Request history

Updated Aug 30, 2026·3 min read

Every request a simulation answers leaves a record: when it came in, what it asked, what went back, and who answered. History is the log you open when a client misbehaves, an integration fails, or you need to see the exact request and the exact answer.

Browse the log

Open History under Recordings in the sidebar and pick a service. Each row is one request: time, method, the full path with its query string, the status, the source, and how long the answer took.

The History view: one service's requests, newest first
  • Source says who answered: mock is the simulation, upstream is your real backend behind it.
  • The time range menu narrows the list: the last 15 minutes, hour, 24 hours, 7 days, or a custom range.
  • Entries load newest first; Load older pages further back.

The same columns exist as response headers on the wire: X-Mockzilla-Source and X-Mockzilla-Duration on every response the simulation sends.

The detail view

Click a row for the whole exchange: request headers and body on top, response headers and the pretty-printed body below, with source, duration and the exact timestamp in the header.

One request opened: headers, bodies, source and duration

Masked headers

Credentials do not belong in a log. Header values on the mask list are masked when the entry is recorded, keeping only the last four characters, ***********4f2a style, so the original value is never stored at all. Authorization, Cookie, Set-Cookie and X-Api-Key are masked from the start; add your own by name, or as a prefix pattern like X-Internal-*. Matching is case-insensitive.

Recording and retention

Recording is on by default, with a one-hour window: entries older than the retain duration expire on their own. History is a rolling window for debugging, not an archive.

The global default applies to every service; Override global gives one service its own recording switch, retain duration and mask list.

A service overriding the global default: record, retain and the mask list

Turn recording off before a load test. Logging every request adds a little work to each call, and a test that fires thousands of identical requests fills the window with rows that tell you nothing; the numbers you care about live in the load tool anyway.

Like other simulation settings, changes here ship with the next deploy.

Was this page helpful?