VM Template for Citrix Provisioning Server

A rather odd bug with VirtualCenter is the inability to deploy VM's with a SCSI controller but without a disk... even creating a template from a VM that has a SCSI controller but no disk results in a template with no SCSI controller (you then have to convert it to a VM, re-add the SCSI controller, then convert back to template. Only to find the SCSI controller stripped during the deploy from template. There's 10 minutes of my life I'll never get back).

If you've been reading this blog for any length of time you'll know I'm not exactly a fan of Citrix XenServer, but Citrix Provisioning Server on the hand is _very_ cool (albeit prohibitively expensive and not without it's drawbacks... another post maybe). For the uninitiated, Provisioning Server streams a disk via the network. But obviously the streamed disk needs to be access via a disk controller (duh!). Which is why I want to create a diskless template in the first place.

Before firing up the trusty PowerShellified version of Notepad2, i did a cursory search of the VI Toolkit forum to see if someone had done this already... and surprise surprise, Cool Hand LucD had done my work for me.

So yeh, now all I need to do is deploy my scsiless and diskless template with new-vm and call the function from LucD, and all is well in the world - big ups to LucD.

/me makes W shape with fingers on one hand while simultaneously pounding fist on chest with the other

Future ESXi Version Call to Arms!

While there wasn't a lot of concrete details that came out of VMworld, one thing is for certain - VMware are working on a _lot_ of new functionality for their next major release. Of particular relevance to us at vinternals is the host profiles feature, as it provides the same functionality as Statelesx, but is done in a much more user friendly way. Which is fine - it's not like we threw in out jobs and tried to position statelesx as a commercial product, and in some ways it's a validation that our idea was a good one.

But VMware need to ensure they implement the other half of the solution, ala the client initiator. Without this, there is still an unnecessary manual task of joining a host to a cluster in the first place. You may think this isn't such a big deal, but if your hosts are PXE booted then a client initiator is an absolute requirement.

Fortunately for us, there is at least one senior guy at VMware who gets all this - Lance Berc (I get a laugh eveytime I see his 'novice' status in the VMware forums - for fucks sake, this is the guy who wrote the original esxtop). So far most attention has been on his post-build configuration scripts, but Lance put his C skills to work and released the crucial client initiator piece as well as a 'midwife' which essentially performs the same duties as statelesx. Put all this together and you have a truly automated, scalable, liquid environment - the likes of which is usually only seen in compute clusters. Except this time it's a _virtual_ environment.

VMware seem to be taking notice of the need for cluster wide configuration automation with host profiles and the distributed virtual switch, we as customers need to make sure they understand the need for the client initiator functionality as well. The 'midwife' functionality would also be nice if it was wrapped into VirtualCenter, but if it isn't then maybe Statelesx will have some longevity after all, even if it is somewhat more limited to automating the application of a host-profile that is defined elsewhere.

So here's the call to arms. We as customers have the strongest sway with regards to product features. And with a new version of the platform in development, now is the time to start hitting our account managers inboxes with requests for this functionality. Hell, just send 'em a link to this post if it makes it easier - the important thing is to push this through the account channels, not Lance (you'd just be preaching to the converted), and the time to act is now!

VM Reconfiguration via the VI Toolkit for Windows

Reconfiguring VM's via the API is something I've had to dabble in lately, due to some.... err.. "interesting" behaviour with deploying VM's from templates. Such as the inability to deploy a VM from a template that has a SCSI controller but no disk - useful if using LUN based snapshots, where the .vmdk already exists and you want to move said vmdk into the same directory as the VM after it's been created (and thus can't use the -Diskpath option of New-VM as the path doesn't exist yet). Or if you want to create VM's for use with Citrix Provisioning server, which don't require a local disk but do require a SCSI controller.

While this operation is uglier than I'd like it (most things that involve Get-View and specifications are uglier than I'd like them :-), I found this very good paper which explains the VI object model as applied to VM hardware reconfiguration _much_ better than the SDK documentation does.

Sure the examples are in Perl, but the theory is the same. Combine the document with a listing of all posts by Carter, LucD and halr9000 in the VI Toolkit for Windows community and there's nothing you can't do!

Hyper-V Server - Microsoft turns platform arse-about

After a combination of holidays and letting the VMworld blog storm pass, vinternals is back!

For those who may not understand the phrase "arse-about", it means backwards. Which is exactly what Microsoft have done with the upcoming release of Hyper-V Server. The marketing around this product seems to have been a bit lost, so let's recap exactly what this product is (and is not)

- Hyper-V Server is NOT running Windows Server Core 2008 in the parent partition
- Hyper-V Server is aimed at small / dev environments, not enterprises
- Hyper-V Server is running little more than the Windows Server driver model and the virtualisation stack in the parent partition
- Hyper-V Server has the same underlying hypervisor and virtual bus architecture as the Hyper-V found in full-blown Windows Server 2008

So why have Microsoft got the platform ass-about (US English :-)? Because their own best practice recommends keeping the parent partition as lean as possible from an application perspective. So why the hell aren't they keeping the parent partition as lean as possible from an OS perspective? Windows Driver model + virtualisation stack - that's what I want in an ENTERPRISE Microsoft virtualisation platform! I don't want to be running something in the parent partition that is subject to patches like this one! But instead Hyper-V Server is being positioned as the red haired step child of the Microsoft virtualisation offerings.

I'll still take a look at Hyper-V Server when it comes out... there are some fundamental infrastructure design differences that may not be so obvious to those of us who have been designing VMware infrastructures for a while, that may still be applicable to Hyper-V Server... sounds like a good topic for SAAAP :-)

Qumranet Acquisition - what does it mean for SPICE?

Sure I'm the 100th blogger to chime in on the Red Hat acquisition of Qumranet, however one thing that seems to be escaping the press is the completely proprietary nature of the SPICE protocol, which is really what made Qumranet stand out from the other KVM based solutions on the market (that and the fact that they were the sole commercial sponsor of the KVM project, the same that Parallels is the sponsor of OpenVZ).

I met Moshe Bar earlier this year, and asked him if SPICE was intrinsic to KVM, and if not why not license it to other vendors in the VDI space. His reply was that they could indeed re-engineer and license SPICE to any vendor, however doing so would pretty much kill the rest of their offering. If Red Hat open source SPICE, there could be some interesting developments in the remote display protocol / broker space indeed. I don't know how many of you have seen SPICE running live, but it's pretty darn impressive!

PowerAlarm Manager Script

Check out Shyam's entry in the (now closed) VI Toolkit for Windows Scripting Contest, PowerAlarm Manager. It allows you to create alarms on any item in the VC inventory based on any perf metric. This is way cool - I've never quite understood why this functionality is not natively available. The Ready Time metric for example, is very useful as I'm sure anyone reading this blog knows. But we've never been able to create alarms based on it until now, thanks to Shyam!

So here's hoping he gets a placing in the top 3, although with some very good entries in there it will be tough. We'll post back any results as they come to hand, but in the meantime you should check it out and please report any bugs back to us!

Statelesx 1.1.0 Released!

Yes yes, hard to believe but we have had a great response from the initial release from people all over the globe, including some improvements that we thought should go in straight away!

In this release, we have added:

- hosts that contact the appliance via the python script now are placed into an "unregistered hosts" section, making it very easy to add them to clusters (ie no more typing in hostname and UUID)

- hosts that are defined in clusters can be configured via a single button click in the web UI, so no python script is needed on the host. Yes, that means you can now configure ESXi boxes. You still need VirtualCenter though, and it's obviously not a completely hands off process since you still need a single button click to initiate the configuration, but we're working on that.

- web interface now located at http://[IP.Address.of.Appliance]/

Credentials are the same as the initial release (root / root for console, statelesx / statelesx for web interface). The new md5 sum is:

98aaea31137048338a547a56dd2a65c2 statelesx-1.1.0-ovf.zip

The python script also has a minor update, now 2 parameters are sent across:

#!/usr/bin/python
import socket
import re
import sys
#Open esx.conf, read only
esxconf = open("/etc/vmware/esx.conf", "r")
#Define regular expressions
getName = re.compile("/adv/Misc/HostName.=.(.*)")
getUUID = re.compile ("/system/uuid.=.(.*)")
#Search esx.conf for matches
for line in esxconf.readlines():
    hostname = getName.findall(line)
    for name in hostname:
        name = name.strip('"')
    UUID = getUUID.findall(line)
    for uuid in UUID:
        uuid = uuid.strip('"')
#Close esx.conf
esxconf.close()
#Define statelesx server address, port and msg
host = sys.argv[1]
port = 1974
msg = name + ":" + uuid
#Open socket
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
#Connect socket
s.connect((host,port))
#Send msg
print "Sending " + msg
s.send(msg)
# Close socket
s.close()


We hope you enjoy this release, feel free to mail us with any feature requests / bug reports or leave them in comments section of this post!