Originally posted by: hebau

We have a 2-Node-HACMP-Cluster with 2 IBM Storage Units DS4300.
Each Cluster-Node has 2 disks hdisk0 and hdisk1 (rootvg, mirrored).
DS4300-A contains the logical drives hdisk2, hdisk3 and hdisk4
DS4300-B contains the logical drives hdisk5, hdisk6 and hdisk7
and
hdisk2 and hdisk5 are mirrored by LVM and build VolumeGroup vg01
hdisk3 and hdisk6 are mirrored by LVM and build VolumeGroup vg02
hdisk4 and hdisk7 are mirrored by LVM and build VolumeGroup vg03.
The DS4300 Units are connected to the cluster-Nodes via
disk-array-controller, array-router and fibre channel adapter.
We have the following problem:
All disks are available on both nodes except hdisk7, which is not available on Node1.
When the VolumeGroup vg03 is activated on Node1, then the mirror (hdisk4, hdisk7) burst.

On Node1 it looks like
root@:~> lsdev -C|grep disk
hdisk0 Available 09-08-00-5,0 16 Bit LVD SCSI Disk Drive
hdisk1 Available 09-08-00-8,0 16 Bit LVD SCSI Disk Drive
hdisk2 Available 05-08-02 1722-600 (600) Disk Array Device
hdisk3 Available 04-08-02 1722-600 (600) Disk Array Device
hdisk4 Available 05-08-02 1722-600 (600) Disk Array Device
hdisk5 Available 05-08-02 1722-600 (600) Disk Array Device
hdisk6 Available 04-08-02 1722-600 (600) Disk Array Device
hdisk7 Defined 05-08-02 2102-F10 Disk Array Device

Shows disk array router "dar" :
root@:~> lsdev -C|grep dar
dar0 Available 1722-600 (600) Disk Array Router
dar1 Available 1722-600 (600) Disk Array Router

Shows disk array controller "dac" :
root@:~> lsdev -C|grep dac
dac0 Available 04-08-02 1722-600 (600) Disk Array Controller
dac1 Available 04-08-02 1722-600 (600) Disk Array Controller
dac2 Available 05-08-02 1722-600 (600) Disk Array Controller
dac3 Available 05-08-02 1722-600 (600) Disk Array Controller

Shows fibre channel adapter "fcs"
root@:~> lsdev -C|grep fcs
fcs0 Available 04-08 FC Adapter
fcs1 Available 05-08 FC Adapter.
The cabling of the disks is correct, but the configuration manager cfgmgr could not switch the state of hdisk7 to available.
How can we get hdisk7 available ?

Thanks in advance
#AIX-Forum

Originally posted by: Juredd1

There are other ways I am sure but if it were mine server I would try

mkdev -l hdisk7

and if that doesn't make it available I would do the follow;

rmdev -dl hdisk7

Then run

cfgmgr

If the machine can see it this should bring it back to an available state and I would expect it to still be hdisk7.

Justin
#AIX-Forum

Originally posted by: hebau

Thanks for your reply.
That was also my first idea. But I am afraid that I have to reconfigure the VolumeGroup vg03 within the cluster configuration, when I remove hdisk7 from the system first and add it to the system afterwards.
Do you think that the reconfiguration is not necessary?
#AIX-Forum

Originally posted by: Juredd1

Unless you have something configured within HACMP that is tied directly to the disk such as disk heartbeat or something. Even then if it comes back with the same device name I don't think that would be a problem.

So I do not think a reconfiguration of the VG in the cluster will be required if you rmdev the device and run cfgmgr to get it back.

It's been a while since I have configured HA but that is my thoughts.
#AIX-Forum