VMware vSphere

 View Only
  • 1.  vMotion not working as expected

    Posted Sep 17, 2018 09:13 AM

    Hi

    We configured new 2 X ESXi (6.7) hosts, assigned Datastore to both the hosts and prepared the VCSA VM on the Datastore. We enabled HA & DRS. For testing we enabled maintenance mode on one of the ESXi host where the VCSA VM is running, we expect the VM to migrate (vMotion) to the other host automatically, but getting warning message saying "Insufficient vSphere HA failover resources". How can we correct this

    Regards



  • 2.  RE: vMotion not working as expected
    Best Answer

    Posted Sep 17, 2018 09:20 AM

    Two things to start of with -

    1) For the VMs to move automatically when the host is placed in maintenance mode, the DRS should be set to Fully Automated mode.

    2) With just two hosts in the cluster, do not use vSphere Admission Control feature. Maybe that is why it is complaining about 'Insufficient vSphere HA failover resources' once one of the host is put into maintenance mode.

    Cheers,

    Supreet



  • 3.  RE: vMotion not working as expected

    Posted Sep 17, 2018 09:36 AM

    Add more Hosts or reconfigure your Admission Control.

    If you specify that the cluster have to tolerate 1 host failure or 50% capacity you cant reduce the number of active hosts to 1 and expect that this single host give you redundancy.

    Regards,

    Joerg



  • 4.  RE: vMotion not working as expected

    Posted Sep 17, 2018 09:54 AM

    We corrected it by Disabling Admission Control. So how can we fine tune HA / DRS settings



  • 5.  RE: vMotion not working as expected

    Posted Sep 17, 2018 10:22 AM

    You can find more details in the resource management guide - https://docs.vmware.com/en/VMware-vSphere/6.7/vsphere-esxi-vcenter-server-67-resource-management-guide.pdf​. Do let us know should you have any specific queries.

    Cheers,

    Supreet



  • 6.  RE: vMotion not working as expected

    Posted Sep 17, 2018 12:20 PM

    By default, HA Admission Control will "reserve" 50% of the resources for HA reasons (in two node cluster; in three node cluster it will be 33% by default). The idea is to be able to successfully assign resources during the failover process (in case of two hosts you should not allocate more than 50% of total available resources to satisfy the possible failover requirements).

    The exact configuration depends on your workloads and protection requirements; in your situation, it might be more useful to disable HA Admission Control at all because the "costs" of those reserved resources will be quite high.

    There is no direct correlation between HA / DRS and Admission control. You can fine-tune HA / DRS with Admission Control disabled, but keep in mind that during the unplanned failure of the host your environment might not have enough resources to run all your VMs without impacting the performance