ProLiant Servers - Netservers

Have you seen this before on a DL365

 
Yvo
Occasional Contributor

Have you seen this before on a DL365

We recently received an order of approximately 50 DL365 G1s with 6 146GB 10K SAS drives using a Smart Array P400i. Three of the machines would experience a 'Red Screen of Death' as an Illegal OpCode when it completed the BIOS configuration (after the NIC).

If you do, I believe I found a way around the issue and I hope to see HP implement a fix in a future sytem rom update for the DL365 or Smart Array P400i.

Below is what I sent HP as an update to my three cases.

Test specs:
3x HP DL365 with Smart Array P400i with 6x 146GB 10K SAS drive configuration. 4 on the primary cable and 2 on the secondary cable.
All machines have the latest Smart Array P400i ROM (version 4.06)
All machines have the latest SystemRom (A10 from May)

Issue:
Machine would "Red Screen" right after the NIC's BIOS had loaded and it needed to POST. It would Red Screen with any option given (Boot off of CD, Boot off of HD, Boot off of NIC) with the exception of entering the NIC's configuration screen. NIC is a Broadcom Netextreme II.

Cause:
The cause of the red screen was to be determined to be caused by adding or changing the physical drive configuration of the RAID controller. For example adding a drive (we only added in pairs, so 2 or 4) would cause a red screen. Unplugging the secondary cable and removing the drives would cause the machine to boot properly. It would ONLY do this if an operating system had not yet been installed AND the previous reboot a RAID array was created using the Smart Array's System Rom configuration. Similar issues have been resolved in the past on other HP products (see at end of report).

Workaround:
The following work around was used on three machines that experienced a Red Screen of Death (RSOD).
1. Power down machine
2. Unplug SECONDARY SAS serial cable (if you look at the machine from the front, its the connection closest to you, not the one in the back).
3. Power up machine
4. Go into RAID bios (F8) and delete all arrays. It will complain on bootup that its missing some drives.
5. Create a new RAID array with just the first two drives (4 in total will show).
6. Let machine finish booting and verify non existance of RSOD
7. Power down
8. Plug SAS serial cable back in
9. Go into RAID bios and create a new 4 drive RAID 1+0 array.
10. Let machine finish booting and verify non existance of RSOD
11. Install Operating System (in our case Red Hat Enterprise Linux 4 update 5)

Summary:
Similiar problems have cropped up in previous HP machines (http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&objectID=c00624605 and http://h18023.www1.hp.com/support/files/server/us/download/25263.html). Very often it was related to a "corrupt" MBR or boot sector on the RAID array. I can only hypothesize that this is a similar problem in a newer machine. I believe that deleting the corrupt RAID array while the drives were out of commission solved the issue. I believe the issue is related to the secondary SAS port being used, however further testing by HP needs to be done to confirm this. We plan on ordering approximately 100 DL365 G1s in the near future and we'd hope to receive either an official HP patch or approved patch for use on the DL365s in order to avoid this problem again.

3 REPLIES 3
James ~ Happy Dude
Honored Contributor

Re: Have you seen this before on a DL365

Hello Yvo,

Strange indeed; You mean to say these RSOD are MUCH before the actual OS installed ?? WOW !

I thought BSOD/PSOD/RSOD are only possible after an OS installation; the later ones with Vista. Your information is much appreciated.

Regards,
James.
Yvo
Occasional Contributor

Re: Have you seen this before on a DL365

Yes the RSOD was on a box that didn't even have an OS installed on the hard drive. However the HP tech is under the impression that either the RAID controller or System Rom is using the Linux kernel (if so, they are violating GPL as it isn't mentioned anywhere).

I have indeed seen a RSOD on a Linux installation before (different machine where the RSOD was caused by a dying Pentium II proc) and on a Vista installation due to a bad driver.
James ~ Happy Dude
Honored Contributor

Re: Have you seen this before on a DL365

Yea... Even I think, Firmware are written on Linux platform;(I think). Same reasons Y SmartStart or Firmware Maintenance CD are coded in Linux.

Interesting !!!