@xen-orchestra/backups 0.36.1 → 0.38.0

package/_runners/_vmRunners/IncrementalXapi.js

@@ -0,0 +1,175 @@
+'use strict'
+
+const findLast = require('lodash/findLast.js')
+const keyBy = require('lodash/keyBy.js')
+const mapValues = require('lodash/mapValues.js')
+const vhdStreamValidator = require('vhd-lib/vhdStreamValidator.js')
+const { asyncMap } = require('@xen-orchestra/async-map')
+const { createLogger } = require('@xen-orchestra/log')
+const { pipeline } = require('node:stream')
+
+const { IncrementalRemoteWriter } = require('../_writers/IncrementalRemoteWriter.js')
+const { IncrementalXapiWriter } = require('../_writers/IncrementalXapiWriter.js')
+const { exportIncrementalVm } = require('../../_incrementalVm.js')
+const { Task } = require('../../Task.js')
+const { watchStreamSize } = require('../../_watchStreamSize.js')
+const { AbstractXapi } = require('./_AbstractXapi.js')
+const { forkDeltaExport } = require('./_forkDeltaExport.js')
+const isVhdDifferencingDisk = require('vhd-lib/isVhdDifferencingDisk')
+const { asyncEach } = require('@vates/async-each')
+
+const { debug } = createLogger('xo:backups:IncrementalXapiVmBackup')
+
+const noop = Function.prototype
+
+exports.IncrementalXapi = class IncrementalXapiVmBackupRunner extends AbstractXapi {
+  _getWriters() {
+    return [IncrementalRemoteWriter, IncrementalXapiWriter]
+  }
+
+  _mustDoSnapshot() {
+    return true
+  }
+
+  async _copy() {
+    const baseVm = this._baseVm
+    const vm = this._vm
+    const exportedVm = this._exportedVm
+    const fullVdisRequired = this._fullVdisRequired
+
+    const isFull = fullVdisRequired === undefined || fullVdisRequired.size !== 0
+
+    await this._callWriters(writer => writer.prepare({ isFull }), 'writer.prepare()')
+
+    const deltaExport = await exportIncrementalVm(exportedVm, baseVm, {
+      fullVdisRequired,
+    })
+    // since NBD is network based, if one disk use nbd , all the disk use them
+    // except the suspended VDI
+    if (Object.values(deltaExport.streams).some(({ _nbd }) => _nbd)) {
+      Task.info('Transfer data using NBD')
+    }
+
+    const differentialVhds = {}
+    // since isVhdDifferencingDisk is reading and unshifting data in stream
+    // it should be done BEFORE any other stream transform
+    await asyncEach(Object.entries(deltaExport.streams), async ([key, stream]) => {
+      differentialVhds[key] = await isVhdDifferencingDisk(stream)
+    })
+    const sizeContainers = mapValues(deltaExport.streams, stream => watchStreamSize(stream))
+
+    if (this._settings.validateVhdStreams) {
+      deltaExport.streams = mapValues(deltaExport.streams, stream => pipeline(stream, vhdStreamValidator, noop))
+    }
+    deltaExport.streams = mapValues(deltaExport.streams, this._throttleStream)
+
+    const timestamp = Date.now()
+
+    await this._callWriters(
+      writer =>
+        writer.transfer({
+          deltaExport: forkDeltaExport(deltaExport),
+          differentialVhds,
+          sizeContainers,
+          timestamp,
+          vm,
+          vmSnapshot: exportedVm,
+        }),
+      'writer.transfer()'
+    )
+
+    this._baseVm = exportedVm
+
+    if (baseVm !== undefined) {
+      await exportedVm.update_other_config(
+        'xo:backup:deltaChainLength',
+        String(+(baseVm.other_config['xo:backup:deltaChainLength'] ?? 0) + 1)
+      )
+    }
+
+    // not the case if offlineBackup
+    if (exportedVm.is_a_snapshot) {
+      await exportedVm.update_other_config('xo:backup:exported', 'true')
+    }
+
+    const size = Object.values(sizeContainers).reduce((sum, { size }) => sum + size, 0)
+    const end = Date.now()
+    const duration = end - timestamp
+    debug('transfer complete', {
+      duration,
+      speed: duration !== 0 ? (size * 1e3) / 1024 / 1024 / duration : 0,
+      size,
+    })
+
+    await this._callWriters(writer => writer.cleanup(), 'writer.cleanup()')
+  }
+
+  async _selectBaseVm() {
+    const xapi = this._xapi
+
+    let baseVm = findLast(this._jobSnapshots, _ => 'xo:backup:exported' in _.other_config)
+    if (baseVm === undefined) {
+      debug('no base VM found')
+      return
+    }
+
+    const fullInterval = this._settings.fullInterval
+    const deltaChainLength = +(baseVm.other_config['xo:backup:deltaChainLength'] ?? 0) + 1
+    if (!(fullInterval === 0 || fullInterval > deltaChainLength)) {
+      debug('not using base VM becaust fullInterval reached')
+      return
+    }
+
+    const srcVdis = keyBy(await xapi.getRecords('VDI', await this._vm.$getDisks()), '$ref')
+
+    // resolve full record
+    baseVm = await xapi.getRecord('VM', baseVm.$ref)
+
+    const baseUuidToSrcVdi = new Map()
+    await asyncMap(await baseVm.$getDisks(), async baseRef => {
+      const [baseUuid, snapshotOf] = await Promise.all([
+        xapi.getField('VDI', baseRef, 'uuid'),
+        xapi.getField('VDI', baseRef, 'snapshot_of'),
+      ])
+      const srcVdi = srcVdis[snapshotOf]
+      if (srcVdi !== undefined) {
+        baseUuidToSrcVdi.set(baseUuid, srcVdi)
+      } else {
+        debug('ignore snapshot VDI because no longer present on VM', {
+          vdi: baseUuid,
+        })
+      }
+    })
+
+    const presentBaseVdis = new Map(baseUuidToSrcVdi)
+    await this._callWriters(
+      writer => presentBaseVdis.size !== 0 && writer.checkBaseVdis(presentBaseVdis, baseVm),
+      'writer.checkBaseVdis()',
+      false
+    )
+
+    if (presentBaseVdis.size === 0) {
+      debug('no base VM found')
+      return
+    }
+
+    const fullVdisRequired = new Set()
+    baseUuidToSrcVdi.forEach((srcVdi, baseUuid) => {
+      if (presentBaseVdis.has(baseUuid)) {
+        debug('found base VDI', {
+          base: baseUuid,
+          vdi: srcVdi.uuid,
+        })
+      } else {
+        debug('missing base VDI', {
+          base: baseUuid,
+          vdi: srcVdi.uuid,
+        })
+        fullVdisRequired.add(srcVdi.uuid)
+      }
+    })
+
+    this._baseVm = baseVm
+    this._fullVdisRequired = fullVdisRequired
+  }
+}

package/_runners/_vmRunners/_Abstract.js

@@ -0,0 +1,95 @@
+'use strict'
+
+const { asyncMap } = require('@xen-orchestra/async-map')
+const { createLogger } = require('@xen-orchestra/log')
+const { Task } = require('../../Task.js')
+
+const { debug, warn } = createLogger('xo:backups:AbstractVmRunner')
+
+class AggregateError extends Error {
+  constructor(errors, message) {
+    super(message)
+    this.errors = errors
+  }
+}
+
+const asyncEach = async (iterable, fn, thisArg = iterable) => {
+  for (const item of iterable) {
+    await fn.call(thisArg, item)
+  }
+}
+
+exports.Abstract = class AbstractVmBackupRunner {
+  // calls fn for each function, warns of any errors, and throws only if there are no writers left
+  async _callWriters(fn, step, parallel = true) {
+    const writers = this._writers
+    const n = writers.size
+    if (n === 0) {
+      return
+    }
+
+    async function callWriter(writer) {
+      const { name } = writer.constructor
+      try {
+        debug('writer step starting', { step, writer: name })
+        await fn(writer)
+        debug('writer step succeeded', { duration: step, writer: name })
+      } catch (error) {
+        writers.delete(writer)
+
+        warn('writer step failed', { error, step, writer: name })
+
+        // these two steps are the only one that are not already in their own sub tasks
+        if (step === 'writer.checkBaseVdis()' || step === 'writer.beforeBackup()') {
+          Task.warning(
+            `the writer ${name} has failed the step ${step} with error ${error.message}. It won't be used anymore in this job execution.`
+          )
+        }
+
+        throw error
+      }
+    }
+    if (n === 1) {
+      const [writer] = writers
+      return callWriter(writer)
+    }
+
+    const errors = []
+    await (parallel ? asyncMap : asyncEach)(writers, async function (writer) {
+      try {
+        await callWriter(writer)
+      } catch (error) {
+        errors.push(error)
+      }
+    })
+    if (writers.size === 0) {
+      throw new AggregateError(errors, 'all targets have failed, step: ' + step)
+    }
+  }
+
+  async _healthCheck() {
+    const settings = this._settings
+
+    if (this._healthCheckSr === undefined) {
+      return
+    }
+
+    // check if current VM has tags
+    const tags = this._tags
+    const intersect = settings.healthCheckVmsWithTags.some(t => tags.includes(t))
+
+    if (settings.healthCheckVmsWithTags.length !== 0 && !intersect) {
+      // create a task to have an info in the logs and reports
+      return Task.run(
+        {
+          name: 'health check',
+        },
+        () => {
+          Task.info(`This VM doesn't match the health check's tags for this schedule`)
+        }
+      )
+    }
+
+    await this._callWriters(writer => writer.healthCheck(), 'writer.healthCheck()')
+  }
+}

package/_runners/_vmRunners/_AbstractRemote.js

@@ -0,0 +1,86 @@
+'use strict'
+const { Abstract } = require('./_Abstract')
+
+const { getVmBackupDir } = require('../../_getVmBackupDir')
+const { asyncEach } = require('@vates/async-each')
+const { Disposable } = require('promise-toolbox')
+
+exports.AbstractRemote = class AbstractRemoteVmBackupRunner extends Abstract {
+  constructor({
+    config,
+    job,
+    healthCheckSr,
+    remoteAdapters,
+    schedule,
+    settings,
+    sourceRemoteAdapter,
+    throttleStream,
+    vmUuid,
+  }) {
+    super()
+    this.config = config
+    this.job = job
+    this.remoteAdapters = remoteAdapters
+    this.scheduleId = schedule.id
+    this.timestamp = undefined
+
+    this._healthCheckSr = healthCheckSr
+    this._sourceRemoteAdapter = sourceRemoteAdapter
+    this._throttleStream = throttleStream
+    this._vmUuid = vmUuid
+
+    const allSettings = job.settings
+    const writers = new Set()
+    this._writers = writers
+
+    const RemoteWriter = this._getRemoteWriter()
+    Object.entries(remoteAdapters).forEach(([remoteId, adapter]) => {
+      const targetSettings = {
+        ...settings,
+        ...allSettings[remoteId],
+      }
+      writers.add(new RemoteWriter({ adapter, config, healthCheckSr, job, vmUuid, remoteId, settings: targetSettings }))
+    })
+  }
+
+  async _computeTransferList(predicate) {
+    const vmBackups = await this._sourceRemoteAdapter.listVmBackups(this._vmUuid, predicate)
+    const localMetada = new Map()
+    Object.values(vmBackups).forEach(metadata => {
+      const timestamp = metadata.timestamp
+      localMetada.set(timestamp, metadata)
+    })
+    const nbRemotes = Object.keys(this.remoteAdapters).length
+    const remoteMetadatas = {}
+    await asyncEach(Object.values(this.remoteAdapters), async remoteAdapter => {
+      const remoteMetadata = await remoteAdapter.listVmBackups(this._vmUuid, predicate)
+      remoteMetadata.forEach(metadata => {
+        const timestamp = metadata.timestamp
+        remoteMetadatas[timestamp] = (remoteMetadatas[timestamp] ?? 0) + 1
+      })
+    })
+
+    let chain = []
+    const timestamps = [...localMetada.keys()]
+    timestamps.sort()
+    for (const timestamp of timestamps) {
+      if (remoteMetadatas[timestamp] !== nbRemotes) {
+        // this backup is not present in all the remote
+        // should be retransfered if not found later
+        chain.push(localMetada.get(timestamp))
+      } else {
+        // backup is present in local and remote : the chain has already been transferred
+        chain = []
+      }
+    }
+    return chain
+  }
+
+  async run() {
+    const handler = this._sourceRemoteAdapter._handler
+    await Disposable.use(await handler.lock(getVmBackupDir(this._vmUuid)), async () => {
+      await this._run()
+      await this._healthCheck()
+    })
+  }
+}

package/_runners/_vmRunners/_AbstractXapi.js

@@ -0,0 +1,257 @@
+'use strict'
+
+const assert = require('assert')
+const groupBy = require('lodash/groupBy.js')
+const ignoreErrors = require('promise-toolbox/ignoreErrors')
+const { asyncMap } = require('@xen-orchestra/async-map')
+const { decorateMethodsWith } = require('@vates/decorate-with')
+const { defer } = require('golike-defer')
+const { formatDateTime } = require('@xen-orchestra/xapi')
+
+const { getOldEntries } = require('../../_getOldEntries.js')
+const { Task } = require('../../Task.js')
+const { Abstract } = require('./_Abstract.js')
+
+class AbstractXapiVmBackupRunner extends Abstract {
+  constructor({
+    config,
+    getSnapshotNameLabel,
+    healthCheckSr,
+    job,
+    remoteAdapters,
+    remotes,
+    schedule,
+    settings,
+    srs,
+    throttleStream,
+    vm,
+  }) {
+    super()
+    if (vm.other_config['xo:backup:job'] === job.id && 'start' in vm.blocked_operations) {
+      // don't match replicated VMs created by this very job otherwise they
+      // will be replicated again and again
+      throw new Error('cannot backup a VM created by this very job')
+    }
+
+    this.config = config
+    this.job = job
+    this.remoteAdapters = remoteAdapters
+    this.scheduleId = schedule.id
+    this.timestamp = undefined
+
+    // VM currently backed up
+    const tags = (this._tags = vm.tags)
+
+    // VM (snapshot) that is really exported
+    this._exportedVm = undefined
+    this._vm = vm
+
+    this._fullVdisRequired = undefined
+    this._getSnapshotNameLabel = getSnapshotNameLabel
+    this._isIncremental = job.mode === 'delta'
+    this._healthCheckSr = healthCheckSr
+    this._jobId = job.id
+    this._jobSnapshots = undefined
+    this._throttleStream = throttleStream
+    this._xapi = vm.$xapi
+
+    // Base VM for the export
+    this._baseVm = undefined
+
+    // Settings for this specific run (job, schedule, VM)
+    if (tags.includes('xo-memory-backup')) {
+      settings.checkpointSnapshot = true
+    }
+    if (tags.includes('xo-offline-backup')) {
+      settings.offlineSnapshot = true
+    }
+    this._settings = settings
+    // Create writers
+    {
+      const writers = new Set()
+      this._writers = writers
+
+      const [BackupWriter, ReplicationWriter] = this._getWriters()
+
+      const allSettings = job.settings
+      Object.entries(remoteAdapters).forEach(([remoteId, adapter]) => {
+        const targetSettings = {
+          ...settings,
+          ...allSettings[remoteId],
+        }
+        if (targetSettings.exportRetention !== 0) {
+          writers.add(new BackupWriter({ adapter, config, healthCheckSr, job, vmUuid: vm.uuid, remoteId, settings: targetSettings }))
+        }
+      })
+      srs.forEach(sr => {
+        const targetSettings = {
+          ...settings,
+          ...allSettings[sr.uuid],
+        }
+        if (targetSettings.copyRetention !== 0) {
+          writers.add(new ReplicationWriter({  config, healthCheckSr, job, vmUuid: vm.uuid, sr, settings: targetSettings}))
+        }
+      })
+    }
+  }
+
+  // ensure the VM itself does not have any backup metadata which would be
+  // copied on manual snapshots and interfere with the backup jobs
+  async _cleanMetadata() {
+    const vm = this._vm
+    if ('xo:backup:job' in vm.other_config) {
+      await vm.update_other_config({
+        'xo:backup:datetime': null,
+        'xo:backup:deltaChainLength': null,
+        'xo:backup:exported': null,
+        'xo:backup:job': null,
+        'xo:backup:schedule': null,
+        'xo:backup:vm': null,
+      })
+    }
+  }
+
+  async _snapshot() {
+    const vm = this._vm
+    const xapi = this._xapi
+
+    const settings = this._settings
+
+    if (this._mustDoSnapshot()) {
+      await Task.run({ name: 'snapshot' }, async () => {
+        if (!settings.bypassVdiChainsCheck) {
+          await vm.$assertHealthyVdiChains()
+        }
+
+        const snapshotRef = await vm[settings.checkpointSnapshot ? '$checkpoint' : '$snapshot']({
+          ignoreNobakVdis: true,
+          name_label: this._getSnapshotNameLabel(vm),
+          unplugVusbs: true,
+        })
+        this.timestamp = Date.now()
+
+        await xapi.setFieldEntries('VM', snapshotRef, 'other_config', {
+          'xo:backup:datetime': formatDateTime(this.timestamp),
+          'xo:backup:job': this._jobId,
+          'xo:backup:schedule': this.scheduleId,
+          'xo:backup:vm': vm.uuid,
+        })
+
+        this._exportedVm = await xapi.getRecord('VM', snapshotRef)
+
+        return this._exportedVm.uuid
+      })
+    } else {
+      this._exportedVm = vm
+      this.timestamp = Date.now()
+    }
+  }
+
+  async _fetchJobSnapshots() {
+    const jobId = this._jobId
+    const vmRef = this._vm.$ref
+    const xapi = this._xapi
+
+    const snapshotsRef = await xapi.getField('VM', vmRef, 'snapshots')
+    const snapshotsOtherConfig = await asyncMap(snapshotsRef, ref => xapi.getField('VM', ref, 'other_config'))
+
+    const snapshots = []
+    snapshotsOtherConfig.forEach((other_config, i) => {
+      if (other_config['xo:backup:job'] === jobId) {
+        snapshots.push({ other_config, $ref: snapshotsRef[i] })
+      }
+    })
+    snapshots.sort((a, b) => (a.other_config['xo:backup:datetime'] < b.other_config['xo:backup:datetime'] ? -1 : 1))
+    this._jobSnapshots = snapshots
+  }
+
+  async _removeUnusedSnapshots() {
+    const allSettings = this.job.settings
+    const baseSettings = this._baseSettings
+    const baseVmRef = this._baseVm?.$ref
+
+    const snapshotsPerSchedule = groupBy(this._jobSnapshots, _ => _.other_config['xo:backup:schedule'])
+    const xapi = this._xapi
+    await asyncMap(Object.entries(snapshotsPerSchedule), ([scheduleId, snapshots]) => {
+      const settings = {
+        ...baseSettings,
+        ...allSettings[scheduleId],
+        ...allSettings[this._vm.uuid],
+      }
+      return asyncMap(getOldEntries(settings.snapshotRetention, snapshots), ({ $ref }) => {
+        if ($ref !== baseVmRef) {
+          return xapi.VM_destroy($ref)
+        }
+      })
+    })
+  }
+
+  async copy() {
+    throw new Error('Not implemented')
+  }
+
+  _getWriters() {
+    throw new Error('Not implemented')
+  }
+
+  _mustDoSnapshot() {
+    throw new Error('Not implemented')
+  }
+
+  async _selectBaseVm() {
+    throw new Error('Not implemented')
+  }
+
+  async run($defer) {
+    const settings = this._settings
+    assert(
+      !settings.offlineBackup || settings.snapshotRetention === 0,
+      'offlineBackup is not compatible with snapshotRetention'
+    )
+
+    await this._callWriters(async writer => {
+      await writer.beforeBackup()
+      $defer(async () => {
+        await writer.afterBackup()
+      })
+    }, 'writer.beforeBackup()')
+
+    await this._fetchJobSnapshots()
+
+    await this._selectBaseVm()
+
+    await this._cleanMetadata()
+    await this._removeUnusedSnapshots()
+
+    const vm = this._vm
+    const isRunning = vm.power_state === 'Running'
+    const startAfter = isRunning && (settings.offlineBackup ? 'backup' : settings.offlineSnapshot && 'snapshot')
+    if (startAfter) {
+      await vm.$callAsync('clean_shutdown')
+    }
+
+    try {
+      await this._snapshot()
+      if (startAfter === 'snapshot') {
+        ignoreErrors.call(vm.$callAsync('start', false, false))
+      }
+
+      if (this._writers.size !== 0) {
+        await this._copy()
+      }
+    } finally {
+      if (startAfter) {
+        ignoreErrors.call(vm.$callAsync('start', false, false))
+      }
+
+      await this._fetchJobSnapshots()
+      await this._removeUnusedSnapshots()
+    }
+    await this._healthCheck()
+  }
+}
+exports.AbstractXapi = AbstractXapiVmBackupRunner
+
+decorateMethodsWith(AbstractXapiVmBackupRunner, {
+  run: defer,
+})

package/_runners/_vmRunners/_forkDeltaExport.js

@@ -0,0 +1,12 @@
+'use strict'
+
+const { mapValues } = require('lodash')
+const { forkStreamUnpipe } = require('../_forkStreamUnpipe')
+
+exports.forkDeltaExport = function forkDeltaExport(deltaExport) {
+  return Object.create(deltaExport, {
+    streams: {
+      value: mapValues(deltaExport.streams, forkStreamUnpipe),
+    },
+  })
+}

package/{writers/FullBackupWriter.js → _runners/_writers/FullRemoteWriter.js}

@@ -1,13 +1,13 @@
 'use strict'
 
-const { formatFilenameDate } = require('../_filenameDate.js')
-const { getOldEntries } = require('../_getOldEntries.js')
-const { Task } = require('../Task.js')
+const { formatFilenameDate } = require('../../_filenameDate.js')
+const { getOldEntries } = require('../../_getOldEntries.js')
+const { Task } = require('../../Task.js')
 
-const { MixinBackupWriter } = require('./_MixinBackupWriter.js')
+const { MixinRemoteWriter } = require('./_MixinRemoteWriter.js')
 const { AbstractFullWriter } = require('./_AbstractFullWriter.js')
 
-exports.FullBackupWriter = class FullBackupWriter extends MixinBackupWriter(AbstractFullWriter) {
+exports.FullRemoteWriter = class FullRemoteWriter extends MixinRemoteWriter(AbstractFullWriter) {
   constructor(props) {
     super(props)
 
@@ -26,15 +26,17 @@ exports.FullBackupWriter = class FullBackupWriter extends MixinBackupWriter(Abst
     )
   }
 
-  async _run({ timestamp, sizeContainer, stream }) {
-    const backup = this._backup
+  async _run({ timestamp, sizeContainer, stream, vm, vmSnapshot }) {
     const settings = this._settings
-
-    const { job, scheduleId, vm } = backup
+    const job = this._job
+    const scheduleId = this._scheduleId
 
     const adapter = this._adapter
-
-    // TODO: clean VM backup directory
+    let metadata = await this._isAlreadyTransferred(timestamp)
+    if (metadata !== undefined) {
+      // @todo : should skip backup while being vigilant to not stuck the forked stream
+      Task.info('This backup has already been transfered')
+    }
 
     const oldBackups = getOldEntries(
       settings.exportRetention - 1,
@@ -47,14 +49,14 @@ exports.FullBackupWriter = class FullBackupWriter extends MixinBackupWriter(Abst
     const dataBasename = basename + '.xva'
     const dataFilename = this._vmBackupDir + '/' + dataBasename
 
-    const metadata = {
+    metadata = {
       jobId: job.id,
       mode: job.mode,
       scheduleId,
       timestamp,
       version: '2.0.0',
       vm,
-      vmSnapshot: this._backup.exportedVm,
+      vmSnapshot,
       xva: './' + dataBasename,
     }