Virtual Machine Manager won't allow memory allocation over 1024

[PCK01]

Hi all,

I'm running Fedora 11 (2.6.30.10-105.2.4.fc11.i686.PAE) 8 GB memory on a AMD Phenom(tm) II X4 965 Processor, 4 core server I just built.

The problem I have run into is any virtual machine that has >1024 MB memory assigned to it will not boot. As along as it has 1024 MB or less it boots fine.

The symptom you see is a very quick message "Booting from hard disk" the screen then goes blank, you can't login through SSH, and can't ping the network address so I know it's not some crazy video problem hiding the boot process.

The attachment are messages from a failed boot attempt.



Thanks
Pete

[jpollard]

Hmmm..

One thing I notice is:

Jan 26 20:25:29 mail2 kernel: WARNING: This combination of AMDprocessors is not suitable for SMP.

and yet you have 2 CPUs started....

You may need to check the BIOS on your server system and make sure it accepts
both CPUs for SMP. Something may be disabled here.

As far as errors, only the two WARNING message indicate a problem. The system
actually came up, with a functioning user mode - mail got started, and it looks like
pcscd got started (smart card service daemon).

Now the system does look like it may have hung... but that could be due to the
warning about the two processors.

Have you tried defining the VM with only one CPU?

I admit to not having any other ideas... It should just work.

[PCK01]

Interesting huh......

I have tried defining the VM with only 1 processor and this always happens whenever the memory assigned is >1024 MB, regardless of the # of processors assigned. I have 4 VM running and if I assign >1024 MB of memory to any of them, that particular VM won't boot successfully.

The mobo in question is a Gigabyte 790XTA-UD4 with the AMD Phenom(tm) II X4 965 Processor, 4 core prosessor. I am running the latest BIOS on the mobo.

The other interesting thing here is when I tested assigning 1024MB compared to 1025MB, the time stamp jumped 8 hrs on the logs as seen here:
#########################

Jan 26 14:11:13 mail2 auditd[1132]: Error sending signal_info request (Operation not supported)
Jan 26 14:11:13 mail2 auditd[1132]: The audit daemon is exiting.
Jan 26 14:11:13 mail2 kernel: audit(1264543873.848:1731): audit_pid=0 old=1132 auid=4294967295 ses=4294967295 res=1
Jan 26 22:15:39 mail2 kernel: imklog 3.22.1, log source = /proc/kmsg started.
Jan 26 22:15:39 mail2 kernel: Initializing cgroup subsys cpuset
Jan 26 22:15:39 mail2 kernel: Initializing cgroup subsys cpu

############################

I'm at a loss, but it has to be code related.........

---------- Post added at 04:18 PM CST ---------- Previous post was at 04:14 PM CST ----------

I need to clairify something....

When I assigned 1025MB of memory, the timestamp in the logs jumped 8 hrs.
When I reverted back to 1024MB memory, the timestamp in the logs returned to correct time......

[PCK01]

Anyone have any ideas on this? I'm totally stumped on this one........

[PCK01]

Problem resolved.....

The problem turned out to be either the AMD Phenom(tm) II X4 965 Processor, and or the Gigabyte XTA-UD4 mobo combination. Nothing I tried made any difference until I replaced both the processor and mobo.

I went with an Intel Core I7-860 processor with 8GB Corsair Memory on a ASUS P7P55D DELUXE mobo. I've got 5 VM's running now, this thing screams........