% zelta-backup(8) | System Manager's Manual
zelta backup, zelta sync - Perform a recursive replication operation.
zelta backup [-bcdDeeFhhLMpuVw] [-iIjnpqRtTv] [initiator] source-endpoint target-endpoint
zelta sync [-bcdDeeFhhLMpuVw] [-iIjnpqRtTv] [initiator] source-endpoint target-endpoint
zelta clone [-d depth] source-dataset target-dataset
zelta backup and zelta sync attempt to intelligently replicate snapshots from a source ZFS dataset endpoint to a target. zelta backup optimizes for complete backups of all snapshots by default using a careful LBYL strategy, appropriate for typical backup jobs. zelta sync optimizes for efficiency using an EAFP strategy, appropriate for time-sensitive operations or controlled environemnts. Endpoints may be remotely accessible via SSH. As with zfs receive, the target dataset endpoint must not exist or be an replica of the source.
Zelta is designed for simplicity and safety and is suitable for a backup server replicating datasets from many systems. To ensure safe operation, the following default options are set for new replication targets:
These defaults, as well as snapshot naming scheme and many other assumptions, can be modified with arguments or via the environment (see zelta.env.example for more information).
See the manuals for zfs-send(8) and zfs-receive(8) for detail on pass-through options listed below.
[-bcdDeeFhhLMpuVw] [--zfs-send-option]
: Pass any unambiguous dashed or double-dashed option to all zfs send and zfs receive operations. Note that some options, such as -s, are ambiguous and must be set using the environment instead. Some options, such as -I, work differently in zelta <backup|sync>, and are described in detail below.
-I This is the default mode for zelta backup. Replicate intermediate incremental streams from the source and target's matching dataset to the newest. If the target is new, all avaialble source snapshots will be replicated.
-i: This is the default mode for zelta sync. Only replicate the latest stream from the source to the target. If the target is new, only the latest source snapshot will be replicated.
-n, --dryrun
: Run with dry run mode. Don't replicate, but show zfs <clone|create|get|receive|send> commands that would be run. Note that zelta match, zfs list, and zfs send -n which are used to determine the anticipated replication operation will not be displayed.
-j,--json
: Produce JSON output for the replication job, suppressing all other output. Stream names and error messages will be included in lists inside the JSON block. Incremental streams will be listed in the format: @earliest-snapshot::dataset@latest-snapshot
-p, --progress
: Attempt to use a progress viewer. pv will be used by default, otherwise dd status=progress will be used.
-q Reduce verbosity.
-V Increase verbosity.
-R Not recommended. Sets --depth=1 and passes -R to the zfs send. See zfs-send(8) for details.
-d depth, --depth depth
: Limits the depth of all Zelta operations.
Note that the same command can be used for new and existing target datasets.
Local synchronization: Synchronize a dataset and all of its snapshots from a local source dataset to a local target dataset, creating a snapshot only if necessary to get the latest data.
zelta backup tank/source/dataset tank/target/dataset
Migrate remote data to localhost: Create a snapshot and replicate it from a remote source to a local target, only if the source has new written data.
zelta sync -ss remote_host:tank/source/dataset tank/target/dataset
Dry Run: Display the zfs send and zfs receive commands without executing them.
zelta backup -n tank/source/dataset tank/target/dataset
ssh(1), zelta(8), zelta-match(8), zelta-policy(8), zfs(8)
Daniel J. Bell bellta@belltower.it