What causes Activity stuck issue?

Upstream data source or sink is unresponsive, causing the activity to wait indefinitely for a response. Resource contention or capacity limits on the compute cluster processing the activity. A pagination loop in a REST or query-based activity where the continuationToken is not being passed correctly, causing infinite retries. Network interruption between Fabric Data Factory and the connected data source or gateway that does not trigger a timeout

How do I fix Activity stuck issue?

Step 1: Compare the current run duration against historical baselines for the same activity — if it is significantly longer with no throughput metrics incrementing, treat it as stuck.. Step 2: Cancel the stuck activity run from the Fabric Data Factory monitoring UI or via the REST API (DELETE /runs/{runId}/cancel).. Step 3: If using a query API with pagination, verify your code passes the full continuationToken string on every subsequent request and does not stop on empty result pages while the token is non-null.. Step 4: Retry the pipeline run manually and monitor the first few minutes for any throughput activity; if it stalls again at the same point, investigate the source or sink connectivity.. Step 5: Check gateway logs (if an on-premises data gateway is involved) and Fabric capacity utilization to rule out resource exhaustion as the root cause before escalating.

High severityrefreshMicrosoft Fabric →

Power BI Refresh Error:
Activity stuck issue

What does this error mean?

A pipeline activity appears to be running indefinitely with little or no visible progress, indicating it may be deadlocked or hung. This differs from a normal long-running activity by the absence of meaningful throughput or state changes.

Common causes

1Upstream data source or sink is unresponsive, causing the activity to wait indefinitely for a response
2Resource contention or capacity limits on the compute cluster processing the activity
3A pagination loop in a REST or query-based activity where the continuationToken is not being passed correctly, causing infinite retries
4Network interruption between Fabric Data Factory and the connected data source or gateway that does not trigger a timeout

How to fix it

1Step 1: Compare the current run duration against historical baselines for the same activity — if it is significantly longer with no throughput metrics incrementing, treat it as stuck.
2Step 2: Cancel the stuck activity run from the Fabric Data Factory monitoring UI or via the REST API (DELETE /runs/{runId}/cancel).
3Step 3: If using a query API with pagination, verify your code passes the full continuationToken string on every subsequent request and does not stop on empty result pages while the token is non-null.
4Step 4: Retry the pipeline run manually and monitor the first few minutes for any throughput activity; if it stalls again at the same point, investigate the source or sink connectivity.
5Step 5: Check gateway logs (if an on-premises data gateway is involved) and Fabric capacity utilization to rule out resource exhaustion as the root cause before escalating.

Frequently asked questions

How do I know if an activity is truly stuck versus just slow?

Compare real-time throughput metrics (rows read/written, bytes transferred) in the activity monitoring detail pane. A slow activity shows incremental progress; a stuck activity shows zero or flat-lined metrics for an extended period despite a 'Running' status.

Will canceling and retrying a stuck activity cause data duplication?

It depends on whether your sink supports idempotent writes or upsert logic. For sinks without built-in deduplication, check whether any partial data was written before canceling and clean it up before retrying to avoid duplicates.