feat(backups-cli): dont report compressed XVAs as broken (#4642)

Detect compressed XVAs and don't try to validate them with tar heuristic.

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

@@ -44,43 +44,62 @@ const asyncMap = curryRight((iterable, fn) =>
 
 const filter = (...args) => thisArg => thisArg.filter(...args)
 
-// TODO: better check?
+const isGzipFile = async fd => {
+  // https://tools.ietf.org/html/rfc1952.html#page-5
+  const magicNumber = Buffer.allocUnsafe(2)
+  assert.strictEqual(
+    await fs.read(fd, magicNumber, 0, magicNumber.length, 0),
+    magicNumber.length
+  )
+  return magicNumber[0] === 31 && magicNumber[1] === 139
+}
 
+// 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:
+// either THOUGH THEY MAY HAVE BEEN COMPRESSED FILES:
 // - 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 => {
+const isValidTar = async (size, 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)
+}
+
+// TODO: find an heuristic for compressed files
+const isValidXva = async path => {
   try {
     const fd = await fs.open(path, 'r')
     try {
       const { size } = await fs.fstat(fd)
-      if (size <= 1024 || size % 512 !== 0) {
+      if (size < 20) {
+        // neither a valid gzip not tar
         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)
+      return (await isGzipFile(fd))
+        ? true // gzip files cannot be validated at this time
+        : await isValidTar(size, fd)
     } finally {
       fs.close(fd).catch(noop)
     }
   } catch (error) {
     // never throw, log and report as valid to avoid side effects
-    console.error('isValidTar', path, error)
+    console.error('isValidXva', path, error)
     return true
   }
 }
@@ -245,7 +264,7 @@ async function handleVm(vmDir) {
   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))) {
+    if (!(await isValidXva(path))) {
       console.warn('Potential broken XVA', path)
       console.warn('')
     }