I'm running glusterfs compiled from source on a two Centos 5.5 64bit hosts with samba cluster. Kernel oplocks disabled in samba ("kernel oplocks = no" in smb.conf)
With glusterfs-3.1.2, when more than one applications try set oplock on same  file (opened with ReadOnly) in cluster filesystem, it report the bug "general file error". 
Excerpt from the smbstatus output:

Pid          Uid        DenyMode   Access      R/W        Oplock           SharePath   Name   Time
--------------------------------------------------------------------------------------------------
0:16622      500        DENY_WRITE 0x20089     RDONLY     EXCLUSIVE+BATCH  /smb/1c   db1c/kt/1Cv7.MD   Fri Jan 28 10:11:05 2011
1:31122      652        DENY_WRITE 0x20089     RDONLY     EXCLUSIVE+BATCH         /smb/1c   db1c/kt/1Cv7.MD   Fri Jan 28 08:54:29 2011

With glusterfs-3.1.1 oplock is working fine. It changes state from "EXCLUSIVE+BATCH" to "LEVEL_II" and the error does not occur:

Pid          Uid        DenyMode   Access      R/W        Oplock           SharePath   Name   Time
--------------------------------------------------------------------------------------------------
1:2140       657        DENY_WRITE 0x20089     RDONLY     LEVEL_II         /smb/1c   db1c/kt/1Cv7.MD   Fri Jan 28 09:03:19 2011
0:17088      626        DENY_WRITE 0x20089     RDONLY     LEVEL_II         /smb/1c   db1c/kt/1Cv7.MD   Fri Jan 28 09:26:52 2011

I am currently working on version 3.1.1 with patch 6033. So far so good, and I really like, but I would like to in future releases of this error due to incorrect work with the oplock has not appeared. Since I believe that the use of glusterfs with samba - a very promising trend, despite the many criticisms glusterfs as file system for samba clusters.