Common problem, occurs when RMAN doesn’t know
where old archive redo log files are located
Due to switching from user managed to RMAN
Due to not using RMAN to delete old files
“loss of archived log compromises recoverability”...
Yikes!
RMAN> crosscheck archivelog all;
7
Terminating RMAN Processes





8
Sometimes necessary when backup or
restore hangs
Try control + C
Manually kill process from operating system
Kill RMAN session from SQL*Plus
Query v$session_longops to see if the job is
actually making progress
Diagnosing NLS Character Set Issues



ORA-12705 error
Sometimes caused by a mismatch between
NLS character set on client and server
Sometimes caused by an erroneously set
NLS-related operating system parameter
SQL> select value from v$nls_parameters
where parameter = 'NLS_CHARACTERSET';
9
Logging RMAN Output
Extremely helpful for debugging/diagnosing issues
 From OS prompt:
$ rman target / log=rman_output.log

From within RMAN:
RMAN> spool log to rman_output.log
RMAN> set echo on
RMAN> backup database;
RMAN> spool log off;

10
Viewing RMAN Command History
Use v$rman_output
SQL> select sid, recid, output
from v$rman_output
order by recid;
 Holds 32,768 rows

11
Enabling Debug Output

You can turn on debugging in several
different ways:
–
–
–
–
12
From the OS prompt
When allocating a channel
When configuring a channel
From the RMAN command-line prompt
Enabling Debug Output
Enable all debugging
$ rman target / debug=all log=rman_output.log

Enables debugging just for I/O activities:
$ rman target / debug=io

13
Information Types of Debugging





14
io
sql
plsql
rcvman
rpc
Enabling Granular Time Reporting
Oracle Support will often recommend that you
enable this before sending them output
 Allows you to view down to the second when
operations occurred
 Authors recommend that you always enable this
 Korn shell example:
$ export NLS_DATE_FORMAT='dd-mon-yyyy
hh24:mi:ss'

15
Working with Oracle Support


Some problems will require help from Oracle
Support
To facilitate response:
–
–
–
–
–
16
Test case that illustrates step-by-step details to reproduce
the problem
Complete RMAN script or command(s) that were run
Complete RMAN output log
Database alert.log file
Remote Diagnostic Agent (RDA) output (optional)
Resolving RMAN Compatibility Issues
Ideally, all databases and catalog will be at
the same level
 Reality, almost never the case
 Most up to date source RMAN compatibility
matrix information in MetaLink note 73431.1
 Checking the recovery catalog version (if
using):
SQL> select * from rcver;

17
Dealing with an ORA-19511 Error





18
Vexing issue
Caused by MML mis-configuration
Caused by incorrectly set OS variables
Work with your MML vendor to resolve
Check sbtio.log file
Dealing with an ORA-27211 Error




19
Vexing issue
Usually related to MML not being correctly
installed
Work with MML vendor to resolve
Look in log and trace files for more
information
Dealing with an ORA-04031 Error





20
Enabled I/O slaves
RMAN will throw this if it can’t allocate
enough memory
Setup a large pool area
Use ASMM
See MetaLink note 73354.1
Managing Files in an ASM Environment
Background: Technical editor wrote this material in
the book, often received questions regarding ASM at
Oracle Support
 Using ASM (Automatic Storage Management)
 Need to view RMAN files
 Use ASMCMD
 Command line utility
$ asmcmd –p
 Unix like commands

21