Fedora Core1 - SMP kernel booting problems

[Geoff Collis]

I have (re)booting problems with 6 brand new Dell 2650 systems, so they all
have the latest firmware/bios. I have omitted all the details for the sake
of brevity.

 

Each system has:

- two Intel(R) Xeon(TM) CPU 2.80GHz

- 6GB Memory

- 4 internel drives (2 x 36GB root h/w mirror, 2 x 72GB drives unused h/w
mirror)

 

They are installed from a using a minimum configuration plus a few extra
RPMs, and have an LVM Root configuration.

 

  [root at broken root]# mount

  /dev/sda1 on / type ext3 (rw)

  none on /proc type proc (rw)

  none on /dev/pts type devpts (rw,gid=5,mode=620)

  usbdevfs on /proc/bus/usb type usbdevfs (rw)

  /dev/Volume00/home on /home type ext3 (rw)

  none on /dev/shm type tmpfs (rw)

  /dev/Volume00/tmp on /tmp type ext3 (rw)

  /dev/Volume00/usr on /usr type ext3 (rw)

  /dev/Volume00/var on /var type ext3 (rw)

  /dev/Volume00/public1 on /opt/public1 type ext3 (rw)

  /dev/Volume00/public2 on /opt/public2 type ext3 (rw)

  /dev/Volume00/log on /log type ext3 (rw)

 

The systems boot/reboot without issue on a non-SMP kernel, 20+ reboots
without any problems.

 

With a SMP kernel it hangs on reboot after I do a "shutdown -r now", but not
every time and not predictably. For example, a system may reboot
successfully 5 times in a row, but then refuse to reboot on the sixth
attempt. Booting off a non-SMP kernel (which always works) and then
rebooting with a SMP kernel usually gets the system up and running but not
always.

 

The reboot seems to hang at one of the following points in the boot
sequence:

- enabling swap

- configuring the loopback interface

- configuring IPTABLES.

  

This affects all the Kernel releases I have tried so far: 

 - 2.4.22-1.2199.nptlsmp, 2.4.22-1.2197.nptlsmp, 2.4.22-1.2115.nptlsmp

 

I have also tried the "noapic" boot option; this seems to help some, but not
all the time.

 

I have seen various *issues* with respect to the Xenon hyper-threading, is
this the cause? Is it LVM with SMP, or the fact that swap is an LVM volume?

 

This affects all of these systems in the same way, so I doubt it is h/w
related, but this does not eliminate it being a BIOS/firmware issue. What
can I try next? Am I going to better off if I buy a supported RHEA release
or am I going to have the same problems?

 

--------

Geoff Collis, Kelkoo

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/fedora-list/attachments/20040817/81b6497b/attachment-0001.htm>