05-05-2016 06:19 AM
Environment
Cluster nodes = 2
SFHA version = 6.1.1
OS = RedHat 5.9
Query
In case of a two node cluster, node2 (passive node) is already down and service group is online on node1 (active node), due to some reason node1 (active node) got rebooted. After sometime the active node restarted back successfully but the service does not came online. However we need to do a manual intervention i.e need to run a command gabconfig -c -x. After triggering the gabconfig -c -x command, the service group started coming online the resources. I want that resources should come online if the node is restarted back without manual intervention.
05-09-2016 02:50 AM
The requirement for gab seeding is for data protection.
How should VCS know that node 2 is really down and not just 'network partitioned' ?
VCS startup and data protection through seeding is well-documented in VCS Admin Guide.
05-09-2016 02:52 AM
Hi,
Manual run gabconfig -c -x is by design if one node down, another node restart and back.
since gab need confrim all member there, otherwise, split brain may occur.
05-17-2016 04:43 AM
Thanks guys for your kind response. My problem is resolved after changing in the /etc/gabtab file
gabconfig -c -n1
05-17-2016 06:15 AM
NO!!!! You should NOT do that unless the 2nd node is permanently down and you have not intention of fixing it.
If so - perform the necessary steps to remove the node from the cluster config and turn it into a one-node cluster.
As explained - the total number of nodes in the cluster should be up and seeded before VCS will start up.
This is by design to prevent split brain.
I hope you are taking 100% responsibility if split brain occurs as a result of your actions.....
05-17-2016 10:36 PM
Thanks Marianne for your words
I can understand your concern. We have configured 2 heartbeats and 1 low priority heartbeat on public network (public NIC is a bounded/teamed NIC) , so the failure chances are 1 in million for split brain.