provisioning VMs in stretched cluster Fails with the following error "0 witness disks available, needed at least 1 witness disks"
Article ID: 369486
Updated On:
Products
Issue/Introduction
When attempting to provision new objects on a vSAN Stretched Cluster it fails with error "0 witness disks available, needed at least 1 witness disks".
In clomd.log you see the following:
2024-06-05T17:37:31.277Z error clomd[7321785] [Originator@6876 opID=1804475830] CLOM_
2024-06-05T17:37:31.277Z error clomd[7321785] [Originator@6876 opID=1804475830] CLOMChangeObjectPolicy: Cluster doesn't have resources to satisfy this policy
2024-06-05T17:49:31.628Z error clomd[7321785] [Originator@6876 opID=1804475873] CLOM_
2024-06-05T17:49:31.628Z error clomd[7321785] [Originator@6876 opID=1804475873] CLOMChangeObjectPolicy: Cluster doesn't have resources to satisfy this policy
Cause
Stretch clusters require a witness appliance for the witness components to reside on. If the witness appliance is in a faulty state then vSAN can't layout the required components for an object to meet compliance.
Resolution
Check vSAN Skyline Health to ensure the cluster is in a healthy state. We shouldn't see any alerts pertaining to the witness appliance being network isolated or failed disks.
Prior to making any changes to the environment ensure you have current & valid backups, validate all current objects status is only in a reduced available state and no inaccessible objects exist. Do not make any changes to the environment if inaccessible objects exist as it could lead to potential data loss. If inaccessible objects exist in the environment open a case with VMware by Broadcom support.
If the witness appliance is network isolated follow KB vSAN Witness appliance partitioned from stretch cluster to get the witness back into the cluster.
If the witness appliance has failed disks follow the below steps:
1. Investigate the failed disk/disk-group by looking at the vmkernel.log, vobd.log, or vmkwarning.log to see why the disk is failed. You may even need to check the physical host the witness appliance resides to ensure there are no issues with the backing storage as the witness appliance is just a VM. If an issue is found with the underling storage please correct that before continuing.
2. If no issues are found with the backing storage reboot the witness appliance and verify if the disk-group comes back online.
3. If the disk-group does not comeback online after a reboot delete and recreate the witness disk-group.
Physical Witness
Treat a physical witness as a normal node with a disk-group issue.
Feedback
