Free AT0-001 practice test: 30 questions with answers

In Bash, each part of a pipeline, including the while loop here, runs in a subshell, so the parent shell keeps its original value of 0. The other options are wrong because Bash arithmetic works on untyped variables, loops can modify variables, and a read-only variable would produce an error rather than a silent 0.

Checking inputs inside the function stops an empty or unexpected value from reaching a destructive command. Removing the check, defaulting to the root directory, or raising privileges all increase the risk of deleting the wrong data.

A bounded loop combines an upper limit on attempts with an early exit on success, which meets both requirements. An unbounded loop or recursion without a base case can hang the pipeline, and a single attempt never waits for readiness.

Merging main into the feature branch keeps the existing commits and adds a merge commit, so no history is rewritten. A hard reset discards the branch's work, force-pushing main over the feature branch destroys remote history, and recreating the branch by hand loses its commit history.

A feature branch isolates in-progress work so the shared main branch remains deployable until the change passes review. Branches do not store secrets or grant host access, and release tags, not feature branches, are what identify a version for redeployment.

Idempotency means running the same desired-state task again leaves the system unchanged once the target state is reached. Immutability describes replacing rather than modifying resources, while concurrency safety and versioning describe different concerns.

Immutable infrastructure replaces servers from a known image rather than changing them after launch, which reduces configuration drift. Mutable infrastructure is the approach the team stopped using, while drift detection and canary deployment solve different problems.

Parse errors after a merge are often caused by conflict markers or indentation changed during conflict resolution, so inspecting and linting the file near the reported line is the right first step. A longer timeout does not fix syntax, and rewriting or removing the file skips the diagnosis and risks losing reviewed changes.

Configuration drift describes live systems that no longer match the declared desired state, often because of manual or out-of-band changes. The other options describe a rebuild, a release strategy, and a concurrency control, none of which explains a mismatch with the repository.

Declarative configuration describes what the end state should be and leaves the sequence of actions to the tool, which is what this file does. The imperative model lists commands step by step, while ad hoc console changes and event-driven triggers are not defined by a desired-state file.

In a pull model, nodes check in on a schedule and retrieve their own desired state, so the controller does not need to reach into each node. The other options describe push-based control or wrongly claim that pull-based systems cannot use declarative files.

GET retrieves a representation of a resource and is the usual mapping for the CRUD Read operation. POST, PUT, and DELETE map to Create, Update, and Delete, and each of them changes server state, which this workflow must avoid.

Clients validate a server certificate by building a chain up to a trusted root, so the server must send its intermediate certificates. Disabling verification exposes the job to impersonation, and changing the HTTP method or the timeout has no effect on certificate validation.

A 403 status means the server knows who is calling but refuses the action because that identity is not authorized. An expired token normally produces 401 Unauthorized, a malformed body is a 400-class error, and an unreachable gateway produces a connection failure or timeout rather than an HTTP status.

Scheduled reconciliation and version-controlled definitions both expose and correct differences between the declared and live state. Undocumented hotfixes and manual rebuilds from notes are common sources of drift rather than ways to prevent it.

PATCH sends only the changes and leaves other fields as they are, which matches a partial update. PUT replaces the whole resource, POST creates a new one, and deleting then reading removes the record instead of editing it.

An artifact repository stores versioned build outputs so that the same tested package or image can be promoted through environments without being rebuilt. Editable source folders, overwritten test results, and database copies are not artifacts that deployments are meant to reuse.

Rollback depends on keeping earlier artifacts and having a repeatable step that redeploys one of them. A rebuild from the latest commit may carry the same fault, schema edits do not restore the old version, and deleting a release without redeploying leaves the service without a working version.

The on key declares the triggering events for a workflow. The jobs, runs-on, and steps keys define what runs, where it runs, and the commands inside a job, so each of them answers a different question from the one asked.

A Jenkinsfile holds the pipeline definition as code, so it can be versioned, reviewed, and branched with the application. A Dockerfile builds container images, a Makefile is a build tool file that should not hold credentials, and settings.gradle configures Gradle builds rather than Jenkins agents.

When a push succeeds but no build starts, the event delivery from the repository host to the CI server is the first thing to verify. Adding tests, re-cloning locally, and force-pushing do not address whether the trigger reached the CI server.

Continuous integration focuses on merging changes frequently and validating each one with automated builds and tests. Continuous deployment goes further by releasing each change to production, while infrastructure as code and drift detection address different concerns.

Masking hides secret values in logs, and run-time injection from a secret store keeps the values out of code and repositories. Printing environment variables exposes secrets in logs, and committing environment files stores real values in the repository where they can leak.

A canary release exposes a limited group of users to the new version so that problems appear early with limited impact. Blue-green switches all traffic at a cutover, rolling replaces instances in batches, and recreate stops everything before starting the new version.

Blue-green keeps the previous environment intact, so a rollback is a traffic switch rather than a rebuild. Blue-green does need a second environment, individual instance updates describe rolling deployments, and fixed-percentage routing describes canary releases.

A rolling deployment updates instances in batches, so capacity stays available during the change and no second full environment is needed. Blue-green requires duplicate capacity, recreate causes downtime, and a feature flag does not replace the running instances.

Least privilege limits what a pipeline can do if its credentials leak, and short-lived workload identities avoid long-lived secrets. Shared administrator keys, keys committed in code, and owner-level grants all widen the impact of a compromise.

An SLA is the customer-facing agreement, often with stated remedies, and MTTR is an average restoration time. An SLO is an internal reliability target rather than a contract, and MTTR is measured in time, not as a percentage.

An error budget is the unreliability the SLO allows, and when it is spent, common practice is to pause risky changes and prioritize stability. Lowering the target to refill the budget, ignoring it, or disabling monitoring all undermine the reliability goal.

Feature flags separate deploying code from enabling a feature, so the change is live in production but hidden from most users until it is turned on. Per-tester branches, duplicate environments, and random nightly copies do not give a controlled release of visibility.