[Ocfs2-users] ORA-19870 and ORA-19502 During RMAN restore to OCFS2 filesystem

Luis Freitas lfreitas34 at yahoo.com
Thu Jul 24 12:40:04 PDT 2008


Ed,

   My two cents here.

   I have some 3Gb datafiles here and restore to OCFS2 using rman with no problems. But I am using Linux 32bits and the backup is read from veritas netbackup so it is not exactly the same environment.

   Also, 1409*8192 gives about 11Mb, so you are well under 2Gb when the problem occurs. The fact that it is occuring on the larger files may be a coincidence. 

   You may be running into some operating system error or some other issue. Check /var/log/messages and dmesg to see if something is appearing. Are your ext3 filesystem also on the SAN?

    Also, just a hunch, check if you are reaching the maximum AIO operations, the default on RedHat 4.0 is way low and this caused us some headaches when we implemented 10g, I never looked how this is set on RH 5.0.

    The current aio pending operations shows on /proc/sys/fs/aio-nr, the maximum is on /proc/sys/fs/aio-max-nr, if you are hitting the maximum AIO try to increase it, we set this to 1000000 here on /etc/sysctl.conf (fs.aio-max-nr = 1000000). 

Regards,
Luis

--- On Wed, 7/23/08, Ed Gulakowski <edgulak at gmail.com> wrote:
From: Ed Gulakowski <edgulak at gmail.com>
Subject: [Ocfs2-users] ORA-19870 and ORA-19502 During RMAN restore to OCFS2 filesystem
To: ocfs2-users at oss.oracle.com
Date: Wednesday, July 23, 2008, 11:48 PM


 
Hi,
 
When attempting to restore to LINUX RHEL5 - OCFS2 filesystem received the following error during RMAN restore for nearly all of the datafiles attempted to restore with exception of a couple of smaller datafiles which were smaller < 2GB.

 
ORA-19870: error reading backup piece /db/dumps/TR1_1/rmanbackup/TR1_88_1
ORA-19502: write error on file "/db/devices/db1/PR2/pr2_1/pr2.data1", blockno 1409 (blocksize=8192)
ORA-27061: waiting for async I/Os failed
Linux-x86_64 Error: 5: Input/output error
 
Tried various Oracle init parameters for filesystem i/o SETALL, ASYNCH, DIRECT and also disk_asynch_io settings.
 
After this failure tried to used a non ocfs2 filesytem (ext3) for the source RMAN backup files, however the error still occurred.  For testing purposes specified  a non ocfs2 filesystem ( ext3 ) as restore target filesystem and the RMAN restore completed successfully without any errors.

 
 
Let me know if you have any suggestions.
Thanks,
Ed
 
 
 
 
_______________________________________________
Ocfs2-users mailing list
Ocfs2-users at oss.oracle.com
http://oss.oracle.com/mailman/listinfo/ocfs2-users


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://oss.oracle.com/pipermail/ocfs2-users/attachments/20080724/1e069e4e/attachment.html 


More information about the Ocfs2-users mailing list