To prevent further occurrences we:
- have limited the number of retries
- have reduced queue depth for long running tasks
- will expand our liveliness checks to kill instances that are blocked by too long running requests or deadlocks.

After purging queues and terminating long running tasks our skills service is up and running again.

We have identified the root cause. Due to long running requests not being terminated automatically our infrastructure was not scaling properly leading to capacity issues.

Our entire skills service is overloaded at the moment, we are investigating the cause. Please bear with us while you receive timeouts.

Our skills service is currently unavailable.