Host shows the vSphere HA status as Network Partitioned

Products

VMware vCenter Server 8.0 VMware vCenter Server 7.0

Issue/Introduction

A host reports the vSphere HA status as Network Partitioned
In the fdm.log file of the network partitioned host, you see these entries:

info 'Cluster' opID=SWI-8948711d] [ClusterManagerImpl::MainLoop] curState 1 lastState 3
info 'Election' opID=SWI-eb460f87] Startup: Got AmMaster
0 info 'Election' opID=SWI-eb460f87] [ClusterElection::ChangeState] Startup => SlaveConnecting : StartupStateFunc
'Cluster' opID=SWI-eb460f87] Change state to SlaveConnecting:20224569933819
verbose 'Cluster' opID=SWI-8948711d] [ClusterManagerImpl::CheckElectionState] Transitioned from Startup to SlaveConnecting
info 'Cluster' opID=SWI-eb460f87] [ClusterManagerImpl::ConnectToMaster] Connecting to master host-###@ #.#.#.#:8182
info 'Cluster' opID=SWI-8948711d] [ClusterManagerImpl::MainLoop] curState 3 lastState 1
warning 'Libs' opID=SWI-2e80d2bf] SSL_VerifyX509: Certificate verification is disabled, so connection will proceed despite the error
verbose 'Cluster' opID=SWI-eb460f87] [ClusterManagerImpl::VerifyHost] Thumbprint match ##:##:##:##:##:##:##:##:##:##:##:##:##:##:##:##:##:##:##:## for host host-###
'Cluster' opID=SWI-eb460f87] [ClusterManagerImpl::ConnectToMaster] Connected to master @ host-###
2012-01-23T15:51:55.105Z [FFE48B90 info 'Election' opID=SWI-eb460f87] Slave to host @ #.#.#.#
2012-01-23T15:51:55.105Z [FFE48B90 info 'Election' opID=SWI-eb460f87] [ClusterElection::ChangeState] SlaveConnecting => Slave : SlaveConnectingStateFunc
2012-01-23T15:51:55.105Z [FFE48B90 info 'Cluster' opID=SWI-eb460f87] Change state to Slave:20224569933819
2012-01-23T15:51:55.105Z [FFE07B90 verbose 'Cluster' opID=SWI-8948711d] [ClusterManagerImpl::CheckElectionState] Transitioned from SlaveConnecting to Slave
'Cluster' opID=SWI-8948711d] [ClusterManagerImpl::MainLoop] curState 4 lastState 3
'Election' opID=SWI-eb460f87] [ClusterElection::ChangeState] Slave => Startup : Lost master
'Cluster' opID=SWI-eb460f87] Change state to Startup:0
verbose 'Cluster' opID=SWI-8948711d] [ClusterManagerImpl::CheckElectionState] Transitioned from Slave to Startup
info 'Message'] Destroying connection
In the fdm.log file on the host that is listed as the master for the HA enabled cluster, you see these entries :

verbose 'Cluster' opID=SWI-14c14582 '[ClusterManagerImpl::IsBAdIP] #.#.#.# is bad ip
verbose 'Cluster' opID=SWI-14c14582 '[ClusterManagerImpl::InvalidCredentialsIP::IsBadIP #.#.#.#  is in the bad ip manager
warning 'Election' opID=SWI-14c14582ReadMsg [60 times] Received messge from bad ip #.#.#.#  - dropping
verbose 'Cluster' opID=SWI-14c14582 '[ClusterManagerImpl::Verify Host] Thumbprint mismatch (##:##:##:##:##:##:##:##:##:##:A3:7F:
verbose 'Cluster' opID=SWI-14c14582 '[ClusterManagerImpl::InvalidCredentialsIP::SetBadIP] Blacklisting ip address #.#.#.#  for 60 seconds
warning 'Cluster Slave host-124 has invalid credentials - closing connection

Environment

VMware vCenter Server 7.0

VMware vCenter Server 8.0

Cause

This issue occurs when the SSL certificate thumbprint presented to the master host is not what the master host is expecting. This is indicated by the thumbprint mismatch error in the fdm.log file of the master host.

Resolution

To resolve this issue:

Disable HA.
1. In the vCenter Server inventory, right-click the cluster and click Edit Settings.
2. Deselect the Turn On vSphere HA option.
3. Click OK.
4. Wait for all hosts to reconfigure HA in Recent Tasks.
Ensure that SSL Certificate Checking is enabled.
1. In the vSphere Web Client, navigate to the vCenter Server instance.
2. Click the Manage tab.
3. Under Settings, click General.
4. Click Edit and select SSL settings.
Select vCenter requires verified host SSL certificates. If there are hosts that require manual validation, these hosts appear in the host list at the bottom of the dialog.
Click OK.
Determine the host thumbprint for each host that requires validation.
1. Log in to the direct console. For more information, see the Using the ESXi Shell section
2. Click View Support Information in the System Customization menu. The thumbprint is displayed in the right column.
  
  Notes:
  - If you do not have access to the direct console, connect a vSphere Client that has not installed the hosts certificate directly to the host. When it prompts for certificate confirmation, click View Certificate > Details and then scroll down to Thumbprint.
  - If your issue is occurring because the SSL Thumbprints do not match, when you click OK, all listed hosts disconnect from vCenter Server. Reconnect each host to refresh the SSL thumbprints. This requires the root password.
Compare the thumbprint you obtained from the host with the thumbprint listed in the vCenter Server Settings dialog.
If the thumbprints match, select the check box for the host.
Click OK. Hosts that you have not selected are now disconnected.
Reconnect the host to vCenter Server.

To reconnect the host:
1. Right-click the disconnected host and click Connect.
2. When prompted, enter host's credentials to connect the host back to vCenter Server.
  
  You see a popup that mentions the SHA1 thumbprint of the SSL certificate for the host.
3. Click Yes. The host is now connected to vCenter Server.
Enable HA.

To enable HA:
1. In the vCenter Server inventory, right-click the cluster and click Edit Settings.
2. Select the Turn On vSphere HA option.
3. Click OK.
4. Wait for all hosts to reconfigure HA in Recent Tasks.

The host should now show the vSphere HA status as connected (slave) or connected (master), depending on the election results when HA was enabled.

Additional Information

Note : Found error "vSphere HA detected that this host is in a different network partition than the master to which vCenter server is connected", corrected NIC Teaming policy from Originating Port ID to IP Hash since rest of the hosts are configured to use IP Hash.