"vsphere client could not connect to" after upgrade from ESXi 3.5 to 4.1

[WDTElwood]

I have an odd problem that is just baffling me. I have two of the following:

- HP DL360 G5 servers with 12GB of RAM

- Both Are hooked to EMC AX150 SAN where my VM's are stored using Q-Logic 4050C-E cards via iscsi

- Both are at BIOS Revision P58

- both have exact matching BIOS settings

- My AX-150 is a Part Number 100-561-402 Rev:A02

I have been running along happily at 3.5 but wanted to upgrade to version 4 for a variety of reasons. Initially, I tried this upgrade and was able to ping the actually machine hardare box, but when i tried to connect with the client, I got the following message:

vsphere client could not connect to x.x.1.6.

I was a bit baffled and just for fun I restarted the box again and noticed a quick red massage flash by at nearly the end of the boot sequence. It went by so fast I had to make a movie of it and look at the frame. The error I was getting was:

"The Execute Disable/No Execute CPU feature is not enabled for this machine"

I tracked this problem down, wondered how it could possibly affect connectivity, but adjusted the bios accordingly and it went away. I still could not connect, even though I could ping the hardware box.

At this point, I decided to go back to 3.5 and bring my machines back online and research the issue. When 4.1 was released, I decided to try it again. I did the install from a CD and poof, everything worked. I decided to wait a week to do the second machine and that is what I tackled this past weekend. To make a long story short, the second upgrade wouldn't let me connect. It gave the "vsphere client could not connect to x.x.1.6" message. I thought maybe I made a bios mistake so I restarted the working 4.1 box and went into the bios - everything was matched on both machines...and then the trouble started.

I exited bios and now I couldn't connect to anything, not even the machine that had been working fine at 4.1 for a week! I tried restarts, verified settings, everything! And then, by accident, I shutdown the original 4.1 box that was working instead of restarting. When I powered it back up, I tried to connect again and amazingly connectivity had returned.

So, what I can't figure out is why do I lose the ability to connect if I restart a hardware box on 4.1, but gain it back if I do a full shutdown?

Any help is greatly appreciated! If any further details are needed, I will provided whatever is asked for.

Thanks

Eric

Edit: added AX140 and QLA card specifics above

[jamesbowling]

Well, the first issue I would see is that the AX150 is not supported on 4.X. It is only supported on ESX(i) 3.5u5 and below. I saw something like this happen at one of my company's sites where the engineer on site was testing a HP DL360G5 with 4.1 installed and had it connected to the AX150. We saw all sorts of odd behavior. Once we removed that host everything was good.

If you found this at all helpful please award points by using the correct or helpful buttons! Thanks!

[WDTElwood]

Thank you for your response!

I did look at the HCL again and it seems the AX150 is not listed in the HCL for ESXi:

AX150 HCL Doc

So, in theory, I should have been having problems all along and yet I have been fine until the 4.1 upgrade. And,as I stated, it's only if I try and restart the actual HP box that I can't connect. If I shut it down and power back up, I seem fine and everything purrs along.

I have added some more info up top about my hardware details.

[jamesbowling]

I am not saying that your 3.5 hosts would have any issues but specifically the 4.1 host causing the issues all the way around by being connected to the AX150. We had the same exact issues and VMware Support came back to us stating that it was the 4.1 host we added to the AX150. Just as you said, as soon as you reboot it is fine. I am willing to bet that it goes back to bad after a while though, correct?

If you found this at all helpful please award points by using the correct or helpful buttons! Thanks!

[WDTElwood]

I will be testing it again this weekend. It runs fine now and ist actually performing better than when it was 3.5 but I want a large maintenance window to confirm it all again.

What I'll check is to confirm that:

1. I lose connectivity if I simply restart the HP Hardware.

2. After losing it, that I get it back if I power down the HP hardware and then bring it back up.

I guess I should add that at no time has my AX150 been powered down during any of this just to take that variable out.