diff options
| author | Jacob Anawalt <jlanawalt@gmail.com> | 2010-10-11 18:53:00 -0600 | 
|---|---|---|
| committer | intrigeri <intrigeri@boum.org> | 2010-10-12 10:49:34 +0200 | 
| commit | 7355535ee18efc472f5a598b8474e8d05765d6f1 (patch) | |
| tree | 8b7f48baa35b014974d2eaaf6f4d2e109264f2ce /handlers/pgsql.helper.in | |
| parent | cc35262c2a0a8153e7bbecef1af02ddf12869254 (diff) | |
| download | backupninja-7355535ee18efc472f5a598b8474e8d05765d6f1.tar.gz backupninja-7355535ee18efc472f5a598b8474e8d05765d6f1.tar.bz2  | |
Enable pg_dump format option.
The format option of pg_dump enables tar and custom archive file formats in
addition to the default plain-text file containing SQL commands.
When either the tar or custom format are selected the behaviour of database=all is changed to no longer dump a single file via pg_dumpall. Instead pg_dumpall
is called once to export the "global" data (roles & tablespaces) and then
pg_dump is called once for each non-template table in the database.
To support the GZIP and GZIP_OPTS variables in backupninja and to give the
default --rsyncable gzip compression flag a chance at working on a PostgreSQL
backup, the custom output is forced to not use compression. Instead compression
is done via a pipe to gzip. Hopefully this benefits rsync and rdiff-backup
style backups for reduced backup and storage costs that outweigh the
restoration ones.
Diffstat (limited to 'handlers/pgsql.helper.in')
| -rw-r--r-- | handlers/pgsql.helper.in | 30 | 
1 files changed, 30 insertions, 0 deletions
diff --git a/handlers/pgsql.helper.in b/handlers/pgsql.helper.in index ff1cfd4..82e6b48 100644 --- a/handlers/pgsql.helper.in +++ b/handlers/pgsql.helper.in @@ -67,6 +67,21 @@ pgsql_wizard() {        pgsql_compress="compress = no"     fi +   # pg_dump format, defaults to plain, custom is recommended by PostgreSQL +   menuBox "$pgsql_title" "Choose a pg_dump format:" \ +      plain "Default plain-text sql script, use with psql." \ +      tar "More flexible than the plain, use with pg_restore." \ +      custom "The most flexible format, use with pg_restore." +   if [ $? = 0 ]; then +      result="$REPLY" +      case "$result" in +         "tar") pgsql_format="format = tar";; +         "custom") pgsql_format="format = custom";; +         *) pgsql_format = "format = plain";; +      esac +   fi +          +     # write config file     get_next_filename $configdirectory/20.pgsql     cat >> $next_filename <<EOF @@ -97,7 +112,22 @@ $pgsql_databases  # if yes, compress the pg_dump/pg_dumpall output.  $pgsql_compress +# format = < plain | tar | custom > (default = plain) +# plain -  Output a plain-text SQL script file with the extension .sql. +#          When dumping all databases, a single file is created via pg_dumpall. +# tar -    Output a tar archive suitable for input into pg_restore. More  +#          flexible than plain and can be manipulated by standard Unix tools  +#          such as tar. Creates a globals.sql file and an archive per database. +# custom - Output a custom PostgreSQL pg_restore archive. This is the most +#          flexible format allowing selective import and reordering of database +#          objects at the time the database is restored via pg_restore. This +#          option creates a globals.sql file containing the cluster role and +#          other information dumped by pg_dumpall -g and a pg_restore file +#          per selected database. See the pg_dump and pg_restore man pages. +$pgsql_format +  ### You can also set the following variables in backupninja.conf: +# PSQL: psql path (default: /usr/bin/psql)  # PGSQLDUMP: pg_dump path (default: /usr/bin/pg_dump)  # PGSQLDUMPALL: pg_dumpall path (default: /usr/bin/pg_dumpall)  # PGSQLUSER: user running PostgreSQL (default: postgres)  | 
