help in cluster startup

[rajesh73]

Hi,

Plesae find the below steps

Procedure-1

*************

1.db001#cmhaltpkg 

2.db001#cmhaltnode db001

3.db001#cmrunnode db001

4.db001#cmrunpkg -n db001 apps1

procedure-2

************

1.db001#cmhaltpkg apps1

2.db001#cmhaltcl

3.db001#cmruncl

4.db001#cmrunpkg -n db001 apps1

hi form the above two procedure , please confirm which procedure is correct

my conclustion the above two procedure perform same action. please confirm

P.S. This thread has been moved from  HP-UX>General to Serviceguard. -HP Forum Moderator

[Patrick Wallek]

Procedure 1:

Step 1 -- This halts the package, though you normally would specify the package name to halt.

Step 2 -- The halts cluster services on the NODE db001.  This does not affect any other nodes in the cluster.

Step 3 -- This starts cluster services on the NODE db001.  This does not affect any other nodes in the cluster.

Step 4 -- This starts the package apps1 on the node db001.

Procedure 2:

Step 1 -- This halts the package apps1.

Step 2 -- This halts the ENTIRE cluster.  This will stop cluster services on ALL NODES in the cluster.

Step 3 -- This starts cluster services on ALL NODES in the cluster.

Step 4 -- This starts the package apps1 on node db001.

To summarize -- Procedure 1 will only affect node db001 (assuming that is where the package apps1 is running).  Procedure 2 will affect ALL NODES IN THE CLUSTER.

cmhaltcl will fail if there are still packages running in the cluster, unless you specify the '-f' option to force the cluster to shut down after halting all packages.

[rajesh73]

Thanks patrick

so we will follow the step1 , i thing this is the safe procedure

[rajesh73]

hi

one more doubt i need to install the kernel patch in 2 node cluster so which procedure is best for this activity.

[Patrick Wallek]

If you are going to install the patch on both nodes, I go ahead and shut down cluster services on both nodes in the cluster with the cmhaltcl command.

Then, once you reboot both nodes the cluster should restart automatically.

[rajesh73]

---If you are going to install the patch on both nodes, I go ahead and shut down cluster services on both nodes in the cluster with the cmhaltcl command.

we run this command in node1the cluster will stop two nodes or only one node.

command ---cmhaltcl

we run this command in node1 the cluser will start two nodes or only one node.

command - cmruncl

[Patrick Wallek]

>>we run this command in node1the cluster will stop two nodes or only one node.

>>command ---cmhaltcl

>>we run this command in node1 the cluser will start two nodes or only one node.

>>command - cmruncl

I thought I answered that above.

# man cmhaltcl

<snip>

DESCRIPTION
cmhaltcl causes all nodes in a configured cluster to stop their cluster daemons, optionally either halting all packages or applications in the process or detaching all modular style failover packages and Multi Node packages that are running in the cluster.

# man cmruncl

<snip>

DESCRIPTION
cmruncl causes all nodes in a configured cluster or all nodes specified to start their cluster daemons and form a new cluster.

[rkjha_hp]

"cmhaltcl" will stop the cluster on all nodes in cluster.

If you want to do patch/kernel change in one node only, just stop cluster on that particular node using "cmhaltnode".

All your package will be running on avavilable node during this and you can make change in this cluster stopped node.

rgds,