In my earlier post (https://ervikrant06.wordpress.com/2014/08/30/vmware-interview-questions-part-1/) I shared some interview questions but I feel need to dig more on vSphere HA topic hence came up with this post.
Q : What is FDM ?
A : FDM stands for Fault domain manager. Its having two communication with hostd and vCenter server. vCenter gets the VM state information from FDM.
Q : What is master node in cluster is responsible for ?
A : Basically master role plays all the role like the primary node in vSphere 4 cluster.
a) Restarting failed Virtual machines
b) Exchanging state with vCenter.
c) Monitor the state of slaves.
Q : What if the master node fails ?
A : If the master node fails reelection will occur. From the Hosts that is participating in the election host with highest number of datastores will be elected as master. If two or more hosts have same number of datastores the one with the highest Managed Object ID will be chosen.
Q : How many host failure types can be detected in HA ?
A : a) Host fails
b) Host becomes network isolated
c) Host loses network connectivity with master host.
Q : What is datastore heartbeating ?
A : Prior to vSphere 4.0 HA will kick off the restart of VM even if only the management network of host was isloated and VMs are still running. This was unnecessary issue. To remediate that new level of resilence is added to make distinction between failed host and an isolated or partitioned host.
Q : How many datastore by default is chosen for datastore heartbeating ?
A : Two datastores by default are chosen for datastore heart-beating. You can change this value with help of das.heartbeatDsPerHost parameter.
Q : How datastore heartbeating works ?
A : HA leverages the existing VMFS file system locking mechanism. The locking mechanism uses a region called “heartbeat region. Per host one file is created, HA will check if whether heartbeat region is updated or not.
Q : What is the difference between isolated and partition state ?
A : Isolated is not receiving any election traffic but partition is receiving it.
Isolated can’t ping the isolation address but partition can.
Similar point is :
In both cases host is not able to receive heartbeat from master
Q : What happens whey you disconnect a host from a cluster ?
A : All VMs registered to that host are unprotected by vSphere HA. Means they are no longer in control of HA.
Q : What happens when we select the disabled VM restart priority for VM in HA ?
A : If disabled setting is selected, vSphere HA is disabled for VM, which means it is not restarted on other ESXi hosts if host fails on which currently its residing. If there is an OS level issue and vSphere HA VM/Application Monitoring is enabled then reboot of OS will trigger but VM will remain on same host.
Q : What is difference between shutdown and power off the VM ?
A : Shutdown operation preserve the state of VM. Its better than powering off the VM in some cases.
Shutdown operation require vmware tools to be installed for VM.
Shutdown operation require more time than poweroff.
Q : When the split brain condition will occur in vSphere HA cluster ?
A : If Host isolation response is disabled and one host got isolated and it has running VMs. As the VMs are not having access to disk locks. VMs got failed over to another host even though the original instances of the VM remain running on the isolated host. Now failed over instances of VM start working by regaining disk locks.
Q : What changes occur if we enable vSAN in HA cluster ?
A : By default HA cluster uses managment network of HA traffic. If vSAN is configured it will use vSAN dedicated network for HA traffic. This change doesn’t occur automatically, HA reconfiguration on hosts require to bring it into effect.