In this simple example, the journal file is written directly to the directory pointed to by the :queue tag in the hepdb.names file. As the server is the database master, the :queue and :todo tags point to the same directory. Whilst the journal file is being written, the reserved prefix zz is used. As soon as it is complete, the file is renamed to have the correct database prefix so that the server, which can of course handle several databases for the same experiment, can identify which database file is to be updated.
After the update has been processed, the master server sends the new
journal file to all slave servers.
In this case the update is first sent to the service machine CDHEPDB. This
machine receives the file and transfers it to the todo directory of the
appropriate server on hepdb. Once the update has been processed, the new
journal file will be written to a tovm directory and transferred back.
This file will then be sent to the local slave server on CERNVM and any
remote servers on Bitnet nodes.
Updating a database residing on node hepdb from CERNVM