SQL Datasource

A SQL datasource is a connection definition for a database outside the platform. Once defined, scripts, reports and saved queries reach it by name.

Fields

Field	Description
Name	The datasource’s name; scripts and queries call it by this (max 100 characters)
Description	Description (max 255 characters)
URL	The JDBC connection address (max 2,048 characters)
Username	The user name (max 50 characters)
Password	The password (max 50 characters)
Is Default	Whether this is the project’s default datasource

The password is not shown back

The list shows whether a password is set, not the password itself. It is stored encrypted and never handed back to be read. When you edit the record the password field comes up empty: leaving it empty keeps the existing password, and it changes only when you type a new one.

Lose the encryption key and the password goes with it

This password is protected by the installation’s attribute-encryption key. If that key is lost during a restore the password cannot be decrypted and has to be entered again by hand — it was never stored in plaintext, so there is nothing to recover. Detail: Backup.

The default datasource

The datasource marked Is Default is where a query that does not name one goes:

ins.runSql('SELECT ...');                    // the default datasource
ins.runSql('ProductionDB', 'SELECT ...');    // by name

Mark only one datasource in a project as the default. Marking a second does not unmark the first, and a call that names no datasource then cannot tell which one it means. The clone button is the easiest way into that state, because it copies every field including Is Default — after cloning, clear the flag on the copy.

Allowed hosts

Datasource addresses are subject to an allow-list, given as ins.datasource.allowed-hosts. It takes comma-separated hostnames (db.example.com), wildcard subdomains (*.example.com) and CIDR blocks (10.0.0.0/8).

Case	Result
List empty (the default)	No address is permitted
Address matches the list	Permitted
Address does not match	Rejected
Loopback, 0.0.0.0 or a link-local address	Rejected whatever the list says

This is deliberate: until the setting is made no datasource works at all, because the alternative would leave a script free to send queries to every address the server can reach.

When the host is taken out of a JDBC address, the user info, the path and SQL Server’s semicolon parameters are stripped. A PostgreSQL address that lists several hosts for high availability needs all of them on the list. Oracle TNS format (an address with no //) is not supported and cannot be saved.

The address is checked in two places: when you save (a rejected address comes back as a form error) and when a query runs.

Timeouts

Limit	Default	When it applies
Connect	10 seconds	A datasource that is down fails fast
Query	60 seconds	The time one query may take

They sit inside the script timeout

These limits are counted within the script execution time, and that is deliberate: a datasource that stops answering surfaces as a datasource error rather than as a timeout that kills the script — so the log says why. If you raise the query timeout, keep it below the script’s own limit, or that distinction is lost.

The connection pool is reused

One connection pool is opened per datasource and reused by later queries. Editing or deleting the record closes that pool and drops it, so a new address or password takes effect on the next query by itself. No restart is needed.

Related

• SQL Query — saved queries
• SQL Database — browsing the schema and running a query
• InfluxDB Datasource — the time-series equivalent
• Script Engine — using ins.runSql