[fredric]

Hi,

I have two p505 servers with one VIOS server on each server.
I'm trying to setup a HACMP cluster between two lpar's, one on each VIOS server. The lpar's are running AIX 5.3. Each physical server have one fc-hba.
A SVC-SAN disk is shared to both VIOS-servers. I can see the shared disk on both VIOS-servers. The disk is presented as hdisk2, hdisk3 and vpath0 (multi pathing).

I'm able to "dedicate"/map the disk on one of the VIOS-servers to a lpar(vhost). But when I'm trying to do the same onthe other VIOS-server, I got errors.
On VIOS-server #1
>mkvdev -vdev vpath0 -vadapter vhost1 -dev c1_vpath0
works fine, the disk is presented on the lpar/vhost.

On VIOS-server #2 (there are other lpars on the VIOS)
>mkvdev -vdev vpath0 -vadapter vhost3 -dev c1_vpath0

Some error messages may contain invalid information
for the Virtual I/O Server environment.
Method error (/usr/lib/methods/cfg_vt_scdisk):
0514-012 Cannot open a file or device.

If I removed the reservation on the disk, I'm able to "dedicate" the disk. But this will mess up the disk on the first lpar.

My question, how do I setup the shared disk on the both VIOS-servers?

Best Regards,
Fredric

[ross.mather]

In this case you'll need to set the persistent_reserve attribute of the underlying hdisks to no_reserve before you can allocate the disks to the LPARS.

Commands are:
chdev -dev hdisk2 -attr reserve_policy =no_reserve -perm
chdev -dev hdisk2 -attr reserve_policy =no_reserve

Then allocate the vpath to the appropriate LPARS.

I haven't done this myself so I guess you may need to do this on the vpath as well.

[fredric]

Hi,
thanks for the response.
I tried you suggestion (done it before as well) and got following errors.

$ chdev -dev hdisk2 -attr reserve_policy=no_reserve -perm

Some error messages may contain invalid information
for the Virtual I/O Server environment.
Method error (/etc/methods/chgdisk):
0514-017 The following attributes are not valid for the
specified device:
reserve_policy


got the same error msg on all hdisk/vpath (hdisk2,hdisk3,vpath0) and both VIOS.

My config:
VIOS #1
$ lspv
NAME PVID VG STATUS
hdisk0 0001c9ba5cf87178 rootvg active
hdisk1 0001c9ba5cf87445 rootvg active
hdisk2 none None
hdisk3 none None
vpath0 0001c9ba664a26f7 None


from oem_setup_env:
# lsvpcfg
vpath0 (Avail pv ) 600507680181008B3000000000000009 = hdisk2 (Avail ) hdisk3 (Avail )

VIOS #2
$ lspv
NAME PVID VG STATUS
hdisk0 0002980a27235dcb datavg active
hdisk1 0002980a8af63a8c rootvg active
hdisk2 none None
hdisk3 none None
vpath0 0001c9ba664a26f7 None
$

from oem_setup_env:
#lsvpcfg
vpath0 (Avail pv ) 600507680181008B3000000000000009 = hdisk2 (Avail ) hdisk3 (Avail )
#

//Fredric

[Administrator]

Out the virtual device (c1_vpath0) into a defined state first. You can't execute chdev on the physical device as long as the virtual device is available.

[fredric]

Thanks for the hint, but I'm still get the same error msg. for all disk devices (hdisk2, hdisk3 and vpath0)

$ lsdev|grep path
c1_vpath0 Defined Virtual Target Device - Disk
vpath0 Available Data Path Optimizer Pseudo Device Driver
$
$ lsdev|grep hdisk
hdisk0 Available 16 Bit LVD SCSI Disk Drive
hdisk1 Available 16 Bit LVD SCSI Disk Drive
hdisk2 Available SAN Volume Controller Device
hdisk3 Available SAN Volume Controller Device


$ chdev -dev hdisk2 -attr reserve_policy=no_reserve

Some error messages may contain invalid information
for the Virtual I/O Server environment.
Method error (/etc/methods/chgdisk):
0514-017 The following attributes are not valid for the
specified device:
reserve_policy

//Fredric

[ross.mather]

I'm thinking now that the reserve_policy attribute is only valid within MPIO. for SDD its probably another name with different functionality.

At the very least try: breaking all the connections of the disks to the LPARS at VIO level. Break the persistant reserve, and then try to set the reserve_policy,

[fredric]

Hi,

could the output from "lsdev -dev hdisk2 -attr" help you?

$ lsdev -dev hdisk2 -attr
attribute value description user_settable
clr_q no Device CLEARS its Queue on error True
location Location Label True
lun_id 0x0 Logical Unit Number ID False
max_transfer 0x40000 Maximum TRANSFER Size True
node_name 0x500507680100129c FC Node Name False
pvid none Physical volume identifier False
q_err yes Use QERR bit True
q_type simple Queuing TYPE True
qfull_dly 20 delay in seconds for SCSI TASK SET FULL True
queue_depth 20 Queue DEPTH True
rw_timeout 60 READ/WRITE time out value True
scbsy_dly 20 delay in seconds for SCSI BUSY True
scsi_id 0x10000 SCSI ID False
start_timeout 180 START unit time out value True
ww_name 0x500507680130129c FC World Wide Name False

and the same for vpath0
______________________________________________
$ lsdev -dev vpath0 -attr
attribute value description user_settable
active_hdisk hdisk3/600507680181008B3000000000000009/fscsi0 Active hdisk False
active_hdisk hdisk2/600507680181008B3000000000000009/fscsi0 Active hdisk False
policy df Scheduling Policy True
pvid 0001c9ba664a26f70000000000000000 Physical volume identifier False
qdepth_enable yes Queue Depth Control True
serial_number 600507680181008B3000000000000009 LUN serial number False
unique_id 33213600507680181008B300000000000000904214503IBMfc p Device Unique Identification False


//Fredric

[fredric]

Solved!

Installed SDDPCM instead of SDD!

The HACMP cluster is now up and running!

Thanks!

//Fredric