jordi.serra-torrens@mongodb.com pointed out that there may be a very improbable but still relevant bug after the changes committed under SERVER-69555.

The migration coordinator is asking the recipient to unset the pending field (in case of commit) or delete the range deletion document (in case of abort).

Since targeting is not done anymore by migrationId, it may happen the following:

• An update/delete on the recipient succeeds, but the request gets stuck in the network and is also retried after 1 hour.
• After 1 hour, a new migration involving the same range may be happening, and the update/deletion would be wrongly applied.