Tạo Snapshot và Backup với OpenStack

Giới thiệu

Trong bài này, chúng ta sẽ tìm hiểu cách tạo snapshot và backup với OpenStack Nova và Glance APIs, bao gồm polling image status và metadata management.

Snapshot vs Backup trong OpenStack

Snapshot (createImage)

Nova Action: createImage

Tạo image từ server hiện tại
Không có rotation
Metadata: source: 'portal-manual'
Dùng cho: Manual snapshot từ user

Backup (createBackup)

Nova Action: createBackup

Tạo backup với rotation support
Tự động xóa backup cũ dựa trên rotation count
Metadata: source: 'portal-auto'
Dùng cho: Scheduled backup tự động

Flow Tạo Snapshot

Step-by-Step Flow

1. Get scoped token (Keystone)
   ↓
2. Get server details (Nova GET /servers/{id})
   ↓
3. Trigger snapshot (Nova POST /servers/{id}/action)
   Body: { createImage: { name, metadata } }
   ↓
4. Extract image ID from Location header
   ↓
5. Poll image status (Glance GET /v2/images/{id})
   - Status: queued → saving → active
   - Timeout: 30 phút
   - Interval: 5 giây
   ↓
6. Update metadata (Glance PATCH /v2/images/{id})
   - Add OS metadata (os_distro, os_version, os_type)
   ↓
7. Return image ID và details

Implementation

Create Server Snapshot

typescript

export const createServerSnapshotOpenStack = async function (params: {
  serverId: string;
  snapshotName: string;
  location?: string;
  osImageFromInet?: string;
}) {
  const { serverId, snapshotName, location, osImageFromInet } = params;

  // 1. Get scoped token
  const { token } = await getScopedToken(location);

  // 2. Build endpoints
  const computeBase = buildEndpoint('compute', location);
  const imageBase = buildEndpoint('image', location);

  // 3. Get server details
  const nova = createAxios(computeBase, { 'X-Auth-Token': token });
  const serverRes = await nova.get(`/servers/${encodeURIComponent(serverId)}`);
  const server: any = serverRes.data?.server || serverRes.data;

  // 4. Prepare OS metadata
  let osMetadata: any = {};
  if (osImageFromInet) {
    osMetadata = {
      os_distro: osImageFromInet, // e.g., "Ubuntu 20.04"
    };
  }

  // 5. Trigger snapshot via Nova
  const actionUrl = `/servers/${encodeURIComponent(serverId)}/action`;
  const actionBody = {
    createImage: {
      name: snapshotName,
      metadata: {
        source: 'portal-manual', // Manual snapshot
        ...osMetadata,
      },
    },
  };

  const actionRes = await nova.post(actionUrl, actionBody, {
    validateStatus: (s) => s === 202
  });

  // 6. Extract image ID from Location header
  const locationHeader = String(
    actionRes.headers['location'] ||
    actionRes.headers['Location'] ||
    ''
  );
  let imageId = '';
  if (locationHeader) {
    const match = locationHeader.match(/\/v2\/.+\/images\/([^/?#]+)/) ||
                  locationHeader.match(/images\/?([^/?#]+)/);
    imageId = match ? match[1] : '';
  }

  // 7. Poll image status via Glance
  const glance = createAxios(imageBase, { 'X-Auth-Token': token });
  const startedAt = Date.now();
  const timeoutMs = 30 * 60 * 1000; // 30 minutes
  const intervalMs = 5000; // 5 seconds
  let status = 'queued';
  let details: any = null;

  while (Date.now() - startedAt < timeoutMs) {
    try {
      if (imageId) {
        const r = await glance.get(`/v2/images/${encodeURIComponent(imageId)}`);
        details = r.data;
        status = String(details.status || details.state || '').toLowerCase();
      } else {
        // Fallback: search by name
        const r = await glance.get('/v2/images', {
          params: { name: snapshotName }
        });
        const items: any[] = (r.data?.images || [])
          .filter((i: any) => i.name === snapshotName);
        if (items.length > 0) {
          details = items[0];
          imageId = String(details.id || details.image_id || '');
          status = String(details.status || '').toLowerCase();
        }
      }

      logger.info(
        `OpenStack snapshot '${snapshotName}' status: ${status}` +
        `${imageId ? ` (id: ${imageId})` : ''}`
      );

      if (status === 'active') break;

      if (
        status === 'killed' ||
        status === 'deleted' ||
        status === 'error' ||
        status === 'failed'
      ) {
        throw new Error(`Snapshot failed with status: ${status}`);
      }
    } catch (e: any) {
      logger.warn(`Polling image status failed: ${e?.message || e}`);
    }

    await new Promise((r) => setTimeout(r, intervalMs));
  }

  if (status !== 'active') {
    throw new Error(
      `Timeout waiting for snapshot to become active. ` +
      `Last status: ${status}`
    );
  }

  // 8. Update metadata in Glance (optional but recommended)
  if (imageId && Object.keys(osMetadata).length > 0) {
    try {
      const patchOps = [];
      if (osMetadata.os_distro) {
        patchOps.push({
          op: 'add',
          path: '/os_distro',
          value: osMetadata.os_distro
        });
      }
      // ... other metadata fields

      if (patchOps.length > 0) {
        await glance.patch(
          `/v2/images/${encodeURIComponent(imageId)}`,
          patchOps,
          {
            headers: {
              'Content-Type': 'application/openstack-images-v2.1-json-patch'
            }
          }
        );
      }
    } catch (e: any) {
      logger.error(`Could not update snapshot metadata: ${e?.message}`);
      // Don't throw - metadata update is optional
    }
  }

  return { imageId, imageName: snapshotName, details };
};

Create Server Backup

Backup với Rotation

typescript

export const createServerBackupOpenStack = async function (params: {
  serverId: string;
  backupName: string;
  backupType?: string;
  rotation?: number;
  location?: string;
  osImageFromInet?: string;
}) {
  const { serverId, backupName, backupType = 'daily', rotation, location, osImageFromInet } = params;

  // 1. Get scoped token
  const { token } = await getScopedToken(location);

  // 2. Build endpoints
  const computeBase = buildEndpoint('compute', location);
  const imageBase = buildEndpoint('image', location);

  // 3. Prepare OS metadata
  let osMetadata: any = {};
  if (osImageFromInet) {
    osMetadata = {
      os_distro: osImageFromInet,
    };
  }

  // 4. Trigger backup via Nova
  const nova = createAxios(computeBase, { 'X-Auth-Token': token });
  const actionUrl = `/servers/${encodeURIComponent(serverId)}/action`;
  const actionBody = {
    createBackup: {
      name: backupName,
      backup_type: backupType,
      rotation: rotation,
      metadata: {
        source: 'portal-auto', // Scheduled backup
        ...osMetadata,
      },
    },
  };

  const actionRes = await nova.post(actionUrl, actionBody, {
    validateStatus: (s) => s === 202
  });

  // 5. Extract image ID và poll status (tương tự snapshot)
  // ... (same polling logic as snapshot)

  return { imageId, imageName: backupName, details };
};

Polling Image Status

Image Status States

Status	Description
`queued`	Image creation queued
`saving`	Image is being saved
`active`	Image is ready ✅
`killed`	Image creation killed ❌
`deleted`	Image deleted ❌
`error`	Error occurred ❌
`failed`	Creation failed ❌

Polling Pattern

typescript

const startedAt = Date.now();
const timeoutMs = 30 * 60 * 1000; // 30 minutes
const intervalMs = 5000; // 5 seconds

while (Date.now() - startedAt < timeoutMs) {
  const image = await glance.get(`/v2/images/${imageId}`);
  const status = image.status?.toLowerCase();

  // Success case
  if (status === "active") {
    break; // Image ready
  }

  // Error cases
  if (["killed", "deleted", "error", "failed"].includes(status)) {
    throw new Error(`Image creation failed: ${status}`);
  }

  // Continue polling
  await new Promise((r) => setTimeout(r, intervalMs));
}

// Check timeout
if (status !== "active") {
  throw new Error(`Timeout waiting for image. Last status: ${status}`);
}

Best Practices

Timeout hợp lý: 30 phút cho large servers
Interval không quá ngắn: 5 giây để tránh spam API
Fallback search: Nếu không có imageId từ Location header
Error handling: Catch và log polling errors, không throw ngay

Metadata Management

Snapshot Metadata Structure

typescript

{
  source: 'portal-manual' | 'portal-auto',
  os_distro: 'Ubuntu 20.04',
  os_version: '20.04',
  os_type: 'linux'
}

Update Metadata trong Glance

typescript

const patchOps = [
  { op: "add", path: "/os_distro", value: "Ubuntu 20.04" },
  { op: "add", path: "/os_version", value: "20.04" },
  { op: "add", path: "/os_type", value: "linux" },
];

await glance.patch(`/v2/images/${imageId}`, patchOps, {
  headers: {
    "Content-Type": "application/openstack-images-v2.1-json-patch",
  },
});

Lý do update metadata:

Nova metadata có thể không được lưu đầy đủ
Glance metadata là source of truth
Dễ dàng query và filter sau này

Error Handling

Common Errors

1. Server Not Found

typescript

try {
  const serverRes = await nova.get(`/servers/${serverId}`);
} catch (error: any) {
  if (error.response?.status === 404) {
    throw new Error(`Server not found: ${serverId}`);
  }
  throw error;
}

2. Image Creation Failed

typescript

if (["killed", "deleted", "error", "failed"].includes(status)) {
  throw new Error(`Snapshot failed with status: ${status}`);
}

3. Timeout

typescript

if (status !== "active") {
  throw new Error(
    `Timeout waiting for snapshot to become active. ` + `Last status: ${status}`
  );
}

4. Missing Image ID

typescript

if (!imageId) {
  // Fallback: search by name
  const r = await glance.get("/v2/images", {
    params: { name: snapshotName },
  });
  // ...
}

Testing

Unit Test Example

typescript

describe('createServerSnapshotOpenStack', () => {
  it('should create snapshot successfully', async () => {
    const result = await createServerSnapshotOpenStack({
      serverId: 'test-server-id',
      snapshotName: 'test-snapshot',
      location: 'HCM',
      osImageFromInet: 'Ubuntu 20.04'
    });

    expect(result.imageId).toBeDefined();
    expect(result.imageName).toBe('test-snapshot');
  });

  it('should handle timeout', async () => {
    // Mock long-running image creation
    // ...
    await expect(
      createServerSnapshotOpenStack({...})
    ).rejects.toThrow('Timeout');
  });
});

Summary

Key Points

Snapshot vs Backup
- Snapshot: createImage, manual, no rotation
- Backup: createBackup, scheduled, with rotation
Polling Pattern
- Poll every 5 seconds
- Timeout after 30 minutes
- Handle all status states
Metadata Management
- Store source (portal-manual vs portal-auto)
- Store OS info (os_distro, os_version)
- Update via Glance PATCH API
Error Handling
- Check server exists
- Handle image creation failures
- Handle timeout
- Fallback search by name

Next Steps

Trong bài tiếp theo, chúng ta sẽ tìm hiểu về:

Schedule System - Cron-based backup scheduler

Last Updated: 2025-01-25
Previous: 02. Authentication
Next: 04. Schedule System

Tạo Snapshot và Backup với OpenStack ​

Giới thiệu ​

Snapshot vs Backup trong OpenStack ​

Snapshot (createImage) ​

Backup (createBackup) ​

Flow Tạo Snapshot ​

Step-by-Step Flow ​

Implementation ​

Create Server Snapshot ​

Create Server Backup ​

Backup với Rotation ​

Polling Image Status ​

Image Status States ​

Polling Pattern ​

Best Practices ​

Metadata Management ​

Snapshot Metadata Structure ​

Update Metadata trong Glance ​

Error Handling ​

Common Errors ​

1. Server Not Found ​

2. Image Creation Failed ​

3. Timeout ​

4. Missing Image ID ​

Testing ​

Unit Test Example ​

Summary ​

Key Points ​

Next Steps ​

Tạo Snapshot và Backup với OpenStack

Giới thiệu

Snapshot vs Backup trong OpenStack

Snapshot (createImage)

Backup (createBackup)

Flow Tạo Snapshot

Step-by-Step Flow

Implementation

Create Server Snapshot

Create Server Backup

Backup với Rotation

Polling Image Status

Image Status States

Polling Pattern

Best Practices

Metadata Management

Snapshot Metadata Structure

Update Metadata trong Glance

Error Handling

Common Errors

1. Server Not Found

2. Image Creation Failed

3. Timeout

4. Missing Image ID

Testing

Unit Test Example

Summary

Key Points

Next Steps