Labels

Thursday, June 16, 2011

High Availability

High Availability
VM HA monitors the esx failure – if any failure happens the VM in it are restarted in another host in the cluster.
  1. Decrease downtime and improve the availability
  2. It takes at least 3 min to power up the VM in another host
  3. Ping 1 sec once to another esx
  4. If 15 sec fails it assumes it is dead

Procedure
1. HA
   - 1.1-host monitoring status
   - 1.2-admission control [allow / disallow]
   - 1.3-Admission control policy

        1.3.1=> Host failure tolerates [1]
      1.3.2 => % of cluster resource reserved for failover spare capacity [25 %]
      1.3.3 => specifies a failover host [^]

 2. CLUSTER DEFAULT SETTING
         - 2.1 => VM restart priority
         - 2.2 => host isolation response

3. VM MONITORING
     - 3.1 => VM Monitoring Status [enable / not]
    - 3.2 => default cluster setting 
Monitoring Sensitivity [low --- High] [HA will restart VM if the heartbeat between the host and the VM has not been received within a 30 second interval. HA restarts the VM after each of the first 3 failures every hour]
 4. EVC
 5. SWAP file location

How HA works
  • First 5 hosts are primary hosts then all subsequent hosts are secondary hosts.
  • The very first host is the primary master
  • Usually VC communicates to primary master
  • Primary master maintain and replicate the cluster info to all the hosts.
  • Primary master – decides where to restart the VM
                                               - keeping track of failed restart attempt
                                               - determining when it is appropriate to keep trying to restart the VM
  •  If VCenter down, HA will work because the configuration stored in the aam folder of the ESX. Only DRS won’t work

List primary hosts
$ cat /var/log/vmware/aam/aam_config_util_listnodes.log 

Secondary host promoted after primary is: • maintenance mode • disconnected • removed - not on failure


Host isolated: no heartbeat for 12 seconds, then cannot ping isolation addresses. Isolation response: • Power off • Leave powered on • Shut down (default). However Host Failure is only after 1 5 seconds.
Admission Control types:
• Host
• Resource Pool
• HA (only HA admission control can be disabled)
Admission Control:
Rules if VMs can power on when they violate availability constraints at HA failover. Actions that change a reservation must satisfy admission control.
HA Admission Ctrl policies:
Host Failures Cluster Tolerates (1 -4 hosts),
 % of Cluster Resources (up to 50%),
 Specify a Failover Host.
  Policy Factors
   • Resource fragmentation
   • Flexibility
   • VM diversity
Slot size: represents VM CPU & memory resources needed for any powered on VM. Distorted by large VM reservations. Avoided with advanced attributes das.slotCpuInMHz or das.slotMemInMB

  • Slot size :- maximum resource of a VM
  • Admission control: checks the resource availability- if success allows to start the VM.
  • Affinity rules : keep the specified VM’s together or separate the VM


- Incoming port: TCP/UDP 8042-8045 - Outgoing port: TCP/UDP 2050-2250
Config file is
1. /opt/vmware/aam
2. /opt/vmware/aam /bin/….
3. /opt/LGT0aam512/config/*

Logs:
 HA logs: /var/log/vmware/aam/ aam_config_util_*.log
Service
  #service vmware-aam stop 


To install and uninstall the HA agent
    1. /opt/vmware/aam/bin/vmware-aam-ha-uninstall.sh
    2. #rm –rf /opt/vmware/aam
  Log files>>>>>>>> /opt
          - /opt/vmware

1 comment:

  1. hi, this is very useful .please try to updated vsphere 5.x version HA working nature

    ReplyDelete