Configuring DPCOM disk

Edit online

In shared storage pool (SSP), the IBM Power servers are used for storage virtualization. The SSP Cluster uses network interfaces for communication between nodes. The network nodes might go down during network maintenance, switch maintenance, or server maintenance. In such situations, the communication breaks between clients and nodes. To troubleshoot this problem, you can use the Disk Ping Communication (DPCOM) technology to keep the clients running and applications available during network interruptions.

You can define a DPCOM disk in the SSP cluster. The metadata is copied from repository disk to the DPCOM disk. When the repository disk is not available or all other network interfaces are down, the node starts communication using the DPCOM disk. After the network interfaces are available, the node stops using the DPCOM interface and starts communication through the regular network and node. A node is in degraded mode when it is using disk communication for the primary interface.

Using a DPCOM cluster the in an SSP cluster

You need an additional disk that is shared with all Virtual I/O Servers. You can define the additional disk as a DPCOM disk. To view the additional disk that is shared with all Virtual I/O Servers, run the following command:

ksysmgr query viodisk vios=callalilyv1,callalilyv2,tomatov5,tomatov6

An output that is similar to the following screen is displayed:

DiskName        Size               VioDiskID
-------------------------------------------------------------------------------------------------
hdisk7           32822            VioDiskID 1

Adding a DPCOM disk to an existing cluster

Complete the following procedure to add a DPCOM disk to an existing cluster in the Virtual I/O Server.

To login to the Virtual I/O Sever, run the following command:
```
ssh tomatov5
```

To check free disks in the Virtual I/O Sever, run the following command:

lspv

An output that is similar to the following screen is displayed:

NAME             PVID                       VG               STATUS
hdisk0           PVID 1                     rootvg           active
hdisk5           PVID 2                     None
hdisk6           PVID 3                     caavg_private    active
hdisk7           PVID 4                     None

To add a DPCOM disk to the cluster, run the following command:
```
cluster -addcompvs hdisk7
```
An output that is similar to the following screen is displayed:
```
Add com disk(s) successful.
```

To view the DPCOM disk that is added to the cluster, run the following command:

lspv

An output that is similar to the following screen is displayed:

NAME             PVID                       VG               STATUS
hdisk0           PVID 1                     rootvg           active
hdisk5           PVID 2                     None
hdisk6           PVID 3                     caavg_private    active
hdisk7           PVID 4                     comdisk_vg       active

The hdisk7 (comdisk_vg) is added as DPCOM disk.

Verifying the DPCOM disk connection

Choose one of the following options to verify the DPCOM connection status.

If all Virtual I/O Servers are running and the network connection is working, the DPCOM connection will be in the Inactive or Standby state. To view the DPCOM disk connection state, run the following command:

pooladm dump cnxs

An output that is similar to the following screen is displayed:

ClusterName=ClusterName1  MyNodeName=Node_Name1:

CONNECTIONS WITH SERVERS:

Server: name=<>ServerName> nodeId=<NodeID> lease state=VALID
        lease phase=FIRST ip=<>IP (LEADER)
 Net     [0]: Lease state=VALID phase=FIRST cnx=ACTIVE IPAddr=<IPAddress> (PrimaryIP)
             Lease renew=<Lease_renew> LRSent=4 LRAcked=4
             MsgSent=<MsgSent> MsgRcvd=<MsgRcvd> avgRespTime: 0 sec 129079 nsec
             avgDeliverTime: 0 sec 100271 nsec maxDeliverTime: 0 sec 25668188 nsec
             ConnSwitchMsg sent=0 ackd=0 markDlvrd=0 lastAckedTxnIdByServer=0 CSServerNo=0
             cnxCheckPending=0
 DiskCom [1]: Lease state=VALID phase=FIRST cnx=STANDBY
             Lease renew=<Lease_renew> LRSent=<LRSent> LRAcked=<LRSent>
             MsgSent=0 MsgRcvd=310 avgRespTime: 0 sec 0 nsec
             avgDeliverTime: 0 sec 0 nsec maxDeliverTime: 0 sec 0 nsec
             ConnSwitchMsg sent=0 ackd=0 markDlvrd=0 lastAckedTxnIdByServer=0 CSServerNo=0
             cnxCheckPending=0

If any of the Virtual I/O Servers is not running and the network connection is not working, the DPCOM connection becomes Active. To view the DPCOM connection state, run the following command:

pooladm dump cnxs

An output that is similar to the following screen is displayed:

ClusterName=KSYS_hm_env19_1  MyNodeName=callalilyv2.ausprv.stglabs.ibm.com:

CONNECTIONS WITH SERVERS:

Server: name=<ServerName> nodeId=<NodeID> lease state=VALID
        lease phase=FIRST ip=<IP> (LEADER)
 Net     [0]: Lease state=EXPIRED phase=SECOND cnx=NOT_AVAILABLE IPAddr=<IPAddress> (PrimaryIP)
             Lease renew=266747 LRSent=13 LRAcked=0
             MsgSent=266747 MsgRcvd=533474 avgRespTime: 0 sec 171712 nsec
             avgDeliverTime: 0 sec 106413 nsec maxDeliverTime: 0 sec 20005887 nsec
             ConnSwitchMsg sent=0 ackd=0 markDlvrd=0 lastAckedTxnIdByServer=266125 CSServerNo=0
             cnxCheckPending=0
 DiskCom [1]: Lease state=VALID phase=FIRST cnx=ACTIVE
             Lease renew=847 LRSent=164 LRAcked=164
             MsgSent=683 MsgRcvd=1696 avgRespTime: 0 sec 890524 nsec
             avgDeliverTime: 0 sec 740053 nsec maxDeliverTime: 0 sec 30279064 nsec
             ConnSwitchMsg sent=1 ackd=1 markDlvrd=0 lastAckedTxnIdByServer=266125 CSServerNo=0
             cnxCheckPending=0

When the VIOS network is down, the DPCOM connection becomes active and the cluster internally communicates through the DPCOM disk.

Removing a DPCOM disk from the cluster

To remove a DPCOM disk from an existing SSP cluster, run the following command:

cluster -rmcompvs <hdisk#>