Files
lmxopcua/docs/GalaxyRepository.md
Joseph Doherty 3b2defd94f Phase 0 — mechanical rename ZB.MOM.WW.LmxOpcUa.* → ZB.MOM.WW.OtOpcUa.*
Renames all 11 projects (5 src + 6 tests), the .slnx solution file, all source-file namespaces, all axaml namespace references, and all v1 documentation references in CLAUDE.md and docs/*.md (excluding docs/v2/ which is already in OtOpcUa form). Also updates the TopShelf service registration name from "LmxOpcUa" to "OtOpcUa" per Phase 0 Task 0.6.

Preserves runtime identifiers per Phase 0 Out-of-Scope rules to avoid breaking v1/v2 client trust during coexistence: OPC UA `ApplicationUri` defaults (`urn:{GalaxyName}:LmxOpcUa`), server `EndpointPath` (`/LmxOpcUa`), `ServerName` default (feeds cert subject CN), `MxAccessConfiguration.ClientName` default (defensive — stays "LmxOpcUa" for MxAccess audit-trail consistency), client OPC UA identifiers (`ApplicationName = "LmxOpcUaClient"`, `ApplicationUri = "urn:localhost:LmxOpcUaClient"`, cert directory `%LocalAppData%\LmxOpcUaClient\pki\`), and the `LmxOpcUaServer` class name (class rename out of Phase 0 scope per Task 0.5 sed pattern; happens in Phase 1 alongside `LmxNodeManager → GenericDriverNodeManager` Core extraction). 23 LmxOpcUa references retained, all enumerated and justified in `docs/v2/implementation/exit-gate-phase-0.md`.

Build clean: 0 errors, 30 warnings (lower than baseline 167). Tests at strict improvement over baseline: 821 passing / 1 failing vs baseline 820 / 2 (one flaky pre-existing failure passed this run; the other still fails — both pre-existing and unrelated to the rename). `Client.UI.Tests`, `Historian.Aveva.Tests`, `Client.Shared.Tests`, `IntegrationTests` all match baseline exactly. Exit gate compliance results recorded in `docs/v2/implementation/exit-gate-phase-0.md` with all 7 checks PASS or DEFERRED-to-PR-review (#7 service install verification needs Windows service permissions on the reviewer's box).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-17 13:57:47 -04:00

144 lines
11 KiB
Markdown

# Galaxy Repository
`GalaxyRepositoryService` reads the Galaxy object hierarchy and attribute metadata from the System Platform Galaxy Repository SQL Server database. This data drives the construction of the OPC UA address space.
## Connection Configuration
`GalaxyRepositoryConfiguration` controls database access:
| Property | Default | Description |
|----------|---------|-------------|
| `ConnectionString` | `Server=localhost;Database=ZB;Integrated Security=true;` | SQL Server connection using Windows Authentication |
| `ChangeDetectionIntervalSeconds` | `30` | Polling frequency for deploy change detection |
| `CommandTimeoutSeconds` | `30` | SQL command timeout for all queries |
| `ExtendedAttributes` | `false` | When true, loads primitive-level attributes in addition to dynamic attributes |
| `Scope` | `Galaxy` | `Galaxy` loads all deployed objects. `LocalPlatform` filters to the local platform's subtree only |
| `PlatformName` | `null` | Explicit platform hostname for `LocalPlatform` filtering. When null, uses `Environment.MachineName` |
The connection uses Windows Authentication because the Galaxy Repository database is local to the System Platform node and secured through domain credentials.
## SQL Queries
All queries are embedded as `const string` fields in `GalaxyRepositoryService`. No dynamic SQL is used.
### Hierarchy query
Returns deployed Galaxy objects with their parent relationships, browse names, and template derivation chains:
- Joins `gobject` to `template_definition` to filter by relevant `category_id` values (1, 3, 4, 10, 11, 13, 17, 24, 26)
- Uses `contained_name` as the browse name, falling back to `tag_name` when `contained_name` is null or empty
- Resolves the parent using `contained_by_gobject_id` when non-zero, otherwise falls back to `area_gobject_id`
- Marks objects with `category_id = 13` as areas
- Filters to `is_template = 0` (instances only, not templates)
- Filters to `deployed_package_id <> 0` (deployed objects only)
- Returns a `template_chain` column built by a recursive CTE that walks `gobject.derived_from_gobject_id` from each instance through its immediate template and ancestor templates (depth guard `< 10`). Template names are ordered by depth and joined with `|` via `STUFF(... FOR XML PATH(''))`. Example: `TestMachine_001` returns `$TestMachine|$gMachine|$gUserDefined|$UserDefined`. The C# repository reader splits the column on `|`, trims, and populates `GalaxyObjectInfo.TemplateChain`, which is consumed by `AlarmObjectFilter` for template-based alarm filtering. See [Alarm Tracking](AlarmTracking.md#template-based-alarm-object-filter).
- Returns `template_definition.category_id` as a `category_id` column, populated into `GalaxyObjectInfo.CategoryId`. The runtime status probe manager filters this down to `CategoryId == 1` (`$WinPlatform`) and `CategoryId == 3` (`$AppEngine`) to decide which objects get a `<Host>.ScanState` probe advised. Also used by `LmxNodeManager.BuildHostedVariablesMap` to identify Platform/Engine ancestors during the hosted-variables walk.
- Returns `gobject.hosted_by_gobject_id` as a `hosted_by_gobject_id` column, populated into `GalaxyObjectInfo.HostedByGobjectId`. This is the **runtime host** of the object (e.g., which `$AppEngine` actually runs it), **not** the browse-containment parent (`contained_by_gobject_id`). The two are often different — an object can live in one Area in the browse tree but be hosted by an Engine on a different Platform for runtime execution. The node manager walks this chain during `BuildHostedVariablesMap` to find the nearest `$WinPlatform` or `$AppEngine` ancestor so subtree quality invalidation on a Stopped host reaches exactly the variables that were actually executing there. Note: the Galaxy schema column is named `hosted_by_gobject_id` (not `host_gobject_id` as some documentation sources guess). See [MXAccess Bridge — Per-Host Runtime Status Probes](MxAccessBridge.md#per-host-runtime-status-probes-hostscanstate).
### Attributes query (standard)
Returns user-defined dynamic attributes for deployed objects:
- Uses a recursive CTE (`deployed_package_chain`) to walk the package inheritance chain from `deployed_package_id` through `derived_from_package_id`, limited to 10 levels
- Joins `dynamic_attribute` on each package in the chain to collect inherited attributes
- Uses `ROW_NUMBER() OVER (PARTITION BY gobject_id, attribute_name ORDER BY depth)` to pick the most-derived definition when an attribute is overridden at multiple levels
- Builds `full_tag_reference` as `tag_name.attribute_name` with `[]` appended for arrays
- Extracts `array_dimension` from the binary `mx_value` column (bytes 13-16, little-endian int32)
- Detects historized attributes by checking for a `HistoryExtension` primitive instance
- Detects alarm attributes by checking for an `AlarmExtension` primitive instance
- Excludes internal attributes (names starting with `_`) and `.Description` suffixes
- Filters by `mx_attribute_category` to include only user-relevant categories
### Attributes query (extended)
When `ExtendedAttributes = true`, a more comprehensive query runs that unions two sources:
1. **Primitive attributes** -- Joins through `primitive_instance` and `attribute_definition` to include system-level attributes from primitive components. Each attribute carries its `primitive_name` so the address space can group them under their parent variable.
2. **Dynamic attributes** -- The same CTE-based query as the standard path, with an empty `primitive_name`.
The `full_tag_reference` for primitive attributes follows the pattern `tag_name.primitive_name.attribute_name` (e.g., `TestMachine_001.AlarmAttr.InAlarm`).
### Change detection query
A single-column query: `SELECT time_of_last_deploy FROM galaxy`. The `galaxy` table contains one row with the timestamp of the most recent deployment.
## Why deployed_package_id Instead of checked_in_package_id
The Galaxy maintains two package references for each object:
- `checked_in_package_id` -- The latest saved version, which may include undeployed configuration changes
- `deployed_package_id` -- The version currently running on the target platform
The queries filter on `deployed_package_id <> 0` because the OPC UA server must mirror what is actually running in the Galaxy runtime. Using `checked_in_package_id` would expose attributes and objects that exist in the IDE but have not been deployed, causing mismatches between the OPC UA address space and the MXAccess runtime.
## Platform Scope Filter
When `Scope` is set to `LocalPlatform`, the repository applies a post-query C# filter to restrict the address space to objects hosted by the local platform. This reduces memory footprint, MXAccess subscription count, and address space size on multi-node Galaxy deployments where each OPC UA server instance only needs to serve its own platform's objects.
### How it works
1. **Platform lookup** -- A separate `const string` SQL query (`PlatformLookupSql`) reads `platform_gobject_id` and `node_name` from the `platform` table for all deployed platforms. This runs once per hierarchy load.
2. **Platform matching** -- The configured `PlatformName` (or `Environment.MachineName` when null) is matched case-insensitively against the `node_name` column. If no match is found, a warning is logged listing the available platforms, and the address space is empty.
3. **Host chain collection** -- The filter collects the matching platform's `gobject_id`, then iterates the hierarchy to find all `$AppEngine` (category 3) objects whose `HostedByGobjectId` equals the platform. This produces the full set of host gobject_ids under the local platform.
4. **Object inclusion** -- All non-area objects whose `HostedByGobjectId` is in the host set are included, along with the hosts themselves.
5. **Area retention** -- `ParentGobjectId` chains are walked upward from included objects to pull in ancestor areas, keeping the browse tree connected. Areas that contain no local descendants are excluded.
6. **Attribute filtering** -- The set of included `gobject_id` values is cached after `GetHierarchyAsync` and reused by `GetAttributesAsync` to filter attributes to the same scope.
### Design rationale
The filter is applied in C# rather than SQL because the project convention `GR-006` requires `const string` SQL queries with no dynamic SQL. The hierarchy query already returns `HostedByGobjectId` and `CategoryId` on every row, so all information needed for filtering is already in memory after the query runs. The only new SQL is the lightweight platform lookup query.
### Configuration
```json
"GalaxyRepository": {
"Scope": "LocalPlatform",
"PlatformName": null
}
```
- Set `Scope` to `"LocalPlatform"` to enable filtering. Default is `"Galaxy"` (load everything, backward compatible).
- Set `PlatformName` to an explicit hostname to target a specific platform, or leave null to use the local machine name.
### Startup log
When `LocalPlatform` is active, the startup log shows the filtering result:
```
GalaxyRepository.Scope="LocalPlatform", PlatformName=MYNODE
GetHierarchyAsync returned 49 objects
GetPlatformsAsync returned 2 platform(s)
Scope filter targeting platform 'MYNODE' (gobject_id=1042)
Scope filter retained 25 of 49 objects for platform 'MYNODE'
GetAttributesAsync returned 4206 attributes (extended=true)
Scope filter retained 2100 of 4206 attributes
```
## Change Detection Polling
`ChangeDetectionService` runs a background polling loop that calls `GetLastDeployTimeAsync` at the configured interval. It compares the returned timestamp against the last known value:
- On the first poll (no previous state), the timestamp is recorded and `OnGalaxyChanged` fires unconditionally
- On subsequent polls, `OnGalaxyChanged` fires only when `time_of_last_deploy` differs from the cached value
When the event fires, the host service queries fresh hierarchy and attribute data from the repository and calls `LmxNodeManager.RebuildAddressSpace` (which delegates to incremental `SyncAddressSpace`).
The polling approach is used because the Galaxy Repository database does not provide change notifications. The `galaxy.time_of_last_deploy` column updates only on completed deployments, so the polling interval controls how quickly the OPC UA address space reflects Galaxy changes.
## TestConnection
`TestConnectionAsync` runs `SELECT 1` against the configured database. This is used at service startup to verify connectivity before attempting the full hierarchy query.
## Key source files
- `src/ZB.MOM.WW.OtOpcUa.Host/GalaxyRepository/GalaxyRepositoryService.cs` -- SQL queries and data access
- `src/ZB.MOM.WW.OtOpcUa.Host/GalaxyRepository/PlatformScopeFilter.cs` -- Platform-based hierarchy and attribute filtering
- `src/ZB.MOM.WW.OtOpcUa.Host/GalaxyRepository/ChangeDetectionService.cs` -- Deploy timestamp polling loop
- `src/ZB.MOM.WW.OtOpcUa.Host/Configuration/GalaxyRepositoryConfiguration.cs` -- Connection, polling, and scope settings
- `src/ZB.MOM.WW.OtOpcUa.Host/Domain/PlatformInfo.cs` -- Platform-to-hostname DTO