Vladimír Čunát

We didn't post here, but the fcntl calls are wrong. Setter takes an `int`, not a pointer. And `getter` takes nothing, returns value. EDIT: that's how `man fcntl.2` reads to...

Nit: why design the interface with `size_t` when we're only able to implement `int`? EDIT: especially note that the (current) implementation silently overflows to non-sensical values.

Accepting negative numbers? Well, if you do an `unsigned int` or `size_t` and pass a too large value, you can end up passing a negative number to the syscall. So...

We don't have a separate negative cache. The implementation would need to pass through all records in the cache and determine which would be viewed as negative (including all positive...

We do have settings for TTL bounds (lower and upper). So adding a separate option for "negative" entries – I'd find that reasonable and not complicated.

Note that each domain.com has control over their TTLs, so they can tweak the upper bounds in any way they want...

As it was always hanging, I rolled back the rhea's `machines` file, so that scheduler wouldn't try to run these jobs.

Running those jobs doesn't seem to work. They hang and time out. (I observed it several days ago again.)

I don't think that's currently doable in a reasonable way with Knot Resolver.

You could run (some) requests as `.STUB` with cache turned off, if I recall right. Though that's slightly hacky, too.