• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

vSphere4 environment in a box

sabregen

sabregen

Fully [H]
Joined
Jun 6, 2005
Messages
19,501
Anyone else planning on doing this? I am starting my deployment, as I type this. I recently moved from 20years of PC usage to a Mac...and the old PC was actually a laptop, but qualifies far more as a DTR system. Since the machine has been returned to me, I thought it'd be great (since we have a shitload of NFR's that we're not using) to go ahead and have a mobile demo environment to show customers. This machine is perfect, as it has it's own big screen, plenty of power, and tons of connectivity (in case I need to use it in a presentation envinronment with a projector, or something).

Laptop Specs:

Sager NP9262
Intel Core 2 Quad Q6600 @ 3.0GHz
4GB PC2-6400
3x 160GB 5400RPM Fujitsu SATA in RAID-5 on Intel ICH7R
17" 1920x1200 WUSXGA Glossy
ABGN 2.4 + 5GHz wireless
DVI/VGA/S-Video/Component/Composite
8x Dual Layer DVD Burner
20,000,000 lbs
1 sec battery life

Host OS is Vista x64 Ultimate, running latest build of Workstation 6.5. Here's the two catches that I've found so far:

vShpere4 requires at least 2GB of RAM to boot (ESX 3.5 U4 would boot with only 512MB of RAM)

vSphere4 doesn't seem to want to let me (at initial installation) not provision a VMFS3 partition out of the boot device. ESX 3.5 could be installed a 5GB virtual SCSI disk. vShpere4 is requiring at least 9.5GB...which sucks. I'd imagine this wouldn't be the case, if the vSphere4 host was acutally on a SAN of some sort, and could see multiple LUNs, but that's not what I am going to do.

The guide to follow is here:
http://www.petri.co.il/running-vmware-esx-and-esxi-in-workstation-on-your-desktop-pc.htm

So far, other than the two things that I have mentioned so far, I haven't found any other signifcant changes that need to be considered when deploying the ESX4 host inside of workstation 6.5. I'll keep this updated as the progress continues.
 
Oh, BTW... I have done this before, but I think it bears repeating, as the last time I did this was about 6 months ago, right before I passed my VCP. Plus, I think a lot of my clients would like to see this stuff in action, instead of just conceptualizing it entirely on a whiteboard. Seeing sometimes makes the light bulb go off, where as a whiteboard might not. Here's the preliminary plan:

* 2x ESX4 VMs (fully licensed for all Enterprise level features)
* Windows Server 2003 R2 Standard Edition (x86) running VC stuff, DNS, AD implementation (as a separate VM, not hosted on the ESX boxes/openfiler)
* OpenFiler iSCSI Target VM (will host all VMs on it's provisioned disk)

Oh, and since I know you are going to read this, lopoetve, I seem to recall last time having some difficulty getting the double-nested VMs running with their files on the OpenFiler VM to be an issue. However, I can't recall now what the problem was, or what the fix was. Care to shed any light on that, if you can?
 
Last edited:
You have to add a line to the VM's vmx file or else it won't let you virtualize a system under virtualization.
 
a line to the double-nested VM's .vmx, or the ESX4 host .vmx file? got a link, or the line?
 
sabregen said:
a line to the double-nested VM's .vmx, or the ESX4 host .vmx file? got a link, or the line?
Click to expand...

On the VMX for ESX (running under another hypervisor) you add these:

ethernet0.virtualDev = "e1000"
monitor.virtual_exec = "hardware"
monitor_control.restrict_backdoor = "true"

No clue if that works on 4.0. I haven't tried 4.0 in a lab box like that. I've done it with 3.5 for my portable SRM lab.
 
in vSphere 4, the only one of those lines that doesn't exist by default in the .vmx under Workstation 6.5 is the monitor_control.restrict_backdoor = "true"
 
I'm no networking genius (we have a CCIE on staff, for that), so can someone double check this for me?

ESX1.tempdomain.local:
2 nics
Service Console 1: 10.10.10.1 / 255.255.255.0
Service Console 2: 10.10.10.10 / 255.255.255.0
VmKernel Port: 10.10.20.1 / 255.255.255.0 (marked for FT and VMotion)
 
Last edited:
I am getting ping test timeouts to my ESX4 hosts. command line execution is as follows:

ping -t 10.10.10.1

results seem to be a response for 8 pings, then a failure to respond for 6 pings. The pattern then repeats. Anyone got any ideas?

esx1 and esx2 are up. OpenFiler is being provisioned now. I am approaching the limitations of the system RAM installed. As OpenFiler is installing, current RAM usage is at 3190MB. It will be interesting to see what the RAM usage is like after OpenFiler is actually up, and serving out disk. Anyone have any ideas on OpenFiler as a Workstation 6.5 VM should have for RAM allocation? It's currently set as an "Other 64-bit Linux Guest" VM, using 384MB of RAM.
 
Last edited:
Looks like this page has all of the information necessary to deploy the latest version of OpenFiler. Note that I am creating two Hard drives for the OpenFiler VM. One will be for boot, and one for the VMFS3 storage.

http://www.techhead.co.uk/how-to-configure-openfiler-v23-iscsi-storage-for-use-with-vmware-esx

OpenFiler boot disk is 2GB, VMFS3 ESX4 target is 40GB. OpenFiler IP information:

hostname: openfiler.testdomain.local
10.10.10.50 / 255.255.255.0
Single NIC
 
Last edited:
Heh, I was wondering why the ESX VMs couldn't see the iSCSI target. Well, if you are in a different IP range, it's a bit hard to see the iSCSI target from the ESX hosts. I added a second NIC to the OpenFiler VM:

10.10.20.3 / 255.255.255.0

Dynamic discovery worked just fine. Volume has been added to each of the hosts. Now I will need to deploy the Win2k3 server for vServer functionality.
 
doesn't work, don't bother yet. You won't be able to boot a VM.
edit: the local vmfs is for the COS vm too :)
 
Here's the config, so far:

OpenFiler VM inside of Workstation 6.5:
Single CPU
384MB of RAM
2GB boot drive, default partitions
40GB secondary (this is the VMFS3 target for the ESX4 hosts)
NIC1: 10.10.10.50 / 255.255.255.0 (bridged, replicating state)
NIC2 & NIC3: Bonded interfaces: 10.10.20.3 / 255.255.255.0 (bridged, replicating state)

Windows 2k3 Enterprise x86 VM:
Single CPU
384MB of RAM
20GB Boot drive
NIC1: 10.10.10.3 (bridged, replicating state)
NIC2: DHCP (bridged, replicating state)
functioning as 2k3 AD Primary PDC, DNS Server, and RRAS is turned on (routing between interfaces)

ESX1:
Single CPU (Intel VT exposed)
2GB of RAM
10GB boot drive default partitions
Service Console 1: 10.10.10.1 / 255.255.255.0 (bridged, replicating state)
Service Console 2: 10.10.10.10 / 255.255.255.0 (bridged, replicating state)
VmKernel: 10.10.20.1 / 255.255.255.0 (bridged, replicating state, marked for FT and VMotion, dual bonded GbE NICs)

ESX1:
Single CPU (Intel VT exposed)
2GB of RAM
10GB boot drive default partitions
Service Console 1: 10.10.10.2 / 255.255.255.0 (bridged, replicating state)
Service Console 2: 10.10.10.11 / 255.255.255.0 (bridged, replicating state)
VmKernel: 10.10.20.2 / 255.255.255.0 (bridged, replicating state, marked for FT and VMotion, dual bonded GbE NICs)
 
sabregen said:
you should fix this. :D what is the cause?
Click to expand...

don't know.

I should correct myself - it boots, takes about 4 hours to get past the bios screen, and then takes about another 4 to actually get to the PXE boot portion. Never mind actually booting an OS.

I'm trying it on a Nehalem soon - I'll report if it works better.
 
Shit. vCenter won't run on an AD DC? what kind of shit is this!? I'm hoping the company will spring for the new sager laptop:

Intel Core i7 2.93 w/HT
12GB of RAM
3x HDD's RAID-0/1/5
2x GTX 280m 1GB in SLi

That'd be killer for a mobile station
 
sabregen said:
Shit. vCenter won't run on an AD DC? what kind of shit is this!? I'm hoping the company will spring for the new sager laptop:

Intel Core i7 2.93 w/HT
12GB of RAM
3x HDD's RAID-0/1/5
2x GTX 280m 1GB in SLi

That'd be killer for a mobile station
Click to expand...

Really? That's news to me!
 
This brings up a whole new set of problems, to be quite honest (the fact that I can't run a vServer on a DC). Is this the case as well for vCenter 2.5, as compared to vCenter4? Our current internal infrastructure is that we have had (for some time) a PDC/BDC 2k3 AD setup. They were P2V migrated, and in the interest of not complicating the application load on those machines, VCServer 2.5 was deployed on an XP VM. I never tested the ability or inability to install it on a DC of any type. Can it be done?

Alternatively, we are bringing up a new set of servers for hosting solution, and the intent would be to run a single copy of Windows Server 2k3, acting as a PDC as a VM, and the BDC as a physical machine. We are doing this because we intend to leverage the AD auth'd roles of users to ESX4 for management purposes, for the clients. If we do this, do we need to have a separate Windows instance (in addition to the PDC & BDC) to run vCenter4? As an option to that, can we do an SLDAP primary/backup config (keeps us from buying Windows licenses, as we have a Premier Partnership with RedHat) for user ahtentication and role assignment? What are the potential "gotchas" for LDAP auth of users in this scenario?

On a side note, VMWare has to know that people are running these type of configs in a box...what is the hold up? for testing purposes, you'd think they'd release something for this purpose. heard anything, there?
 
Uh...this should be simple, but I can't figure out why I can't do something in the vSphere client:

I accidentally set ESX1 hostname to esx1.testdomain.local instead of tempdomain.local. I have ESX2 set correctly, and the AD server is tempdomain.local, as well as OpenFiler. However, when I go into the vSphere client, and go to the Service console under vswitch0, and try to edit the DNS settings, everything is greyed out. As I intend to have licensed HA, DRS, etc on this cluster, DNS resolution could be a big issue. I have already edited the /etc/hosts file for each of the ESX host machines, but if they are going to be controlled via vCenter, I would imagine this may create an issue.

How do I fix this?

Note that I have tried changing the default gateway device from vSwitch0 to vSwitch1, closing the vSphere Client session, and coming in on the Service Console 2 port, and it still would not let me change it.
 
Shit. Googling "vCenter4 on AD PDC" brings up this thread as hit #1....wonderful
 
Arg. I finally got vCenter machine built on another copy of 2k3 Enterprise 32bit. The first installation of the machine failed because I allowed the vCenter installer to use the local system account, which would not allow for the DB module to load up, and vCenter depends on that service to start, so it was an unending loop. I just rebuilt the VM this time, though, and it worked (providing the proper username/password for the DB portion fixed the issue).

More concerning is that this test will likely go no further. There is MAJOR disk contention going on with all of the basic services launched:

2x ESX4 servers
1x AD PDC
1x vCenter Server
1x OpenFiler iSCSI target

This is without any double nested VMs, at all. RAM usage is at about 3600MB/4096MB available. Everything just fucking crawls. I think if I was running SSDs in RAID-0, this would not be an issue, but on the current 5400RPM disks, all of the little seeks,reads,writes are just killing the host machine. Here's the utterly FUBAR layers that are surrounding the disk issues:

3x Fujitsu 5400RPM 2.5SATA w/8MB cache
Intel ICH7R SATA controller
Windows Vists x64 Ultimate + ICH7R driver calculating RAID-5 parity (which is using host machine CPU + RAM)
NTFS File system for Host OS
VMWare Workstation 6.5
- 10GB ESX1 containered file system (ESX1 will page out to swap...which is a triple nested file system)
- 10GB ESX2 containered file system (ESX2 will page out to swap, which is a triple nested file system)
- 2GB OpenFiler boot LUN (will also page out to swap, triple nested)
- 40GB OpenFiler iSCSI target LUN (will not page out, triple nested)
- 10GB 2k3 AD PDC (triple nested, will page out)
- 10GB vCenter (triple nested, will page out)

And on top of all of that, all of the host OS likely is paging out to disk while all of this is going on, as well, because of the high RAM usage from loading all 5 of the VMs. I can bring up about 4 VMs before everything just goes to it's knees. If I wait for all of the VMs to come up, the connection from the vSphere client times out while waiting for a response from the vCenter VM.

The RAM uage on the system is really high, and I can't install more than 4GB of RAM on this machine...but it would be enough for a proof of concept box, if the disk contention didn't just kill everything.

Anyone else have any ideas on how to relieve the disk contention? Maybe a nice, big and fast SSD that interfaces with Expresscard? The laptop has the following expansion options:

Expresscard 34/52 slot, 4x USB 2.0, and the 3x 2.5 SATA drive bays that are all currently being used by the 160GB Fujitsu 5400RPM drives in RAID-5, running the entire setup, thus far.
 
Okay, so I have the vCenter VM up and running. AD DC and vCenter are activated, and updated. vCenter has been licensed, and both ESX hosts have been licensed. Hosts have been added into vCenter view with the vSphere4 client. I have also double checked DNS configs, etc. Everything was ready to roll, until I got to launching the first nested VM. I found that I needed to add the following to the nested VM (XP Home) running inside of the virtualized ESX4 host, with it's disk living on the iSCSI VM:

vmx.allowNested = "TRUE"

Once I had that in the .VMX file for the nested VM, it powered on, and I no longer got the "Unable to run an VM in a VM" message. However, once it was powered on, I was not able to get a console to the VM. Here's the screenshot of the vSphere4 client connected to the vCenter4 VM, managing the ESX1 VMs. The console is attempting to be launched to view the VM running inside of the ESX VM. Here it is, but I have been unable to get past this:

Untitled-23.jpg
 
Can you ping and resolve the name it's trying to connect to? Usually that means it's a DNS problem.
 
All names respond to a ping, and all hosts are in the DNS config on the AD DC, and are capable of being looked up in reverse, as well.
 
try straight to the host. Also I wuldnt bother with ad. Just do host files for Dns. Improves the speed of the name resolution which will help later.
 
I will try ViClient straight to the host, and see if that works. However, the idea is that this entire physical box will be pulled out at clients offices to give them the Demo of vCenter's capabilities. Direct to host will negate the existence of the vCenter VM in the environment, and make the entire effort pointless

HOST file has been modified to reflect the environment.
 
attempt to get console view straight from vsphere4 client, to the host also failed, with the same error message. I have verified that none of the VMs have firewall enabled, at all. The host machine also has it's firewall disabled.
 
from PuTTY:

220 VMWare Authentication Daemon Version 1.10: SSL Required, ServerDaemonProtocol: SOAP, MKSDisplayProtocol: VNC

Popup window: Connection closed by remote host

Birdged networking on all single and double nested VMs, all replicating state
 
hmmm.
Well, the traffic is getting htere - has to be, since the telnet session worked. Hrm.
 
It's obviously rejecting the session for a security reason....but what reason?
 
that's not a security issue - that's us not connecting right.

Hang on, I'm building VI4 in a box and my system is slower :p she's a going as fast as she can!
 
Works fine for me. No AD / No DNS, using host files only. ~shrug~

I'm guessing DNS isn't working right.

edit1: host files on esx hosts / vc server both
edit2: I didn't have to do the allow.nested thing either :confused:
 
Last edited:
host files on each ESX machine have been created, and double checked. there is no host file on the vCenter VM, amybe I should look into that, and just nuke the AD PDC, and run RRAS (NAT) onthe vCenter VM (it's providing outside connectivity tot he virtual environment for NTP, Windows Updates, etc).

would require a slight rebuild on the vCenter VM, currently, it is the only machine that's a member of the AD Domain.

No connection issues from the host OS to the vCenter? Are your layers as follows, as well?:

Windows host OS > Workstation 6.5 > vCenter VM (managing 2x ESX4 VMs)

ESX4 VMs have a Vm running on them, stored on iSCSI VM

Authentication chain:

vsphere4 client > vCenter VM > ESX host > console session to VM deployed on ESX4 VM

?
 
You must log in or register to reply here.
Back
Top