Exam Bundle

A delete operation produces the expected removal for its key pertains to a different control (Configure AUTO CDC delete handling with the correct `APPLY AS DELETE WHEN` condition) and a different need (Delete events must remove the current SCD1 record according to a source operation field). It cannot establish that Use the full composite natural key in the AUTO CDC `KEYS` clause met the stated goal.

This signal belongs to the adjacent case “Auditors require the prior attribute values and validity intervals when a dimension changes,” whose remedy is Configure AUTO CDC to store SCD Type 2 history. The observed result for the present requirement is Updates affect only rows matching the complete business key.

A late lower-sequence update leaves the newer target state intact pertains to a different control (Use the event sequence column in AUTO CDC `SEQUENCE BY`) and a different need (Events may arrive out of order and must not overwrite a newer record with stale values). It cannot establish that Use the full composite natural key in the AUTO CDC `KEYS` clause met the stated goal.

The alternative Plan a new checkpoint and controlled state reinitialization or migration has a distinct purpose—A data engineer changes stateful grouping keys and the existing checkpoint is incompatible. It would not establish the required outcome (The stream restarts every day and must continue without rereading committed input); the necessary verification concerns After restart the offsets resume from checkpointed progress.

This action solves a different issue: “A stream must deliver at-least-once microbatches to an external system safely.” Its normal signal is Retry of one microbatch does not duplicate the externally committed records. Here the requested behavior is “The stream restarts every day and must continue without rereading committed input,” so that signal is not sufficient.

The alternative Give each logical streaming query its own checkpoint location has a distinct purpose—Two independent streaming queries were configured to share the same checkpoint path. It would not establish the required outcome (The stream restarts every day and must continue without rereading committed input); the necessary verification concerns After restart the offsets resume from checkpointed progress.

That observation is a valid indicator for “Two independent streaming queries were configured to share the same checkpoint path” when using Give each logical streaming query its own checkpoint location. The scenario instead tests “The stream restarts every day and must continue without rereading committed input”; the relevant result is After restart the offsets resume from checkpointed progress.

That observation is a valid indicator for “A stream must deliver at-least-once microbatches to an external system safely” when using Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key. The scenario instead tests “The stream restarts every day and must continue without rereading committed input”; the relevant result is After restart the offsets resume from checkpointed progress.

State store size stabilizes as sufficiently old event-time state expires pertains to a different control (Set a suitable event-time watermark for the use case) and a different need (A stateful aggregation grows without bound due to no late-event cutoff). It cannot establish that Configure a durable, exclusive checkpoint location for the query met the stated goal.

Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key belongs to the separate case “A stream must deliver at-least-once microbatches to an external system safely,” where Retry of one microbatch does not duplicate the externally committed records would be expected. That differs from the condition in this stem: “Two independent streaming queries were configured to share the same checkpoint path.”

The alternative Set a suitable event-time watermark for the use case has a distinct purpose—A stateful aggregation grows without bound due to no late-event cutoff. It would not establish the required outcome (Two independent streaming queries were configured to share the same checkpoint path); the necessary verification concerns Each query maintains independent offsets and state without checkpoint collisions.

The alternative Plan a new checkpoint and controlled state reinitialization or migration has a distinct purpose—A data engineer changes stateful grouping keys and the existing checkpoint is incompatible. It would not establish the required outcome (Two independent streaming queries were configured to share the same checkpoint path); the necessary verification concerns Each query maintains independent offsets and state without checkpoint collisions.

The restarted query initializes the intended state schema without checkpoint errors pertains to a different control (Plan a new checkpoint and controlled state reinitialization or migration) and a different need (A data engineer changes stateful grouping keys and the existing checkpoint is incompatible). It cannot establish that Give each logical streaming query its own checkpoint location met the stated goal.

This signal belongs to the adjacent case “A stream must deliver at-least-once microbatches to an external system safely,” whose remedy is Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key. The observed result for the present requirement is Each query maintains independent offsets and state without checkpoint collisions.

State store size stabilizes as sufficiently old event-time state expires would demonstrate the separate action Set a suitable event-time watermark for the use case, which addresses “A stateful aggregation grows without bound due to no late-event cutoff.” It does not verify the requirement “Two independent streaming queries were configured to share the same checkpoint path” following Give each logical streaming query its own checkpoint location.

The alternative Give each logical streaming query its own checkpoint location has a distinct purpose—Two independent streaming queries were configured to share the same checkpoint path. It would not establish the required outcome (A data engineer changes stateful grouping keys and the existing checkpoint is incompatible); the necessary verification concerns The restarted query initializes the intended state schema without checkpoint errors.

The alternative Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key has a distinct purpose—A stream must deliver at-least-once microbatches to an external system safely. It would not establish the required outcome (A data engineer changes stateful grouping keys and the existing checkpoint is incompatible); the necessary verification concerns The restarted query initializes the intended state schema without checkpoint errors.

This action solves a different issue: “A stateful aggregation grows without bound due to no late-event cutoff.” Its normal signal is State store size stabilizes as sufficiently old event-time state expires. Here the requested behavior is “A data engineer changes stateful grouping keys and the existing checkpoint is incompatible,” so that signal is not sufficient.

State store size stabilizes as sufficiently old event-time state expires would demonstrate the separate action Set a suitable event-time watermark for the use case, which addresses “A stateful aggregation grows without bound due to no late-event cutoff.” It does not verify the requirement “A data engineer changes stateful grouping keys and the existing checkpoint is incompatible” following Plan a new checkpoint and controlled state reinitialization or migration.

Retry of one microbatch does not duplicate the externally committed records would demonstrate the separate action Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key, which addresses “A stream must deliver at-least-once microbatches to an external system safely.” It does not verify the requirement “A data engineer changes stateful grouping keys and the existing checkpoint is incompatible” following Plan a new checkpoint and controlled state reinitialization or migration.

This signal belongs to the adjacent case “The stream restarts every day and must continue without rereading committed input,” whose remedy is Configure a durable, exclusive checkpoint location for the query. The observed result for the present requirement is The restarted query initializes the intended state schema without checkpoint errors.

This action solves a different issue: “Two independent streaming queries were configured to share the same checkpoint path.” Its normal signal is Each query maintains independent offsets and state without checkpoint collisions. Here the requested behavior is “A stateful aggregation grows without bound due to no late-event cutoff,” so that signal is not sufficient.

The alternative Plan a new checkpoint and controlled state reinitialization or migration has a distinct purpose—A data engineer changes stateful grouping keys and the existing checkpoint is incompatible. It would not establish the required outcome (A stateful aggregation grows without bound due to no late-event cutoff); the necessary verification concerns State store size stabilizes as sufficiently old event-time state expires.

Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key belongs to the separate case “A stream must deliver at-least-once microbatches to an external system safely,” where Retry of one microbatch does not duplicate the externally committed records would be expected. That differs from the condition in this stem: “A stateful aggregation grows without bound due to no late-event cutoff.”

The restarted query initializes the intended state schema without checkpoint errors pertains to a different control (Plan a new checkpoint and controlled state reinitialization or migration) and a different need (A data engineer changes stateful grouping keys and the existing checkpoint is incompatible). It cannot establish that Set a suitable event-time watermark for the use case met the stated goal.

This signal belongs to the adjacent case “A stream must deliver at-least-once microbatches to an external system safely,” whose remedy is Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key. The observed result for the present requirement is State store size stabilizes as sufficiently old event-time state expires.

Each query maintains independent offsets and state without checkpoint collisions pertains to a different control (Give each logical streaming query its own checkpoint location) and a different need (Two independent streaming queries were configured to share the same checkpoint path). It cannot establish that Set a suitable event-time watermark for the use case met the stated goal.

The alternative Set a suitable event-time watermark for the use case has a distinct purpose—A stateful aggregation grows without bound due to no late-event cutoff. It would not establish the required outcome (A stream must deliver at-least-once microbatches to an external system safely); the necessary verification concerns Retry of one microbatch does not duplicate the externally committed records.

Configure a durable, exclusive checkpoint location for the query is appropriate when “The stream restarts every day and must continue without rereading committed input”; it would be verified by After restart the offsets resume from checkpointed progress. It does not directly satisfy “A stream must deliver at-least-once microbatches to an external system safely,” whose relevant acceptance evidence is Retry of one microbatch does not duplicate the externally committed records.

Give each logical streaming query its own checkpoint location is appropriate when “Two independent streaming queries were configured to share the same checkpoint path”; it would be verified by Each query maintains independent offsets and state without checkpoint collisions. It does not directly satisfy “A stream must deliver at-least-once microbatches to an external system safely,” whose relevant acceptance evidence is Retry of one microbatch does not duplicate the externally committed records.

State store size stabilizes as sufficiently old event-time state expires would demonstrate the separate action Set a suitable event-time watermark for the use case, which addresses “A stateful aggregation grows without bound due to no late-event cutoff.” It does not verify the requirement “A stream must deliver at-least-once microbatches to an external system safely” following Use `foreachBatch` with idempotent writes keyed by microbatch ID or business key.

This signal belongs to the adjacent case “A data engineer changes stateful grouping keys and the existing checkpoint is incompatible,” whose remedy is Plan a new checkpoint and controlled state reinitialization or migration. The observed result for the present requirement is Retry of one microbatch does not duplicate the externally committed records.

That observation is a valid indicator for “Two independent streaming queries were configured to share the same checkpoint path” when using Give each logical streaming query its own checkpoint location. The scenario instead tests “A stream must deliver at-least-once microbatches to an external system safely”; the relevant result is Retry of one microbatch does not duplicate the externally committed records.