CISAM 4.10 and NFS

We're using CISAM 5.00.UD1 under Solaris (SunOS 5.3). When several processes are simultaneously reading a NFS-mounted database using CISAM, the database becomes corrupted and all but one of the processes will block forever waiting on a lock. This seems to be a curruption of the database files, because any future processes will also block forever until the database is recreated from scratch. This is Informix's internal locking; we do not use any CISAM locking ourselves because we don't modify the database with these particular processes. (See the stack listing at the end.) So does anyone have any hints? Is CISAM unsafe over NFS? Do we need to upgrade to a newer version? Please respond via email, because I don't read this group regularly. - Bob (rlister@megatest.com) (debugger) where [1] _fcntl(0x8, 0x7, 0x5a38fc, 0xe2c280, 0x0, 0xfc442b80), at 0xef6a9120 [2] lksemaphore(0x21, 0x0, 0xe2c280, 0x1, 0x0, 0x1), at 0x3e599c [3] isenter(0xe2c110, 0x29, 0x5fc6d0, 0x30, 0x5c6c00, 0x30), at 0x3dca00 [4] isread(0x0, 0x61d0d0, 0x2, 0xe2c120, 0x0, 0x0), at 0x3dd7fc