1. According to the article, 000005973.
- Stopped Oracle
- Tried to umount /dev/odm
- Mounted it again.
- Started Oracle
- However, it ended up failing to bring Oracle online.
2. Then looked up the difference between this file system and other file systems in /etc/fstab and at OS level.
It is because it was shown that Oracle complained on the file system, '/symctst1 /oradata/control01.ctl'
----------------------------------------------------------------------------------------------------
# mount -v |egrep "symctst1|symctst2"
/dev/vx/dsk/SymcDg1/vol10 on /symctst1 type vxfs delaylog,cio <<< This is the problem!
/dev/vx/dsk/SymcDg2/vol11 on /symctst2 type vxfs delaylog, qio
# cat /etc/fstab |egrep "symctst1|symctst2"
/dev/vx/dsk/SymcDg1/vol10 /symctst1 vxfs suid, cio 0 0 <<< This is the problem!
/dev/vx/dsk/SymcDg2/vol11 /symctst2 vxfs suid 0 2
----------------------------------------------------------------------------------------------------
3. Thereafter, it was able to find out the big difference between symctst1(abnormal) and other file system(normal)
- The file system, symctst1 was mounted with "cio" option.
- Whereas, the other file system were mounted with "qio" by default.
- It is because most of all, the vxfs file system must be supported under ODM with "qio".
- However, it seems that there should have been some change in the entry of /etc/fstab due to some reason by someone else, which is very severe and critical.
4. So therefore, /etc/fstab file was modified correctly on /symctst1 with qio option.
----------------------------------------------------------------------------------------------------
[ BEFORE ]
/dev/vx/dsk/SymcDg1/vol10 /symctst1 vxfs suid,cio 0 0
[ AFTER ]
/dev/vx/dsk/SymcDg1/vol10 /symctst1 vxfs suid 0 2
----------------------------------------------------------------------------------------------------
5. Then Oracle DBA confimed Oracle engine got brought up without a file.
Applies To
[ VERSION OF OS/PACKAGE ]
- HPUX11v3
- SFHA5.0 for HPUX11v3