Command line reference

Every lakana command and sub-command, with its arguments and options, in the order lakana --help lists them. The options of the first section go before the command name, as in lakana --output json whoami.

Options for every command

Usage: lakana [OPTIONS] COMMAND [ARGS]...

lakana login

Usage: lakana login [OPTIONS]

Verify a credential and save it as a profile.

lakana logout

Usage: lakana logout [OPTIONS]

Remove a saved profile (and its token) from the profile file.

lakana whoami

Usage: lakana whoami [OPTIONS]

Show who the configured credential authenticates as.

lakana sql

Usage: lakana sql [OPTIONS] {query}

Run SQL and print (or save) the result, or write it into a table.

With --into T --mode create --expire-after N the new table expires N days after its last write or read.

A query that stays queued for 3 seconds prints a line on standard error that says why ('Waiting: ...'), and again when the reason changes; the result still goes to standard output only. Nothing is printed about waiting with --output json.

lakana deploy

Usage: lakana deploy [OPTIONS] [repo]

Deploy REPO to an environment; exit 1 if the deploy fails.

lakana rollback

Usage: lakana rollback [OPTIONS] [repo]

Move REPO's deployment back without fetching from git.

lakana undeploy

Usage: lakana undeploy [OPTIONS] [repo]

Remove REPO's deployment and the identity its runs use; tables and history stay.

lakana orphans

Usage: lakana orphans [OPTIONS]

List tables earlier project versions produced that no current one does.

lakana run

Usage: lakana run [OPTIONS] [repo]

Run REPO's deployed project version; exit 0 on success, 1 otherwise, 7 if one is active.

lakana search

Usage: lakana search [OPTIONS] {TEXT}

Find tables, columns, descriptions, runs and people that match TEXT.

Matching ignores case and finds TEXT anywhere in a name or description. Exact matches come first, then names that start with TEXT, then names that contain it, then descriptions. Runs are found by the first four or more characters of their id, their job or their repository. You see only what you may read. Use --output json for every field.

lakana init

Usage: lakana init [OPTIONS] [directory]

Create a Lakana project that deploys unchanged.

lakana profile

Usage: lakana profile [OPTIONS] COMMAND [ARGS]...

Manage saved profiles (local only).

lakana profile list

Usage: lakana profile list [OPTIONS]

List saved profiles; tokens are never shown.

lakana profile use

Usage: lakana profile use [OPTIONS] {name}

Make NAME the default profile.

lakana profile remove

Usage: lakana profile remove [OPTIONS] {name}

Delete profile NAME and its token.

lakana env

Usage: lakana env [OPTIONS] COMMAND [ARGS]...

Manage environments.

lakana env create

Usage: lakana env create [OPTIONS] {name}

Create environment NAME (with bronze/silver/gold namespaces).

Write the storage root the one way Lakana compares locations: a lower-case scheme (s3, not s3a), one slash between path segments and none doubled after file:// (file:///data, not file:////data), for file an absolute path and no host (not file://data or file://localhost/data), no . or .. segments, no percent escapes and no *, ?, [ or ]. One trailing slash is fine. Another spelling is refused, and the message names the one to use.

lakana env list

Usage: lakana env list [OPTIONS]

List environments you can see, or with --deleted the deleted ones.

A deleted environment keeps its tables' files until purge_after (the server's undo period, 7 days by default); until then lakana env restore NAME brings it back.

lakana env get

Usage: lakana env get [OPTIONS] {name}

Show environment NAME.

lakana env delete

Usage: lakana env delete [OPTIONS] {name}

Delete environment NAME. Its tables' files are kept for the undo period, then deleted.

Within the server's undo period (LAKANA_TABLES__UNDO_DAYS, 7 days by default), lakana env restore NAME brings the environment back with its tables, their data and their grants. Until then the name stays taken, and lakana env list --deleted shows until when. Views and external tables are removed; an external table's files are never deleted. --purge skips the period.

Refused (409) while any repository is deployed to it: run lakana undeploy REPO --env NAME for each first. Refused while protected.

lakana env restore

Usage: lakana env restore [OPTIONS] {name}

Restore the deleted environment NAME, with the tables, data and grants it had.

Works until the end of the undo period that lakana env list --deleted shows, and not after lakana env delete NAME --purge. A table dropped before the environment was deleted stays dropped; restore it afterwards with lakana table undrop. Needs MANAGE_ENVIRONMENT (deleting the environment revoked the grants given on it, so usually an administrator's).

lakana env set-storage

Usage: lakana env set-storage [OPTIONS] {name}

Change environment NAME's storage options or credentials secret.

While the environment has a storage secret, changing the options also needs the right to manage secrets in it (the ADMIN role), because an option such as s3.endpoint decides where the secret is sent. --option replaces all the options, so pass every one the environment should have. Attaching a secret approves the options it has then.

lakana env protect

Usage: lakana env protect [OPTIONS] {name}

Protect environment NAME: lock writes down to deployed runs.

A protected environment refuses every write outside runs of the deployed code (SQL, SDK, CLI and Iceberg REST clients alike) whatever the caller's grants; choose the identity a deployed job runs as with lakana deployments set-run-as. A holder of MANAGE_ENVIRONMENT can make one audited write with --break-glass --reason. It also refuses deletion and storage changes.

lakana env unprotect

Usage: lakana env unprotect [OPTIONS] {name}

Remove protection from environment NAME.

lakana env retention

Usage: lakana env retention [OPTIONS] {name}

Show environment NAME's retention, or set it with --days or turn it off with --off.

The retention applies to every managed table in the environment that sets none of its own (lakana table retention). Such a table expires N days after its last activity, and never sooner than N days after this command, so turning it on gives every table its full period. Expired tables are dropped by the worker's clean-up and can be restored with lakana table undrop during the undo period. Off unless set. Needs MANAGE_ENVIRONMENT; refused for a protected environment, whose tables never expire.

lakana env orphaned-locations

Usage: lakana env orphaned-locations [OPTIONS] {name}

List, or with --delete remove, table directories older releases left without a record.

Releases before location records could leave table directories under managed/ that no table and no record names: tables dropped before the upgrade, creates that failed, environments deleted before their tables could be restored. The worker's clean-up only counts them. Without --delete this is a dry run: it lists each directory with its files, size and newest file, and deletes nothing. A directory with a file newer than --older-than (30 days by default) is never deleted. --delete asks first (--yes skips the question), checks every directory again, and keeps one that a table, a create or a dropped table now uses. NAME may be a deleted environment. Needs MANAGE_ENVIRONMENT on the whole instance (the ADMIN role), because the storage root may be shared. After restoring the catalog from a backup, keep --older-than above the backup's age: a table created after the backup has no record either.

lakana ns

Usage: lakana ns [OPTIONS] COMMAND [ARGS]...

Manage namespaces.

lakana ns create

Usage: lakana ns create [OPTIONS] {path}

Create namespace PATH (dotted, e.g. bronze.eu).

lakana ns list

Usage: lakana ns list [OPTIONS]

List namespaces.

lakana ns get

Usage: lakana ns get [OPTIONS] {path}

Show namespace PATH.

lakana ns delete

Usage: lakana ns delete [OPTIONS] {path}

Delete namespace PATH.

lakana table

Usage: lakana table [OPTIONS] COMMAND [ARGS]...

Manage tables and load data.

lakana table create

Usage: lakana table create [OPTIONS] {name}

Create managed table NAME (namespace.table).

With --expire-after N the table expires N days after its last write or read: the worker's clean-up then drops it, and lakana table undrop restores it during the undo period.

lakana table create-external

Usage: lakana table create-external [OPTIONS] {name}

Register external table NAME over files under raw/.

lakana table list

Usage: lakana table list [OPTIONS] {namespace}

List the tables in NAMESPACE, or with --dropped the tables dropped from it.

A dropped table keeps its files until purge_after (the server's undo period, 7 days by default); until then lakana table undrop restores it. A table that its retention lets expire within the server's warning window (7 days by default) shows when, in expires.

lakana table get

Usage: lakana table get [OPTIONS] {name}

Show table or view NAME, its columns, and a view's SQL and references.

A table whose first build did not finish, or failed a check, is shown as not published; one that keeps the data that failed a check (on_failure: keep), as kept. A table that its retention lets expire within the server's warning window says when. An artifact is shown with its name and a pointer to lakana artifact versions list.

lakana table health

Usage: lakana table health [OPTIONS] {name}

Show table NAME's last maintenance, the newest result of each check, and who reads it.

Readers are counted from the queries that succeeded, per person and day. You see them only with the ADMIN role. A Python task's reads of its inputs count too. Use --output json for every field.

lakana table describe

Usage: lakana table describe [OPTIONS] {name}

Describe table NAME or one of its columns, or reset the description.

The text you set is shown instead of the description: in the asset's header, and a deploy never replaces it. --reset goes back to the header's text. You need to own the table, or hold the right to drop it.

lakana table delete

Usage: lakana table delete [OPTIONS] {name}

Drop table NAME. Its files are kept for the undo period, then deleted.

Within the server's undo period (LAKANA_TABLES__UNDO_DAYS, 7 days by default), lakana table undrop NAME restores the table with its data and its grants, row filters and column masks. --purge skips the period: the worker deletes the files at its next clean-up. An external table's files are never deleted.

lakana table undrop

Usage: lakana table undrop [OPTIONS] {name}

Restore the dropped table NAME, with its data, grants, row filters and column masks.

Works until the end of the undo period that lakana table list NAMESPACE --dropped shows. Needs the rights to drop and to create tables in the namespace. When NAME was dropped more than once, the newest comes back; restore an older one with --as.

lakana table append

Usage: lakana table append [OPTIONS] {name} {file}

Append FILE (Parquet, CSV or Arrow) to table NAME.

The table's schema is fetched first. A CSV is parsed straight into the table's column types; Parquet and Arrow are cast to them. Nullability is checked too, and a column that doesn't fit fails here, before anything is sent.

lakana table replace

Usage: lakana table replace [OPTIONS] {name} {file}

Replace every row of table NAME with FILE (Parquet, CSV or Arrow).

Without --evolve-schema the file is conformed to the table's columns first, exactly as table append does. With it, the file is sent as it is and the table takes its schema.

lakana table maintain

Usage: lakana table maintain [OPTIONS] {name}

Expire table NAME's old snapshots, delete its orphaned files, and remove failed-check tags.

A snapshot kept under the tag of a failed check is removed with that tag once it is older than 7 days. A snapshot that a run still in progress needs to set its table back is kept.

lakana table lineage

Usage: lakana table lineage [OPTIONS] {name}

Show table NAME's lineage as a tree (or JSON with --output json).

lakana table retention

Usage: lakana table retention [OPTIONS] {name}

Show when managed table NAME expires, or set it with --days, --on or --never.

An expired table is dropped by the worker's clean-up and can be restored with lakana table undrop during the undo period. Its last activity is the later of its last write (or other change) and its last read through Lakana's queries. Tables in a protected environment and tables a deployed project builds never expire.

Without an option this shows the table's retention. The table's owner may set any. Otherwise, making it expire sooner (or at all) needs the right to drop the table; making it expire later, or --never, needs the right to write it.

lakana table expiring

Usage: lakana table expiring [OPTIONS]

List the managed tables that their retention lets expire soon, soonest first.

Each row says when, the rule (and whether the table or its environment sets it) and who set it. Keep a table with lakana table retention NAME --never, or give it more days.

lakana artifact

Usage: lakana artifact [OPTIONS] COMMAND [ARGS]...

Manage artifacts: versioned files such as models.

lakana artifact upload

Usage: lakana artifact upload [OPTIONS] {name} {path}

Store PATH (a file or a directory) as a new version of artifact NAME (namespace.name).

Everything in a directory is stored, including hidden files such as .env: point the command at a directory that holds only the artifact. Only files the store does not have yet are sent. Links, devices and Windows-unsafe names are refused before anything is sent; the message names the path.

lakana artifact list

Usage: lakana artifact list [OPTIONS]

List the artifacts you may read, with their current version and size.

lakana artifact drop

Usage: lakana artifact drop [OPTIONS] {name}

Drop artifact NAME (namespace.name): every version, its files and its grants.

This cannot be undone. The name becomes free; a new artifact of that name starts at version 1. The files are removed inside the request, so an artifact with thousands of files can take minutes; if the server fails part-way, Lakana's periodic clean-up finishes the job. Needs DROP_TABLE on the artifact, which a task never has. In a protected environment pass --break-glass --reason.

lakana artifact download

Usage: lakana artifact download [OPTIONS] {name} {directory}

Download a version of artifact NAME (namespace.name) into DIRECTORY.

DIRECTORY must not exist, or be empty. The files appear there only when all of them have arrived and match the manifest; if anything goes wrong nothing is left at DIRECTORY (a directory that did not exist is not there, one that existed is empty again) and the message says what to do. An empty directory is never replaced: it keeps its place and its mode, and the files are moved into it at the end. Nothing is written through a link.

The copy is private to you, because the artifact's contents are access-controlled on the server: files are 0600 and directories 0700 (less your umask), nothing is executable. To open the copy to others run chmod -R a+rX DIR. (A read-only copy, files 0400 and directories 0500, is what a task gets as its input; remove one with chmod -R u+w DIR first.)

lakana artifact versions

Usage: lakana artifact versions [OPTIONS] COMMAND [ARGS]...

Show the versions of an artifact: when each was made, its size, its file count, its digest, and whether it was deleted.

lakana artifact versions list

Usage: lakana artifact versions list [OPTIONS] {name}

List the versions of artifact NAME (namespace.name), newest first.

A deleted version is listed too, with when it was deleted; its bytes are gone. Version numbers can have gaps. If an upload said the server did not confirm its commit, look here for a version with the manifest_sha256 it names.

On a narrow terminal the table leaves out columns, the digest last, and says which; the JSON form always has them all and the whole digest: lakana --output json artifact versions list NAME.

lakana artifact versions delete

Usage: lakana artifact versions delete [OPTIONS] {name} {version}

Delete one version of artifact NAME (namespace.name). This cannot be undone.

The version's files are removed from Lakana's storage, except a file another version still has. A bucket that keeps old object versions, and a backup, still have them. The version stays in versions list, marked deleted, with who deleted it and when; its number is never reused. The command says how many files it freed and how many it kept, and why: another version has them, an open upload names them, or Lakana's periodic clean-up decides about them (for example for a large version, a manifest the catalog does not know after a restore, a damaged manifest or a storage failure). Needs DROP_TABLE on the artifact, which a task never has. In a protected environment pass --break-glass --reason.

lakana artifact uploads

Usage: lakana artifact uploads [OPTIONS] COMMAND [ARGS]...

List and withdraw uploads that were begun and not finished. lakana artifact upload withdraws its own when it fails; use list to find one a killed process left behind and abort to withdraw it.

lakana artifact uploads list

Usage: lakana artifact uploads list [OPTIONS] [name]

List your own open uploads, newest first: the id, the artifact, when it began and expires.

Only your own: another person's uploads are never listed. An upload that was begun and never finished (after a killed process, a lost connection or a power cut) holds one of its artifact's 16 upload places until it expires, one hour by default. Withdraw one with lakana artifact uploads abort NAME UPLOAD_ID. The list does not say how many files an upload has sent.

lakana artifact uploads abort

Usage: lakana artifact uploads abort [OPTIONS] {name} {upload_id}

Withdraw upload UPLOAD_ID of artifact NAME (namespace.name) that was never finished.

lakana artifact upload does this by itself when it fails or is interrupted, so you rarely need it. An artifact allows 16 open uploads at once; an abandoned one frees its place when it is withdrawn or after the upload lifetime (one hour by default). Only the person (or service account) that began the upload can withdraw it, and only while it is open. Files it already stored stay until Lakana's periodic clean-up finds that nothing needs them.

lakana query

Usage: lakana query [OPTIONS] COMMAND [ARGS]...

Submit queries and fetch their results step by step.

lakana query list

Usage: lakana query list [OPTIONS]

List the queued and running queries of an environment.

Running queries come first, then queued ones in the order they will be taken, with the reason each waits. You see your own queries; with MANAGE_ENVIRONMENT on the environment, everyone's, and --submitter narrows them to one principal's. The SQL and the error text are not listed: read them with 'lakana query status ID'. A long list prints a "More results" line with the command that shows the next page.

lakana query submit

Usage: lakana query submit [OPTIONS] {query}

Submit SQL without waiting; prints the query id.

lakana query status

Usage: lakana query status [OPTIONS] {query_id}

Show a query's status; for a queued query, waiting_for says why it has not started.

lakana query wait

Usage: lakana query wait [OPTIONS] {query_id}

Wait for a query to finish.

After 3 seconds in the queue, a line on standard error says why the query waits ('Waiting: ...'), and again when the reason changes. Ctrl-C cancels the query, also one of someone else's if you hold MANAGE_ENVIRONMENT.

lakana query cancel

Usage: lakana query cancel [OPTIONS] {query_id}

Cancel a queued query, or ask the worker to stop a running one.

lakana query result

Usage: lakana query result [OPTIONS] {query_id}

Fetch a finished query's result.

lakana principal

Usage: lakana principal [OPTIONS] COMMAND [ARGS]...

Manage principals.

lakana principal create

Usage: lakana principal create [OPTIONS] {name}

Create principal NAME.

lakana principal list

Usage: lakana principal list [OPTIONS]

List principals, newest first.

--search TEXT lists those whose name contains TEXT, in any letter case. A long list prints a "More results" line with the command that shows the next page.

lakana principal get

Usage: lakana principal get [OPTIONS] {name}

Show principal NAME.

lakana principal deactivate

Usage: lakana principal deactivate [OPTIONS] {name}

Deactivate principal NAME; its credentials stop working.

lakana group

Usage: lakana group [OPTIONS] COMMAND [ARGS]...

Manage groups.

lakana group create

Usage: lakana group create [OPTIONS] {name}

Create group NAME.

lakana group list

Usage: lakana group list [OPTIONS]

List groups, newest first.

--search TEXT lists those whose name contains TEXT, in any letter case. A long list prints a "More results" line with the command that shows the next page.

lakana group delete

Usage: lakana group delete [OPTIONS] {name}

Delete group NAME.

lakana group members

Usage: lakana group members [OPTIONS] {name}

List group NAME's members.

lakana group add-member

Usage: lakana group add-member [OPTIONS] {name} {principal}

Add PRINCIPAL to group NAME.

lakana group remove-member

Usage: lakana group remove-member [OPTIONS] {name} {principal}

Remove PRINCIPAL from group NAME.

lakana token

Usage: lakana token [OPTIONS] COMMAND [ARGS]...

Manage personal access tokens.

lakana token issue

Usage: lakana token issue [OPTIONS] {principal} {name}

Issue a personal access token named NAME for PRINCIPAL; the token is shown once.

lakana token list

Usage: lakana token list [OPTIONS] {principal}

List PRINCIPAL's tokens (never their secrets).

lakana token revoke

Usage: lakana token revoke [OPTIONS] {principal} {key_id}

Revoke PRINCIPAL's token KEY_ID.

lakana task-token

Usage: lakana task-token [OPTIONS] COMMAND [ARGS]...

Mint and revoke task tokens.

lakana task-token mint

Usage: lakana task-token mint [OPTIONS] {principal}

Mint a task token for PRINCIPAL; the token is shown once.

PRINCIPAL is your own name, unless you are an instance admin.

lakana task-token revoke

Usage: lakana task-token revoke [OPTIONS] {key_id}

Revoke task token KEY_ID.

lakana grant

Usage: lakana grant [OPTIONS] COMMAND [ARGS]...

Manage grants.

lakana grant create

Usage: lakana grant create [OPTIONS]

Grant a role or action on a securable to a principal or group.

lakana grant list

Usage: lakana grant list [OPTIONS]

List active grants, newest first; without a filter, every grant.

Filters combine. --principal and --group list the grants made to them (not those a principal has through its groups: lakana access explain says that). --action lists the grants that give an action, by itself or through a role. --env, with --ns and --table, lists the grants on exactly that object. A long list prints a "More results" line with the command that shows the next page.

The table names who each grant is for in subject (user alice, service ci-bot, group analysts); the API sends ids only, so the names cost one request per principal and group until all are found, and a caller that may not list them sees the ids. A column with nothing in it is left out. When the terminal is too narrow the table leaves out kind, namespace, table, environment and action, in that order, and folds a long subject onto more lines, so that the id that grant revoke needs always stays whole, and a line under the table shows the revoke command; --output json always has every field.

A grant of MANAGE_GRANT, MANAGE_GROUP or MANAGE_TOKEN as a single action has the marker no effect in a note column, which never drops (a narrow terminal leaves out table after the others, so that the id and the marker stay whole), and a line under the table explains it, followed by the command that revokes each such grant with its whole id (the first five; --output json has all of them and carries the full sentence). Revoke such a grant, and give the ADMIN role to whoever should keep the ability.

lakana grant revoke

Usage: lakana grant revoke [OPTIONS] {grant_id}

Revoke grant GRANT_ID.

Take the id from grant list, which keeps it in the table even on a narrow terminal.

lakana access

Usage: lakana access [OPTIONS] COMMAND [ARGS]...

Explain who may do what.

lakana access explain

Usage: lakana access explain [OPTIONS] {principal}

Show what PRINCIPAL may do on an object, and through which grant, role or group.

One row per action that applies to the object: allowed is the answer of the same check the server makes when the principal tries, and via names the grants that give it (a role, a single action, or either through a group). You may explain yourself; explaining someone else needs the ADMIN role on the instance.

lakana row-filter

Usage: lakana row-filter [OPTIONS] COMMAND [ARGS]...

Manage row filters.

lakana row-filter create

Usage: lakana row-filter create [OPTIONS] {table}

Restrict which rows of TABLE (namespace.table) a principal or group sees.

lakana row-filter list

Usage: lakana row-filter list [OPTIONS]

List active row filters, optionally filtered.

The table names who each row filter is for in subject. When the terminal is too narrow it leaves out environment, id and namespace, in that order; --output json always has every field.

lakana row-filter revoke

Usage: lakana row-filter revoke [OPTIONS] {row_filter_id}

Revoke row filter ROW_FILTER_ID.

Take the id from row-filter list; on a narrow terminal the table leaves it out, so widen the terminal or use --output json.

lakana column-mask

Usage: lakana column-mask [OPTIONS] COMMAND [ARGS]...

Manage column masks.

lakana column-mask create

Usage: lakana column-mask create [OPTIONS] {table} {column}

Mask COLUMN of TABLE (namespace.table) for a principal or group.

lakana column-mask list

Usage: lakana column-mask list [OPTIONS]

List active column masks, optionally filtered.

The table names who each column mask is for in subject. When the terminal is too narrow it leaves out environment, id and namespace, in that order; --output json always has every field.

lakana column-mask revoke

Usage: lakana column-mask revoke [OPTIONS] {column_mask_id}

Revoke column mask COLUMN_MASK_ID.

Take the id from column-mask list; on a narrow terminal the table leaves it out, so widen the terminal or use --output json.

lakana secret

Usage: lakana secret [OPTIONS] COMMAND [ARGS]...

Manage write-only secrets.

lakana secret create

Usage: lakana secret create [OPTIONS] {name}

Create secret NAME; its values can never be read back.

Name its use with --index, --repository or --own-env to bind it as it is created; that needs the right to manage secrets, as secret bind does.

lakana secret list

Usage: lakana secret list [OPTIONS]

List secrets (names and metadata only).

format_version 1 marks a secret stored in the original format; set it again with lakana secret rotate to store it in the current one.

use says what a secret may be used for: a package index or a repository (use_target names it), or its environment's storage and webhooks. A secret with no use serves its environment only; lakana secret bind gives it one.

lakana secret rotate

Usage: lakana secret rotate [OPTIONS] {name}

Replace secret NAME's values.

lakana secret bind

Usage: lakana secret bind [OPTIONS] {name}

Bind secret NAME to the one use it may serve; needs the right to manage secrets.

lakana secret revoke

Usage: lakana secret revoke [OPTIONS] {name}

Revoke secret NAME.

lakana repo

Usage: lakana repo [OPTIONS] COMMAND [ARGS]...

Register and manage git repositories.

lakana repo add

Usage: lakana repo add [OPTIONS] {name}

Register repository NAME.

lakana repo list

Usage: lakana repo list [OPTIONS]

List repositories.

lakana repo show

Usage: lakana repo show [OPTIONS] {name}

Show repository NAME.

lakana repo update

Usage: lakana repo update [OPTIONS] {name}

Change repository NAME; only the options given change.

lakana repo remove

Usage: lakana repo remove [OPTIONS] {name}

Delete repository NAME and its deploy history.

lakana index

Usage: lakana index [OPTIONS] COMMAND [ARGS]...

Register and manage package indexes for Python assets.

lakana index add

Usage: lakana index add [OPTIONS] {name}

Register package index NAME.

lakana index list

Usage: lakana index list [OPTIONS]

List package indexes.

lakana index show

Usage: lakana index show [OPTIONS] {name}

Show package index NAME.

lakana index update

Usage: lakana index update [OPTIONS] {name}

Change package index NAME; only the options given change.

lakana index remove

Usage: lakana index remove [OPTIONS] {name}

Delete package index NAME.

lakana deploys

Usage: lakana deploys [OPTIONS] COMMAND [ARGS]...

Deploy history.

lakana deploys list

Usage: lakana deploys list [OPTIONS]

List deploy history, newest first.

A long list prints a "More results" line with the command that shows the next page.

lakana deploys show

Usage: lakana deploys show [OPTIONS] {deploy_id}

Show deploy DEPLOY_ID.

lakana deployments

Usage: lakana deployments [OPTIONS] COMMAND [ARGS]...

Current deployments and the identity each one runs as.

lakana deployments list

Usage: lakana deployments list [OPTIONS]

Show the current project version per repository.

lakana deployments set-run-as

Usage: lakana deployments set-run-as [OPTIONS] [repo]

Bind (or with --clear remove) the service account REPO's runs write as.

lakana project-version

Usage: lakana project-version [OPTIONS] COMMAND [ARGS]...

Immutable project versions.

lakana project-version show

Usage: lakana project-version show [OPTIONS] {project_version_id}

Show project version ID: its assets, the package indexes its lock uses and the DAG as a tree.

lakana runs

Usage: lakana runs [OPTIONS] COMMAND [ARGS]...

Runs of deployed project versions.

lakana runs list

Usage: lakana runs list [OPTIONS]

List runs, newest first.

A long list prints a "More results" line with the command that shows the next page.

lakana runs show

Usage: lakana runs show [OPTIONS] {run_id}

Show run RUN_ID, its tasks and its notifications; waiting_for says why a task waits.

lakana runs logs

Usage: lakana runs logs [OPTIONS] {run_id}

Print a run's task logs.

lakana runs cancel

Usage: lakana runs cancel [OPTIONS] {run_id}

Request cancellation of RUN_ID.

lakana runs resume

Usage: lakana runs resume [OPTIONS] {run_id}

Resume RUN_ID's failed, upstream_failed and cancelled tasks in a new run.

lakana scheduler

Usage: lakana scheduler [OPTIONS] COMMAND [ARGS]...

The scheduler role.

lakana scheduler status

Usage: lakana scheduler status [OPTIONS]

Show the scheduler holder and queue lengths (admin).

lakana worker

Usage: lakana worker [OPTIONS] COMMAND [ARGS]...

The workers of this instance.

lakana worker list

Usage: lakana worker list [OPTIONS]

List the registered workers and what each is doing (admin).

A worker is alive while its heartbeat is recent and lost after that; a lost worker's row stays for an hour, and the query or deploy it held is ended by the scheduler. The column takes is the work the worker takes: all, interactive (ad-hoc queries) or pipeline (deploys, pipeline tasks and table maintenance).

lakana job

Usage: lakana job [OPTIONS] COMMAND [ARGS]...

Jobs: named selections of a project, started by hand, by the API or by a schedule.

lakana job list

Usage: lakana job list [OPTIONS]

List the jobs of the deployed projects: state, triggers, next run and last run.

lakana job show

Usage: lakana job show [OPTIONS] {name}

Show job NAME: its state, its triggers with the next run, and its last run.

lakana job run

Usage: lakana job run [OPTIONS] {name}

Start job NAME now; if its project is busy, the start waits and begins when it is free.

Exit 0 when the run succeeds, 1 when it fails, is cancelled or the timeout elapses. When the start waits, the command prints what it waits for and exits 0: that means accepted, not run; a script that needs the result follows the run it names with lakana runs show.

lakana job pause

Usage: lakana job pause [OPTIONS] {name}

Pause job NAME's schedule; a waiting start still happens and a start by hand still works.

lakana job resume

Usage: lakana job resume [OPTIONS] {name}

Resume job NAME's schedule from now; the times missed while paused do not fire.

lakana job notify-test

Usage: lakana job notify-test [OPTIONS] {name}

Send a test message to job NAME's webhook and wait up to 30 seconds for the result.

The message carries clearly fake run data. Exit 0 when it was delivered; 1 when the webhook refused it or was not reached (the reason is printed), and also 1 when it is still being tried after the wait (look later with lakana job notification). Needs the deploy right.

lakana job notification

Usage: lakana job notification [OPTIONS] {delivery_id}

Show one notification delivery, such as the one job notify-test queued.

lakana check

Usage: lakana check [OPTIONS] COMMAND [ARGS]...

Data quality checks: the results of the checks a run made.

lakana check list

Usage: lakana check list [OPTIONS]

List check results, newest first. With --run, the results of one run.

The run column shows 8 characters of the run id; --output json has the whole id.

lakana check rows

Usage: lakana check rows [OPTIONS] {RUN} {CHECK}

Show the rows a check found in a run.

These are the rows this check found in that run, as they were then, also after the table was set back.

It runs as a query of yours, under your own rights, row filters and column masks: you see the rows the check found that you may read, and every other table the check's SQL reads is read as it is now. It needs the right to read the run and to submit queries, and the right to read every table the check's SQL reads: without it, the command refuses and shows neither rows nor SQL.

The rows stay readable for seven days after a failed blocking check; then lakana table maintain removes the snapshot, and the number of rows remains in lakana check list. A check that read no snapshot (a view, or SQL that does not read its table) has no rows to show; the message says where to run its SQL yourself.

Prints a table like lakana sql; --out writes every row to a file instead.

lakana audit

Usage: lakana audit [OPTIONS] COMMAND [ARGS]...

Read the audit log.

lakana audit list

Usage: lakana audit list [OPTIONS]

List audit events, newest first.

--output json shows each event's detail. Query text and column values in someone else's events are hidden unless you hold READ_QUERY_TEXT: a query's SQL text shows as its SHA-256 and length, an incremental write's watermark as {"hidden": true}. Your own events show in full. A page that shows other people's events, or --principal-id of someone else, is itself recorded in the audit log (ACTIVITY_VIEWED).

lakana server

Usage: lakana server [OPTIONS] COMMAND [ARGS]...

Run Lakana on this machine (local only).

lakana server start

Usage: lakana server start [OPTIONS]

Start Lakana on this machine: migrate, then the server and worker.

Stop it with Ctrl-C, or from another terminal with 'lakana server stop'.

lakana server status

Usage: lakana server status [OPTIONS]

Report whether a local-mode instance is running.

Running means the data directory's lock is held (even before any pidfile exists, or while /healthz is slow), or the recorded server PID is alive and answers /healthz.

lakana server stop

Usage: lakana server stop [OPTIONS]

Stop this machine's local instance, as Ctrl-C in its terminal does.

Asks the instance in the data directory to shut down and waits up to 30 seconds for it to leave. Nothing running is not an error. It stops only the instance on this machine: a Compose stack stops with 'docker compose down' in its bundle directory.

lakana server reset

Usage: lakana server reset [OPTIONS]

Delete this machine's local instance's data, so the next start is a first run.

Deletes the database, the tables' files and the secrets key of the instance in the data directory, and nothing else: your project files are never touched. It refuses while the instance runs, and in any directory that is not a local instance's data directory.

lakana bundle

Usage: lakana bundle [OPTIONS] COMMAND [ARGS]...

Generate a team deployment bundle.

lakana bundle init

Usage: lakana bundle init [OPTIONS] [directory]

Generate compose.yaml and .env for a team deployment, and save a profile.