VirtualClouds.info

Today I was playing with a scripted installation of ESXi 5.1 in my lab. I noticed that IPv6 has been enabled by default. And because I don’t use IPv6 I tried to disabled it via the DCUI.

When you disable IPv6 you need to reboot the ESXi host before the changes are applied as you can see in the screenshot bellow. No big deal I can do this later when I’m done. So I exit the IPv6 configuration screen and applied the network configuration changes.

And as soon as the changes are applied I had a little surprise when I saw a reboot screen

So be aware that you don’t change this setting on a production host during office hours

Lately when I was trying to build my vLab for the upcoming VCAP exams, I had a little challenge to get HA working on my two node cluster. Because this vLab is running on top of my existing vSphere 5 environment, I have assigned no more than 2GB of vRAM to each ESXi 4.1 host.

Installation and the initial configuration went all fine, until I tried to add the ESXi host to the vCenter Server; I received a warning message “Cannot complete the configuration of the HA agent on the host”.

Okay, that’s not much information but after a quick look in the Task & Events tab I saw that the installation of the HA agent won’t start because the ramdisk can’t be created, see the screenshot below.

The next step I have taken is a quick look at the ESXi host itself, when installing the HA agent you can track the installation process by viewing the /var/log/vmware/vpx-iupgrade.log.

As mentioned above here, the vCenter gives an error message that the ramdisk can’t be created. On the ESXi host you can see the same message but with some extra interesting information, it states that there is no more space to create the ramdrive.

Okay, that’s cool because according to the VMware documentation 2GB should be enough. However VMware says this in there documentation (ESXi Installable and vCenter Server Setup Guide, page 13): “2GB RAM minimum. For upgrades, 3GB RAM is required if the ESXi host is managed by vCenter Server”.

So is it 2GB or 3GB when the ESXi host is managed by the vCenter Server? Well when I see the above error I will believe it should be more than 2GB So I shutdown the ESXi hosts and add 1GB extra to the virtual machine, so in total the virtual ESXi hosts will have 3GB of RAM available.

Now that the ESXi hosts have more RAM I tried to add the hosts to the vCenter Server and the HA/DRS Cluster, and it actually worked, the hosts are now added to vCenter and the HA agent is active

Today a co-worker called me and told me they where testing something out and they stumbled on a little issue. Here’s what they did, they where trying to start 125 VMs at once on an ESXi 4.1 host to do some load testing.

This al works just fine when they use a single vCPU for each virtual machine, they could start all 125 VMs at once, no problems at all.

However when they where doing the same test but now with an extra vCPU (so in total they assigned 2 vCPUs for each virtual machine) they couldn’t start more than 100 VMs.

The other 25 where unable to start and the received a error message that the system could not power on the VMs (look at the screenshot below).

Weird? well not really, especially if you look at the error description. You could already guess what’s going wrong here, it looks that they have reached some limit in the amount of used vCPUs on the ESXi host.

To solve this little issue you need to change the maximum number of vCPUs to for example 512 vCPUs.

To increase the maximum number of vCPUs to 512:

1. Launch the vSphere Client.
2. Click the host.
3. Click the Configuration tab and then go to Advanced Settings.
4. Go to Vmkernel > Boot > maxVCPUS.
5. Modify the parameter to 512.

Check out this VMware KB article to find other ways to change the maximum number of vCPUs.

Thanks, goes to Krishan Khoesial who has worked this issue out.

UPDATE: Big thanks to Eric Sloof, Arne Fokkema, Alan Renouf and William Lam. They have pointed my at three additional ways to turn on SSH, I have added them to this post to make it even more complete, thanks guys!!!

Yeah I know, there are many blog post out there that describe how to enable SSH on an ESXi host, so why write another one? Well actually I have two reasons for it, first there are several ways to enable SSH and secondly I want to talk about the consequences when you enable SSH, because I think it need some more attention than just show you how to enable it.

So I decided to dig a little bit further in this matter, and provide you with some extra information, and hopefully some guidance to leave SSH turned on or off, but we will get back on that subject later in this post.

Like most of you already know it’s possible to access an ESX (classic) host through SSH with help of Putty or some other kind of SSH application, and login to the Service Console.

But with the introduction of VMware ESXi this have changed a little bit, because the ESXi host won’t have a Service Console, there’s no way to login with SSH, or is it? Lucky for us ESXi still have a so called Busybox (a very small “Service Console” but very limited in functionality) so with some tweaks it is still possible to use SSH for accessing an ESXi’s host, but remember it’s not the full blown Service Console like ESX have.

Since the release of vSphere 4.1 it is possible to enable SSH aka “Remote Tech Support” via the vSphere Client to make it a lot easier to enable SSH, more about his in option two.

Read more…

Before we start with configuring CDP on a ESX(i) hosts, let’s first take a look at what CDP realy is?

According to Cisco; The Cisco Discovery Protocol (CDP) is primarily used to obtain protocol addresses of neighboring devices and discover the platform of those devices. CDP can also be used to show information about the interfaces your router uses. CDP is media- and protocol-independent, and runs on all Cisco-manufactured equipment including routers, bridges, access servers, and switches.

Okay, now thats more clear (I hope so ), lets take a look at how we can use CDP within ESX(i) because, hopefully as you all know by now, VMware ESX(i) is capable of using the Cisco Discovery Protocol (CDP).

By default the CDP configuration is in Listen only mode, this means that the ESX(i) host can only recieve CDP information like for example; on what switchport and switch it is connected, type of switch, VLAN, routing support etc. This is the same on a standard vSwitch as well as the distributed dvSwitch.

This is fine for an Administrator if he wants to know on what switchport an ESX(i) host is connected, but perhaps the network folkes also want to know the same information only from their perspective. Because the default setting is on “Listen”, the ESX(i) hosts won’t send any CDP information back to the switch, so if a network guy logs in his switch and execute the command: “show cdp neighbors” he won’t get any information that there is an ESXi hosts with a vSwitch or dvSwtich connected to a switch port.

So to enable CDP on “both” sides we need to configure CDP to Listen and Advertise at the same time so both parties have the information they need.  But before we start configuring CDP, lets first take a look at what configuration options we have available.

Read more…

Because off the daylight saving time changes here in Europe I think it would be nice if I posted a quick ”how to” about NTP configuration on a ESXi hosts.

There are many documents out there that describes why you need a reliable time source like Timekeeping in a VMware Virtual Machines.

However not every company have their own timeserver (NTP server) somewhere in their datacenter, so they must rely on a public time source. Lucky for us there is a good and reliable public time source project called the “NTP Pool Project“.

To setup a NTP source on a ESXi host I use the vMA (Virtual Management Assistant) and use the vifcg-ntp command. Bellow I will add the NTP server called nl.pool.ntp.org to the ESXi hosts esx01.virtualclouds.info.

Read more…

HP made there customized VMware ESX 4.1 and ESXi 4.1 available for downloading.

Check out the following sites to download the ESX 4.1 and the ESXi 4.1 ISO  files.

Read more…

As you might know it’s possible to roll back an ESXi “image” to a previous version. For example you’re updated a ESXi host with a newer version of ESXi and for some reason the host is not running correctly, it’s possible to re-install the host or to roll back the ESXi image to a previous version that’s stored on the alternate boot bank of the SD/USB/HDD disk. Here’s how you do this.

Read more…