│   │   ├── SuggestIndexer.java     # Main indexing API
│   │   ├── contents/               # Content parsers and readers
│   │   └── writer/                 # Index writing strategies
│   ├── converter/                  # Text conversion utilities
│   ├── normalizer/                 # Text normalization
│   ├── analysis/                   # Custom analyzers

```

Then change the field `image.tag` in `old_values.yaml` file with MinIO image tag you want to use. Now update the chart using

```bash
helm upgrade -f old_values.yaml my-release minio/minio
```

Default upgrade strategies are specified in the `values.yaml` file. Update these fields if you'd like to use a different strategy.

### Configuration

Refer the [Values file](./values.yaml) for all the possible config fields.

當然，同一台機器上除了你的應用之外，通常也會有「其他行程」在執行。

有個有趣的細節是，每個行程的「CPU 使用率」百分比會隨時間大幅「變動」，但「記憶體（RAM）」通常維持相對「穩定」。

如果你的 API 每次做的計算量相近，且客戶很多，那「CPU 使用率」也可能「相對穩定」（而不是快速上下起伏）。

### 複本與擴展的工具與策略範例 { #examples-of-replication-tools-and-strategies }

要達成這些有很多種作法。我會在後續章節（例如談到 Docker 與容器時）介紹更具體的策略。

主要的限制是：必須有「單一」元件負責處理「公開 IP」上的「連接埠」。接著它必須能把通信「轉發」到被複製的「行程／workers」。

以下是一些可能的組合與策略：

- Uvicorn 搭配 `--workers`

        this.bitSize(),
        that.bitSize());
    checkArgument(
        this.strategy.equals(that.strategy),
        "BloomFilters must have equal strategies (%s != %s)",
        this.strategy,
        that.strategy);
    checkArgument(
        this.funnel.equals(that.funnel),
        "BloomFilters must have equal funnels (%s != %s)",
        this.funnel,

3. **Persistent Handle State**: Requires reliable state persistence mechanism
4. **Backward Compatibility**: Must maintain compatibility with SMB2/SMB1

### Mitigation Strategies
1. Implement features behind configuration flags
2. Provide graceful fallback mechanisms
3. Extensive testing with various server configurations
4. Phased rollout with beta testing

## Resource Requirements

    // Statistics
    private final AtomicLong totalRequests = new AtomicLong(0);
    private final AtomicLong totalChannelsCreated = new AtomicLong(0);

    /**
     * Load balancing strategies for distributing operations across channels.
     */
    public enum LoadBalancingStrategy {
        /** Round-robin distribution */
        ROUND_ROBIN,
        /** Least connections */
        LEAST_CONNECTIONS,

3. **Parent-Child Relationships**: Implement hierarchical leases
4. **Lease Sharing**: Support lease sharing across handles

      return count == 0 ? EMPTY : new ImmutableLongArray(array, 0, count);
    }
  }

  // Instance stuff here

  // The array is never mutated after storing in this field and the construction strategies ensure
  // it doesn't escape this class
  @SuppressWarnings("Immutable")
  private final long[] array;

  /*
   * TODO(kevinb): evaluate the trade-offs of going bimorphic to save these two fields from most

        VersionRange range2 = VersionRange.createFromVersionSpec("1.1");
        VersionRange mergedRange = range1.restrict(range2);
        // TODO current policy is to retain the original version - is this correct, do we need strategies or is that
        // handled elsewhere?
        //        assertEquals( "1.1", mergedRange.getRecommendedVersion().toString(), CHECK_VERSION_RECOMMENDATION );

- we need to deal with a shared context for plugins, like the guarded mojos
- we need to deal with plugins as core application logic which can also interact with plugins

* Along with this comes the testing strategies that make this work
* The repository model where plugins can be stored and cataloged
* Bridging this into an OSGi system: if this could be done then we can basically take over p2

TODO