HPVM Unable to create more guests due to lack of RAM

Hi All,
There are few threads regarding this subject of being unable to create more guests due to lack of RAM. So I am aware how the sum works.. add 8.5% to whatever is allocated, be that the host or guest. But I'm not sure if I have a hardware issue with memory or what I see is correct as I am new to HPVM. I have 98GB of memory on a quad CPU system. I have started up 33 guests all configured with 1024M RAM, now when I try to start another it comes back with an error..

HPVM guest sfd-ec3 configuration problems:
Warning 1: There is insufficient free host memory for this guest.
These problems may prevent HPVM guest sfd-ec3 from starting.
hpvmstart: Unable to continue.

I'll list a couple of other things which will hopefully help you to help me.. please?
# hpvmstatus -s
[HPVM Server System Resources]

    Processor speed = 1667 Mhz
    Total physical memory = 98273 Mbytes
    Total number of processors = 4
    Available memory = 732 Mbytes
    Available swap space = 47415 Mbytes
    Maximum vcpus for an HP-UX virtual machine = 4
    Maximum vcpus for a Windows virtual machine = 4
    Maximum vcpus for a Linux virtual machine = 4
    Available entitlement for a 1 way virtual machine = 1667 Mhz
    Available entitlement for a 2 way virtual machine = 1667 Mhz
    Available entitlement for a 3 way virtual machine = 1250 Mhz
    Available entitlement for a 4 way virtual machine = 916 Mhz

# swapinfo -t
Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 8388608 0 8384740 0% 0 - 1 /dev/vg00/lvol2
dev 76808192 0 76775384 0% 0 - 1 /dev/vg00/swap2
reserve - 36601000 -36601000
memory 100632240 51258792 49373448 51%
total 185829040 87859792 97932572 47% - 0 -

For me it appears that I have lots of RAM available, well at least 49GB.

Any ideas?

Thanks very much, Eric

I have never had the chance to work on anything above 11.11 (...) that said,you have not mentionned which OS version nor Architecture, but it seems you are on itanium platform...
Have you considered allocating more swap space? If you believe you have enough RAM, that may solve your issue...

---------- Post updated at 13:14 ---------- Previous update was at 12:59 ----------

looking at your configuration, it seems you are at the limits where the system might start swapping and if it occurs, the way you are currently configured will deeply impact your performances, so try to allocate if you can your swap on different LUNs/VG and make them same size so if swapping occurs you will have all swap spaces being used at the same time (over more controllers...).

Hi there.. it's actually 11.23, which is new to me to be honest I tried adding a bit more swap on disk to see if the o/s would release some RAM, but it didn't, not a bit. I also found that the SFM processes (EMS) was holding over 25GB RAM, so I stopped that, but that made no difference.. grabbing at straws really. I've now built another system, exactly the same spec and run up the same amount of VM's and it still has 54GB of free RAM, so it looks like a hardware fault with RAM.

Thanks for replying to my thread

Have you tried to see what gpm/glance shows?
I also used to use ipcs to see who were being hogs...
The idea was lets say you found oracle or a dba holding suspicious quantity of memory, you would ask that folk to leave and check if the resource was given back or not... if not you use ipcrm...