N.B. The use of UCX is not recommended. The following section remains for historical reasons only.
At the time of writing, DEC's UCX product still does not provide an NFS client. In this case one can mount a VMS directory on node hepdb. This is done today for CPLEAR.
As is the case for Multinet NFS, one must map a Unix UID/GID pair to a VMS username. In addition, a binding must be made been a VMS directory and a Unix style file name.
This can be done as follows:
Binding a VMS directory to a Unix name
$ UCX UCX> BIND UXCP05$DKA300: /vxcplear UCX> show bind Logical filesystem Pathname UXCP05$DKA300: /vxcplear UCX>
Mapping a UID/GID pair to a VMS username
$ UCX UCX> ADD PROXY CDCPLEAR /UID=102 /GID=1 /HOST=hepdb.cern.ch
Note that UCX treats hostnames as case sensitive.
Finally, one must start the UCX NFS server. This involves
Modifying the UCX NFS startup command file
$ ! ...
$ !
$ ! Set the following UID and GIDs
$ !
$ DEFINE/SYSTEM/EXE/NOLOG UCX$NFS00000000_GID 1
$ DEFINE/SYSTEM/EXE/NOLOG UCX$NFS00000000_UID 0
$ !
$ ! ...
$ !
$ ! Comment out the following line
$ ! RUN SYS\$SYSTEM:UCX\$SERVER\_NFS.EXE
$ !
$ ! The following section contains NFS process quota that is required by
$ ! manual startup. Please uncomment the following lines and comment out
$ ! the "RUN" command above, if you choose to manually start NFS.
$ !
$ RUN SYS\$SYSTEM:UCX\$SERVER\_NFS.EXE/DETACH -
/OUTPUT=NLA0: -
/ERROR='P1' -
/AST\_LIMIT=512 -
/BUFFER\_LIMIT=200000 -
/EXTENT=20000 -
/FILE\_LIMIT=1024 -
/IO\_BUFFERED=400 -
/IO\_DIRECT=200 -
/QUEUE\_LIMIT=64 -
/ENQUEUE\_LIMIT=3000 -
/MAXIMUM\_WORKING\_SET=20000 -
/PAGE\_FILE=20000 -
/PRIORITY=8 -
/PRIVILEGES=(BYPASS,SYSPRV) -
/UIC=[1,4] -
/NORESOURCE
$ !
$EXIT:
$ EXIT
The file system is now ready for mounting on hepdb.
Extract from /etc/filesystems for /vxcplear
/vxcplear:
dev = /vxcplear/cdcplear
vfs = nfs
nodename = uxcp05
mount = true
options = bg,hard,intr