LVM2 Error when creating snapshot

I get this error whenever I create a bunch of snapshots. I have 36 filesystems in my vg. If I walk through them creating a snapshot on each one, the last 4 fail. They fail with the below error. I did the order backwards and got the same results so its not dependant on the lv I'm snapshotting, it appears to be dependant upon the number of snapshots I'm creating.

/usr/sbin/lvcreate -L 1G -p r -s -n save_perpet2-20060619161740 /dev/svg/save_perpet2lv
device-mapper: reload ioctl failed: Cannot allocate memory Failed to suspend origin save_perpet2lv

It creates the snapshot devices in /dev/mapper.

ls -l /dev/mapper/svg-save_perpet2--20060619161740* brw-rw---- 1 root disk 253, 159 Jun 20 15:19 /dev/mapper/svg-save_perpet2--20060619161740 brw-rw---- 1 root disk 253, 158 Jun 20 15:19 /dev/mapper/svg-save_perpet2--20060619161740-cow

but it doesn't create any devices in /dev/svg.

It also is available to query with lvdisplay.

lvdisplay /dev/svg/save_perpet2-20060619161740
--- Logical volume ---

  LV Name                /dev/svg/save_perpet2-20060619161740
  VG Name                svg
  LV UUID                oP2nDj-aa7p-dS3Q-4VdH-gk8r-x5C8-Akg78K
  LV Write Access        read/write
  LV Status              available
  # open                 0
  LV Size                1.00 GB
  Current LE             256
  Segments               1
  Allocation             inherit
  Read ahead sectors     0
  Block device           253:159

I can't do anything with that volume other than remove it though.

I'm running RHEL 4 U 3 with the beta hugemem kernel installed. I installed the beta kernel as the request of redhat support.

device-mapper-1.02.02-3.0.RHEL4
udev-039-10.12.EL4
kernel-hugemem-2.6.9-39.EL

Any insight would be appreciated.

Sadly, you are getting about as much response as I got to both a posting here and a bug filed with redhat about this issue.

This appears to be a "mis-use" of a particular kernel memory allocation function that is used to reallocate a pool until it reaches the maximum size that that particular function is capable of allocating. I don't recall the details at the moment however. The bug in redhat's bugzilla has a pretty good accounting IIRC.

b.

My other computer is your Microsoft Windows server.

Brian J. Murrell