feat(backups-cli): lowlevel tool to help with backups (#4556)

.eslintrc.js

@@ -21,7 +21,7 @@ module.exports = {
 
   overrides: [
     {
-      files: ['cli.js', '*-cli.js', 'packages/*cli*/**/*.js'],
+      files: ['cli.js', '*-cli.js', '**/*cli*/**/*.js'],
       rules: {
         'no-console': 'off',
       },

@xen-orchestra/backups-cli/index.js

@@ -0,0 +1,369 @@
+#!/usr/bin/env node
+
+const args = process.argv.slice(2)
+
+if (
+  args.length === 0 ||
+  /^(?:-h|--help)$/.test(args[0]) ||
+  args[0] !== 'clean-vms'
+) {
+  console.log('Usage: xo-backups clean-vms [--force] xo-vm-backups/*')
+  // eslint-disable-next-line no-process-exit
+  return process.exit(1)
+}
+
+// remove `clean-vms` arg which is the only available command ATM
+args.splice(0, 1)
+
+// only act (ie delete files) if `--force` is present
+const force = args[0] === '--force'
+if (force) {
+  args.splice(0, 1)
+}
+
+// -----------------------------------------------------------------------------
+
+const assert = require('assert')
+const lockfile = require('proper-lockfile')
+const { default: Vhd } = require('vhd-lib')
+const { curryRight, flatten } = require('lodash')
+const { dirname, resolve } = require('path')
+const { DISK_TYPE_DIFFERENCING } = require('vhd-lib/dist/_constants')
+const { pipe, promisifyAll } = require('promise-toolbox')
+
+const fs = promisifyAll(require('fs'))
+const handler = require('@xen-orchestra/fs').getHandler({ url: 'file://' })
+
+// -----------------------------------------------------------------------------
+
+const asyncMap = curryRight((iterable, fn) =>
+  Promise.all(
+    Array.isArray(iterable) ? iterable.map(fn) : Array.from(iterable, fn)
+  )
+)
+
+const filter = (...args) => thisArg => thisArg.filter(...args)
+
+// TODO: better check?
+
+// our heuristic is not good enough, there has been some false positives
+// (detected as invalid by us but valid by `tar` and imported with success),
+// either:
+// - these files were normal but the check is incorrect
+// - these files were invalid but without data loss
+// - these files were invalid but with silent data loss
+//
+// FIXME: the heuristic does not work if the XVA is compressed, we need to
+// implement a specific test for it
+//
+// maybe reading the end of the file looking for a file named
+// /^Ref:\d+/\d+\.checksum$/ and then validating the tar structure from it
+//
+// https://github.com/npm/node-tar/issues/234#issuecomment-538190295
+const isValidTar = async path => {
+  try {
+    const fd = await fs.open(path, 'r')
+    try {
+      const { size } = await fs.fstat(fd)
+      if (size <= 1024 || size % 512 !== 0) {
+        return false
+      }
+
+      const buf = Buffer.allocUnsafe(1024)
+      assert.strictEqual(
+        await fs.read(fd, buf, 0, buf.length, size - buf.length),
+        buf.length
+      )
+      return buf.every(_ => _ === 0)
+    } finally {
+      fs.close(fd).catch(noop)
+    }
+  } catch (error) {
+    // never throw, log and report as valid to avoid side effects
+    console.error('isValidTar', path, error)
+    return true
+  }
+}
+
+const noop = Function.prototype
+
+const readDir = path =>
+  fs.readdir(path).then(entries => {
+    entries.forEach((entry, i) => {
+      entries[i] = `${path}/${entry}`
+    })
+
+    return entries
+  })
+
+// -----------------------------------------------------------------------------
+
+// chain is an array of VHDs from child to parent
+//
+// the whole chain will be merged into parent, parent will be renamed to child
+// and all the others will deleted
+async function mergeVhdChain(chain) {
+  assert(chain.length >= 2)
+
+  const child = chain[0]
+  const parent = chain[chain.length - 1]
+  const children = chain.slice(0, -1).reverse()
+
+  console.warn('Unused parents of VHD', child)
+  chain
+    .slice(1)
+    .reverse()
+    .forEach(parent => {
+      console.warn('  ', parent)
+    })
+  force && console.warn('  merging…')
+  console.warn('')
+  if (force) {
+    // `mergeVhd` does not work with a stream, either
+    // - make it accept a stream
+    // - or create synthetic VHD which is not a stream
+    return console.warn('TODO: implement merge')
+    // await mergeVhd(
+    //   handler,
+    //   parent,
+    //   handler,
+    //   children.length === 1
+    //     ? child
+    //     : await createSyntheticStream(handler, children)
+    // )
+  }
+
+  await Promise.all([
+    force && fs.rename(parent, child),
+    asyncMap(children.slice(0, -1), child => {
+      console.warn('Unused VHD', child)
+      force && console.warn('  deleting…')
+      console.warn('')
+      return force && handler.unlink(child)
+    }),
+  ])
+}
+
+const listVhds = pipe([
+  vmDir => vmDir + '/vdis',
+  readDir,
+  asyncMap(readDir),
+  flatten,
+  asyncMap(readDir),
+  flatten,
+  filter(_ => _.endsWith('.vhd')),
+])
+
+async function handleVm(vmDir) {
+  const vhds = new Set()
+  const vhdParents = { __proto__: null }
+  const vhdChildren = { __proto__: null }
+
+  // remove broken VHDs
+  await asyncMap(await listVhds(vmDir), async path => {
+    try {
+      const vhd = new Vhd(handler, path)
+      await vhd.readHeaderAndFooter()
+      vhds.add(path)
+      if (vhd.footer.diskType === DISK_TYPE_DIFFERENCING) {
+        const parent = resolve(dirname(path), vhd.header.parentUnicodeName)
+        vhdParents[path] = parent
+        if (parent in vhdChildren) {
+          const error = new Error(
+            'this script does not support multiple VHD children'
+          )
+          error.parent = parent
+          error.child1 = vhdChildren[parent]
+          error.child2 = path
+          throw error // should we throw?
+        }
+        vhdChildren[parent] = path
+      }
+    } catch (error) {
+      console.warn('Error while checking VHD', path)
+      console.warn('  ', error)
+      if (error != null && error.code === 'ERR_ASSERTION') {
+        force && console.warn('  deleting…')
+        console.warn('')
+        force && (await handler.unlink(path))
+      }
+    }
+  })
+
+  // remove VHDs with missing ancestors
+  {
+    const deletions = []
+
+    // return true if the VHD has been deleted or is missing
+    const deleteIfOrphan = vhd => {
+      const parent = vhdParents[vhd]
+      if (parent === undefined) {
+        return
+      }
+
+      // no longer needs to be checked
+      delete vhdParents[vhd]
+
+      deleteIfOrphan(parent)
+
+      if (!vhds.has(parent)) {
+        vhds.delete(vhd)
+
+        console.warn('Error while checking VHD', vhd)
+        console.warn('  missing parent', parent)
+        force && console.warn('  deleting…')
+        console.warn('')
+        force && deletions.push(handler.unlink(vhd))
+      }
+    }
+
+    // > A property that is deleted before it has been visited will not be
+    // > visited later.
+    // >
+    // > -- https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/for...in#Deleted_added_or_modified_properties
+    for (const child in vhdParents) {
+      deleteIfOrphan(child)
+    }
+
+    await Promise.all(deletions)
+  }
+
+  const [jsons, xvas] = await readDir(vmDir).then(entries => [
+    entries.filter(_ => _.endsWith('.json')),
+    new Set(entries.filter(_ => _.endsWith('.xva'))),
+  ])
+
+  await asyncMap(xvas, async path => {
+    // check is not good enough to delete the file, the best we can do is report
+    // it
+    if (!(await isValidTar(path))) {
+      console.warn('Potential broken XVA', path)
+      console.warn('')
+    }
+  })
+
+  const unusedVhds = new Set(vhds)
+  const unusedXvas = new Set(xvas)
+
+  // compile the list of unused XVAs and VHDs, and remove backup metadata which
+  // reference a missing XVA/VHD
+  await asyncMap(jsons, async json => {
+    const metadata = JSON.parse(await fs.readFile(json))
+    const { mode } = metadata
+    if (mode === 'full') {
+      const linkedXva = resolve(vmDir, metadata.xva)
+
+      if (xvas.has(linkedXva)) {
+        unusedXvas.delete(linkedXva)
+      } else {
+        console.warn('Error while checking backup', json)
+        console.warn('  missing file', linkedXva)
+        force && console.warn('  deleting…')
+        console.warn('')
+        force && (await handler.unlink(json))
+      }
+    } else if (mode === 'delta') {
+      const linkedVhds = (() => {
+        const { vhds } = metadata
+        return Object.keys(vhds).map(key => resolve(vmDir, vhds[key]))
+      })()
+
+      // FIXME: find better approach by keeping as much of the backup as
+      // possible (existing disks) even if one disk is missing
+      if (linkedVhds.every(_ => vhds.has(_))) {
+        linkedVhds.forEach(_ => unusedVhds.delete(_))
+      } else {
+        console.warn('Error while checking backup', json)
+        const missingVhds = linkedVhds.filter(_ => !vhds.has(_))
+        console.warn(
+          '  %i/%i missing VHDs',
+          missingVhds.length,
+          linkedVhds.length
+        )
+        missingVhds.forEach(vhd => {
+          console.warn('  ', vhd)
+        })
+        force && console.warn('  deleting…')
+        console.warn('')
+        force && (await handler.unlink(json))
+      }
+    }
+  })
+
+  // TODO: parallelize by vm/job/vdi
+  const unusedVhdsDeletion = []
+  {
+    // VHD chains (as list from child to ancestor) to merge indexed by last
+    // ancestor
+    const vhdChainsToMerge = { __proto__: null }
+
+    const toCheck = new Set(unusedVhds)
+
+    const getUsedChildChainOrDelete = vhd => {
+      if (vhd in vhdChainsToMerge) {
+        const chain = vhdChainsToMerge[vhd]
+        delete vhdChainsToMerge[vhd]
+        return chain
+      }
+
+      if (!unusedVhds.has(vhd)) {
+        return [vhd]
+      }
+
+      // no longer needs to be checked
+      toCheck.delete(vhd)
+
+      const child = vhdChildren[vhd]
+      if (child !== undefined) {
+        const chain = getUsedChildChainOrDelete(child)
+        if (chain !== undefined) {
+          chain.push(vhd)
+          return chain
+        }
+      }
+
+      console.warn('Unused VHD', vhd)
+      force && console.warn('  deleting…')
+      console.warn('')
+      force && unusedVhdsDeletion.push(handler.unlink(vhd))
+    }
+
+    toCheck.forEach(vhd => {
+      vhdChainsToMerge[vhd] = getUsedChildChainOrDelete(vhd)
+    })
+
+    Object.keys(vhdChainsToMerge).forEach(key => {
+      const chain = vhdChainsToMerge[key]
+      if (chain !== undefined) {
+        unusedVhdsDeletion.push(mergeVhdChain(chain))
+      }
+    })
+  }
+
+  await Promise.all([
+    unusedVhdsDeletion,
+    asyncMap(unusedXvas, path => {
+      console.warn('Unused XVA', path)
+      force && console.warn('  deleting…')
+      console.warn('')
+      return force && handler.unlink(path)
+    }),
+  ])
+}
+
+// -----------------------------------------------------------------------------
+
+asyncMap(args, async vmDir => {
+  vmDir = resolve(vmDir)
+
+  // TODO: implement this in `xo-server`, not easy because not compatible with
+  // `@xen-orchestra/fs`.
+  const release = await lockfile.lock(vmDir)
+  try {
+    await handleVm(vmDir)
+  } catch (error) {
+    console.error('handleVm', vmDir, error)
+  } finally {
+    await release()
+  }
+}).catch(error => console.error('main', error))

@xen-orchestra/backups-cli/package.json

@@ -0,0 +1,27 @@
+{
+  "bin": {
+    "xo-backups": "index.js"
+  },
+  "bugs": "https://github.com/vatesfr/xen-orchestra/issues",
+  "dependencies": {
+    "@xen-orchestra/fs": "^0.10.1",
+    "lodash": "^4.17.15",
+    "promise-toolbox": "^0.13.0",
+    "proper-lockfile": "^4.1.1",
+    "vhd-lib": "^0.7.0"
+  },
+  "engines": {
+    "node": ">=8.16.1"
+  },
+  "homepage": "https://github.com/vatesfr/xen-orchestra/tree/master/@xen-orchestra/backups-cli",
+  "name": "@xen-orchestra/backups-cli",
+  "repository": {
+    "directory": "@xen-orchestra/backups-cli",
+    "type": "git",
+    "url": "https://github.com/vatesfr/xen-orchestra.git"
+  },
+  "scripts": {
+    "postversion": "npm publish --access public"
+  },
+  "version": "0.0.0"
+}

packages/vhd-lib/src/createSyntheticStream.js

@@ -13,7 +13,7 @@ import {
 import { fuFooter, fuHeader, checksumStruct } from './_structs'
 import { test as mapTestBit } from './_bitmap'
 
-export default async function createSyntheticStream(handler, path) {
+export default async function createSyntheticStream(handler, paths) {
   const fds = []
   const cleanup = () => {
     for (let i = 0, n = fds.length; i < n; ++i) {
@@ -24,7 +24,7 @@ export default async function createSyntheticStream(handler, path) {
   }
   try {
     const vhds = []
-    while (true) {
+    const open = async path => {
       const fd = await handler.openFile(path, 'r')
       fds.push(fd)
       const vhd = new Vhd(handler, fd)
@@ -32,11 +32,18 @@ export default async function createSyntheticStream(handler, path) {
       await vhd.readHeaderAndFooter()
       await vhd.readBlockAllocationTable()
 
-      if (vhd.footer.diskType === DISK_TYPE_DYNAMIC) {
+      return vhd
-        break
+    }
+    if (typeof paths === 'string') {
+      let path = paths
+      let vhd
+      while ((vhd = await open(path)).footer.diskType !== DISK_TYPE_DYNAMIC) {
+        path = resolveRelativeFromFile(path, vhd.header.parentUnicodeName)
+      }
+    } else {
+      for (const path of paths) {
+        await open(path)
       }
-
-      path = resolveRelativeFromFile(path, vhd.header.parentUnicodeName)
     }
     const nVhds = vhds.length
 

yarn.lock

@@ -11165,6 +11165,15 @@ prop-types@^15.5.10, prop-types@^15.5.4, prop-types@^15.5.6, prop-types@^15.5.7,
     object-assign "^4.1.1"
     react-is "^16.8.1"
 
+proper-lockfile@^4.1.1:
+  version "4.1.1"
+  resolved "https://registry.yarnpkg.com/proper-lockfile/-/proper-lockfile-4.1.1.tgz#284cf9db9e30a90e647afad69deb7cb06881262c"
+  integrity sha512-1w6rxXodisVpn7QYvLk706mzprPTAPCYAqxMvctmPN3ekuRk/kuGkGc82pangZiAt4R3lwSuUzheTTn0/Yb7Zg==
+  dependencies:
+    graceful-fs "^4.1.11"
+    retry "^0.12.0"
+    signal-exit "^3.0.2"
+
 protocol-buffers-encodings@^1.1.0:
   version "1.1.0"
   resolved "https://registry.yarnpkg.com/protocol-buffers-encodings/-/protocol-buffers-encodings-1.1.0.tgz#f3905631106669b85381bad47a336add7d206873"
@@ -12408,6 +12417,11 @@ ret@~0.1.10:
   resolved "https://registry.yarnpkg.com/ret/-/ret-0.1.15.tgz#b8a4825d5bdb1fc3f6f53c2bc33f81388681c7bc"
   integrity sha512-TTlYpa+OL+vMMNG24xSlQGEJ3B/RzEfUlLct7b5G/ytav+wPrplCpVMFuwzXbkecJrb6IYo1iFb0S9v37754mg==
 
+retry@^0.12.0:
+  version "0.12.0"
+  resolved "https://registry.yarnpkg.com/retry/-/retry-0.12.0.tgz#1b42a6266a21f07421d1b0b54b7dc167b01c013b"
+  integrity sha1-G0KmJmoh8HQh0bC1S33BZ7AcATs=
+
 reusify@^1.0.0:
   version "1.0.4"
   resolved "https://registry.yarnpkg.com/reusify/-/reusify-1.0.4.tgz#90da382b1e126efc02146e90845a88db12925d76"