0 Replies Latest reply on Nov 6, 2019 4:25 AM by ivanivanov1

    VM restart interval in vSphere 6.7 HA

    ivanivanov1 Lurker

      Hi all.

      Lab environment: vxRail 4.7.3 (3-node cluster), vSphere 6.7, vSAN 6.7, HA enabled.

       

      I took a look on HA behaviour in lab env and got a bit confused.

       

      Case:

      - only one VM with "highest" restart priority is located on the "slave" host.

      - host gets failed (powered off by me, for example)

       

      Due to official docs and "vSphere 6.x HA Deepdive" book (https://www.vmgu.ru/ext/books/vsphere-ha-deepdive.pdf )

      i expected to see following:

      1. After 15s - host is declared dead,

      2. Test VM is restarted on one of the live hosts.

       

      IRL i got 80s delay between 1 and 2:

      2019-11-05T15:34:48.754Z info vpxd[06540] [Originator@6876 sub=MoHost opID=FdmWaitForUpdates-vim.ClusterComputeResource:domain-c8-71d6bbe9] VC state for host host-44 (initialized -> initialized), FDM state (FDMUnreachable -> Dead), src of state (null -> host-48)

      2019-11-05T15:36:08.332Z info vpxd[03742] [Originator@6876 sub=InvtVm opID=HB-host-21@789-4d57227f] dasRestart set to true for VM (cfgPath: ds:///vmfs/volumes/vsan:52be8d3b8fa4262b-5ca52f2655c8af9b/b83ec15d-7268-7694-0b34-34800d427f20/ubuntu-test.vmx, moId: vm-81, vpxaVmid: 5)

       

      General questions:

      - what exactly happens in those 80 seconds?

      - can i change this delay?