Operating System - HP-UX
1827435 Members
4825 Online
109965 Solutions
New Discussion

Re: Cluster Lock and Disk

 
SOLVED
Go to solution
sahir_1
Frequent Advisor

Cluster Lock and Disk

Could anybody please let me know what is Cluster Lock and its function?

What is cluster lock and cluster lock server?

Regards,
Sahir
14 REPLIES 14
Jeeshan
Honored Contributor

Re: Cluster Lock and Disk

The cluster lock disk (used only in HP-UX clusters) is a disk that can be written to by all members of the cluster. When a node obtains the cluster lock, this disk is marked so that other nodes will recognize the lock as â taken.â This mark will survive an off-on power cycle of the disk device unlike SCSI disk reservations. A lock disk may be used for clusters of up to four nodes.

Cluster lock server uses the same thing, but it's uses only when you have more than 2 node cluster. It's called Quorum server.
a warrior never quits
Michael Steele_2
Honored Contributor

Re: Cluster Lock and Disk

Hi

A cluster lock disk if a boolean flag that gets set to 'on' when the first node up in a cluster grags or forms the cluster first. Its a race contention preventor. If this flag was not set then the 2nd node would form a 2nd cluster, thereby creating what is known as 'split brain', or, two nodes both listed as primary nodes and both trying to control the cluster.
Support Fatherhood - Stop Family Law
melvyn burnard
Honored Contributor

Re: Cluster Lock and Disk

Take a read of this document, it should give you a good overview:
http://docs.hp.com/en/B3936-90078/B3936-90078.pdf
My house is the bank's, my money the wife's, But my opinions belong to me, not HP!
AVV
Super Advisor

Re: Cluster Lock and Disk

Hi,

Lock disk is the disk uses where there is a chance to split the nodes in equal size (eg. 2 node cluster, one fails to communicate with other and both of them tries to form a cluster). To avoid the system to be in equal polling situation, the one which gets the lock disk first will form the cluster and the other will perform a TOC and tries to join the existing cluster when it comes up.

Am I giving the correct explaination ?
Michael Steele_2
Honored Contributor

Re: Cluster Lock and Disk

"...and the other will perform a TOC and tries to join the existing cluster when it comes up..."

When a box TOC's it dumps its core to crash. This often takes 20 minutes. Are you sure you want to say this?
Support Fatherhood - Stop Family Law
Johnson Punniyalingam
Honored Contributor
Solution

Re: Cluster Lock and Disk

Cluster lock disk

The Cluster lock is a disk located in a volume group shared by all the nodes in the cluster . The cluster lock volume group and the pshyical volumer are identified in the Cluster configuration file,

The cluster lock is used as a tiebraker only for the situation in which the running cluster fails and atempt to refrom the cluster, the cluster is split init two subcluster of equal size , Each sub cluster will attempt to acquire the cluster lock.

the sub cksuter that gets the cluster lock will form the cluster , The subcluster that gets the cluster lock qill form the new cluster , The Sub cluster that does not get the cluster lock will panic in order to allow existing application to failover to

the new cluster which has been formed

Hope This Helps. :)

Problems are common to all, but attitude makes the difference
Michael Steele_2
Honored Contributor

Re: Cluster Lock and Disk

Again, not panic, not TOC, not reboot.

During a panic, or TOC, or reboot MC/SG initiates a failover to another node.

This does not happen during cluster formation. The losing node only gives up.

It does not reboot. It does not panic. It does not TOC.

You two learn these impressive buzz words and through them around like dollar bills in a stripper's G string.
Support Fatherhood - Stop Family Law
melvyn burnard
Honored Contributor

Re: Cluster Lock and Disk

@Michael
Actually, If a cluster experiences a situation that would result in a 50/50 split, then a tie-breaker must be used to prevent "split-brain" syndrome.
If there is a tie-breaker method, be it cluster lock disk/LockLun/Quorum Server/arbitration node, then one half of th ecluster will gain the "lock", and continue, the nodes in the "half" that do NOT get the lock WILL TOC. This is to enforce a package takeover by the surviving nodes.

This has been standard operating procedure since the inception of Serviceguard.
My house is the bank's, my money the wife's, But my opinions belong to me, not HP!
Michael Steele_2
Honored Contributor

Re: Cluster Lock and Disk

Melvin

What's the difference between a TOC and a RS? The crash dump initiated by the CMCLD daemon.

You can not find a citation or reference of a 20 minute crash dump during a cluster formation.

No where in the MC/SG admin guide does TOC come up except when communication has been lost and then the cmcld initates a TOC
Support Fatherhood - Stop Family Law
sahir_1
Frequent Advisor

Re: Cluster Lock and Disk

Thanks to all for yours great responce.

How can we plan the configuration of a cluster Disk?
Which is the confguration file?
What is default configuration incase I am not setting any specific Disk.?
Johnson Punniyalingam
Honored Contributor

Re: Cluster Lock and Disk

Cluster lock is obligatory for 2-node cluster only. If a cluster has more than 2 nodes, lock disk is optional.

In your scenario, if one of the nodes fail or it is removed for maintenance - it will be better to have a lock disk or quorum server.


All information of the Cluster Lock disk will be provided in cluster configuration ascii file you will get the Cluster Lock Disk information

Example :-

vi /etc/cmcluster/cmclconfig.ascii
FIRST_CLUSTER_LOCK_VG /dev/vgcllock
FIRST_CLUSTER_LOCK_PV /dev/dsk/c2t0d1

For more Information please read the below documents

http://docs.hp.com/en/B3936-90117/ch03s02.html#babbgfdh

Hope all your Question will be answered by reading above documents
Problems are common to all, but attitude makes the difference
Torsten.
Acclaimed Contributor

Re: Cluster Lock and Disk

Regarding the TOC operation (from SG manual):

The most dramatic response to a failure in a Serviceguard cluster is an HP-UX TOC or
INIT, which is a system reset without a graceful shutdown (normally referred to in
this manual simply as a system reset). This allows packages to move quickly to another
node, protecting the integrity of the data.
A system reset occurs if a cluster node cannot communicate with the majority of cluster
members for the predetermined time, or under other circumstances such as a kernel
hang or failure of the cluster daemon (cmcld).

Hope this helps!
Regards
Torsten.

__________________________________________________
There are only 10 types of people in the world -
those who understand binary, and those who don't.

__________________________________________________
No support by private messages. Please ask the forum!

If you feel this was helpful please click the KUDOS! thumb below!   
sahir_1
Frequent Advisor

Re: Cluster Lock and Disk

Thanks all..

Thanks johnson for the document.


I am closing this..

Thanks & Regards,
sahir
sahir_1
Frequent Advisor

Re: Cluster Lock and Disk

Thank you all...