Example of configuration on two VAX systems

Let us take the example of two VAX systems: VXLNFB and AXPALS. In this example, the master server will run on the node VXLNFB. To achieve this, we must configue the file hepdb.names so that the :queue and :todo tags point to the same directory, as shown below.

Example names file for master server

:nick.config
:list.ct rt gt
:log.DISK\$MF:[KLOEDB.LOG]
:queue.DISK\$MF:[KLOEDB.TODO]
:todo.DISK\$MF:[KLOEDB.TODO]
:save.DISK\$MF:[KLOEDB.SAVE]
:bad.DISK\$MF:[KLOEDB.BAD]
:loglevel.3
:wakeup.60
:servers.axpals

:nick.ct
:file.DISK\$MF:[KLOEDB.DATABASE]CALRT.DBS
:desc.Calibration Database for Test Beam 1994
:servers.axpals

:nick.rt
:file.DISK\$MF:[KLOEDB.DATABASE]RUNRT.DBS
:desc.RUN CONDITION Database for Test Beam 1994
:servers.axpals

:nick.gt
:file.DISK\$MF:[KLOEDB.DATABASE]GEORT.DBS
:desc.Geometry Database for Test Beam 1994
:servers.axpals

:nick.axpals
:userid.kloedb
:node.axpals
:poll.yes

The names file for the slave server is similar, as shown below.

Example names file for slave server

:nick.config
:list.ct rt gt
:log.DISK\$MF:[KLOEDB.LOG]
:queue.DISK\$MF:[KLOEDB.TOVXLNFB]
:todo.DISK\$MF:[KLOEDB.TODO]
:save.DISK\$MF:[KLOEDB.SAVE]
:bad.DISK\$MF:[KLOEDB.BAD]
:loglevel.3
:wakeup.60
:servers.

:nick.ct
:file.DISK\$MF:[KLOEDB.DATABASE]CALRT.DBS
:desc.Calibration Database for Test Beam 1994
:servers.vxlnfb

:nick.rt
:file.DISK\$MF:[KLOEDB.DATABASE]RUNRT.DBS
:desc.RUN CONDITION Database for Test Beam 1994
:servers.vxlnfb

:nick.gt
:file.DISK\$MF:[KLOEDB.DATABASE]GEORT.DBS
:desc.Geometry Database for Test Beam 1994
:servers.vxlnfb

In the above examples, it is assumed that CDMOVE will run on the same node as the master server, e.g. VXLNFB.

Making an update on a node with a MASTER server

• If an update is made on the node where the master server is running, it is written into the directory DISK$MF:[KLOEDB.TODO].
• The master server will process this update and make a copy in the directory DISK$MF:[KLOEDB.TOAXPALS].
• A seperate process, CDMOVE takes the files in DISK$MF:[KLOEDB.TOAXPALS], connects to the node AXPALS and transfers the files to the directory DISK$MF:[KLOEDB.TODO].
• It will then be processed by the CDSERV process on this node and the local copy of the database updated.

Making an update on a node with a SLAVE server

• If an update is made on the node where the slave server is running, it will be written into the directory DISK$MF:[KLOEDB.TOVXLNFB].
• The CDMOVE process running on VXLNFB will connect at periodic intervals and transfer any pending files, as we have specified the tag :poll.yes, the CDMOVE Alternatively, one may specify :receive.yes, in which case updates will only be transferred from AXPALS if the CDMOVE server has made a connection to transfer updates from VXLNFB to AXPALS. If neither tag is specified, updates will never be transferred from AXPALS to VXLNFB.

Running CDMOVE on the slave node

Normally, one would run the CDMOVE process on the same node as the master server. As stated above, this requires that ZSERV has been correctly configured on all slave nodes.

If this is not possible for some reason, e.g. if UCX is running on the slave node, then CDMOVE can be run on the slave node. This, in turn, requires that ZSERV is correctly configured on the master node.

In this case the names file on the slave node must be modified as shown below. The changes are indented for clarity, although this is purely cosmetic.

Example names file for slave server with changes for CDMOVE

:nick.config
:list.ct rt gt
:log.DISK\$MF:[KLOEDB.LOG]
:queue.DISK\$MF:[KLOEDB.TOVXLNFB]
:todo.DISK\$MF:[KLOEDB.TODO]
:save.DISK\$MF:[KLOEDB.SAVE]
:bad.DISK\$MF:[KLOEDB.BAD]
:loglevel.3
:wakeup.60
   :servers.VXLNFB

:nick.ct
:file.DISK\$MF:[KLOEDB.DATABASE]CALRT.DBS
:desc.Calibration Database for Test Beam 1994
:servers.vxlnfb

:nick.rt
:file.DISK\$MF:[KLOEDB.DATABASE]RUNRT.DBS
:desc.RUN CONDITION Database for Test Beam 1994
:servers.vxlnfb

:nick.gt
:file.DISK\$MF:[KLOEDB.DATABASE]GEORT.DBS
:desc.Geometry Database for Test Beam 1994
:servers.vxlnfb

   :nick.VXLNFB
   :userid.kloedb
   :node.VXLNFB
   :poll.yes