Make get_syzygy_binaries.py resistant to cloud failures.

build/get_syzygy_binaries.py

 #!/usr/bin/env python
 # Copyright 2014 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 """A utility script for downloading versioned Syzygy binaries."""
 
 import cStringIO
 import hashlib
 import errno
@@ skipping 232 matching lines @@
     if os.path.exists(p) and _DirIsEmpty(p):
       _LOGGER.debug('Deleting empty directory "%s".', p)
       if not dry_run:
         _RmTree(p)
 
   return deleted
 
 
 def _Download(url):
   """Downloads the given URL and returns the contents as a string."""
-  response = urllib2.urlopen(url)
-  if response.code != 200:
-    raise RuntimeError('Failed to download "%s".' % url)
-  return response.read()
+  _LOGGER.debug('Downloading "%s".', url)
+  retries = 3
+  while retries > 0:
+    code = 0
+    try:
+      response = urllib2.urlopen(url)
+      if response.code == 200:
+        return response.read()
+      code = response.code
+    except urllib2.HTTPError, e:
+      code = e.code
+
+    _LOGGER.debug('Download failed with a %d response, retrying.',
+                  code)
+    retries -= 1

[Nico, 2015/07/10 15:04:39]

Trying again immediately likely won't work. Try sleeping 4s after the first try,
8s after the second or something like that (
https://en.wikipedia.org/wiki/Exponential_backoff etc)

[chrisha, 2015/07/14 17:51:38]

Err, yup. Meant to do that.

+
+  raise RuntimeError('Failed to download "%s".' % url)
 
 
 def _InstallBinaries(options, deleted={}):
   """Installs Syzygy binaries. This assumes that the output directory has
   already been cleaned, as it will refuse to overwrite existing files."""
   contents = {}
   state = { 'revision': options.revision, 'contents': contents }
   archive_url = _SYZYGY_ARCHIVE_URL % { 'revision': options.revision }
   if options.resources:
     resources = [(resource, resource, '', None)
                  for resource in options.resources]
   else:
     resources = _RESOURCES
   for (base, name, subdir, filt) in resources:
     # Create the output directory if it doesn't exist.
     fulldir = os.path.join(options.output_dir, subdir)
     if os.path.isfile(fulldir):
       raise Exception('File exists where a directory needs to be created: %s' %
                       fulldir)
     if not os.path.exists(fulldir):
       _LOGGER.debug('Creating directory: %s', fulldir)
       if not options.dry_run:
         os.makedirs(fulldir)
 
-    # Download the archive.
-    url = archive_url + '/' + base
-    _LOGGER.debug('Retrieving %s archive at "%s".', name, url)
-    data = _Download(url)
+    # Download and unzip the archive. Try this a few times if it fails, as the
+    # cloud storage server occasionally flakes and sends incomplete data.
+    retries = 0
+    while True:
+      url = archive_url + '/' + base
+      _LOGGER.debug('Retrieving %s archive at "%s".', name, url)
+      data = _Download(url)
 
-    _LOGGER.debug('Unzipping %s archive.', name)
-    archive = zipfile.ZipFile(cStringIO.StringIO(data))
+      try:
+        _LOGGER.debug('Unzipping %s archive.', name)
+        archive = zipfile.ZipFile(cStringIO.StringIO(data))
+      except zipfile.BadZipfile, e:
+        # If retried too often then let the error continue.
+        retries += 1
+        if retries == 3:
+          raise e
+       
+        # Otherwise retry the download to see if the archive comes down clean
+        # this time.
+        _LOGGER.debug('Bad zip file, retrying download.')
+        continue
+
+      # Successfully unzipped the archive, so continue to extract files.
+      break
+
+    # Extract desired contents and install to disk.
     for entry in archive.infolist():
       if not filt or filt(entry):
         fullpath = os.path.normpath(os.path.join(fulldir, entry.filename))
         relpath = os.path.relpath(fullpath, options.output_dir)
         if os.path.exists(fullpath):
           # If in a dry-run take into account the fact that the file *would*
           # have been deleted.
           if options.dry_run and relpath in deleted:
             pass
           else:
@@ skipping 11 matching lines @@
   return state
 
 
 def _ParseCommandLine():
   """Parses the command-line and returns an options structure."""
   option_parser = optparse.OptionParser()
   option_parser.add_option('--dry-run', action='store_true', default=False,
       help='If true then will simply list actions that would be performed.')
   option_parser.add_option('--force', action='store_true', default=False,
       help='Force an installation even if the binaries are up to date.')
+  option_parser.add_option('--no-cleanup', action='store_true', default=False,
+      help='Allow installation on non-Windows platforms, and skip the forced '
+           'cleanup step.')
   option_parser.add_option('--output-dir', type='string',
       help='The path where the binaries will be replaced. Existing binaries '
            'will only be overwritten if not up to date.')
   option_parser.add_option('--overwrite', action='store_true', default=False,
       help='If specified then the installation will happily delete and rewrite '
            'the entire output directory, blasting any local changes.')
   option_parser.add_option('--revision', type='string',
       help='The SVN revision or GIT hash associated with the required version.')
   option_parser.add_option('--revision-file', type='string',
       help='A text file containing an SVN revision or GIT hash.')
@@ skipping 72 matching lines @@
   options = _ParseCommandLine()
 
   if options.dry_run:
     _LOGGER.debug('Performing a dry-run.')
 
   # We only care about Windows platforms, as the Syzygy binaries aren't used
   # elsewhere. However, there was a short period of time where this script
   # wasn't gated on OS types, and those OSes downloaded and installed binaries.
   # This will cleanup orphaned files on those operating systems.
   if sys.platform not in ('win32', 'cygwin'):
-    return _RemoveOrphanedFiles(options)
+    if options.no_cleanup:
+      _LOGGER.debug('Skipping usual cleanup for non-Windows platforms.')
+    else:
+      return _RemoveOrphanedFiles(options)
 
   # Load the current installation state, and validate it against the
   # requested installation.
   state, is_consistent = _GetCurrentState(options.revision, options.output_dir)
 
   # Decide whether or not an install is necessary.
   if options.force:
     _LOGGER.debug('Forcing reinstall of binaries.')
   elif is_consistent:
     # Avoid doing any work if the contents of the directory are consistent.
@@ skipping 20 matching lines @@
   # Install the new binaries. In a dry-run this will actually download the
   # archives, but it won't write anything to disk.
   state = _InstallBinaries(options, deleted)
 
   # Build and save the state for the directory.
   _SaveState(options.output_dir, state, options.dry_run)
 
 
 if __name__ == '__main__':
   main()