starlingx/config
Code Issues Proposed changes

Support an Alternative Backup Restore Procedure

Browse Source 
This code change is to support system backup and restore
without reinstalling storage nodes.

The config_controller command is modified as follows:

  A new option is added to system restore command
  "config_controller --restore-system <name>" with default set
  to exclude-storage-reinstall.

  config_controller --restore-system <include-storage-reinstall |
      exclude-storage-reinstall> <name>

Add a new command "tidy_storage_post_restore" to be run after
the restore. It scans Cinder/Glance and rbd backend for any
discrepancy between the two and generates a user action log file.

This code change has been tested on a virtual box as well as
hardware lab.

Story: 2004184
Task: 27672
Change-Id: I5e64fde70b977ea4bc3a5927bcbd852a393caec2
Signed-off-by: Wei Zhou <wei.zhou@windriver.com>
This commit is contained in:
Wei Zhou
2018-10-26 12:00:43 -04:00
parent 2b3da19436
commit 2b6f92da1b
14 changed files with 736 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

578
controllerconfig/controllerconfig/controllerconfig/tidy_storage.py Normal file
View File

@@ -0,0 +1,578 @@
"""
Copyright (c) 2015-2018 Wind River Systems, Inc.
SPDX-License-Identifier: Apache-2.0
"""
import numpy as np
import os
import subprocess
import sys
import textwrap
import time
from keystoneclient.auth.identity import v3
from keystoneauth1 import session as ksc_session
from cinderclient.v3 import client as cinder_client_v3
from glanceclient import Client
from cinderclient import utils as c_utils
from controllerconfig.common import log
from controllerconfig.common.rest_api_utils import get_token
from controllerconfig.common.exceptions import TidyStorageFail
LOG = log.get_logger(__name__)
KEYSTONE_AUTH_SERVER_RETRY_CNT = 60
KEYSTONE_AUTH_SERVER_WAIT = 1 # 1sec wait per retry
search_opts = {'all_tenants': 1}
class OpenStack(object):
def __init__(self):
self.admin_token = None
self.conf = {}
self.cinder_client = None
self.glance_client_v1 = None
self.glance_client_v2 = None
try:
self.conf['admin_user'] = os.environ['OS_USERNAME']
self.conf['admin_pwd'] = os.environ['OS_PASSWORD']
self.conf['admin_tenant'] = os.environ['OS_PROJECT_NAME']
self.conf['auth_url'] = os.environ['OS_AUTH_URL']
self.conf['region_name'] = os.environ['OS_REGION_NAME']
self.conf['user_domain'] = os.environ['OS_USER_DOMAIN_NAME']
self.conf['project_domain'] = os.environ['OS_PROJECT_DOMAIN_NAME']
except KeyError:
LOG.error("Please source openstack service credentials file.")
raise TidyStorageFail("Please source openstack credentials file.")
def __enter__(self):
if not self._connect():
raise Exception('Failed to connect')
return self
def __exit__(self, exc_type, exc_val, exc_tb):
self._disconnect()
def __del__(self):
self._disconnect()
def _connect(self):
""" Connect to an OpenStack instance """
if self.admin_token is not None:
self._disconnect()
# Try to obtain an admin token from keystone
for _ in range(KEYSTONE_AUTH_SERVER_RETRY_CNT):
self.admin_token = get_token(self.conf['auth_url'],
self.conf['admin_tenant'],
self.conf['admin_user'],
self.conf['admin_pwd'],
self.conf['user_domain'],
self.conf['project_domain'])
if self.admin_token:
break
time.sleep(KEYSTONE_AUTH_SERVER_WAIT)
return self.admin_token is not None
def _disconnect(self):
""" Disconnect from an OpenStack instance """
self.admin_token = None
@property
def get_cinder_client(self):
if not self.cinder_client:
auth = v3.Password(auth_url=self.conf['auth_url'],
username=self.conf['admin_user'],
password=self.conf['admin_pwd'],
user_domain_name=self.conf['user_domain'],
project_name=self.conf['admin_tenant'],
project_domain_name=self.conf['project_domain'])
self.cinder_client = cinder_client_v3.Client(
session=ksc_session.Session(auth=auth),
auth_url=self.conf['auth_url'],
endpoint_type='internalURL',
region_name="RegionOne")
return self.cinder_client
@property
def get_glance_client(self):
if not self.glance_client_v1 or not self.glance_client_v2:
auth = v3.Password(auth_url=self.conf['auth_url'],
username=self.conf['admin_user'],
password=self.conf['admin_pwd'],
user_domain_name=self.conf['user_domain'],
project_name=self.conf['admin_tenant'],
project_domain_name=self.conf['project_domain'])
self.glance_client_v1 = Client(
'1', session=ksc_session.Session(auth=auth))
self.glance_client_v2 = Client(
'2', session=ksc_session.Session(auth=auth))
return self.glance_client_v1, self.glance_client_v2
def show_help():
print ("Usage: %s <user_action_log_file>" % sys.argv[0])
print textwrap.fill(
"Tidy storage post system restore. Check user actions "
"in the generated user_action_log_file.", 80)
def tidy_storage(result_file):
"""
Search Glance images DB and rbd images pool for any discrepancy
between the two.
- If an image is in Glance images DB but not in rbd images pool,
list the image and suggested actions to take in a log file.
- If an image is in rbd images pool but not in Glance images DB,
create a Glance image in Glance images DB to associate with the
backend data. List the image and suggested actions to take in a log
file.
Search Cinder volumes DB and rbd cinder-volumes pool for any discrepancy
between the two.
- If a volume is in Cinder volumes DB but not in rbd cinder-volumes
pool, set the volume state to "error". List the volume and suggested
actions to take in a log file.
- If a volume is in rbd cinder-volumes pool but not in Cinder volumes
DB, remove any snapshot(s) assoicated with this volume in rbd pool and
create a volume in Cinder volumes DB to associate with the backend
data. List the volume and suggested actions to take in a log file.
- If a volume is in both Cinder volumes DB and rbd cinder-volumes pool
and it has snapshot(s) in the rbd pool, re-create the snapshot in
Cinder if it doesn't exist.
Clean up Cinder snapshots DB if the snapshot doesn't have backend data.
"""
with OpenStack() as client:
# Check Glance images
print("Scanning Glance images in DB and rbd images pool...\n")
try:
g_client_v1, g_client_v2 = client.get_glance_client
image_l = g_client_v2.images.list()
image_id_l = [image['id'].encode('utf-8') for image in image_l]
output = subprocess.check_output(
["rbd",
"ls",
"--pool",
"images"],
stderr=subprocess.STDOUT)
except subprocess.CalledProcessError:
LOG.error("Failed to access rbd images pool")
raise TidyStorageFail("Failed to access rbd images pool")
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to list Glance images")
rbd_image_l = [i for i in output.split('\n') if i != ""]
print("Images in Glance images DB: %s \n" % image_id_l)
print("Images in rbd images pool: %s \n" % rbd_image_l)
in_glance_only = np.setdiff1d(image_id_l, rbd_image_l)
in_rbd_image_only = np.setdiff1d(rbd_image_l, image_id_l)
print("Images in Glance images DB only: %s \n" % in_glance_only)
print("Images in rbd images pool only: %s \n" % in_rbd_image_only)
if in_rbd_image_only.size != 0:
output = subprocess.check_output(
["grep",
"fsid",
"/etc/ceph/ceph.conf"],
stderr=subprocess.STDOUT)
ceph_cluster = [i.strip() for i in output.split('=')
if i.find('fsid') == -1]
fields = dict()
for image in in_rbd_image_only:
try:
img_file = 'rbd:images/{}'.format(image)
output = subprocess.check_output(
["qemu-img", "info", img_file], stderr=subprocess.STDOUT)
fields['disk_format'] = 'qcow2'
for line in output.split('\n'):
if 'file format:' in line:
fields['disk_format'] = line.split(':')[1].strip()
break
fields['name'] = 'found-image-%s' % image
fields['id'] = image
fields['container_format'] = 'bare'
fields['location'] = \
'rbd://{}/images/{}/snap'.format(ceph_cluster[0],
image)
print ("Creating a Glance image %s ...\n " % fields['name'])
g_client_v1.images.create(**fields)
except subprocess.CalledProcessError:
LOG.error("Failed to access rbd image %s" % image)
raise TidyStorageFail("Failed to access rbd image")
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to create glance image")
# Check cinder volume snapshots. Do it before "cinder manage"
# operation as "cinder manage" does not support keeping the same
# volume id.
print("Scanning Cinder snapshots in DB and rbd cinder-volumes "
"pool...\n")
try:
c_client = client.get_cinder_client
snap_l = c_client.volume_snapshots.list(search_opts=search_opts)
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to get Cinder snapshots")
snaps_no_backend_vol = []
for snap in snap_l:
print ("Check if volume snapshot %s has backend " % snap.name)
try:
output = subprocess.check_output(
["rbd", "ls", "--pool", "cinder-volumes"],
stderr=subprocess.STDOUT)
except subprocess.CalledProcessError:
LOG.error("Failed to access rbd cinder-volumes pool")
raise TidyStorageFail(
"Failed to access rbd cinder-volumes pool")
found_vol = False
for line in output.split('\n'):
if snap.volume_id in line:
found_vol = True
break
if found_vol:
volume = 'cinder-volumes/volume-{}'.format(snap.volume_id)
try:
output = subprocess.check_output(
["rbd", "snap", "list", volume],
stderr=subprocess.STDOUT)
keep_snap = False
for line in output.split('\n'):
if snap.id in line:
keep_snap = True
break
except subprocess.CalledProcessError:
LOG.info("Failed to list snapshots for volume %s in "
"rbd cinder-volumes pool"
% snap.volume_id)
raise TidyStorageFail("Failed to list snapshots in rbd.")
if not keep_snap:
try:
print ("Volume snapshot %s has no backend data. "
"Deleting it from Cinder...\n" % snap.name)
c_client.volume_snapshots.delete(c_utils.find_resource(
c_client.volume_snapshots, snap.id), force=True)
except Exception as e:
LOG.exception(e)
raise TidyStorageFail(
"Failed to delete volume snapshot")
else:
# Volume snapshot that doesn't have backend volume cannot
# be deleted. If the backend volume is restored later, then
# the snapshot can be deleted. So for now we will add these
# snapshots in the user action log.
snaps_no_backend_vol.append(snap)
# Check Cinder volumes
print("Scanning Cinder volumes in DB and rbd cinder-volumes pool...\n")
try:
volume_l = c_client.volumes.list(search_opts=search_opts)
v_t_d = c_client.volume_types.default()
avail_zones = c_client.availability_zones.list()
pools = c_client.pools.list()
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to get Cinder volume info")
if pools:
host = pools[0].name
if v_t_d is None:
v_t_d = 'ceph'
else:
v_t_d = v_t_d.name
cinder_volume_l = [i.id.encode('utf-8') for i in volume_l]
if avail_zones:
avail_z = avail_zones[0].zoneName
try:
output = subprocess.check_output(
["rbd", "ls", "--pool", "cinder-volumes"],
stderr=subprocess.STDOUT)
except subprocess.CalledProcessError:
LOG.error("Failed to access rbd cinder-volumes pool")
raise TidyStorageFail("Failed to access rbd cinder-volumes pool")
rbd_volume_l = [i[7:] for i in output.split('\n') if i != ""]
print("Volumes in Cinder volumes DB: %s \n" % cinder_volume_l)
print("Volumes in rbd pool: %s \n" % rbd_volume_l)
in_cinder_only = np.setdiff1d(cinder_volume_l, rbd_volume_l)
in_rbd_volume_only = np.setdiff1d(rbd_volume_l, cinder_volume_l)
in_cinder_and_rbd = np.intersect1d(cinder_volume_l, rbd_volume_l)
print("Volumes in Cinder volumes DB only: %s \n" % in_cinder_only)
print("Volumes in rbd pool only: %s \n" % in_rbd_volume_only)
print("Volumes in Cinder volumes DB and rbd pool: %s \n"
% in_cinder_and_rbd)
for vol_id in in_rbd_volume_only:
volume = 'cinder-volumes/volume-{}'.format(vol_id)
try:
# Find out if the volume is a bootable one
output = subprocess.check_output(
["rbd", "info", volume],
stderr=subprocess.STDOUT)
bootable = False
for line in output.split('\n'):
if 'parent: images/' in line:
bootable = True
break
# Find out if the volume has any snapshots.
print("Checking if volume %s has snapshots...\n" % vol_id)
output = subprocess.check_output(
["rbd", "snap", "list", volume], stderr=subprocess.STDOUT)
snap_l = [item.strip() for item in output.split(' ')
if item.find('snapshot-') != -1]
# Returned volume id (vol.id) will be different from vol_id
try:
vol = c_client.volumes.manage(
host=host,
ref={'source-name': 'volume-%s' % vol_id},
name='found-volume-%s' % vol_id,
description='manage a volume',
volume_type=v_t_d,
availability_zone=avail_z,
bootable=bootable)
print("Creating volume found-volume-%s in Cinder...\n"
% vol_id)
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to manage volume")
try:
for snap in snap_l:
# Manage a snapshot for a managed volume is not
# supported in rbd. So we just remove the snapshot.
# Remove the snapshot
print (textwrap.fill(
"Removing snapshot %s from volume %s "
"in rbd...\n" % (snap, vol_id), 76))
del_snap = '{}@{}'.format(volume, snap)
output = subprocess.check_output(
["rbd", "snap", "unprotect", del_snap],
stderr=subprocess.STDOUT)
output = subprocess.check_output(
["rbd", "snap", "rm", del_snap],
stderr=subprocess.STDOUT)
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to manage volume snapshot")
except subprocess.CalledProcessError:
LOG.error("Failed to access volume %s in cinder-volumes pool"
% vol_id)
raise TidyStorageFail("Failed to access rbd image")
for vol in in_cinder_only:
try:
c_client.volumes.reset_state(
c_utils.find_volume(c_client, vol), state='error')
print("Setting state to error for volume %s \n" % vol)
except Exception as e:
LOG.error("Failed to update volume to error state for %s"
% vol)
raise TidyStorageFail("Failed to update volume to error state")
# For volumes that are in Cinder volumes DB and rbd cinder-volumes
# pool, we check if any volume snapshot needs to be re-created
try:
c_s_l = c_client.volume_snapshots.list(search_opts=search_opts)
cinder_snap_l = ['snapshot-{}'.format(snap.id) for snap in c_s_l]
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to get Cinder snapshots")
for vol_id in in_cinder_and_rbd:
volume = 'cinder-volumes/volume-{}'.format(vol_id)
try:
# Find out if the volume has any snapshots.
print("Checking if volume %s has snapshots...\n" % vol_id)
output = subprocess.check_output(
["rbd", "snap", "list", volume],
stderr=subprocess.STDOUT)
snap_l = [item.strip() for item in output.split(' ')
if item.find('snapshot-') != -1]
for snap in snap_l:
if snap not in cinder_snap_l:
print ("Creating volume snapshot found-%s "
"in Cinder...\n" % snap)
c_client.volume_snapshots.manage(
volume_id=vol_id,
ref={'source-name': snap},
name='found-%s' % snap,
description='manage a snapshot')
except subprocess.CalledProcessError:
LOG.error("Failed to access snapshot for volume %s"
% vol_id)
raise TidyStorageFail("Failed to access volume snapshot")
except Exception as e:
LOG.exception(e)
raise TidyStorageFail("Failed to manage Cinder snapshot")
try:
with open(result_file, 'w') as f:
f.write('\n%s\n' % ('-' * 80))
f.write(textwrap.fill(
"Following images are found in Ceph images pool but "
"not in Glance. These images were created after the "
"system backup was done. If you do not want to keep "
"them, you can delete them by "
"\"glance image-delete <id>\" command.", 80))
f.write("\n\n")
f.write('{0[0]:<40}{0[1]:<50}\n'.format(['ID', 'NAME']))
image_l = g_client_v2.images.list()
for image in image_l:
if image['name'].find("found-image") != -1:
f.write('{0[0]:<40}{0[1]:<50}\n'.format(
[image['id'].encode('utf-8'), image['name']]))
f.write("\n")
f.write('\n%s\n' % ('-' * 80))
f.write(textwrap.fill(
"Following images are found in Glance without backend "
"data associated with. These images were deleted after "
"the system backup was done. You can delete them by "
"\"glance image-delete <id>\" command or follow the B&R "
"document to restore the image.", 80))
f.write("\n\n")
f.write('{0[0]:<40}{0[1]:<50}\n'.format(['ID', 'NAME']))
image_l = g_client_v2.images.list()
for image in image_l:
if (in_glance_only.size != 0 and
image['id'].encode('utf-8') in in_glance_only):
f.write('{0[0]:<40}{0[1]:<50}\n'.format(
[image['id'].encode('utf-8'), image['name']]))
f.write("\n")
f.write('\n%s\n' % ('-' * 80))
f.write(textwrap.fill(
"Following volumes are found in Ceph cinder-volumes "
"pool but not in Cinder. These volumes were created "
"after the system backup was done. If you do not want "
"to keep them you can delete them by "
"\"cinder delete <id>\" command.", 80))
f.write("\n\n")
f.write('{0[0]:<40}{0[1]:<50}\n'.format(['ID', 'NAME']))
volume_l = c_client.volumes.list(search_opts=search_opts)
for volume in volume_l:
if volume.name.find("found-") != -1:
f.write('{0[0]:<40}{0[1]:<50}\n'.format(
[volume.id.encode('utf-8'), volume.name]))
f.write("\n")
f.write('\n%s\n' % ('-' * 80))
f.write(textwrap.fill(
"Following volumes are found in Cinder without backend "
"data associated with. These volumes were deleted "
"after the system backup was done. You can delete them "
"by \"cinder delete <id>\" command or follow the B&R "
"document to restore the cinder volume.", 80))
f.write("\n\n")
f.write('{0[0]:<40}{0[1]:<50}\n'.format(['ID', 'NAME']))
volume_l = c_client.volumes.list(search_opts=search_opts)
for volume in volume_l:
if (in_cinder_only.size != 0 and
volume.id in in_cinder_only):
f.write('{0[0]:<40}{0[1]:<50}\n'.format(
[volume.id.encode('utf-8'), volume.name]))
f.write("\n")
f.write('\n%s\n' % ('-' * 80))
f.write(textwrap.fill(
"Following volume snapshots are found in Ceph but not in "
"Cinder. These volume snapshots were created after the "
"system backup was done. If you do not want to keep them "
"you can delete them by \"cinder snapshot-delete <id>\" "
"command.", 80))
f.write("\n\n")
f.write('{0[0]:<40}{0[1]:<50}\n'.format(['ID', 'NAME']))
snap_l = c_client.volume_snapshots.list(
search_opts=search_opts)
for snap in snap_l:
if snap.name.find("found-") != -1:
f.write('{0[0]:<40}{0[1]:<50}\n'.format(
[snap.id.encode('utf-8'), snap.name]))
f.write("\n")
f.write('\n%s\n' % ('-' * 80))
f.write(textwrap.fill(
"Following volume snapshots are found in Cinder without "
"backend volumes. If you want to delete them, you can do "
"so by \"cinder snapshot-delete <id>\" after backend "
"volumes are restored.", 80))
f.write("\n\n")
f.write('{0[0]:<40}{0[1]:<50}\n'.format(['ID', 'NAME']))
for snap in snaps_no_backend_vol:
f.write('{0[0]:<40}{0[1]:<50}\n'.format(
[snap.id.encode('utf-8'), snap.name]))
f.write("\n\n")
except IOError:
raise TidyStorageFail("Failed to open file: %s" % result_file)
def main():
if (len(sys.argv) < 2 or
sys.argv[1] in ['--help', '-h', '-?']):
show_help()
exit(1)
log.configure()
result_file = sys.argv[1]
try:
open(result_file, 'w')
except IOError:
raise TidyStorageFail("Failed to open file: %s" % result_file)
exit(1)
tidy_storage(result_file)