Once a retrieve action is triggered on a cold storage blob, the cold storage addon pulls every hour for a retrieval status.
Once the blob is retrieved, it remains available/downloadable for a given amount of time (24h by default). The blob then returns to a cold storage state and is not downloadable anymore.
If the server couldn't pull for the retrieval status (because of a network outage, shutdown, etc.) within the default 24 hours then we end up in a corrupted state where the document is in a perpetual retrieval state.
We must fix this by checking the ongoing flag during the pull to s3 status.