AIX Raw Logical Volume issue

[Cyclone]

Hello all,

I am doing a DR test and I'm restoring my TSM server on new hardware. Everything has been going well but I have reached the section of the DRM where I am recreating my disk pool. My disk pool is pretty simple. Its a 1000GB pool that is made of 10x 100GB raw logical volumes. The DRM Plan in section "PRIMARY.VOLUMES.REPLACEMENT" tells me to create new 'unique' named volumes. Ok, no problem. I created 10 new logical volumes with the exact specs as the original ones, but I just appended a 'b' on the end of each lv name. So 'lv100' is named 'lv100b' on the new server.

I see the old volumes listed in the restored TSM db. So I updated them as 'Destroyed' and then tried to create the 10 new volumes to add to the disk storage pool. But I receive the following error message.

tsm: RPS1500>def vol RPS.DISK1 '/dev/rlv100b' acc=READW f=104736 wait=yes
ANR2404E DEFINE VOLUME: Volume /dev/rlv100b is not available.
ANS8001I Return code 14.

tsm: RPS1500>def volume RPS.DISK1 /dev/rlv100b acc=READW f=104736 wait=yes
Session established with server RPS1500: AIX
  Server Version 6, Release 2, Level 2.2
  Server date/time: 07/26/12   17:38:58  Last access: 07/26/12   17:20:38

ANR2404E DEFINE VOLUME: Volume /dev/rlv100b is not available.
ANS8001I Return code 14.

I checked in AIX and everything looks correct.

$ lsvg -l tsmstg_vg
tsmstg_vg:
LV NAME             TYPE       LPs     PPs     PVs  LV STATE      MOUNT POINT
lv100b              raw        3273    3273    10   closed/syncd  N/A
lv101b              raw        3273    3273    10   closed/syncd  N/A
lv102b              raw        3273    3273    10   closed/syncd  N/A
lv103b              raw        3273    3273    10   closed/syncd  N/A
lv104b              raw        3273    3273    10   closed/syncd  N/A
lv105b              raw        3273    3273    10   closed/syncd  N/A
lv106b              raw        3273    3273    10   closed/syncd  N/A
lv107b              raw        3273    3273    10   closed/syncd  N/A
lv108b              raw        3273    3273    10   closed/syncd  N/A
lv109b              raw        3273    3273    10   closed/syncd  N/A

I check against the original system and everything looks the same (except that the LV State is open on the working system, but I assume that it because TSM is actively using the rlv there). I have tried a few variations on the "def vol" command, but It still hasn't worked. Does anything stick out? I also checked the permissions on the /dev/rlv100? files and they match up to the source system.

Thanks for responses.

Note: Here is a listing of the old volumes that have not yet been removed from the disk pool.

tsm: RPS1500>q vol devc=disk

Volume Name                  Storage         Device         Estimated       Pct      Volume
                             Pool Name       Class Name      Capacity      Util      Status
------------------------     -----------     ----------     ---------     -----     --------
/dev/rlv100                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv101                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv102                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv103                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv104                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv105                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv106                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv107                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv108                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line
/dev/rlv109                  RPS.DISK1       DISK               0.0 M       0.0     Off-Line

 

[moon-buddy]

On the original installation, do you have the same RAW disks?

I was made to understand that disks presented to TSM should be 'cooked' disks.

 

[Cyclone]

Yes, the source system is using 'raw' logical volumes and has been doing so for well over a year and a half.

 

[Mikey D]

Did you change the ownership of the RAW logical volumes to match the user ID and group that's running the TSM instance?

 

[Cyclone]

Yes, I did check, they are both the same

Source:

tsm2:/ #ls -l /dev/rlv*
crw-rw----    1 root     system       45,  1 Jul 30 14:19 /dev/rlv100
crw-rw----    1 root     system       45,  2 Jul 30 14:19 /dev/rlv101
crw-rw----    1 root     system       45,  3 Jul 30 14:20 /dev/rlv102
crw-rw----    1 root     system       45,  4 Jul 30 14:21 /dev/rlv103
crw-rw----    1 root     system       45,  5 Jul 30 14:21 /dev/rlv104
crw-rw----    1 root     system       45,  6 Jul 30 14:21 /dev/rlv105
crw-rw----    1 root     system       45,  7 Jul 30 14:21 /dev/rlv106
crw-rw----    1 root     system       45,  8 Jul 30 14:27 /dev/rlv107
crw-rw----    1 root     system       45,  9 Jul 30 14:17 /dev/rlv108
crw-rw----    1 root     system       45, 10 Jul 30 14:19 /dev/rlv109

DR Host:

$ ls -l /dev/rlv*
crw-rw----    1 root     system       45,  1 Jul 26 11:56 /dev/rlv100b
crw-rw----    1 root     system       45,  2 Jul 26 11:56 /dev/rlv101b
crw-rw----    1 root     system       45,  3 Jul 26 11:56 /dev/rlv102b
crw-rw----    1 root     system       45,  4 Jul 26 11:56 /dev/rlv103b
crw-rw----    1 root     system       45,  5 Jul 26 11:56 /dev/rlv104b
crw-rw----    1 root     system       45,  6 Jul 26 11:56 /dev/rlv105b
crw-rw----    1 root     system       45,  7 Jul 26 11:56 /dev/rlv106b
crw-rw----    1 root     system       45,  8 Jul 26 11:56 /dev/rlv107b
crw-rw----    1 root     system       45,  9 Jul 26 11:56 /dev/rlv108b
crw-rw----    1 root     system       45, 10 Jul 26 11:56 /dev/rlv109b

Does TSM have to format the Raw LVs before they can be used?

 

[moon-buddy]

Yes, I did check, they are both the same

Source:

tsm2:/ #ls -l /dev/rlv*
crw-rw----    1 root     system       45,  1 Jul 30 14:19 /dev/rlv100
crw-rw----    1 root     system       45,  2 Jul 30 14:19 /dev/rlv101
crw-rw----    1 root     system       45,  3 Jul 30 14:20 /dev/rlv102
crw-rw----    1 root     system       45,  4 Jul 30 14:21 /dev/rlv103
crw-rw----    1 root     system       45,  5 Jul 30 14:21 /dev/rlv104
crw-rw----    1 root     system       45,  6 Jul 30 14:21 /dev/rlv105
crw-rw----    1 root     system       45,  7 Jul 30 14:21 /dev/rlv106
crw-rw----    1 root     system       45,  8 Jul 30 14:27 /dev/rlv107
crw-rw----    1 root     system       45,  9 Jul 30 14:17 /dev/rlv108
crw-rw----    1 root     system       45, 10 Jul 30 14:19 /dev/rlv109

DR Host:

$ ls -l /dev/rlv*
crw-rw----    1 root     system       45,  1 Jul 26 11:56 /dev/rlv100b
crw-rw----    1 root     system       45,  2 Jul 26 11:56 /dev/rlv101b
crw-rw----    1 root     system       45,  3 Jul 26 11:56 /dev/rlv102b
crw-rw----    1 root     system       45,  4 Jul 26 11:56 /dev/rlv103b
crw-rw----    1 root     system       45,  5 Jul 26 11:56 /dev/rlv104b
crw-rw----    1 root     system       45,  6 Jul 26 11:56 /dev/rlv105b
crw-rw----    1 root     system       45,  7 Jul 26 11:56 /dev/rlv106b
crw-rw----    1 root     system       45,  8 Jul 26 11:56 /dev/rlv107b
crw-rw----    1 root     system       45,  9 Jul 26 11:56 /dev/rlv108b
crw-rw----    1 root     system       45, 10 Jul 26 11:56 /dev/rlv109b

Does TSM have to format the Raw LVs before they can be used?

I believe it does - that is why I said 'cooked' volumes.

 

[Cyclone]

I guess that I thought that the "def vol" command was going to format the raw lv. I saw some references to a dsmfmt command, but I could not located it on my system. Not in server bin.

 

[chad_small]

You dont "format" raw logical volumes and using the DEFINE VOLUME with format size is wrong. You create the LV in the OS then DEFINE IT ONLY ..... DO NOT USE FORMATSIZE in the DEFINE and it will work. The silly thing about TSM is that when you create a storage pool volume "file" what it is really doing is creating a RAW LOGICAL VOLUME within its own filesystem handler. The beauty of a RAW LV is that you don't have to deal with all the hassle of creating the JFS filesystem and then formating the diskpool volume within it.

 

[Cyclone]

Thank for responding Chad. I did create the raw LVs using mklv in the OS. But for some reason I still can not add those LVs to my diskpool. I dropped the sizing off the "define volume" and the command still can't see the LVs. Any ideas on what I am doing wrong?

tsm: RPS1500>def vol RPS.DISK1 /dev/rlv100b access=readw wait=yes
ANR2404E DEFINE VOLUME: Volume /dev/rlv100b is not available.
ANS8001I Return code 14.

tsm: RPS1500>def vol RPS.DISK1 '/dev/rlv100b' access=readw wait=yes
ANR2404E DEFINE VOLUME: Volume /dev/rlv100b is not available.
ANS8001I Return code 14.

tsm: RPS1500>def vol RPS.DISK1 '/dev/lv100b' access=readw wait=yes
ANR2404E DEFINE VOLUME: Volume /dev/lv100b is not available.
ANS8001I Return code 14.

 

[moon-buddy]

Try a COOKED volume

To be honest, I never used RAW volumes for disk pools.

 

[rmazzon]

Mikey D asked earlier if the owner/group of the volumes was the same as the account running the dsmserv binary. You replied by listing out the volumes on both systems. So I think the question still remains, is TSM running under the root account?

 

[chad_small]

Yeah looks like a permission issue. With 6.1 not running under the root ID you have to set permissions on the LV accordingly.

 

[Cyclone]

I guess that this is where my inexperience with raw LVs come in. Do I just chmod tsmuser:tsmgroup /dev/rlv100* for these raw LVs, or is there somewhere else that I set permissions. The only reason that I ask, is that the permissions for the /dev/rlv* devices on the DR system do currently match what they are on the source server.

Thank you.

 

[moon-buddy]

I guess that this is where my inexperience with raw LVs come in. Do I just chmod tsmuser:tsmgroup /dev/rlv100* for these raw LVs, or is there somewhere else that I set permissions. The only reason that I ask, is that the permissions for the /dev/rlv* devices on the DR system do currently match what they are on the source server.

Thank you.

If indeed this is a permission issue, all you need to do is 'chown tsmuser:tsmgroup /dev/rlv100*'

 

[Cyclone]

Ok, thanks. I will try this out. Gotta love a DR system to play with.

 

[Cyclone]

Follow up. Sure enough, once I set the ownership of the raw LVs, I could add the volumes to the storage pool. I guess that on the source system that the server process is executed by root, so it has the authority to access those volumes on that server.

Thanks again to everyone who posted here.