Browse Prior Art Database

Method of notifying failure LPAR information to other LPARs in cluster through Hypervisor Disclosure Number: IPCOM000249515D
Publication Date: 2017-Mar-02
Document File: 3 page(s) / 59K

Publishing Venue

The Prior Art Database


Disclosed is the method of notifying the Failure LPAR (Logical Partition) Information to other LPAR in a Clustered environment via Hypervisor in which the Failed LPAR will inform the hypervisor via a system h-call with its network information to enable its hypervisor acquire the failed lpar network details and send notification messages to all other LPARS within the cluster about the Failed LPAR status information using a packet exhange information table maintained within hypervisor. This prevents the cluster LPAR resources being outage and thus helps in other LPARs in the cluster to take over the cluster control and resources to maintain the cluster active.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 53% of the total text.


Method of notifying failure LPAR information to other LPARs in cluster through Hypervisor

In cluster environment, nodes (LPARs) comes from different hypervisors. When a node crashes which is holding the Tie breaker disk (storage lock), it causes the cluster outage. In this situation, the other node in the cluster should get notified about the Tie-breaker node failure. So that other nodes can preempt the Tie breaker disc and cluster can be active state.

In cluster environment, all the nodes in cluster has communication between other LPARs. When a LPAR is holding a Tie-breaker disc, it acquires storage lock and maintains the application resources. The application resources in cluster should be acquired by other LPAR when Tie-breaker LPAR is not reachable/crashes. So that applications will be available 24/7.

If there is no communication between Tie-breaker LPAR and other LPARs in the cluster, the resources will not be acquired and cluster will be on outage state. This causes application to go down. When Tie-breaker LPAR crashes, it should send a notification message to other LPARs in the cluster to acquire the resources.

The notification mechanism can be implemented with the help of Hypervisor.

1) When LPAR crashes, it makes a H-CALL to hypervisor with LPAR’s network details to acquire the LPAR’s Network.

2) Hypervisor acquires LPAR’s network and send a notification message to all the LPARs in cluster in following manner.

a. Hypervisor maintains the information of LPAR’s which are there in the cluster in form of table with IP address dynamically.

b. Once H-call is validated by Hypervisor, the table will be updated with IP address of the crashed LPAR (Tie-breaker) to send notifications to other nodes in the cluster.

c. Hypervisor acquires the IP address and snoops ICMP packets which are received to the LPAR.

d. Hypervisor creates failure notification information to provide response to incoming ICMP packets.

e. Hypervisor maintains packet exchange information between failure node and other nodes in cluster in form of a table dynamically.

f. The packet exchange information table will be updated as soon as the information send to other nodes....