def add_ceph_mon(cluster_name, minions):
conf_file = _CEPH_CLUSTER_CONF_DIR + "/" + cluster_name + "/" + cluster_name + ".conf"
config = ConfigParser.RawConfigParser()
config.read(conf_file)
public_network = IPNetwork(config.get("global", "public network"))
check_minion_networks(minions, public_network)
used_mon_ids = set([id.strip() for id in config.get("mon", "mon initial members").split(",")])
unused_mon_ids = list(set(_MON_ID_LIST) - used_mon_ids)
unused_mon_ids.sort()
mon_id_map, monitors = _get_mon_id_map(unused_mon_ids, minions)
mon_initial_members = list(used_mon_ids) + list(monitors)
mon_initial_members.sort()
config.set("mon", "mon initial members", ", ".join(mon_initial_members))
_config_add_monitors(config, monitors)
with open(conf_file, "wb") as f:
config.write(f)
pillar_data = _add_ceph_mon_pillar_data(mon_id_map, cluster_name, monitors)
pillar = {"usm": pillar_data}
out = run_state(local, minions, "add_ceph_mon", expr_form="list", kwarg={"pillar": pillar})
if out:
return out
return sync_ceph_conf(cluster_name, minions)
def AddMon(cluster_name, minions, ctxt=""):
# convert list of minions to below dict
# {MINION_ID: {'public_ip': IP_ADDRESS,
# 'cluster_ip': IP_ADDRESS}, ...}
d = {}
for m in minions:
d.update({m['Node']: {'public_ip': m['PublicIP4'],
'cluster_ip': m['ClusterIP4']}})
minion_set = minions
minions = d
conf_file = (_CEPH_CLUSTER_CONF_DIR + '/' + cluster_name + '/' +
cluster_name + '.conf')
config = ConfigParser.RawConfigParser()
config.read(conf_file)
public_network = IPNetwork(config.get('global', 'public network'))
_check_minion_networks(minions, public_network)
used_mon_ids = set([id.strip() for id in config.get(
'mon', 'mon initial members').split(',')])
unused_mon_ids = list(set(_MON_ID_LIST) - used_mon_ids)
unused_mon_ids.sort()
mon_id_map, monitors = _get_mon_id_map(unused_mon_ids, minions)
mon_initial_members = list(used_mon_ids) + list(monitors)
mon_initial_members.sort()
config.set('mon', 'mon initial members', ', '.join(mon_initial_members))
_config_add_monitors(config, monitors)
with open(conf_file, 'wb') as f:
config.write(f)
pillar_data = _add_ceph_mon_pillar_data(mon_id_map, cluster_name, monitors)
pillar = {'skyring': pillar_data}
local = salt.client.LocalClient()
out = run_state(local, minions, 'add_ceph_mon', expr_form='list',
kwarg={'pillar': pillar})
if out:
log.error('%s-add_mon failed for %s. error=%s' %
(ctxt, minion_set, out))
raise Exception('add_mon failed for %s. error=%s' %
(minion_set, out))
out = sync_ceph_conf(cluster_name, minions)
if out:
log.error("%s-sync_ceph_conf failed to %s. error=%s" %
(ctxt, minions, out))
raise Exception("sync_ceph_conf failed to %s. error=%s" %
(minions, out))
return True
def verify_config(cls, config):
prepend_root_dirs = []
for config_key in ("log_file", ):
if urllib.parse.urlparse(config.get(config_key, "")).scheme == "":
prepend_root_dirs.append(config_key)
if prepend_root_dirs:
salt.config.prepend_root_dir(config, prepend_root_dirs)
salt.utils.verify.verify_env(
[str(pathlib.Path(config["log_file"]).parent)],
running_username(),
pki_dir=config.get("pki_dir") or "",
root_dir=config["root_dir"],
)
def verify_config(cls, config):
salt.utils.verify.verify_env(
[str(pathlib.Path(config["log_file"]).parent)],
running_username(),
pki_dir=config.get("pki_dir") or "",
root_dir=config["root_dir"],
)
def list_server_logs(self, request, fqdn):
client = salt.client.LocalClient(
config.get('cthulhu', 'salt_config_path'))
results = client.cmd(fqdn, "log_tail.list_logs", ["."])
if not results:
return Response(status=status.HTTP_503_SERVICE_UNAVAILABLE)
return Response(sorted(results[fqdn]))
def get_server_log(self, request, fqdn, log_path):
lines = request.GET.get('lines', 40)
client = salt.client.LocalClient(
config.get('cthulhu', 'salt_config_path'))
results = client.cmd(fqdn, "log_tail.tail", [log_path, lines])
if not results:
return Response(status=status.HTTP_503_SERVICE_UNAVAILABLE)
else:
return Response({'lines': results[fqdn]})
def _get_cluster_uuid_from_name(cluster_name):
configfile = "/etc/ceph/%s.conf" % (cluster_name)
if not os.path.isfile(configfile):
raise Error("Cluster confg file does not exist:'%s'" % configfile)
config = ConfigParser.ConfigParser()
config.read(configfile)
try:
fsid = config.get("global","fsid")
except ConfigParser.NoOptionError:
raise Error("Cluster confg file does not sewt fsid:'%s'" % configfile)
return fsid
def _get_cluster_uuid_from_name(cluster_name):
configfile = "/etc/ceph/%s.conf" % (cluster_name)
if not os.path.isfile(configfile):
raise Error("Cluster confg file does not exist:'%s'" % configfile)
config = ConfigParser.ConfigParser()
config.read(configfile)
try:
fsid = config.get("global", "fsid")
except ConfigParser.NoOptionError:
raise Error("Cluster confg file does not sewt fsid:'%s'" % configfile)
return fsid
def _get_cluster_name_from_uuid(cluster_uuid):
output = None
dir_content = os.listdir("/etc/ceph/")
for file_name in dir_content:
if file_name[-5:] != ".conf":
continue
fullpath = os.path.join("/etc/ceph/", file_name)
config = ConfigParser()
config.read(fullpath)
try:
fsid = config.get("global","fsid")
if fsid is not None:
output = file_name[:-5]
except:
continue
return output
def apply_fuse_config(overrides, defaults=None):
if defaults is None:
defaults = DEFAULT_FUSE_OPTS
config = defaults.copy()
if overrides:
config.update(overrides)
# set up the extension_modules location from the cachedir
config['extension_modules'] = (
config.get('extension_modules') or
os.path.join(config['cachedir'], 'extmods')
)
return config
def _get_cluster_name_from_uuid(cluster_uuid):
output = None
dir_content = os.listdir("/etc/ceph/")
for file_name in dir_content:
if file_name[-5:] != ".conf":
continue
fullpath = os.path.join("/etc/ceph/", file_name)
config = ConfigParser()
config.read(fullpath)
try:
fsid = config.get("global", "fsid")
if fsid is not None:
output = file_name[:-5]
except:
continue
return output
def retrieve_grains(self, request, fqdn):
salt_config = salt.config.client_config(
config.get('cthulhu', 'salt_config_path'))
pillar_util = salt.utils.master.MasterPillarUtil(
fqdn,
'glob',
use_cached_grains=True,
grains_fallback=False,
opts=salt_config)
try:
# We (ab)use an internal interface to get at the cache by minion ID
# instead of by glob, because the process of resolving the glob
# relies on access to the root only PKI folder.
cache_grains, cache_pillar = pillar_util._get_cached_minion_data(
fqdn)
return Response(cache_grains[fqdn])
except KeyError:
return Response(status=status.HTTP_404_NOT_FOUND)
def get_cluster_log(self, request, fsid):
"""
Retrieve the cluster log from one of a cluster's mons (expect it to be in /var/log/ceph/ceph.log)
"""
lines = request.GET.get('lines', 40)
# Resolve FSID to name
name = self.client.get_cluster(fsid)['name']
# Resolve FSID to list of mon FQDNs
servers = self.client.server_list_cluster(fsid)
# Sort to get most recently contacted server first; drop any
# for whom last_contact is None
servers = [s for s in servers if s['last_contact']]
servers = sorted(servers,
key=lambda t: dateutil_parse(t['last_contact']),
reverse=True)
mon_fqdns = []
for server in servers:
for service in server['services']:
service_id = ServiceId(*(service['id']))
if service[
'running'] and service_id.service_type == MON and service_id.fsid == fsid:
mon_fqdns.append(server['fqdn'])
client = salt.client.LocalClient(
config.get('cthulhu', 'salt_config_path'))
log.debug("LogTailViewSet: mons for %s are %s" % (fsid, mon_fqdns))
# For each mon FQDN, try to go get ceph/$cluster.log, if we succeed return it, if we fail try the next one
# NB this path is actually customizable in ceph as `mon_cluster_log_file` but we assume user hasn't done that.
for mon_fqdn in mon_fqdns:
results = client.cmd(mon_fqdn, "log_tail.tail",
["ceph/{name}.log".format(name=name), lines])
if results:
return Response({'lines': results[mon_fqdn]})
else:
log.info("Failed to get log from %s" % mon_fqdn)
# If none of the mons gave us what we wanted, return a 503 service unavailable
return Response("mon log data unavailable",
status=status.HTTP_503_SERVICE_UNAVAILABLE)
def _lookup_ifaces(self, servers):
"""
Resolve the frontend/backend addresses (known
by cthulhu via Ceph) to network interfaces (known by salt from its
grains).
"""
salt_config = salt.config.client_config(
config.get('cthulhu', 'salt_config_path'))
pillar_util = salt.utils.master.MasterPillarUtil(
'',
'glob',
use_cached_grains=True,
grains_fallback=False,
opts=salt_config)
def _lookup_one(server):
log.debug(">> resolving grains for server {0}".format(
server['fqdn']))
fqdn = server['fqdn']
cache_grains, cache_pillar = pillar_util._get_cached_minion_data(
fqdn)
server['frontend_iface'] = None
server['backend_iface'] = None
try:
grains = cache_grains[fqdn]
if server['frontend_addr']:
server['frontend_iface'] = self._addr_to_iface(
server['frontend_addr'], grains['ip_interfaces'])
if server['backend_addr']:
server['backend_iface'] = self._addr_to_iface(
server['backend_addr'], grains['ip_interfaces'])
except KeyError:
pass
log.debug("<< resolving grains for server {0}".format(
server['fqdn']))
# Issue up to this many disk I/Os to load grains at once
CONCURRENT_GRAIN_LOADS = 16
p = gevent.pool.Pool(CONCURRENT_GRAIN_LOADS)
p.map(_lookup_one, servers)
def _salt_key(self):
return Key(salt.config.master_config(config.get('cthulhu', 'salt_config_path')))
def AddOSD(cluster_name, minions, ctxt=""):
# convert minions dict to below dict
# {MINION_ID: {'public_ip': IP_ADDRESS,
# 'cluster_ip': IP_ADDRESS,
# 'devices': {DEVICE: FSTYPE, ...}}, ...}
d = {minions['Node']: {'public_ip': minions['PublicIP4'],
'cluster_ip': minions['ClusterIP4'],
'devices': {
minions['Device']: minions['FSType'],
}}}
minions = d
conf_file = (_CEPH_CLUSTER_CONF_DIR + '/' + cluster_name + '/' +
cluster_name + '.conf')
config = ConfigParser.RawConfigParser()
config.read(conf_file)
public_network = IPNetwork(config.get('global', 'public network'))
if config.has_option('global', 'cluster network'):
cluster_network = IPNetwork(config.get('global', 'cluster network'))
else:
cluster_network = None
public_network, cluster_network = _check_minion_networks(
minions, public_network, cluster_network, check_cluster_network=True)
pillar_data = {}
for minion, v in minions.iteritems():
pillar_data[minion] = {'cluster_name': cluster_name,
'cluster_id': config.get('global', 'fsid'),
'devices': v['devices']}
pillar = {'skyring': pillar_data}
local = salt.client.LocalClient()
out = run_state(local, minions, 'prepare_ceph_osd', expr_form='list',
kwarg={'pillar': pillar})
if out:
log.error("%s-prepare_osd failed for %s. error=%s" %
(ctxt, minions, out))
raise Exception("prepare_osd failed for %s. error=%s" %
(minions, out))
for minion, v in minions.iteritems():
count = 0
found = False
failed_devices = []
while count < 6:
out = local.cmd(minion, 'cmd.run_all', ['ls -l /dev/disk/by-parttypeuuid'])
time.sleep(15)
for key, value in v['devices'].iteritems():
val_to_check = key.split('/')[-1]
found = False
for line in out[minion]["stdout"].splitlines():
if val_to_check in line:
found = True
if key in failed_devices:
failed_devices.remove(key)
break
if not found:
if key not in failed_devices:
failed_devices.append(key)
break
if found:
break
count += 1
if len(failed_devices) != 0:
log.error("%s-prepare_osd failed for %s" % (ctxt, failed_devices))
raise Exception("prepare_osd failed for %s" % failed_devices)
out = local.cmd(minions, 'cmd.run_all', ['ceph-disk activate-all'],
expr_form='list')
osd_map = {}
failed_minions = {}
for minion, v in out.iteritems():
osds = []
if v.get('retcode') != 0:
failed_minions[minion] = v
continue
for line in v['stdout'].splitlines():
if line.startswith('=== '):
osds.append(line.split('=== ')[1].strip())
break
osd_map[minion] = osds
config.set('global', 'cluster network', cluster_network)
for minion, osds in osd_map.iteritems():
name = _get_short_hostname(minion)
for osd in osds:
config.add_section(osd)
config.set(osd, 'host', name)
config.set(osd, 'public addr', minions[minion]['public_ip'])
config.set(osd, 'cluster addr', minions[minion]['cluster_ip'])
with open(conf_file, 'wb') as f:
config.write(f)
out = sync_ceph_conf(cluster_name, minions)
if out:
log.error("%s-sync_cepH-conf failed for %s. error=%s" %
(ctxt, minions, out))
#raise Exception("sync_ceph_conf failed for %s. error=%s" %
# (minions, out))
if failed_minions:
log.error('%s-add_osd failed. error=%s' % (ctxt, failed_minions))
raise Exception('add_osd failed. error=%s' % failed_minions)
return osd_map
def bind(self):
log.info("%s bind..." % self.__class__.__name__)
self._server.bind(config.get('cthulhu', 'rpc_url'))
self._bound = True
from calamari_common.config import CalamariConfig
import salt.config
# A config instance for use from within the manager service
config = CalamariConfig()
# A salt config instance for places we'll need the sock_dir
salt_config = salt.config.client_config(config.get('cthulhu', 'salt_config_path'))
def bind(self):
log.info("%s bind..." % self.__class__.__name__)
self._server.bind(config.get('cthulhu', 'rpc_url'))
self._bound = True
def deploy(source, update_type, versions, **kwargs):
'''
Updates all installed binary packages of the source package
to the specified version.
source : Name of the source package
update_type : tool | library and others, see doc/readme.txt
versions : A dictionary of distros and the version to be installed,
e.g. jessie : 1.0-1.
If the distro isn't used, no update is performed
'''
pending_restarts_pre = set()
pending_restarts_post = set()
blacklisted_packages = []
installed_distro = grains['oscodename']
if versions.get(installed_distro, None) == None:
log.info("Update doesn't apply to the installed distribution (" + installed_distro + ")")
return {}
if os.path.exists("/etc/debdeploy-minion.conf"):
config = ConfigParser.ConfigParser()
config.read("/etc/debdeploy-minion.conf")
if config.has_section("blacklist-" + installed_distro):
if config.has_option("blacklist-" + installed_distro, source):
blacklisted_packages = [x.strip() for x in config.get("blacklist-" + installed_distro, source).split(",")]
log.info("Packages blacklisted for upgrades: " + str(blacklisted_packages))
# Detect all locally installed binary packages of a given source package
# The only resource we can use for that is parsing the /var/lib/dpkg/status
# file. The format is a bit erratic: The Source: line is only present for
# binary packages not having the same name as the binary package
installed_binary_packages = []
for pkg in deb822.Packages.iter_paragraphs(file('/var/lib/dpkg/status')):
# skip packages in deinstalled status ("rc" in dpkg). These are not relevant for
# upgrades and cause problems when binary package names have changed (since package
# installations are forced with a specific version which is not available for those
# outdated binary package names)
installation_status = pkg['Status'].split()[0]
if installation_status == "deinstall":
continue
if pkg.has_key('Package') and pkg.get('Package') in blacklisted_packages:
log.info('Package ' + pkg.get('Package') + ' has been blacklisted for installation')
continue
# Source packages which have had a binNMU have a Source: entry with the source
# package version in brackets, so strip these
# If no Source: entry is present in /var/lib/dpkg/status, then the source package
# name is identical to the binary package name
if pkg.has_key('Source') and re.sub(r'\(.*?\)', '', pkg['Source']).strip() == source:
installed_binary_packages.append({pkg['Package'] : versions[installed_distro]})
elif pkg.has_key('Package') and pkg['Package'] == source:
installed_binary_packages.append({pkg['Package'] : versions[installed_distro]})
log.debug("Installed binary packages for " + source + ": " + str(installed_binary_packages))
if len(installed_binary_packages) == 0:
log.info("No binary packages installed for source package " + source)
return {}
if update_type == "library":
pending_restarts_pre = Checkrestart().get_programs_to_restart()
log.debug("Packages needing a restart prior to the update:" + str(pending_restarts_pre))
old = list_pkgs()
log.warn("Refreshing apt package database")
log.info("Refreshing apt package database")
__salt__['pkg.refresh_db']
apt_call = install_pkgs(installed_binary_packages)
new = list_pkgs()
if update_type == "library":
pending_restarts_post = Checkrestart().get_programs_to_restart()
log.debug("Packages needing a restart after to the update:" + str(pending_restarts_post))
old_keys = set(old.keys())
new_keys = set(new.keys())
additions = []
removals = []
updated = []
restarts = []
new_restarts = []
if update_type == "library":
restarts = list(pending_restarts_post)
new_restarts = list(pending_restarts_post.difference(pending_restarts_pre))
for i in new_keys.difference(old_keys):
additions.append[i]
for i in old_keys.difference(new_keys):
removals.append[i]
intersect = old_keys.intersection(new_keys)
modified = {x : (old[x], new[x]) for x in intersect if old[x] != new[x]}
log.info("Newly installed packages:" + str(additions))
log.info("Removed packages: " + str(removals))
log.info("Modified packages: " + str(modified))
log.info("Packages needing a restart: " + str(restarts))
log.info("New packages needing a restart: " + str(new_restarts))
r = {}
r["additions"] = additions
r["removals"] = removals
r["updated"] = modified
r["new_restart"] = new_restarts
r["restart"] = restarts
r["aptlog"] = str(apt_call['stdout'])
r["apterrlog"] = str(apt_call['stderr'])
r["aptreturn"] = apt_call['retcode']
jobid = kwargs.get('__pub_jid')
with open("/var/lib/debdeploy/" + jobid + ".job", "w") as jobfile:
pickle.dump(r, jobfile)
return r
def add_ceph_osd(cluster_name, minions):
"""
:: minions = {MINION_ID: {'public_ip': IP_ADDRESS,
'cluster_ip': IP_ADDRESS,
'host_name': HOSTNAME,
'devices': {DEVICE: FSTYPE, ...}}, ...}
"""
conf_file = _CEPH_CLUSTER_CONF_DIR + "/" + cluster_name + "/" + cluster_name + ".conf"
config = ConfigParser.RawConfigParser()
config.read(conf_file)
public_network = IPNetwork(config.get("global", "public network"))
if config.has_option("global", "cluster network"):
cluster_network = IPNetwork(config.get("global", "cluster network"))
else:
cluster_network = None
public_network, cluster_network = check_minion_networks(
minions, public_network, cluster_network, check_cluster_network=True
)
pillar_data = {}
for minion, v in minions.iteritems():
pillar_data[minion] = {
"cluster_name": cluster_name,
"cluster_id": config.get("global", "fsid"),
"devices": v["devices"],
}
pillar = {"usm": pillar_data}
out = run_state(local, minions, "prepare_ceph_osd", expr_form="list", kwarg={"pillar": pillar})
if out:
return out
out = local.cmd(minions, "cmd.run_all", ["ceph-disk activate-all"], expr_form="list")
osd_map = {}
failed_minions = {}
for minion, v in out.iteritems():
osds = []
if v.get("retcode") != 0:
failed_minions[minion] = v
continue
for line in v["stdout"].splitlines():
if line.startswith("=== "):
osds.append(line.split("=== ")[1].strip())
break
osd_map[minion] = osds
config.set("global", "cluster network", cluster_network)
for minion, osds in osd_map.iteritems():
name = minions[minion].get("host_name", utils.get_short_hostname(minion))
for osd in osds:
config.add_section(osd)
config.set(osd, "host", name)
config.set(osd, "public addr", minions[minion]["public_ip"])
config.set(osd, "cluster addr", minions[minion]["cluster_ip"])
with open(conf_file, "wb") as f:
config.write(f)
sync_ceph_conf(cluster_name, minions)
return failed_minions
def get_queue_size():
credentials = pika.PlainCredentials(RABBIT_USERNAME, RABBIT_PASSWORD)
connection = pika.BlockingConnection(
pika.ConnectionParameters(RABBIT_HOST, RABBIT_PORT, RABBIT_VHOST,
credentials))
channel = connection.channel()
result = channel.queue_declare(queue=HOST_ID)
return result.method.message_count
if __name__ == "__main__":
Watcher()
config = ConfigParser.SafeConfigParser()
config.read('dockerm-minion.conf')
HOST_ID = salt.config.minion_config('/etc/salt/minion')['id']
RABBIT_USERNAME = config.get('rabbitmq', 'RABBIT_USERNAME')
RABBIT_PASSWORD = config.get('rabbitmq', 'RABBIT_PASSWORD')
RABBIT_HOST = config.get('rabbitmq', 'RABBIT_HOST')
RABBIT_PORT = config.getint('rabbitmq', 'RABBIT_PORT')
RABBIT_VHOST = config.get('rabbitmq', 'RABBIT_VHOST')
while 1:
queue_size = get_queue_size()
if queue_size > 10:
thread_num = 5
else:
thread_num = 2
if queue_size > 0:
t = []
for i in range(thread_num):
t.append(threading.Thread(target=main, args=()))
t[i].setDaemon(True)
from calamari_common.config import CalamariConfig
import salt.config
# A config instance for use from within the manager service
config = CalamariConfig()
# A salt config instance for places we'll need the sock_dir
salt_config = salt.config.client_config(
config.get('cthulhu', 'salt_config_path'))
def _salt_key(self):
return Key(
salt.config.master_config(config.get('cthulhu',
'salt_config_path')))
def AddMon(cluster_name, minions, ctxt=""):
# convert list of minions to below dict
# {MINION_ID: {'public_ip': IP_ADDRESS,
# 'cluster_ip': IP_ADDRESS}, ...}
d = {}
for m in minions:
d.update({
m['Node']: {
'public_ip': m['PublicIP4'],
'cluster_ip': m['ClusterIP4']
}
})
minion_set = minions
minions = d
conf_file = (_CEPH_CLUSTER_CONF_DIR + '/' + cluster_name + '/' +
cluster_name + '.conf')
config = ConfigParser.RawConfigParser()
config.read(conf_file)
public_network = IPNetwork(config.get('global', 'public network'))
_check_minion_networks(minions, public_network)
used_mon_ids = set([
id.strip()
for id in config.get('mon', 'mon initial members').split(',')
])
unused_mon_ids = list(set(_MON_ID_LIST) - used_mon_ids)
unused_mon_ids.sort()
mon_id_map, monitors = _get_mon_id_map(unused_mon_ids, minions)
mon_initial_members = list(used_mon_ids) + list(monitors)
mon_initial_members.sort()
config.set('mon', 'mon initial members', ', '.join(mon_initial_members))
_config_add_monitors(config, monitors)
with open(conf_file, 'wb') as f:
config.write(f)
pillar_data = _add_ceph_mon_pillar_data(mon_id_map, cluster_name, monitors)
pillar = {'skyring': pillar_data}
local = salt.client.LocalClient()
out = run_state(local,
minions,
'add_ceph_mon',
expr_form='list',
kwarg={'pillar': pillar})
if out:
log.error('%s-add_mon failed for %s. error=%s' %
(ctxt, minion_set, out))
raise Exception('add_mon failed for %s. error=%s' % (minion_set, out))
out = sync_ceph_conf(cluster_name, minions)
if out:
log.error("%s-sync_ceph_conf failed to %s. error=%s" %
(ctxt, minions, out))
raise Exception("sync_ceph_conf failed to %s. error=%s" %
(minions, out))
return True
def AddOSD(cluster_name, minions, ctxt=""):
# convert minions dict to below dict
# {MINION_ID: {'public_ip': IP_ADDRESS,
# 'cluster_ip': IP_ADDRESS,
# 'devices': {DEVICE: FSTYPE, ...}}, ...}
d = {
minions['Node']: {
'public_ip': minions['PublicIP4'],
'cluster_ip': minions['ClusterIP4'],
'devices': {
minions['Device']: minions['FSType'],
}
}
}
minions = d
conf_file = (_CEPH_CLUSTER_CONF_DIR + '/' + cluster_name + '/' +
cluster_name + '.conf')
config = ConfigParser.RawConfigParser()
config.read(conf_file)
public_network = IPNetwork(config.get('global', 'public network'))
if config.has_option('global', 'cluster network'):
cluster_network = IPNetwork(config.get('global', 'cluster network'))
else:
cluster_network = None
public_network, cluster_network = _check_minion_networks(
minions, public_network, cluster_network, check_cluster_network=True)
pillar_data = {}
for minion, v in minions.iteritems():
pillar_data[minion] = {
'cluster_name': cluster_name,
'cluster_id': config.get('global', 'fsid'),
'devices': v['devices']
}
pillar = {'skyring': pillar_data}
local = salt.client.LocalClient()
out = run_state(local,
minions,
'prepare_ceph_osd',
expr_form='list',
kwarg={'pillar': pillar})
if out:
log.error("%s-prepare_osd failed for %s. error=%s" %
(ctxt, minions, out))
raise Exception("prepare_osd failed for %s. error=%s" % (minions, out))
for minion, v in minions.iteritems():
count = 0
found = False
failed_devices = []
while count < 6:
out = local.cmd(minion, 'cmd.run_all',
['ls -l /dev/disk/by-parttypeuuid'])
time.sleep(15)
for key, value in v['devices'].iteritems():
val_to_check = key.split('/')[-1]
found = False
for line in out[minion]["stdout"].splitlines():
if val_to_check in line:
found = True
if key in failed_devices:
failed_devices.remove(key)
break
if not found:
if key not in failed_devices:
failed_devices.append(key)
break
if found:
break
count += 1
if len(failed_devices) != 0:
log.error("%s-prepare_osd failed for %s" % (ctxt, failed_devices))
raise Exception("prepare_osd failed for %s" % failed_devices)
out = local.cmd(minions,
'cmd.run_all', ['ceph-disk activate-all'],
expr_form='list')
osd_map = {}
failed_minions = {}
for minion, v in out.iteritems():
osds = []
if v.get('retcode') != 0:
failed_minions[minion] = v
continue
for line in v['stdout'].splitlines():
if line.startswith('=== '):
osds.append(line.split('=== ')[1].strip())
break
osd_map[minion] = osds
config.set('global', 'cluster network', cluster_network)
for minion, osds in osd_map.iteritems():
name = _get_short_hostname(minion)
for osd in osds:
config.add_section(osd)
config.set(osd, 'host', name)
config.set(osd, 'public addr', minions[minion]['public_ip'])
config.set(osd, 'cluster addr', minions[minion]['cluster_ip'])
with open(conf_file, 'wb') as f:
config.write(f)
out = sync_ceph_conf(cluster_name, minions)
if out:
log.error("%s-sync_cepH-conf failed for %s. error=%s" %
(ctxt, minions, out))
#raise Exception("sync_ceph_conf failed for %s. error=%s" %
# (minions, out))
if failed_minions:
log.error('%s-add_osd failed. error=%s' % (ctxt, failed_minions))
raise Exception('add_osd failed. error=%s' % failed_minions)
return osd_map
if args.config:
# load config from file
with open(args.config, 'r') as stream:
config = ruamel.yaml.YAML().load(stream)
elif select.select([
sys.stdin,
], [], [], 0.0)[0]:
# expect wheelhouse yaml structured metadata
config = ruamel.yaml.YAML().load(sys.stdin)
jobs = args.jobs if args.jobs else jobs
Toolbox.merge_dict(config, config_override)
# Make it happen
if len(jobs) > 1 and len(config) > 1:
if config.get('engine', '') == 'salt':
# trigge Salt wheel
SaltWheel(config, jobs=jobs).runner()
elif True:
pass
#TODO, should we go over some errors?
#RED_ERROR = termcolor.colored('FATAL ERROR:', 'red')
#if args.debug:
#else:
# try:
# run(args)
# except (errors.UserException, errors.BuildError) as exc:
# print(RED_ERROR, exc.args[0], file=sys.stderr)
# sys.exit(exc.CODE)