# `Rheo.Backoff`
[🔗](https://github.com/thanos/rheo/blob/v1.0.0/lib/rheo/backoff.ex#L1)

Exponential backoff schedule shared by `Rheo.Group` and `Rheo.Producer`.

A runtime holds the current delay in its state (`0` when healthy). Each
consecutive fetch failure doubles the delay from `min_ms/0` up to `max_ms/0`;
a successful fetch resets it to `0`. The schedule is fixed (100 ms … 5_000 ms)
so coordinators share the same pacing without per-instance config.

## Examples

    iex> Rheo.Backoff.next(0)
    100

    iex> Rheo.Backoff.next(100)
    200

    iex> Rheo.Backoff.next(4_000)
    5000

    iex> Rheo.Backoff.min_ms()
    100

    iex> Rheo.Backoff.max_ms()
    5000

# `max_ms`

```elixir
@spec max_ms() :: pos_integer()
```

Largest delay in milliseconds.

## Examples

    iex> Rheo.Backoff.max_ms()
    5000

## Returns

`5000`.

# `min_ms`

```elixir
@spec min_ms() :: pos_integer()
```

Smallest non-zero delay in milliseconds.

## Examples

    iex> Rheo.Backoff.min_ms()
    100

## Returns

`100`.

# `next`

```elixir
@spec next(non_neg_integer()) :: pos_integer()
```

Next delay after `current_ms`.

Pass `0` when there is no backoff yet (healthy). Otherwise doubles
`current_ms` and clamps at `max_ms/0`.

## Examples

    iex> Rheo.Backoff.next(0)
    100

    iex> Rheo.Backoff.next(100)
    200

    iex> Rheo.Backoff.next(Rheo.Backoff.max_ms())
    5000

## Returns

A positive integer delay in milliseconds.

---

*Consult [api-reference.md](api-reference.md) for complete listing*