Help - virtual center loses connection to esx host machine

[dkaiserlt]

Can someone possibly identify what's going on between my vSphere vCenter server and ESX4 host server. For some reason, the vcenter server is reporting that it is unable to connect to the ESX host. The alarm is "Host connection and ower state" is in an alert state and in the events log i see a "Host is not responding." In virtual center, if I click on my esx host and click connect, then it re-establishes connectivity and everything looks fine for a min or two until it loses connectivity again. This process repeats itself when I try to reconnect to my ESX host via vCenter server.

I am able to connect to the ESX host directly and i don't get dropped.

Running VMware ESX 4.0.0, 175625 and vCenter Server 4.0.0, 162856

Thanks everyone.

[Datto]

Three possibilities to consider:

1) A duplex mismatch problem someplace (in NIC or in switch port) -- in the VIC, make sure your virtual NICs are set to auto negotiate (assuming your switch ports are also set to auto-negotiate)

2) A firewall port that's not open on the ESX host (Configuratoin tab/Security Profile/Properties -- see if the VMware VCenter Agent port is open in the list)

3) Some kind of anti-virus/firewall installed on the VCenter server is interferring with VC to ESX communication.

Datto

[dkaiserlt]

Datto,

Thanks for replying.

1. No duplex mismatch problem that I can see. Everything is configured for auto negotiate and all NICs are showing 1000 Ful.

2. Confirmed that the vCenter Agent port is opened.

3. I even tried disabling antivirus on the vCenter server but it didn't help.

I configured another ESX host and installed the vSphere client and it works just fine, able to manage using vCenter. So it somewhat rules out an issue with the vCenter Server.

Any other ideas or suggestions?

[Patrick_Donker]

I too have this issue. Out of the blue vCenter loses its connection to the ESX host. I can reconnect to the server, but after a while it disconnects again.

If I connect to the server directly all goes well. I have reinstalled vcenter on another server but the issue pops up on that one also.

Looks like something is wrong on the ESX host....but what? I've inspected logs, buI cannot find anything.

Any clues?

Thanks

[rschmidt1300]

I am also having this issue running my hosts and vCenter Server under workstation 6.5. If anyone has any suggestions I would really appreciate them.

Thanks

Rob

[sr_whq]

And the same to me. All NICs of the ESX host (3) are on 1000Mbit FD, all switch ports as well. There is no AV / FW software of any kind on the VCenter Server (W2K8 w. SQL Express). Also the VCenter Agent port is open (@Datto: to ask for that is senseless in my opinion, 'cause the OP already told that the can reconnect to the ESX host).

I restarted the VCenter Agent on the ESX host ("service vmware-vpxa restart"), the VCenter Service on the VCenter Server and both the hosts - no change at all. If I connect directly to the ESX host (from the VCenter Server host), everything is just fine. If I (re-)connect to the VCenter Server (also from the VCenter Server host), the connection is lost after a differing 30 and 90 seconds.

I have installed both available update bundles yesterday, but there is absolutely no change in behavior. If I close Vclient and re-start it with connection to the VCenter Server, the connection is established at 1st but shows an old build number for the ESX host. Every time. If I dis- and re-connect the ESX host, the correct version is displayed.

I am currently evaluating the software for a large customer and for your information, dear VMware professionals, this problem is a show stopper.

EDIT-1: It seems that the VCenter Setup misses something when configuring the integrated firewall of Windows Server 2008. After completely disabling it the connection is by now established for over 30 minutes, which is new to me. I suppose that VCenter switches / randomizes ports or something, when talking to the ESX host?

GREAT how easy things are resolved sometimes.

BUT: VMware, fix that firewall configuration setup asap, my customer uses the integrated windows firewall and it is an requirement to have VCenter working with integrated firewall enabled. In the meantime I need a complete profile of all connections between VCenter and ESX hosts to work around this issue.

EDIT-2: I verified the behaviour by re-enabling Windows' internal firewall. So the clear result is that there is a bug in VCenter Server's setup routine which configures the firewall to allow VMware traffic.

[PatrickDonker]

For me a reboot of the ESX host helped.

[sr_whq]

Patrick, that's good to hear...but I'm curious: On which platform do you use VCenter Server, W2K, W2K3 or W2K8? ServicePack? Which database?

[dkaiserlt]

Hi Guys,

I appologize for the delay in getting back to everyone regarding my experience, my problem just went away after a week or two of posting the original message. I haven't made any changes to either vCenter or ESX to get it working, just came in one day and checked vCenter which showed the server was still connected.... very strange. Anyhow, I'm running vCenter on Windows 2008 Std 64 bit.

[PatrickDonker]

I ran it on a w2k3 sp2 when the problem first started. I thought it had to do with the server it ran on; a really old banger. So I installed it on a brand new Sun x2270 with w2k8, but the problem still occurred. I'm not convinced it is a vC problem because I had to install some fixes on the ESX, and after a reboot the problem disappeared. So I think it has to do with some sort of bug on ESX.

[suchittt]

i have resovled this issue by using windows firewall exception technique.

you must have windows fireallwall powered on machine where  your vcenter server is installed.

In this scenario windows firewall blocks communication between vcenter server and esx host after certain time.

I have experienced similar problem like you. every time i had to connect it after 2-3 mins. it was showing 'host connection failure' in events of vcenter.

Resolution:  Login to vcenter server host machine--> go to control pannel --> windows firewall--> open firewall settings and go to exception.

--> in exception add program--> browse for vcenter server installaed location in program files--> select the vpxd.exe file :at path c:\program files\vmware\infrastructure\virtualcenter server\vpxd.exe---> click ok to add exception.---> and here we go.

you will have connectivity to esx hosts automaticaly in vcenter console.

Please give feedback if it works for you. 

[iree75]

Had this issue on vc 4.1 on top of Win2k8.

Looking at the windows firewall settings, it shows that our network was defined as a "public" network. While you could change the firewall to allow all of the virtual center connections in the "public" profile, I just set the network connection type to what it should be, "home or work - private". The firewall rules for vc by default are set to allow in this profile. All of my esx servers reconnected.

-F