vOllfried

Orphaned Port Groups after migrating to distributed vSwitch

Oliver Antwerpen — Mon, 28 Nov 2011 10:18:25 +0000

I came across this today and it took me quite a while to figure out what happened. I migrated a vSphere 5 environment from vNetwork Standard Switch to vNetwork Distributed Switch. Afterwards I deleted all old Standard Switches, but I had two Portgroups in Home->Inventory->Networking with Virtual Machines connected to them. Strange, as the vSwitches were gone.
After checking logs and playing around I found that these VMs had a snapshot. After deleting the snapshots the orphaned Port Groups disappeared, so I can now head off to HP Discover in Vienna.

vCenter Installation fails with “Error 28038. Setup cannot create vCenter Server Directory Services instance.”

Oliver Antwerpen — Mon, 15 Mar 2010 20:26:05 +0000

I ran into this problem today while installing VMware vCenter Server 4.0 Update 1 on Windows Server 2008 SP2. In addition, the error said “Refer to C:\\Users\...\AppData\Local\Temp\...\jointool-0.log for more details."

Unfortunately, there was no jointool-0.log. I tried several steps from Duncan’s Blog Post, and also from VMware’s KB articles 1015887 1010938 and 1013822, all with no success. Removing the server from the domain and rejoining it did not help, setting permissions on the root-drive had no effect and even creating the AD LDS instance for vCenter was unsuccessful.

After doing some more research I came across the fact, the using a semicolon “;” in the databases password could cause errors – and that was the solution. After changing the vpxuser’s password on the database installation went throug and I had (nearly) no problem.

Exclude unused NICs from hp management tools

Oliver Antwerpen — Thu, 18 Feb 2010 10:00:30 +0000

Your new Server has more NICs than you need? No problem. As long as they are not connected to a vSwitch VMware won’t bother with them. But if you install HP’s management agents and take a look at the System Management Homepage or integrate your ESX into System Insight Manager, you will see a warnings for each NIC that has no link.

I wrote a small script that will disable these unused NICs:
disable_unused_nics.sh
Here’s how to install it:

Save this file as /usr/local/bin/disable_unused_nics.sh
Edit the file an adjust the unused NIC numbers
Make it executeable with “chmod +x /usr/local/bin/disable_unused_nics.sh”
Open /etc/init.d/rc.local and add the line “/usr/local/bin/disable_unused_nics.sh” at the end of the file

Now, when you reboot the host, or manually call /etc/init.d/rc.local you will see the following:

Afteer that, check your SMH:

That’s all.

Using a FC Tape Library inside a Virtual Machine

Oliver Antwerpen — Wed, 17 Feb 2010 10:00:57 +0000

A costumer uses a HP MSL6000 Tape Library for his Backups. The library is direct attached to a server via SCSI, but now we are moving towards VMware. What to do? Use a NSR and get that thing into his SAN.

This is what it looks like: You can plug two drives even in the small 6030, but – more interesting – a NSR. A Network Storage Router is a device, that connects SCSI devices to a SAN. Easily spoken, you just unplug the SCSI cable from your server, plug it into the NSR, connect the other end of the NSR to your SAN switch and just configure the device. But, there is one problem I ran into: The drives just were not recognized by the ESX hosts. The solution for that is quite easy, if you just know it: A tape has to be loaded into each drive to get it discovered.

That means you need to follow a startup procedure like this:

Load tapes to all drives in your library (via Webinterface or Control Panel)
Boot up ESX hosts or rescan FC HBAs
Create Raw Device Mappings to you virtual Backup Server
Boot the VM and check device access
Unload tapes from drives

You will also need to follow these steps when you reboot your VM or host.

This is what it looks like:

Does your ESX Server really get its time via NTP?

Oliver Antwerpen — Mon, 15 Feb 2010 10:00:04 +0000

If you want to check NTP on your ESX host, don’t rely on vSphere Client! Even if it looks okay it might not be working.

These days I installed some new ESX4 hosts at a customer’s site, ESX 4.0 Update1 to be accurate. As you might know, you can configure an NTP server during the installation phase. We did that, installation went straightforward and was done within a couple of minutes.

After the installation and getting the host into the new vCenter I wanted to show the customer that NTP really works. So i broght up the Configuration->Time Service Tab and we saw that the correct NTP server was configured and the services was running.

Then I went down to the service console and had a look at the time service:

[root@vm0011 ~] ntpq -p
localhost: timed out, nothing received
***Request timed out

When you look at /etc/ntp.conf you see these lines:

restrict default kod nomodify notrap nopeer noquery
restrict -6 default kod nomodify notrap nopeer noquery

Comment these lines out or delete them an restart the service with

/etc/init.d/ntp restart

After that check the status like this:

[root@vm0011 ~] ntpq -p
==============================================================================
 ntp1.rz-ip.net  192.53.103.104   2 u    4   64    1    1.098   -0.305   0.001
 ntp2.rz-ip.net  192.53.103.108   2 u    3   64    1    0.725   -0.562   0.001

Great, now it’s working!

Strange problem with MSA1000 and VMware

Oliver Antwerpen — Tue, 02 Feb 2010 21:30:17 +0000

I am just investigating a strange problem on a MSA1000 and ESX3.0. VMFS seems to become corrupt, even though I can write to it.Yes, I know. The MSA1000 is old and it’s ugly. But – quite some customers still use it. One of them is even worse and uses it together with the integrated MSA SAN-Switch 2/8, that’s an eight-port 2Gbit FC Swicth that’s plugged into the back of the Chassis. Attached to the MSA1000 there is a MSA30 which holds some HDDs for more space.

The disks in the MSA1000 are devided into four arrays each holding a single LUN. The MSA30′s HDDs are a forth array with a forth LUN. The ESX-Servers use LUN1+4, LUN2+3 are connected to windows servers.

All of a sudden all VMs on LUN1 threw errors and we were unable to start them up. VMware could not open some of the files residing in that datastore, VMs could be cloned to that datastore successfully, biut not be powered om, the VMDKs were corrupt.

We restartet the MSA, recreated array and LUN, updates ESX to 3.0.3, recreated VMFS, still the same behaviour. I am runnning out of ideas, we’ll see what comes up next. Right now the VMs reside on an iSCSI target hosted by a linux server.