<a href="https://github.com/containous/traefik/releases" class="external-link" target="_blank">下载 Traefik</a>，这是一个二进制文件，需要解压文件，并在 Terminal 中直接运行。

然后创建包含如下内容的 `traefik.toml` 文件：

```TOML hl_lines="3"
[entryPoints]
  [entryPoints.http]
    address = ":9999"

[providers]
  [providers.file]
    filename = "routes.toml"
```

这个文件把 Traefik 监听端口设置为 `9999`，并设置要使用另一个文件 `routes.toml`。

/// tip | "提示"

      if (object instanceof Multiset.Entry) {
        Entry<?> entry = (Entry<?>) object;
        Object element = entry.getElement();
        int entryCount = entry.getCount();
        if (entryCount != 0) {
          // Safe as long as we never add a new entry, which we won't.
          // (Presumably it can still throw CCE/NPE but only if the underlying Multiset does.)

      if (object instanceof Multiset.Entry) {
        Entry<?> entry = (Entry<?>) object;
        Object element = entry.getElement();
        int entryCount = entry.getCount();
        if (entryCount != 0) {
          // Safe as long as we never add a new entry, which we won't.
          // (Presumably it can still throw CCE/NPE but only if the underlying Multiset does.)

                    jarOutputStream.addJarEntry(JarFile.MANIFEST_NAME, manifestFile)
                }
                val visited = linkedSetOf<ClassDetails>()
                for (classDetails in classGraph.entryPoints) {
                    visitTree(classDetails, classesDir, jarOutputStream, visited)
                }
            }
        } catch (exception: Exception) {

    accepts a `return_attention_scores` argument. When set to True, the layer
    returns the attention scores as an additional output argument.
*   Adds `tf.metrics.log_cosh` and `tf.metrics.logcosh` API entrypoints with the
    same implementation as their `tf.losses` equivalent.
*   For Keras model, the individual call of `Model.evaluate` uses no cached data
    for evaluation, while `Model.fit` uses cached data when `validation_data`