roadmap
- 01
- 02
- 03
- 04
- 05
- 06
- 07
- 08
Prepare for Gatling interview questions on writing load tests as code in Java, Kotlin, Scala or JavaScript. Covers workload theory, injection profiles, feeders and checks, assertions and CI runs.
156 of 156 questions in your plan: 0 known, 0 in progress, 0 due, 0 skipped, 156 new. 0 cut, outside your plan.
Gatling questions turn up in QA, SDET and performance-engineering loops, usually inside a round on performance testing, and in backend loops on JVM teams that keep load tests beside the service code. The round tends to start from a workload you are asked to express: two minutes in which nobody arrives and then a burst of users at once, or 300 users held for ten minutes and brought down to 50. What the interviewer listens for is whether you reach for an open step such as constantUsersPerSec or a closed one such as constantConcurrentUsers, and what each one does to the load when the service slows down.
Follow-ups move to the data and the verdict. A login simulation stops partway through because its CSV feeder, left on the default queue strategy, ran out of rows. A captured order id is missing, and both the request that should have saved it and the one that sends it fail. A build step has exited 0 for months because the simulation never registered an assertion. Candidates separate on whether they can say what the numbers mean, not only write the DSL: why a throttled run can finish below its target rate, why a percentile3 budget depends on a setting outside the simulation, and what you take on by hand when the Community Edition runs on several machines at once.
The opening module is theory that is not Gatling's own, and it comes first because the DSL is an expression of it. Load modeling separates an arrival rate from a count of concurrent users, the distinction every injection step encodes; run shapes say what a spike or a soak is for; and reading results treats percentiles, failed requests and pass rules before the report puts them in front of you. The saturation-signals topic adds the load, stress and soak vocabulary many performance rounds open with.
Gatling itself starts with the scenario rather than the load. Chained builders cover the SDK each language gets and the loops, pauses and groups a scenario is assembled from. Session values follow, because a virtual user that behaves like a real one needs records fed from a file and fields captured from one response to send on the next.
Only then does injection arrive, and the opening theory returns as code: open arrival steps, closed concurrency steps, and assembling a whole profile from staircases, the setUp call and a throttle. The plan then asks what a finished run says, through the assertion chain and its exit code and the HTML report. It closes outside the simulation file, on launching it from a build tool or a pipeline and on how Gatling stands against the other load generators a team might pick.
When Gatling is one tool among several in a QA or SDET posting, narrow level to the range you are applying at and raise importance to must know; a tool-level question usually lands in the scripting and injection modules, and the theory can wait unless open and closed workload models are new to you. For a performance-engineering role, switch study order to by topics and work from the top, since the senior questions on throttling, percentiles and run verdicts rest on that opening theory. Someone moving over from a thread-group tool should give injection the most time: Gatling declares a load profile as a chain of injection steps instead of sizing a pool of threads. The plan teaches one load generator, and a performance round often asks you to compare it with [k6](/roadmaps/k6) or [JMeter](/roadmaps/jmeter). The test design, API and UI automation questions asked in the same loop belong to the [QA Engineer roadmap](/roadmaps/qa-engineer), which carries this tool beside them; for a Java automation role, the [Java SDET roadmap](/roadmaps/java-sdet) does the same.
When Gatling is one tool among several in a QA or SDET posting, narrow level to the range you are applying at and raise importance to must know; a tool-level question usually lands in the scripting and injection modules, and the theory can wait unless open and closed workload models are new to you. For a performance-engineering role, switch study order to by topics and work from the top, since the senior questions on throttling, percentiles and run verdicts rest on that opening theory. Someone moving over from a thread-group tool should give injection the most time: Gatling declares a load profile as a chain of injection steps instead of sizing a pool of threads.
The plan teaches one load generator, and a performance round often asks you to compare it with k6 or JMeter. The test design, API and UI automation questions asked in the same loop belong to the QA Engineer roadmap, which carries this tool beside them; for a Java automation role, the Java SDET roadmap does the same.