Boot from san with open-e DSS

[kbs]

What is the blocksize of the LUN? If it is 4096 (default) then Linux will have problems to boot from this LUN. Try to create a LUN with a blocksize of 512, it worked for my servers. I Have three VMWare servers booting from DSS SAN.

Hello,
I am trying to implement boot from san with open-e DSS v1.32. I added a qlogic qla2342 with latest bios in the open-e box and set it to target mode in de open-e gui. I created a Lun and added the wwn's of one port of the 2342 in the open-e box and of the qlogic qla2340 which is in the server I want to boot from san. On my san switch I create a zone for the 2 ports.

When I boot the machine the bios on the qlogic gla2340 sees the lun on the open-e box. When I want to install ESX the installer exits with error "the installer exited unexpectately). I tried to install in text mode but same thing.

I also tried to install windows server 2003 on that lun, but after the lun is formated and the server reboots to grafical mode I only see a blinking cursor.

Does anyone can help me?
thx

[paul2kl]

Question for Kbs

You mention that you have and FC san working with ESX server, there are a few of us that are having issues getting this to work. Could you let us know what FC cards you are using, and what switch if you are using one.

We are seeing issues with the 2340, 2310F and 2342's. We are not trying to boot from San, just want to store VM's on it. I can create a LUN that is seen by ESX, but once it is formated with VMFS, the troubles start. Unable to create a new VM, We see lots of Timeout retries on both ESX and DSS. Then the DSS server has to be powered off and on again.
So if you could let me know what cards you have working would be good, also what bios you are running on them.

[kbs]

Sorry, we use Vmware GSX 3 Server on a Redhat Enterprise Linux (Version 3). But I think the ESX Server is based on a hardened RHEL too. Whe I tried to install REHL, I had the same proplem like iddeabvba with the installer terminating with 'the installer exited unexpectately'. Then I tried to boot the server with REHL Version 5 and the installer came up with the error that the blocksize of 4096byte blocks is not supported and the he wants to reinitialize the LUNs. After setting back the LUNs to 512byte blocks it was no problem to install the system and everything is working fine till now. But I think it is a different problem to yours.
I use is a QLA 2340 in the DSS. All other servers use Emulex LP 8000/9000 (1 GBit) controllers. They are connected with a Brocade Silkworm Switch. Two of the servers are for development and one for production use. If I have time next week I will try to install the ESX Server on a forth machine, maybe I can tell you more after that.

[kbs]

Hi,

I finally managed to install an ESX Server 3.0.2. I had no Problems installing an Windows 2003 R2 Server on a DSS Lun. I use this Configuration:

DSS: QLA2340, 20GB FC LUN blocksize 512, HBA Alias
ESX: Emulex LP9002, Boot from SCSI, DSS LUN for VM

If you need more information, you are free to contact me.

[paul2kl]

Hi,

I finally managed to install an ESX Server 3.0.2. I had no Problems installing an Windows 2003 R2 Server on a DSS Lun. I use this Configuration:

DSS: QLA2340, 20GB FC LUN blocksize 512, HBA Alias
ESX: Emulex LP9002, Boot from SCSI, DSS LUN for VM

If you need more information, you are free to contact me.

Hi KBS,

I have sent you an email with a few questions to see if we can get this working properly, as there are a few of us with the same issue as myself. Would be great if we can compare notes to see if we can locate the real issue.

Paul

[kbs]

Hi,

when I installed the VM (Win2003 Server) yesterday everthing was fine. I tried to power on, power off, suspend and resume the VM and it was OK. Today I rebooted the ESX server and after powering on the VM again I had the same problems like you. This is a very strange behavior. A lot of kernel exceptions (aacraid, scsi_tgt, scst_fileio, qla2x00tgt) on the DSS and all other servers lost their fc luns. So I had to restart the dss and all other servers. I have a similar system with a Redhat Enterprise Linux and VMWare GSX Server 3. These servers even boot from the DSS SAN and have no problems. I think there are some advanced configuration parameters on the ESX. If I find the time I will try to change them maybe it helps. But actually it is problematic because DSS serves for productive systems.

[paul2kl]

Hi,

KBS yes you are seeing the same issues as me. I was able to sometimes leave it over night and it would be ok... But if i restarted ESX the same issues as you would occur. Or if i just shut down the VM's running from the LUN, and then tried to boot them 30 mins later, DSS had errors, and loads of busy retry errors on ESX. I would have to power off the DSS server and ESX server to get things back again. I am currently trying a beta of 1.37, but my inital test shows the same issues. I am going to rebuild ESX and DSS servers from scratch and then test it all again. Do you have any sugestions on which ESX settings you may wish to change, as i will try them out to see how it goes.

Thanks,

Paul