Difference between backup and archiving

Archiving is a process of moving inactive or infrequently accessed data from primary storage to other tiers of storage. Usually, it is moving from expensive to cheaper storage. For example, from expensive SAN lun to a NAS device or optical media. The primary information is moved to other media in this case. Backup will create a second copy of the information in another media such as tapes. So archiving the data does not eliminate the need to backup the data.

You can setup rule to archive you data. It maybe based on last access time, file type, etc. Once the file are written to secondary storage, a stub (file tag) is left on the primary storage to point to the secondary storage. Thus, space is freed up from the primary storage. When user accesses the file, the archiving software will retrieve the file from secondary storage. The stub has all the info about the location of file on the secondary storage.

Therefore, administrator should have a backup strategy on backup because archving does not keep a second copy of the file. Depending on the secondary storage, they can backup the secondary medias or running incremental backup on the primary storage. Check and make sure the backup software is compatible with the archiving software. Some of the backup software are not extended attributes aware and will cause a lot of issues.

Installation of VMWare

I try to install VMWare ESX Server on my HP DL360. I got it from ebay for CDN 800 with 4GB of RAM. Installation is very similar to Linux. However, I got a strange error message about checking BIOS settings.

vmkernel: 0:00:00:00.000 cpu0) ALERT: Chipset: 433: no PCI entries in MPS table - check BIOS settings

I remember I change the BIOS settings for OS to Linux. Doing some research on Google and find this VMWare doc. After change the OS back to Windows 2000/2003, the error message disappears.

Details
HP ProLiant servers (and pre-merger Compaq ProLiant servers) require specific BIOS settings in order to be fully compatible with VMware ESX Server. Use of BIOS settings with values other than those identified below might result in operational instability.

If the MPS Table Mode setting within the ProLiant BIOS is not properly configured for ESX Server, the BIOS might not fully populate the PCI interrupt routing entries in the MP Configuration Table (part of the Intel MultiProcessor Specification). An incomplete MP Configuration Table might result in failure of an initial installation of ESX Server, failure to complete the boot process, or instability during normal operation.

Solution
ESX Server requires a BIOS MPS Table Mode setting of Full Table APIC. With the exception of the specific systems referenced below, the following BIOS settings must be applied in order if available:

System Options > OS Selection: Select Windows 2000.
Advanced Options > MPS Table Mode: Select Full Table APIC.

When presented with multiple Windows options (Windows 2000, Windows Server 2003, Windows .NET, and so on) select Windows 2000. If both BIOS settings are available and can be modified, both must be set correctly. You should confirm these settings after any BIOS upgrade operation.

For the HP ProLiant DL 740 and DL 760 servers specifically, set the OS Selection BIOS setting to Linux to allow for proper operation when hyperthreading is enabled. The remainder of the BIOS settings, and the order in which they are applied, are as specified above.

If any BIOS setting is incorrect, you might see one or more messages in /var/log/messages or /var/log/vmkwarning on the Service Console similar to the following:

May 11 22:36:30 esx101 vmkernel: 0:00:00:00.00 ALERT: Chipset: 303: no PCI entries - Check BIOS Settings
May 11 22:36:30 esx101 vmkernel: 0:00:00:00.000 cpu0) ALERT: Chipset: 433: no PCI entries in MPS table - check BIOS settings
May 11 22:36:30 esx101 vmkernel: 0:00:00:00.000 cpu0) WARNING: IOAPIC: 986: no PCI entry for busID 2 busIRQ 8 - trying ISA irq 15

You might also see unnecessary hardware interrupt sharing, which is explained in detail in knowledge base article 1290 at http://kb.vmware.com/kb/1290.

You might need to use the ROM-Based Setup Utility (RBSU), SmartStart CD, or other BIOS configuration tool to make these adjustments. Hewlett-Packard's Web site includes extensive documentation on how to adjust BIOS settings on HP and Compaq systems, including generalized RBSU guides which may be found at http://search.hp.com/query.html?qt=title%3A%22ROM-Based+Setup+Utility+User+Guide%22.

The "OS Selection" section of the HP RBSU User Guide notes that, in certain circumstances, adjusting the OS Selection BIOS setting might result in an automatic change of the MPS Table Mode BIOS setting (emphasis added):

The OS Selection option allows you to select the primary OS for the server. A list of supported OSs for the server displays on the menu, with the ROM defaulting to Microsoft Windows 2000 when the server supports it. Default server Multi Processor (MP) settings and PCI Hot Plug reservation settings are automatically set based on the OS selected.

Although an automatic change might occur, the individual BIOS settings should be confirmed explicitly. For additional assistance in making these changes, please contact HP Technical Support.

Windows 2000 in VMWare Server 1.06

I rebuilt my machine with Windows 2000 and installed VMWare Server 1.06. After that, I would like to create a VM with Windows 2000 for browsing and software testing. After completing installation and reboot, bluescreen comes up with inaccessbile boot device error. I figure out I have setup the HDD with LSI controller. To fix the issue, go to lsilogic website and download the Windows 2000 driver for LSI 20320-R driver. Press F6 when booting from bootable CD / Windows 2000 bootdisk 1. Insert the driver FDD when you are asked. The same applies to XP and Windows 2003.

Microsoft cluster with VMWare Server 1.06

Recently, I have changed job, and a lot of the works are related to Microsoft cluster. Setup is not difficult at all. The most important is to edit the vmx files (VM offline) for the two nodes. Add these lines

scsi1.sharedBus = virtual
disk.locking = "false"
diskLib.dataCacheMaxSize = “0”

On the second node, make sure you select using existing virtual disk and choose the right vmdk file (quorum disk created with node 1). Put the quorom disk on bus 1. If you put the quorum disk in other bus, change the scsi[x].sharedBus = virtual.

It makes our life a lot easy with testing and trying software in cluster environment.

How to test BES email

We have trouble running SP4 update for our BES server. It stuck at updating the mdss database which is useless to us. However, it requires you updating it. Eventually, I talk to Blackberry support. Uninstall and reinstall SP4 without mdss. Now, it works flawlessly. However, I am not a BES user. How can I test it? Send an email to a BES user with subject confirm and the < > around the subject . If it is sent to the user Blackberry, you will receive a reply with subject "Blackberry Delivery Confirmation". Then you can tell if the email is delivered to Blackberry.

MDSCS? MDSS?

Some of the users would like to access files from file server behind the firewall from Blackberry. We have BES running behind firewall. Of course, port 3101 is opened on the firewall.

We download the Mobile File Manager from www.rovemobile.com. As long as IPPP service book appears in your Blackberry, you should have access to your network. If MDSCS service installed correctly on your BES server, it should be pushed down to the blackberry.

MDSS is for software designed by your developers. So, don't get confused. Only MDS Connection service is required for this software, not MDSS - Apache Tomcat Service. You don't need MDSS unless you develope your own software for Blackberry running on BES.

Rovemobile had a few nice tools available. However, the transfer rate is still expensive.

Create VM in ESX 3.0

I have trouble deploying VM from from templates with customization option. I have talked to VM support, and they suggest to run all the updates first. However, it is not a consideration at this time because of the VLAN project. The workaround

1. First step is the same "Deploy VM from this template"

2. Complete all the necessary steps and choose "Do not Customize" for guest customization

3. Once VM created, start VM

4. Login to the Windows 2003 Server, download the sysprep tools "http://www.petri.co.il/download_windows_2003_sp1_sysprep.htm"

5. Unzip the sysprep tool. You will find several files including setupmgr (edit existing sysprep.inf such as computer name...to suit the settings of the VM). In case you have not created sysprep.inf, run setupmgr.

6. Copy the sysprep.inf and all the unzip files in the sysprep tool to the VM "C:\sysprep\"

7. Edit sysprep.inf. Make sure “ExtendOemPartition=1” (it will extend the partition to all the available space)

8. Then run sysprep.

9. Power down VM.

10. Now, modify the size of VM (sorry, you can only increase after version 3.0). Login to ESX server through console.

11. Run “vmkfstools –X 12G /vmfs/volumes/VMName/vmname.vmdk”
(In this case, the VM VMName will be extended to 12G. Now powerup the VM. You should see the size of C drive increasing to 12G. For existing VM, if you want to increase the size of C, do not use sysprep because this will re-generate the SIDs.)

Add domain user to local admin group through Group Policy

Using Security group
1. In Active Directory, created a Global Group with Security group type called LocalAdmin.
2. Included the users as members of the LocalAdmin.
3. Created a GPO.
4. Edited the GPO by clicking Computer Configuration -> Windows Settings -> Security Settings -> Restricted Groups. Add GroupName "Administrators".
5. Then Add domain admins and DOMAIN\LocalAdmin as member of this group
6. It will take some time for the group policy to take effect. One draw back of this method: If a Restricted Groups policy is defined and Group Policy is refreshed, any current member not on the Restricted Groups policy members list is removed. This can include default members, such as administrators. For example, if you had added DOMAIN\user1 to local admin in the past, DOMAIN\user1 will be removed from the local admin group of the machine.

If you do not prefer central management, then create a group policy. (from http://www.tutorials-win.com can't find the posting any more)
1. Create a startup script in the policy with this command. NET localgroup Administrators /add "Domain\Domain Group" (if you just want to add specific user, change "Domain\Domain Group" to Domain\user)
2. make sure the workstations placed in the correct OU
3. Domain Group has to be less than 30 chars. You cannot add more than one domain group to the same local group. This is a limitation.

HP C-Class Blades single point of failure

We powered down the SAN for maintenance about a week ago. Prior to that, we have to power down all the servers connected directly to the SAN first. We have a C-class blade enclosure with five servers currently. After the SAN upgrade was completed, we tried to power the enclosure back up. Guess what. The onboard administrator of enclosure failed to start up. We attempted to power up the server and it failed. We contacted HP support. The answer was a hardware failure. The part arrived at UPS store close to Pearson airport. We figured out we went to there to pick it up instead of waiting for them to deliver. Replace the onboard admin and all servers power up. Now, update with latest firmware and reload the config from backup. If you don't have config file backup, it won't take long to re-config it. So, make sure you save the config before power down the whole enclosure.

The HP support told me the servers would keep running if the onboard admin failed while the servers were powered on. However, if I powered down the servers, you could not power them back up until onboard admin was replaced. One thing forget to ask was "can I replace onboard admin while the servers are running".

Activesync Attempt with Exchange gives 0x80072EE2 error

This issue was resolved few months ago actually. What happens was one of the directors always had trouble sync with the Exchange Server. However, he didn't want to use Blackberry. He had 5 different WM phone at home. He tried all of them and still had the same issue. Everytime, he had to power cycle (turn off the phone and take the battery out) the phone, then it will sync again. After a while, it will not work again.

Eventually, we found out he plugged the phone to laptop to sync the notes through USB. Our internal Exchange Server ip is 10.x.x.x. The webmail address he used to sync wirelessly is a public address of course. Based on the research found in internet, it has to do with Windows Mobile DNS cache not being cleared properly when USB cable is unplugged. When plugged into the desktop to sync inside the office, it will use the internal address for Exchange server. After unplugged from USB, it will cause issue of course because the DNS will point to internal address of Exchange server instead of the webmail address. Workaround....powercycle after unplug from a machine in the internal network.

Persistent Binding with VMWare

Persistent Binding is highly recommended in SAN environment. However, this is not required for ESX Server 3. ESX Server 3 uses UUID's written to the LVM of the disk. For luns containing VMFS volumes and the ESX Server boot disks, persistent binding is no longer required.

You can check UUID in Linux with following command.
ls -l /dev/disk/by-uuid

Version of OS on Windows Mobile Phone

Have some time today to re-arrange all the notes and printout. One of them is about the version of Windows Mobile Phone on your pocket PC. I believe it is from Pocket PC Central.

Version Number and OS Version

3.0.9348 Pocket PC 2000
3.0.11171 Pocket PC 2002
4.20.xxxx Windows Mobile 2003
4.21.xxxx Windows Mobile 2003 SE
5.1.xxxx Windows Mobile 5
5.2.xxxx Windows Mobile 6