UBERF-9543: Restore-all tool (#8132)

Signed-off-by: Andrey Sobolev <haiodo@gmail.com>

.vscode/launch.json

@@ -497,6 +497,26 @@
       "cwd": "${workspaceRoot}/dev/tool",
       "protocol": "inspector"
     },
+    {
+      "name": "Debug restore-all tool",
+      "type": "node",
+      "request": "launch",
+      "args": ["src/__start.ts", "restore-all"],
+      "env": {
+        "ACCOUNTS_URL": "http://localhost:3000",
+        "STORAGE": "minio|localhost?accessKey=minioadmin&secretKey=minioadmin",
+        // "DB_URL": "mongodb://localhost:27017",
+        "DB_URL": "postgresql://root@host.docker.internal:26258/defaultdb?sslmode=disable",
+        "SERVER_SECRET": "secret",
+        "BUCKET_NAME":"dev-backups"
+      },
+      "runtimeArgs": ["--nolazy", "-r", "ts-node/register"],
+      "showAsyncStacks": true,
+      "sourceMaps": true,
+      "outputCapture": "std",
+      "cwd": "${workspaceRoot}/dev/tool",
+      "protocol": "inspector"
+    },
     {
       "name": "Debug Github integration",
       "type": "node",

dev/tool/src/index.ts

@@ -78,10 +78,11 @@ import { buildStorageFromConfig, createStorageFromConfig, storageConfigFromEnv }
 import { program, type Command } from 'commander'
 import { addControlledDocumentRank } from './qms'
 import { clearTelegramHistory } from './telegram'
-import { diffWorkspace, updateField } from './workspace'
+import { backupRestore, diffWorkspace, updateField } from './workspace'
 
 import core, {
   AccountRole,
+  DOMAIN_BLOB,
   generateId,
   getWorkspaceId,
   isActiveMode,
@@ -697,6 +698,70 @@ export function devTool (
       }
     )
 
+  program
+    .command('restore-all')
+    .description('Restore workspaces to selected region DB...')
+    .option('-t|--timeout [timeout]', 'Timeout in days', '60')
+    .option('-r|--region [region]', 'Timeout in days', '')
+    .option('-w|--workspace [workspace]', 'Force backup of selected workspace', '')
+    .option('-d|--dry [dry]', 'Dry run', false)
+    .action(async (cmd: { timeout: string, workspace: string, region: string, dry: boolean, account: string }) => {
+      const { txes, dbUrl } = prepareTools()
+
+      const bucketName = process.env.BUCKET_NAME
+      if (bucketName === '' || bucketName == null) {
+        console.error('please provide butket name env')
+        process.exit(1)
+      }
+
+      const token = generateToken(systemAccountEmail, getWorkspaceId(''))
+      const workspaces = (await listAccountWorkspaces(token, cmd.region))
+        .sort((a, b) => {
+          const bsize = b.backupInfo?.backupSize ?? 0
+          const asize = a.backupInfo?.backupSize ?? 0
+          return bsize - asize
+        })
+        .filter((it) => cmd.workspace === '' || cmd.workspace === it.workspace)
+
+      for (const ws of workspaces) {
+        const lastVisitDays = Math.floor((Date.now() - ws.lastVisit) / 1000 / 3600 / 24)
+
+        toolCtx.warn('--- restoring workspace', {
+          url: ws.workspaceUrl,
+          id: ws.workspace,
+          lastVisitDays,
+          backupSize: ws.backupInfo?.blobsSize ?? 0,
+          mode: ws.mode
+        })
+        if (cmd.dry) {
+          continue
+        }
+        try {
+          const st = Date.now()
+          await backupRestore(
+            toolCtx,
+            dbUrl,
+            bucketName,
+            ws,
+            (dbUrl, storageAdapter) => {
+              const factory: PipelineFactory = createBackupPipeline(toolCtx, dbUrl, txes, {
+                externalStorage: storageAdapter,
+                usePassedCtx: true
+              })
+              return factory
+            },
+            [DOMAIN_BLOB]
+          )
+          const ed = Date.now()
+          toolCtx.warn('--- restoring complete', {
+            time: ed - st
+          })
+        } catch (err: any) {
+          toolCtx.error('REstore of f workspace failedarchive workspace', { workspace: ws.workspace })
+        }
+      }
+    })
+
   program
     .command('backup-all')
     .description('Backup all workspaces...')

dev/tool/src/workspace.ts

@@ -16,16 +16,28 @@
 
 import contact from '@hcengineering/contact'
 import core, {
+  DOMAIN_TX,
+  getWorkspaceId,
   type BackupClient,
+  type BaseWorkspaceInfo,
   type Class,
   type Client as CoreClient,
   type Doc,
-  DOMAIN_TX,
+  type MeasureContext,
   type Ref,
   type Tx,
-  type WorkspaceId
+  type WorkspaceId,
+  type WorkspaceIdWithUrl
 } from '@hcengineering/core'
 import { getMongoClient, getWorkspaceMongoDB } from '@hcengineering/mongo'
+import { createStorageBackupStorage, restore } from '@hcengineering/server-backup'
+import {
+  createDummyStorageAdapter,
+  wrapPipeline,
+  type PipelineFactory,
+  type StorageAdapter
+} from '@hcengineering/server-core'
+import { createStorageFromConfig, storageConfigFromEnv } from '@hcengineering/server-storage'
 import { connect } from '@hcengineering/server-tool'
 import { generateModelDiff, printDiff } from './mdiff'
 
@@ -95,3 +107,57 @@ export async function updateField (
     await connection.close()
   }
 }
+
+export async function backupRestore (
+  ctx: MeasureContext,
+  dbURL: string,
+  bucketName: string,
+  workspace: BaseWorkspaceInfo,
+  pipelineFactoryFactory: (mongoUrl: string, storage: StorageAdapter) => PipelineFactory,
+  skipDomains: string[]
+): Promise<boolean> {
+  const storageEnv = process.env.STORAGE
+  if (storageEnv === undefined) {
+    console.error('please provide STORAGE env')
+    process.exit(1)
+  }
+  if (bucketName.trim() === '') {
+    console.error('please provide butket name env')
+    process.exit(1)
+  }
+  const backupStorageConfig = storageConfigFromEnv(storageEnv)
+
+  const storageAdapter = createStorageFromConfig(backupStorageConfig.storages[0])
+
+  const workspaceStorage = createDummyStorageAdapter()
+  const pipelineFactory = pipelineFactoryFactory(dbURL, workspaceStorage)
+
+  try {
+    const storage = await createStorageBackupStorage(
+      ctx,
+      storageAdapter,
+      getWorkspaceId(bucketName),
+      workspace.workspace
+    )
+    const wsUrl: WorkspaceIdWithUrl = {
+      name: workspace.workspace,
+      uuid: workspace.uuid,
+      workspaceName: workspace.workspaceName ?? '',
+      workspaceUrl: workspace.workspaceUrl ?? ''
+    }
+    const result: boolean = await ctx.with('restore', { workspace: workspace.workspace }, (ctx) =>
+      restore(ctx, '', getWorkspaceId(workspace.workspace), storage, {
+        date: -1,
+        skip: new Set(skipDomains),
+        recheck: false,
+        storageAdapter: workspaceStorage,
+        getConnection: async () => {
+          return wrapPipeline(ctx, await pipelineFactory(ctx, wsUrl, true, () => {}, null), wsUrl)
+        }
+      })
+    )
+    return result
+  } finally {
+    await storageAdapter.close()
+  }
+}

server/backup/src/service.ts

@@ -451,7 +451,7 @@ export async function doRestoreWorkspace (
       restore(ctx, '', getWorkspaceId(ws.workspace), storage, {
         date: -1,
         skip: new Set(skipDomains),
-        recheck: true,
+        recheck: false, // Do not need to recheck
         storageAdapter: workspaceStorageAdapter,
         cleanIndexState,
         getConnection: async () => {

server/workspace-service/src/index.ts

@@ -63,6 +63,10 @@ export function serveWorkspaceAccount (
     process.exit(1)
   }
 
+  if (process.env.MIGRATION_CLEANUP !== 'true') {
+    console.log('Migration cleanup is not set, so move to regions will not clean old DB.')
+  }
+
   const backup =
     wsOperation === 'all+backup'
       ? {

server/workspace-service/src/service.ts

@@ -15,6 +15,7 @@
 import {
   type BaseWorkspaceInfo,
   type BrandingMap,
+  DOMAIN_BLOB,
   type Data,
   type MeasureContext,
   type Tx,
@@ -70,8 +71,6 @@ import {
 import { buildStorageFromConfig, storageConfigFromEnv } from '@hcengineering/server-storage'
 import { createWorkspace, upgradeWorkspace } from './ws-operations'
 
-const dbCleanTreshold = 256 // Cleanup workspaces if less 256mb
-
 export interface WorkspaceOptions {
   errorHandler: (workspace: BaseWorkspaceInfo, error: any) => Promise<void>
   force: boolean
@@ -498,8 +497,7 @@ export class WorkspaceWorker {
         await sendEvent('migrate-clean-started', 0)
         await this.sendTransactorMaitenance(token, { name: workspace.workspace })
 
-        const sz = workspace.backupInfo?.backupSize ?? 0
-        if (sz <= dbCleanTreshold) {
+        if (process.env.MIGRATION_CLEANUP === 'true') {
           try {
             await this.doCleanup(ctx, workspace, false)
           } catch (err: any) {
@@ -651,7 +649,7 @@ export class WorkspaceWorker {
         opt.backup.bucketName,
         pipelineFactory,
         workspaceStorageAdapter,
-        ['blob'],
+        [DOMAIN_BLOB],
         true,
         (_p: number) => {
           if (progress !== Math.round(_p)) {