Proliant DL380R4/SmartArray 6i hangs on boot

[toomas_aas]

I seem to have rendered my Proliant DL380R4 unbootable. Here's what I did to achieve this:

1. Upgraded the system ROM to latest version (P51, 12/02/2004).

2. Rebooted the system. The operating system booted up fine.

3. Rebooted once more and changed two settings in BIOS:

3a Under Advanced Options, set 'MP table mode' from 'Full APIC' to 'Disabled'
3b Changed the PXE boot for NIC1 from 'enabled' to 'disabled'.

Having done that and rebooted, the system comes up to a point where it starts to initialize the hard disks, and remains there.

This system has integrated Smart Array 6i controller and 6 drives in duplex configuration: 2 36 GB drives (RAID1) on channel 1 and 4 146 GB drives (RAID0+1) on channel 2. When the system powers up, the hard drives settle down to a state where the green hard disk icon is lit on all drives, and the green arrow 'disk access' icon is lit on the 36 GB drives. After waiting ca 20 minutes, the disk access icon on 36 GB drives goes off and disk access icon on 146 GB drives turns on. It's been sitting there for more than a hour now. I'm going to let it sit until morning (it's 9pm now), but I'm not sure what is happening or what should I do if it remains in this state infinitely. As far as I can see I can't enter the BIOS setup or boot from SmartStart CD, because these happen only *after* the SmartArray controller has been initialized...

[Doug de Werd · Accepted solution]

You could totally reset CMOS/NVRAM. It is switch 6 on the system board (check out the label on the under side of the top after you remove it).

- power off the server. take the cover off
- set switch 6 to ON. power on server
- let it run through POST. power off server
- change switch 6 back to OFF. power on server
- Press F10 to go into ROM Based Setup Utility

My guess is that your problem is with the APIC Mode. If it was working OK before, and you didn't have a good reason to change it, I would leave it alone.

Thanks,
Doug

[toomas_aas]

Yes, clearing NVRAM helped the system back up. I'm sure you are right that changing the APIC mode was what caused my problem. All the disk activity was red herring - I guess the controller was just doing it's regular data scrubbing since nobody was there to tell it to do something else.

As to why I changed this setting, that's a long story but since you asked ;-)

I am running SuSE Linux Standard Server 8 on this machine and top constantly showed extremely high CPU load on this server (85-90% round the clock). Most of this load was classified by top as 'System'. And vmstat showed an extremely high number of interrupts (90000 interrupts/sec). Looking at 'cat /proc/interrupts' I noticed that vast majority of them are ACPI interrupts. So I went looking for a way to make ACPI behave or turn it off if it refuses to.

First, as mentioned, I updated the system ROM to latest version. That brought no change.

Then I looked through BIOS settings and this setting (actually it's called MPS Table Mode) was the only one that looked even remotely related. However, as we found out, that only made things worse.

Finally I just added 'acpi=off' to kernel load command in GRUB configuration and the interrupts are gone now.

Next week I'll update the Smart Array ROM to latest, but for tonight I've had enough excitement.