Server down

[teser]

My Solaris 2.6 server went down with a white screen and wont come back up.

As far as I know there was not any files near 100%. NOw to get this server back up
I need to get my 2.6 Solaris CD and do the following?
[tt]
boot cdrom -s

Do file system checks and (hopefully) automatically fix my filesystem using:
#fsck /dev/rdsk/c0t.........

If all filesystems are okay then do the following on each file(/var, /usr, root etc..)?
Also I do have a copy of prior filesystem space (df -k).
#newfs /dev/rdsk/c0t3d0s0

#fsck /dev/rdsk/c0t3d0s0

#mount /dev/dsk/c0t3d0s0 /a
#cd /a
#ufsrestore rvf /dev/rmt/0
#rm restoresymtable
#cd /
#umount /a
#fsck /dev/rdsk/c0t3d0s0
install bootblk for root
reboot[tt]

Is there any format of the hardrive or checking to see if the hardrive is okay??
Please advise if I am doing all this okay because it has been some time since I have done this.

Thanks

 

[tuka]

1. What do you mean that it doesn't come back up?

Is anything displayed after the system is powered?

if the system does display and you can get it to the ok prompt then type at the ok prompt: probe-scsi-all
ARe all the devices seen?
If it try to boot from the system disk what type of message are you getting.

Please let me know and we will continue.

 

[teser]

The server has a loud whistling sound coming from the internal hardrive with
message on white screen:
[tt]
Error for Comman: load/start/stop
Request Block 0
Vendor: Seagate[/tt]

The system does come up after powering it off but only stays on for about 30 mins.
Also the /var filesystem is filling up fast! It is now up to 90%. I assume a log file of message errors is filling it up?

So I need to put in a new drive and then format it and then reconfigure system etc..?

Please advise.

 

[tuka]

Are you sure which disk is bad? (example cXtXdXsX)

Let me know and I can give you instructions on replacing this drive.
If you have a spare drive of equal size or larger than the bad drive AND you can connect the new drive into the system
you could possibly do a dd from one drive to the other.

You mention the problem start within 30 minutes after re-powered. The disk fails after running for a while.

It is possible that the system could be power off and cooled down, the new disk inserted along with the existing disk(s and then repowered and the system rebooted. The new drive would need configuring into the system. Then you could do a dd from the bad disk to the new disk before the bad disk pukes. You might do something to insure more cooling is provided to the system while this is being done.

 

[teser]

Yes it is definetly one bad drive because I took it out and turned the system on and the loud whistling sound went away.

What is dd? Please advise how I can use it.

If I keep the old drive in there the /var will fill up to 100% because something (probably log files) is filling it up.

From my knowledge I assume I need the 2.6 Solaris cd and do a complete new install and reconfiuration of the system with the new hardrive?

Please advise.

 

[tuka]

What is the system type?
Is it the system disk

 

[teser]

Sun sparc 20, 2.6 , and it is the system disk.

 

[tuka]

If turning the system off for a while allows the system to come up and run for 30 minutes to 1 hours I would try to dd to a new disk. Here are the steps. I presume the system disk is c0t0d0 and the other disk is c0t1d0. If this isn't the case then use the correct devices.

1. While the system is up you can run this command to zero out the messages file.

# cat /dev/null >/var/adm/messages
2. Run init 0 or shutdown -i0
3. the system should be at the ok prompt
4. at the ok prompt type ok probe-scsi
this will print out the disk drives (cxtxdx)
5. Turn the power off and remove the good disk.
6. Install a new disk of equal or greater than the bad
disk into removed slot.
***insure the new disk has the same target as the
good disk. This is the cxtxdx (t=target #)
7. Let the system stay off and keep the cover off to allow
more cooling. Off for at least 15-20 minutes.
8. power on and boot cdrom -s
9. After system is up
10. cd /bin
11. run this command **insure the correct targets are used*
12. dd if=/dev/rdsk/c0t0d0s2 of=/dev/rdsk/c0t1dos2 bs=256k

/dev/rdsk/c0t0d0s2= should be the system disk
/dev/rdks/c0t1d0s2= should be the new disk

c0=controller 0
t0=target 0 ( should be the boot disk)
t1=target 1 (maybe the good/new disk target if not
please adjust this number from the info
you got from the probe-scsi)
s2= the whole disk

The dd command will take all the data from the old system
disk to the new disk and make the new disk bootable.

***If the dd finishes then shutdown and power off the system.

Remove the new disk and re-installed the original drive.
Remove the bad disk and replace with the new disk
***insure the new disk and the old disk has the same
target number.

power up and boot

I am leaving for the day, I will be back at 0500hrs 0n 1/18

If the dd doesn't work then we will have to try another method

 

[tuka]

I forgot to add in the earlier msg. After booting up single from the cd, fsck each partition prior to the dd.

The partition can be found by running format and select partition and print the partition.

 

[teser]

Thanks Tuka,

I will give it a try on Tuesday. ONe question: The disk that is bad is the system disk with root , /var etc... Can I do the dd with the bad disk copied to the new disk?

Also if this doesnt work , can I just put in another disk I have from another Sun Sparc 20 (2.6 OS) and just restore the backups to that disk using single user mode?
Or do I have to format the drive and reconfigure the system?

I originally had two internal drives with the one that is bad. The good one had just the database on it. The bad one (with root etc...system stuff) is partitioned (has two targets).

 

[tuka]

You mentioned the bad disk is partitioned with two targets.
How can a disk has two targets? You can have two LUN sharing one target.

Please send me a copy of the /etc/vfstab and the df -k,
if you can.

# /etc/vfstab >vfstab.out (this will place the content into a file you can send.)
# df -k >df-k.out

Yes you can ufsrestore from a backup tape BUT you must know the file systems and their size that were put on the original disk. A new disk must be partitioned for the slices and then newfs prior to the ufsrestore.

 

[tuka]

You had asked what is dd. dd command copies the input file
to output. It is used as a backup tool.

Please send me a copy of format along with the other files I requested.