Vxfencing fails to start with error "ERROR V-11-2-4045 At least three coordination points must be defined" when configuring CP servers on cloned VMs

book

Article ID: 100042354

calendar_today

Updated On:

Description

Error Message

When setting up server-based I/O fencing on VCS using the installer and choosing option 1:

# /opt/VRTS/install/installer -fencing

Fencing configuration
     1)  Configure Coordination Point client based fencing
     2)  Configure disk based fencing
     3)  Configure majority based fencing

at the end of the configuration, vxfen fails to start with the following message:

Starting Fencing on node01  ............................................................ Failed
Could not start VxFEN on node01

The following error is logged in /var/VRTSvcs/log/vxfence.log:

VXFEN vxfenconfig ERROR V-11-2-4045 At least three coordination points must be defined

 

Cause

If the VMs used as CP servers have been deployed by cloning them from a source VM, they will have the same UUID of the source.
Vxfen needs the CP servers to have different unique UUID. Hosts with the same UUID will be recognised as the same host, hence vxfen will complain about not reaching the minimum number of CP servers configured.
The UUID is stored in file /etc/VRTScps/db/current/cps_uuid on each CP server.

Example:

Two of the following three CP servers have the same UUID, so vxfen will recognize only two CP servers available:

[cpserver01]# cat /etc/VRTScps/db/current/cps_uuid       
{baf51372-5b0e-11e6-a982-59613f77f06a}
[cpserver02]# cat /etc/VRTScps/db/current/cps_uuid        
{baf51372-5b0e-11e6-a982-59613f77f06a}
[cpserver03]# cat /etc/VRTScps/db/current/cps_uuid       
{d664067a-23a0-11e8-a033-d8917947e812}

 

Resolution

- Remove the contents of the DB directory on the cloned CP server, which is located  in /etc/VRTScps/db/
- Remove file /etc/vxcps.conf on the cloned CP server
- Reconfigure the CP server using the installer, or manually (please refer to the VCS install and configuration guide)