cannot create filesystem using cluster file system

[Hank Huggins]

I have configured Storage Foundation Cluster file system 5.0 on HPUX 11.23 with 2 nodes in the cluster.
# cfscluster status

Node : sddbblng
Cluster Manager : running
CVM state : running
No mount point registered with cluster configuration


Node : sddplalt
Cluster Manager : running
CVM state : running
No mount point registered with cluster configuration

I have created a shared disk group and volume:
# vxdg list
NAME STATE ID
sdddg enabled,shared,cds 1175711630.15.sddplalt
# vxprint
Disk group: sdddg

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0
dg sdddg sdddg - - - - - -

dm c6t0d0 c6t0d0 - 104820608 - - - -

v vol1 fsgen ENABLED 52428800 - ACTIVE - -
pl vol1-01 vol1 ENABLED 52428800 - ACTIVE - -
sd c6t0d0-01 vol1-01 ENABLED 52428800 0 - - -

when I try to make the filesystem I get this error:
# mkfs -F vxfs /dev/vx/rdsk/sdddg/vol1 50G
UX:vxfs mkfs: ERROR: V-3-20003: Cannot open /dev/vx/rdsk/sdddg/vol1: No such device or address
# ls -al /dev/vx/rdsk/sdddg
total 0
drwxr-xr-x 2 root root 96 Apr 5 19:09 .
drwxr-xr-x 3 root root 96 Apr 5 19:09 ..
crw------- 1 root root 68 0x002ee0 Apr 5 19:09 vol1

Any ideas what I am doing wrong?

HH

[Jason Ray]

Whenever I create volumes and such for CFS filesystems, I've been creating the volume and filesystem in VEA but choosing not to mount it.

That seems to work for me. I would remove the volume and recreate it in VEA but don't mount and see if that works.

Then I just use the:
cfsmntadm add /mount/point all=rw

[Steven E. Protter]

Shalom,

You say you are using the Foundation Cluster File System.

Yet your mkfs command is vxfs.

This confuses me. Is this a new file system or just a means to use vxfs in shared mode, which I'm not aware of.

Normally I use newfs to do a filesystem, its a pretty wrapper for mkfs.

What documentation are you using here? Links please.

SEP

[Hank Huggins]

Using newfs does not seem to make much difference:
# newfs -F vxfs /dev/vx/rdsk/sdddg/vol1
UX:vxfs newfs: ERROR: V-3-20113: Cannot open /dev/vx/rdsk/sdddg/vol1: No such device or address

[Sameer_Nirmal]

I suspect the diskgroup configuration problem which is related to the sub-disk underneath it.
See line
sd c6t0d0-01 vol1-01 ENABLED 52428800 0 ---
I would expect the DG name for "c6t0d0-01 " instead.

I would re-create the DG configuration from the scratch. I would follow the steps given the I&C manual. If you have the manual, there is a script shown for creating the DG , Shared volume and CFS.

[Hank Huggins]

I think I see what you are saying about the DG. I recreated it and got the same error when running newfs. I noticed the subdisk does not show as active in the vxprint output:
# vxprint
Disk group: sdddg

TY NAME ASSOC KSTATE LENGTH PLOFFS STATE TUTIL0 PUTIL0
dg sdddg sdddg - - - - - -

dm c6t0d0 c6t0d0 - 104820608 - - - -

v sddshare fsgen ENABLED 52428800 - ACTIVE - -
pl sddshare-01 sddshare ENABLED 52428800 - ACTIVE - -
sd c6t0d0-01 sddshare-01 ENABLED 52428800 0 - - -

I am using an EMC CX3-20 array for my storage. Here the ioscan output for the "disk":
disk 5 0/4/1/0.1.5.0.0.0.0 sdisk CLAIMED DEVICE DGC CX3-20WDR0
/dev/dsk/c6t0d0 /dev/rdsk/c6t0d0

[Sameer_Nirmal]

In vxprint there won't be "ACTIVE" status for sd but KSTATE should ENABLED. You can check if the sd is associated with the EMC disk using
# vxprint -htr

What do you see in
# vxdg list
# cfsdgadm display -v

I guess there is DMP enabled. Have you installed the recommended patches for SFC?

[Hank Huggins]

I just got off the phone with Symantec support and they found I had the local-activation mode on the DG set incorrectly. I think you were heading me in the right direction. I set the local-activation to shared-write and the cluster-actv-mode to sw for both nodes and I can create the filesystem.

Thanks for your help.

[Joshua Laden]

I have found these cheat sheets to be very helpful. This was written by one the senior SG gurus.