Exam Bundle

That observation is a valid indicator for “A CLI script must use a production workspace rather than a developer's default profile” when using Pass the intended Databricks CLI profile or host configuration. The scenario instead tests “A monitoring identity should not have permissions to edit production jobs”; the relevant result is The identity reads run status but fails job mutation requests.

That observation is a valid indicator for “A monitoring script needs the state of a specific current job run” when using Call the supported Jobs API run detail endpoint using its run ID. The scenario instead tests “A monitoring identity should not have permissions to edit production jobs”; the relevant result is The identity reads run status but fails job mutation requests.

The response identifies pending, running and terminal update states would demonstrate the separate action Query the supported pipelines API for update details, which addresses “A nightly dashboard should report active production pipeline update status.” It does not verify the requirement “A monitoring identity should not have permissions to edit production jobs” following Grant read/monitor privileges only to the service principal.

Inspect streaming state operator metrics and batch execution durations belongs to the separate case “A microbatch duration spikes after an increase in state cardinality,” where State rows and processing time increase together for affected batches would be expected. That differs from the condition in this stem: “A stream seems healthy but event processing falls farther behind incoming data.”

Inspect watermark position and dropped-late-record progress counters is appropriate when “Late-data drops rise after a watermark configuration change”; it would be verified by Dropped record count reflects events older than the configured cutoff. It does not directly satisfy “A stream seems healthy but event processing falls farther behind incoming data,” whose relevant acceptance evidence is The arrival rate exceeds processing capacity during the backlog window.

This action solves a different issue: “A streaming sink occasionally retries the same batch after transient errors.” Its normal signal is Repeated batch IDs are visible without duplicate accepted logical records. Here the requested behavior is “A stream seems healthy but event processing falls farther behind incoming data,” so that signal is not sufficient.

Dropped record count reflects events older than the configured cutoff would demonstrate the separate action Inspect watermark position and dropped-late-record progress counters, which addresses “Late-data drops rise after a watermark configuration change.” It does not verify the requirement “A stream seems healthy but event processing falls farther behind incoming data” following Compare input rate and processed-rows rate in streaming progress.

The restarted source offset starts after prior committed progress would demonstrate the separate action Compare before/after checkpointed offsets and processing progress, which addresses “A query restarts and must be proven to resume from its prior committed offset.” It does not verify the requirement “A stream seems healthy but event processing falls farther behind incoming data” following Compare input rate and processed-rows rate in streaming progress.

Repeated batch IDs are visible without duplicate accepted logical records pertains to a different control (Correlate batch IDs and sink commit logs across retries) and a different need (A streaming sink occasionally retries the same batch after transient errors). It cannot establish that Compare input rate and processed-rows rate in streaming progress met the stated goal.

The alternative Correlate batch IDs and sink commit logs across retries has a distinct purpose—A streaming sink occasionally retries the same batch after transient errors. It would not establish the required outcome (A microbatch duration spikes after an increase in state cardinality); the necessary verification concerns State rows and processing time increase together for affected batches.

Compare before/after checkpointed offsets and processing progress belongs to the separate case “A query restarts and must be proven to resume from its prior committed offset,” where The restarted source offset starts after prior committed progress would be expected. That differs from the condition in this stem: “A microbatch duration spikes after an increase in state cardinality.”

Compare input rate and processed-rows rate in streaming progress belongs to the separate case “A stream seems healthy but event processing falls farther behind incoming data,” where The arrival rate exceeds processing capacity during the backlog window would be expected. That differs from the condition in this stem: “A microbatch duration spikes after an increase in state cardinality.”

Dropped record count reflects events older than the configured cutoff pertains to a different control (Inspect watermark position and dropped-late-record progress counters) and a different need (Late-data drops rise after a watermark configuration change). It cannot establish that Inspect streaming state operator metrics and batch execution durations met the stated goal.

That observation is a valid indicator for “A query restarts and must be proven to resume from its prior committed offset” when using Compare before/after checkpointed offsets and processing progress. The scenario instead tests “A microbatch duration spikes after an increase in state cardinality”; the relevant result is State rows and processing time increase together for affected batches.

Repeated batch IDs are visible without duplicate accepted logical records pertains to a different control (Correlate batch IDs and sink commit logs across retries) and a different need (A streaming sink occasionally retries the same batch after transient errors). It cannot establish that Inspect streaming state operator metrics and batch execution durations met the stated goal.

Inspect streaming state operator metrics and batch execution durations belongs to the separate case “A microbatch duration spikes after an increase in state cardinality,” where State rows and processing time increase together for affected batches would be expected. That differs from the condition in this stem: “Late-data drops rise after a watermark configuration change.”

Compare before/after checkpointed offsets and processing progress belongs to the separate case “A query restarts and must be proven to resume from its prior committed offset,” where The restarted source offset starts after prior committed progress would be expected. That differs from the condition in this stem: “Late-data drops rise after a watermark configuration change.”

This action solves a different issue: “A streaming sink occasionally retries the same batch after transient errors.” Its normal signal is Repeated batch IDs are visible without duplicate accepted logical records. Here the requested behavior is “Late-data drops rise after a watermark configuration change,” so that signal is not sufficient.

This signal belongs to the adjacent case “A streaming sink occasionally retries the same batch after transient errors,” whose remedy is Correlate batch IDs and sink commit logs across retries. The observed result for the present requirement is Dropped record count reflects events older than the configured cutoff.

That observation is a valid indicator for “A microbatch duration spikes after an increase in state cardinality” when using Inspect streaming state operator metrics and batch execution durations. The scenario instead tests “Late-data drops rise after a watermark configuration change”; the relevant result is Dropped record count reflects events older than the configured cutoff.

That observation is a valid indicator for “A query restarts and must be proven to resume from its prior committed offset” when using Compare before/after checkpointed offsets and processing progress. The scenario instead tests “Late-data drops rise after a watermark configuration change”; the relevant result is Dropped record count reflects events older than the configured cutoff.

Inspect watermark position and dropped-late-record progress counters is appropriate when “Late-data drops rise after a watermark configuration change”; it would be verified by Dropped record count reflects events older than the configured cutoff. It does not directly satisfy “A streaming sink occasionally retries the same batch after transient errors,” whose relevant acceptance evidence is Repeated batch IDs are visible without duplicate accepted logical records.

This action solves a different issue: “A stream seems healthy but event processing falls farther behind incoming data.” Its normal signal is The arrival rate exceeds processing capacity during the backlog window. Here the requested behavior is “A streaming sink occasionally retries the same batch after transient errors,” so that signal is not sufficient.

The alternative Inspect streaming state operator metrics and batch execution durations has a distinct purpose—A microbatch duration spikes after an increase in state cardinality. It would not establish the required outcome (A streaming sink occasionally retries the same batch after transient errors); the necessary verification concerns Repeated batch IDs are visible without duplicate accepted logical records.

Dropped record count reflects events older than the configured cutoff pertains to a different control (Inspect watermark position and dropped-late-record progress counters) and a different need (Late-data drops rise after a watermark configuration change). It cannot establish that Correlate batch IDs and sink commit logs across retries met the stated goal.

That observation is a valid indicator for “A stream seems healthy but event processing falls farther behind incoming data” when using Compare input rate and processed-rows rate in streaming progress. The scenario instead tests “A streaming sink occasionally retries the same batch after transient errors”; the relevant result is Repeated batch IDs are visible without duplicate accepted logical records.

State rows and processing time increase together for affected batches pertains to a different control (Inspect streaming state operator metrics and batch execution durations) and a different need (A microbatch duration spikes after an increase in state cardinality). It cannot establish that Correlate batch IDs and sink commit logs across retries met the stated goal.

Correlate batch IDs and sink commit logs across retries belongs to the separate case “A streaming sink occasionally retries the same batch after transient errors,” where Repeated batch IDs are visible without duplicate accepted logical records would be expected. That differs from the condition in this stem: “A query restarts and must be proven to resume from its prior committed offset.”

Inspect streaming state operator metrics and batch execution durations is appropriate when “A microbatch duration spikes after an increase in state cardinality”; it would be verified by State rows and processing time increase together for affected batches. It does not directly satisfy “A query restarts and must be proven to resume from its prior committed offset,” whose relevant acceptance evidence is The restarted source offset starts after prior committed progress.

The alternative Inspect watermark position and dropped-late-record progress counters has a distinct purpose—Late-data drops rise after a watermark configuration change. It would not establish the required outcome (A query restarts and must be proven to resume from its prior committed offset); the necessary verification concerns The restarted source offset starts after prior committed progress.

State rows and processing time increase together for affected batches pertains to a different control (Inspect streaming state operator metrics and batch execution durations) and a different need (A microbatch duration spikes after an increase in state cardinality). It cannot establish that Compare before/after checkpointed offsets and processing progress met the stated goal.

That observation is a valid indicator for “A stream seems healthy but event processing falls farther behind incoming data” when using Compare input rate and processed-rows rate in streaming progress. The scenario instead tests “A query restarts and must be proven to resume from its prior committed offset”; the relevant result is The restarted source offset starts after prior committed progress.

Dropped record count reflects events older than the configured cutoff would demonstrate the separate action Inspect watermark position and dropped-late-record progress counters, which addresses “Late-data drops rise after a watermark configuration change.” It does not verify the requirement “A query restarts and must be proven to resume from its prior committed offset” following Compare before/after checkpointed offsets and processing progress.