Solaris

Our SunFire 480 - one of the boot disks failed. It is mirrored with VxVm. I followed the exact steps that is detailed in the veritas and sun guides (I can not post the URL since I don't have 5 posts yet.)

After removing the failed disk, I ran devfsadm -C -c disk and it still shows the device links under /dev/dsk and /dev/rdsk. So, I can not put in the new drive and the server is running with only one boot disk (without any mirrors).

So, what is the best way to get rid of the device links for the failed disk under /dev/dsk and /dev/rdsk?

Code:

devfsadm -C

Should have re-created the links...
Perhaps some outputs from your terminal session would help...

What outputs are you looking for?

What makes you think that you need to delete the device node??

If a disk fails and you replace it with another in the same slot then the replacement disk will need/use the same device node.

If you are sure that you must delete the device node (and then recreate it if you are using the same slot) then you can force deletion with:

Code:

# rm -f /dev/dsk/<device node>

but I don't think that you need to do that.

devfsadm can be used (in this context) to (re)create device nodes that you need to use.

Also, devfsadm will not create a device node for a disk that is not yet in the machine.

480 crashed and we brought it up using the rootmirror. Now, we are trying to replace the first drive. We ran the following steps before pulling out the disk:
1. vxdiskadm - Step #4 - Remove a disk for replacement
2. luxadm remvove_disk c1t0d0s2 - this step gave an error saying that invalid path
3. So, I had to do a luxadm with a force flag like this: - > luxadm -v remove_device -F /dev/rdsk/c1t0d0s2

(This step also gave an error after pulling out the disk saying Error: Unexpected Fibre Channel topology found. - /dev/rdsk/c1t0d0s2.)


4. devfsadm -C -c disk - this should have cleaned all the device tree links for c1t0d0 under /dev/dsk and /dev/rdsk - but it did not remove it.


What is the best option to get rid of the device links under /dev/dsk and /dev/rdsk for this removed drive?

As I already said in post#4, if you really think it necessary, you can delete a device node by using:

Code:

# rm -f /dev/dsk/<device node>

You should also remove the same device node from /dev/rdsk and /dev/cfg

Does it cause any other trouble if I do forced rm -f to delete the device nodes? Anything you can think of?