Connections

A connection is a saved way to reach one source: a PostgreSQL database, a Redis server, a folder of your own files or a server over SSH. Catalogs, tables, query consoles, pipelines and dashboards all work through connections.

Each kind of source has a page of its own: Files, PostgreSQL, Redis and SSH.

The list of connections

The Connections column is on the left of the Data space. The Pipelines space shows the same list.

• Press a connection’s name to open its catalog. It opens in the column beside the list, as a tab named after the connection.
• Press a connection’s icon to open its window, where it is edited.
• + at the head of the column adds a connection.
• The magnifier turns the head into a filter by name. Esc or the cross closes it.
• The arrow at the end of the head reads the list again.
• The button at the top left of the screen narrows the column to icons and widens it again. The narrow column still opens catalogs and still has +.

A dot on each icon says what the last check found:

• green: the source answered;
• red: the source did not answer, or refused;
• yellow: the connection is switched off;
• grey: it has not been checked yet.

Connections are checked while the workspace is open. How often is set on the Preferences page.

Adding a connection

1. Press + at the head of Connections. The New connection window opens.
2. Pick a kind of source. The kinds are grouped on the left: SQL Databases, Key-Value Stores, Files & Storage. The search box finds a kind by its name or by what it can do.
3. The panel on the right says what the kind is, what you can see and change through it, and has Guide, which opens its page of this guide.
4. Press Create connection, or double-click the kind. The connection window opens for it.
5. Fill in the fields, press Test Connection, then OK to save and close, or Apply to save and keep the window open. Back returns to the list of kinds without saving.

Where this server can run containers, the panel also offers to start a new server of that kind on this machine: Start PostgreSQL here, Start TimescaleDB here, Start Redis here or Start SSH server here. The window that opens asks How:

• automatically: the server is named for you, gets a password nobody has to choose and comes filled with sample data, so there is something to look at;
• with your own choices: a name, a user, a database, a port on this machine, and for an SSH server whether you sign in with a password or one of your SSH keys.

Either way the connection is made for you. How the start is going is shown in the activity bell at the bottom right. When the server is ready, a window shows where to reach it from your own tools and its password, once. Copy the password then if you need it elsewhere.

A button that is greyed out says why when you point at it: this installation may not run containers at all.

The connection window

The window of a connection opens when you add one or press its icon. Several can be open at once, so you can read one connection while filling in another. Closing a window with changes that were not saved asks first.

• Name is how the connection is listed. It is required.
• Tab color colours the connection’s tabs and its row in the list.
• In use switches the connection off without deleting it. A switched-off connection keeps its settings, cannot be opened and has a yellow dot.
• URL is the connection string composed from the fields. The password is never part of it. Paste a connection string into it and the fields are filled from it. The copy button next to it copies the string.
• Test Connection tries to connect with what the form holds now, saved or not. The answer appears in a small popup above the button.
• Delete removes the connection. Only its owner has it, and it asks first.

The window has three pages at least.

Connection

How to reach the source. The fields depend on the kind of source and are described on its page. A choice of Authentication changes which fields are shown.

Options

Keep answers says how long a catalog level and a page of rows may be reused before the source is read again.

• As this source decides is the default. The label says how long that is.
• Never keep answers reads the source every time.
• Keep for … sets a time of your own, up to a day.

Pressing a reload button always reads the source, whatever is chosen here. A small mark beside the reload button says Live while the answer on screen is fresh from the source, and shows a clock once it is one that was kept. Hover it to see how old it is.

A source that keeps nothing says so, and the field is switched off.

Permissions

• Consoles decides what a query console may do on this connection: Read only, Ask before writing (the default) or Writing allowed. It applies to everybody the connection is shared with.
• Allow sending data to third parties decides whether data read through this connection may be sent to an outside service, such as an AI model provider. It is off unless you switch it on.
• Who else may reach it lists the people the connection is shared with and the requests waiting for an answer. Share gives it to somebody.

Sharing a connection

A connection belongs to the person who made it. The owner can give other people one of three levels:

• Read: open the catalog, read rows, download and chart them;
• Edit rows: also change, add and delete rows;
• Change structure: also change columns, indexes, keys and tables.

A level covers the whole connection, not one table. Only the owner can delete the connection, change its settings, back it up or share it further.

A shared connection appears in the other person’s list. Its window shows who shared it and at what level, and its fields cannot be changed.

A table can also be shared as a link: Share this view in the row of commands above a table makes a link to that table with its filters and columns. A link gives no access by itself. Somebody who opens it without access to the connection can ask for it, and the request appears on the owner’s Permissions page and in their activity bell. The owner chooses the level when approving it.

Opening a catalog

Pressing a connection opens its catalog: what the source holds, level by level or as a tree, depending on the kind of source. A catalog has a row of its own commands at the top, and a menu of actions on what is ticked. The commands depend on the kind of source and are described on its page.

Things that appear in every catalog:

• ticking a node and opening the actions menu acts on what is ticked, or on the level you are on when nothing is;
• Details says what a node is: its size, its dates, what it can be opened as;
• Copy and Cut pick nodes up, and Paste puts them where you are standing, in the same connection or in another one;
• opening a node opens it in a tab of its own in the work column, as rows, as a document, as an image and so on;
• the catalog can be duplicated beside itself, or paired with the catalog of another connection in one window, to move things between the two.