Compilation of the most important options from --help. Every agent option can also be used as key=value in the configuration file.
A.1 bkp_agent#
| Option | Meaning |
|---|---|
--source <path> | Source: file or block device (Windows: \\.\C:, \\.\PhysicalDrive<n>, a disk better by its ID disk-id:…; Linux: /dev/disk/by-id/…) |
--out <dir> | local sink: the store lies under <dir>/<store>.cbs (allowed several times) |
--target <host:port>, --target-dir <name> | Sink on a bkp_target; the directory must be in the ACL entry of the key |
--tls-ca <path> | pinned certificate of the server |
--api-key file:<path>|kms:<url> | API key; plain text only in the configuration file |
--sink <name> | write into the store of the section [sink:<name>] (allowed several times; replaces the sinks of the profile) |
--name <name> | Name of the backup, stored in every run; also selects the profile of the configuration |
--store <name> | Name of the store (default default) |
--type inc|full | inc (default): normal run; full: new store generation |
--store-generation none|<n>d|<n>m | automatically a new generation if the current one is older (default none) |
--chunk-size <avg> | average chunk size of a new store or a new generation (power of two, default 256K) |
--hash sha256|blake3|xxh3-128 | ID hash of a new store (default blake3); for an existing one it must match |
--threads <n>, --chunk-threads <n> | Workers (hash and compression) or segments chunked at the same time |
--chunk-probe yes|no | Chunk size measurement on the side (default yes) |
--compression zstd|lz4|none, --level <n>, --max-level <n> | Compression per new chunk; zstd level (default 3) and upper limit of the adaptive increase |
--verify | read the backup back from the store after it has finished (local stores only) |
--no-fs-bitmap | also back up free areas with ext2/3/4/NTFS (bit-identical copy) |
--retention <spec>, --retention-class <c> | Retention of this backup (<n>d|w|m|y, forever, gfs:<rules>) and class |
--worm none|extend|generation | WORM locks of a new store; in addition --worm-margin, --worm-max, --worm-retention-max |
--index-dir <dir> | put the index of the store outside the store |
--background, --background-min/-max <pct> | Self-throttling under foreign load |
--no-progress | provide no progress for bkp_monitor |
--snapshot auto|on|off | Snapshot behaviour (Linux LVM, Windows VSS) |
--snapshot-size <n> | Space for changes with thick snapshots (10%, 512M, 2G) |
--pre-snapshot <cmd>, --post-snapshot <cmd> | Hooks before and after the snapshot |
--hook-timeout <s> | Time limit for hooks (default 60) |
--keep-snapshot | leave the snapshot after the backup (for bkp_verifier) |
--vss-mode copy|full | Windows only: VSS_BT_COPY (default) or VSS_BT_FULL |
--vss-persistent auto|off | Windows only: try a persistent shadow copy for a resumable network backup (default auto) |
--net-retry <minutes> | after a connection loss retry this long (default 30, 0 = give up at once) |
--sink-stall-seconds <s> | drop an optional sink that holds up the backup this long (default 300) |
--resume auto|off | Handling of an interrupted run (default auto) |
--resume-max-age <hours> | an interrupted run whose last activity lies further back is not continued (default 24) |
--resume-dir <dir> | where interrupted runs are noted (default /var/lib/bkp_agent/resume or %ProgramData%\AegisGrid\resume); another directory gets a reference there |
--resume-pending | start and finish all interrupted runs again (for system start); remove runs idle for too long together with their snapshot |
--resume-cleanup | only remove runs idle for too long together with their snapshot, start nothing (bkp_sched calls it daily) |
--discard-interrupted | no backup: discard an interrupted run of this job (snapshot, state file); used by bkp_sched stop |
--result-file <file> | Result as JSON (status, run, full/incremental, sizes); used by bkp_sched |
--config-path <file> | Configuration file (default /etc/bkp/bkp_agent.cfg) |
--path <path> | back up a file or folder instead of a device (several times, up to 64; <folder>/* only the files in it) |
--exclude <pattern> | leave out what matches (several times); --default-excludes yes|no (default yes) |
--one-file-system yes|no | do not enter other mounted file systems (default yes) |
--ads yes|no, --change-journal yes|no | Windows only: also back up alternate data streams, use the NTFS change journal (default yes each) |
--ransom-check yes|no | Ransomware warning (default yes); --ransom-ratio <pct> (30), --ransom-factor <n> (5), --ransom-hold-days <n> (30) |
A.2 bkp_restore#
| Option | Meaning |
|---|---|
--store <dir> | Store directory (local) or with --server the store name |
--server <host:port>, --server-dir <name> | Restore directly from bkp_target |
--tls-ca, --api-key, --net-retry | Certificate, key and reconnection for the server connection |
--sink <name>, --config-path <file> | read from the sinks of a profile (selection and order) |
--archive <file|-> | from an archive (.cba) instead of a store; - reads from stdin |
--name <name> [--at <time>] | newest run of this name [up to this point in time] |
--run <run> | exactly this run (name from --list, UUID or run_group) |
--disk <slot>, --allow-incomplete | one disk of a VM run; also an incomplete VM run |
--list, --show-run | show runs or show the chosen run as JSON, restore nothing |
--list (without --name, --sink, --store, --server) | all backups of all profiles from bkp_agent.cfg, each once with its sinks; unreachable sinks quietly below the table |
--sinks | all sinks from bkp_agent.cfg and sinks.cfg: kind, place, reachable (with reason), which profiles use them; --json <file> for programs |
--target <path> | Target file or target device (is overwritten) |
--target-format raw|vhdx | Target file as a raw image (default) or as a new, growing VHDX (Windows; default with the extension .vhdx, at least 3 MB); an existing file is never overwritten |
--partition <n> | restore only partition n of a disk backup, onto a volume or into a file |
--new-disk-id auto|yes|no | new ID (GPT GUID or MBR signature) for the restored disk: auto only if the original is still in the machine |
--target-expect size=<bytes>,id=<id> | the target must still be the chosen one (size and ID), otherwise nothing is written (exit 6); set by the scheduler |
--decrypt-key file:<path> | Key of an encrypted store |
--threads <n> | Number of workers |
--assume-zeroed | Target contains only zeros; do not write holes |
--strict | --damage-file <path> | abort at the first damage | write damaged areas as JSON |
--index-dir <dir> | Index lies outside the store |
--background | Self-throttling under foreign load |
--no-progress | provide no progress |
--files ls|restore|find|versions|diff|zip|mount | File backups: show, restore, search, versions, changes, ZIP, show as a folder |
--to <folder>, --original | Target of --files restore (or archive with zip, folder with mount) |
--overwrite never|older|always, --as-version | existing files; put a version next to the current one |
--run2 <run>, --since/--until <time>, --paths-from <file> | Comparison run for diff; period for find; further paths |
--image-volumes, --image-mount, --volume <volume> | files from disk and VM runs (Linux as root, Windows as administrator) |
--idle <minutes> | end the view of --files mount/--image-mount after this much idle time |
--viewer <SID> | Windows: this user account may read the view of --files mount/--image-mount (e.g. in Explorer) |
--clear-anomaly | lift the protection after a ransomware warning |
A.3 bkp_sched#
| Command | Meaning |
|---|---|
list [<pattern>] [--status <s>] [--full] [--json] [--watch [s]] | all jobs in a table: status, next and last run, duration, after= |
show <job> [--json] | one job in detail: command, profile, sinks, schedule, next start, last run, state and the command for restoring |
preview <job> [--count <n>] | the next start times of a job |
run <job>, start <job>, stop <job> | run a job at once (in the foreground or in the background) or cancel it |
pause, resume, throttle, hold <job>… | --all | pause, continue, throttle or hold running jobs |
A.4 bkp_target#
| Option | Meaning |
|---|---|
--storage <dir> | Root directory of the stores |
--tls-cert <path>, --tls-key <path> | Certificate and private key of the server |
--acl-file <path> | ACL: <api_key> <dir1>[,<dir2>,...] per line |
--bind <address>, --port <n> | Listen address and port (default port 9990) |
--drain-timeout <s> | Waiting time for running sessions when ending (default 300) |
--index-dir <dir> | Indexes of the stores under <dir>/<store-uuid>.lmdb (mandatory with NFS/SMB) |
--client-retention extend|no | Whether clients may change the retention of their backups: extend (default) only extend, no not at all |
--sched-config <file>, --no-sched | Configuration of the target scheduler, or switch it off |
--sched-validate, --sched-list, --sched-preview <job>, --sched-run <job>, --sched-stop <job> | check the scheduler, show it, show the next starts of a job, start a job at once, cancel a job |
--sched-pause <job> [--for] [--reason], --sched-resume <job> | Pause or continue a running job |
A.5 bkp_maintenance#
| Command | Meaning |
|---|---|
store-info --store <d> [--exclusive] [--fragmentation] [--json] | Content, deduplication, last maintenance, chunk recommendation |
store-info --store <d> --names | who every backup name belongs to and which machines have runs |
prune (--search-dir <d> | --store <d>) [--name <n>] [--dry-run] | delete expired runs and data no longer needed |
delete --store <d> --run <run> [--dry-run] | delete a run and the data no other needs |
scrub (--store <d> | --search-dir <d>) [--decrypt-key file:<f>] | read and check all objects |
repair --store <d> --from <copy>|sink:<n>|server:<host:port> ... | replace the damage of the last scrub from a copy |
mirror (--store <d> | --search-dir <d>) --to <d>|server:<host:port> ... | copy a store into a second one (only what is missing) |
mirror --reencrypt --store <d>|sink:<n> --to <d>|sink:<n> [--to-key <k>] ... | copy runs with another key or ID hash into a store |
move --store <d> [--name <n>] --to <d>|server:<host:port> [--dry-run] | move a store or the runs of single names (rename or copy, then delete) |
export --store <d> (--run <r> | --name <n> [--at <t>]) --to <file|dir|-> | write a run as an archive |
import --archive <file|-> --to <d> [--dry-run] | bring an archive back into a store or check it |
reindex --store <d> | rebuild the index from the packs |
acl-add --acl-file <f> --dir <a>[,<b>] [--key <k>] | create an ACL entry; without --key a random key is created |
tls-init --cert <f> --key <f> [--cn <n>] [--days <n>] [--force] | create a self-signed ECDSA certificate (P-256) |
encrypt-init --key-file <f> [--force] | create an encryption key |
hwid [--json] | show the hardware fingerprint of this machine |
A.6 bkp_monitor, bkp_dump, bkp_restore_legacy#
| Call | Meaning |
|---|---|
bkp_monitor [--pid <pid>] | interactive display |
bkp_monitor --dump --pid <pid> [--interval <ms>] [--wait <s>] | one tab-separated line per interval |
bkp_dump <store> [--run <run> [--extents] [--limit <n>]] | show store, runs and manifest of a run (default 50 lines, 0 = all) |
bkp_restore_legacy list|restore|prune|delete ... | Backups in the format of version 1 (Chapter 27) |
A.7 bkp_updater and bkp_license#
| Call | Meaning |
|---|---|
bkp_updater --status | --check | --fetch <dir> | --update | --rollback | State, check, download, install, switch back |
bkp_updater --adopt <dir> | take over an existing installation |
bkp_updater --install-task | --remove-task | Windows: create or remove the task |
--config, --url, --channel, --public-key, --ca-file, --proxy, --state-dir | override the configuration |
--dry-run, --force, --to <v>, --allow-downgrade, --restart-now | Control of --update |
bkp_license status | show activation, validity of the installation certificate and license, for a trial version with remaining time, renewal and points |
bkp_license activate [--code-file <file>] [--server <url>] [--role <role>] | activate with the activation code at the license server (without --code-file it asks for the code) |
bkp_license offline-request [--out <file>] [--role <role>] | write the request for the activation |
bkp_license offline-apply <file> | apply the answer of the support |
bkp_license renew [--if-due] [--quiet] [--server <url>] | renew license and installation certificate at the license server; --if-due only when due (used by scheduler and updater); on a public target measure the customers' storage first |