You are operating a Cloud Dataflow streaming pipeline. The pipeline aggregates events from a Cloud Pub/ Sub subscription source, within a window, and sinks the resulting aggregation to a Cloud…

Question

Options

• AAn alert based on a decrease of subscription/num_undelivered_messages for the source and a rate of change increase of instance/storage/used_bytes for the
• BAn alert based on an increase of subscription/num_undelivered_messages for the source and a rate of change decrease of instance/storage/used_bytes for
• CAn alert based on a decrease of instance/storage/used_bytes for the source and a rate of change increase of subscription/num_undelivered_messages for the
• DAn alert based on an increase of instance/storage/used_bytes for the source and a rate of change decrease of subscription/num_undelivered_messages for

• A
  4% (2)
• B
  73% (40)
• C
  7% (4)
• D
  16% (9)

Explanation

Option B is correct because it identifies the two symptoms of a failing pipeline: when the pipeline stops consuming Pub/Sub messages, the undelivered message backlog (subscription/num_undelivered_messages) will increase - and since nothing is being written to Cloud Storage, the rate of change of instance/storage/used_bytes will decrease toward zero. Together, these two conditions confirm the pipeline has stalled.

Option A is wrong because a decrease in undelivered messages means the pipeline is successfully draining the queue (healthy behavior), and an increase in storage growth also signals healthy output - these would alert you when things are working fine, not when they're broken.

Option C is wrong because it reverses the metric labels: "decrease of storage bytes" as the source metric and "increase of undelivered messages for the source" swaps the context incorrectly - Cloud Storage is the sink, not the source, and the framing conflates the two roles.

Option D is wrong for the same inverse-logic reason as A - an increase in storage bytes means data is being written, and a decrease in undelivered messages means Pub/Sub is being consumed; both indicate a healthy pipeline, not a broken one.

Memory tip: Think "pile up and dry up" - if the pipeline breaks, messages pile up in Pub/Sub (backlog rises) and output dries up in Storage (write rate falls). Alert on both together.

No community discussion yet for this question.