Step Type: Wait For Condition

The Wait For Condition step polls a configurable data source on a fixed interval until a user-supplied expression evaluates truthy, or a timeout is reached. It is the standard way to pause a procedure until something happens on an external system before moving on.

When to Use

• Wait for a target server to finish ingesting records migrated by an earlier step (e.g. a ServiceNow Import Set transform that runs asynchronously)
• Wait for an external job — a build, an export, a backup — to report success via a status endpoint
• Wait for a record count on a target table to reach an expected value before continuing
• Wait for any other condition you can express as a formula against data PB can fetch

For a one-shot API call where you do not need to retry, use Call API Endpoint instead.

Configuration

The step has three configuration sections.

Source (the Check)

Pick what PB fetches on each polling iteration. Available check types:

Check type	What it fetches
API Endpoint	Calls a named endpoint on a connection's adaptor. The endpoint's response is available in the until expression.
Record Query	Runs a filter query against a connection/table and exposes the returned records.
Record Count	Runs a count query against a connection/table and exposes the integer count.
Expression	Evaluates the until expression directly with no fetch. Useful for time-only waits or for expressions that consult only procedure variables.

Until (the Condition)

A formula expression that PB evaluates on every iteration after the fetch. When it evaluates truthy, polling stops and the step completes successfully. The variables available to the expression are:

• All procedure variables in scope
• The keys exposed by the selected check (e.g. response, records, count)

Examples:

• count >= 100 — wait until at least 100 records are present
• response["state"] == "complete" — wait until an API status field flips to "complete"
• len(records) > 0 and all(r["status"] == "ready" for r in records) — wait until all matching records are ready

Timing and Error Handling

Setting	Default	Description
poll_interval_seconds	5	How often to poll. Clamped to a 1-second minimum at execute time.
max_wait_seconds	300	Total wait budget before timeout handling kicks in. Capped at 3600 (one hour).
on_timeout	fail	What to do if the condition is not met before max_wait_seconds elapses. fail raises an error; continue logs a warning and proceeds with result=False.
on_fetch_error	retry	What to do if the check itself raises (e.g. a 5xx response). retry swallows the exception and waits for the next poll; fail re-raises immediately.

A broken until expression is always fatal — PB does not retry on expression evaluation errors, because a misconfigured expression would otherwise loop silently forever.

Output Variables

The Wait For Condition step can populate output variables with the following calculations:

Calculation	Type	Description
result	bool	True if the condition was met, False on timeout-with-continue.
attempts	int	Number of poll iterations performed.
elapsed	float	Total seconds elapsed in the polling loop.
context	dict	The last context dict returned by the check (response body, records, count, etc.).

Example

A typical use case is waiting for a ServiceNow Import Set transform to finish before reading the migrated records:

1. Source: API Endpoint check pointing at a status endpoint
2. Until: response["state"] == "loaded"
3. Poll interval: 10 seconds
4. Max wait: 600 seconds (10 minutes)
5. On timeout: fail
6. Output variables: capture attempts and elapsed for the execution report

Related Documentation

• Call API Endpoint — one-shot API calls without retry
• Iterate — for looping over a known collection rather than polling for a condition