Proliant DL740 and DL760G2 hangs sporadically running SUSE Linux ES 8.0 SP3

[Dirk Quacken]

I just want to share this information in case someone has the same problem.

A Proliant DL740 or DL760G2 may hang completly in certain conditions:

- The server is running United Linux 1.0 SP3 / SUSE Linux ES 8.0 SP3
- the VGA Kernel Parameter is changed from vga=7xx to vga=normal
- massive text based output is listed on the text console (e.g. cat /var/log/messages)

Workaround:

Verify that your server shows this failure. To reproduce the failure just type in the following command on a text based console: "cat /var/log/message". Depending on the size of the message log you have to repeat the command serveral times within a few second. If you have a DL760G2 or DL740 running SLES 8.0 SP3 and vga=normal activated, your server will crash immediatly. I was able to reproduce the failure with the following kernel: 2.4.21-138-smp, 2.4.21-168-smp and 2.4.21-261-smp. Other kernels were not tested but might show this issue. Different Bios versions, hyperthreading disabled and acpi=off also shows the failure . So far the only workaround I know is: Change the vga parameter to "vga=788" or "vga=791" or start Linux with failsafe settings (but this is not a suitable workaround).

To change the vga parameter, simply edit the file menu.lst from Grub or the lilo.conf (depending on what you choose as boot loader) and change the vga parameter from "vga=normal" to "vga=788" or "vga=791". Reboot the server to activate the new parameter.

HP is currently investigating this problem.