Data Exportation Technicalities
This web page gives a
few technical hints on database exportation and importation. For more details,
one should refer to the detailed documentation
on Data Distribution Tools.
Anonymous FTP server :
Some subsets of data are available via anonymous
ftp on ftp-babar.slac.stanford.edu.
They are located in the directory: /babar/databases/export.
Conditions and configuration databases :
Snapshots of the conditions and configuration databases
are done on a regular basis. They all go in the following subdirectory
trees:
-
/babar/databases/export/snapshots/online/date/backup_type
-
/babar/databases/export/snapshots/online/date/backup_type
where "date"
is the date of the full DB backup.
and "backup_type"
is
either "full" for full backup or "incdate"
for incremental backups with respect to the corresponding "full" backup.
It can also be "gz"
for compressed versions.
How to install the conditions and configuration
databases in a remote site :
If you start from scratch :
-
Install a fresh fedration from a release directory
:
gmake database.import BYPASS_CONDITIONS_LOAD=yes BYPASS_CONFIG_LOAD=yes
Copy all the conditions and configuration export
files in a scratch directory starting from a full backup
Load the files in your database
BdbDmnImport -workdir the_scratch_dir export_file_name.export or .tdf
If you want to update an exiting database :
-
Set correctly your OO_FD_BOOT variable to point to
the federation bootfile
-
Copy all the conditions and configuration export
files in a scratch directory starting from a full backup
-
Load the files in the federation
BdbDmnImport -replace -workdir the_scratch_dir export_file_name.tdf
The -replace option will "force" the loading of the
database files even if a database with the same name already exists in
the federation.
Important :
If there is one full backup and some incrementals,
you should load the full backup first and the incrementals in the chronological
order.
Importing Collections :
The available collections are in the directory :
"Collections/version_number"
.
One should make sure to have an up to date database schema before trying
to load a collection.
The current version of the schema is stored in
: $BFROOT/Bdb/Reference/current/ and
it can be loaded in the database with the following command : (See : HN
posting from Dave Quarrie )
gmake database.schemaload SCHEMA_DIR=$BFROOT/Bdb/Reference/current
Copy the collections export files in a scratch directory
Load the files in the federation
BdbCollImport -force -workdir the_scratch_directory export_file_name.export or .tdf
The -force option is mandatory as a given exported
database file may have been already loaded in the database.
Exporting collections :
After setting the correct OO_FD_BOOT, the command
to export a collection from a database is the following
BdbCollExport -wait -workdir some_scratch_space -exclude raw:rec collection_list
For a long list of collections, it is possible to
write their names in a files and to do :
BdbCollExport -wait -workdir some_scratch_space -exclude raw:rec `cat file_name`
If there is a read lock on the database, the -wait
option will force the system to wait until the lock is removed
One can use the -lftp
option if the scratch space resides
on a file system which is not seen by the PUD daemon.
The -item options
allows to export only certain types of data (-item aod:tag will only export
aod and tag).
Massive exportation :
Massive exportation is working in the same way as
collection based exportation, but it provides the possibility to export
all the data since a given date (option -since).
The exporting command is BdbDistCenter
and is working in the same way as BdbCollExport.
|
