When working with Azure Blob Storage and other cloud services, it’s common to run into timing issues, a file may be created by a service, but it’s not instantly available for retrieval. In my case, Azure Speech-to-Text generates a JSON transcription file in Blob Storage, but the file isn’t ready immediately after the transcription request completes. If you call Get blob content (V2) too early, the action fails and your entire flow stops. In this guide, we’ll explore how to use Power Automate Do Until to handle timing issues when working with Azure Blob Storage and other cloud services
To solve this, I implemented a retry loop using Power Automate’s Do Until control and a Boolean variable. This ensures the flow patiently waits until the file exists, but without introducing long, static delays.
Flow logic
How It Works
Initialize a Boolean Variable
At the start, I create a variablevarJSONReadyset tofalse. This will be the flag that controls our loop.Do Until Control
The loop runs untilvarJSONReadyequalstrue. Inside the loop, we try to get the blob content.Attempt to Get the Blob
If the file exists, Get blob content (V2) succeeds.
When successful, we immediately set
varJSONReadytotrueso the loop ends.If it fails (file not yet created), the flow moves to the delay step.
Delay
I use a very short delay (1 second in this example) before retrying. This keeps the process fast and responsive without hammering the service too aggressively.
To make this work correctly, the Delay action is configured to run only ifGet blob content (V2):
- has failed
- is skipped
- has timed out
This is done by opening the … menu on the Delay action, selecting Configure run after, unchecking is successful, and enabling only those three states. This ensures the delay only runs when the blob isn’t ready, skipping any wait time if the file is already available.
Why This Is Better Than Static Delays
Instead of adding a fixed 30-second pause, this loop checks as often as needed and proceeds immediately once the file is ready. This makes the flow faster in most cases while still being reliable.
Reusability Beyond Blob Storage
This retry‑until‑success pattern comes in clutch whenever an action bombs out because the target resource isn’t provisioned or committed yet. In Power Automate terms, think of those times when a connector call returns a 404, 409, or “not found” simply because the backend hasn’t finished writing the file, record, or response. Instead of hard‑coding a Delay() and hoping for the best, the loop keeps polling the endpoint or storage location until it’s actually ready, then continues down the flow without throwing a fail and killing the run.
Examples:
- Waiting for a SharePoint document to finish generating after a file conversion process before moving or processing it.
- Polling an API endpoint until it returns a “completed” status instead of “in progress” when running long operations like report generation or data export.
- Checking if a record exists in Dataverse before attempting to update it, avoiding “record not found” errors.
- Verifying that an email attachment has been saved in a monitored folder before starting a parsing step.
- Waiting for a large file to finish uploading to Azure Blob Storage before triggering a downstream process.
- Ensuring a new user account has propagated in Azure AD before assigning licenses or groups.
