LUND to be present for 2 nodes Cluster

[Syedfurquan]

Hello Expert,

I need to implement 2 nodes cluster on hp-ux 11.31 OS. We have currently running HP Serviceguard cluster in the production and need to prepare the same setup on new server as well.

I have confusion on the LUN need to be created. When I see cmviewcl it shows me 2 package running 1 on each node respectively.

So for the new server I need to create two LUN's and shared on both the server or I can create only 1 LUN?

some of the O/P from curent production server as follows:

# cmviewcl -v

CLUSTER        STATUS
cluster1       up

  NODE           STATUS       STATE
  kcbsrv01       up           running

    Cluster_Lock_LVM:
    VOLUME_GROUP          PHYSICAL_VOLUME       STATUS
    /dev/vglock           /dev/disk/disk18      up

    Network_Parameters:
    INTERFACE    STATUS                     PATH                NAME
    PRIMARY      down (disabled) (IP only)  0/4/2/0             lan0
    PRIMARY      up                         0/4/2/1             lan1
    STANDBY      up                         0/7/1/0/6/0         lan2

    PACKAGE        STATUS           STATE            AUTO_RUN    NODE
    imalp1         up               running          enabled     kcbsrv01

      Policy_Parameters:
      POLICY_NAME     CONFIGURED_VALUE
      Failover        configured_node
      Failback        manual

      Script_Parameters:
      ITEM       STATUS   MAX_RESTARTS  RESTARTS NAME
      Service    up       Unlimited     0        imalp1

      Node_Switching_Parameters:
      NODE_TYPE    STATUS       SWITCHING    NAME
      Primary      up           enabled      kcbsrv01 (current)
      Alternate    up           enabled      kcbsvr

      Other_Attributes:
      ATTRIBUTE_NAME                          ATTRIBUTE_VALUE
      Style                                   legacy
      Priority                                no_priority

  NODE           STATUS       STATE
  kcbsvr         up           running

    Cluster_Lock_LVM:
    VOLUME_GROUP          PHYSICAL_VOLUME       STATUS
    /dev/vglock           /dev/disk/disk18      up

    Network_Parameters:
    INTERFACE    STATUS                     PATH                NAME
    PRIMARY      down (disabled) (IP only)  0/4/2/0             lan0
    PRIMARY      up                         0/4/2/1             lan1
    STANDBY      up                         0/7/1/0/6/0         lan2

    PACKAGE        STATUS           STATE            AUTO_RUN    NODE
    imalp2         up               running          enabled     kcbsvr

      Policy_Parameters:
      POLICY_NAME     CONFIGURED_VALUE
      Failover        configured_node
      Failback        manual

      Script_Parameters:
      ITEM       STATUS   MAX_RESTARTS  RESTARTS NAME
      Service    up       Unlimited     0        imalp2

      Node_Switching_Parameters:
      NODE_TYPE    STATUS       SWITCHING    NAME
      Primary      up           enabled      kcbsvr (current)
      Alternate    up           enabled      kcbsrv01

      Other_Attributes:
      ATTRIBUTE_NAME                          ATTRIBUTE_VALUE
      Style                                   legacy
      Priority                                no_priority
#


Also What will be the size of vglock disk and will it also be shared on both the server?

On NODE1:

/dev/sybvgha2/lvol1       10256384 4994228 4933277   50% /sybvgha2

on NODE2:

/dev/sybvgha1/lvol1        511180800 199585525 292120685   41% /sybvgha1

Thanks for any help in advance.

Regards,

[sujit kumar singh]

Hi,

the VG lock disk is already specified as lock PV as seen in the cmviewcl -v O/P shared.

What is the need to create another lock disk here?

Ask the SAN admin to allocate/mask the NEW LUN to both the nodes and get going.

If You need to change the configuration from Lock PV to a Lock LUN, Lock LUN as small as 100KB would be sufficient,

But Please make sure that lock PV or Lock LUN updates are done only after having good understanding of the cluster locking and changes to be made to lock PV/LUN, else this can cause the cluster to be fully unstable with Split Brain if one of the node fails when there is no functional cluster lock PV or LUN.

For cluster lock PV and LV, the PV that can be used may belong to an existing VG on the systems, where as the lock LUN being defined cant be used for LVM and VxVM as well as all data on lock LUN is destroyed when service guard creates that as a Lock LUN.

[Primesh Abeysinghe]

Hi,

If you want to create two package running in both nodes you need to create two LUN and add the same to the packages

also you may need configure volume groups as "shared VGs" and

lock LUN should be small like 100KB but in today storage systems you can not create such a small ones, so 1GB would be more practical.

lock LUN is know as "Quorum disk" need only one for cluster that single LUN has to presented to both nodes

hope this is good enough for you to start.

Regards,

Primesh