Race between active job failure and its recovery task leaving link disconnected

When an active job fails while its recovery task is still running, the failing job will assume recovery is still on-going. At the same time, the recovery task will assume the job did not fail. This will result in keeping the link in a disconnected state and new data not being ingested.

The workaround is to manually disconnect and reconnect the link or call the analytics Cluster Restart API to retrigger the link recovery task.