next up previous contents index
Next: Master and slave Up: HEPDB callable routines Previous: Creating a new

Managing the database servers

     

Once the database file has been created, the server must be configured for this file. This is done using a NAMES file, as follows.

Names file entries for a database file (hepdb.names)

:nick.config
            :list.ge au
            :log./hepdb/cplear/logs
            :queue./hepdb/cplear/queue
            :todo./hepdb/cplear/todo
            :save./hepdb/cplear/save
            :bad./hepdb/cplear/bad
            :loglevel.3
            :wakeup.60
            :servers.cernvm vxcpon hepdb

:nick.ge
            :file./hepdb/cplear/database/geo.dbs
            :servers.vxcpon hepdb cernvm
            :desc.Geometry database for the CPLEAR experiment

:nick.au
            :file./hepdb/cplear/database/aux.dbs
            :servers.vxcpon cernvm
            :desc.Auxiliary database for the CPLEAR experiment
            :read.*
            :write.phr cpb

:nick.hepdb
            :userid.cdcplear
            :node.hepdb
            :localq./hepdb/l3/todo

:nick.vxcpon
            :userid.cdcplear
            :node.vxcpon
            :queue.disk$db:[cdcplear.todo]
            :protocol.tcpip
            :localq./hepdb/cplear/tovxcpon

:nick.cernvm
            :userid.cdcplear
            :node.cernvm
            :queue./hepdb/cplear/tocernvm
            :protocol.tcpip
            :localq./hepdb/cplear/tocernvm

The various tags in the preceding names file have the following meanings.

CONFIG
Configuration details for the server, as follows.
LIST
A list of two character database prefixes
LOG
The directory where the server logs are written
QUEUE
The directory where new updates are placed by HEPDB clients
TODO
The directory scanned by the HEPDB servers for updates to process. In the case of MASTER servers, the todo and queue directories are the same. In the case of SLAVE servers, these queues are different.
BAD
The directory where the server places bad updates. Bad updates are files for which the corresponding database cannot be found, or updates which cannot be successfully processed by the database server.
SAVE
The directory where the server saves updates after processing
LOGL
The log level for the server
WAKEUP
The wakeup interval in seconds for the server
SERVERS
This is the or of the list of servers for the individual databases. The database servers are responsible for moving updates to the local queues for the remote servers. A separate process, CDMOVE, is responsible for moving the processes between different systems.
prefix
The two character database prefix, e.g. aa.
FILE
The full name of the database file. For VM/CMS systems, the syntax is <user.address>filename.filetype
DESC
A comment string identifying the database and/or its purpose
SERVERS
The list of remote servers for this database. Each node in this list must also be in the list for the :nick.config entry.
READ
A list of users who may read the database. An asterisk grants read access to all users. If this tag is not present, read access control is not performed.
WRITE
A list of users who may update the database. Users with write access automatically gain read access. If this tag is not present, write access control is not performed.
server
The nickname of the servers, e.g. aa1.
USERID
Userid under which the server runs on the remote node
NODE
Node on which the server runs
QUEUE
Input queue on the remote node
PROTOCOL
Method by which updates are transmitted
LOCALQ
The local directory where updates are written pending transmission to the remote node. This may, in fact, be the same as QUEUE, e.g. when the directory is accessible via NFS or AFS.   




next up previous contents index
Next: Master and slave Up: HEPDB callable routines Previous: Creating a new


Janne Saarela
Tue May 16 10:31:09 METDST 1995