1 Reply Latest reply on Jul 24, 2020 8:13 AM by TheBobkin

    1 more standalone host is required. Some objects will become inaccessible or non-compliant with their storage policy

    InvisibleAdmin Lurker

      I had a 12-node stretch-cluster with two fault domains. This was an active/passive with no VMs running on the Failover domain. I have deleted the Failover Fault Domain (now its hosts show as "standalone) and now wish to decommission those hosts, but when I try to put them into Maintenance Mode (and choose full data migration), I get this: 1 more standalone host is required. Some objects will become inaccessible or non-compliant with their storage policy

       

      Right now, I have six hosts in a "preferred" Fault Domain and six hosts which are standalone (having deleted the Failover Fault Domain)

       

      What I need to do is decomission the six hosts that were formerly in the Failover Fault Domain. How do I do this?

       

      THX

        • 1. Re: 1 more standalone host is required. Some objects will become inaccessible or non-compliant with their storage policy
          TheBobkin Virtuoso
          VMware EmployeesvExpert

          Hello InvisibleAdmin,

           

          "This was an active/passive with no VMs running on the Failover domain."

          VMs not running on one site doesn't mean they don't have components stored there - if you are/were using the default Stretched-cluster Storage Policy (RAID1 across sites) or any other Stretched-cluster policy then this isn't really 'active/passive' from a storage perspective.

           

          From what you have described there are 2 likely causes, either you are:

          1. Unstretching this cluster incorrectly and you still have data with Stretched-cluster policies applied - checking for such Objects just on the stretched Storage Policy(ies) is not sufficient as this only shows VMs/Objects in current vCenter inventory - check with the output from 'esxcli vsan debug object list' (--all required if 6.7 U3).

          Proper process for unstretching a cluster:

          Convert a Stretched Cluster to a Standard vSAN Cluster

          2. Have some data with a Storage Policy of FTT=2,FTM=RAID6 - this requires 6 'standalone' nodes (e.g. not in Fault Domains) for compliance and thus if you only had 6 applicable nodes then you can't place one in Maintenance Mode with 'Full Data Evacuation' option as you wouldn't have anywhere applicable to move the data to (as all other nodes already have components and thus moving the data to any other node would violate the Storage Policy).

           

          Bob