VMware ESX host can disconnect from VirtualCenter & have a problem connecting to it again for many reasons. Below I will cover few ways to diagnose the problem, few common resolutions to this issue.

In these cases usually the first thing you want to do is to log on through ssh, vcli, or directly on the server console & check the following log files for errors:

Vmkernel – /var/log/vmkernel
Vmkernel Warnings – /var/log/vmkwarning
ESX Server host agent log – /var/log/vmware/hostd.log
Service Console – /var/log/messages
VirtualCenter agent – /var/log/vmware/vpx

Note: it would be great if you get to know what each of the above log files is for.  VMwareWolf has a great post explaining these log files that can be found at: http://www.vmwarewolf.com/which-esx-log-file/

Usually reading the log files above with enough knowledge you will be able to tell what is wrong, or at least get a clue of what to look for. Below are few further things to check for.

At the service console make sure the mgmt-vmware service is running by using the following command:

service mgmt-vmware status

If the mgmt-VMware service is stopped try to start it by running:

service mgmt-vmware start

Some other time, the service might be running but not doing what its suppose to do. For that there is always no loss in restarting the mgmt-vmware service as it does not cause any downtime. The command to restart the mgmt-vmware service is:

service mgmt-vmware restart

Next step will be to try to restart the following services:

service vmware-vpxa restart
service vmware-vmkauthd restart
service xinetd restart

If after restarting all of the above services, you still could not connect you should consider to kill -9 the pid for the vmware-hostd & then start the mgmt-vmware service again.

If all of the above did not work & you are running a VMware 3.5 U1-U4, you might be facing a memory leakage problem that was resolved in VMware 3.5U5 & above. Of course upgrading the server to 3.5U5 or above can resolve the problem, but that might be not an option. To check if this bug is the case, run esxtop & check the memory usage of the Pegasus (cimserver). If that is quite high, then there is a huge chance its the cause of your problem & you should consider looking at the following KB. Pegasus (cimserver) memory leaks reported in ESX 3.5 Update 2 and later

If none of the above work, you might want to consider reinstalling the vpxa & hostd rpm using the following commands:

for vpxa reinstallation:

#rpm -qa | grep -i vpxa (check current vpxa version)
the output will be something like below:
VMware-vpxa-X.X.X-yyyyy
Try reinstalling the package by:
#rpm -Uvh VMware-vpxa-X.X.X-yyyyy

for hostd reinstallation:

#rpm -qa | grep -i hostd (check current hostd version)
the output will be something like below:
vmware-hostd-esx-xxx.rpm
Try reinstalling the package by:
#rpm -Uvh <vmware-hostd-esx-xxx.rpm>

I hope one of the above methods help you resolve your ESX host connectivity to Virtual Center. If you have any further method that I have missed, please share it in the comments area below.