News

vmpmotion.pl updated

It took me a while, but I finally figured out why vmpmotion.pl stopped working on the RCLI that coincides with VMware ESX 3.5.0 Update 2. It turns out that a regular expression that I was passing was no longer valid. In any case if you go download the script now to perform command line storage vmotions it should work just as it did in the older versions of the RCLI.

Labels: vmpmotion svmotion

Posted by Dominic Rivera at Saturday, October 18, 2008.

Backup Comparison Updated with lastet Versions

Many things were announced at or before VMworld, Backup Tools were no different. Each major vendor has new versions of their software and this update to the backup guide looks at these new versions. Specifically version 2.0 from Veeam, 3.1.18 from PHD, and 3.4.2 from VizionCore.

Look for more updates to this guide as new tools and versions of the products are released.

Visit: http://vmprofessional.com/index.php?content=esx3backups

Posted by Texiwill at Wednesday, September 24, 2008.

snapalert.pl updated for ESX 3.5.0 Update 2 RCLI

A couple of readers wrote in to tell me that the new RCLI ( from Update 2 ) was incompatible with snapalert.pl. I had actually already changed how it worked ( it now scans an entire VirtualCenter vs a single Datacenter ) and the new version worked without any modifications in the new RCLI. The new version is available in the same place as the old one.

Enjoy.

Posted by Dominic Rivera at Saturday, September 13, 2008.

VMProfessional releases: vmpimlscan.pl

One of the downsides of moving from VMware ESX "Classic" to the svelte VMware ESXi Installable/Embedded is the loss of functionality of the service console, and the abiltiy to install 3rd party agents. When looking at the problem in depth, the only 3rd party software that was essential to how I worked on a day to day basis was the HP SIM agents which provided some hardware alerting that was hard to reproduce without them. To work around this problem, I developed a lightweight perl script (vmpimlscan.pl) to alleviate the burden of this loss. In a nutshell, vmpimlscan.pl has the ability to gather all of the host information from VirtualCenter, and then talk to the ilo interfaces of those hosts and read the IML ( Integrated Management Log ) and parse the output looking for hardware failures. The script can then send the alerts through email ( or an external program like openview ) to a ticketing system for automatic alerts and possibly automatically generated tickets. The script will also be a part of Amir Sharif's VMWorld 2008 session PO1520: Managing VMware ESXi in the Datacenter. I've now published the script so that it is available for download.

Unfortunately, I won't be able to make the VMworld 2008 session due to a scheduling conflict, but I will be around VMworld 2008, so if you have questions about it and will be in Vegas next week drop me a line.

Posted by Dominic Rivera at Saturday, September 13, 2008.

ESX Lolcat

One of my co-workers made this as we ( they actually, I'm out on paternity leave ) worked on fixing the bug in ESX 3.5.0 Update 2 where the build times out on August 12th causing VMs not to power on, be VMotioned etc ( general havoc ). The cat is funny, but the problem... not so much.

Posted by Dominic Rivera at Wednesday, August 13, 2008.

VMware Enhanced VMotion Compatibility

With VMware ESX 3.5.0 Update 2, the VMware guys slipped in a feature named Enhanced VMotion Compatibility that will really prove to make virtual machines portable between hardware generations, potentially making hardware upgrades seamless to your end users.

Enhanced VMotion Compatibility ( EVC for short ) is turned on at the cluster level, and it basically sets up a baseline ( lowest common denominator of CPU instructions ) for that cluster so that the entire cluster only has a certain instruction set available for use, and masks out any new instructions that might be added in a new processor family. You must stay within the same processor vendor ( ie, AMD or Intel ) and you must have the same processor features turned on in the BIOS ( such as AMD-V and NX for AMD and XD and VT for intel ). The functionality is brilliant, but here is the rub: all of the VMs in the cluster need to be powered off to turn on this functionality. Even if all of your hardware is exactly the same.

The workaround for this is the following: Create a new cluster, enable EVC on the new cluster and then one by one put your hosts into maintenance mode evacuating all of your VMs and put the host into the new cluster, then manually VMotioning all of your VMs from one cluster to another. Using this workaround you lose a few things though: Cluster performance history, migration history, HA and DRS settings will have to be manually copied over and then lastly and most annoyingly all of your affinity and anti-affinity rules will need to be rebuilt on the new cluster.

The lesson to be learned here is that if you're setting up a new cluster, make sure that you enable Enhanced VMotion Compatibility from the start. It's a minor amount of work at the beginning of a cluster that can save you a ton of time later on and there is really no drawback to enabling it.

Labels: VMware ESX 3.5.0 Update 2 EVC Enhanced VMotion Compatibility

Posted by Dominic Rivera at Saturday, August 09, 2008.

ESXi Embedded and HA

I have a feeling that not a lot of people have run into this issue yet, but as ESXi becomes more mainstream, this problem will crop up more and more. You see when you have an ESXi Embedded host that has no storage ( most likely a blade ) and you try to add that host to a HA cluster you get an error message similar to the following:


HA agent has an error : Host in HA Cluster must have userworld swap enabled

It turns out that you can't add an ESX host without swap enabled into an HA cluster, so you need to turn to the advanced options of the host and set ScratchConfig.ConfiguredScratchLocation to point at a valid datastore with more than 1GiB of space free. You can see the fix in this VMware KB Article