Mysterious problem - HP DL380 G7 Server - ESXi Installs but wont boot

[digglife]

Anyone ever see this? I have a new DL380 G7 Server, I have 8 disks and have it setup with 2 RAID10 arrays (4 disks each). ESXi 4 DVD is from an ISO I downloaded from HP (and have used in other servers like DL380 G6 without issue). The install goes through to the end, asks me to reboot and then I remove the DVD and reboot. The SmartArray P410 controller is set as the boot device (there is only one in the system) and I even try F11 during boot and tell it to boot from HD ... screen goes blank for a sec and the server then tries the other boot devices (network, etc).

I am really puzzled. During installation the installer sees both logical drives just fine. I picked LogicalDrive0 to install to.

Thoughts? I am stumped.

[DSTAVERT]

Since your servers are designed to do so I would install ESXi to the internal SD card (need to add the SD card).

[digglife]

I am not entirely convinced of the reliability of SD Cards or USB to boot from. I would prefer to risk myself with a P410 controller handling the booting and managing the drives that are used for boot.

Anyway, when it tries to boot I see a mesage about a non-system disk or disk error.

I have run the diagnostics 'Boot disk test', it shows:

Boot Disk Status: Pass - Disk Present

Boot Disk Ready: Pass - Disk Ready

Verify Boot Sector: Pass - Valid OS Boot Sector Present

[a_p_]

On different LSI based controllers I had issues in the past, where the order of the logical disks created on the RAID controller was not the same as presented in the ESX(i) setup and therefore even with the first logical disk selected to install the hypervisor, it was installed on the second one. I didn't have the time to search for the cause of this yet. I ended up to install ESXi again on the "second" disk then.

André

[DSTAVERT]

There is next to no writing to the SD card or USB disk. ESXi loads directly into RAM. Once an hour there is a configuration backup from one boot partition to the alternate boot partition.

If you absolutely positivly need to use hard disks I would create a virtual LUN using the smartstart CD Array Configuration Utility. Make the Vitual LUN 5GB. That will give you a 1GB partition for ESXi and 4GB scratch.

All firmware is up to date? Disk controller battery is installed and charged. Try booting / installing the other array.

[samansalehi]

Please download the iso from vmware.com, not embedded one, and see if it will install on the server or not?

or would u please try to update all firmwares to server, Firmware ISO file 9.20

use this link:

http://h18004.www1.hp.com/products/servers/management/core-management-100.html

on downloads tab it has the latest update for servers proliant.

if it is useful, would please mark it as "Helpful" or "Correct"?

Saman

[MHAV]

hey digglife,

you need to install the lastest firmware for the DL380G7. Use the ESXi4.1 ISO from VMware to install the Server.

BIOS 2010.12.01
Firmware 9.20 (B)

http://h20000.www2.hp.com/bizsupport/TechSupport/DriverDownload.jsp?lang=de&cc=de&prodNameId=4091432...

ESXi4.0 can cause a problem not knowing the vmnic0 this will be fixed with ESXi4.1

Regards

Mike

[digglife]

My worry is that hardware monitors, etc wo't work if I use the vanilla ISO from VMWare instead of using the HP one to install.

I will try downloading the update CD.

[digglife]

Hmm. vmnic issues would be bad, the server has 4 in it. Would other nics be affected as well (vmnic 1, 2, 3)? I dont need all 4 for now.

All of our servers here are ESXi 4.0u2, and vCenter server is 32 bit, so I dont think ESXi 4.1 could be added to vCenter. There is a project to update vCenter, but that is on hold for now due to costs.

So, I would like to stil with ESXi 4.0U2 if at all possible. If all else fails, I will have to try 4.1. Again my preference would be to use the HP ISO for 4.1.

[J1mbo]

As mentioned above, the ESXi installer is dangerous when used with machines with LSI controllers with more than one LUN.  It will pick the LUN it writes to at random regardless of which LUN is selected on screen, potentially over-writing your data in the process.  I have shown this repeatably on Dell PE2950 and PE710 servers.

I would suggest reconfiguring the machine with only one LUN for now, installing ESXi, and then creating the remainder once it's up-and-running.

HTH

[digglife]

Incidentally, should I get no luck on HD install, is the 4GB SD option the only one for this server from HP? I see part number: 580387-B21

I suppose ESXi 4/4.1 will install in 4GB?

[LarryBlanco2]

U can load the vanilla ESXi 4.0 from VMware and then load up the CIM bundle from HP and update the server using esxupdate command.  This way would will be able to still get the monitors.

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=us&prodTypeId=15...

Copy the zip bundle locally onto the esx server. 

Then SSH into the server or DCUI and tech support mode.

and issue.

cd sbin

esxupdate --bundle=hp-esxi4.0u2X-bundle-1.4.zip --install

Larry B

[a_p_ · Accepted solution]

I agree with J1mbo's suggestion. This is a good way to make sure you install the OS/ESXi on the intended logical volume.

Another possible way to make sure you install to the correct logical volume is to created the volumes with slightly different sizes. (I usually create the first one with only 10 GB for ESXi.) This way you will be able to pick the LUN in the installer based on it's size.

André

[digglife]

Thanks for all of the input so far. I am downloading the latest firmware update DVD (its 955MB). I will delete the logical drives and create a single one of 5-8 GB and try to install on that once the firmware is updated.

Incidentally, is the default cache ratio of 25% Read / 75% Write good enough for ESXi servers. This server will be hosting webservers (no database or heavily IO dependant servers).

[a_p_]

... create a single one of 5-8 GB ...

LUN sizes are still limited to 2 TB minus 512 Bytes with ESX(i) 4.

I think you can start with that cache ratio. Once you see that you may need more read cache you can change that in the future.

André

[samansalehi]

Dear Friend, did u try Live chat with HP Specilist?

http://h50203.www5.hp.com/HPISWeb/customer/product.aspx?option=activechat&origin=acc&region=&pnid=&p...

If it is useful, Please mark it as "Helpful".

[digglife]

Thanks for all the help.

The issue as identified by a few of you was that ESXi Installer was confused about which LUN to write to. I removed them all and created a single small one for ESXi to install to. That worked, and it booted. I then created LUNs for each datastore I required. Working fine now. I did ensure all firmware was up to date too.

Many Thanks to all!!!

[lostjames]

This is really useful to know.

I have installed ESXi since 3.5,4.0 and 4.1 on several HP DL360's G5 and G6.

I have always just created one RAID 10 partition using the existing disks (typically 6x145GB) which gives a Local DataStore of approx 410GB.

Never had a problem, though my process has always been

Maintenance CD to patch (now seems defunct from July 2009)

Create Disk Volume

Install ESXi from HP Website.

I have never had a problem so this was interesting to read, thanks for posting.

Glad you got this fixed.  

[digglife]

Issue was resolved after firmware update for P400, using a single LUN for installation, and iLO disabling.

This issue did not manifest itself in G6 servers with multiple datastores of same size, etc. Although those were using different controllers.

[dkleeman2]

I have also had this problem but it was not solved by only using one LUN/logical disk. I found that the server would not boot to the hard disk while I had a blank SD card mounted in the SD card reader on the motherboard. Manually changing the boot order so that the hard disk was listed above the USB drive allowed the server to boot.

I have not tried physically removing the SD card as this is not practical in this instance.