Fibre Channel - device LUN cleanup on Solaris

 

Procedure to assist in removing Fiber Channel devices (on Solaris 10)

Based on: Symatec's Veritas Technote 150451,   Veritas VxVM 5 release Notes,  and Breno Sobral Storage Blog.

Gathering all the device information for failing devices and formulating the appropriate commands can be cumbersome.  We have an in-house tool to assist in the process that is non-destructive.  The tool will show the commands that you should use.  To use the tool simple run as root:    /install/veritas/vxvm/fc_show

 

Prereq for Veritas:    Remove device from VxFS/VxVM first . 

    If devices are under Veritas conrol:

  • first unmount the affected underlying VxFS filesystems . 
  • then remove associated disk from VxVM volume manager:   "vxdisk rm" on the device

 

General Procedure to remove devices SAN-attached storage

Description
When storage devices that present multiple luns to Solaris[TM] through a Storage Area Network(SAN) have some of those luns removed or made unavailable, Solaris device entries will still exist for those luns.

However Solaris may report "missing" or "failing" states for those luns. 
This document explains how to clean up the device entries and hence remove the error condition which is 
caused by Solaris trying to access the unavailable luns. 
This applies to Solaris 8 and Solaris 9 using the Sun StorEdge[TM] San Foundation Kit (SFK), 
also known as the Leadville driver stack. This document is specific to SAN-attached fibre channel storage and 
does not apply to direct-attached fibre channel storage.

Steps to Follow
The following commands will be presented:

- cfgadm -c configure [ap_id]
- cfgadm -al -o show_FCP_dev
- cfgadm -o unusable_FCP_dev -c unconfigure [ap_id]
- devfsadm -C
- ("luxadm -e offline <device path>" may also be needed)
The following output shows a system with 4 dual-pathed luns which are SAN-attached to a Solaris host:

# cfgadm -al -o show_FCP_dev

Ap_Id Type Receptacle Occupant Condition
c2 fc-fabric connected configured unknown
c2::50060e8004274d20,0 disk connected configured unknown
c2::50060e8004274d20,1 disk connected configured unknown
c2::50060e8004274d20,2 disk connected configured unknown
c2::50060e8004274d20,3 disk connected configured unknown
c3 fc-fabric connected configured unknown
c3::50060e8004274d30,0 disk connected configured unknown
c3::50060e8004274d30,1 disk connected configured unknown
c3::50060e8004274d30,2 disk connected configured unknown
c3::50060e8004274d30,3 disk connected configured unknown

# format

Searching for disks...done

AVAILABLE DISK SELECTIONS:

(output omitted for clarity)

4. c2t50060E8004274D20d0 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,0
5. c2t50060E8004274D20d1 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,1
6. c2t50060E8004274D20d2 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,2
7. c2t50060E8004274D20d3 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,3
8. c3t50060E8004274D30d0 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,0
9. c3t50060E8004274D30d1 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,1
10. c3t50060E8004274D30d2 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,2
11. c3t50060E8004274D30d3 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,3

In this example, using the native tools of the storage device, we will remove all of the odd numbered luns. Here the storage is a Sun StorEdge[TM] 9990, so we used Storage Navigator to remove the lun mappings from the host.

The following output shows the same system after the luns have been removed:

# cfgadm -al -o show_FCP_dev

Ap_Id Type Receptacle Occupant Condition
c2 fc-fabric connected configured unknown
c2::50060e8004274d20,0 disk connected configured unknown
c2::50060e8004274d20,1 disk connected configured failing
c2::50060e8004274d20,2 disk connected configured unknown
c2::50060e8004274d20,3 disk connected configured failing
c3 fc-fabric connected configured unknown
c3::50060e8004274d30,0 disk connected configured unknown
c3::50060e8004274d30,1 disk connected configured failing
c3::50060e8004274d30,2 disk connected configured unknown
c3::50060e8004274d30,3 disk connected configured failing

# format

Searching for disks...done

AVAILABLE DISK SELECTIONS:

(output omitted for clarity)

4. c2t50060E8004274D20d0 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,0
5. c2t50060E8004274D20d1 <drive type unknown>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,1
6. c2t50060E8004274D20d2 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,2
7. c2t50060E8004274D20d3 <drive type unknown>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,3
8. c3t50060E8004274D30d0 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,0
9. c3t50060E8004274D30d1 <drive type unknown>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,1
10. c3t50060E8004274D30d2 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,2
11. c3t50060E8004274D30d3 <drive type unknown>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,3

We can now see above, that "cfgadm -al -o show_FCP_dev" shows the "failing" state and format shows the device as "<drive type unknown>".

The first step in removing these devices is to change the state shown in the cfgadm output from "failing" to "unusable". This is done with the following command:

# cfgadm -c configure c2 c3
# cfgadm -al -o show_FCP_dev

Ap_Id Type Receptacle Occupant Condition
c2 fc-fabric connected configured unknown
c2::50060e8004274d20,0 disk connected configured unknown
c2::50060e8004274d20,1 disk connected configured unusable
c2::50060e8004274d20,2 disk connected configured unknown
c2::50060e8004274d20,3 disk connected configured unusable
c3 fc-fabric connected configured unknown
c3::50060e8004274d30,0 disk connected configured unknown
c3::50060e8004274d30,1 disk connected configured unusable
c3::50060e8004274d30,2 disk connected configured unknown
c3::50060e8004274d30,3 disk connected configured unusable

Possible extra step:

If the devices remaining in a "failing" state according to the above output from cfgadm, and they do not move to an "unusable" state after running "cfgadm -c configure" as shown above, then the following command can also be tried:

# luxadm -e offline /dev/dsk/c3t50060E8004274D30d3s2

(i.e. "luxadm -e offline <device path for LUN in 'failing' state from cfgadm>)

Then re-run the previous cfgadm command (cfgadm -al -o show_FCP_dev) to check that the LUN state has changed from "failing" to "unusable". This luxadm command should then be repeated for each LUN which was previously shown in the "failing" state by cfgadm. Then carry on with the process below.

--oOo--

Now that the state of the inaccessible luns has been changed to "unusable" in the output from cfgadm, we can remove those entries from the list with the following command:

# cfgadm -o unusable_FCP_dev -c unconfigure c2::50060e8004274d20
# cfgadm -o unusable_FCP_dev -c unconfigure c3::50060e8004274d30

 

- If you tried remove device and you got some errors below, you'll use this:

# cfgadm -o unusable_FCP_dev -c unconfigure c3::50060e8004274d30
cfgadm: Library error: failed to offline: /devices/scsi_vhci/ssd@g600015d00005cc00000000000000f166
                    Resource                             Information     
------------------------------------------------  -------------------------
/dev/dsk/c6t600015D00005CC00000000000000F166d0s2  Device being used by VxVM

# cfgadm -f -o unusable_FCP_dev -c unconfigure c3::50060e8004274d30

 

# cfgadm -la -o show_FCP_dev

Ap_Id Type Receptacle Occupant Condition
c2 fc-fabric connected configured unknown
c2::50060e8004274d20,0 disk connected configured unknown
c2::50060e8004274d20,2 disk connected configured unknown
c3 fc-fabric connected configured unknown
c3::50060e8004274d30,0 disk connected configured unknown
c3::50060e8004274d30,2 disk connected configured unknown

# format

Searching for disks...done

AVAILABLE DISK SELECTIONS:

(output omitted for clarity)

4. c2t50060E8004274D20d0 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,0
5. c2t50060E8004274D20d2 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1/fp@0,0/ssd@w50060e8004274d20,2
6. c3t50060E8004274D30d0 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,0
7. c3t50060E8004274D30d2 <HITACHI-OPEN-V-5002 cyl 1711 alt 2 hd 15 sec 512>
/pci@23c,600000/SUNW,qlc@1,1/fp@0,0/ssd@w50060e8004274d30,2

Now we see that the luns are no longer displayed in the format listing.

Even though the output of the format command looks good, there are still entries for the removed devices in /dev/disk and /dev/rdsk. These can be removed if desired, by using the devfsadm command.

# ls /dev/dsk/c2t50060E8004274D20d*

/dev/dsk/c2t50060E8004274D20d0s0 /dev/dsk/c2t50060E8004274D20d2s4
/dev/dsk/c2t50060E8004274D20d0s1 /dev/dsk/c2t50060E8004274D20d2s5
/dev/dsk/c2t50060E8004274D20d0s2 /dev/dsk/c2t50060E8004274D20d2s6
/dev/dsk/c2t50060E8004274D20d0s3 /dev/dsk/c2t50060E8004274D20d2s7
/dev/dsk/c2t50060E8004274D20d0s4 /dev/dsk/c2t50060E8004274D20d3s0
/dev/dsk/c2t50060E8004274D20d0s5 /dev/dsk/c2t50060E8004274D20d3s1
/dev/dsk/c2t50060E8004274D20d0s6 /dev/dsk/c2t50060E8004274D20d3s2
/dev/dsk/c2t50060E8004274D20d0s7 /dev/dsk/c2t50060E8004274D20d3s3
/dev/dsk/c2t50060E8004274D20d1s0 /dev/dsk/c2t50060E8004274D20d3s4
/dev/dsk/c2t50060E8004274D20d1s1 /dev/dsk/c2t50060E8004274D20d3s5
/dev/dsk/c2t50060E8004274D20d1s2 /dev/dsk/c2t50060E8004274D20d3s6
/dev/dsk/c2t50060E8004274D20d1s3 /dev/dsk/c2t50060E8004274D20d3s7
/dev/dsk/c2t50060E8004274D20d1s4 /dev/dsk/c2t50060E8004274D20d4s0
/dev/dsk/c2t50060E8004274D20d1s5 /dev/dsk/c2t50060E8004274D20d4s1
/dev/dsk/c2t50060E8004274D20d1s6 /dev/dsk/c2t50060E8004274D20d4s2
/dev/dsk/c2t50060E8004274D20d1s7 /dev/dsk/c2t50060E8004274D20d4s3
/dev/dsk/c2t50060E8004274D20d2s0 /dev/dsk/c2t50060E8004274D20d4s4
/dev/dsk/c2t50060E8004274D20d2s1 /dev/dsk/c2t50060E8004274D20d4s5
/dev/dsk/c2t50060E8004274D20d2s2 /dev/dsk/c2t50060E8004274D20d4s6
/dev/dsk/c2t50060E8004274D20d2s3 /dev/dsk/c2t50060E8004274D20d4s7

# devfsadm -C
# ls /dev/dsk/c2t50060E8004274D20d*

/dev/dsk/c2t50060E8004274D20d0s0 /dev/dsk/c2t50060E8004274D20d2s0
/dev/dsk/c2t50060E8004274D20d0s1 /dev/dsk/c2t50060E8004274D20d2s1
/dev/dsk/c2t50060E8004274D20d0s2 /dev/dsk/c2t50060E8004274D20d2s2
/dev/dsk/c2t50060E8004274D20d0s3 /dev/dsk/c2t50060E8004274D20d2s3
/dev/dsk/c2t50060E8004274D20d0s4 /dev/dsk/c2t50060E8004274D20d2s4
/dev/dsk/c2t50060E8004274D20d0s5 /dev/dsk/c2t50060E8004274D20d2s5
/dev/dsk/c2t50060E8004274D20d0s6 /dev/dsk/c2t50060E8004274D20d2s6
/dev/dsk/c2t50060E8004274D20d0s7 /dev/dsk/c2t50060E8004274D20d2s7

 

Syndicate

Syndicate content