Redo Corruption Errors During Redo Transport [ID 386417.1]

Applies to:

Oracle Server - Enterprise Edition - Version: 9.2.0.1.0 to 9.2.0.7.0
This problem can occur on any platform.

Symptoms

When shipping redo to a standby database, the following errors are reported in the primary alert.log:

Tue Aug 1 13:01:11 2006
ARC0: I/O error 272 archiving log 3 to 'STANDBY'
Tue Aug 1 13:01:11 2006
Errors in file /u01/app/oracle/admin/PRIM/bdump/PRIM_arc0_3329.trc:
ORA-00272: error writing archive log
Tue Aug 1 13:01:25 2006
ARC0: Completed archiving log 3 thread 1 sequence 11333
Tue Aug 1 13:06:46 2006
ARC1: Begin FAL archive (thread 1 sequence 11327 destination STANDBY
Creating archive destination LOG_ARCHIVE_DEST_2: 'STANDBY'
Tue Aug 1 13:06:47 2006
Errors in file /u01/app/oracle/admin/PRIM/bdump/PRIM_arc1_3331.trc:
ORA-03114: not connected to ORACLE
ARC1: FAL archive failed, see trace file.
ARCH: FAL archive failed. Archiver continuing

The ORA-272 occurs whether using ARCH or LGWR mode of redo transport and during FAL service requests.

 

On the standby side, one or both of the following types of errors may be reported by RFS:

Error #1:
--------------
Tue Aug 1 13:00:44 2006
Errors in file /u01/app/oracle/admin/STDBY/udump/STDBY_rfs_17699.trc:
ORA-00600: internal error code, arguments: [kcrrrfswda.6], [4], [], [], [], [] , [], []
< snip text >
Tue Aug 1 13:00:46 2006
Errors in file /u01/app/oracle/admin/STDBY/udump/STDBY_rfs_17699.trc:
ORA-00367: checksum error in log file header
ORA-00311: cannot read header from archived log
ORA-00334: archived log: '/u02/arch/STDBY/STDBY_0000011330.arch'
ORA-27091: skgfqio: unable to queue I/O
ORA-27072: skgfdisp: I/O error
SVR4 Error: 2: No such file or directory
Additional information: 1



Error #2:
-----------------
Tue Aug 1 13:00:46 2006
Errors in file /u01/app/oracle/admin/STDBY/udump/STDBY_rfs_17701.trc:
RFS: Possible network disconnect with primary database
< snip text >
Tue Aug 1 13:00:46 2006
Errors in file /u01/app/oracle/admin/STDBY/udump/STDBY_rfs_17701.trc:
ORA-00367: checksum error in log file header
ORA-00311: cannot read header from archived log
ORA-00334: archived log: '/u02/arch/STDBY/STDBY_0000011331.arch'
ORA-27091: skgfqio: unable to queue I/O
ORA-27072: skgfdisp: I/O error
SVR4 Error: 2: No such file or directory
Additional information: 1
ORA-12592: TNS:bad packet

Changes

A firewall has been added to the network between the primary and standby sites or already exists.  The firewall may be Cisco-based and have a feature called 'fixup' enabled which may or may not be configured fully.

Database auditing may have also been enabled or is currently in use per the initialization parameter AUDIT_TRAIL.

Cause

When auditing is enabled, queries or DML to the SYS.AUD$ table may trigger larger then normal TCP packets.  When these packets go through the firewall, the 'fixup' can possibly modify the packet incorrectly and cause it to be corrupted which then breaks Oracle Net.

Solution

Several workarounds are available which must be evaluated and tested individually based on the requirements of each environment.

Listed with the most commonly used at the top:

1. Turn off database auditing.
   
   
2. If auditing must be maintained and the 'fixup' feature cannot be disabled, work with the network administrators to evaluate the MTU for the firewall and size the Oracle TCP packets to be below the current setting. The SDU/TDU settings are configured via the Oracle Net files on both primary and standby.  See the References section below for related content.
   
   
3. Disable the 'fixup' feature and/or consult with the firewall vendor regarding patches/fixes to the feature.
   
   
4. Disable the firewall.
   

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/11134237/viewspace-684341/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/11134237/viewspace-684341/