def ha_relation_changed():
# Check that we are related to a principle and that
# it has already provided the required corosync configuration
if not get_corosync_conf():
log('Unable to configure corosync right now, deferring configuration',
level=INFO)
return
if relation_ids('hanode'):
log('Ready to form cluster - informing peers', level=DEBUG)
relation_set(relation_id=relation_ids('hanode')[0], ready=True)
else:
log('Ready to form cluster, but not related to peers just yet',
level=INFO)
return
# Check that there's enough nodes in order to perform the
# configuration of the HA cluster
if len(get_cluster_nodes()) < int(config('cluster_count')):
log('Not enough nodes in cluster, deferring configuration', level=INFO)
return
relids = relation_ids('ha') or relation_ids('juju-info')
if len(relids) == 1: # Should only ever be one of these
# Obtain relation information
relid = relids[0]
units = related_units(relid)
if len(units) < 1:
log('No principle unit found, deferring configuration', level=INFO)
return
unit = units[0]
log('Parsing cluster configuration using rid: %s, unit: %s' %
(relid, unit),
level=DEBUG)
resources = parse_data(relid, unit, 'resources')
delete_resources = parse_data(relid, unit, 'delete_resources')
resource_params = parse_data(relid, unit, 'resource_params')
groups = parse_data(relid, unit, 'groups')
ms = parse_data(relid, unit, 'ms')
orders = parse_data(relid, unit, 'orders')
colocations = parse_data(relid, unit, 'colocations')
clones = parse_data(relid, unit, 'clones')
locations = parse_data(relid, unit, 'locations')
init_services = parse_data(relid, unit, 'init_services')
else:
log('Related to %s ha services' % (len(relids)), level=DEBUG)
return
if True in [ra.startswith('ocf:openstack') for ra in resources.values()]:
apt_install('openstack-resource-agents')
if True in [ra.startswith('ocf:ceph') for ra in resources.values()]:
apt_install('ceph-resource-agents')
if True in [ra.startswith('ocf:maas') for ra in resources.values()]:
try:
validate_dns_ha()
except MAASConfigIncomplete as ex:
log(ex.args[0], level=ERROR)
status_set('blocked', ex.args[0])
# if an exception is raised the hook will end up in error state
# which will obfuscate the workload status and message.
return
log('Setting up access to MAAS API', level=INFO)
setup_maas_api()
# Update resource_parms for DNS resources to include MAAS URL and
# credentials
for resource in resource_params.keys():
if resource.endswith("_hostname"):
res_ipaddr = get_ip_addr_from_resource_params(
resource_params[resource])
resource_params[resource] += (
' maas_url="{}" maas_credentials="{}"'
''.format(config('maas_url'), config('maas_credentials')))
write_maas_dns_address(resource, res_ipaddr)
# NOTE: this should be removed in 15.04 cycle as corosync
# configuration should be set directly on subordinate
configure_corosync()
try_pcmk_wait()
failure_timeout = config('failure_timeout')
configure_cluster_global(failure_timeout)
configure_monitor_host()
configure_stonith()
# Only configure the cluster resources
# from the oldest peer unit.
if is_leader():
log('Setting cluster symmetry', level=INFO)
set_cluster_symmetry()
log('Deleting Resources' % (delete_resources), level=DEBUG)
for res_name in delete_resources:
if pcmk.crm_opt_exists(res_name):
if ocf_file_exists(res_name, resources):
log('Stopping and deleting resource %s' % res_name,
level=DEBUG)
if pcmk.crm_res_running(res_name):
pcmk.commit('crm -w -F resource stop %s' % res_name)
else:
log('Cleanuping and deleting resource %s' % res_name,
level=DEBUG)
pcmk.commit('crm resource cleanup %s' % res_name)
# Daemon process may still be running after the upgrade.
kill_legacy_ocf_daemon_process(res_name)
pcmk.commit('crm -w -F configure delete %s' % res_name)
log('Configuring Resources: %s' % (resources), level=DEBUG)
for res_name, res_type in resources.items():
# disable the service we are going to put in HA
if res_type.split(':')[0] == "lsb":
disable_lsb_services(res_type.split(':')[1])
if service_running(res_type.split(':')[1]):
service_stop(res_type.split(':')[1])
elif (len(init_services) != 0 and res_name in init_services
and init_services[res_name]):
disable_upstart_services(init_services[res_name])
if service_running(init_services[res_name]):
service_stop(init_services[res_name])
# Put the services in HA, if not already done so
# if not pcmk.is_resource_present(res_name):
if not pcmk.crm_opt_exists(res_name):
if res_name not in resource_params:
cmd = 'crm -w -F configure primitive %s %s' % (res_name,
res_type)
else:
cmd = ('crm -w -F configure primitive %s %s %s' %
(res_name, res_type, resource_params[res_name]))
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
if config('monitor_host'):
cmd = ('crm -F configure location Ping-%s %s rule '
'-inf: pingd lte 0' % (res_name, res_name))
pcmk.commit(cmd)
else:
# the resource already exists so it will be updated.
code = pcmk.crm_update_resource(res_name, res_type,
resource_params.get(res_name))
if code != 0:
msg = "Cannot update pcmkr resource: {}".format(res_name)
status_set('blocked', msg)
raise Exception(msg)
log('Configuring Groups: %s' % (groups), level=DEBUG)
for grp_name, grp_params in groups.items():
if not pcmk.crm_opt_exists(grp_name):
cmd = ('crm -w -F configure group %s %s' %
(grp_name, grp_params))
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Master/Slave (ms): %s' % (ms), level=DEBUG)
for ms_name, ms_params in ms.items():
if not pcmk.crm_opt_exists(ms_name):
cmd = 'crm -w -F configure ms %s %s' % (ms_name, ms_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Orders: %s' % (orders), level=DEBUG)
for ord_name, ord_params in orders.items():
if not pcmk.crm_opt_exists(ord_name):
cmd = 'crm -w -F configure order %s %s' % (ord_name,
ord_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Clones: %s' % clones, level=DEBUG)
for cln_name, cln_params in clones.items():
if not pcmk.crm_opt_exists(cln_name):
cmd = 'crm -w -F configure clone %s %s' % (cln_name,
cln_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
# Ordering is important here, colocation and location constraints
# reference resources. All resources referenced by the constraints
# need to exist otherwise constraint creation will fail.
log('Configuring Colocations: %s' % colocations, level=DEBUG)
for col_name, col_params in colocations.items():
if not pcmk.crm_opt_exists(col_name):
cmd = 'crm -w -F configure colocation %s %s' % (col_name,
col_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Locations: %s' % locations, level=DEBUG)
for loc_name, loc_params in locations.items():
if not pcmk.crm_opt_exists(loc_name):
cmd = 'crm -w -F configure location %s %s' % (loc_name,
loc_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
for res_name, res_type in resources.items():
if len(init_services) != 0 and res_name in init_services:
# Checks that the resources are running and started.
# Ensure that clones are excluded as the resource is
# not directly controllable (dealt with below)
# Ensure that groups are cleaned up as a whole rather
# than as individual resources.
if (res_name not in clones.values()
and res_name not in groups.values()
and not pcmk.crm_res_running(res_name)):
# Just in case, cleanup the resources to ensure they get
# started in case they failed for some unrelated reason.
cmd = 'crm resource cleanup %s' % res_name
pcmk.commit(cmd)
for cl_name in clones:
# Always cleanup clones
cmd = 'crm resource cleanup %s' % cl_name
pcmk.commit(cmd)
for grp_name in groups:
# Always cleanup groups
cmd = 'crm resource cleanup %s' % grp_name
pcmk.commit(cmd)
# All members of the cluster need to be registered before resources
# that reference them can be created.
if len(get_member_ready_nodes()) >= int(config('cluster_count')):
log('Configuring any remote nodes', level=INFO)
remote_resources = configure_pacemaker_remote_resources()
stonith_resource = configure_pacemaker_remote_stonith_resource()
resources.update(remote_resources)
resources.update(stonith_resource)
configure_resources_on_remotes(resources=resources,
clones=clones,
groups=groups)
else:
log('Deferring configuration of any remote nodes', level=INFO)
for rel_id in relation_ids('ha'):
relation_set(relation_id=rel_id, clustered="yes")
# Inform peers that local configuration is complete and this member
# is ready
for rel_id in relation_ids('hanode'):
relation_set(relation_id=rel_id, member_ready=True)
def test_get_member_ready_nodes(self, get_node_flags):
utils.get_member_ready_nodes()
get_node_flags.assert_called_once_with('member_ready')
def test_get_member_ready_nodes(self, get_node_flags):
utils.get_member_ready_nodes()
get_node_flags.assert_called_once_with('member_ready')
def ha_relation_changed():
# Check that we are related to a principle and that
# it has already provided the required corosync configuration
if not get_corosync_conf():
log('Unable to configure corosync right now, deferring configuration',
level=INFO)
return
if relation_ids('hanode'):
log('Ready to form cluster - informing peers', level=DEBUG)
relation_set(relation_id=relation_ids('hanode')[0], ready=True)
else:
log('Ready to form cluster, but not related to peers just yet',
level=INFO)
return
# Check that there's enough nodes in order to perform the
# configuration of the HA cluster
if len(get_cluster_nodes()) < int(config('cluster_count')):
log('Not enough nodes in cluster, deferring configuration',
level=INFO)
return
relids = relation_ids('ha') or relation_ids('juju-info')
if len(relids) == 1: # Should only ever be one of these
# Obtain relation information
relid = relids[0]
units = related_units(relid)
if len(units) < 1:
log('No principle unit found, deferring configuration',
level=INFO)
return
unit = units[0]
log('Parsing cluster configuration using rid: %s, unit: %s' %
(relid, unit), level=DEBUG)
resources = parse_data(relid, unit, 'resources')
delete_resources = parse_data(relid, unit, 'delete_resources')
resource_params = parse_data(relid, unit, 'resource_params')
groups = parse_data(relid, unit, 'groups')
ms = parse_data(relid, unit, 'ms')
orders = parse_data(relid, unit, 'orders')
colocations = parse_data(relid, unit, 'colocations')
clones = parse_data(relid, unit, 'clones')
locations = parse_data(relid, unit, 'locations')
init_services = parse_data(relid, unit, 'init_services')
else:
log('Related to %s ha services' % (len(relids)), level=DEBUG)
return
if True in [ra.startswith('ocf:openstack')
for ra in resources.values()]:
apt_install('openstack-resource-agents')
if True in [ra.startswith('ocf:ceph')
for ra in resources.values()]:
apt_install('ceph-resource-agents')
if True in [ra.startswith('ocf:maas')
for ra in resources.values()]:
try:
validate_dns_ha()
except MAASConfigIncomplete as ex:
log(ex.args[0], level=ERROR)
status_set('blocked', ex.args[0])
# if an exception is raised the hook will end up in error state
# which will obfuscate the workload status and message.
return
log('Setting up access to MAAS API', level=INFO)
setup_maas_api()
# Update resource_parms for DNS resources to include MAAS URL and
# credentials
for resource in resource_params.keys():
if resource.endswith("_hostname"):
res_ipaddr = get_ip_addr_from_resource_params(
resource_params[resource])
resource_params[resource] += (
' maas_url="{}" maas_credentials="{}"'
''.format(config('maas_url'),
config('maas_credentials')))
write_maas_dns_address(resource, res_ipaddr)
# NOTE: this should be removed in 15.04 cycle as corosync
# configuration should be set directly on subordinate
configure_corosync()
try_pcmk_wait()
configure_cluster_global()
configure_monitor_host()
configure_stonith()
# Only configure the cluster resources
# from the oldest peer unit.
if is_leader():
log('Setting cluster symmetry', level=INFO)
set_cluster_symmetry()
log('Deleting Resources' % (delete_resources), level=DEBUG)
for res_name in delete_resources:
if pcmk.crm_opt_exists(res_name):
if ocf_file_exists(res_name, resources):
log('Stopping and deleting resource %s' % res_name,
level=DEBUG)
if pcmk.crm_res_running(res_name):
pcmk.commit('crm -w -F resource stop %s' % res_name)
else:
log('Cleanuping and deleting resource %s' % res_name,
level=DEBUG)
pcmk.commit('crm resource cleanup %s' % res_name)
# Daemon process may still be running after the upgrade.
kill_legacy_ocf_daemon_process(res_name)
pcmk.commit('crm -w -F configure delete %s' % res_name)
log('Configuring Resources: %s' % (resources), level=DEBUG)
for res_name, res_type in resources.items():
# disable the service we are going to put in HA
if res_type.split(':')[0] == "lsb":
disable_lsb_services(res_type.split(':')[1])
if service_running(res_type.split(':')[1]):
service_stop(res_type.split(':')[1])
elif (len(init_services) != 0 and
res_name in init_services and
init_services[res_name]):
disable_upstart_services(init_services[res_name])
if service_running(init_services[res_name]):
service_stop(init_services[res_name])
# Put the services in HA, if not already done so
# if not pcmk.is_resource_present(res_name):
if not pcmk.crm_opt_exists(res_name):
if res_name not in resource_params:
cmd = 'crm -w -F configure primitive %s %s' % (res_name,
res_type)
else:
cmd = ('crm -w -F configure primitive %s %s %s' %
(res_name, res_type, resource_params[res_name]))
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
if config('monitor_host'):
cmd = ('crm -F configure location Ping-%s %s rule '
'-inf: pingd lte 0' % (res_name, res_name))
pcmk.commit(cmd)
else:
# the resource already exists so it will be updated.
code = pcmk.crm_update_resource(res_name, res_type,
resource_params.get(res_name))
if code != 0:
msg = "Cannot update pcmkr resource: {}".format(res_name)
status_set('blocked', msg)
raise Exception(msg)
log('Configuring Groups: %s' % (groups), level=DEBUG)
for grp_name, grp_params in groups.items():
if not pcmk.crm_opt_exists(grp_name):
cmd = ('crm -w -F configure group %s %s' %
(grp_name, grp_params))
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Master/Slave (ms): %s' % (ms), level=DEBUG)
for ms_name, ms_params in ms.items():
if not pcmk.crm_opt_exists(ms_name):
cmd = 'crm -w -F configure ms %s %s' % (ms_name, ms_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Orders: %s' % (orders), level=DEBUG)
for ord_name, ord_params in orders.items():
if not pcmk.crm_opt_exists(ord_name):
cmd = 'crm -w -F configure order %s %s' % (ord_name,
ord_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Clones: %s' % clones, level=DEBUG)
for cln_name, cln_params in clones.items():
if not pcmk.crm_opt_exists(cln_name):
cmd = 'crm -w -F configure clone %s %s' % (cln_name,
cln_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
# Ordering is important here, colocation and location constraints
# reference resources. All resources referenced by the constraints
# need to exist otherwise constraint creation will fail.
log('Configuring Colocations: %s' % colocations, level=DEBUG)
for col_name, col_params in colocations.items():
if not pcmk.crm_opt_exists(col_name):
cmd = 'crm -w -F configure colocation %s %s' % (col_name,
col_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
log('Configuring Locations: %s' % locations, level=DEBUG)
for loc_name, loc_params in locations.items():
if not pcmk.crm_opt_exists(loc_name):
cmd = 'crm -w -F configure location %s %s' % (loc_name,
loc_params)
pcmk.commit(cmd)
log('%s' % cmd, level=DEBUG)
for res_name, res_type in resources.items():
if len(init_services) != 0 and res_name in init_services:
# Checks that the resources are running and started.
# Ensure that clones are excluded as the resource is
# not directly controllable (dealt with below)
# Ensure that groups are cleaned up as a whole rather
# than as individual resources.
if (res_name not in clones.values() and
res_name not in groups.values() and
not pcmk.crm_res_running(res_name)):
# Just in case, cleanup the resources to ensure they get
# started in case they failed for some unrelated reason.
cmd = 'crm resource cleanup %s' % res_name
pcmk.commit(cmd)
for cl_name in clones:
# Always cleanup clones
cmd = 'crm resource cleanup %s' % cl_name
pcmk.commit(cmd)
for grp_name in groups:
# Always cleanup groups
cmd = 'crm resource cleanup %s' % grp_name
pcmk.commit(cmd)
# All members of the cluster need to be registered before resources
# that reference them can be created.
if len(get_member_ready_nodes()) >= int(config('cluster_count')):
log('Configuring any remote nodes', level=INFO)
remote_resources = configure_pacemaker_remote_resources()
stonith_resource = configure_pacemaker_remote_stonith_resource()
resources.update(remote_resources)
resources.update(stonith_resource)
configure_resources_on_remotes(
resources=resources,
clones=clones,
groups=groups)
else:
log('Deferring configuration of any remote nodes', level=INFO)
for rel_id in relation_ids('ha'):
relation_set(relation_id=rel_id, clustered="yes")
# Inform peers that local configuration is complete and this member
# is ready
for rel_id in relation_ids('hanode'):
relation_set(relation_id=rel_id, member_ready=True)