GBase Database(GBase 8c): Why Three Deployment Layouts Have Different Configuration Fields

GBase Database (GBase 8c) 的三种部署布局共享通用的配置结构，但它们并不完全使用相同的字段、端口分配或组件数量。这种区别很容易被忽视。配置文件看起来可能足够相似，以至于让人想直接复制，但使用错误的部署示例可能会导致意想不到的拓扑结构和安装问题。关键在于理解最小化布局、单高可用组布局和默认分布式布局是三个独立的配置基准，而不是同一个通用模板的三种变体。

1. Configuration Fields: What Exists and What Does Not? Start with field presence. The minimal and single-HA-group examples include: az for DN instances. proxy_port for CN and DN instances. The official default distributed example includes neither field. This is not necessarily an omission or an incomplete configuration. Each example is designed for its own deployment layout. When adapting a GBase Database configuration, do not assume that every field in one example must also appear in another.

1. 配置字段：哪些存在，哪些不存在？ 从字段的存在性开始。最小化和单高可用组示例包括：用于 DN 实例的 az，以及用于 CN 和 DN 实例的 proxy_port。官方的默认分布式示例则不包含这两个字段。这并不一定是遗漏或配置不完整。每个示例都是为其特定的部署布局而设计的。在调整 GBase Database 配置时，不要假设一个示例中的每个字段都必须出现在另一个示例中。

2. Port Assignments Are Layout-Specific The differences are not limited to field presence. Port assignments also vary. GTM agent_port: Single-HA-group example: 36667 and 46667. Default distributed example: 8001 and 8002. DN port assignments: In the default distributed example: DN instance Port dn1_1 15432 dn1_2 15433 dn1_3 15433 dn2 instances 20010 These values belong to their respective configuration examples. Do not blindly reuse port assignments when moving between deployment layouts. Always verify port availability and the actual configuration requirements for each host.

2. 端口分配是特定于布局的 差异不仅限于字段的存在性，端口分配也各不相同。GTM agent_port：单高可用组示例为 36667 和 46667；默认分布式示例为 8001 和 8002。DN 端口分配：在默认分布式示例中，DN 实例 dn1_1 为 15432，dn1_2 为 15433，dn1_3 为 15433，dn2 实例为 20010。这些值属于各自的配置示例。在不同部署布局之间切换时，不要盲目复用端口分配。务必验证端口可用性以及每台主机的实际配置要求。

3. Compare the Three GBase Database Deployment Layouts The component counts make the architectural differences easier to understand. Component Minimal Single HA Group Default Distributed gha_server 1 2 1 coordinator (CN) 1 3 2 DN groups 1 1 2 DN instances per group 1 3 3 The pattern is clear: Minimal: One DN group with one instance, providing the smallest deployment configuration. Single HA Group: Adds DN replicas and more coordinator instances. Default Distributed: Expands the DN layer to two groups, each with three instances, while using different CN and gha_server counts. These are architectural differences, not simply different numbers of machines.

3. 比较三种 GBase Database 部署布局 组件数量使架构差异更容易理解。组件包括：最小化布局（gha_server: 1, CN: 1, DN 组: 1, 每组 DN 实例: 1）、单高可用组布局（gha_server: 2, CN: 3, DN 组: 1, 每组 DN 实例: 3）、默认分布式布局（gha_server: 1, CN: 2, DN 组: 2, 每组 DN 实例: 3）。模式很明确：最小化布局提供最小的部署配置；单高可用组布局增加了 DN 副本和更多的协调器实例；默认分布式布局将 DN 层扩展为两组，每组三个实例，同时使用不同的 CN 和 gha_server 数量。这些是架构上的差异，而不仅仅是机器数量的不同。

4. Why Copying Between Layouts Is Risky A common deployment shortcut is to take an existing gbase.yml, replace IP addresses and hostnames, and adjust component counts. The problem is that changing counts does not automatically make the field structure correct. For example: Keeping az or proxy_port when adapting a configuration to a layout whose official example does not include them. Removing fields without understanding their architectural purpose. Reusing GTM or DN port assignments from another deployment. Changing the number of DN instances without verifying the intended replica structure. Missing fields can alter the declared topology. Extra fields should not be assumed to be harmless or automatically ignored. GBase Database executes the configuration that is declared. Engineers remain responsible for ensuring that the configuration matches the intended deployment architecture.

4. 为什么在布局之间复制配置是有风险的 一种常见的部署捷径是直接拿现有的 gbase.yml，替换 IP 地址和主机名，并调整组件数量。问题在于，更改数量并不能自动使字段结构变得正确。例如：在适配到官方示例不包含 az 或 proxy_port 的布局时保留这些字段；在不了解架构目的的情况下删除字段；复用其他部署的 GTM 或 DN 端口分配；在未验证预期副本结构的情况下更改 DN 实例数量。缺失的字段可能会改变声明的拓扑结构。不应假设多余的字段是无害的或会被自动忽略。GBase Database 会执行所声明的配置。工程师仍需负责确保配置与预期的部署架构相匹配。

5. The Better Approach: Start with the Correct Configuration Example For GBase Database(GBase 8c), use this workflow: Identify the target layout: minimal, single-HA-group, or default distributed. Select the corresponding official gbase.yml example. Replace IP addresses and hostnames section by section. Update installation directories according to the target environment. Verify component counts, field presence, and port assignments. Review the final configuration before running gha_ctl install. Avoid treating one deployment example as a universal template and adapting it by adding or deleting fields. The three examples are designed for different scenarios. Each should be treated as a complete baseline for its intended layout.

5. 更好的方法：从正确的配置示例开始 对于 GBase Database (GBase 8c)，请使用以下工作流程：确定目标布局（最小化、单高可用组或默认分布式）；选择相应的官方 gbase.yml 示例；分段替换 IP 地址和主机名；根据目标环境更新安装目录；验证组件数量、字段存在性和端口分配；在运行 gha_ctl install 之前审查最终配置。避免将一个部署示例视为通用模板并通过添加或删除字段来调整它。这三个示例是为不同场景设计的，每个示例都应被视为其预期布局的完整基准。

6. Three Templates, One Simple Rule The most important lesson is that GBase Database deployment configuration is layout-dependent. The minimal layout, single-HA-group layout, and default distributed layout share a common structure, but their field sets and component arrangements are not interchangeable. Remember three things: Field presence can differ between deployment layouts. Port assignments and component counts are layout-specific. The official example for the target architecture should always be the starting point. Three deployment layouts, three configuration baselines. Pick the right one before changing a single IP address.

6. 三个模板，一个简单的规则 最重要的一点是，GBase Database 的部署配置取决于布局。最小化布局、单高可用组布局和默认分布式布局共享通用结构，但它们的字段集和组件排列是不可互换的。请记住三件事：不同部署布局之间的字段存在性可能不同；端口分配和组件数量是特定于布局的；目标架构的官方示例应始终作为起点。三种部署布局，三个配置基准。在更改任何 IP 地址之前，请先选择正确的布局。