How-to guides

A how-to guide solves one task by the shortest path, for someone who knows what they want.

Pages:

• Projects and deploys: start a project, deploy it, run it, set up CI and write a first Python asset. The layout, the keys and the rules are in the reference.
• Check your data: state in SQL what must be true of a table, and what happens to the table when it is not.
• Run a job on a schedule: declare a job and its cron schedule, bind a run principal, pause it, start it from CI, and what happens when the project is busy or the platform was down.
• Get told when a job fails: send a message to your chat tool or a webhook when a run fails and when the job works again, test it, and verify its signature.
• Deploying with Compose (team mode): run a team instance with Docker Compose.
• Run Lakana for other people: install, hand over, back up, upgrade, watch and separate instances, in the order an operator needs them.
• Is my instance healthy?: what to look at (services, workers, queue, database, storage, disk), which log lines matter, and what to collect before you ask a question.
• Give a second person access: create a user, grant the smallest set to query one environment, issue a token, and take access away.
• Why may (or may not) someone do this?: see what a principal may do on an object, and through which grant, role or group.
• Upgrade to the release with artifacts: the migration, the order and the way back.
• Manage secrets: see how each stored secret is encrypted and set an older one again.
• Use a private package index: install your team's own package in Python assets, with a password that stays out of the project.
• Train a scikit-learn model and score with it: store a model as an artifact and read it in another asset.
• Inspect and download artifact versions: see what an artifact holds and copy a version to your machine.
• Limit how many versions are kept: keep only the newest versions of an artifact.
• Delete an artifact version: remove one version or a whole artifact; Lakana cannot undo it.
• Show some people only some rows: hide the rows that fail a condition, and what Lakana refuses when you create a filter.
• Mask a column for some people: show a replacement value instead of a column, and what masks do not protect.
• Describe a table: write a description for a table and its columns, in the project or by hand, and go back to the project's text.
• Drop and restore a table: drop a table, restore it with its data and access rules during the undo period, or delete its files now; delete and restore a whole environment the same way.
• Clean up files left by older releases: list the table directories in storage that an older release left behind without a record, and delete them.
• Let tables expire: let a development table, or every table of an environment, be dropped once nobody has used it for a while; see what expires soon, and keep a table.
• Find a table: search tables, columns, descriptions, runs and people from one text, and what you are allowed to see.
• Find a run, a query or a deploy: filter the lists by status, trigger, job, time, submitter or repository, and page through a long one.
• Stop a query: cancel one with Ctrl-C or a command, who may, and what else ends a query.
• Why is my query or task waiting?: read the reason a query or a task has not started and what to do about it.
• List and withdraw an open upload: free the place that an upload left by a killed process holds.

Pages are added here as they are written.