Exam Bundle

Retries do not duplicate outputs and batch duration meets the SLA would demonstrate the separate action Make `foreachBatch` sink operations idempotent and right-size write throughput, which addresses “A streaming sink repeatedly retries due to expensive non-idempotent external writes.” It does not verify the requirement “Streaming progress indicates no input backlog but high end-to-end event latency” following Examine source arrival timestamps and downstream processing or serving delays.

State rows stabilize while expected within-window matches remain would demonstrate the separate action Add appropriate watermarks and bounded join time ranges, which addresses “Stateful joins consume excessive memory because event histories never expire.” It does not verify the requirement “Streaming progress indicates no input backlog but high end-to-end event latency” following Examine source arrival timestamps and downstream processing or serving delays.

Processed rows per second eventually exceed incoming rate pertains to a different control (Reduce per-batch work or add supported compute capacity rather than merely shorten triggers) and a different need (Microbatch duration exceeds configured trigger interval for several minutes). It cannot establish that Examine source arrival timestamps and downstream processing or serving delays met the stated goal.

This action solves a different issue: “Several independent downstream consumers need different checkpoints.” Its normal signal is One slow consumer does not advance another's state erroneously. Here the requested behavior is “Only a fraction of an upstream Delta table changes each hour,” so that signal is not sufficient.

The alternative Deduplicate/merge idempotently by stable key and sequence has a distinct purpose—A CDF pipeline applies duplicate changed rows under retry. It would not establish the required outcome (Only a fraction of an upstream Delta table changes each hour); the necessary verification concerns The incremental job reads only applicable changes since checkpoint.

Perform a controlled full resync and establish a fresh version checkpoint belongs to the separate case “A consumer is far behind and its required CDF versions were removed by retention,” where The target matches source after resync and resumes incremental updates would be expected. That differs from the condition in this stem: “Only a fraction of an upstream Delta table changes each hour.”

That observation is a valid indicator for “Several independent downstream consumers need different checkpoints” when using Store a durable version watermark per consumer. The scenario instead tests “Only a fraction of an upstream Delta table changes each hour”; the relevant result is The incremental job reads only applicable changes since checkpoint.

Target state reflects the updated row once without double-counting would demonstrate the separate action Filter or reconcile change types to apply only intended current values, which addresses “A CDF update includes both preimage and postimage records.” It does not verify the requirement “Only a fraction of an upstream Delta table changes each hour” following Consume Change Data Feed instead of reprocessing the complete source table.

The target matches source after resync and resumes incremental updates pertains to a different control (Perform a controlled full resync and establish a fresh version checkpoint) and a different need (A consumer is far behind and its required CDF versions were removed by retention). It cannot establish that Consume Change Data Feed instead of reprocessing the complete source table met the stated goal.

Store a durable version watermark per consumer belongs to the separate case “Several independent downstream consumers need different checkpoints,” where One slow consumer does not advance another's state erroneously would be expected. That differs from the condition in this stem: “A CDF update includes both preimage and postimage records.”

Deduplicate/merge idempotently by stable key and sequence belongs to the separate case “A CDF pipeline applies duplicate changed rows under retry,” where A replay leaves the downstream target logically unchanged would be expected. That differs from the condition in this stem: “A CDF update includes both preimage and postimage records.”

Perform a controlled full resync and establish a fresh version checkpoint is appropriate when “A consumer is far behind and its required CDF versions were removed by retention”; it would be verified by The target matches source after resync and resumes incremental updates. It does not directly satisfy “A CDF update includes both preimage and postimage records,” whose relevant acceptance evidence is Target state reflects the updated row once without double-counting.

That observation is a valid indicator for “A CDF pipeline applies duplicate changed rows under retry” when using Deduplicate/merge idempotently by stable key and sequence. The scenario instead tests “A CDF update includes both preimage and postimage records”; the relevant result is Target state reflects the updated row once without double-counting.

The incremental job reads only applicable changes since checkpoint pertains to a different control (Consume Change Data Feed instead of reprocessing the complete source table) and a different need (Only a fraction of an upstream Delta table changes each hour). It cannot establish that Filter or reconcile change types to apply only intended current values met the stated goal.

The target matches source after resync and resumes incremental updates would demonstrate the separate action Perform a controlled full resync and establish a fresh version checkpoint, which addresses “A consumer is far behind and its required CDF versions were removed by retention.” It does not verify the requirement “A CDF update includes both preimage and postimage records” following Filter or reconcile change types to apply only intended current values.

Consume Change Data Feed instead of reprocessing the complete source table is appropriate when “Only a fraction of an upstream Delta table changes each hour”; it would be verified by The incremental job reads only applicable changes since checkpoint. It does not directly satisfy “A consumer is far behind and its required CDF versions were removed by retention,” whose relevant acceptance evidence is The target matches source after resync and resumes incremental updates.

Deduplicate/merge idempotently by stable key and sequence belongs to the separate case “A CDF pipeline applies duplicate changed rows under retry,” where A replay leaves the downstream target logically unchanged would be expected. That differs from the condition in this stem: “A consumer is far behind and its required CDF versions were removed by retention.”

Filter or reconcile change types to apply only intended current values belongs to the separate case “A CDF update includes both preimage and postimage records,” where Target state reflects the updated row once without double-counting would be expected. That differs from the condition in this stem: “A consumer is far behind and its required CDF versions were removed by retention.”

That observation is a valid indicator for “A CDF pipeline applies duplicate changed rows under retry” when using Deduplicate/merge idempotently by stable key and sequence. The scenario instead tests “A consumer is far behind and its required CDF versions were removed by retention”; the relevant result is The target matches source after resync and resumes incremental updates.

That observation is a valid indicator for “Several independent downstream consumers need different checkpoints” when using Store a durable version watermark per consumer. The scenario instead tests “A consumer is far behind and its required CDF versions were removed by retention”; the relevant result is The target matches source after resync and resumes incremental updates.

This signal belongs to the adjacent case “Only a fraction of an upstream Delta table changes each hour,” whose remedy is Consume Change Data Feed instead of reprocessing the complete source table. The observed result for the present requirement is The target matches source after resync and resumes incremental updates.

Filter or reconcile change types to apply only intended current values is appropriate when “A CDF update includes both preimage and postimage records”; it would be verified by Target state reflects the updated row once without double-counting. It does not directly satisfy “Several independent downstream consumers need different checkpoints,” whose relevant acceptance evidence is One slow consumer does not advance another's state erroneously.

This action solves a different issue: “Only a fraction of an upstream Delta table changes each hour.” Its normal signal is The incremental job reads only applicable changes since checkpoint. Here the requested behavior is “Several independent downstream consumers need different checkpoints,” so that signal is not sufficient.

Deduplicate/merge idempotently by stable key and sequence belongs to the separate case “A CDF pipeline applies duplicate changed rows under retry,” where A replay leaves the downstream target logically unchanged would be expected. That differs from the condition in this stem: “Several independent downstream consumers need different checkpoints.”

This signal belongs to the adjacent case “Only a fraction of an upstream Delta table changes each hour,” whose remedy is Consume Change Data Feed instead of reprocessing the complete source table. The observed result for the present requirement is One slow consumer does not advance another's state erroneously.

Target state reflects the updated row once without double-counting pertains to a different control (Filter or reconcile change types to apply only intended current values) and a different need (A CDF update includes both preimage and postimage records). It cannot establish that Store a durable version watermark per consumer met the stated goal.

A replay leaves the downstream target logically unchanged pertains to a different control (Deduplicate/merge idempotently by stable key and sequence) and a different need (A CDF pipeline applies duplicate changed rows under retry). It cannot establish that Store a durable version watermark per consumer met the stated goal.

The alternative Filter or reconcile change types to apply only intended current values has a distinct purpose—A CDF update includes both preimage and postimage records. It would not establish the required outcome (A CDF pipeline applies duplicate changed rows under retry); the necessary verification concerns A replay leaves the downstream target logically unchanged.

This action solves a different issue: “A consumer is far behind and its required CDF versions were removed by retention.” Its normal signal is The target matches source after resync and resumes incremental updates. Here the requested behavior is “A CDF pipeline applies duplicate changed rows under retry,” so that signal is not sufficient.

Store a durable version watermark per consumer belongs to the separate case “Several independent downstream consumers need different checkpoints,” where One slow consumer does not advance another's state erroneously would be expected. That differs from the condition in this stem: “A CDF pipeline applies duplicate changed rows under retry.”

That observation is a valid indicator for “A consumer is far behind and its required CDF versions were removed by retention” when using Perform a controlled full resync and establish a fresh version checkpoint. The scenario instead tests “A CDF pipeline applies duplicate changed rows under retry”; the relevant result is A replay leaves the downstream target logically unchanged.

Target state reflects the updated row once without double-counting would demonstrate the separate action Filter or reconcile change types to apply only intended current values, which addresses “A CDF update includes both preimage and postimage records.” It does not verify the requirement “A CDF pipeline applies duplicate changed rows under retry” following Deduplicate/merge idempotently by stable key and sequence.

This signal belongs to the adjacent case “Several independent downstream consumers need different checkpoints,” whose remedy is Store a durable version watermark per consumer. The observed result for the present requirement is A replay leaves the downstream target logically unchanged.