quay/tools/auditancestry.py

import logging
import json

from data.database import Image, ImageStorage, Repository, User, configure
from data import model
from app import app, storage as store


logger = logging.getLogger(__name__)
logging.basicConfig(level=logging.DEBUG)

configure(app.config)

# Turn off debug logging for boto
logging.getLogger('boto').setLevel(logging.CRITICAL)


query = (Image
         .select(Image, ImageStorage, Repository, User)
         .join(ImageStorage)
         .switch(Image)
         .join(Repository)
         .join(User)
         .where(ImageStorage.uploading == False))

bad_count = 0
good_count = 0

def resolve_or_create(repo, docker_image_id, new_ancestry):
  existing = model.get_repo_image(repo.namespace_user.username, repo.name, docker_image_id)
  if existing:
    logger.debug('Found existing image: %s, %s', existing.id, docker_image_id)
    return existing
  else:
    # we need to find some storage to link it to
    try:
      to_link = (ImageStorage
        .select()
        .join(Image)
        .where(Image.docker_image_id == docker_image_id)
        .get())
      logger.debug('Linking to storage: %s' % to_link.uuid)
      created = Image.create(docker_image_id=docker_image_id, repository=repo,
                             storage=to_link, ancestors=new_ancestry)
      logger.debug('Created image: %s' % created)
      return created
    except ImageStorage.DoesNotExist:
      msg = 'No image available anywhere for storage: %s in namespace: %s'
      logger.error(msg, docker_image_id, repo.namespace_user.username)
      raise RuntimeError()


def all_ancestors_exist(ancestors):
  if not ancestors:
    return True

  found_count = len(list(Image
    .select()
    .where(Image.id << ancestors)))
  return found_count == len(ancestors)


cant_fix = []
for img in query:
  try:
    with_locations = model.get_repo_image(img.repository.namespace_user.username,
                                          img.repository.name, img.docker_image_id)
    ancestry_storage = store.image_ancestry_path(img.storage.uuid)
    if store.exists(with_locations.storage.locations, ancestry_storage):
      full_ancestry = json.loads(store.get_content(with_locations.storage.locations,
                                                   ancestry_storage))[1:]
      full_ancestry.reverse()

      ancestor_dbids = [int(anc_id) for anc_id in img.ancestors.split('/')[1:-1]]

      if len(full_ancestry) != len(ancestor_dbids) or not all_ancestors_exist(ancestor_dbids):
        logger.error('Image has incomplete ancestry: %s, %s, %s, %s', img.id, img.docker_image_id,
                     full_ancestry, ancestor_dbids)

        fixed_ancestry = '/'
        for ancestor in full_ancestry:
          ancestor_img = resolve_or_create(img.repository, ancestor,
                                           fixed_ancestry)
          fixed_ancestry += str(ancestor_img.id) + '/'

        img.ancestors = fixed_ancestry
        img.save()

        bad_count += 1
      else:
        good_count += 1
    else:
      bad_count += 1

  except RuntimeError:
    cant_fix.append(img)

  logger.debug('Bad: %s Good: %s Can\'t Fix: %s', bad_count, good_count,
               len(cant_fix))

for cant in cant_fix:
  logger.error('Unable to fix %s in repo %s/%s', cant.id, cant.repository.namespace_user.username,
               cant.repository.name)