One may avoid running a local database server on a given node by accessing the database files over the network. This is the recommended procedure for Unix systems at CERN. To enable this, one should first mount the /hepdb file system as shown below.
Mounting the /hepdb file system on a machine running Unix
mount hepdb:/hepdb /hepdb
Should your experiment require access to HEPDB from one of the CORE/SHIFT/CSF systems, please contact your CORE representative and ask them to perform the above action on the nodes in question.
On a VAX/VMS system that has the NFS client software installed, as is the case on VXCERN, the following commands are issued at system startup time.
Mounting the /hepdb file system on a machine running VMS
$ !
$ ! Mount the file system if not already done
$ !
$ if f$trnlm("HEPDB").eqs."" then NFSMOUNT/soft HEPDB::"/hepdb" HEPDB
The HEPDB software automatically uses C I/O to access remote database files on VMS systems. This is because VAX Fortran does not recognise the file structure of the remote Unix database file but is in any case completely transparent to the user.
It is currently recommended that the update directory reside on the local VMS system. This is because Multinet NFS requires that VMS UICs are mapped to Unix UID and GID pairs on the remote node, even if the remote directory is world writable (or writable by others in Unix parlence). On VXCERN only a single UIC is mapped to a valid UID/GID pair on node hepdb. A job runs under this UIC to move the update files between the local and remote file systems.