How edge caching works
Hit/miss, freshness, query strings and the difference between domain and path lifetimes.
When caching is enabled the proxy keeps a copy of each cacheable response keyed by host + path (and, unless you opt out, the query string). A subsequent request for the same key is answered from the edge in well under a millisecond without touching your origin — that is a cache hit. A miss fetches from the origin, stores the result and serves it.
Freshness
How long a copy stays fresh is the cache lifetime. The domain-wide default is max_age_seconds; individual paths can override it with path_rules. When a copy expires the next request re-fetches from the origin.
Enabling caching
Caching has a global master switch (add_caching, on by default) and is shaped per domain by the lifetime and path rules. Turn the master off to run the proxy as a pure pass-through.