“General fault caused by file” when deleting files in Vmware ESX 3.5 (Build 207095)

You are browsing a datastore in ESX in the objective to delete some files or folders.

When trying to delete the files or folders in the storage, you get the following error:

“General fault caused by file

/”

image002

image003

This error actually means that the file (or a file in the folder) is being used by another process.

To make sure, have a look to your configurations:

· Is the file being deleted a vmdk file that is mounted on a virtual machine?

· Is the file being deleted an iso file being mounted on a virtual machine?

Once you have checked your settings, you will have probably found the file is in use.

Ensure yourself you really don’t need to use the file, remove the blocking configuration, then delete the file.

 

15 Steps to Migrate a Windows Server 2003 MSCS cluster to Vmware ESX

We had a legacy application running on an MSCS cluster, and because of some process reasons, I was given 5 days to migrate it to another equipment. The challenge was to keep the same installations and configuration, but just release the out-of-support boxes and SAN.

The boxes were IBM x346 servers and the storage SUN Storedge 6920. There were 6 different SAN volumes connected to the cluster, each less than 100GB.

I directly thought of virtualizing the server as we have a set of Vmware infrastructure ESX 3.5 servers. They are not connected to any SAN.

Through Google, I found two KBs from Vmware: http://kb.vmware.com/kb/1002661 and http://www.vmware.com/pdf/vsphere4/r40/vsp_40_mscs.pdf but still was not sure on how to correctly set the virtualized environment copied with same cluster resources.

Continuing my research, I found this particular doc from a blessed one who still wants to be called evil, I don’t know why.

http://communities.vmware.com/docs/DOC-9080

Anyway, it helped me to build my own procedure as follow:

1. Move all cluster resources to Physical Node1 and import OS partition and any non-shared (non-Quorum) drives using VMWare Converter. Choose not to install VMWare tools at this time.
Here are the disks I had at the beginning:

2. Import Physical Node2 too in same conditions.

3. Start the VMs:

a. Make sure the network is disconnected and turn on virtualized nodes

b. Install VMWare Tools without VSS option, restart the VMs (I found that it helped to correctly have the new NICs to configure)

c. Clean up nodes IP addresses values in the registry, but leave all IP addresses that concern the cluster itself (you may just find the different entries with your nodes IP addresses to complete this, and delete the value)

d. Assign the IP addresses to the virtualized nodes

4. From a VI client, create the folder to host the cluster disk volumes (i.e: “ClusterDisks” was created directly in the root of the available storage)

5. Use a shell tool to access the ESX server and create the cluster disk volumes using command lines (I usually use putty)

a. Make sure of the virtual machines files path by typing command (I assume you have enough rights):

vmware-cmd –l

This gives:

[root@ESX05 root]# vmware-cmd -l

/vmfs/volumes/4ae8e9f9-5ab87043-a544-*****dd27e16/vm1/vm1.vmx

/vmfs/volumes/4ae8e9f9-5ab87043-a544-*****dd27e16/vm2/vm2.vmx

/vmfs/volumes/4ae8e9f9-5ab87043-a544-*****dd27e16/vm3/vm3.vmx

b. Create the quorum and other cluster disk volumes using a similar command to this (“1g” represents the disk size):

vmkfstools -c 1g -d eagerzeroedthick -a lsilogic “/vmfs/volumes//vmfs/volumes/4ae8e9f9-5ab87043-a544-*****dd27e16/ClusterDisks/Disk_Q.vmdk”

Creating disk ‘/vmfs/volumes/4ae8e9f9-5ab87043-a544-*****dd27e16/QUORUMS/Disk_Q.vmdk’ and zeroing it out…

Create: 100% done.

In my case, I used this opportunity to extend the disk volumes.

6. With the VMs off, ON EACH VM:

a. add HD to VM, but choose SCSI (1:0) which will create a new SCSI Controller

b. On SCSI (1:0), change to LSI logic, and set to “Virtual”

c. Add all your other cluster disks to the same SCSI controller (1:1, 1:2, etc.)

7. Failover cluster to Physical Node1, turn off Physical Node2

8. Bring online Node2 VM

9. If needed, in the Windows OS DiskMgmt.msc console, rescan the disks to see the Cluster shared drives. Then format the drive and assign the same drive letter as the old Quorum drive.

10. Use Robocopy to copy quorum and all other cluster disk volumes from Physical Node1 to Node2 VM.

robocopy.exe v: %clusterdisk% /copyall /s /e /sec /R:1 /W:1

11. From Windows 2000 resource kit (or a download), use DUMPCFG to get the disk signature of the physical Quorum drive from one of the physical nodes

>dumpcfg_x86.exe

[System Information]

Computer Name: ***********03

Cluster name (DNS): ***********03.*********.***

Cluster name (NetBIOS): **********03

System Root (install directory): C:WINDOWS

OS: Windows 2000 Server

Service Pack: Service Pack 2

Product: Windows 2000 Advanced Server is installed. [Note that it show Windows 2000 even if your install is Windows 2003!]

Terminal Services is installed.

[DISKS]

Disk Number: 1

Signature: 00D9FC05

…

12. From the VM, overwrite the signature with dumpcfg command (note that the Disk Number may be different on the Physical Node compared to the new disks on VM, so be careful)

dumpcfg -s %signature% %disk number%

ex: dumpcfg -s 00D9FC05 2

13. Shutdown Physical Node1

14. Restart cluster services or the Node2 VM… if all goes well the cluster should come up

15. Bring online the Node1 VM

YOU ARE DONE!!!

Notes:

I encountered many issues during the Conversion. The following links may be helpful if you are in the same situation.

The VMWare Converter Fails at Reconfiguration step:

http://networkadminkb.com/kb/Knowledge%20Base/VMWare/P2V%20of%20Windows%202000%20Reconfiguration%20failed.aspx

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1006296

The sysprep error on reconfiguration fixed:

http://vmetc.com/2009/08/18/how-to-add-sysprep-to-vmware-vcenter-for-vm-customizations-vmware-converter-also/

http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&docType=kc&externalId=1005593&sliceId=1&docTypeID=DT_KB_1_1&dialogID=31097134&stateId=0%200%2031095900

http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=1005593

An interesting post highlighting that chkdsk c: /f would be helpful before converting the SOURCE machine :

http://kb.vmware.com/selfservice/microsites/search.do?cmd=displayKC&externalId=1006559

 

Performance Chart : ESX 3.5 and vCenter 2.5 have different options

When it comes to monitor performance on ESX servers, VMware has had the good idea to integrate charts that can help you check CPU, Memory, System (uptime) and Disk health.

But using VI client through ESX 3.5 or through vCenter 2.5 to check them, you would have different levels of information.

Let’s have a look.

In the VI Client, to reach the performance charts, you should click on “Inventory” then browse on the left side of the window to the ESX server name or IP.

When the server is highlighted, on the right side of the window, you will find “Performance” tab. From there, you can see the real-time statistics by default.

You can notice the “Change chart options” link: click on it to see the performance chart configuration options.

Here you have a screenshot of that options on a VI Client connected to an ESX 3.5 server:

As you can see, you can only view PastDay stats or Real-time stats.

Now let’s see the screenshot on vCenter 2.5 server:

You have much more possibilities: Past Week, Past Month, Past Year and even Customized stats !

So as you can see, the level of information you can get for your servers is different, whether they are clustered or not.

 

Unable to install Windows XP on ESX 3.5

On ESX 3.5 update 5, you are trying to install Windows XP on a virtual machine.
When starting XP installation, you notice that the setup fails with an error: “Setup did not find any hard disk drive installed in your computer”.
This is a known issue. To install Windows XP on an ESX virtual machine, you should first load the hard disk drivers using a floppy disk in the virtual machine.
I have not created any floppy disk image myself, so I suggest you download them from a website I found when looking for a solution.

Please download either the LSI or BUSLogic controller from that website, load it to the virtual machine as a floppy disk (in the Virtual Machine settings).
You may then start the installation and when prompted to press F6, please do so to add some additional drivers. The installation should work after that.

I did use the LSI Logic controller perfectly.

 

Add a SAN volume to an ESX server

Type the following commands, connected to the ESX through a SSH console:

kudzu

esxcfg-rescan vmhba0 (run this twice)

esxcfg-rescan vmhba1 (run this twice)

—Detects new hardware.—

esxcfg-vmhbadevs

—list the devs—

fdisk -l

fdisk /dev/sd?

—? represents the right device; select the new disk /dev/sdc most like it will be sdc or sdd —

p

—print the partition info for the disk—

n

—initiate new part def —

p

—primary type—

1

—partition 1—

enter

—start at block 1 default—

enter

—end at last block default—

t

—set type—

1

—select partition 1—

fb

—set type to fb = vmfs—

w

—write out the table—

esxcfg-rescan vmhba0

esxcfg-rescan vmhba1

—rescan again—

—example of expected result….—

—Device Boot Start End Blocks Id System—

—/dev/sdc1 * 1 65270 524281211 fb Unknown —