kudu-backup.1

.TH KUDU-BACKUP 1 "August 2025" "Version 1.0" "User Commands"
.SH NAME
kudu-backup \- Simplify backing up and restoring Apache Kudu tables
.SH SYNOPSIS
.B kudu-backup
<backup|restore> [options] <space separated table names>
.SH DESCRIPTION
.B kudu-backup
is a utility for managing Apache Kudu table backups and restores.
It simplifies the process of creating and restoring backups by automatically searching for the required JAR and Kudu master addresses if they are not specified.

You must specify either
.I backup
or
.I restore
as the first command-line parameter. Otherwise you will see the help message.

Tables to operate on are specified as a space-separated list at the end of the command. If no tables are specified, you will see the help message.

Additional options can be passed directly to
.B spark-submit
.nf 
by\~appending\~--\~and the extra options.
.ni
If no options are provided, reasonable defaults will be used.
.TP
\- The jarPath will be automatically searched for.
.TP
\- The kuduMasterAddresses will be read from /etc/kudu/conf/master.conf.
.TP
\- Users that are allowed to administer the Kudu cluster will be real from /etc/kudu/conf/master.conf.

.SH OPTIONS
.SS Common Options
.TP
.BR --jarPath " <path>"
Path to kudu-backup jar file. If not defined, the jar will be automatically searched for.
.TP
.BR --kuduMasterAddresses " <addresses>"
Comma-separated list of Kudu master addresses. If not defined, it will be read from /etc/kudu/conf/master.conf.
.TP
.BR --rootPath " <path>"
Directory to store backup. Can be any Spark-compatible path. Defaults to hdfs:///kudu_backups.
.TP
.B --failOnFirstError
Fail the job as soon as a single operation fails.
.TP
.B --help
Show help message and exit.

.SS Backup Options
.TP
.BR --fromMs " <ms>"
Unix timestamp that defines the start time of an incremental backup.
.TP
.BR --timestampMs " <ms>"
Unix timestamp in milliseconds since the epoch to execute scans at.
.TP
.BR --scanBatchSize " <bytes>"
Maximum number of bytes returned by the scanner in each batch (default: 20MB).
.TP
.BR --scanRequestTimeoutMs " <ms>"
Maximum duration (ms) for each scan request (default: 30000 ms).
.TP
.BR --keepAlivePeriodMs " <ms>"
Period (ms) at which to send keep-alive requests to the tablet server.
.TP
.B --forceFull
Force a full backup even if one already exists (default: false).

.SS Restore Options
.TP
.BR --newDatabaseName " <name>"
If set, replaces the existing database name or adds a new one.
.TP
.BR --tableSuffix " <suffix>"
Adds the suffix to the restored table names.
.TP
.BR --timestampMs " <ms>"
Unix timestamp in milliseconds that defines the latest time to use when selecting restore candidates.
.TP
.B --createTables
Create the tables during restore. Do not specify this if the target tables already exist.
.TP
.B --removeImpalaPrefix
Removes the "impala::" prefix from restored table names.
.TP
.B --restoreOwner
Restores table ownership when creating new tables. Otherwise, the current user will be used.

.SH USAGE
You must specify either
.B backup
or
.B restore
as the first argument. This sets the class to be used for the rest of the command.

kudu-backup can be run with minimal arguments, or with explicit options for full control.

Tables must be specified as a space-separated list at the end of the command.

Additional options can be passed directly to 
.B spark-submit
by appending\~'\fB--\fR'\~and then the desired options.
.SH EXAMPLES
.TP
.B "# Backup multiple tables"
.PP
.nf
kudu-backup backup --failOnFirstError --forceFull my_first_table my_second_table
.fi
.TP
.B "# Restore multiple tables with specific options"
.PP
.nf
kudu-backup restore --removeImpalaPrefix --restoreOwner my_first_table my_second_table
.fi
.TP
.B "# Pass additional options to spark-submit"
.PP
.nf
kudu-backup backup my_table -- --failOnFirstError true spark.executor.memory=4g
.fi
.SH FILES
.TP
.I /etc/kudu/conf/master.conf
The default file used to discover Kudu master addresses, if not provided via --kuduMasterAddresses.
Also used to find Kudu superusers.
.TP
.I /usr/odp/<version>/kudu/jars
The folder containing bundled Kudu jars.
The <version> directory is your ODP version, which can be found by running
.nh 
"\fBodp\-select\fR\~--version"
.ny
.SH SEE ALSO
.TP
.BR odp-select (1)
Tool to select and display the current ODP version.

.TP
.BR Kudu
Apache Kudu documentation:
.IR https://kudu.apache.org/1.17.0/docs/

.TP
.BR spark-submit
.nf
Spark job submission tool. See:
.ni
.IR https://spark.apache.org/docs/latest/submitting-applications.html
.SH AUTHOR
Jeffrey Smith <jeffrey.smith@acceldata.io>
.SH COPYRIGHT
Copyright (C) 2025 Acceldata. All rights reserved.