TableCore
Sign in

Your first connection

The shortest path from a fresh install to a server you can query. Everything here has a fuller treatment in Connections — this page is the part you need before your first query works.

Add the connection

  1. In the Explorer on the left, choose New connection.
  2. Give it a Connection name. This is what you will see in the explorer, so name it after the environment rather than the host — Orders (staging) beats mongodb-3.
  3. Put the server's URI in Connection string. A new connection is pre-filled with mongodb://localhost:27017, which is the right answer more often than it looks.
  4. Choose Test Connection. TableCore reaches the server and tells you whether it answered before you commit to anything.
  5. Save.

The connection string accepts the standard MongoDB forms, including the mongodb+srv:// seed list that Atlas gives you:

mongodb://localhost:27017
mongodb://reader@db.internal.example:27017/?authSource=admin
mongodb+srv://reader@cluster0.abcde.mongodb.net/?retryWrites=true

You can leave the password out of the URI and put it in the separate MongoDB password field instead. That is the better habit: the password then never appears in a field you might screenshot, and it is stored the same way either way.

Where the connection string goes

Not into a settings file. settings.json keeps the profile's name, colour and write protection; the connection string and every password go to your operating system's credential store. Credentials and secrets documents this per platform, including what you will see in your own keychain.

Mark production read-only while you are here

The connection dialog has a Writing to this connection setting. Set anything you would be upset to damage to Read only now, before you have opened anything on it.

Read-only means what it says inside TableCore: no document edit, no drop, no rename, no restore, no reshape, and a { update: … } command typed into the editor is refused too rather than quietly running.

It guards against the wrong window, not against an attacker

This is a client-side guard, not a permission. TableCore runs on your computer, and anyone who can edit its settings can turn it off. For a connection that cannot write no matter what, ask your administrator for a read-only MongoDB user. Read-only connections explains exactly what each level stops.

If you work with several environments, give each one a Colour tag. The explorer row, the query tab's location chip, the edge of the result card and the selection inside the grid all carry it, which is the cheapest protection against running the right query on the wrong server. A connection with no tag keeps the ordinary neutral selection, so the colour always means something.

Open a collection

Expand the connection in the explorer. TableCore loads its databases, then the collections under whichever database you open. Selecting a collection opens a query tab against it.

The row under each connection tells you where it stands — not tested, connecting, connected or unavailable — so a server that has gone away looks different from one you have not opened yet.

When it does not connect

What you seeUsually means
Connection refusedNothing is listening on that host and port, or a firewall drops the route before it arrives
Authentication failedThe user, password or authSource does not match. authSource is the database the account was created in, which is often admin and rarely the one you are querying
Timed out on an Atlas clusterYour current IP address is not in the cluster's access list
TLS or certificate errorsThe server requires TLS and the URI does not ask for it, or its certificate is not trusted by your machine
A replica set that resolves but never connectsThe set advertises hostnames your machine cannot resolve. This is normal for a set configured with internal names and reached from outside

Every failure is reported with its technical detail attached, and you can copy that detail out of the dialog. Secrets are redacted before it is shown, so a copied diagnostic is safe to paste into a bug report.

Next

You have a server. Now run a query against it.