I have one big problem! Today, I do not log in my server which contains a production Oracle Database, however, it still runs. No console, no telnet, no SSH is to be allowed. I do not understand, really.
Google searching, I found that, one action can cause it. It is described like "Some one changed owner one of root's directory to the other, it may be /etc, /sbin..., and it caused no-one to connect the server".
The problem I traced it by copied file automatically before its accidents is " #chown -R oracle:dba /etc"
And then, I do not understand how to fix it. Said my friend "Do not reboot it, because it can be died".
Step 1: Boot by CDROM
Step 2: List device by format command line, to display all of devices, the first is often boot disk.
Step 3: Mount this disk
Step 4: Re-change the owner of /etc from oracle to root
Step 5: Unmount the disk
Step 6: Reboot.
Your procedure may work, but remember that format will give you slice 2 of each disk and root is probably slice 0.
c1t1d0s2
c1t1d0s0
Next remember that if you mount the disk on /mnt, you need to fix /mnt/etc not /etc. Also chmod -R changed all file and subdirectories under /etc not just /etc.
This will not work if the drive is mirrored. After you mount root, cat /mnt/etc/vfstab to see if if root is really what you think it is. If it is mirrored, you will need a much more complex operation to unmirror root. Then if you get it working, you can resync the mirrors.
Definitely wrong. But I have no experience with a V890 with raid 0 (or even a v890 at all) so I don't know the correct procedure. But first of all, step 4 is wrong. When you boot from cdrom, you get a little /etc created for you. This /etc is not your problem. After you mount to /a you will have /a/etc and this is what needs fixing. But you should do the mount and then simply
cat /a/etc/vfstab
and post this. Then someone who understands your setup better than I may be able to help.
I have to tell you that I would not try for too long to fix this. It might be easier to simply rebuild the box.
I have just reboot in single mode successfully, but I get one trouble. My server is running with SAN, and when I use the statment
#echo | format
this show like
AVAILABLE DISK SELECTIONS:
0. c1t0d0 <SUN36G cyl 24620 alt 2 hd 27 sec 107>
/pci@8,600000/SUNW,qlc@4/fp@0,0/ssd@w21000004cf6f1b1f,0
1. c1t1d0 <SUN36G cyl 24620 alt 2 hd 27 sec 107>
/pci@8,600000/SUNW,qlc@4/fp@0,0/ssd@w21000004cf6efe8d,0
2. c5t600A0B80001323B80000000040F2051Dd0
Now, I have solved this problem, it's caused by my mistake.
I am a newbie, so that, instead of booting -s to direct load /dev/dsk/Disk_root but I boot by cdrom -s.
Fixed by[/b] chgrp [b]command line :D. When I used
#chown -R root:sys /etc
This show the errors (I am sorry but I did not remember exactly) and I could not change the owner.