NAME
gbak -- Backup/restore Firebird database
SYNOPSIS
gbak -b [backup option ...] [common option ...] source target
gbak -c|-r [restore option ...] [common option ...] source target
DESCRIPTION
gbak is the tool for managing Firebird database backup and restore.
In backup mode source is an existing database path (may include server
name) and target is backup file name. It could be also a special device
name like /dev/stdout (to redirect backup data to STDOUT) or /dev/st to
put the backup on tape.
In restore mode source is existing backup file (or special device like
/dev/stdin for STDIN or /dev/st to get the backup from tape) and target
is path to firebird database and may include server name.
All options can be abbreviated. Below both full versions of options and
the shortest possible versions are given.
COMMON OPTIONS
-user username
User name to use when connecting to database. Overrides environment
variable. See ENVIRONMENT.
-ro[le] role_name
SQL role to use when connecting to database. Overrides environment
variable. See ENVIRONMENT.
-pas[sword] password
Password to use when connecting to database. Overrides environment
variable. See ENVIRONMENT.
-se[rvice]
Use services manager when talking to a database.
-v[erify]
Report each action.
Displays verbose information about what metadata is being processed,
what table is being backed up/restores and how many records are
processed.
-y path
Redirect status/error messages to the given path.
-z
Report full version number before doing anything else.
BACKUP OPTIONS
-b[ackup_database]
Turn on backup mode.
-co[nvert]
Convert external tables into tables.
Normally external tables are ignored during backup and restore. With
-convert option external tables are included in the backup as any other
regular table. When such a backup is restored, these previously
external tables are created as regular tables.
-fa[ctor] num
Blocking factor. Output is written in block of num bytes. Useful when
backing up to a tape.
-g[arbage_collect]
Inhibit garbage collection.
Normally Firebird runs a garbage collector when reading through the
tables. Since the backup process reads all the tables, this forces
garbage collection of the whole database. In case you make the backup
just to restore it over the existing database this garbage collection
is unnecessary. Another reason to use this options is if a garbage
collection is engaged by other means - either by using gfix or the
automatic sweep. Using this switch speeds up the backup process and
also can help in some cases of backing up damaged databases.
-ig[nore]
Ignore records with bad checksums.
This option should be used only when backing up damaged databases.
-l[imbo]
Ignore transactions in limbo.
This switch causes the backup process to ignore transactions in limbo
state.
-m[etadata_only]
Backup metadata only.
With this option only the metadata is written to the backup. This
includes all tables/views/triggers/indexes/stored
procedures/domains/generators etc. No table data is included in backup.
Such a backup is useful for creating an "empty" copy of a database.
-nt
Use non-transportable backup format.
By default a "transportable" backup format is used. This means you can
backup a database on little-endian machine and restore it on big-endian
machine and vice versa. By using -nt the backup is only useful on
machines with the same "endianness".
-ol[d_descriptions]
save old-style metadata descriptions
RESTORE OPTIONS
-c[reate_database]
Turn on restore mode.
target database will be created and must not already exist.
-r[ecreate_database] [o[verwrite]]
Like -create_database, but by adding overwrite one can force
replacement of the target database.
-rep[lace_database]
Like -create_database, but the target is overwritten if it exists.
-bu[ffers] num
When creating target database, ignore information about database
buffers stored in the backup and use specified number of buffers.
-e[xpand]
Normally, Firebird uses RLE compression for strings when storing them
on disk. This option turns this compression off.
-i[nactive]
Restore indexes in inactive state. This can be used to restore a
backup, containing data inconsistent with foreign key/unique
constraints.
-k[ill]
Do not create shadows when restoring.
Normally and database shadows are created during restore. This option
inhibits this.
-mo[de] access
Restore database with given access.
Valid values for access are read_only and read_write. Default is to
restore the database with its original access mode.
-n[o_validity]
Restore database without its validity constraints. This includes
foreign key, unique and check constraints. Useful for restoring a
backup, containing inconsistent data.
-o[ne_at_a_time]
Commit restoring transaction after restoring each table.
-p[age_size] num
Override page size stored in the backup. Valid values for num are
1024, 2048, 4096, 8192, 16384 and 32768.
-use_[all_space]
Normally, Fireburd reserves some space on each data page for further
use. This reserved space is used for newly inserted data and for
keeping older versions of the data. Having space reserved for this
purpose "near" to the data speeds up modifications. If the database
will used mainly for read operations, specifying -use_all_space will
save some space.
This option is particularly useful when restoring a read-only database,
since read-only databases do not need any additional space for back-
versions.
RESTORING TO MULTIPLE-FILE DATABASE
Previously, Firebird was not able to work with 64-bit file pointers
thus limiting databases to (about) 2GB of size per database.
Since 2GB is not much of data since years, there is a mechanism for
spreading the database into multiple files. This way you can have
multiple 2GB files, containing all your data.
When restoring to multiple-file database the target argument is of
form: file_1 pages_1 file_2 pages_2 ... file_N-1 pages_N-1 file_N
Each pages_n specifies at most how many pages to put in file_n. Last
in the list is a filename without page limit - it will contain all the
pages not fit in file_N-1. Note that pages_n is measured in pages, not
bytes so the maximum possible number depends on page size.
Imagine you have a filesystem (or an ancient OS) which only supports
32-bit file pointers. To play safe, you decide to split your database
on 2GB files. If the page size for the database is 8192, then each file
can have up to 2*1024*1024*1024/8192 = 262144 pages.
NOTE
Note that on a recent OS and filesystem Firebird can use 64-bit file
pointers rendering multi-file databases unnecessary.
ENVIRONMENT
As all other Firebird utilities, gbak accepts following environment
variables:
ISC_USER default value for -user option
ISC_ROLE default value for -role option
ISC_PASSWORD
default value for -password option
EXAMPLES
In all examples -user, -role and -password options are omitted for
clarity. In a real world situation they (or their corresponding
enviromnemt variables) should be used.
Note that filename extensions used here are just recommended. Using
unified extensions scheme helps guess file type just by looking at its
extension.
Here are some commonly used extensions:
.fdb Firebird database
.gdb Firebird database, legacy extension from the days when
Firebird was Interbase. gdb actually comes from Grotton
database, named after the company that created the software
back in 1984.
.fdb.2 Second file of multi-file database
.fdb.3 Third file of multi-file database
.fdb.N N-th file of multi-file database
.fbk Firebird backup file
.gbk Legacy extension for backup file
.fbk.gz Gzipped backup file
Backup a database into a compressed format:
gbak -b db-srv:/database.fdb /dev/stdout | gzip > /file.fbk.gz
Restore a database into new filename:
zcat /file.fbk.gz | gbak -c /dev/stdin db-srv:/new-database.fdb
AUTHOR
This manpage was written by Damyan Ivanov <dam@modsoftsys.com> for
Debian GNU/Linux but may be used by others. Permission is granted to
use this document, with or without modifications, provided that this
notice is retained. If we meet some day, and you think this stuff is
worth it, you can buy me a beer in return.
COPYRIGHT
(C) 2004,2007 Damyan Ivanov