How to migrate VMWare ESXi boot LUN and RDM to a new SAN system using free tools

Disclaimer: Use this at your own risk. I provide no guarantees that this will work for you or set your SAN/VMWare ablaze. I suggest to test it with less risky resources.


To copy ESXi boot lun you'll need two ESXi hosts. Host #1 Source. Host #2 - assist host.

• Put one in maintenance mode and ensure that this host has one throw away powered off the virtual machine.
• Mount the boot from the old system you wish to clone/copy with a known Lun ID (let's say 7)
• Mount the blank LUN (same sized) from a new SAN also with known Lun ID (let's say 8)
• Re-scan storage adapter (skip VMFS discovery for the quicker process)
• SSH to the ESXi host in maintenance mode 
• I generally like to browse to the folder of temp VM.
• Issue two commands: 
  esxcfg-mpath -L |grep L7 
  esxcfg-mpath -L |grep L8
• Make a note of "naa." device IDs for both LUNs - you will need them in a second

• vmkfstools -z /vmfs/devices/disks/naa.of_source_lun   /vmfs/volumes/location _of_temp_ vm/name_of_lun_to_clone.vmdk

• vmkfstools -i /vmfs/volumes/location _of_temp_ vm/name_of_lun_to_clone.vmdk  -d rdm:/vmfs/devices/disks/naa.of_destination_lun   name_of_lun_to_clone2.vmdk

DO NOT RESCAN storage yet or you will cause a purple screen!

Disconnect both luns from 2nd host, connect the new Lun to the source host as LUN ID 1

change boot policy on the host to boot from new LUN
reboot it validate boot

you're done.

Intruducing OpenFiler

Hey Guys,
Wow, I didn’t post anything new since August – Damn - I’m lazy :-)
I already mentioned FreeNas/OpenFiler software in some older my posts about low-cost DIY NAS boxes, but I guess its about time I try it myself in real environment.
Long story short, our IT director has decided to migrate all servers to the blade system, so this left us with spare regular “old-school” servers, most of them ProLiant DL360 and DL380.
One particular one –DL380 G3 which used to be company’s central SQL, now was standing idle for a few months or so, but now the time has come for the big – REPURPOSE !!
The original plan was to install Windows 2008 server, but after half-day spent I gave up on this idea. Lack: DVD drive, 64bit support, and no official drivers support was the final nail in the coffin (I did manage to initiate install from pre-existing 2003, but past-install OS never could complete the boot)
Going back to the 2003 server just seemed way to plain vanilla boring.
I said to my boss – Let us try OpenFiler – and he said – fire away !!
Today and later on I’d like to share some of my experiences and discoveries thru this mini-project from standpoint of mostly windows systems admin. Yea – I support ESX and Netapp – but most of command line changes are done by vendor’s support stuff :)
I promise to post soon about the install process.

How to find out the Brocade Switch model Number remotely

Quite simple actually, but not as straightforward as one would expect… I guess it’s part of McData heritage.
Open web browser at http://hostname/SwitchInfo.html
Scroll down till you see in List of Ports: switchType:    xx.x
Use table below to “decipher” the Model series
1.x = Brocade SilkWorm 1000 series
2.x = DS-16B / Brocade SilkWorm 2800 series
3.x = DS-8B / Brocade SilkWorm 2400 series
4.x = Brocade SilkWorm 20x0;
5.x = Brocade SilkWorm 22x0;
6.x = DS-16B / Brocade SilkWorm 3800 Series (Includes parity checking on all paths)
7.x = SCB
9.x = DS-16B2 / Brocade SilkWorm 3800 series
10.x = ED-12000B / Brocade SilkWorm 12000 switch blade
12.x = DS-32B2 / Brocade SilkWorm 3900 series
16.x = DS-8B2 / Brocade SilkWorm 3200 series
17.x =  Brocade SilkWorm 3800VL series
18.x =  Brocade SilkWorm 3000 series
21.x = ED-24000B / Brocade 24000 switch blade
22.x=  Brocade SilkWorm 3016 2GB Embedded Switch for IBM and Intel Blade Servers
26.x = DS-16B3 / Brocade 3850 series
27.x = DS-8B3 / Brocade 3250 series
29.x=  Brocade SilkWorm 4012 4GB Embedded Switch for HP Blade Servers
32.x = DS-4100B / Brocade Silkworm 4100 series
33.x =  Brocade SilkWorm 3014 2GB Embedded Switch for Dell, Fujitsu-Siemens and Hitachi Blade Servers
34.x = DS-220B /  Brocade 200E
36.x =  PB-48000-18i / Brocade SilkWorm FR4-18i 48000 Director Blade
37.x =  Brocade SilkWorm 4020 4GB Embedded Switch for IBM and Intel Blade Servers
38.0 = AP-7420
40.x= Fibre Channel Routing (FCR) Front Domain
41.x= Fibre Channel Routing (FCR) Xlate Domain
42.2 = ED-48000B / Brocade Silkworm 48000 series
43.x =  Brocade SilkWorm 4024 4GB Embedded Switch for NEC Blade Servers
44.x =  DS-4900 / Brocade SilkWorm 4900 series
45.x =  Brocade SilkWorm 4016 4GB Embedded Switch for Hitachi Blade Servers
46.x =  MP-7500B / Brocade SilkWorm 7500 series
47.x =  PB-48K-16IP / Brocade SilkWorm FC4-16IP 48000 Director Blade
51.x =  Brocade SilkWorm 4018 4GB Embedded Switch for Blade Servers
55.x =  Brocade SilkWorm FA4-18 48000 Director Blade
55.2 =  AP7600B / Brocade SilkWorm 7600 series
58.x =  DS-5000B / Brocade SilkWorm 5000 series

Fix: My annoying Boot from San Issue with BL45p Blade (Update)

My company recently bought 3 "new" HP blades - a nice addition for existing HP p-Class chassis. Why new in quotes - since technically the models are no longer current in HP model line, and they had to be bought from external vendor we don't usually use - but all purposes they are new for us.

BL20p G3 were pretty easy to setup as boot from out NetApp 3020 SAN, but the bl45p G1 was stubborn as an donkey.

There problem was during POST the Qlogic HBA didn't recognized the LUN and therefore , skipped on loading the HBA's bios = No boot.
However in Qlogic config utility (the CTRL-Q) I had all settings done correctly and the fiber scan shows the LUN correctly.

It took me 5 days of troubleshooting, trying different configs and getting ideas and opinions from colleagues and experts friends.

Nothing Helped! ;-(

Until a colleague given me an idea - to connect the same fiber/lun to different blade and only one as free was bl25 - despite totally different hardware (AMD vs Intel - BSOD pretty much guaranteed) it was still important test to see if its the fabric/lun/config issue. BL25 booted from lun perfectly every time (as well as BSOD every time as well)

A total surprise to me was a solution which I got from 2nd level supporter at HP after convincing them I need to escalate the issue and It's not 3rd party problem -

CLEAR BIOS TO DEFAULT SETTINGS !

It looks like part of Cmos settings got corrupted, but there was no indication in bios for that....

Update: Looks like after all it was problem with bad fiber…. :(

 

Technorati Tags: HP,Qlogic,Fiber,Blade,Netapp,P-Class,SAN

Microsoft Finally Gets Clustering Right........

Microsoft's recent release of Windows Server 2008 and its highly anticipated improved clustering and network load balancing seems to have been worth the long wait. It appears that after years of research and dedication to the problem, building a cluster that is supposed to work as it should, has finally paid off.

In testing the clustering failover in our own dev network, we discovered a rich feature set to provide for more stability and high availability in clustered environments. The most important to my company will be the multi-site cluster technology.

We have an HA site about 15 miles away with 10Gbps fiber (*1) between them to replicate. With this new technology Microsoft has developed, we can now easily put our HA boxes in another data center. This is a feature we have never been able to have in production for lack of Microsoft's tools to cluster the servers in another site properly.

Now when IBM allows snapshot copy on a block by block level through their 2145 San Volume Controller, we will be in business!! (*2)

Related Links:
http://www.microsoft.com/windowsserver2008/en/us/high-availability.aspx

(*) Editor Comments:
1 - DARN!!
2 - Netapp SnapMirror already does that :-)

Technorati Tags: Microsoft,Server,Fiber,Clustering,Network,failover