The nbgpbackup_config.yaml file

The backup and restore commands for Greenplum require the use of a plug-in configuration file in yaml format. The example script creates a plug-in configuration file when the backup policy runs.

An example yaml file is installed with the NetBackup for Greenplum plug-in at the following location:

/usr/openv/netbackup/ext/db_ext/greenplum/scripts/nbgpbackup_config.yaml

If you want to run gpbackup or gprestore from the command line you must create your own nbgpbackup_config.yaml file.

Note: Do not save custom scripts in the scripts directory. Subsequent NetBackup upgrades can overwrite your site's scripts.

Supported options for the nbgpbackup_config.yaml file

The following options are supported for the nbgpbackup_config.yaml file if you need to customize how backups or restores run or improve performance.

Table: Options in the nbgpbackup_config.yaml file

Option	Description	Backup option	Restore option
buffer_size	The size of the shared memory buffer to use, in bytes. It should be a multiple of 512 bytes. The default is `512`. This setting also applies when you use the option incremental_optimize_restore.	Yes	Yes
coordinator	The host name of the coordinator host.	No	Yes
coordinator_hosts	A comma-separated list of host names. This list is used for a redirected restore to another client to find files at restore time.	No	Yes
end_image_timeout	The number of seconds for a NetBackup job to wait for another file before it ends the image. The default is 15 seconds.	Yes	No
error_shutdown_timer	The number of seconds for a nbgpbackup worker process to stay alive without doing anything. The default is 36000 seconds.	Yes	No
incremental_optimize_restore	Enables an optimized restore that creates a job per image. No disk cache is used. The default is off. When set to on, this option requires the coordinator option.	No	Yes
keyword	A keyword that can be associated with NetBackup images. Corresponds to NBBSA_KEYWORD.	Yes	No
number_of_buffers	The number of shared memory buffers to use per shared memory segment. The default is 30. This setting also applies when you use the option incremental_optimize_restore.	Yes	Yes
number_of_streams	The number of NetBackup backup streams to use per host. See Best practices for Greenplum backups.	Yes	No
optimize_restore	Enables an optimized restore which caches entire images to disk so the subsequent file restores are faster. Options are `on` or `off`. The default is `off`. Requires that optimized_restore_directory is also specified. The NetBackup for Greenplum agent makes every attempt to clean up the `cache` directory. However, if there are errors during the restore you may have to manually clean up the cache directory on all segment hosts.	No	Yes
optimize_restore_directory	The parent directory to use for an optimized restore. The directory that is specified is created if it doesn't exist. Any parent directories are not created. For example, if `/a/b/c` is specified then nbgpbackup tries to create the `c` directory if it isn't present. It does not try to create `/a/b`. The directory is deleted after a successful restore.	No	Yes
policy	The NetBackup policy to use for backup. This option is required for a backup. Corresponds to NBBSA_POLICY.	Yes	No
read_delay	The number of milliseconds to wait before NetBackup evaluates if a shared memory buffer is full. The default is 10 milliseconds.	Yes	Yes
restore_timeout	The number of seconds for an incremental optimized restore to wait for an empty buffer. The default is 300 seconds. If you adjust the Client read timeout setting in the media server host properties, it is recommended that you also set the `restore_timeout` to the same value.	No	Yes
schedule	The NetBackup schedule to use for a backup. Required for a backup. Corresponds to NBBSA_SCHEDULE.	Yes	No
server	The NetBackup primary server to use. If not specified then the first server that is listed in the `bp.conf` is used. Corresponds to BSA_SERVICE_HOST.	Yes	Yes
segment_hosts	A comma-separated list of host names. This list is used for a redirected restore to another client to find files at restore time.	No	Yes
server_read_timeout	The number of seconds that bpbrm on the media server waits before it times out. This setting overrides the `client_read_timeout` value on the media server that is configured for bpbrm. The default value is `0` seconds, which instructs NetBackup to use the `client_read_timeout` value for the media server.	Yes	Yes
shutdown_process_timer	The number of seconds for any nbgpbackup worker processes to stay alive after they think all backup operations are complete. The default is `2` seconds.	Yes	No
write_delay	The number of milliseconds to wait before NetBackup verifies that a shared memory buffer is empty. The default is `15`.	Yes	Yes

The nbgpbackup_config.yaml file

Feedback

Feedback