sqb load

Runs source loader functions and writes data into their target tables using the configured write strategy.

Usage

sqb --project-dir <path> load [flags]

Flags

Flag	Description
--select, -s	Select specific sources or loaders by name
--exclude	Exclude specific sources or loaders
--reload	Pass is_reload=True to loader functions
--concurrency	Number of worker connections (default: from settings or 1)
--start-cursor	Override the start cursor; an integer value sets start_cursor_int, anything else is read as an ISO timestamp and sets start_cursor_ts
--end-cursor	Override the end cursor, typed the same way as --start-cursor
--start-cursor-ts	Override start cursor for timestamp-based loaders (ISO format)
--end-cursor-ts	Override end cursor for timestamp-based loaders (ISO format)
--start-cursor-int	Override start cursor for integer-based loaders
--end-cursor-int	Override end cursor for integer-based loaders
--json	Output results as JSON
--json-output	Write JSON results to a file path
--vars	Override project variables with a JSON object (--vars '{"key": "value"}')
--warehouse <name>	Snowflake warehouse for this invocation; overrides the target’s build warehouse group and the connection warehouse

Examples

# Load all managed sources
sqb load

# Load a specific source
sqb load --select raw_customers

# Load with concurrency
sqb load --concurrency 4

# Reload all sources (full refresh behavior)
sqb load --reload

# Override cursor bounds
sqb load --start-cursor "2026-05-01T00:00:00"

Execution order

Loaders are executed in DAG topological order based on depends_on declarations. Independent loaders run concurrently when --concurrency is greater than 1.

Intermediate loaders (those referenced only via depends_on without a direct source binding) run first, followed by the source-bound loaders that depend on them.

Output

Load ready  2 selected

Sources (2)
├── raw__customers
└── raw__orders

Execution  sqb load  (concurrency: 1)

Connecting to duckdb...
✓ Warehouse connected  duckdb  (0.05s)

  1/2  source    raw__customers                 OK 0.04s  rows=5
  2/2  source    raw__orders                    OK 0.03s  rows=10

✓ Completed successfully  PASS=2  WARN=0  FAIL=0  SKIP=0  TOTAL=2  (0.15s)

Auto-load during builds

sqb build automatically loads managed sources before building dependent models. This is controlled by the auto_load_sources setting (default: true) and the --load / --no-load / --reload flags:

# Default: auto-load is on
sqb build

# Explicitly skip loading
sqb build --no-load

# Force reload
sqb build --reload

With loading off, through --no-load or auto_load_sources = false, no loader runs, even for a source selected by name such as sqb build --no-load --select raw_orders. Managed sources in the selection stay in the plan, and their audits check the data they already hold, at the end of the build if no model reads them. A source outside the selection is neither loaded nor audited.

See Loaders for full documentation on write strategies, the loader context API, and source deferral.