[internal] Add sds-common-scheduler-extender v2

[krpsh123]

Description

New API Endpoint: /api/v1/volumes/filter-prioritize

Overview

Direct API endpoint for filtering and scoring LVMVolumeGroups based on available space. Designed for external consumers like CSI drivers that need to select the best LVG for volume placement.

Endpoint

Path	Method	Description
/api/v1/volumes/filter-prioritize	POST	Filter LVGs by available space and return scored results

Request

Content-Type

application/json

Request Body

{
  "lvgs": [
    {
      "name": "lvg-worker-1",
      "thinPoolName": "thin-pool-1"  // Required only for thin volumes
    },
    {
      "name": "lvg-worker-2",
      "thinPoolName": "thin-pool-1"
    }
  ],
  "volume": {
    "name": "pvc-abc123",
    "size": 10737418240,  // Size in bytes (10 Gi)
    "type": "thick"       // "thick" or "thin"
  }
}

Request Fields

Field	Type	Required	Description
lvgs	array	Yes	List of candidate LVMVolumeGroups
lvgs[].name	string	Yes	LVMVolumeGroup resource name
lvgs[].thinPoolName	string	For thin	Thin pool name (required when volume.type = "thin")
volume.name	string	Yes	Volume name (used for space reservation in cache)
volume.size	int64	Yes	Requested volume size in bytes
volume.type	string	Yes	Volume type: "thick" or "thin"

Response

Success Response (200 OK)

{
  "lvgs": [
    {
      "name": "lvg-worker-1",
      "score": 8
    },
    {
      "name": "lvg-worker-2", 
      "score": 5
    }
  ]
}

Response Fields

Field	Type	Description
lvgs	array	Filtered and scored LVGs, sorted by score (highest first)
lvgs[].name	string	LVMVolumeGroup name
lvgs[].score	int	Score from 1-10 based on available space
error	string	Error message (only present on error)

Empty Result

If no LVGs have sufficient space:

{
  "lvgs": []
}

Behavior

1. Filtering: LVGs without enough available space are excluded
2. Scoring: Remaining LVGs are scored 1-10 based on free space percentage
3. Reservation: Space is reserved in cache for all filtered LVGs (to prevent over-allocation during concurrent requests)
4. Sorting: Results are sorted by score (highest to lowest)

Error Responses

Status	Condition
400 Bad Request	Empty lvgs list
400 Bad Request	Invalid volume data (empty name or size ≤ 0)
400 Bad Request	Invalid volume type
400 Bad Request	Missing thinPoolName for thin volumes
405 Method Not Allowed	Non-POST request
500 Internal Server Error	Internal processing error

Example Usage

Thick Volume Request

curl -X POST https://scheduler-extender:8000/api/v1/volumes/filter-prioritize \
  -H "Content-Type: application/json" \
  -d '{
    "lvgs": [
      {"name": "lvg-worker-1"},
      {"name": "lvg-worker-2"}
    ],
    "volume": {
      "name": "pvc-12345",
      "size": 5368709120,
      "type": "thick"
    }
  }'

Thin Volume Request

curl -X POST https://scheduler-extender:8000/api/v1/volumes/filter-prioritize \
  -H "Content-Type: application/json" \
  -d '{
    "lvgs": [
      {"name": "lvg-worker-1", "thinPoolName": "data-pool"},
      {"name": "lvg-worker-2", "thinPoolName": "data-pool"}
    ],
    "volume": {
      "name": "pvc-67890",
      "size": 10737418240,
      "type": "thin"
    }
  }'

Notes

• LVG must exist in scheduler-extender's cache (populated by LVG watcher controller)
• Space reservation expires after configured timeout (default: 30 seconds)
• Score calculation uses logarithmic scale based on free space percentage

Why do we need it, and what problem does it solve?

What is the expected result?

Checklist

• The code is covered by unit tests.
• e2e tests passed.
• Documentation updated according to the changes.
• Changes were tested in the Kubernetes cluster manually.