I'm having the same issue, but I'm on running ESXi 4.0.0 releasebuild 171294 and vSphere 4.0.0. I'm doing a POC and had been running for about 30 days with no problems, but came in this last Monday and notice that the ESXi host had disconnected from vSphere. I tried to connect, but it kept telling me that there were networking issues or the management network may need to be restarted.
Restarting the management network on the host works since I can then reconnect from vSphere, but it disconnects again anywhere from 90 seconds to 2 minutes later. I noticed that running "test management network", the pings always are "ok" for the gateway, but fail half the time for the DNS servers. Ping is testing layer 1 so that should be a physical problem somewhere, right?
I only have 1 NIC in the host system and the host is only running 1 active VM provisioned through View. While I'm having the host disconnect and reconnect problems, the user isn't having any problems at all so the VM isn't experiencing any network problems that I can tell.
I spent time working through the problem last night checking network cables, connections and did replace some dodgy looking cables, but didn't change the symptoms at all. One thing that keeps bugging me though is that I had this running nonstop for 30 days and then our IT admin did work in the server room over the weekend and on Monday the problem was noticed. But the only place I notice the problem is the connection between vSphere and ESXi host, the View connection server is fine
and the only thing I could do to keep the host connected to vSphere was to run a continuous ping from the vSphere server to the host IP. So far, the host has been connected all night and into this morning.
What is different about the Management Network vs the VM network since they are both going through the same virtual switch and the same NIC?
Any ideas would be greatly appreciated!
Rafael
