Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations strongm on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

V480 failed drive

Status
Not open for further replies.

kHz

MIS
Dec 6, 2004
1,359
US
Yesterday a drive failed in a server, but the server was having other problems so moved both drives to another V480 that wasn't in use. Moved both drives both wouldn't boot. Finally removed hdd0 and moved hdd1 into hdd0 slot, rebuilt /etc/path_to_inst and everything is fine.

I now have a replacement drive that Sun sent. It will go into the other slot. V480s have hot-swappable scsi drives so I can slide the new drive into the slot. But what do I need to do for Solaris to recognize it? I think it is devfsadm but not entirely certain. And would it be with -C -c ?

Thanks.
 
I presume you wouldn't need to do a devfsadm because the device path for the old disk would normally be identical to the new disk. Stick it in and try accessing it using the existing device entries I guess?

Annihilannic.
 
You will probably be able to see the drive after swapping it without running devfsadm since it will occupy the same scsi target as the original. Just be cautious when you do- hot swapping can sometimes be a little flaky. If you have the luxury to do so take the system down before installing the replacement disk, if not, go for it, but to be on the safe side do it at a time when you can afford to have a little downtime if something goes wrong- whenever you plug a scsi device into a live system, there's always a (very) small chance of hanging the bus- it's a very tiny risk, I've only seen it happen maybe once or twice in the past 10 years (and that was on older enterprise class boxes).

If you need to do a devfsadm, you can use devfsadm -c disk
which will only reconfigure for disks. The uppercase C flag is for cleaning up unused device links. You couls always just run devfsadm with no options- unless you've been moving expansion cards around for no apparent reason, you won't break anything and will get the same results although the command may run several seconds longer than if you specify just disks.



Vincent Esposito
 
KHZ;

Last week we tested a couple things in the lab and found that just replaceing the fibre drive does not change the WWN info.

Since this is fibre drive you need to remove the path/WWN of the old drive. If this was scsi you could just pop in the new drive and go but since this is fibre you need to make sure the new drive/WWN replaces the old drive info.

So say if you do an;

ls -al /dev/dsk/c0t1d0s0 you will get the path with the WWN assigned to it. If I remove this drive and just install then run ls -al /dev/dsk/c0t1d0s0 you will still see the WWN of the old drive. You need to atleast run devfsadm -C -c disks so the wwn is updated for that drive target.

You need to take into account also the state of the bad or failing drive. I try to use the luxadm remove_device command for removing fibre as it removes the dev entries for the old drive. I know you can use the luxadm command on V880 so I would assume(hate to assume) that you can also use it on the v480.

If the drive is amber luxadm remove_device may not work. So if I wanted to replace the drive I would remove it from the system, install the new drive wait for it to spin up and run devfsadm -C -c disks.

If the drive is giving errors, but running I would umount the file system, run luxadm remove_device on bad drive, insert the new device, run luxadm insert_device.

Of course you will need to take into account if you are using Veritas or Disk Suite there are certain steps to follow.


Thanks

CA
 
boot from cdrom
ok> boot cdrom -s

fsck the boot disk
# fsck -y /dev/rdsk/c1t0d0s0

mount rootslice
# mount /dev/dsk/c1t0d0s0 /a

save old path_to_inst
# mv /a/etc/path_to_inst /a/etc/path_to_inst.orig

remove old devices
# rm /a/dev/rdsk/c* ; /a/dev/dsk/c* ; /a/dev/rmt* ; /a/dev/cfg/c*

rebuild path_to_inst
# devfsadm-r /a -p /a/etc/path_to_inst

unmount root slice and reboot
# umount /a ; init 6

 
CA, thanks for the info, I appreciate it. Do you happen to know what the difference in IO speed and throughput is for fibre disks in a V480 vs SCSI? Is there that much of an increase? Why didn't they just use SCSI drives?
 
KHZ,

I could not find that much. As fibre vs scsi goes there is enough of a difference to matter. Also with the fibre you can multipath which increases the throughput. You will generally get half your bandwidth which is actual throughput.

check this link out and search by throughput gives you some info.


TABLE 3-5 FC-AL Features and Advantages
FC-AL Features Advantages
Supports 100-Mbyte per second data transfer
rate (200 Mbytes per second with dual
porting).
High throughput meets the demands of current
generation high-performance processors and
disks.


also check out doc 43501 gives some scsi info.

In my position I don't deal to much with the bandwidth and throughput topic. Last class I took that really had any info on it was A5X00 training. I should really read up on the latest stuff.

Thanks

CA
 
Reading the pdf confirmed what I thought. The backplane needs an optional FC-AL HBA to control the second loop. Without the second HBA, dual-loop capability is lost along with the increased bandwidth and redundancy.

And without that dual-loop I don't see the benefit of those drives over SCSI (160 MB/s).

Thanks for the link. I'll read the infodoc, too.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top