OTCA — Maintaining and Debugging Observability Pipelines
10% of the OTCA exam. Sample questions below; the full library has 12 questions tagged to this domain.
Sample questions on Maintaining and Debugging Observability Pipelines
Maintaining and Debugging Observability Pipelines
Q1. Why does each OpenTelemetry telemetry payload include a `schema_url` field?
Reveal answer and explanations
ATo declare the semantic-convention schema version used by the producer
Correct. `schema_url` identifies the version of the semantic conventions that the producer used, letting consumers apply schema transformations or migrate between conventions cleanly.
BTo point consumers to the Collector's runtime configuration file
Incorrect. The Collector's configuration is local; it is not referenced via `schema_url` in telemetry.
CTo advertise the gRPC endpoint that downstream consumers should use
Incorrect. Endpoint selection is configured separately, not via `schema_url`.
DTo act as the OAuth2 token issuer URL the backend uses for client authentication and authorization
Incorrect. Authentication is unrelated to the schema URL.
Maintaining and Debugging Observability Pipelines
Q2. After upgrading the SDK, a backend complains about an unknown attribute key on incoming telemetry. The team notices `schema_url` advanced to a newer semantic-convention version. What is the appropriate response?
Reveal answer and explanations
ARoll back the SDK upgrade because semantic conventions must never change once adopted
Incorrect. Schema evolution is expected; the mechanism exists to handle it gracefully.
BSet `OTEL_SDK_DISABLED=true` in production to bypass the conflict
Incorrect. Disabling the SDK stops telemetry entirely and does not address schema evolution.
CStrip `schema_url` from all telemetry to silence the warning
Incorrect. Stripping `schema_url` removes the very metadata that lets consumers translate; it does not solve the conflict.
DUse the new `schema_url`'s files to translate attributes downstream, or pin the producer to the older schema
Correct. `schema_url` advertises the schema version, and OpenTelemetry Schema Files describe transformations between versions so consumers can either translate forward/backward or coordinate the upgrade window between producers and consumers.
Maintaining and Debugging Observability Pipelines
Q3. An async worker reads jobs from a Kafka topic and the resulting spans appear as new root spans rather than children of the producer. What is the most likely cause?
Reveal answer and explanations
AThe consumer is using OTLP/HTTP rather than OTLP/gRPC for telemetry export to the Collector
Incorrect. The transport between SDK and Collector does not affect end-to-end context propagation through a message bus.
BTrace context is not propagated through Kafka headers from producer to consumer
Correct. To stitch producer and consumer spans, the producer must inject trace context into Kafka message headers and the consumer must extract it; missing this step yields disconnected root spans on the consumer side.
CThe Collector is using `tail_sampling` instead of `probabilistic_sampler`
Incorrect. The choice of sampler does not decide whether spans are linked through Kafka.
DThe consumer's system clock is ahead of the producer's by several seconds
Incorrect. Clock skew affects timestamps but not parent-child relationships.
Maintaining and Debugging Observability Pipelines
Q4. A Collector is restarted and a backend was unreachable during the entire window. Which configuration would have minimized data loss across the restart?
Reveal answer and explanations
AIncreasing `sending_queue.queue_size` to a very large in-memory value
Incorrect. In-memory queues, however large, vanish on restart.
BEnabling `sending_queue.storage` with a `file_storage` extension
Correct. A persistent sending queue backed by the `file_storage` extension writes queued telemetry to disk so a Collector restart does not lose buffered data; the queue replays once the backend recovers.
CDisabling `retry_on_failure` so the Collector exits cleanly
Incorrect. Disabling retries makes data loss worse, not better.
DSwitching to the `debug` exporter during the outage
Incorrect. The debug exporter writes to stdout and does not address durability.
Maintaining and Debugging Observability Pipelines
Q5. What does the `zpages` extension provide?
Reveal answer and explanations
AA web UI for editing the Collector's YAML configuration at runtime through an admin dashboard
Incorrect. `zpages` does not edit configuration; it is read-only diagnostics.
BA bearer-token authentication scheme used by the Collector's admin API
Incorrect. Authentication is provided by separate extensions like `bearertokenauth`.
CAn alternative receiver that replaces the standard `otlp` receiver
Incorrect. It is not a receiver; it is an extension for diagnostics.
DBuilt-in HTTP diagnostic pages for traces, pipelines, and component status
Correct. `zpages` exposes HTTP pages such as `/debug/tracez` and `/debug/pipelinez` for inspecting recent traces, pipelines, and component health from the Collector itself.
Maintaining and Debugging Observability Pipelines accounts for 10% of the OTCA exam. Expect questions that test recall of terminology and the ability to read short scenarios — not deep configuration. Use the sample questions above as difficulty calibration; if any feel hard, the rest of our 12-question domain bank will close those gaps.