Recently, I conducted environmental upgrade from version 5.1 to version 5.5U2. After picking up versions of all components of vCenter came time to upgrade ESXi hosts. In my case it was carried out by using the Update Manager. All machines have been done within two days. The next step was to upgrade the distribution switches. First side up – all OK. Then the production side. Upgrade went without any problem, but after a while I received an alert that one host can not synchronize all information from vCenter.
Synchronization problem has been solved by forcing resynchronization settings.
Losing network packets was very irregular and resulted in periodic changes in the status of the host. From time to time I received information that the host does not respond. At first I started to browse logs vCenter and affected this issue host. Nothing indicated any problem. The only information on which I could find concerned losing hartbeats between vCenter and host. After several changes of compatible drivers that have not produced the desired effect decided to open a support request in VMware.
How is it with this type of requests, have been asked to send logs and accurate description of the problem. VMware also did not find anything in the logs indicating that by this problem. I was, of course, also be asked to perform several recommended steps that could help. But nothing brought no result.
After more than a week and performed various operations came across a strange situation from the Web Client, namely when checking the configuration of vDS noticed that the tab “Related Objects -> Hosts” does not show all servers. Yes, there lacked the host, who had a problem with communication.
It was very strange because from the vSphere client host was visible from the switch and all port groups, and from the Web Client was not visible from the port groups.
I decided to throw a problematic host from all the distribution switches and adding it again. The only problem was the managment port group that I needed to migrate to a vSS and after the entire operation move it back.
After this all the hosts were visible from the vSphere client and Web Client, and most importantly, communication back to normal.
Support VMware was very surprised by this problem, and I’m very happy that the problem no longer exists.